Characterization of Trypanosoma brucei gamma-glutamylcysteine synthetase, an essential enzyme in the biosynthesis of trypanothione (diglutathionylspermidine).

The parasitic protozoan Trypanosoma brucei maintains redox balance by synthesizing a conjugate of glutathione and spermidine termed trypanothione. The first committed step in the biosynthesis of glutathione, and thereby trypanothione, is catalyzed by the enzyme γ-glutamylcysteine synthetase (γGCS). We have cloned and sequenced the 2037-base pair gene coding for the catalytic subunit of T. brucei γGCS. T. brucei γGCS appears to be encoded by a single copy gene. A transcript of about 2.3 kilobases was observed in procyclic trypomastigotes. The deduced amino acid sequence of 679 amino acids shares 45, 41, and 36% sequence identity with mammalian, Caenorhabditis elegans, and yeast γGCS, respectively. The T. brucei γGCS gene was expressed in E. coli; the purified 77.4-kDa enzyme catalyzes the ligation of L-Glu to L-Cys with a kcat of 10 s−1, confirming that the gene encodes the functional catalytic subunit of γGCS. The apparent Km values measured for the three natural substrates L-Glu, L-Cys, and ATP are 0.24, 0.69, and 0.07 mM, respectively. Unlike the mammalian enzyme, L-α-aminobutyrate (apparent Km = 10 mM) is a poor substitute for L-Cys in the T. brucei γGCS-catalyzed reaction. T. brucei γGCS is feedback-inhibited by glutathione (apparent KI = 1.1 mM), and it is inactivated by cystamine and buthionine sulfoximine. The kinetic properties of recombinant T. brucei γGCS suggest that the substrate binding pocket and the mechanism of enzyme regulation differ from the mammalian enzyme, providing evidence that T. brucei γGCS could be a selective chemotherapeutic target for the treatment of trypanosomiasis.

The parasitic protozoa Trypanosoma brucei is the causative agent of African sleeping sickness in humans and nagana in cattle (1). As current chemotherapy is unsatisfactory, metabolic differences between trypanosomes and their mammalian hosts are being characterized to elucidate new potential drug targets in the parasite. Major differences have been found in the utilization of the tripeptide thiol glutathione (GSH). 1 Mammals rely on the antioxidant GSH to protect against oxidative injury by peroxides or free radicals and detoxification of xenobiotics (2,3). In place of GSH, trypanosomes utilize trypanothione (diglutathionylspermidine), a conjugate of GSH and spermidine, to maintain the redox balance of the cells (4). Trypanothione is synthesized in four steps via the synthesis of GSH and its subsequent conjugation to spermidine. GSH is synthesized by two enzymes, which are in common with mammalian cells, while the conjugation of GSH to spermidine is catalyzed by two trypanosome-specific enzymes (5).
The first and rate-limiting step in the biosynthesis of GSH is catalyzed by ␥-glutamylcysteine synthetase (␥GCS; Reaction 1).
L-Glutamate ϩ L-cysteine ϩ ATP ␥GCS 3 L-␥-glutamyl-L-cysteine ϩ ADP ϩ P i REACTION 1 A specific inhibitor of ␥GCS, buthionine sulfoximine, cures or prolongs survival of mice infected with T. brucei, implicating ␥GCS as a potential drug target (6). The effectiveness and selectivity of buthionine sulfoximine against T. brucei infection suggests that trypanosomes are more sensitive to GSH depletion than are mammalian cells. Trypanosomes have been reported to lack catalase (7), which in mammals breaks down hydrogen peroxide in the peroxisome. Consequently, trypanosomes possess an intracellular hydrogen peroxide concentration higher than that found in mammalian cells, which may account for the detrimental effect of GSH depletion (8). The effect of oxidative stress on T. brucei survival is also illustrated by the finding that lysis of the cattle variant of T. brucei by a component in human serum is likely to be mediated by H 2 O 2 (9). Despite the importance of ␥GCS to trypanosome viability, the enzyme has never been characterized.
Mammalian ␥GCS consists of a catalytic or heavy subunit (70 kDa) and a regulatory or light subunit (30 kDa) (10). The isolated heavy subunit retains all of the catalytic activity including feedback inhibition by GSH (11); the regulatory subunit modulates the activity of the catalytic subunit by increasing the affinity of the enzyme for L-Glu and decreasing its affinity for GSH (12). The catalytic subunit of ␥GCS has been cloned from rat (2), human (13), Saccharomyces cerevisiae (14), Schizosaccharomyces pombe (15), Caenorhabditis elegans (16), Escherichia coli (17), and Arabidopsis thaliana (18). Because the E. coli and plant enzymes share minimal sequence identity with the other eukaryotic enzymes, it has been suggested that they are unrelated (17,18).
It is not known if the regulatory subunit, which has been cloned from both human and rat kidney (19,20), is a component of the nonmammalian enzymes. The other eukaryotic ␥GCS which has been purified and characterized is from the worm, Ascaris suum, and this enzyme was isolated as a 70-kDa monomer, suggesting that, if a regulatory domain had been present, it was not tightly associated with the catalytic subunit (21).
In order to characterize structural and functional differences between parasite and host ␥GCS we undertook the cloning of the T. brucei ␥GCS gene. A partial clone was obtained by PCR using degenerate oligonucleotide primers designed to conserved sequences found in the mammalian and yeast enzymes. A full-length clone of the T. brucei ␥GCS gene was obtained from a genomic library and encodes a 679-amino acid protein (molecular mass ϭ 77.4 kDa) sharing 45, 41, and 36% sequence identity with the mammalian catalytic subunit, C. elegans, and yeast ␥GCS, respectively. Recombinant T. brucei ␥GCS was expressed in E. coli and purified. Kinetic characterization of T. brucei ␥GCS suggests that mediation of enzyme activity by a regulatory domain is unlikely to be necessary in vivo.

EXPERIMENTAL PROCEDURES
Materials-Reagents for the enzyme assay were purchased from Sigma, Ni 2ϩ -agarose was purchased from Qiagen, and TEV protease was purchased from Life Technologies, Inc.
DNA and RNA Isolation from T. brucei-Genomic DNA and total RNA was isolated as described (22,23) from procyclic 427 or 366D T. brucei cells cultured in SDM-79 medium or modified Steiger's medium, respectively (24).
Library Construction and Screening-T. brucei genomic DNA from procyclic 366D cells was digested with a mixture of AluI, HaeIII, and RsaI under conditions which produced partial digestion. The DNA was methylated with EcoRI methylase, ligated to EcoRI linkers, and digested with EcoRI. The DNA was applied to a Bio-Rad Bio-Gel DNA XL column for size selection of 2-10-kb DNA. The DNA was ligated to Zap II arms (digested with EcoRI and dephosphorylated), and packaged with Gigapack II Gold packaging mix as recommended by Stratagene. The unamplified library contained 2.5⅐10 5 plaque-forming units with an average insert size of 2 kb. The library was amplified as described by the manufacturer (Stratagene).
The cloned BamHI/EcoRI PCR fragment was isolated by restriction digestion and agarose gel electrophoresis, labeled with [ 32 P]dCTP by the random primer method (Boehringer Mannheim), and used to screen approximately 6.2⅐10 5 plaques from the amplified T. brucei genomic library. Phage were transferred to nylon membranes and hybridized to the labeled probe at high stringency (hybridization conditions: 42°C in 5 ϫ SSPE, 1 ϫ Denhardt's, solution, 1% SDS, 50% formamide). Positive plaques were isolated and replated until single hybridizing plaques were obtained. Plasmids were rescued from these phage by co-infection with R408 helper phage as recommended by the manufacturer (Stratagene).
DNA Sequencing and Analysis-DNA sequencing was done using the dideoxynucleotide chain-termination method (25). Comparative sequences were obtained from GenBank and Swiss-Prot sequence data bases. Sequence alignment and analysis was performed using the Pileup, Profilemake, and Profilesearch programs from the GCG Wisconsin Sequence Analysis Software Package.
Analysis of Nucleic Acids-T. brucei genomic DNA (6 -10 g) was digested with various restriction endonucleases, electrophoresed on 1.0% agarose gels, and transferred to nylon membranes. Total RNA (50 g) was fractionated on a 1.0% formaldehyde agarose gel before transfer to a polyvinylidene difluoride membrane (26). DNA and RNA blots were hybridized with the random primer 32 P-labeled PCR fragment (10 -50 ng) and washed as described above for library screening. The DNA and RNA blots were autoradiographed at Ϫ80°C for various periods of time.
Dot Blot Analysis-Serial dilution's of genomic DNA were immobilized on nylon membranes and probed with equal length 32 P-labeled DNA fragments from either the ␥GCS gene or the ornithine decarboxylase gene (27). The probes were also hybridized to Dot Blots containing plasmid DNA to control for variations in specific radioactivity. The Dot Blots were analyzed using a PhosphorImager.
Expression of ␥GCS in E. coli-The open reading frame of ␥GCS from Tb12.1 was cloned into a modified pET-11d vector (28). As described previously for ornithine decarboxylase (29), this vector directs the expression of the foreign gene from the T7 promoter with an N-terminal His 6 -tag to facilitate purification. The His 6 -tag can be removed after purification by digestion with TEV protease (30). An oligonucleotide (5Ј-GTCCTAATTCGGTACCCAGAA-3Ј) was designed to incorporate an NcoI site into Tb12.1 at the 5Ј end of the ␥GCS gene by Kunkel mutagenesis (31). Mutagenesis was confirmed by DNA sequence analysis. Because of the presence of an internal NcoI in the ␥GCS, the gene was cloned into the NcoI/HindIII sites of the expression vector in two pieces. The 5Ј end of T. brucei ␥GCS expressed from this construct (pTbGCS) is M-H 6 -AENLYFQGAMGLL. The underlined amino acids represent the TEV protease site (30), and the residues in italics are the first 4 amino acids of T. brucei ␥GCS.
Enzyme Purification-E. coli BL21/DE3 cells (30) containing pTb-GCS were grown from single colonies at 37°C in Luria-Bertani medium to A 600 ϭ 0.8, and expression of the protein was induced by addition of isopropyl-␤-D-thiogalactopyranoside to a final concentration of 0.2 mM. The temperature was lowered to 30°C, and the cells were harvested 4 -6 h after induction. Ni 2ϩ -agarose column chromatography was performed as described previously (29). The His-tag was removed by proteolysis with His-tagged TEV protease (Life Technologies, Inc.), and ␥GCS without the tag was separated from tagged enzyme and TEV protease by reapplication to Ni 2ϩ -agarose. Untagged enzyme was recovered in the flow through. Approximately 4 mg of purified ␥GCS was obtained per liter of culture. The protein eluted from the Ni 2ϩ -agarose column was concentrated and applied to a Hi/Load Superdex 200 16/60 gel filtration column (Pharmacia Biotech Inc.) in 50 mM Tris⅐HCl, pH 8.0, 100 mM NaCl, and 5 mM MgCl 2 .
Enzyme Assays-␥GCS activity was followed at 37°C using a spectrophotometric assay which couples ADP production to NADH reduction via pyruvate kinase and lactate dehydrogenase as described (12). Briefly, 5 units of type III rabbit muscle pyruvuate kinase (Sigma) and 10 units of type II rabbit muscle lactic dehydrogenase were mixed with buffer (100 mM Tris⅐HCl, pH 8.0, 150 mM KCl, 20 mM MgCl 2 , 2 mM phosphoenolpyruvate, 0.2 mM NADH) and ␥GCS substrates. The reaction was initiated by the addition of ␥GCS. L-␣-Aminobutyric acid was used in place of L-Cys unless specified. For specific activity measurements the concentrations of ATP, L-␣-aminobutyric acid, and L-Glu were 5, 50, and 10 mM, respectively. For determination of K m , two substrates were held at saturating levels, while the third was varied by 10-fold around the putative K m . ␥GCS concentration ranged from 1 to 5 g/ml. Control reactions were done in the absence of L-Glu to confirm that the ATP hydrolysis was ␥GCS-dependent. The time-dependent inhibitors, buthionine sulfoximine and cystamine, were incubated with ␥GCS for 10 min at concentrations of 1 and 10 mM, and of 0.5 and 5 mM, respectively. The method of Lowry (32) was used to determine protein concentrations. Michaelis-Menten parameters were calculated using the k cat program (Biometallics, Inc.).
Amino Acid Sequencing-Purified protein was subjected to SDS-PAGE and electroblotted to polyvinylidene difluoride paper as described elsewhere (33). Edman degradation was performed on an Applied Biosystems model 470A automated sequencer using standard manufacturer's chemicals and programming.

RESULTS
Isolation and Characterization of Clones-Degenerate oligonucleotide primers were designed to two amino acid sequence regions conserved between the rat kidney catalytic subunit (34) and S. cerevisiae ␥GCS (14), corresponding to the motif MGF-GMG (residues 240 -245 of rat ␥GCS) and to HFENIQST-NWQT (residues 396 -407 of rat ␥GCS). A 500-bp fragment was amplified from the primers using T. brucei genomic DNA and cloned into the BamHI and EcoRI sites of pBluescript. Sequence analysis verified that this DNA fragment encoded an ORF with sequence similarity to rat and yeast ␥GCS. The cloned PCR fragment was labeled with 32 P and used to screen a T. brucei genomic library (trypanosome genes lack introns) (35). Seven clones were isolated which contained overlapping DNA fragments of 1.5-5.8 kb. Clone Tb12.1 (insert size 5.5 kb) contained a full-length 2037-bp ORF which was sequenced on both strands. The ORF is predicted to encode a 679-amino acid polypeptide of 77.4 kDa (Fig. 1).
Comparative Data Base Analysis of T. brucei ␥GCS-The deduced T. brucei ␥GCS amino acid sequence shares 36 -45% sequence identity with the catalytic subunits of ␥GCS from rat (34), human (13), C. elegans (16), S. cerevisiae (14), and S. pombe (15), with 146 invariant residues identified among all six sequences. A unique 60-amino acid insertion is found in T. brucei ␥GCS at position 242. An alignment of representative eukaryotic ␥GCS sequences is displayed in Fig. 2A. Two other ␥GCS sequences have been reported for E. coli and A. thaliana; however, it has been suggested that they may not be related to the mammalian and yeast enzymes (18).
Analysis of the function of conserved amino acid residues in the ␥GCS family has been limited by the small number of organisms for which sequence information was available. The cloning of the T. brucei gene allows us to identify sequence motifs which have been conserved among organisms with diverse evolutionary backgrounds; T. brucei is thought to have originated as one of the most primitive eukaryotic lineages (36). A profile was constructed from the mammalian, yeast, C. elegans, and T. brucei sequences and was used to search the Swissprot data base. A. thaliana ␥GCS was identified by this search (Z score, 3.18), whereas E. coli ␥GCS was not. A. thaliana ␥GCS shares an average sequence identity with the other eukaryotic enzymes of 18% over the full-length of the sequence. Notably, it shares 35 out of the 146 invariant residues from the other sequences, with the region between residues 315 and 365 displaying the greatest similarity (Fig. 2B). Previous analysis had not identified this region as one of the conserved sequence blocks (18). Included in the invariant residues are Cys 319 and seven positively charged amino acids (T. brucei residues Arg 190 , Arg 191 , Arg 377 , Arg 474 , Lys 476 , Arg 491 , Arg 625 ).
Nucleic Acid Analysis-The ␥GCS gene structure and organization within the T. brucei genome was mapped by Southern blot analysis. T. brucei genomic DNA was digested with various restriction enzymes and probed with the labeled PCR fragment as described above (Fig. 3). The probe hybridizes to only a single band when DNA was digested with enzymes that do not cut within its sequence (EcoRI, BamHI, and SacI), whereas the probe hybridizes to two fragments when the DNA was cut with an enzyme that cuts at a single site within the probe (XbaI). These results are consistent with the profile of a single copy gene. Gene copy number was also tested by Dot Blot analysis of genomic DNA. Dot Blots were probed with equal length DNA fragments from either the ␥GCS gene or the ornithine decarboxylase gene, a known single copy gene (27), as described under "Experimental Procedures." The signal intensities (corrected for variations in probe specific activity) and amount of DNA which could be detected were the same for both probes providing further evidence that T. brucei ␥GCS is a single copy gene (data not shown).
RNA blot analysis confirmed that the trypanosomal ␥GCS gene is actively transcribed. A single 2.3-kb transcript was observed in total RNA isolated from procyclic 427 cells (Fig. 4).
Kinetic Characterization of ␥GCS-The ␥GCS gene contained in Tb12.1 was subcloned into an expression vector which utilizes the T7 promoter to direct the expression of ␥GCS fused to an N-terminal six-histidine tag followed by the TEV protease cleavage site (30). The soluble protein was purified from the E. coli cell extracts as described in Experimental Procedures. A single band of approximately 75 kDa was observed by SDS-PAGE analysis (Fig. 5). The identity of the band was confirmed to be T. brucei ␥GCS by N-terminal amino acid sequencing. The apparent molecular mass of T. brucei ␥GCS estimated by chro- The nucleotide sequence is that of the sense strand. The coding sequence begins at nucleotide ϩ1 and ends at nucleotide ϩ2037. The single letter code for amino acids is used. The stop codon is underlined. The GenBank accession no. is U56818. matography on the Superdex 200 column is 75 kDa, suggesting that it is a monomer in solution.
The purified recombinant T. brucei ␥GCS catalyzes the ligation of L-Glu with L-␣-aminobutyric acid or L-Cys in the presence of ATP with a specific activity of 7.6 mol/min/mg of protein, which corresponds to a k cat of 9.8 s Ϫ1 . The specific activity is similar to the activity reported for the rat catalytic subunit (16 units/mg) (12,20) or for the single subunit enzyme purified from A. suum (18 units/mg) (21). For T. brucei ␥GCS the K m app for L-Glu is 6-and 75-fold lower than for the rat holoenzyme complex and the rat catalytic subunit, respectively, while it is 4-fold lower than for the A. suum enzyme (Table I). In contrast, the K m app for L-␣-aminobutyric acid is 10 -30-fold higher than reported for the rat or the A. suum enzymes. Additionally, the selectivity of T. brucei ␥GCS for L-Cys over L-␣-aminobutyrate (measured by the ratio of K m app for L-Cys to K m app for ␣-aminobutyrate) is 4-and 20-fold greater than that observed for the rat or A. suum catalytic subunits, respectively ( Table I). The K m app for ATP is 0.071 Ϯ 0.01 mM, similar to that reported for the rat holoenzyme (12,20). Like rat ␥GCS (37), T. brucei ␥GCS is inactivated by both cystamine (87% inhibition was observed at 0.5 mM) and buthionine sulfoximine (66% inhibition was observed at 1 mM), and it is feedback-inhibited by GSH. GSH inhibition is competitive with respect to L- Glu   FIG. 3. Southern blot analysis of T. brucei genomic DNA. T. brucei genomic DNA (6 -10 g) was digested with the indicated restriction endonuclease, electrophoresed, and blotted. The blot was probed at high stringency with the 32 P-labeled cloned PCR fragment (10 -50 ng). with a K I app ϭ 1.1 Ϯ 0.2 mM slightly higher than the reported cellular concentrations of GSH in T. brucei (38). DISCUSSION We have cloned the T. brucei ␥GCS gene which codes for a 679-amino acid, 77.4-kDa protein. Although many trypanoso-mal genes are found in multiple copies and in tandem arrays (39), Southern and comparative Dot Blot analyses suggest that T. brucei ␥GCS is a single copy gene. Northern blot analysis demonstrates that the T. brucei ␥GCS gene is transcribed in procyclic trypanosomes and expression of the ␥GCS gene in E. coli confirms that the T. brucei gene encodes the functional catalytic subunit of ␥GCS.
T. brucei ␥GCS shares a 36 -45% amino acid sequence identity with the other eukaryotic enzymes from mammals (rat and human), yeast (S. cerevisiae and S. pombe) and C. elegans. Profile analysis which included the T. brucei ␥GCS sequence identified A. thaliana but not E. coli ␥GCS in the search, suggesting that A. thaliana ␥GCS may be structurally related to the other eukaryotic enzymes. Of the amino acids which are invariant in all of the eukaryotic enzymes, 7 invariant positively charged amino acids and a single Cys were identified. The structures of other enzymes which bind GSH (e.g. GSH S-transferase (40) and GSH reductase (41)) reveal that Arg residues are found as common components of the substrate and GSH carboxylate binding sites, suggesting that the conserved Arg residues maybe involved in the binding of L-Glu and L-Cys by ␥GCS.
A role for a conserved Cys is suggested by the observation that the eukaryotic enzymes, including as we have demonstrated, T. brucei ␥GCS, are inactivated by cystamine. Cystamine inactivation of ␥GCS is reversed by dithiothreitol providing evidence that the mechanism of inactivation is through disulfide bond formation to a Cys residue (42)(43)(44). L-Glu protects the enzyme from cystamine inactivation suggesting that the susceptible Cys residue is in or near the active site (45). As the only conserved Cys residue, Cys 319 is likely to mediate the response to cystamine. The reaction catalyzed by ␥GCS is thought to proceed by transfer of the ␥-phosphate of ATP to L-Glu to form a ␥-glutamylphosphate intermediate; this intermediate is attacked by L-Cys to form products (46). Thus the ␥-phosphate of ATP must be positioned in close proximity to the L-Glu binding site. While there is no universal ATP binding site, a glycine-rich loop (e.g. the P-loop) is a common component of phosphate binding sites (47,48). ␥GCS does not contain a classic P-loop motif but it does contain a conserved glycine-rich sequence which directly precedes the invariant Cys and is characterized by the motif M(A/G)FGMGXXCLQ.
The substrate specificity profile of recombinant T. brucei ␥GCS differs significantly from the profiles reported for the rat or A. suum enzymes and suggests it will be possible to design selective inhibitors of the parasite enzyme. T. brucei ␥GCS has a stronger preference for L-Cys over ␣-aminobutyrate when compared to the rat enzyme, while the A. suum enzyme has a slight preference for ␣-aminobutyrate. These results suggest that the T. brucei binding pocket for L-Cys must differ from the binding pockets of the other enzymes.
In addition to these differences, the T. brucei ␥GCS catalytic subunit has a much greater apparent binding affinity for L-Glu  than the rat catalytic subunit (Table I). In the rat holoenzyme, the effect of the regulatory subunit is to increase the affinity of the catalytic subunit for L-Glu by reducing the K m app from 18 mM to 1 mM, and to moderate the feedback inhibition by GSH by increasing the K I app from 2 to 10 mM (12,20). Our results demonstrate that T. brucei ␥GCS does not require a regulatory subunit to create a high affinity L-Glu binding site. While for T. brucei ␥GCS the apparent K I for GSH is similar to that reported for the rat catalytic subunit and lower than that for the rat holoenzyme (Table I), the intracellular concentration of GSH has also been reported to be lower in T. brucei cells (0.4 mM) (38) relative to mammalian cells (5 mM) (2). Thus, with respect to their cellular environments, monomeric T. brucei ␥GCS should be as efficient as the holoenzyme complex of the mammalian enzyme. While our results do not rule out a second subunit component of T. brucei ␥GCS, they do lead to the conclusion that the role of any such regulatory subunit must be markedly different from that reported for the rat enzyme. The most striking difference between the T. brucei and rat ␥GCS sequences is the unique 60-amino acid insertion in the T. brucei enzyme at position 242. Perhaps this insertion is involved in replacing the role of the regulatory subunit in promoting high affinity binding to L-Glu. Further analysis of both the recombinant and native T. brucei ␥GCS will allow us to address these questions.