Trypanothione Synthesis in Crithidia Revisited*

In Crithidia fasciculata the biosynthesis of trypanothione (N1,N8-bis(glutathionyl)spermidine; reduced trypanothione), a redox mediator unique to and essential for pathogenic trypanosomatids, was assumed to be achieved by two distinct enzymes, glutathionylspermidine synthetase and trypanothione synthetase (TryS), and only the first one was adequately characterized. We here report that the TryS of C. fasciculata, like that of Trypanosoma species, catalyzes the entire synthesis of trypanothione, whereas its glutathionylspermidine synthetase appears to be specialized for Gsp synthesis. A gene (GenBank™ accession number AY603101) implicated in reduced trypanothione synthesis of C. fasciculata was isolated from genomic DNA and expressed in Escherichia coli as His-tagged or Nus fusion proteins. The expression product proved to be a trypanothione synthetase (Cf-TryS) that also displayed a glutathionylspermidine synthetase, an amidase, and marginal ATPase activity. The dual specificity of the Cf-TryS preparations was not altered by removal of the tags. Steady-state kinetic analysis of Cf-TryS yielded a pattern that was compatible with a concerted substitution mechanism, wherein the enzyme forms a ternary complex with Mg2+-ATP and GSH to phosphorylate GSH and then ligates the glutathionyl residue to glutathionylspermidine. Limiting Km values for GSH, Mg2+-ATP, and glutathionylspermidine were 407, 222, and 480 μm, respectively, and the kcat was 8.7 s-1 for the TryS reaction. Mutating Arg-553 or Arg-613 to Lys, Leu, Gln, or Glu resulted in marked reduction or abrogation (R553E) of activity. Limited proteolysis with factor Xa or trypsin resulted in cleavage at Arg-556 that was accompanied by loss of activity. The presence of substrates, in particular of ATP and GSH alone or in combination, delayed proteolysis of wild-type Cf-TryS and Cf-TryS R553Q but not in Cf-TryS R613Q, which suggests dynamic interactions of remote domains in substrate binding and catalysis.

the most neglected diseases in terms of affordable, safe, and efficient therapy (www.accessmed-msf.org).
A common feature of trypanosomatids is their ability to transform glutathione into trypanothione (T(SH) 2 ), 1 which is the bisglutathionyl derivative of spermidine. T(SH) 2 has been shown to be of pivotal importance for the vitality and virulence of Trypanosoma brucei brucei and Leishmania donovani by knock-out or knock-down of trypanothione reductase, the enzyme regenerating T(SH) 2 from its oxidized cyclic disulfide form (1,2). More recently, the significance of T(SH) 2 in the defense against oxidative stress and for the maintenance of viability and proliferation in T. brucei was corroborated by knocking down trypanothione synthetase (TryS) by double strand RNA interference (3). The indispensability of T(SH) 2 is, in part at least, explained by its specific action on tryparedoxins (TXN), which are multipurpose oxidoreductases related to thioredoxin. T(SH) 2 -reduced TXN is the reducing substrate of a peroxiredoxin-type peroxidase (TXN-Px) in all trypanosomatids (4,5) and also of glutathione peroxidase homologs most recently discovered in Trypanosoma cruzi (6,7) and T. brucei (8,9). In trypanosomatids the T(SH) 2 /TXN/TXN-Px system is evidently the functional substitute for the GSH/glutathione peroxidase system (10) that is characteristic of the mammalian hosts. Accordingly, also knock-down of TXN and TXN-Px by double strand RNA interference leads to impaired viability and peroxide resistance of T. brucei (11). Although trypanosomatids are also equipped with a typical thioredoxin (12), TXN appears to substitute for thioredoxin in ribonucleotide reduction (13), which may explain the growth arrest observed upon depletion of T(SH) 2 (3) or trypanothione reductase (1).
Although the metabolic pathways using T(SH) 2 are being elucidated at considerable speed, the equally important biosynthesis of T(SH) 2 has remained a matter of debate and confusion since its discovery (14). The building blocks of trypanothione were soon clarified (15); it is formed from spermidine and two molecules of GSH with consumption of two ATP (Fig. 1). The source of spermidine, however, differs between species. In T. brucei, it is derived from endogenous ornithine, which is decarboxylated to putrescine and further processed to spermidine by aminopropyl transfer from S-adenosylmethionine. Accordingly, the ornithine decarboxylase inhibitor DL-␣-difluoromethylornithine proved to be efficacious in African trypanosomiasis (18) but not in the Latin American T. cruzi and the fish pathogen Trypanosoma granulosum, which lack ornithine decarboxylase and depend on polyamine uptake (19,20). Species differences * 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 nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AY603101.
ʈ To whom correspondence should be addressed: MOLISA GmbH, Universitä tsplatz 2, D-39106 Magdeburg, Germany. Tel.: 49-391-6718673; Fax: 49-391-6712223; E-mail: l.flohe@t-online.de. are also being discussed for the consecutive glutathionylations of spermidine. A TryS catalyzing both steps of T(SH) 2 synthesis was recently identified in T. cruzi (21), and the homologous enzyme of T. brucei brucei was also shown to synthesize T(SH) 2 from spermidine and GSH (22,23). Moreover, knock-down of T. brucei-TryS by means of RNAi verified the hypothesis that this enzyme is responsible for the entire synthesis of the spermidine-conjugated thiols, glutathionylspermidine (Gsp), and T(SH) 2 (3). Originally T(SH) 2 synthesis in C. fasciculata was also reported to be catalyzed by a single protein (16). Later, however, two distinct enzymes, a glutathionylspermidine synthetase (GspS) and a TryS, could be separated by Smith et al. (17). A GspS devoid of TryS activity was also purified from C. fasciculata by Koenig et al. (24) and produced by heterologous expression by Oza et al. (25). 2 These apparent discrepancies between T(SH) 2 biosynthesis of Crithidia and Trypanosoma species raised concerns whether C. fasciculata was an adequate model organism if the ultimate goal is the design of trypanocidal GspS or TryS inhibitors (21).
Here we report that the putative TryS gene of C. fasciculata indeed encodes a TryS that, like that of the Trypanosoma species, catalyzes both steps of trypanothione synthesis. More-over, the yields obtained by heterologous expression of this gene for the first time enabled an in-depth analysis of such prototype of trypanothione synthetase.

EXPERIMENTAL PROCEDURES
Isolation of the TryS-encoding Gene from C. fasciculata-Oligonucleotides (see Table I) were designed from the DNA sequence available on the NCBI data base for Cf-TryS (GenBank TM accession number AF006615). In order to isolate and amplify the corresponding gene, PCR (50 l) was prepared as follows: 5 units of ProofStart DNA Polymerase (Qiagen, Hilden, Germany), 1ϫ ProofStart PCR buffer (MgSO 4 at a final concentration of 1.5 mM), 300 M dNTPs, 0.5 g of chromosomal DNA from C. fasciculata (kindly provided by S. A. Guerrero, Universidad Nacional del Litoral, Argentina), and 3 M of each primer (Fo Cf-TryS and Re Cf-TryS). The PCR conditions were as follows: 5 min at 95°C, 35 cycles (95°C for 1 min, 60°C for 1 min, and 72°C for 4 min), and a final extension at 72°C for 10 min. PCR products were separated in a 1% agarose gel, purified with MinElute gel extraction kit (Qiagen), and cloned into pCR4-TOPO (TOPO TA cloning kit, Invitrogen). The plasmid thus obtained, pMCTS 1-7, harbors the Cf-TryS coding sequence. Both DNA strands of the insert were sequenced and analyzed by means of the program Vector NTI Suite 6.0 (InforMax Inc., Oxford, UK). The sequence was submitted to NCBI data base (GenBank TM accession number AY603101).
Construction of Expression Vectors-Plasmids of the pET series (15b, 19b and 43b; Novagen, Darmstadt, Germany) were used for heterologous expression. The Cf-TryS gene was engineered by PCR to add suitable restriction sites and tag regions (see Table I). The Cf-TryS gene was cloned into pET 15b and 19b, and NdeI (5Ј end) and BamHI (3Јend)  (16). It proposes the synthesis of trypanothione by a single ATP-dependent enzyme, TryS. This model did not consider the presence of amidase activity that re-converts the GSH spermidine conjugates back to GSH and spermidine nor a distinct GspS. B, scheme of Smith et al. (17). It proposes a stepwise biosynthesis of trypanothione catalyzed by two distinct enzymatic entities, Cf-GspS and Cf-TryS. It also considers the existence of a glutathionylspermidine amidase activity in Cf-GspS. C, novel scheme, like that of Smith et al. (17), considers the existence of Cf-GspS and Cf-TryS but differs in attributing to the Cf-TryS the ability to catalyze entire T(SH) 2 synthesis, as proposed by Henderson et al. (16). Moreover, it considers amidase activity of Cf-TryS and a distinct GspS activity that can synthesize Gsp irrespective of the T(SH) 2 requirement. sites were added. For cloning of Cf-TryS into pET 43b, an EcoRI site (5Ј end) and a His 6 tag followed by a HindIII site (3Ј end) were added. PCR were performed with pMCTS 1-7 as template, proper primer pairs, and TaqPCR Master Mix kit (Qiagen) according to the protocol of the supplier. The PCR program consisted of 5 min at 95°C, 30 cycles (95°C for 1 min, 52°C for 1 min and 72°C for 4 min), and a final extension at 72°C for 10 min. Products were separated on a 1% agarose gel and purified by the gel extraction kit (Qiagen). Using T4-DNA ligase (Roche Applied Science), the Cf-TryS gene was ligated to pET 15b, 19b, and 43b, generating pMC10c15, pMC12c1, and pMC8c5, respectively. For all cloning and subcloning steps, Escherichia coli strain DH5-␣ was used.
Heterologous Expression and Purification of Recombinant Cf-TryS-The plasmids pMC10c15, pMC12c1, and pMC8c5 were heterologously expressed in the E. coli strain Tuner-DE3 grown aerobically in Terrific Broth medium supplemented with 20 g⅐liter Ϫ1 glucose and 200 g⅐ml Ϫ1 ampicillin, at 180 rpm and 37°C. At A 600 nm of 1-1.5, cultures were chilled at 4°C for 15 min and expression of recombinant proteins was induced by addition of 0.5 mM isopropyl 1-thio-␤-D-galactopyranoside. Cultivation was resumed at 25°C and 100 rpm for an additional 16 h. For large scale production, 5 liters of Biostat B fermentor (Braun Biotech, Melsungen, Germany) was used, applying the conditions described above. Cells were harvested by centrifugation at 4000 ϫ g for 15 min at 4°C and resuspended in 50 mM sodium phosphate, 300 mM NaCl, pH 8 (buffer A), at a ratio of 1 g of wet weight pellet per 10 ml of buffer. Cell lysis was achieved by gently shaking with Lysozyme (30 mg % w/v, Fluka, Neu-Ulm, Germany) in the presence of a 1ϫ Complete EDTA-free Protease Inhibitor Mixture (cefpi mixture; Roche Applied Science) for 1 h at 4°C. To accomplish complete cell extraction, the suspension was subjected to two cycles of sonification on ice for 1 min with a Sonoplus Ultrasonic Homogenizer HD 2200 (Bandelin, Berlin, Germany). The lysate was centrifuged twice at 20,000 ϫ g for 20 min at 4°C to remove debris. The His-tagged proteins were purified by chromatography using XK-16/20 columns (Amersham Biosciences) packed with 10 or 25 ml of Ni-NTA Superflow resin (Qiagen). Supernatants were applied to the column pre-equilibrated in buffer A at a flow rate of 0.5 ml⅐min Ϫ1 , the columns were washed with 10 volumes of buffer A containing 10 mM imidazole and eluted with a gradient of 0 -100% of buffer A containing 0.5 M imidazole at a flow rate of 1 ml⅐min Ϫ1 . Ni-NTA fractions containing purified proteins were pooled, concentrated with a Vivaspin 20 PES 50,000 or 100,000 molecular weight cut-off concentrator (Vivascience, Hannover, Germany), dialyzed against 20 mM Tris, pH 7.2, by means of a Sephadex G-25 column (Amersham Biosciences), and concentrated again. In case of the NusA-tagged TryS (expressed from pMC8c5), an additional gel filtration chromatography was necessary (15 mg of protein from concentrated Ni-NTA fractions were applied to an Ultrogel AcA 54 (Chipergen, Biosepra Process Division) column of 2.5 ϫ 85 cm; 20 mM Tris-HCl, pH 7.2; flow rate 0.75 ml⅐min Ϫ1 ). Chromatography was performed by means of a BioLogic System (Bio-Rad). Protein concentrations were determined by the Bio-Rad Protein Assay with bovine serum albumin as standard. The yields of purified proteins from pMC8c5, pMC10c15, and pMC12c1 were 25, 65, and 160 mg per liter of culture medium, respectively. Proteins were sterile-filtered (0.22 m; Millipore) and stored at 4°C until use. Loss of activity was not observed over a period of 6 months.
Cleavage of N-terminal Tags with Enterokinase or Thrombin-Cleavage of N-terminal His or NusA tags was performed by means of thrombin or enterokinase cleavage capture kits (Novagen), respectively. Routinely, 1 unit of the corresponding proteases was used per 0.1 mg of protein at room temperature (RT) overnight. Proteases were removed according to the protocols of the supplier. Cleavage was confirmed by denaturing SDS-PAGE, matrix-assisted laser desorption and ionization time-of-flight (MALDI-TOF), and N-terminal Edman degradation.
Determination of Molecular Masses of Heterologously Expressed Cf-TryS-Recombinant proteins prepared as described above were separated in a 9% SDS-polyacrylamide gel. Gels were stained with ruthenium II tris-bathophenanthroline disulfonate and processed for tandem mass spectrometry analysis according to Rabilloud et al. (26). Molecular weights of recombinant Cf-TryS were estimated from mobilities in a 9% SDS-polyacrylamide gel stained with Coomassie Blue (Amersham Biosciences). Accurate molecular masses were obtained by MALDI-TOF mass spectrometry (Brucker GmbH, Bremen, Germany). The identity of the recombinant proteins was further assessed by N-terminal Edman degradation and MALDI-TOF analysis of tryptic digests. Nano-electrospray MALDI-TOF mass spectrometry was applied to rule out critical contaminants, if necessary.
Site-directed Mutagenesis of Recombinant Cf-TryS-Point mutations of Cf-TryS were generated to replace three different arginine residues (Arg-553, Arg-556, and Arg-613) with Lys, Leu, Gln, or Glu, respectively. For the mutants R613K, R613L, R613Q, R613E, and R553Q a first PCR (35 cycles at 95°C for 1 min, 50°C for 1 min and 72°C for 1 min) was performed with Re CfTS B/H and one of the following modified forward primers: Fo TryS R613K, Fo TryS R613L, Fo TryS R613Q, Fo TryS R613E, and Fo TryS R553Q, respectively (Table I). For the second PCR (identical cycling conditions), the fragments obtained from the first PCR were used as reverse megaprimers, and FoCfTryS-SalI was included as forward primer. To generate mutants R553K, R553L, and R553E and R556K, R556L, R556Q, and R556E, forward and reverse megaprimers were first synthesized by means of two separated PCR (same cycling conditions as above), one with FoCfTryS-SalI and Re TryS R553K, Re TryS R553L, Re TryS R553E, Re TryS R556K, Re TryS R556L, Re TryS R556Q, and Re TryS R556E; the other one using Re CfTS B/H and Fo TryS R553K, Fo TryS R553L, Fo TryS R553E, Fo TryS R556K, Fo TryS R556L, Fo TryS R556Q, and Fo TryS R556E, respectively (Table I). These products were used as primers and as templates for a new PCR (cycling conditions as above but at 55°C annealing temperature) that also contained primers FoCfTryS-SalI and Re CfTS B/H. pMC12c1 was always used as DNA template. All PCRs were performed with Proof Start DNA polymerase (Qiagen). PCR products were separated on a 1% agarose gel and purified using a MinElute Gel extraction kit (Qiagen). Final PCR products were digested with SalI and BamHI and, after renewed purification (QIAquick PCR purification kit from Qiagen), cloned into SalI-BamHI-digested pMC12c1. Ligations were performed with T4-DNA ligase (Roche Applied Science). E. coli Tuner-DE3 cells were transformed with the plasmids by electroporation (E. coli Pulser; Bio-Rad). Correctness of directed mutations was verified  by DNA sequencing of the plasmids. Heterologous expression and purification of mutants was performed as described above for the wildtype reCf-TryS. Substrate Specificity of Wild-type or Mutant reCf-TryS-Specificity was investigated by HPLC analysis of products (24). Enzymatic activities of recombinant proteins (1-10 g of protein for wild-type sequences or 0.05-1 mg of protein for mutant proteins) were incubated at 25°C in a volume of 0.9 ml containing 50 mM bis-Tris-propane, 50 mM Tris, pH 7.5, 5 mM Mg 2ϩ -ATP, 2-5 mM GSH, and 1-10 mM spermidine, 1-10 mM N 8 -acetylspermidine, or 2 mM Gsp, depending on the substrate specificity to be determined. The thiols were reacted with monobrobimane and methansulfonic acid at different time points and stored for 24 h at RT before application to a 92 Nucleosil 300 -5C8 HPLC column. Bimane derivatives of thiols were detected fluorometrically as described (24). GSH, Gsp, and T(SH) 2 were used for calibration (a differentiation between N 1 -and N 8 -glutathionylspermidine was not achieved). Amidase activity was analyzed accordingly by omitting ATP, MgSO 4 , and spermidine and adaptation of the concentrations of Gsp and T(SH) 2 .
Steady-state Kinetics Characterization of reCf-TryS-Kinetic studies of the enzymes were carried out by a continuous spectrophotometric assay monitoring NADH consumption at 340 nm . Therein, ATP hydrolysis is coupled to oxidation of NADH through pyruvate kinase and lactate dehydrogenase.
Reactions were performed at 25°C in 100 mM HEPES-K ϩ , pH 7.2, in a final volume of 0.9 ml. Each reaction contained 0.2 mM NADH, 1 mM phosphoenolpyruvate, 5 mM dithiothreitol, 0.5 mM EDTA, 10 mM MgSO 4 , 2 units⅐ml Ϫ1 pyruvate kinase (Sigma), 2 units⅐ml Ϫ1 lactate dehydrogenase (Sigma), 250 nM reCf-TryS, and varying amounts of ATP, GSH, and either spermidine or N 1 -glutathionylspermidine (Bachem AG, Weil am Rhein, Germany). For the GspS activity, the apparent Michaelis constants (K m ) and maximum initial velocities (V max ) values for each substrate were determined at five different substrate concentrations under fixed concentrations of the co-substrates involved (2.5 mM ATP, 1 mM GSH, or 10 mM spermidine). For TryS activity measurements, the concentrations of each substrate were varied between 0.1, 0.15, 0.3, and 5 mM for ATP; 0.18, 0.22, 0.29, 0.42, and 0.75 mM for GSH; and 0.15, 0.18, 0.22, 0.29, and 0.42 mM for N 1glutathionylspermidine. Each of the measurements at the respective combinations of substrates was performed in triplicate. All reactions were preincubated for 10 min at 25°C and started by the addition of spermidine or N 1 -glutathionylspermidine. Substrate consumption curves were collected with a UV-mc 2 SAFAS spectrophotometer (SAFAS, Monaco) and analyzed with the program Sigma-Plot 8.
The set of primary data was used to deduce K m and V max values and Dalziel coefficients (27) as follows. Apparent K m and V max values for substrate A were obtained from conventional double-reciprocal plots at five fixed concentrations of B or C, respectively ("primary plots," see Fig.  5, B and C). The apparent 1/V max values for [A] ϭ ϱ were then replotted against the reciprocal concentrations of B or C to yield a new set of apparent K m and V max values for infinite concentrations of two substrates ("secondary plots"). Replotting of the ordinate intercepts against the reciprocal third substrate concentration ("tertiary plot") yields the limiting kinetic constants (see Fig. 5D). If these plots are normalized for enzyme molarity ([E]/v instead of 1/v), the Dalziel coefficients can directly be read from slopes or intercepts, as described in detail by Dalziel (27).
Limited Proteolysis of Wild-type or Mutant reCf-TryS-Limited proteolysis by trypsin (0.22 units per 100 g of protein, Sigma) or factor Xa (2 units per 100 g of protein, Novagen) was carried out in 20 mM Tris buffer, pH 7.2, at RT. At the given times, proteolysis was stopped by adding an equal volume of 2ϫ SDS loading buffer and heating for 5 min at 95°C. Alternatively, proteolysis was terminated by addition of 1ϫ cefpi mixture (Roche Applied Science), if samples were to be analyzed for enzymatic activities. The extent of proteolysis was analyzed by SDS-PAGE. For control, samples were processed identically but without adding factor Xa or trypsin to rule out any effects of potential proteolytic contaminants of the preparations ("mock digestion"). The cleavage sites of trypsin and factor Xa were verified by N-terminal sequencing, MALDI-TOF, and tandem mass spectrometry of fragments. In substrate protection experiments, MgSO 4 , ATP, GSH, spermidine, and Gsp were added separately or in combination at concentrations of 5, 1, 5, 0.5, and 0.5 mM, respectively, and incubated for 60 min at RT.

RESULTS
Trypanothione Synthetase, an Enzyme with Multiple Activities-The C. fasciculata gene presumed to encode TryS was isolated and sequenced (GenBank TM accession number Y603101). Its DNA sequence displayed a 98% identity with the one previously reported for this gene (GenBank TM accession number AF006615; see Ref. 28). The predicted translation product conflicted with GenBank TM accession number O60993 (28) at positions 89, where Asn replaces Ser, and 404 -405, where a sequence inversion is found (DE versus ED; Fig. 2). As a rule, efforts to express the unmodified Cf-TryS gene yielded insoluble and inactive products that resisted re-activation, as reported previously by Tetaud et al. (28). We therefore tried to increase the solubility of the expression products by adding histidine tags of different lengths or producing TryS as a fusion protein with a cleavable Nus tag. Fig. 3 demonstrates the results obtained with three E. coli expression systems that yielded satisfactory amounts of TryS (lanes 2 and 4), a substantial proportion being found in the cytosol. The soluble fusion proteins could be purified to apparent homogeneity (Fig. 3, lanes 5) by means of affinity chromatography (see "Experimental Procedures"). All these preparations showed comparable specific TryS activities (Table II). However, they also proved to synthesize trypanothione from spermidine and GSH indicating that they had GspS as well as TryS activities. Removal of the tags did not significantly alter the activity and specificity (Table II).
As demonstrated in Fig. 4, the TryS activity is high enough to prevent accumulation of Gsp, when GSH is present in excess over spermidine (Fig. 4A). A considerable steady-state level of the intermediate Gsp is detected at low GSH and high spermidine concentration (Fig. 4B). The ability of the enzyme to catalyze the first step of the reaction was also tested with N 8 -acetylspermidine, which resulted in the formation of N 1glutathionyl-N 8 -acetylspermidine (not shown). The heterologously expressed Cf-TryS forms proved to have amidase activity as reported previously for GspS from E. coli (29) and C. fasciculata (25,28) and for TryS of T. cruzi (21) and T. brucei (23). The amidase activity of Cf-TryS was, however, less pronounced with T(SH) 2 than with Gsp as substrate (Table III). Furthermore, a marginal ATPase activity could be detected (see below).
Theoretically, the GspS activity of the Cf-TryS preparation could be due to a contamination with GspS from the production strain. In 1975, Tabor and Tabor had reported the isolation of a native GspS from E. coli (30), and its sequence was determined by Bollinger et al. (29). However, we were unable to detect any significant GspS activity in our wild-type production strain E. coli Tuner DE-3. In order to exclude any traces of E. coli GspS that might have been enriched our Cf-TryS preparations by co-purification, the samples were subjected to tryptic cleavage and checked for the presence of fragments indicative of E. coli GspS by nano-electrospray mass spectrometry at the highest sensitivity level. Again, no trace of E. coli GspS was detected. The GspS activity of our preparations thus is an intrinsic characteristic of Cf-TryS.
Steady-state Kinetics-Working out the kinetic pattern of an enzyme that displays four distinct but interdependent activities involving one or three substrates plus a metal ion (Fig. 1) is a special challenge. A selective measurement of the Gsp synthesis by TryS is hardly possible, because the product may be used as substrate for the consecutive reaction either instantly or with some delay as is evident from Fig. 4, A and B. Pertinent kinetic data are based on the overall ATP consumption that only in the very beginning can exclusively be attributed to the GspS reaction. Kinetic parameters thus obtained should therefore be considered as rough approximations. For the GspS activity, the approximate K m values were 1175 Ϯ 562, 52 Ϯ 17, and 7424 Ϯ 461 M for the substrates GSH, Mg 2ϩ -ATP, and spermidine, respectively (Table III). The k cat value for the reaction started with spermidine is 12.9 Ϯ 2.8 s Ϫ1 . It represents the overall k cat for both GspS and TryS activities (Table III). Subtracting the k cat value of the TryS reaction (8.7 s Ϫ1 ; see below) from the overall k cat yields a real k cat for the GspS activity of 4.2 s Ϫ1 . The k cat for the GspS reaction of Cf-TryS is close to those reported for the "real" GspS of C. fasciculata (24,25) and E. coli (29). Cf-TryS, however, mark- edly differs from typical GspS enzymes in having a more than 10 times higher K m for spermidine, which explains that only at high spermidine concentrations the GspS activity starts to compete with the TryS activity (Fig. 4B).
In respect to the TryS activity, the chances to collect unambiguous kinetic data are better. The results may still be compromised by the amidase and ATPase activities and the pronounced substrate inhibition observed at concentrations above 1 mM GSH, but these may be disregarded when ATP consumption is monitored in initial velocity measurements at concentrations below the K i for GSH. Nevertheless, measurements of initial velocities of T(SH) 2 synthesis from Gsp, GSH, and Mg 2ϩ -ATP proved to be complicated by a consistently observed lag phase (Fig. 5A). This delayed resumption of full activity could be shortened by preincubating the enzyme with GSH, but the type of curve shown in Fig. 5A was about the optimum that could be achieved. Initial velocities were therefore taken 250 s after starting the reaction, as indicated in Fig. 5A, and the pertinent substrate levels corrected accordingly. Fig. 5, B and C, shows examples of primary reciprocal plots that represent different patterns. If the reciprocal initial velocity is plotted against 1/[Gsp] at an excess of either Mg 2ϩ -ATP or GSH, parallel slopes are obtained for various fixed GSH or Mg 2ϩ -ATP concentrations, respectively, which is indicative of an enzyme substitution or "ping-pong" mechanism ( Fig. 5B). In contrast, when plotting 1/v against 1/[Mg 2ϩ -ATP], the slopes for distinct fixed GSH levels clearly converge (Fig. 5C), suggesting a central complex mechanism. A corresponding pattern is obtained for the primary plot at various fixed Mg 2ϩ -ATP concentrations and variable GSH concentrations (not shown). In the tertiary plot (Fig. 5D) the regression lines cut the ordinate at infinite concentrations of all substrates, i.e. at the real 1/V max , whereas the abscissa intercepts yield the limiting K m values for the respective variable substrate (Table III). According to the analysis by Dalziel (27) of the enzymatic threesubstrate reactions, the kinetic pattern of the TryS activity from Cf-TryS rules out a central complex mechanism that requires formation of a quaternary complex between the enzyme and all three substrates before the catalysis can proceed, because the term ATP/GSH/Gsp equals zero (Table IV). A "triple transfer mechanism" can equally be ruled out, because the term describing the formation of a central complex between enzyme, ATP, and GSH ( ATP/GSH ) differs from zero. Instead, the kinetic data conform to a "concerted substitution mechanism," which means that two of the substrates form a ternary complex with the enzyme, whereas the third one independently reacts with a modified ("substituted") enzyme to complete the catalytic cycle. The kinetic pattern by itself does not discriminate between the possible alternative sequences of central complex and ping-pong steps. The chemistry of the reaction, however, demands that GSH is first activated by ATP to become ligated to Gsp. Accordingly, the sequence of events may be schematized as in Fig. 6. The enzyme forms a ternary complex with ATP and GSH, wherein GSH becomes activated, likely in the form of a glutathionyl phosphate; upon release of ADP, the enzyme with glutathionyl phosphate bound kinetically behaves like a substituted enzyme that reacts with Gsp at a velocity that is no longer affected by the co-substrates ATP and GSH.
Although the amidase activity of GspS and TryS is considered unrelated to their synthetase activities (21,25), the ATPase activities may be considered as leakiness of the synthetic mechanism. A marginal ATPase activity (1.27 Ϯ 0.28% of the GspS activity) was detected when Cf-TryS was incubated in the pyruvate kinase-lactate dehydrogenase-coupled reaction mixture with ATP alone or in the presence of ATP and spermidine (Gsp could not be tested due to high amidase activity of Cf-TryS). 10 mM GSH doubled the formation of ADP, whereas higher GSH concentrations did not significantly alter the rate (data not shown). This observation points to an intrinsic  ATPase activity of the enzyme that is related to the need to transfer the ␥-phosphate. Its enhancement by GSH likely results from hydrolysis of glutathionyl phosphate (Fig. 6).
Relevant kinetic constants of Cf-TryS, as derived from our analysis, are compiled in Tables III and IV and are compared  with available historical data (Table III). The values for k cat (1/ 0 ) appear not to differ substantially between preparations and species. There is not even a relevant difference in k cat for the TryS and GspS reactions of the three TryS species, and in absolute terms these values compare well with those reported for GspS-type enzymes (6.9 Ϫ15.5 s Ϫ1 for Cf-GspS (24, 25) and 7 s Ϫ1 for E. coli GspS (29)). Most interestingly, however, the K m values of Cf-TryS for spermidine are significantly higher than those for Gsp and those of Cf-GspS for spermidine (59-470 M (24, 25)), whereas the TryS of trypanosomes, which seems to lack a GspS, display lower K m values for spermidine.
Substrate-modulated Proteolytic Cleavage Site-During attempts to cleave N-terminal tags from GspS-and TryS-type proteins, we discovered a noncanonical cleavage site for factor Xa near the C terminus of such proteins. In Cf-TryS it reads KPIVGRVGR NVT (where the underline indicates position 556), which resembles the canonical factor Xa site IEGR. It is conserved in known TryS species with the consensus sequence KPIVGRVG(R or S)NVT and is similar in GspS-type enzymes (see Fig. 2). The site is equally sensitive to tryptic cleavage.
Limited proteolysis with trypsin or factor Xa (Fig. 7) resulted in the cleavage of Cf-TryS with an apparent mass of 85 kDa into a fragment of 74 kDa (Fig. 7, B and C) and a small C-terminal fragment (not shown). Proteolysis correlated with a complete loss of GspS and TryS activities. N-terminal Edman degradation confirmed that the larger fragment corresponded to the N terminus of Cf-TryS sequence ( 1 MGHHHHHHHHH-HSSGHIDDDDKH 23 MASAERVPVS, normal and italic letters mark the tag and the first 10 N-terminal amino acids, respectively; construct pMC10 -15) whereas the smaller one started at Asn-557, as predicted. The molecular masses obtained by MALDI-TOF spectroscopy were 66,643.38 Ϯ 184.99 Da for the larger species (predicted 66,794.56 Da) and 10,473.08 Ϯ 5. 16 Da for the small fragment (predicted 10,786.37 Da). Analysis of tryptic digests from both species by MALDI-TOF (not shown) and site-directed mutagenesis (see below) further corroborated Arg-556 as the cleavage site for factor Xa and trypsin. Related peptide-forming ligases, e.g. D-Ala:D-Ala ligase, glutathione synthetase, and E. coli GspS, were reported to have a flexible "⍀-loop" that is prone to proteolysis when unliganded but protected from cleavage by binding of substrates or inhibitors (29,31,32). Analogous experiments with Cf-TryS revealed that all individual substrates alone or in combination inhibited cleavage of the enzyme, although to a different extent (Fig. 8A). As a rule, spermidine protected poorly, whereas the other substrates (ATP, GSH, and Gsp), in particular in combination, efficiently protected against proteolysis, the combination ATP/ GSH/Gsp being the most efficacious. It may be questioned if the minor effect of Gsp is attributable to the compound itself, because under the experimental conditions it is likely converted to GSH and spermidine due to the amidase activity of Cf-TryS. Therefore, in essence, the substrate protection assays suggest that the protease-sensitive region, upon binding of substrates, in particular of ATP and/or GSH, adopts a conformation that is less accessible to proteolytic attack.
Role of Arginines Located in the Putative ⍀-Loop and the Glycine-rich Motif of the Cf-TryS-Further elaborating on the analogy of Cf-TryS to other ligases, we investigated the arginine residues 553 and 556 of the presumed ⍀-loop for a possible role in binding the negatively charged substrates GSH and/or ATP. In addition, Arg-613 was considered, because it is located in a glycine-rich region (Gly-599 to Gly-621; see Fig. 2) such as an ATP-binding site. Each of the arginines was replaced by lysine, leucine, glutamine, and glutamate to explore the consequences of charge abrogation and inversion on activity (Table  V) and resistance to proteolysis (Fig. 8).
Mutation of Arg-553 dramatically decreased the activity, with the substitution by the negatively charged glutamate being the most detrimental. R553L and R553E were inherently more resistant to proteolysis, whereas R553K was degraded faster. In none of these cases did the presence of substrates had any influence on stability (not shown). The marginally active mutein R553Q instead displayed the same response to substrate exposure as the wild-type enzyme (Fig. 8, A and B), which suggested a similar substrate-induced conformational change despite the lack of turnover. Mutation of Arg-556 rendered the enzyme resistant against an attack by trypsin or factor Xa, which complies with the Arg-556 to Asn-557 bond being the presumed cleavage site. Most interestingly, mutations of this Arg residue by Lys or neutral residues hardly affected activity (R556Q or R556L), and a negative charge (R556E) at this position still left over substantial activity. An inverse picture was obtained when Arg-613 was mutated. None of the muteins had any significant residual activity, and all were readily cleaved at Arg-556. Substrate protection against proteolysis was undetectable in R613Q (Fig. 8D) and R613L (not shown), and only the combined presence of ATP, GSH, and Gsp marginally stabilized the muteins R613K and R613E (not shown). Taken together, the data comply with the assumption of substrate-induced conformational changes similar to those seen in other amide bond-forming ligases (29,(31)(32)(33)(34)(35)(36)(37)(38)(39)(40)(41). Whereas the free enzyme is readily cleaved by trypsin or factor Xa at Arg-556 in the presumed flexible ⍀-loop, substrate binding induces a compact form, in which the ⍀-loop is no longer attacked by proteases. Most interestingly, this substrate-induced compaction appears to be even more pronounced, if substrate turnover is abrogated as in the mutein R553Q (compare Fig. 8, A and B). Inversely, abrogation of a presumed ATP-  (27) is depicted to disclose the tentative physical meaning of the empirical coefficients (reciprocal rate constants of catalytic steps) that were deduced from double-reciprocal plots (Fig. 5, B-D), as outlined under "Experimental Procedures." The kinetic analysis was performed with reCf-TryS containing an N-terminal tag. binding site in the glycine-rich motif (R613X) prevents the substrate-induced conformational change, which suggests that somehow the substrates interact with both, the ⍀-loop and the glycine-rich motif, when inducing the compact enzyme conformation.

DISCUSSION
Enzymatic Redundancies in Trypanothione Synthesis-For the first time, the putative TryS gene of C. fasciculata (Gen-Bank TM accession number AY603101) could be functionally expressed, and thereby the assumption that it encodes an enzyme catalyzing the second step of T(SH) 2 synthesis, i.e. TryS, was verified. So far, our data seem to support the widely accepted view that, in Crithidia as opposed to T. brucei and T. cruzi, T(SH) 2 synthesis is achieved by two distinct enzymes. To our surprise, however, Cf-TryS proved to also catalyze the first step of T(SH) 2 synthesis. In order to make sure that this GspS activity of Cf-TryS was not due to artificial alteration of specificity, different tags were used for isolation, and the tags were removed. None of the enzyme modifications altered the specificity spectrum. A contaminating GspS activity derived from the expression host E. coli could equally be excluded. Therefore, it has to be concluded that Cf-TryS indeed has a dual specificity. In this respect, it resembles the homologous enzymes of T. brucei and T. cruzi. In the Trypanosoma species, however, GspS appears not to be present, and TryS is the only enzyme responsible for T(SH) 2 synthesis (3,21). Because Cf-TryS is also sufficient to catalyze the entire T(SH) 2 synthesis, the biological role of Cf-GspS remains enigmatic. The seemingly superfluous presence of this enzyme in Crithidia points to a distinct role of Gsp that can be synthesized irrespective of T(SH) 2 requirements. Alternatively, GspS may regulate polyamine levels, as has been proposed for the homologous GspS in E. coli (42), which cannot synthesize any T(SH) 2 . The comparatively low K m values for spermidine of GspS of C. fasciculata and E. coli indeed qualify these enzymes for both T(SH) 2independent Gsp synthesis and regulation of polyamine levels.
Mechanistic Aspects of TryS Catalysis-The dual activity of TryS raises mechanistic questions that are not easily answered. When the interim product Gsp has been formed, it may remain bound to the active site to become further processed to T(SH) 2 . In this case a second reaction center has to be postulated that binds and activates GSH by ATP for glutathionylation of the free amino group in Gsp. Alternatively, Gsp might flip within the active site or change its position to become further glutathionylated at the very same reaction center that had glutathionylated spermidine. Finally, a third possibility deserves consideration. The site that generates the activated GSH is flexible enough to ligate the glutathionyl residue to either site of spermidine or Gsp, which are bound essentially in the same way. The scarcity of homologous sequences and the lack of any structural data on TryS-related enzymes preclude discrimination between these alternatives. The substantial differences between the K m values for GSH and Mg 2ϩ -ATP, when determined with spermidine or Gsp as third substrate, argue in favor of two distinct reaction centers; however, the substrateinduced conformational changes, which become obvious from limited proteolysis, leave open alternative interpretations.
The TryS kinetics clearly reveal a concerted/substitution mechanism, whereby GSH and Mg 2ϩ -ATP form a ternary complex with the enzyme to activate GSH for glutathionylation of FIG. 6. Proposed mechanism of the trypanothione synthetase reaction. Steady-state kinetics reveal formation of a ternary complex between E, Mg 2ϩ -ATP, and GSH and a co-substrate-independent reaction of a substituted enzyme with Gsp. Additional circumstantial evidence (see text) suggests that first the ternary complex is formed, likely in an ordered way with ATP being the substrate binding first, which is followed by phosphorylation of GSH and ADP release. The enzyme-bound glutathionyl phosphate (GS-P), no longer in equilibrium with ATP and GSH and considered kinetically equivalent to a substituted enzyme, then reacts with Gsp to yield T(SH) 2 . Gsp (Fig. 6). By analogy to related ligases (e.g. D-alanine:Dalanine ligase, glutamine synthetase, and glutathione synthetase), the activation should consist in a phosphorylation of the glycine carboxyl group of GSH that enables glutathionylation of the terminal amino group(s) of Gsp (or spermidine). The enzyme loaded with glutathionyl phosphate would represent the substituted enzyme that exerts the glutathionylation of Gsp independently of the concentration of the co-substrates, as is reflected in the "ping-pong pattern" observed if the reciprocal Gsp concentration is plotted against inverse velocities at different fixed co-substrate concentrations (e.g. Fig. 5B). The central complex formation between enzyme, Mg 2ϩ -ATP, and GSH is likely ordered. Binding of Mg 2ϩ -ATP facilitates the binding of GSH, as is corroborated by decreasing apparent K m values for GSH with increasing ATP concentrations. In this respect, Cf-TryS resembles functionally related amide bond-forming ligases such as glutamine synthetases (43-45) and D-Ala:Dlactate ligases (32,46) or glutathione synthetases (34,(47)(48)(49) and ␥-glutamyl-cysteine synthetase (48).
The analogy of Cf-TryS catalysis to the processing of GSH by remotely related or unrelated enzymes disclosed some more mechanistic details. The carboxyl groups of GSH, Gsp, and T(SH) 2 are typically bound to arginines and less often to lysines (50), as has been discussed for glutathione peroxidases (51,52), glutathione reductase (53), tryparedoxin (54), and trypanothione reductase (55), for example. The very same residues are also implicated in binding and activating ATP.
After we had discovered, by unintended proteolytic cleavage, that the C-terminal domain is essential for Cf-TryS activity, we could indeed identify two arginine residues (Arg-553 and Arg-613) that were pivotal for activity, and as expected, a charge inversion at these positions (Arg versus Glu exchange) completely abrogated activity. It cannot be deduced with certainty if these residues bind to GSH, Gsp, or ATP, because the residual activities of the various muteins were too small to be kinetically analyzed, and the reactions specificities were not significantly changed (see Table V). The glycine-rich region that harbors Arg-613, however, reminds us of a typical ATPbinding motif, whereas Arg-553 might constitute the anchor for a ␥-glutamylcarboxyl of GSH or Gsp. In analogy to the mechanism of glutathione synthetase and D-alanine:D-alanine ligase, as analyzed by Fan et al. (36), these domains are presumed to be flexible and to interact with each other in the activation of GSH or Gsp, respectively. As had similarly been demonstrated before for glutathione synthetase (31) and E. coli GspS (29), GSH and ATP, particularly when present simultaneously, induced the compact enzyme conformation that is less easily proteolytically cleaved at Arg-556. These findings reveal that Cf-TryS, like other amide bond-forming ligases, forms its catalytic complexes by substrate-induced movement of remote loops. The inherent flexibility of the enzyme structure might also explains its dual activities; the complex catalytic entity might shift its glutathionylphosphate into different positions to acylate N 1 or N 8 of bound spermidine or Gsp, respectively. Therefore, neither a second reaction center nor a re-orientation of initially formed Gsp would be required to achieve the twostep synthesis of T(SH) 2 from spermidine and GSH.
Evidently the conformational shifts cannot be directly related to formation of the enzyme-substrate complexes, as de-   2). b Values in parentheses are relative specific activities referred to the control (wild-type reCf-TryS) set to 100%. c Values with a significance level of p Ͻ 0.05. d Values with a significance level of p Ͻ 0.01, two-tailed Student's t test between mutant (n ϭ 2) and wild type Cf-TryS (n ϭ 2). duced from steady-state kinetics, because each of the substrates by itself is able to induce at least some protection against proteolysis. The additive effects of ATP, GSH, and Gsp on proteolytic stability reveal that these three substrates can bind simultaneously to the enzyme. This finding may be suspected to conflict with the proposed concerted/substitution mechanism that only considers binary and ternary but no quaternary enzyme-substrate complexes. The kinetic analysis, however, only detects enzyme-substrate complexes that affect turnover, whereas a kinetically silent substrate binding may nevertheless induce conformational changes. Indeed, full comprehension of the complex pattern of synergistic and antagonistic effects of substrates on resistance against proteolysis has to await structure elucidation of TryS-type enzymes in different functional states.
Is C. fasciculata Still a Model for Pathogenic Trypanosomatids?-More recently, when the differences in T(SH) 2 synthesis between C. fasciculata and T. brucei and T. cruzi were noticed, the use of C. fasciculata enzymes as targets for drug design was generally questioned (21). The similarity of TryS from Crithidia and Trypanosoma species, however, is sufficiently high to expect results that are largely species-independent. The expression systems for Cf-TryS presented here for the first time allow the preparation of a TryS in reasonable quantities and thus offer a unique chance to analyze further this type of enzymes. Also, the scenario of polyamine glutathionylation of Crithidia may resemble that of the second clinically important genus of the family, Leishmania. Pertinent knowledge on the T(SH) 2 synthesis of these species is lagging behind. Genes presumed to encode the Leishmanial TryS have recently been isolated (Leishmania major, GenBank TM accession number AJ311570, and L. donovani, GenBank TM accession number AJ430873) but not yet functionally characterized. Also, a fragment (GenBank TM accession number AL133443) of a related gene was detected in L. major and suggested to encode a GspS (23), and this may indicate that the genus Leishmania is closer to Crithidia than to Trypanosoma in T(SH) 2 biochemistry, as it is in general. Therefore, it appears premature to abandon the most convenient model organism in the struggle for new therapeutic principles for diseases caused by Kinetoplastida.