The Putative Glutathione Peroxidase Gene of Plasmodium falciparum Codes for a Thioredoxin Peroxidase*

A putative glutathione peroxidase gene (Swiss-Prot accession number Z 68200) of Plasmodium falciparum, the causative agent of tropical malaria, was expressed in Escherichia coli and purified to electrophoretic homogeneity. Like phospholipid hydroperoxide glutathione peroxidase of mammals, it proved to be monomeric. It was active with H2O2 and organic hydroperoxides but, unlike phospholipid hydroperoxide glutathione peroxidase, not with phosphatidylcholine hydroperoxide. With glutathione peroxidases it shares the ping-pong mechanism with infiniteV max and K m when analyzed with GSH as substrate. As a homologue with selenocysteine replaced by cysteine, its reactions with hydroperoxides and GSH are 3 orders of magnitude slower than those of the selenoperoxidases. Unexpectedly, the plasmodial enzyme proved to react faster with thioredoxins than with GSH and most efficiently with thioredoxin of P. falciparum(Swiss-Prot accession number 202664). It is therefore reclassified as thioredoxin peroxidase. With plasmodial thioredoxin, the enzyme also displays ping-pong kinetics, yet with a limiting K m of 10 μm and a k 1′ of 0.55 s− 1. The apparent k 1′ for oxidation with cumene, t-butyl, and hydrogen peroxides are 2.0 × 104 m − 1 s− 1, 3.3 × 103 m − 1s− 1, and 2.5 × 103 m − 1 s− 1, respectively. k 2′ for reduction by autologous thioredoxin is 5.4 × 104 m -1s− 1 (21.2m − 1 s− 1for GSH). The newly discovered enzymatic function of the plasmodial gene product suggests a reconsideration of its presumed role in parasitic antioxidant defense.

Plasmodium falciparum, the causative agent of the most severe form of malaria, reportedly displays glutathione peroxidase (GPx) 1 activity (1). Also, a putative GPx gene was isolated from P. falciparum (2). Complementing enzymes that may constitute a glutathione-dependent antioxidant defense system have also been characterized in Plasmodia species. They can rely on their own GSH biosynthesis (3), and the pertinent key enzyme, ␥-glutamylcysteine synthetase, was identified (4). The parasites regenerate GSH from GSSG more efficiently than their host cells (3), and a plasmodial glutathione reductase was also characterized (5). Plasmodia species require an efficient antioxidant defense system, since they have to survive in a pro-oxidant habitat, the red blood cell. Moreover, they have to overcome the oxidant attack of phagocytes during the critical period between dissemination from, and reinvasion of the host cell (6,7). Within the infected erythrocytes, Plasmodia species appear to enhance oxidative stress, as indicated by the generation of methemoglobin (8) and hydroxyalkenals (9,10). On the other hand, they are known to be sensitive to oxidant killing, as evident from the peroxide nature and pro-oxidant potential of many antimalarial drugs and impaired survival in host cells with disturbed hydroperoxide metabolism (for review see Ref. 11). Plasmodial enzymes involved in the antioxidant defense of the parasite have therefore attracted attention as potential targets for the development of novel antimalarials.
In this context the putative plasmodial GPx gene appeared particularly intriguing. Encoding a sulfur homologue of the mammalian selenoproteins, the gene product should be substantially less efficient than the host cell enzymes. (12,13). The comparatively high sensitivity of Plasmodia species to hydroperoxides could therefore result from the necessity to rely on sulfur-catalyzed hydroperoxide reduction, whereas the host cells make use of the more efficient selenium catalysis. Beyond, the amino acid sequence deduced from the plasmodial GPx gene resembles phospholipid hydroperoxide peroxidase (PHGPx, GPx-4), which displays a degenerate substrate specificity (14). Correspondingly, it could not be taken for granted that the plasmodial GPx homologue is indeed a glutathione peroxidase. The plasmodial GPx gene was therefore heterologously expressed in Escherichia coli, and the protein was purified in sufficient quantities to allow an in-depth functional analysis. As expected, it proved to be a peroxidase acting on a broad spectrum of hydroperoxides with low efficiency. Also, the reaction rates with GSH were surprisingly low. Instead, thioredoxins reduced the enzyme efficiently enough to reclassify this member of the glutathione peroxidase family as a thioredoxin peroxidase and to substitute the acronym PfTPx for the originally introduced glutathione peroxidase of P. falciparum (PfGPx) (2).

EXPERIMENTAL PROCEDURES
Heterologous Expression and Purification of PfTPx-The full-length cDNA encoding the PfTPx was amplified by reverse transcriptasepolymerase chain reaction using asynchronous blood stage RNA and * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The clone was grown in carbenicillin (1 mg/ml)-supplemented LB media at 37°C and 180 rpm to an A 600 of 0.5 and induced with 0.4 mM isopropyl-1-thio-␤-D-galactopyranoside. The culture was grown for an additional 5 h and controlled for expression by SDS-polyacrylamide gel electrophoresis. For routine preparation, cells were harvested 3 h after induction, resuspended in 50 mM Tris, 1 mM dithiothreitol, pH 8.5 (buffer A), disrupted with a French press at the 900 p.s.i., and centrifuged at 18,000 rpm for 10 min. The supernatant was loaded on a Macro-Prep ® High Q Support fitted on a Macro-Prep ® High S Support column with a flow rate of 2 ml/min (Bio-Rad). The column was washed with 10 bed volumes of buffer A. The Macro Prep ® High S Support column alone was eluted with a NaCl gradient (0 -2 M) in the same buffer. Fractions with GPx activity were concentrated by ultrafiltration (Omega Cell, cut-off 10 kDa, Pall Gelman Sciences, An Arbor, MI), loaded onto a Sephacryl S-200 (Amersham Pharmacia Biotech) gel filtration column equilibrated with 0.1 M Tris, 0.1 M NaCl, 5 mM EDTA, pH 7.6, and eluted at a flow rate of 0.5 ml/min. Active fractions were analyzed by SDS-polyacrylamide gel electrophoresis for homogeneity and stored at 4°C.
Characterization of Expression Product-The molecular mass of denatured PfTPx was estimated by silver-stained (15) SDS-polyacrylamide gel electrophoresis using the Phast System TM (Amersham Pharmacia Biotech) with an acrylamide gradient of 8 -25% and a 10-kDa ladder as reference. The molecular mass of native PfTPx was estimated by gel filtration on a Sephadex S-100 column equilibrated with a 0.1 M Tris buffer of pH 7.6 containing 5 mM EDTA at a flow rate of 0.5 ml/min. Chymotrypsinogen, bovine serum albumin, blue dextran, and cytochrome c were co-chromatographed as reference proteins. Protein concentration was determined according to Bradford (16) with the reagent from Bio-Rad, taking bovine serum albumin as the standard. The precise molecular weight was determined by matrix-assisted laser desorption and ionization time of flight (MALDI-TOF) mass spectrometry with a Bruker Reflex II-MALDI-TOF mass spectrometer (Bruker-Franzen-Analytik, Bremen, Germany). For this purpose the protein was precipitated with trichloroacetic acid, washed with the acetone, and resuspended in saturated matrix solution (10 mg/ml sinapinic acid in 40% acetonitrile and 0.1% trifluoroacetic acid). The protein at a final concentration about 50 pmol/l matrix solution was accelerated at 20 kV. Spectra were externally calibrated using bovine serum albumin as standard. Approximately 200 shots were summed for each spectrum. N-terminal sequencing was performed with an Applied Biosystems 494 A sequencer.
Thioredoxin peroxidase activity was determined analogously by replacing GSH by thioredoxin and glutathione reductase by thioredoxin reductase. In each case the particular thioredoxin was coupled to the thioredoxin reductase of the same species. Care was taken that the thioredoxin reductase capacity never became rate-limiting. E. coli thioredoxin was from Sigma-Aldrich, human thioredoxin was kindly provided by Prof. A. Holmgren Uppsala, Sweden, and thioredoxin as well as thioredoxin reductase of P. falciparum were prepared as described elsewhere (20,21).
For kinetic analysis, the glutathione peroxidase assay was performed at GSH concentrations ranging from 3.3 to 30 mM and various hydroperoxides (t-bOOH, H 2 O 2 , cumene hydroperoxides). Thioredoxin peroxidase kinetics were performed with thioredoxin of P. falciparum at concentrations between 1.5 and 5 M. Spontaneous reaction rates were subtracted. Kinetic data were obtained by the single curve progression analysis according to Forstrom et al. (22) and further analyzed as described by Dalziel (23) for bisubstrate reactions.

Heterologous Expression and Purification of PfTPx-E. coli
BL21 (DE3) pLys cells transformed with the pET5a-derived expression plasmid efficiently produced PfTPx upon induction with isopropyl-1-thio-␤-D-galactopyranoside, as indicated by the appearance of a prominent band of ϳ20 kDa. The heterologously expressed protein peaked near 2 h after induction and appeared to remain stable for at least three more hours. The purification scheme applied yielded a product of apparent electrophoretic homogeneity (Fig. 1). The apparent as well the precise molecular mass of 19,750 Da obtained by MALDI-TOF spectrometry was significantly less than the molecular mass of 23953 Da calculated for the deduced amino acid sequence of full-length PfTPx (205 residues). N-terminal sequencing revealed an amino acid sequence covering position 26 -40 of the deduced PfTPx sequence shown in Fig. 2. Within experimental error, the experimentally determined molecular mass complies with that calculated for a sequence covering position 26 to 196 (19,780.8 Da).
Comparison of PfTPx with cGPx and PHGPx-Alignment of PfTPx with mammalian glutathione peroxidases reveals that the plasmodial enzyme is a member of the GPx superfamily yet is more related to PHGPx than to any of the other GPx types (Fig. 2) displaying 64, 55, 51, and 45 identities with PHGPx, cGPx, gastrointestinal GPx, and pGPx, respectively. Also the PfTPx sequence as obtained by heterologous expression in E. coli, corresponds to the homologous stretch of 170 amino acid residues found in mature porcine PHGPx (24). Like PHGPx, PfTPx proved to be monomeric when subjected to Sephadex G-100 gel filtration under nondenaturing conditions. The nativity of the eluting enzyme was verified by GPx activity measurements. It eluted with an apparent molecular mass of 18.9 kDa, whereas no activity could be detected in the fractions corresponding to the molecular mass of its tetrameric congeners, pGPx or cGPx. As is evident from the x-ray structures of cGPx (25) and pGPx (26), subunit contact surfaces in these tetrameric GPx types are essentially built up by residues corresponding to the inserts at positions 121 and 161 of the PfTPx sequence (27). They are missing in both PfTPx and PHGPx, which explains their monomeric nature.
PfTPx, like PHGPx, was more active with cumene hydroperoxide, less active with H 2 O 2 and t-bOOH, and unlike PHGPx, did not at all accept phosphatidylcholine hydroperoxide (data not shown). In terms of molar efficiencies, PfTPx appeared markedly poorer than bovine cGPx coinvestigated as a selenoperoxidase reference standard. Specific GPx activities, as measured under routine conditions, differed by three orders of magnitude (not shown).
The kinetic mechanism for GSH-dependent hydroperoxide reduction by PfTPx was evaluated by means of the single curve progression analysis (22) at various fixed GSH concentrations, which were kept constant over time by regeneration, and a suboptimal concentration of ROOH, which declined over time and correspondingly lead to slowing down of the reaction rate. From these curves, the reciprocal concentrations of ROOH at intervals of 2 s were derived and plotted against the initial velocities at each pertinent time point, as exemplified for the turnover of t-bOOH by GSH in Fig. 3. The data are presented as Dalziel plots (23), in which the reciprocal velocities are multiplied by enzyme molarities to facilitate the extrapolation of meaningful kinetic coefficients, as indicated. As shown in Fig. 3, such primary plots yielded parallel slopes for different concentrations of GSH, as is typical for "enzyme substitution" or "ping-pong" mechanisms. Replotting the reciprocal GSH concentrations against the reciprocal apparent V max for infinite concentrations of t-bOOH yielded a straight line cutting at the ordinate origin (Fig. 4). The same kinetic pattern was displayed by PfTPx with H 2 O 2 and cumene hydroperoxide as long as GSH was used as reducing substrate (Table I). It can be described by a Dalziel equation for two substrate ping-pong mechanisms.
wherein the coefficient ⌽ 0 approximates zero. Accordingly, limiting V max and K m values are infinite. Such lack of enzyme saturation can be due to two distinct catalytic phenomena, either the formation of enzyme-substrate complex is slower than the reaction within the complexes or specific enzymesubstrate complexes not formed at all, as is presumed in Equations 2-4.
In this case, the coefficient ⌽ 1 is defined as the reciprocal rate constant k ϩ1 Ј for the net forward reaction of reduced enzyme with ROOH and depends on the nature of the peroxide (Table  I). k 1ϩ Ј is defined as k ϩ1 Ϫ k Ϫ1 , and may be regarded as k ϩ1 , since the partial reaction shown in Equation 2 should be irreversible. ⌽ 2 is the reciprocal k ϩ2 Ј for the two-step regeneration of the reduced enzyme by GSH according to Equations 3 and 4. Therefore, the physical meaning of k ϩ2 Ј is more complex (Equation 5).
The kinetic pattern of PfTPx is identical to that of the selenium-containing glutathione peroxidases (28 -31). However, the kinetic coefficients ⌽ 1 and ⌽ 2 of PfTPx differ markedly from those of the selenoproteins. Instead, they are similar to those of a sulfur homologue of PHGPx produced by site-directed mutagenesis. Although low efficiency of PfTPx in GSH-dependent hydroperoxide reduction and the sequence similarities to the less GSH-specific pGPx (32) and PHGPx (33,34), its activity with alternative reducing substrates, notably thioredoxins, was investigated. Table II shows that PfTPx indeed accepts thioredoxins of various species. The turnover rates observed with thioredoxin of E. coli and man in the submillimolar range were similar to those measured with 10 mM GSH, and autologous plasmodial thioredoxin at a concentration of only 5 M triggered a significantly faster reaction.
Based on these preliminary findings, the kinetics of PfTPx were analyzed with thioredoxin (Trx) of P. falciparum (PfTrx) and t-bOOH. Fig. 5 reveals that also with PfTrx a ping-pong pattern is observed. The secondary plot (Fig. 6), however, shows a marked difference when compared with the kinetic pattern obtained with GSH as reducing substrate. The slope no longer cuts the ordinate at zero, implying Michaelis-Mententype saturation kinetics with defined ⌽ 2 , V max , and K m values. Also, the ⌽ 2 value for PfTrx is much lower than for GSH (Table  III). This kinetic pattern and the coefficients obtained allow the following conclusions. (i) The reducing substrate does not affect the reaction of the reduced enzyme with the hydroperoxide (Equation 2); ⌽ 1 or k 1 , respectively, are not significantly different for the GSH-and the PfTrx-driven reaction, as expected for a ping-pong mechanism. (ii) Saturation kinetics therefore can only result from a shift of rate constants in the reductive part of catalytic cycle. Obviously, the formation of a complex between the oxidized enzyme and PfTrx occurs faster with k ϩ4 Ϫ k Ϫ4 than its decay with k ϩ5 (Equation 6).
ϩ E red ϩ Trx ox (Eq. 6) (iii) The limiting K m value for the PfTrx is defined by (k Ϫ4 ϩ k ϩ5 )/k ϩ4 . It approximates the dissociation constant K s ϭ k Ϫ4 / k ϩ4 and can be taken as a measure of affinity for PfTrx. (iv) The overall rate determining k cat , then, is k ϩ5 . (v) The high k ϩ2 Ј value and saturation kinetics with a low K m classify PfTrx as a specific substrate of PfTPx.

DISCUSSION
The glutathione peroxidase family of proteins is spread over the whole living kingdom (2,14). The name-giving classical glutathione peroxidase, however, has so far only been detected in vertebrates, where it proved to be a selenoprotein (35,36) and of the related selenoproteins, pGPx, gastrointestinal GPx, and PHGPx, only the latter has been identified in a nonvertebrate species, Schistosoma mansoni (37,38). Genes encoding homologous proteins in which the active site selenocysteine is replaced by cysteine appear to be widely distributed in nature (2,14). Such proteins are commonly addressed as glutathione peroxidases, although their activity has never been systematically analyzed. In fact, some representatives of the family have been discovered in a biological context not reminiscent of an antioxidant function, e.g. the cobalamine-binding protein in E. coli (39), the salt stress-responsive protein in Citrus plants (40), and androgen-responsive epidydymal proteins in mammals (41). Furthermore, any relevant peroxidase activity of such proteins may be questioned considering the low efficiencies of cysteine-containing muteins of GPx and PHGPx.
To our knowledge, the efficiency and specificity of a naturally occurring nonselenium GPx homologue is addressed here for the first time. To this end, we have unfortunately to rely on an heterologously expressed protein, since a purification of PfTPx from P. falciparum has not yet been feasible. In its basic characteristics, however, the PfTPx gene expression product, as here obtained, is presumed to closely resemble authentic PfTPx. Admittedly, the discrepancy between the size of the deduced maximum sequence and the isolated protein is substantial. But this is also observed with the closely related mammalian PHGPx. As in the PfTPx gene, two potential start codons are contained in the PHGPx gene. In this case they are alternatively used in a tissue-specific manner, either leading to mitochondrial or cytosolic localization of the enzyme (42,43). In each case the processed expression product is the same (43), i.e. a protein of about 19 kDa corresponding in size to a PfTPx starting with Met-26. Monitoring the production of PfTPx in E. coli does not reveal any primary product in the 24-kDa region. The electrophoretic mobility of the band showing up upon induction is identical to that of the isolated product (Fig. 1). Obviously, our E. coli production strain has chosen the second ATG start codon, which is also the preferred start codon of the mammalian PHGPx genes. Certainly, the heterologously expressed PfTPx contains all residues constituting the catalytic triad essential for hydroperoxide reduction, i.e. Cys-76, Gln-111, and Trp-169 in homologous position to Gln-81, selenocysteine 46, Trp-136 of porcine PHGPx (13,24). Whether the missing N-terminal extension and the minor C-terminal truncation affects specificity remains to be established. The gain of a new specificity, as described here, is not likely explained by such modifications.
The low efficiency of PfTPx in reducing hydroperoxides does not surprise but raises the question whether such low efficiency peroxidases may be implicated in antioxidant defense at all. The k ϩ1 Ј values for the reaction of PfTPx with t-bOOH, cumene hydroperoxide, and H 2 O 2 reported here come close to the k 1 Ј of the cysteine mutein of porcine PHGPx with phosphatidylcholine hydroperoxide (13). Similar k 1 Ј values were reported for tryparedoxin peroxidase, a structurally unrelated peroxiredoxin also working with sulfur catalysis (44). These observations suggest that, with sulfur catalysis, rate constants near 10 4 M Ϫ1 s Ϫ1 can be reached for the reduction of hydroperoxides by thiols, whereas 10 7 M Ϫ1 s Ϫ1 are commonly observed with selenium catalysis. Rate constants beyond 10 6 M Ϫ1 s Ϫ1 are also reported for heme-catalyzed hydroperoxide reduction (45). Less efficient hydroperoxidases can reasonably be implicated in an- Cutting the ordinate at zero indicates that the term ⌽ 0 approaches zero, which implies that the maximum velocity and K m values of PfTPx are infinite for the pair of substrates investigated. tioxidant defense if their low molar efficiency is compensated for by extreme concentration, as has been proposed for the hydroperoxide detoxification by a peroxiredoxin in trypanosomes (44). In line with these considerations, E. coli overexpressing PfTPx proved to be slightly more resistant to oxidative challenge, 2 which is not surprising in view of PfTPx being the prominent protein in such cells (Fig. 1). At less abundant levels, the ability of PfTPx to balance oxidative stress may be doubted.
The observation that a GPx homologue reacts with thiols other than GSH is not surprising either. Only for the cytosolic GPx has a pronounced specificity for GSH been documented (46). This specificity is considered to be due to basic residues, Arg-57, Arg-102, Arg-184, Arg-185, and Lys-92 in bovine cGPx, which direct the SH group of the substrate to the active-site selenium atom by electrostatic forces (47). These residues are only partially conserved in the gastrointestinal and extracellular isozymes and completely lost in PHGPx-type enzymes. Accordingly, pGPx has been reported to accept thioredoxin and glutaredoxin (32), and PHGPx, in the absence of GSH, can form high molecular weight protein aggregates that are cross-linked by Se-S and/or S-S bridges, a process shown to be of physiological importance in late phases of mammalian sperm maturation (33). It does, however, not react with E. coli thioredoxin and human thioredoxin 1 and 4. 3 In contrast, PfTPx appears to be specialized for interaction with thioredoxin, as evident from the kinetic data reported.
Nevertheless, a competition of GSH with thioredoxin for oxidized PfTPx under in vivo conditions cannot be fully ruled out. Irrespective of uncertainties about the concentrations of GSH and Trx in the various differentiation states of the parasites, the kinetic parameters of PfTPx imply that the GSHdriven reaction falls short under most conditions that could be envisaged to be physiologically relevant, as is easily calculated by means of the rate equation (Equation 1) and the kinetic coefficients (Table III) (5 M), and glutathione reductase as indicator enzyme (5.6 units/ml) was replaced by the autologous thioredoxin reductases (0.5 unit/ml, 0.56 unit/ml, 0.001 unit/ml, respectively). With each of the thioredoxins, the specific rate was at least as high as that observed with GSH at 50 -2000ϫ the concentration. ical level, the maximum velocities achieved with PfTrx are never reached. In fact, the GSH-dependent reaction would break even with that of PfTrx if the GSH levels were raised beyond 26 mM. At high peroxide challenge, however, cellular GSH concentrations of 5-10 mM, which can not be rated as uncommon, may compete with PfTrx if present in concentrations below the K m of 10 M. These consideration are, however, not meant to implicate a pivotal role of PfTPx in GSH-dependent hydroperoxide detoxification in P. falciparum. Irrespective of the donor substrate, the efficiency of PfTPx in hydroperoxide reduction is limited by the low k 1 Ј values. The GSH-dependent hydroperoxide removal by cGPx in the parasite's host cells is faster by orders of magnitude (48). The specificity of PfTPx for thioredoxin points to biological roles distinct from antioxidant defense such as redox regulation of gene expression and differentiation processes, as are attributed to PHGPx and low efficiency peroxiredoxin-type peroxidases in mammalian cells (33, 49 -51).
The search for other candidates that exert antioxidant defense in Plasmodia species therefore remains rewarding. An Fe-superoxide dismutase and catalase are reportedly present in Plasmodia species. The increase of alkyl hydroperoxide reduction by supplementation of cultures of malaria parasites with selenium (52), however, needs clarification. It cannot be ruled out that selenium, when provided as selenite or selenocystine in micromolar concentrations, is unspecifically incorporated into PfTPx to some extent, and selenium incorporation into 1% PfTPx could result in the observed duplication of turnover, taking into account the pronounced difference of rate constants between Se-and non-Se glutathione peroxidases (12,13). Alternatively, also in Plasmodia species real selenoperoxidases might exist that could be typical glutathione peroxidases or peroxiredoxins. Peroxiredoxin-type putative thioredoxin peroxidases have recently been identified in Plasmodia (Gen-Bank TM accession number AF225977 and AF225978) and are currently being characterized. In mammals, a member of this family has been reported to be a non-Se GPx (53), and in amino acidophilic bacteria, such enzymes are selenoproteins (54). Thus, many options to link selenium-enhanced glutathione-dependent peroxide metabolism to antioxidant defense in Plasmodium remain to be explored. PfTPx, being a low efficiency nonselenium peroxidase with a clear preference for the pleiotropic redox regulator thioredoxin, is not the ideal candidate to play this role.