A Second Class of Peroxidases Linked to the Trypanothione Metabolism*

Trypanosoma brucei, the causative agent of African sleeping sickness, has three nearly identical genes encoding cysteine homologues of classical selenocysteine-containing glutathione peroxidases. The proteins are expressed in the mammalian and insect stages of the parasite. One of the genes, which contains a mitochondrial as well as a glycosomal targeting signal has been overexpressed. The recombinant T. brucei peroxidase has a high preference for the trypanothione/tryparedoxin couple as electron donor for the reduction of different hydroperoxides but accepts alsoT. brucei thioredoxin. The apparent rate constantsk 2′ for the regeneration of the reduced enzyme are 2 × 105 m −1s−1 with tryparedoxin and 5 × 103 m −1 s−1 with thioredoxin. No saturation kinetics was observed and the rate-limiting step of the overall reaction is reduction of the hydroperoxide. With glutathione, the peroxidase has marginal activity and reduction of the enzymes becomes limiting with a k 2′ value of 3 m −1 s−1. The T. bruceiperoxidase, in contrast to the related Trypanosoma cruzienzyme, also accepts hydrogen peroxide as substrate. The catalytic efficiency of the peroxidase studied here is comparable with that of the peroxiredoxin-like tryparedoxin peroxidases, which shows that trypanosomes possess two distinct peroxidase systems both dependent on the unique dithiol trypanothione.

Parasitic trypanosomatids are the causative agents of lifethreatening tropical diseases such as American Chagas' disease (Trypanosoma cruzi), African sleeping sickness (T. brucei gambiense and T. b. rhodesiense), Nagana cattle disease (T. congolense, T. b. brucei), and the three manifestations of leishmaniasis. All these protozoa have in common that they possess an unparalleled thiol metabolism (1)(2)(3) where glutathionylspermidine conjugates are the main non-protein thiols. Trypanothione (T(SH) 2 ; N 1 ,N 8 -bis(glutathionyl)spermidine) (4) is kept in the dithiol state by the flavoenzyme trypanothione reductase, which catalyzes the NADPH-dependent reduction of trypanothione disulfide. Several genetic approaches revealed that trypanothione reductase is essential for the parasites (Ref. 5, for a recent review see Ref. 3).
So far the trypanothione metabolism is the only known thiol system of the parasites where it replaces the otherwise ubiquitous glutathione/glutathione reductase couple. We could show that African trypanosomes contain also a classical thioredoxin but the respective specific reductase has not yet been described in any kinetoplastid organism (6). Trypanothione is the donor of reducing equivalents in several vital pathways of the parasites, like reduction dehydroascorbate (7), synthesis of deoxyribonucleotides by ribonucleotide reductase (8), conjugation and export of metals and drugs (9 -11), and the detoxication of hydroperoxides (12)(13)(14). Trypanosomes are exposed to various reactive oxygen species such as superoxide anions, hydrogen peroxide, and products of the host myeloperoxidase system. However, their ability to cope with oxidative stress seems to be surprisingly weak. Although pathogenic trypanosomes possess an iron-containing superoxide dismutase, they lack both catalase and a selenocysteine-containing classical glutathione peroxidase (GPX), 1 which are the main hydroperoxide metabolising enzymes in the mammalian host. In trypanosomatids, the metabolism of hydrogen peroxides is achieved by a unique trypanothione-dependent cascade first described in the insect parasite Crithidia fasciculata (12,13). With NADPH as final electron donor, the reducing equivalents flow from trypanothione onto tryparedoxin and finally a tryparedoxin peroxidase, which then catalyzes the reduction of the hydroperoxide. The latter enzyme is a member of the 2-Cys peroxiredoxin protein family. Components of this pathway have been demonstrated in many trypanosomatid organisms (14) such as C. fasciculata, T. brucei, T. cruzi, Leishmania major, and L. donovani. Distinct mitochondrial tryparedoxin peroxidases have been found in T. cruzi (15) and T. brucei (16).
Classical glutathione peroxidases are selenoproteins, which catalyze the reduction of different hydroperoxides at the expense of glutathione (17)(18)(19). Four distinct selenocysteine-glutathione peroxidases occur in mammals, cytosolic GPX 1, also called classic or cellular GPX, gastrointestinal GPX 2, the GPX 3 found in human plasma, and the phospholipid hydroperoxide glutathione peroxidase GPX 4 (19,20). Classical glutathione peroxidases have so far only been detected in vertebrates and a GPX 4 in the helminth Schistosoma mansoni (21). Genes encoding homologous proteins in which the active site selenocysteine is replaced by cysteine are widely distributed in nature but in most cases catalytic activities have not yet been demonstrated. Recently a putative cysteine-containing glutathione peroxidase has been studied in the malarial parasite Plasmodium falciparum. The recombinant protein proved to be much more efficient with thioredoxin than with glutathione as electron source (22). In T. cruzi the gene of a cysteine homologue of glutathione peroxidases has been cloned and overexpressed and the recombinant protein has been described as glutathione peroxidase (23). The reported activity with glutathione was extremely low, resulting in an apparent V max of 30 milliunits/mg in the presence of 3 mM GSH and a K m for GSH of about 7 mM. Recently the enzyme was shown also to use trypanothione/tryparedoxin as electron source to reduce different hydroperoxides but not hydrogenperoxide (24). The marginal activity of the T. cruzi enzyme with GSH and the fact that the GSH concentration is conspicuously low in trypanosomes prompted us to study the respective peroxidase system in African trypanosomes and to elucidate the physiological electron donor(s). Here we report on the cloning and functional characterization of the T. brucei peroxidase. The enzyme accepts different thiols as substrate but shows a very high preference for the trypanothione/tryparedoxin couple.

EXPERIMENTAL PROCEDURES
Materials-Human thioredoxin reductase and human glutathione reductase were a kind gift from Dr. Heiner Schirmer, Heidelberg. Recombinant T. brucei thioredoxin (6) and tryparedoxin (25) and T. cruzi trypanothione reductase (26)  PCR Amplification and Sequence Analysis of the Peroxidase Locus-Total RNA of T. brucei bloodstream form parasites of strain 449 was reverse transcribed with Omniscript TM Reverse Transcriptase (Qiagen). The cDNA was amplified by 10 cycles of PCR using a spliced leader primer (5Ј-CTATTATTAGAACAGTTTCTGTAC-3Ј) and an oligo(T) primer (95°C, 2 min; 95°C, 1 min; 50°C, 1 min; 72°C, 5 min; 10 cycles; 72°C, 10 min). The 5Ј-region of the peroxidase gene was amplified from the cDNA pool with a gene-specific primer (5Ј-ATCAATGAG-GAAGGATGTAAAGTTCC-3Ј) and the spliced leader primer. For the 3Ј-region a gene-specific primer (5Ј-CTCATGCTGCGTTCATCTCGG-3Ј) and the oligo(T) primer were used (94°C, 2 min; 94°C, 30 s; 50°C, 30 s; 72°C, 3 min; 30 cycles; 72°C, 5 min). Two specific primers (5Ј-GCGCG-GATCCCTGCGTTCATCTCGG-3Ј and 5Ј-GCATGGTACCCAGCGA-CAAATCC-3Ј) were derived from the resulting sequences for gene amplification from genomic DNA of bloodstream strain 449 (94°C, 2 min; 94°C, 30 s; 50°C, 30 s; 72°C, 2 min; 30 cycles; 72°C, 5 min). The PCR product was cloned using the BamHI and KpnI restriction sites (underlined sequences) into the pQE-30 vector (Qiagen) and used as digoxigenin-labeled hybridization probe for screening the P1 library. Labeling and luminescence detection components from Roche Molecular Biochemicals were used as described by the provider. Four positive clones were isolated, digested with PvuII and BamHI, and the three positive fragments obtained were subcloned into the pBluescript KS(ϩ) vector (Stratagene). Both strands of the subclones were completely sequenced by MWG Biotech. The sequence of the complete peroxidase locus later revealed that the 5Ј-and 3Ј-primers used to generate the hybridization probe corresponded to px III and px I or II, respectively, and that the genomic amplification had generated a chimeric product. Therefore for the functional characterization of the peroxidase III (Px III) protein, the authentic gene was cloned and overexpressed as described below.
Preparation of Genomic DNA and Southern Blot Analysis-Genomic DNA of T. brucei strain 449 was prepared using the DNeasy tissue kit (Qiagen). 10 g of DNA was digested with the respective restriction enzyme. The fragments were separated on a 1% agarose gel, blotted onto Hybond TM -Nϩ membrane (Amersham Biosciences), and hybridized with the digoxigenin-labeled peroxidase gene.
Preparation of Poly(A) ϩ mRNA and Northern Blot Analysis-Total RNA from cultured T. brucei cells of procyclic and bloodstream strains 449 (7 ϫ 10 8 cells) was prepared with the RNeasy Kit (Qiagen). Poly(A) ϩ mRNA was enriched from 475 g of total RNA using the Oligotex ® mRNA Kit (Qiagen) and separated on a 1% agarose gel containing 1.8% formaldehyde. Blotting and hybridization were performed as described above.
Heterologous Expression of the Peroxidase III Gene and Purification of the Recombinant Protein-For the production of the recombinant peroxidase III with an N-terminal His 6 tag, the coding region without the initial ATG was amplified by PCR using gene-specific primers (5Ј-GCGCGGATCCCTGCGTTCATCTCGG-3Ј and 5Ј-GCATGGTAC-CGTCCAGAGACGTATTC-3Ј) and genomic DNA from procyclic strain AnTat 1.1 (94°C, 2 min; 94°C, 30 s; 50°C, 30 s; 72°C, 2 min; 35 cycles; 72°C, 5 min, Pfu polymerase). The amplification product was digested with KpnI and BamHI, cloned into the pQE-30 vector (Qiagen), and sequenced. Competent Escherichia coli XL-1 Blue cells were transformed with the pQE-30/px III plasmid. A 5-ml overnight culture was diluted 1:100 in 2ϫ YT medium containing 100 g/ml carbenicillin. The cells were grown at 37°C and 180 rpm to an OD 600 of 0.6. Expression was induced by adding 1 mM isopropyl-␤-D-thiogalactopyranoside overnight at 30°C. After centrifugation, the cells were resuspended in 15 ml of buffer A (50 mM sodium phosphate, 300 mM NaCl, pH 8.0) containing 10 mM imidazole, 150 nM pepstatin, 4 nM cystatin, 100 M phenylmethylsulfonyl fluoride, 1.5 mg lysozyme, 0.15 mg DNase A, and disintegrated by sonification. Following centrifugation, the supernatant was applied onto a 4-ml nickel-nitrilotriacetate matrix Superflow column (Qiagen). The column was washed with 50 ml of 25 mM imidazole in buffer A, and the His-tagged protein was eluted with 250 mM imidazole in buffer A. The protein fractions were analyzed for homogeneity by SDS-polyacrylamide gel electrophoresis and stored at 4°C. The protein concentration was determined using the bicin choninic acid kit (Pierce). For homogenous peroxidase III, a protein concentration of 1 mg/ml corresponds to an ⌬A 280 of 1.
Gel Filtration-The subunit composition of T. brucei peroxidase III was determined by FPLC on a Superose 12 HR 10/30 column (Amersham Biosciences). 160 g of purified recombinant protein was applied onto the column equilibrated in 0.1 M potassium phosphate buffer, 1 mM EDTA, pH 7.0 and run at room temperature with a flow rate of 0.2 ml/min. Bovine serum albumin, rabbit aldolase, horse cytochrome c, and ferritin served as molecular mass standards.
Activity Measurements-Glutathione peroxidase activity was measured at 25°C in a total volume of 90 l in 0.1 M Tris, 5 mM EDTA, pH 7.6, containing 240 M NADPH, 2.5-12 mM GSH, 75 milliunits of human glutathione reductase, 4 M T. brucei peroxidase, and 50 M hydroperoxide. The reaction was followed at 340 nm by measuring the glutathione reductase catalyzed oxidation of NADPH. The total concentration of the hydroperoxide substrate was determined by allowing the reaction to run to completion, and the concentration at each time point was calculated from the initial concentration. The assays were standardized with glutathione peroxidase from bovine erythrocytes (Sigma). Thioredoxin peroxidase activity was assayed as described above except that GSH and glutathione reductase were replaced by 1.5-20 M T. brucei thioredoxin and human thioredoxin reductase. The trypanothione/tryparedoxin-dependent peroxidase activity was measured at a fixed concentration of 100 M trypanothione, 90 milliunits of trypanothione reductase, 0.6 -8.3 M tryparedoxin, and 1 M peroxidase. All other parameters were those described above for the glutathione peroxidase assay. The reactions were started by adding the hydroperoxide substrate (H 2 O 2 , t-butyl hydroperoxide, cumene hydroperoxide). Control assays lacked the peroxidase, tryparedoxin, GSH, thioredoxin, and trypanothione, respectively.
Kinetic Analysis of the Peroxidase-Concentration dependence of the initial velocities was obtained by analyzing the time progression curve of NADPH consumption in a Beckman DU 65 spectrophotometer with data acquisition every 1.5 s. The data were analyzed using the integrated Dalziel equation for a two substrate enzyme reaction (32,33) shown in Equation 1, where [ROOH] is the initial hydroperoxide concentration, [ROH] t the concentration of the product at time t, [Thiol] the concentration of the respective thiol substrate, [E] 0 the total concentration of the peroxidase and ⌽ 0 , ⌽ 1 , and ⌽ 2 are the kinetic Dalziel coefficients.

Analysis of the Genomic Locus of the Trypanosoma brucei
Peroxidase-The genome sequencing project of Trypanosoma brucei revealed several fragments encoding a glutathione peroxidase-related gene. Based on this information, the gene has been cloned from cDNA of bloodstream parasites of strain 449. The 5Ј-region was amplified by PCR using a sequence specific reverse primer and a spliced leader primer, which is the very 5Ј-end of all trypanosomal mRNAs added by transsplicing. The 3Ј-region was amplified with a specific forward primer and a poly(T) primer. The complete coding region of the gene was then amplified from genomic DNA with two gene-specific primers and used as hybridization probe in Southern blot analysis. Digestion of T. brucei genomic DNA (procyclic strain 927-449) with several restriction enzymes revealed the presence of more than one peroxidase gene (Fig. 1A). The combined digestion with PvuII and BamHI resulted in three fragments of Ն0.9 kb, each fragment being larger than the coding region of the gene (about 0.5 kb) used for hybridization. The upper band in the PvuII digest represents two fragments of which one is cleaved in the double digestion with BamHI yielding the 1.2-kb fragment. To elucidate the complete structure of the locus, a P1 library of T. brucei genomic DNA (strain TREU 927/4) on high density filters was screened with the peroxidase probe. Four positive clones were obtained which yielded an identical restriction pattern (data not shown). The three positive fragments of P1 clone 21E1 were cloned and sequenced (EMBL/ GenBank TM accession number AJ298281). The sequence and Southern blot data were in accordance with a genomic locus of 4.4 kb containing three clustered peroxidase genes (px I-px III) (Fig. 1B). The existence of more than three gene copies can be excluded since the BsaI/BamHI fragment of 3.2 kb comprises all three peroxidase genes. Further thorough Southern blot analyses of genomic DNA digested with EcoRI, EcoRV, BamHI, HindIII, Bsp143I, and BsaI fully agreed with the arrangement of three genes depicted in Fig. 1B. Some of the blots showed a few very faint additional bands, which may indicate the existence of distantly related gene(s) outside this locus.
Structural Comparison of the Three T. brucei Thiol Peroxidases-The three protein sequences deduced from the genomic peroxidase locus consist of 166, 169, and 176 amino acid residues, respectively (Fig. 2). The overall sequences are almost identical except for the very N-and C-terminal regions. Peroxidase III, which is characterized here, is the largest one and has a calculated pI value of 9.5. In contrast, the peroxidases I and II have pI values of 6.28 and 8.34, respectively. The very basic pI of peroxidase III is mainly due to its N-terminal extension. The sequence MLRSSRKKMSAA is rich in positively charged and hydroxylated residues and shows homology to other exceptionally short (7-9 amino acid residues) mitochondrial targeting signals of trypanosomatids (34,35). When the gene is overexpressed from the pQE-30/pxIII in E. coli, two protein species are obtained, one with the His 6 tag and the complete N-terminal sequence and a second about a 2-kDa smaller protein, which does not bind to the metal matrix. Both proteins are fully active. N-terminal Edman degradation revealed that the recombinant protein has been cleaved between two basic residues of the RKK motif resulting in a protein starting with (K)KMSAA. The motif is obviously recognized by a bacterial protease. Interestingly, T. brucei possesses an endopeptidase which preferentially cleaves after two basic residues and which has been discussed to play a role in protein processing (36). Px III has also a C-terminal ARL motif, which could be a variation of the peroxisomal and glycosomal PTS 1 targeting signal SKL. Since such variations have been shown to be functional (37), the peroxidase III might be also localized in glycosomes. The structurally closely related T. cruzi peroxidase I possesses a C-terminal ARI sequence and occurs both in the cytosol and in glycosomes (24). The other two T. brucei proteins Px I and Px II lack known signal peptides and are probably located in the cytoplasm of the parasites. Except from the very N-and Cterminal regions, only eight residues are not identical in the three proteins (Fig. 2). Five substitutions are found only in Px I and the other three in the Px II sequence. In any case, only two different residues occur at the respective position. Two of these biases (positions 59 and 86 in Px III) are even present when comparing the peroxidase III sequences from strain TREU 927/4 (27) and strain AnTat 1.1 (30). With the possible exception of lysines 131 and 165, which occur only in the peroxidase II, the substitutions seem to be randomly distributed in accordance with a microheterogeneity of these genes. Taken together, the data suggest very similar or identical catalytic properties for the three peroxidases.
Structural Comparison of the T. brucei Peroxidases with Classical Glutathione Peroxidases-The primary structures of the T. brucei peroxidases described here classify the proteins as members of a large protein family comprising the selenocysteine-containing glutathione peroxidases (GPX). Classical glutathione peroxidases have a strictly conserved catalytic triad composed of a Se-Cys, a Gln, and a Trp residue (Fig. 3). The highest similarity of the T. brucei peroxidases is found with a T. cruzi protein described as glutathione peroxidase I (23,24) where 72% of all residues are conserved (Fig. 3). In the trypanosomal proteins the selenocysteine is replaced by a cysteine residue as it is the case in the protein of the malarial parasite P. falciparum (22) and many functionally not well characterized plant proteins. To other members of the family the highest degree of similarity is found with glutathione peroxidase-like proteins from plants; for instance 49% of all residues are identical in the putative phospholipid hydroperoxide glutathione peroxidase from Arabidopsis thaliana (38). The closest relative of the four human selenocysteine-containing glutathione peroxidases is the phospholipid hydroperoxide GPX 4 (39% identity), followed by the cytosolic GPX 1 (35%), the gastrointestinal GPX 2 (30%), and the extracellular GPX 3 (25%).
The 3-dimensional structures of bovine erythrocyte glutathione peroxidase (GPX 1) and of the human plasma glutathione peroxidase (GPX 3) (39, 40) have been solved in free form and the residues responsible for substrate binding have not yet been elucidated. For the cytosolic bovine GPX 1, several basic residues have been suggested to play a role in glutathione binding (39 -41). None of the residues is conserved in the T. brucei peroxidases and only Arg-103 in the extracellular human enzyme. If the residues are indeed involved in glutathione binding this finding is in accordance with the latter two enzymes not using glutathione as preferred thiol substrate (Ref. 42 and see below).
The Peroxidase Is Expressed in All Developmental Stages of Trypanosoma brucei-Northern blot analysis of poly(A) ϩ RNA from cultured procyclic and bloodstream parasites revealed a single mRNA of about 1 kb in both stages (data not shown). For Western blot analysis, cell lysates of cultured bloodstream and procyclic parasites as well as of long slender and short stumpy forms isolated from mice were prepared. The polyclonal rabbit antibody against the recombinant peroxidase detected a protein band in all stages in accordance with a general function of the peroxidase(s) throughout the life cycle of T. brucei (Fig. 4). Due to the high sequence homology of the three peroxidases, it is not possible to distinguish the expression of the individual proteins. Western blot analysis of isolated mitochondria from procyclic T. brucei revealed a corresponding single protein band (data not shown), which confirms that the very short N-terminal stretch of Px III is indeed a mitochondrial import signal.
The Recombinant Peroxidase III Is a Monomer-The pxIII gene was expressed from the pQE-30/px III vector in E. coli XL-1 blue cells. Purification of the N-terminally His 6 -tagged protein on a nickel ligand column yielded about 15 mg of pure recombinant protein from 1 liter of bacterial culture. On SDS-PAGE, the recombinant peroxidase showed a molecular mass of 21,000, 2 kDa larger than the authentic protein (Fig. 4). Gel filtration on Superose 12 revealed an M r of about 20,000 for the recombinant His-tagged protein in accordance with the enzyme being a monomer as it is the case for porcine phospholipid hydroperoxide GPX 4 (43). Both proteins lack a stretch of about 15 residues (following Thr-132), which has been shown to be a subunit interaction site in the tetrameric GPXs (39).
Different Thiol Systems as Electron Donors for the T. brucei Peroxidase-T. brucei peroxidase III catalyzes the reduction of hydrogen peroxide and organic hydroperoxides in the presence of different thiol substrates. Glutathione, T. brucei thioredoxin, and the trypanothione/T. brucei tryparedoxin couple were studied as electron donors. The kinetic mechanism of the parasite enzyme was evaluated by single curve progression analysis (32,33) at several fixed thiol concentrations. The hydroperoxide concentration was chosen such that the reaction rate was dependent on its concentration. The thiol concentration was kept constant over time by coupling the reaction to the respective reductase, which allowed the reactions to be followed by meas- uring NADPH consumption. The reciprocal initial velocities multiplied by the enzyme concentration were plotted against the reciprocal concentrations of the hydroperoxide substrate. These plots yielded parallel lines (Fig. 5A), which implies an enzyme substitution or ping-pong mechanism as described by the special Dalziel equation for bisubstrate reactions (32) as shown in Equation 2, where ternary complexes between the enzyme and both substrates are not formed. In this equation ⌽ 0 defines [E]/V max , the reciprocal enzyme-normalized limiting maximum turnover; ⌽ 1 and ⌽ 2 are k 1 Ј and k 2 Ј, respectively, the overall reciprocal rate constants of the two half reactions (Fig. 6). Most probably the peroxidase reacts with its substrates in two independent consecutive reactions, oxidation of the reduced peroxidase by ROOH, and subsequent reduction of the oxidized enzyme by the respective thiol. An analogous kinetic has been found and analyzed in detail in the case of the peroxiredoxin-like peroxidases in trypanosomatids (12,13). A replot of the ordinate intercepts, which represent the reciprocal apparent maximum velocities at infinite peroxide concentration, against the reciprocal thiol concentration, yields a straight line that cuts the ordinate at ⌽ 0 . As outlined below, with thioredoxin and the trypanothione/tryparedoxin couple as thiol donors, the T. brucei peroxidase yielded secondary Dalziel plots intersecting the ordinate at or near zero ( Fig. 5B and Table I). The fact that the term ⌽ 0 approaches zero implies that the kinetic does not show typical saturation at high concentrations of both substrates but that the limiting V max and K m values are infinite. The kinetic pattern does not exclude formation of a specific enzyme-substrate complex, but if it is formed its decay cannot be the rate-limiting step in the overall reaction (Fig. 6). The activity of the Px III was first measured in the presence of GSH/NADPH/human glutathione reductase with cumene hydroperoxide and t-butylhydroperoxide as substrates. GSH concentrations between 2.5 and 12 mM were necessary to detect enzymatic activity at all. The primary Dalziel plots at several fixed GSH concentrations showed linear dependence of the reciprocal enzyme activity on the hydroperoxide concentration. Parallel lines were obtained for the different GSH concentrations, in accordance with ternary complexes between the enzyme and the substrates not being formed. The replot of the ordinate intercepts versus the reciprocal GSH concentration yielded lines that intersected the ordinate at distinct ⌽ 0 values. Taking into account the large experimental error when measuring such low activities, K m values of 12 and 47 M were obtained for cumene hydroperoxide and t-butyl hydroperoxide, respectively (Table I). The calculated K m value for glutathione was 16 and 46 mM in the presence of the two hydroperoxides. Since the concentration of free GSH is about 134 and 300 M in bloodstream and procyclic Trypanosoma brucei, respectively (1), glutathione should not play a major role as electron donor for the enzyme and the reactions were not evaluated in further detail.
We then studied T. brucei thioredoxin and the trypanothione/T. brucei tryparedoxin couple for their ability to serve as electron donors of peroxidase III. The assays contained T. brucei thioredoxin, NADPH, human thioredoxin reductase and trypanothione, NADPH, T. cruzi trypanothione reductase, T. brucei tryparedoxin, respectively. At millimolar concentrations of trypanothione, direct reduction of hydroperoxides by the dithiol becomes significant. Therefore the assays with tryparedoxin contained a fixed concentration of 100 M T(SH) 2 where the spontaneous reaction can be neglected. Assays lacking the peroxidase showed that tryparedoxin alone does not accelerate the spontaneous reaction between T(SH) 2 and the hydroperoxide. A second control assay containing all components except tryparedoxin confirmed that trypanothione does not directly reduce the peroxidase. Thioredoxin as well as T(SH) 2 /tryparedoxin were efficient electron donors for the peroxidase (Table I) FIG. 6. The two partial reactions catalyzed by the peroxidase. k 1 Ј is k ϩ1 Ϫ k -1 and may be regarded as k ϩ1 since the partial reaction should be irreversible. k 2 Ј is the overall rate constant for the two-step regeneration of reduced enzyme by the thiol. The reciprocal rate constant k 1 Ј is defined as ⌽ 1 , whereas the reciprocal k 2 Ј is given by ⌽ 2 .
cients could be regarded as zero, so V max and K m values were infinite for both thiol systems. Presuming a ping-pong mechanism for the reaction, the k 1 Ј values obtained for a single hydroperoxide substrate should be independent of the nature of the thiol. As shown in Table I, the k 1 Ј values varied significantly especially with GSH, which may be due to the extremely low activities in the latter system. It is therefore not possible to give an exact judgement of the specificity of the enzyme toward different hydroperoxide substrates. Cumene-OOH and H 2 O 2 appear to be slightly better substrates than t-butyl-OOH.
In contrast to the k 1 Ј values, which are specific for the hydroperoxide substrate, the k 2 Ј values represent the apparent overall rate constants for the regeneration of the reduced enzyme and therefore depend on the thiol. As expected, the k 2 Ј values with different hydroperoxides are in fact almost identical for the particular thiol system. With thioredoxin a k 2 Ј value of 5 ϫ 10 3 M Ϫ1 s Ϫ1 is obtained, while the k 2 Ј value in the T(SH) 2 /Tpx system is about 2 ϫ 10 5 M Ϫ1 s Ϫ1 . These results clearly show that the trypanothione/tryparedoxin couple is by two and five orders of magnitude more efficient as electron donor when compared with thioredoxin and glutathione, respectively. Thus the enzyme, although a structural homologue of the classical glutathione peroxidases, is highly specific for the trypanothione/tryparedoxin couple and represents a new class of tryparedoxin peroxidases (Fig. 7). DISCUSSION The T. brucei peroxidase studied here is a non-selenium homologue of glutathione peroxidases. The parasite enzyme has only marginal activity with glutathione as electron donor but shows strong preference for T. brucei thioredoxin and especially for the trypanothione/T. brucei tryparedoxin couple. Other proteins, which structurally belong to the glutathione peroxidase family but prefer thioredoxin as reducing substrate, are the selenoenzyme human plasma glutathione peroxidase (GPX 3) (42) and the non-selenium P. falciparum thioredoxin peroxidase (22). In the presence of thioredoxin or the trypanothione/tryparedoxin couple, the T. brucei peroxidase does not show saturation kinetics toward different hydroperoxide substrates. It follows ping-pong kinetics with infinite K m and V max values as it is the case with the classical glutathione peroxidases (17) as well as the peroxiredoxin-like peroxidase from C. fasciculata (13). In contrast, with glutathione as thiol substrate, saturation kinetics are observed with a K m value for GSH of about 30 mM and negligible k cat values of 0.05-0.1 s Ϫ1 . The shift toward saturation kinetics implies that one or more rate constants describing the decay of an enzyme-substrate complex are slowed down when glutathione acts as the electron donor. With glutathione as well as with thioredoxin, regeneration of the reduced enzyme species is the rate-limiting half reaction. In contrast, in the presence of the trypanothione/ tryparedoxin couple, the reaction is much faster and reduction of the hydroperoxide substrate becomes limiting. A similar observation has been made for the P. falciparum thioredoxin peroxidase mentioned above, where also with glutathione regeneration of the reduced enzyme species is the slower half reaction, whereas in the presence of thioredoxin, reduction of the peroxide substrate is rate-limiting (22).
The closest known relative of the new T. brucei peroxidase is a T. cruzi protein described as glutathione peroxidase I (GPX I) (34). Despite the structural similarity (the proteins share 72% of all residues) the catalytic properties are quite different. In contrast to the T. brucei peroxidase, the T. cruzi enzyme does not accept H 2 O 2 as substrate (23,24). The marginal activity of the T. brucei peroxidase in the presence of glutathione is comparable with that of the T. cruzi peroxidase (23) for which an apparent V max of 33 nmol/min⅐mg protein (corresponding to a k cat of 0.01 s Ϫ1 ) has been obtained at 3 mM GSH. Taking into account the extremely low turnover of this system and the fact that the glutathione concentration in trypanosomatids is in the 100 -300 M range (1) this activity should not be of physiological significance. Recently the T. cruzi enzyme has been shown to be reduced also by the trypanothione/tryparedoxin couple but the activities were only 8 -15 times higher than those observed with glutathione (24). The low activities can at least partially be explained by the fact that a lower T(SH) 2 concentration was used to keep tryparedoxin reduced. In addition, the activities refer to the overall reaction, which is limited by the reaction between reduced trypanothione and tryparedoxin.
The T. cruzi peroxidase contains a C-terminal tripeptide ARI and occurs both in glycosomes and in the cytosol (24). The T. brucei Px III studied here possesses a C-terminal ARL which indicates also a glycosomal localization. In addition, its Nterminal MLRSSRKKM sequence targets the protein into mitochondria. Thus, the glutathione peroxidase-related peroxidase as well as the peroxiredoxin-like enzyme (16) occur in the mitochondria of T. brucei. The first peroxidase system discovered in trypanosomatids was the cascade composed of trypanothione, trypanothione reductase, tryparedoxin, and the perox- . Trypanothione subsequently reduces tryparedoxin (Tpx ox ) and the reduced tryparedoxin (Tpx red ) then transfers electrons onto the oxidized peroxidase (Px ox ). The reduced Px red finally catalyzes the reduction of the hydroperoxide substrate.

TABLE I
Kinetic coefficients and apparent rate constants of T. brucei peroxidase III for the reduction of hydroperoxides by different thiol substrates The Dalziel coefficients ⌽ 1 and ⌽ 2 for the thioredoxin-and tryparedoxin systems are the mean of at least three independent series. They correspond to the reciprocal of the apparent rate constants k 1 Ј and k 2 Ј for the two half reactions. The extrapolated ⌽ 0 coefficients were zero within the experimental error for these reactions. The ⌽ 1 coefficients for the GSH system were determined for five different GSH concentrations. Limiting K m values for the hydroperoxide and thiol substrate, given as ⌽ 1 /⌽ 0 and ⌽ 2 /⌽ 0 , respectively, were only obtained in the glutathione system. The K m values for cumeneOOH and t-butylOOH were 12 and 47 M, the K m values for GSH were 16 and 46 mM, respectively. iredoxin-type tryparedoxin peroxidase (13,15). T. brucei Px III has very similar kinetic properties and can replace the peroxiredoxin-like peroxidase in this cascade. Both peroxidases reduce their peroxide substrates at rates of 10 4 -10 5 M Ϫ1 s Ϫ1 , which is slower than the subsequent regeneration of the reduced enzyme species. In contrast, in the mammalian selenoenzyme GPX I reduction of the peroxide substrate is extremely fast and the reductive step by glutathione which occurs at a rate of Յ10 5 M Ϫ1 s Ϫ1 becomes limiting (17)(18)(19). Thus on a molecular basis, the parasite peroxidases are only 5-10 times less efficient than the classical selenoenzymes. The overall poor performance of trypanosomatids in coping with oxidative stress can neither be attributed to the ability of the peroxidases to reduce hydroperoxides nor to their regeneration by reduced tryparedoxin. The limiting factor is the regeneration of reduced tryparedoxin by trypanothione (13). As shown for different C. fasciculata tryparedoxins, reduction of tryparedoxin by trypanothione is much slower than the subsequent reduction of the peroxidase by the dithiol protein, which results in a turnover number for the overall reaction of about 5-15 s Ϫ1 (12,44).

CONCLUSIONS
The identification of the new peroxidases demonstrates that two different hydroperoxide metabolizing systems occur in T. brucei, both based on the unique trypanothione metabolism of the parasites. The finding that the T. brucei peroxidase shows preference for the trypanothione/tryparedoxin couple is not surprising since only for the cytosolic classical GPX 1 a high specificity for GSH has been demonstrated (45). In addition, it underlines the difficulty to deduce the catalytic properties of an enzyme solely from its primary structure. The relatively weak performance of both types of trypanothione-dependent peroxidase systems is in accordance with the sensitivity of trypanosomes toward oxidative stress. In addition, it cannot be ruled out that the parasite peroxidases fulfill other tasks completely distinct from the defense against oxidative stress. Even within the classical selenocysteine peroxidases, only GPX 1 unambiguously functions as an antioxidant enzyme, the others have been suggested to play different roles (19). Work is in progress to silence the peroxidase genes in bloodstream T. brucei by RNA interference to get a deeper insight in the physiological function(s) of these proteins.