Intersubunit Interactions in Plasmodium falciparumThioredoxin Reductase*

The thioredoxin redox system is composed of the NADPH-dependent homodimeric flavoprotein thioredoxin reductase (TrxR) and the 12-kDa protein thioredoxin. It is responsible for the reduction of disulfide bridges in proteins such as ribonucleotide reductase and several transcription factors. Furthermore, thioredoxin is involved in the detoxification of hydrogen peroxide and protects the cell against oxidative damage. There exist two classes of TrxRs: the high M r and the lowM r proteins. The well characterizedEscherichia coli TrxR represents a member of the lowM r class of proteins, whereas the mammalian,Caenorhabditis elegans, and Plasmodium falciparum proteins belong to the family of highM r proteins. The primary structure of these proteins is very similar to that of glutathione reductase and lipoamide dehydrogenase. However, the high M r TrxRs possess, in addition to their redox active N-terminal pair of cysteines, a pair of cysteine residues or a selenenylsulfide motif at their C terminus. These residues have been shown to be crucial for the reduction of thioredoxin. In this study we address the question whether the active site residues of P. falciparum TrxR are provided by one or both subunits. Differentially tagged wild-type andPfTrxR mutants were co-expressed in E. coli and the recombinant protein species were purified by affinity chromatography specific for the respective tags of the recombinant proteins. Co-expression of PfTrxR wild-type and mutant proteins resulted in the formation of three different protein species: homodimeric PfTrxR wild-type proteins, homodimeric mutant proteins, and heterodimers composed of onePfTrxR wild-type subunit and one PfTrxR mutant subunit. Co-expression of the double mutant PfTrxRC88AC535A with PfTrxR wild-type generated an inactive heterodimer, which indicates that PfTrxR possesses intersubunit active sites. In addition, the data presented possibly imply a coopertive interaction between both active sites of PfTrxR.

Infection with Plasmodium falciparum, the causative agent of malaria tropica, is responsible for 2-3 million deaths per year. The malaria parasite spends part of its developmental life cycle in human erythrocytes where it is challenged with enhanced oxidative stress. Therefore the parasite needs efficient anti-oxidants to protect itself against damages, such as nucleic acid modifications, lipid peroxidation, or oxidation of thiolcontaining proteins, caused by reactive oxygen species. The thioredoxin redox system, composed of the NADPH-dependent homodimeric thioredoxin reductase (TrxR) 1 and the 12-kDa protein thioredoxin (Trx) confers reduction of protein disulfides, ribonucleotide reductase being the most prominent example (1). Apart from this, thioredoxin interacts with a number of transcription factors in prokaryotic and eukaryotic cells, resulting in modified DNA binding activities and altered gene transcription (2). Another important function is the interaction of reduced thioredoxin with thioredoxin-dependent peroxidases, which detoxify reactive oxygen species and aid in the prevention of oxidative damage in the cell (3).
There exist two classes of TrxRs, the low M r proteins are represented by the 35-kDa Escherichia coli TrxR; the high M r proteins were identified in mammals, P. falciparum, and also the free living nematode Caenorhabditis elegans (4 -7). The two classes of proteins do not only differ in their primary structure but also in their reaction mechanisms. E. coli TrxR achieves catalysis by a conformational change (8,9), whereas high M r TrxRs transfer reducing equivalents to their substrate by an intramolecular dithiol/disulfide interchange between the Nterminal redox active cysteines and the C-terminal pair of cysteines (10 -12). The primary structure of PfTrxR resembles that of glutathione reductase and lipoamide dehydrogenase with respect to the location of the N-terminal redox active cysteine pair, and the occurrence of the C-terminal pair of cysteines is similar to mercuric ion reductase. In mercuric ion reductase it was shown that the C-terminal residues are involved in the catalytic process but are not redox active as in TrxRs (13,14).
To achieve transfer of electrons from the N terminus to the C terminus, the redox active residues should be located in close proximity to each other. There are two possible conformations which allow this transfer: either by the formation of an intermolecular interaction between the two subunits of the dimeric protein or by an intrasubunit interaction where each of the subunits forms its own active site (Fig. 1). In mercuric ion reductase an intermolecular interaction between both subunits occurs (15). In glutathione reductase and lipoamide dehydrogenase the active sites are also formed by an interaction of both subunits; the redox active cysteine residues are provided by one subunit, and the acid/base catalyst is provided by the second subunit (16).
To analyze the organization of the PfTrxR active sites, we decided to co-express wild-type and mutant PfTrxR (PfTrxR-C88A, PfTrxRC535A, or PfTrxRC88AC535A) in E. coli and to separate the recombinant proteins. Using this method we expected to generate homodimeric enzymes consisting of wildtype or mutant subunits only and heterodimeric proteins consisting of one wild-type and one mutant subunit. Co-expressing single or double mutant PfTrxRs with PfTrxR wild-type and characterizing the recombinant protein species should give a clear picture of how the active sites of PfTrxR are formed and answer the question whether the homodimeric protein possesses intersubunit or intrasubunit active sites (Fig. 1).
DNA Manipulation and Nucleotide Sequence Analyses-Routine DNA manipulations and transformations were performed as described by Sambrook et al. (17). The plasmids containing the genetically modified PfTrxR constructs were sequenced using the Sanger dideoxy-chain termination reaction for double-stranded DNA (17).
P. falciparum TrxR Mutants-The PfTrxR mutants PfTrxRC88A and PfTrxRC535A were generated previously (18,19). The double mutant PfTrxRC88AC535A was generated according to Gilberger et al. (18) using the mutagenic oligonucleotides sense 5Ј-GCTAAAGGAGGAGCT-GGGGGTGGAAAATGTGG-3Ј and antisense 5Ј-TCACATTTTCCACCC-CCAGCTCCTCCTTTAGC-3Ј and PfTrxRC88A in pJC40 as a template. The introduction of the second mutation was verified by nucleotide sequence analysis of the entire insert as described above.
Construction of Expression Plasmids-To guarantee co-expression and co-existence of two individual plasmids in one E. coli cell, they need to have distinct origins of replication and selectable markers. We intended to use one plasmid containing an N-terminal Strep-tag and one plasmid containing an N-terminal His-tag, which would result in the formation of differentially tagged recombinant proteins. The His-tag plasmids (pSZ1, pSZ1TrxRC88A, pSZ1TrxRC535A, and pSZ1TrxRC88AC535A) were constructed as shown in Fig. 2. The plasmid pACYC184 (carrying the p15A origin of replication and a chloramphenicol acetyltransferase as selectable marker) was digested with SalI, and the 5Ј-overhang was filled in with T4 DNA polymerase according to standard methods (17). The expression cassette of pJC40 was isolated using DrdI, and the 3Ј-overhang was blunted with T4 DNA polymerase. The SalI-digested plasmid pACYC184 and the isolated pJC40 expression cassette were digested with BspHI and ligated using T4 DNA ligase, resulting in the expression plasmid pSZ1. To obtain the expression plasmids pSZ1TrxRC88A, pSZ1TrxRC535A, pSZ1TrxRC88AC535A, respectively, the PfTrxR mutants PfTrxRC88A, PfTrxRC535A, and PfTrxRC88AC535A were amplified by PCR using Pfu polymerase and plasmid DNA containing the respective mutant PfTrxR as template with the following PCR program: 95°C (1 min) and 30 cycles of 95°C (1 min), 50°C (1 min), and 68°C (3 min). The sense oligonucleotide 5Ј-GCGCCCCGGGCATGTGTAAAGATAAAAACG-3Ј coding for the first 6 amino acids of PfTrxR contained a SmaI restriction site. The antisense oligonucleotide 5Ј-GCGCGGGCCCTTATCCACATTTTCCACCCCC-3Ј To obtain a plasmid that possesses a p15A origin of replication and allows expression of a recombinant protein with an N-terminal His tag, the plasmid pSZ1 was constructed. The expression cassette of pJC40 was subcloned into pACYC184 as described under "Experimental Procedures." encoding the last 6 amino acids of PfTrxR generated an ApaI restriction site to facilitate directional cloning. The expression plasmid pSZ1 and the gel-purified mutant PfTrxR inserts were double digested with SmaI and ApaI and ligated using T4 DNA ligase.
Wild-type PfTrxR was amplified by PCR using Pfu polymerase using the same PCR protocol as described above. The sense oligonucleotide 5Ј-GCGCGCGGTCTCGGCGCTGTAAAGATAAAAACG-3Ј coding for the first 6 amino acids and the antisense oligonucleotide 5Ј-GCGCGCG-GTCTCGTATCATTATCC-ACATTTTCCACC-3Ј encoding the last 5 amino acids of PfTrxR both contained BsaI restriction sites. The purified PCR product (PfTrxRWT) and the expression vector pASK-IBA7 (ColE1 origin of replication, can be selected for with ampicillin) were digested with BsaI and ligated using T4 DNA ligase, resulting in pStrepPfTrxRWT.
Co-expression and Purification of PfTrxR Homo-and Heterodimers-The expression vector pStrepPfTrxRWT was transformed into competent E. coli BLR (DE3). Subsequently the bacteria containing pStrep-PfTrxRWT were made competent and used for co-transformation with pSZ1TrxRC88A, pSZ1TrxRC535A, or pSZ1TrxRC88AC535A (pSZ1Mut), respectively. An overnight culture of the freshly co-transformed bacteria was diluted 1:50 in 3 liters of fresh Luria Bertani-medium containing 50 g ml Ϫ1 ampicillin and 35 g ml Ϫ1 chloramphenicol and grown at 37°C until the A 600 reached 0.5. Expression of wild-type protein (pStrep-PfTrxRWT) was induced using 200 g of anhydrotetracycline per liter of bacterial culture, and the temperature was reduced to 20°C. Three hours later the expression of pSZ1Mut was induced by addition of 500 M isopropyl ␤-D-thiogalactopyranoside, and the bacterial cultures were incubated overnight. The cells were harvested by centrifugation (4000 ϫ g, 10 min). The bacterial pellet was resuspended in buffer W (100 mM Tris/HCl, pH 8.0, 1 mM EDTA), sonicated (Branson Sonifier 250), and the cell lysate was centrifuged at 100,000 ϫ g for 1 h (Centrikon-T 1065, Kontron). The supernatant was applied to a StrepTactin column (2 ml) and washed with 10 ml of buffer W. Elution of the Streptag proteins was performed using 2.5 mM desthiobiotin in buffer W. Flow-through (30 ml) and eluant (5 ml) were dialyzed separately overnight against 2 l of binding buffer (20 mM Tris/HCl, pH 7.9, 5 mM imidazole, 0.5 M NaCl). Subsequently flow-through and StrepTactin eluant were applied separately to Ni 2ϩ -chelating chromatography following the manufacturer's recommendations. The purification resulted in three fractions: 1) PfTrxRWT homodimer (Strep/Strep-tagged), 2) PfTrxR mutant homodimer (His/His-tagged), and 3) PfTrxRWT/PfTrxR mutant heterodimer (Strep/His-tagged) (Fig. 3). Protein concentrations were determined either by using the molar extinction coefficient of 11340 M Ϫ1 cm Ϫ1 for the protein bound FAD at 460 nm (18) or by the method of Bradford (20) using bovine serum albumin as a standard.
The purity of the three enzyme species was analyzed on a 7.5% SDS-PAGE, and the proteins were visualized with Coomassie Brilliant Blue.
Enzyme Assays-The specific activities of homo-(wild-type and mutant PfTrxR) and heterodimeric proteins were determined by the DTNB and thioredoxin reduction assays using E. coli Trx and insulin (19). The DTNB reduction assay (1 ml, 20°C) contained 100 mM potassium phosphate buffer, pH 7.5, 0.2 mg ml Ϫ1 bovine serum albumin, 2 mM EDTA, 2 g of enzyme, 200 M NADPH, and 3 mM DTNB. The change in absorbance was monitored at 412 nm (Uvikon 932, Kontron). The E. coli Trx reduction assay (1 ml, 37°C) consisted of 100 mM HEPES buffer, pH 7.5, 0.2 mg ml Ϫ1 bovine serum albumin, 2 mM EDTA, 2 g of enzyme, 200 M NADPH, 200 g of insulin, and 100 M E. coli Trx. The change in absorbance was measured at 340 nm.

Expression and Purification of PfTrxR Homo-and Heterodimers-
The dimeric high M r TrxRs identified in mammals and P. falciparum possess one redox active pair of cysteine residues in the N-terminal region and one Cys-seleno-Cys or Cys-XXXX-Cys motif at their C terminus, respectively (6,21). It has been shown that both motifs are essential for enzymatic activity with thioredoxin as substrate (10 -12, 19). Even though it has been proposed that there exists a high degree of structural similarity between the high M r TrxRs with glutathione reductase and lipoamide dehydrogenase (10,18), the catalytic mechanism of the TrxRs appears to be much more complicated than that of the latter enzymes. Wang et al. (10,11) have shown that during catalysis reducing equivalents are transferred from FAD to the N-terminal pair of cysteines and subsequently reduce the C-terminal cysteine pair by an intramolecular dithiol/disulfide exchange. To achieve such an interchange the Nand C-terminal residues involved in this process need to be in close proximity. There are two possible arrangements one can envisage for such an interaction. One of them would be an intramolecular interaction and thus the formation of two independent active sites in the homodimeric protein; the second would be an intermolecular interaction where amino acid residues are supplied by both subunits to form the active sites of the enzyme (Fig. 1). To investigate how the active sites of PfTrxR are formed, we co-expressed different PfTrxR-species in E. coli BLR (DE3), separated the generated proteins from each other and determined their specific activities and K m values. Co-expression and subsequent purification of StrepPfTrxRWT with three constructs carrying different active site mutations resulted in the isolation of three different protein species: Strep-tagged wild-type homodimers, His-tagged mutant homodimers, and Strep/His-tagged wild-type/mutant heterodimers (Figs. 3 and 4). The protein concentrations of the three protein species in one preparation, however, was quite different. The homodimeric proteins were expressed at much higher levels than the heterodimeric proteins. There are several explanations for this phenomenon. It is possible that the heterodimers formed during co-expression are not very stable and are degraded more rapidly by the bacterial cell than the homodimeric proteins. Another reason might be the synthesis of monomers that do not dimerize co-translationally and are therefore degraded rapidly by the bacterial cell. It has been shown previously that monomeric mouse glutathione reductase is degraded rapidly by bacterial cells (22). The generation of E. coli glutathione reductase hybrid protein species resulted in high yields of proteins, and there was no mention about a differential expression pattern concerning the different protein species that were generated (23). Similarly, the co-expression of mercuric ion reductase hybrids apparently did not result in different yields of the hybrid proteins (15).
Specific Activities of Recombinant PfTrxR Homo-and Heterodimers-The specific activities of the co-expressions with wild-type and three mutant PfTrxRs were determined using two different assay systems. The results are summarized in Table I. The specific activities of the Strep-tagged wild-type PfTrxR are comparable with those of preparations of nontagged recombinant PfTrxR in both assays (4.5 Ϯ 1.5 units mg Ϫ1 without tag versus 3.9 Ϯ 0.5 units mg Ϫ1 with a Strep-tag using 100 M E. coli thioredoxin and 6.2 Ϯ 0.8 units mg Ϫ1 versus 5.6 Ϯ 1.6 units mg Ϫ1 using DTNB).
The activities of PfTrxRWT and PfTrxRC88A with E. coli thioredoxin are in good accordance with the expected results regardless whether the active sites of PfTrxR are formed by an intersubunit or intrasubunit interaction. The homodimers are either 100% active (PfTrxRWT, 3.9 units mg Ϫ1 ) or show a very low activity of 0.4 unit mg Ϫ1 (PfTrxRC88A). It has been shown previously that the mutation of Cys 88 into alanine resulted in a loss of DTNB and thioredoxin reducing activity and thus that this residue is essential for the catalytic activity of PfTrxR (18). The hybrid protein formed between PfTrxRWT and PfTrxRC-88A has 56% of wild-type activity, which suggests that the protein possesses one fully active catalytic site and one inactive catalytic site. Using DTNB as a substrate the results are similar. The heterodimer formed between wild-type and mutant subunit exhibits similar kinetic parameters to those of the wild-type protein (K m PfTrxRWT/mutant heterodimer: 384 Ϯ 59 M versus K m PfTrxRWT: 429 Ϯ 57 M), indicating that the C88A mutation is only acting on one of the active sites, and no interaction is transmitted across the interface to the second active site.
The alteration of Cys 535 into alanine rendered the protein  inactive toward E. coli thioredoxin, and the DTNB reduction by this mutant was decreased to about 60% of wild-type activity (19). In accordance PfTrxRWT in this co-expression experiment is fully active, and the C535A mutant homodimer is almost completely inactive using E. coli thioredoxin as a substrate and maintains about 60% of its DTNB reducing activity. The heterodimer between PfTrxRWT and PfTrxRC535A subunits maintains only 42% of residual activity rather than 50%, for the reduction of E. coli thioredoxin. This slight discrepancy could possibly be explained by contamination of this protein fraction with inactive homodimers of mutant protein, which may occur during the purification procedure. The results obtained in the DTNB reduction assay are, however, more difficult to explain. The heterodimers between PfTrxRWT and PfTrxRC535A show a residual activity of 55%, which is much lower than the expected 80%, taking into account that one fully active and one slightly impaired catalytic site (60% residual activity) are formed in this enzyme hybrid (Table I). Interestingly the K m value for DTNB reduction is higher than for the wild-type protein (678 Ϯ 1 M versus 429 Ϯ 57 M) and lower than for the PfTrxRC535A homodimer (678 Ϯ 1 M versus 800 Ϯ 22 M). These results suggest that the alteration of Cys 535 in one active site possibly results in subtle changes in the reactivity of the second active site, which may interfere with DTNB and also thioredoxin reduction. An asymmetry in the catalytic activity of the two active sites of PfTrxR occurring during the interchange of electrons between the N-terminal cysteine residues and the C-terminal cysteine residues has been previously proposed by Wang et al. (24), and asymmetry of the active sites occurs in several other members of this enzyme family (25)(26)(27). The data presented here support this hypothesis, and these possible cooperative interactions will be investigated in more detail in future studies.
The key experiment that allows us to draw final conclusions about the formation of the PfTrxR active sites was the coexpression of PfTrxRWT with the double mutant PfTrxRC-88AC535A. Both enzymatic assays show the results expected when the active sites of PfTrxR are formed by an intersubunit interaction between both subunits of the dimeric protein (Fig.  5). The heterodimers between PfTrxRWT and PfTrxRC88AC-535A were inactive and the homodimeric mutant proteins also did not show an appreciable enzymatic activity with thioredoxin (Table I). Therefore it can be concluded that the interaction of the N-terminal cysteine residues Cys 88 and Cys 93 with the protein-bound FAD occurs on one subunit, and the reducing equivalents are then transferred to the C-terminal cysteine residues Cys 535 and Cys 540 of the second subunit. The thiolate anions are most likely stabilized by the acid/base His 509 also provided by the second subunit (18). This conformation resembles those of the glutathione reductase, lipoamide dehydrogenase, and mercuric ion reductase. Perham and co-workers (23,28) have demonstrated a similar interaction between the subunits in E. coli glutathione reductase by separating homo-and heterodimers by engineering an arginine-tag to one of the subunits and co-expressing the arginine-tagged and unmodified proteins in E. coli. Neither of the hybrid proteins formed in this study had any influence on the kinetic parameters, indicating that in glutathione reductase no cooperative effects between the two subunits occur (29). In mercuric ion reductase, which like PfTrxR possesses an additional pair of Cys residues at the C-terminus, intersubunit active sites are formed (15). A cooperative interaction between both active sites has been suggested in mercuric ion reductase (25) similar to our observations on PfTrxR described in Wang et al. (24) and in this study. The generation of heterodimeric PfTrxR protein species will facilitate future studies on the catalytic mechanism of the protein, such as potential cooperative interactions between the active sites during catalysis (25,26,30). In addition the knowledge of the architecture of the PfTrxR catalytic centers allows the overexpression of dominant negative mutants in P. falciparum, which will help to determine the biological role of the protein in the parasites. Ultimately this approach will aid to validate and assess P. falciparum thioredoxin reductase as a target for a chemotherapeutic intervention of malaria.