Cooperativity and pseudo-cooperativity in the glutathione S-transferase from Plasmodium falciparum.

Binding and catalytic properties of glutathione S-transferase from Plasmodium falciparum (PfGST) have been studied by means of fluorescence, steady state and pre-steady state kinetic experiments, and docking simulations. This enzyme displays a peculiar reversible low-high affinity transition, never observed in other GSTs, which involves the G-site and shifts the apparent K(D) for glutathione (GSH) from 200 to 0.18 mM. The transition toward the high affinity conformation is triggered by the simultaneous binding of two GSH molecules to the dimeric enzyme, and it is manifested as an uncorrected homotropic behavior, termed "pseudo-cooperativity." The high affinity enzyme is able to activate GSH, lowering its pK(a) value from 9.0 to 7.0, a behavior similar to that found in all known GSTs. Using 7-chloro-4-nitrobenzo-2-oxa-1,3-diazole, this enzyme reveals a potential optimized mechanism for the GSH conjugation but a low catalytic efficiency mainly due to a very low affinity for this co-substrate. Conversely, PfGST efficiently binds one molecule of hemin/monomer. The binding is highly cooperative (n(H) = 1.8) and occurs only when GSH is bound to the enzyme. The thiolate of GSH plays a crucial role in the intersubunit communication because no cooperativity is observed when S-methylglutathione replaces GSH. Docking simulations suggest that hemin binds to a pocket leaning into both the G-site and the H-site. The iron is coordinated by the amidic nitrogen of Asn-115, and the two carboxylate groups are in electrostatic interaction with the epsilon-amino group of Lys-15. Kinetic and structural data suggest that PfGST evolved by optimizing its binding property with the parasitotoxic hemin rather than its catalytic efficiency toward toxic electrophilic compounds.

Cytosolic glutathione S-transferases (GSTs) 1 are a superfamily of multifunctional enzymes involved in the cellular de-toxification of many endobiotic and xenobiotic compounds (1,2). They are dimeric proteins grouped into at least 10 geneindependent classes, named Alpha, Beta, Delta, Kappa, Pi, Mu, Theta, Zeta, Sigma, and Omega, on the basis of different amino acid sequences and substrate specificities (3)(4)(5)(6)(7)(8)(9)(10)(11). A prominent catalytic activity performed by eucaryotic GSTs (in particular by the Alpha, Mu, Pi, Theta, and Sigma GSTs) is the conjugation of glutathione (GSH) to a number of toxic electrophilic compounds, thus promoting their excretion from the cell (1,2). Besides catalyzing conjugation activity, GSTs are involved in the sequestering and transport of exogenous hydrophobic compounds (i.e. pesticides, herbicides, and antibiotics) and of a large variety of endogenous compounds (i.e. steroids, bilirubin, and ferriprotoporphyrin IX). In contrast to many other organisms, the malarial parasite Plasmodium falciparum possesses only one GST isoenzyme (PfGST). A gene encoding a protein with sequence identity of up to 33% with known GSTs was identified on chromosome 14 of P. falciparum. The three-dimensional structure of PfGST shares several structural features with the Mu class GSTs even though alignments with its members display low sequence identities (20 -33%) (12). Despite this structural similarity with Mu class GSTs, PfGST shows very low or no affinity toward the classical GST substrates. The homodimeric enzyme catalyzes the reaction GSHdependent toward 1-chloro-2,4-dinitrobenzene, ethacrynic acid, o-nitrophenyl acetate and cumene hydroperoxide but shows very low catalytic efficiency in comparison with other GSTs (in particular Pi, Alpha, and Mu class GSTs) (13,14). It has been suggested that PfGST, which represents 1-10% of cellular proteins, may be involved directly in the heme detoxification of the parasite, avoiding the incorporation of the heme molecule into hemozoin, a highly insoluble material (15). In fact, it has been shown that the parasitotoxic hemin inhibits PfGST uncompetitively (K i ϭ 6.5 M) indicating that free hemin can be bound by the enzyme in the micromolar range (13). Furthermore, it is known that chloroquine, one of the most important antimalarial drugs, inhibits hemin catabolism, leading to intracellular hemin accumulation (16). In chloroquine-resistant parasites, PfGST activity significantly increases compared with sensitive strains (17,18). PfGST inhibitors are thus expected to act synergistically with the antimalarial drug chloroquine (which itself is a very weak inhibitor for GST, showing IC 50 Ͼ 200 M), increasing the accumulation of hemin in the parasite and thus enhancing its antiparasitic effects (13).
Starting from these preliminary observations, we have investigated the binding and activation of GSH and the kinetic mechanism of the reaction between GSH and the alternative co-substrate, 7-chloro-4-nitrobenzo-2-oxa-1,3-diazole (NBD-Cl). Moreover, the mechanism of interaction of the hemin with PfGST has been elaborated carefully on using pre-steady state, steady state kinetic, and fluorescence experiments. Our results reveal the existence of two distinct cooperative phenomena that modulate the interaction of this enzyme with GSH and with hemin. Structural intersubunit communication in GST superfamily has been demonstrated previously for GSTP1-1, GSTA1-1, and GSTT2-2 (19 -25). Our results confirm that cooperativity is a mechanism adopted by a number of GST isoenzymes. Because of this cooperative mechanism, GSTs might be able to modulate interactions with different compounds, thereby responding to specific evolutionary pressures.
Enzyme-The recombinant PfGST was expressed in Escherichia coli and purified as described previously (13). The purified form of PfGST was stored at Ϫ20 or 5°C in the presence of 10 mM GSH. Under these conditions, the enzyme was stable for a few weeks. After a Sephadex G-25 chromatography and subsequent incubation of the enzyme at 25°C for 12 h, the 90% inactive enzyme (PfGST i ) was used for reactivation and stopped-flow binding experiments. Protein concentration was calculated assuming an ⑀ 1 mg/ml of 1.1 at 280 nm for PfGST on the basis of the amino acid sequence (26). A molecular mass of 25 kDa/GST subunit was used in the calculation (14).
Spectroscopic Detection of GSH Thiolate-Difference spectra of the GSH thiolate bound to PfGST were obtained with a Kontron double beam Uvikon 940 spectrophotometer thermostatted at 25°C. In a typical experiment, 1 mM GSH was added to the enzyme (15 M) in a phosphate-acetate-borate (50:50:50 mM) buffer, pH 7.0, and the spectra were recorded after 60 min. Spectra were corrected for the contribution of free GSH and free enzyme. The amount of thiolate was calculated by assuming an ⑀ 240 nm of 5000 M Ϫ1 cm Ϫ1 . Experiments at different pH values were carried out using the phosphate-acetate-borate (50:50:50 mM) buffer between pH 5.0 and 7.5. The pK a value of the bound GSH was calculated by fitting the spectral data to Equation 1, where A lim is the limiting thiolate absorbance at 240 nm and at alkaline pH values. Enzymatic Activity-Standard PfGST activity with CDNB as cosubstrate was measured at 25°C in a 0.1 M potassium phosphate buffer, pH 6.5, containing 1 mM GSH and 1 mM CDNB. The activity was assayed spectrophotometrically by following the enzymatic product at 340 nm (⑀ ϭ 9600 M Ϫ1 cm Ϫ1 ). Steady state kinetic experiments with NBD-Cl as co-substrate were performed in 0.1 M sodium acetate buffer, pH 5.0 (25°C), by varying NBD-Cl from 0.2 to 2 mM and GSH from 0.1 to 1 mM over a matrix of 16 substrate concentrations. The enzymatic product absorbs at 419 nm (⑀ ϭ 14,500 M Ϫ1 cm Ϫ1 ).
Fluoride/Chloride Leaving Group Substitution-Kinetic data were obtained at 25°C and at fixed NBD-Cl or NBD-F (0.2 mM) and GSH (1 mM) concentrations in 1 ml (final volume) of 0.1 M sodium acetate buffer, pH 5.0, containing suitable amounts of PfGST. The spontaneous reactions were evaluated under the same experimental condition in the absence of the enzyme.
Stopped-flow Experiments-Rapid kinetic experiments were performed on an Applied Photophysics Kinetics spectrometer stopped-flow instrument equipped with a thermostatted 2-mm light path observation chamber. Binding of GSH to PfGST i was studied at 25°C by rapid mixing of enzyme (30 M in 0.1 M potassium phosphate buffer, pH 6.5) with the same volume of GSH (from 2 to 100 mM) in 0.1 M potassium phosphate buffer, pH 6.5. Binding of GSH was monitored by following the decrease of intrinsic fluorescence of the protein at 333 nm ( ex ϭ 280 nm) and the increase of the absorbance at 240 nm due to the ionization of the bound GSH. Likewise, binding of hemin to PfGST was studied under the same experimental conditions by rapid mixing of 20 M enzyme and 20 mM GSH in 0.1 M potassium phosphate buffer, pH 6.5, with the same volume of a hemin solution (from 10 to 100 M). Binding of hemin was monitored by following the decrease of the intrinsic fluorescence of the protein at 333 nm.
A fresh stock hemin solution was prepared by dissolving hemin chloride in 100 mM NaOH and then centrifuging for 10 min at 15,000 rpm to remove the remaining hemin crystals. Heme concentrations were estimated from the absorbance using ⑀ 385 nm ϭ 58,400 M Ϫ1 cm Ϫ1. This stock hemin solution was diluted to 0.1 mM hemin and 10 mM NaOH and used immediately. No solubility artifacts were observed in control spectrophotometric experiments carried out under experimental conditions similar to those used for the stopped-flow technique or for the fluorometric experiments described below.
Isothermic Binding of GSH and Hemin-Isothermic binding of hemin to PfGST was measured in a single photon counting spectrofluorometer (Fluoromax, S.A. Instrument, Paris) with a sampler holder thermostatted at 25°C. Excitation was at 295 nm, and emission was collected at 333 nm. The affinity of PfGST (2 M) for hemin was determined by measuring the perturbation of the intrinsic fluorescence of the protein following the addition of the inhibitor (from 0.5 to 5 M) in the presence of 10 mM GSH in 0.1 M potassium phosphate buffer, pH 6.5. Fluorescence data were corrected both for dilution and inner filter effects. Similarly, the isothermic binding of GSH to PfGST was studied by incubating PfGST i (2 M) with variable amounts of GSH (from 0.1 to 10 mM) in 0.1 M potassium phosphate buffer, pH 6.5. Fluorescence quenching at 333 nm was measured after 60 min at 25°C.
Inhibition of PfGST by Hemin-Inhibition of PfGST by hemin was studied by adding variable amounts of hemin (from 0.5 to 6 M) with a fixed enzyme concentration (2 M) in 1 ml of 0.1 M potassium phosphate buffer, pH 6.5. After about 10 s, 5 mM GSH and 1 mM CDNB were added for activity measurement. The data obtained by varying the hemin concentration were analyzed to the Hill equation, where v i is the initial velocity observed at a given concentration of hemin, v max is the velocity observed in absence of inhibitor, [S] is the hemin concentration, and n is the Hill coefficient at the hemin concentration corresponding to the half-enzyme saturation. Inhibition data were also analyzed with a two-site Adair model for ligand binding to a homodimeric macromolecule, where K 1 is the dissociation equilibrium constant for hemin binding to the free enzyme, ␣K 1 (ϭK 2 ) represents the dissociation equilibrium constant for hemin to the monoligated enzyme, and ␣ is the adimensional interaction parameter coupling the two functionally linked hemin binding sites. Docking Simulations-Docking simulations were performed with the program Autodock 3.0.5 (27) using the PfGST crystal structure in complex with S-hexylglutathione (Protein Data Bank code 1Q4J), after removal of the hexyl moiety. Polar hydrogen atoms were added geometrically, and Kollman united atom charges were assigned. Docking experiments for hemin were performed with a flexible ligand. Hemin charges were added by the Gasteiger-Marsili method (28). Grids of molecular interactions were calculated in a cubic box, size 33.375 Å, grid spacing 0.375 Å, centered on the oxygen atom of the tyrosine side chain participating in GSH activation (Tyr-9). Docking was performed 100 times using the Lamarkian genetic algorithm with random starting position and conformation, a population size of 50, standard parameters (29), and a maximum of 250,000 energy evaluations and 27,000 generations. The 100 final docked conformations were ranked according to their binding free energy and clustered using a tolerance of 3 Å rootmean-square deviation. The MOLMOL program (30) was used for graphical interpretation and representation of results.

RESULTS AND DISCUSSION
Low-High Affinity Transition-PfGST shows good stability and constant specific activity only when it is stored at 1 mM or higher GSH concentrations. On the contrary, a rapid and biphasic inactivation occurs in GSH-depleted solutions. For example, the specific activity of the enzyme stored in the presence of 10 mM GSH drops to 50% in a few minutes after a Sephadex G-25 chromatography (t1 ⁄2 ϭ 3-4 min), whereas a complete inactivation is reached only after about 48 h of incubation at 25°C (t1 ⁄2 ϭ 12 h). In the presence of traces of GSH, both of the phases become slower, and only the first phase of inactivation can be readily observed ( Fig. 1). At 0.1 mM GSH concentration, the 50% inactivation kinetics displays a t1 ⁄2 of 12 min. The hemi-inactivated enzyme completely recovers the original activity in less than 1 min upon incubation with 10 mM GSH. A complete reactivation is also observed starting from a 90% inactivated enzyme (PfGST i ) using either 1 mM GSH (Fig. 1) or 10 mM GSH. In both cases the reactivation is a monophasic process, but a nonlinear dependence of the reactivation rate on GSH concentration may reveal a cooperative phenomenon (Fig.  1, inset). When the enzyme completely looses its original activity, GSH is unable to trigger an efficient reactivation, indicating that a further irreversible process takes place. Interestingly, S-methylglutathione is also able to trigger the activation of PfG-ST i , and thus the free sulfhydryl group of GSH is not essential for this event. Apart from the possibility that the low-high activity transition may be caused by a structural modification of essential residues involved in catalysis, alternatively, this could be due to structural changes of the G-site and/or of the H-site lowering the affinity for GSH or for the co-substrate. Isothermic binding of GSH to PfGST i , followed on the basis of the intrinsic fluorescence quenching at variable GSH concentrations, indicated that changes of the enzyme affinity for GSH are likely involved in this transition. In fact, a marked sigmoid behavior suggests that GSH binding shifts the enzyme from a low affinity to a high affinity conformation (data not shown). However the fluorescence data did not fit satisfactorily to a cooperative equation, and only an approximate estimation of the affinity constants for GSH has been calculated, i.e. K D Ͼ 200 mM for the low affinity conformation and K D Ͻ 0.2 mM for the high affinity conformation. •, active PfGST (100 M in 5 mM GSH) was diluted up to 2 M enzyme and 0.1 mM GSH concentration in 0.1 M potassium phosphate buffer, pH 6.5, at 25°C. At fixed time points, the residual activity was assayed at 340 nm after the addition of 1 mM GSH and 1 mM CDNB. Activity was calculated within 20 s of the addition of reagents. The continuous line is the best fit to a monoexponential decay, which gives a half-time of inactivation of 19 min. E, PfGST i (45 M) was incubated at 25°C with 1 mM GSH in 0.1 M potassium phosphate buffer, pH 6.5. At different times, aliquots were diluted in the assay mixture containing 1 mM GSH and 1 mM CDNB for activity determinations. The continuous line is the best fit to a monoexponential process, which gives a half-time of reactivation of 16 min. Inset, dependence of the pseudo-first order kinetic constants for PfGST activation on GSH concentration. S.A., specific activity.
Stopped-flow Experiments-The low-high affinity transition triggered by GSH and the binding mechanism of GSH have been detailed by means of stopped-flow experiments following the quenching of the intrinsic fluorescence after rapid mixing PfGST i (30 M in 0.1 M phosphate buffer, pH 6.5) with GSH at variable concentrations (from 100 to 2 mM) (Fig. 2). Interestingly, k obs values do not increase linearly by increasing GSH concentrations, as expected for a simple bimolecular interaction, but follow a clear exponential behavior. It should be noted that reaction half-times span from ϳ200 ms to Ͼ500 s by changing the final GSH concentrations from 50 to 1 mM, respectively. However, all experimental traces do not fit satisfactory to a classical positive cooperative mechanism (Fig. 2, inset). Conversely, k obs values increase linearly on the square of GSH concentration (Fig. 3), suggesting the occurrence of a trimolecular interaction. All experimental data obtained from the stopped-flow experiments, taken together with kinetics of enzyme activation and inactivation observed in the presence of GSH or after its removal, have been rationalized on the basis of a more complicated binding mechanism (Scheme 1). GSH binds to the two PfGST subunits of the inactive enzyme (I-I) sequentially, with identical rate constants. Only when both subunits are saturated with GSH does the inactive enzyme rapidly convert into the active dimeric enzyme (A-A). The conversion of the inactive enzyme with GSH bound into the active enzyme apparently is not reversible. The active enzyme becomes inactive only by following Step 4 (in Scheme 1) to give the GSH-A-I species (which may correspond to the hemi-inactivated enzyme recovered after Sephadex G-25 chromatography). Stopped-flow data fit very well to this mechanism (Fig. 2), and the fitting procedure allowed us to estimate with satisfying precision k 1 , k Ϫ1 , k 2 , and k Ϫ2 values ( Table I). The rate constants for the conversion of the inactive enzyme to the active enzyme (k 3 and k Ϫ3 , as well as k Ϫ5 ) were not resolved by the fitting procedure. In these cases, the fitting procedure can only indicate the upper or lower limits of the rate constants. Rate constants for dissociation of GSH from the active enzyme (k Ϫ4 and k Ϫ5 ) were found experimentally by looking at the rate of the biphasic inactivation due to dilution or removal of GSH from an activate enzyme solution (see Fig. 1). The hysteresis represented by the cyclic reaction in Scheme 1 is due to the very low rate constant for the conversion of the active enzyme into the inactive enzyme in the presence of bound GSH, which causes the apparent irreversibility of reaction 3. Interestingly, the low activity enzyme shows a kinetics of GSH binding (k 1 and k 2 ) that is thousands of times lower than those observed for Alpha, Pi, and Mu class GSTs (31); conversely, the kinetics of GSH release (k Ϫ1 and k Ϫ2 ) is similar. Thus, it appears that in the low affinity conformation, the G-site is partially obscured so that GSH cannot interact at a high rate with its protein counterpart. The overall dissociation constant calculated by the stopped-flow data ranges from 200 mM for the low affinity conformation to 0.18 mM for the high affinity state, close to the values calculated by the preliminary analysis of the fluorescence data at equilibrium.
GSH Activation-Differential spectrophotometry has been used to verify whether PfGST forces the deprotonation of the bound GSH in the same manner that it occurs in most GSTs (2). These experiments, based on the peculiar absorbance of the thiolate anion at 240 nm, were carried out at different pH values and at fixed 1 mM GSH, a concentration that stabilizes the high affinity conformation. As shown in Fig. 4, the high affinity enzyme lowers the pK a of GSH so that detectable amounts of GS Ϫ are found even at neutral pH values. The pH dependence of the absorbance at 240 nm indicates that about 1 mol of thiolate/mol of enzyme subunit is formed at pH 8.5. The apparent pK a of the bound GSH is 7.0, a value 2 units lower than that of the free GSH. Stopped-flow experiments performed by mixing the low affinity enzyme with variable amounts of GSH also indicated that kinetics of thiolate formation are similar to that of the binding process (see Fig. 3); thus, these two events appear synchronous.
Kinetic Mechanism with NBD-Cl as Co-substrate-One peculiar feature of PfGST is its very low specific activity for reactions involving classical GST substrates such as CDNB. In SCHEME 1

TABLE I Kinetic parameters for GSH binding to PfGST
Microscopic rate constants were calculated by fitting stopped-flow data to Scheme 1.
a Upper or lower limits as indicated by the fitting procedure. b Rate constants were found experimentally by following the biphasic inactivation due to dilution or removal of GSH from the enzyme solution. the high affinity conformation, its specific activity is only 1 unit/mg, which is 50 -100 times lower than that found for Alpha, Pi, and Mu class GSTs. As suggested previously (13,14), this low catalytic efficiency could be mainly because of the scarce affinity of the enzyme for CDNB. Our steady state kinetic experiments confirmed that the K m for CDNB is very high (Ͼ Ͼ10 mM), and an accurate determination of kinetic parameters such as K m and k cat was precluded because of the scarce solubility of this compound. A more detailed kinetic analysis was obtained using the alternative co-substrate NBD-Cl. This compound displays strong affinity for Alpha, Pi, and Mu class isoenzymes (K m ϭ 5-20 M) (32) but a 1000-fold lower affinity for PfGST (K m ϭ 2.5 mM) according to the scarce propensity of PfGST to efficiently bind hydrophobic compounds. However, this co-substrate may be used at saturating levels for PfGST, allowing us to evaluate the catalytic competence of this enzyme. On the basis of the reciprocal plots reported in Fig. 5, a turnover number of 15 s Ϫ1 has been calculated, which indicates a good catalytic efficiency under saturating GSH and NBD-Cl concentrations. From the double reciprocal plots varying GSH and NBD-Cl, the kinetic mechanism appears to be ordered, with GSH binding first. The chemical step is probably the rate-limiting event in catalysis. In fact, the substitution of chloride with the better leaving group fluoride in this co-substrate causes a 20-fold increase in the turnover number. This acceleration is similar to the rate increase observed for the spontaneous reaction (18-fold) that is likely rate-limited by a chemical step.
Binding of Hemin-It has been shown that PfGST binds hemin with high affinity, contributing to the protection of the parasite against this toxic compound (13-15). However, as the stoichiometric, mechanical, and structural details of this inter-  Table II. A.U., arbitrary units. SCHEME 2 action have not been clarified, we obtained further insights using a multidisciplinary approach. Binding of hemin has been studied by means of inhibition and fluorescence experiments at equilibrium and also by a stopped-flow approach following the kinetics of intrinsic fluorescence quenching due to hemin binding. As suggested previously (13,14), we also observed that no binding occurs in the absence of GSH; thus the high affinity conformation and GSH bound to the G-site are crucial requirements for this interaction. In the presence of 10 mM GSH, inhibition data indicate that binding is not hyperbolic but follows a sigmoid behavior (Fig. 6). Experimental data fit very well to a positive cooperative mechanism with a Hill coefficient of 1.8. By fitting all inhibition data to the Adair equation, we also obtained an estimation of the dissociation constants for the hemin-enzyme complex in the low affinity conformation (K D1 ϭ 2.8 M) and in the high affinity conformation (K D2 ϭ 0.16 M). Interestingly, replacement of GSH by 10 mM S-methylglutathione changes the binding mechanism. In particular, the process is now hyperbolic and uncooperative with a K D ϭ 2.2 M, close to the value found for the low affinity conformation (data not shown). Thus, the thiolate group of GSH is not involved in a crucial coordination of the iron ion of the ligand but may play an important role in the intersubunit communication triggered by hemin.
Fluorescence data at equilibrium detailed the stoichiometry of this interaction that involves one hemin molecule for one enzyme monomer (Fig. 7). We emphasize that hemin binding causes a dramatic quenching of the intrinsic fluorescence of this protein.
In fact, when two hemin molecules are bound to the dimeric enzyme, the fluorescence intensity drops to 3% of its initial value (Fig. 7, inset). Trp-131 is the sole Trp residue present in the structure of PfGST. The strong energy interaction between this group and the hemin molecule indicates that these groups are close enough to be included in the Förster radius, as will be also confirmed by docking simulations (see below).
Stopped-flow experiments detailed the kinetics and possible mechanism of this interaction. By mixing variable amounts of hemin (from 100 to 10 M) to PfGST in the presence of 10 mM GSH, at least two kinetically distinct phases were observed. The first one, accounting for no more than 10% of the total fluorescence quenching, represents a fast mixing artifact that is observed on mixing PfGST with buffer either in the presence or absence of hemin. It was completed in about 50 ms and excluded from the fitting procedure. The second slower and more prominent perturbation is likely because of the formation of the PfGST-hemin complex. The experimental traces were fitted well to the cooperative Scheme 2 (see Fig. 8). The microscopic rate constants given by the fitting procedure are shown in Table II. Interestingly, the affinities of the low affinity and high affinity binding sites (K D1 and K D2 ) calculated by this kinetic approach are close to the corresponding values coming from the inhibition data. Moreover, the origin of the observed cooperativity must be identified in a 10-fold decrease of the k Ϫ2 value for hemin in the high affinity conformation. In other words, in the high affinity conformation, the hemin is more tightly bound to its protein counterpart, whereas the accessibility for the external hemin does not change appreciably in the two conformations.
Docking Simulations-To localize the protein portion and critical residues involved in the hemin binding, docking simulations were performed with the program Autodock (27). The lowest energy model is reported in Fig. 9. The hemin molecule is located in a cleft between the C-terminal region of helix 4 and the N-terminal part of helix 1 and the C-terminal loop. Fortytwo of the 50 lowest energy models (of 100 different docking trials) bind to the same site, demonstrating the reliability of the proposed structure. The Autodock program assigns a binding free energy of Ϫ12 kcal/mol to the best model. This extremely high affinity is the result of multiple interactions. The protein surface displays a striking complementarity to the hemin molecule, as shown in Fig. 9A. Electrostatic interactions further stabilize the docked molecule; as shown in Fig. 9B, specific interactions are present between the side chain of Lys-15 and the carboxylate groups of the hemin molecule and between Asn-112 and the iron atom. Finally, hydrophobic patches are present on the protein surface in correspondence to the aromatic rings of the porphyrin moiety. On the other hand, a direct interaction between the glutathione thiolate and the hemin iron atom is not present, these two atoms being at ϳ8 Å in the docked structure. It is worth mentioning that the proposed model is compatible with the presence of one hemin molecule/subunit in agreement with the observed 1:1 binding stoichiometry, because no direct interactions are present in the dimeric structure when two ligands are placed in identical positions on each monomer (Fig. 9C).
The docked model provides a rationale for the almost complete quenching of intrinsic fluorescence caused by hemin binding. The single Trp residue of PfGST is located at ϳ16 Å from both hemin molecules docked to the dimeric PfGST. Hemin is an extremely efficient energy acceptor for tryptophan fluorescence (33); the Förster distance calculated for the spectral features of these two chromophores is 34 Å. Because this interaction radius is much higher than the distance observed in the docked structure, our model is compatible with an almost complete quenching of Trp fluorescence (34).
Conclusions-The existence of intersubunit structural communication in GSTs has been a debated question in the past, and for about 30 years GSTs have been thought to be uncooperative enzymes. A first indication for a cooperative interaction in GSTs has been obtained for GSTP1-1, an isoenzyme that shows positive homotropic behavior by GSH binding after point mutation of crucial residues such as Cys-47 and Lys-54 (19,20). Later, it was discovered that even the native enzyme triggers both positive and negative homotropic behavior by GSH binding at high or low temperatures (22). This phenomenon appears to be a particular case of a more general property of GSTP1-1 that uses cooperativity to oppose a complete loss of activity caused by physical or chemical killers (21). Furthermore, half of the site reactivity, an extreme case of negative cooperativity, has been described for the Alpha class GSTA1-1 when it binds selected substrates (24), and also for the Theta class GSTT2-2 in the reaction with menaphtyl sulfate (25). A crucial lock-andkey motif seems to be the structural determinant for cooperativity in GSTs. In GSTP1-1, Tyr-49 (the key motif) is wedged into a hydrophobic pocket of the adjacent subunit formed by specific portions of helix 4 and helix 5 (the lock motif). Structurally similar Tyr or Phe residues play an identical role in Alpha and Mu class GSTs. Although PfGST displays a similar lock-and-key motif, with the only difference that Phe-56 replaces Tyr-49, it shows a peculiar case of cooperativity never observed in other GSTs. In fact, the low-high affinity transition described in the hysteretic mechanism shown in Scheme 1 is triggered by GSH binding, but it cannot be included in a classical homotropic interaction, i.e. binding of the ligand to the low affinity subunit and a subsequent shift of the vacant subunit toward the high affinity conformation. In fact, this transition occurs only when both of the G-sites have simultaneously bound GSH. This phenomenon may be named "pseudo-cooperativity" and could reflect a primordial cooperativity that did not evolve toward a correctly optimized mechanism. Probably, GSH concentrations in P. falciparum are high enough to stabilize the high affinity conformation, and thus the evolutionary pressure to optimize this event could have been quite low.
A completely different cooperativity is observed in the interaction of this enzyme with hemin. This toxic metabolite triggers a true positive homotropic behavior in PfGST, causing a 20-fold increased affinity in the vacant subunit as a consequence of its binding to one subunit. Although all GSTs bind hemin with high affinity, we were not able to find any cooperativity in the hemin interaction with Alpha, Pi, and Mu GSTs (data not shown). Thus, the existence of a homotropic regulation in PfGST for hemin binding is a peculiar event and suggests an important metabolic finality, i.e. the protection of the parasite against this toxic compound. Interestingly, PfGST seems to have evolved to interact efficiently with hemin without a parallel optimization of its catalytic competence toward hydrophobic toxic compounds. In fact, the scarce affinity for classical GST substrates is the prominent cause of its very low catalytic activity, stressed in previously published studies (13,14) and confirmed by all of our present data. Our experiments suggest that PfGST deprotonates the bound GSH efficiently, as observed for the human Alpha, Pi, and Mu GSTs, and also its catalytic competence, evaluated under co-substrate saturating conditions, is similar to that found in the human enzymes. However, the efficient deprotonation of GSH seems to be finalized more to trigger the homotropic binding of hemin rather than to efficiently catalyze the conjugation of GSH to toxic compounds. The disappearance of homotropy when GSCH 3 replaces GSH is convincing evidence in this direction. In summary, the discovery of cooperativity and the possible localization of hemin in the crystal structure, as given by docking simulations, may be useful in designing selective inhibitors of PfGST that lower its ability to bind the parasitotoxic hemin.