Structural flexibility modulates the activity of human glutathione transferase P1-1. Influence of a poor co-substrate on dynamics and kinetics of human glutathione transferase.

Presteady-state and steady-state kinetics of human glutathione transferase P1-1 (EC) have been studied at pH 5.0 by using 7-chloro-4-nitrobenzo-2-oxa-1,3-diazole, a poor co-substrate for this isoenzyme. Steady-state kinetics fits well with the simplest rapid equilibrium random sequential bi-bi mechanism and reveals a strong intrasubunit synergistic modulation between the GSH-binding site (G-site) and the hydrophobic binding site for the co-substrate (H-site); the affinity of the G-site for GSH increases about 30 times at saturating co-substrate and vice versa. Presteady-state experiments and thermodynamic data indicate that the rate-limiting step is a physical event and, probably, a structural transition of the ternary complex. Similar to that observed with 1-chloro-2,4-dinitrobenzene (Ricci, G., Caccuri, A. M., Lo Bello, M., Rosato, N., Mei, G., Nicotra, M., Chiessi, E., Mazzetti, A. P., and Federici, G. (1996) J. Biol. Chem. 271, 16187-16192), this event may be related to the frequency of enzyme motions. The observed low, viscosity-independent kcat value suggests that these motions are slow and diffusion-independent for an increased internal viscosity. In fact, molecular modeling suggests that the hydroxyl group of Tyr-108, which resides in helix 4, may be in hydrogen bonding distance of the oxygen atom of this new substrate, thus yielding a less flexible H-site. This effect might be transmitted to the G-site via helix 4. In addition, a new homotropic behavior exhibited by 7-chloro-4-nitrobenzo-2-oxa-1,3-diazole is found in Cys-47 mutants revealing a structural intersubunit communication between the two H-sites.

) (2) but a "poor" substrate for the Pi and Mu class isoenzymes having k cat NBD-Cl values only 2-4% of those found with CDNB (2). Steady-state experiments performed at pH 5.0 in the presence of NBD-Cl or CDNB indicate that a minimum rapid equilibrium random sequential bi-bi model is sufficient to describe the catalytic mechanism by GST P1-1, similar to that observed with CDNB at pH 6.5 (3). Presteady-state kinetics, thermodynamic data, and viscosity variation experiments suggest that the rate-limiting step with NBD-Cl is, probably, a structural transition of the ternary complex. Interestingly, intra-and intersubunit communications between active sites are triggered by this new co-substrate. These findings have been tentatively explained on the basis of the x-ray crystal structure.
Spectrophotometric Measurements-Steady-state kinetics of GST P1-1 with NBD-Cl as co-substrate were followed spectrophotometrically at 419 nm where the GS-NBD conjugate has its absorption maximum (⑀ 419 nm ϭ 14.5 mM Ϫ1 cm Ϫ1 ) (2). Activity with CDNB was measured at 340 nm where the product absorbs (⑀ 340 nm ϭ 9.6 mM Ϫ1 cm Ϫ1 ) (6). Spectrophotometric measurements were performed with a double-beam UVICON 940 spectrophotometer (Kontron Instruments) equipped with a cuvette holder thermostatted at 25°C. Kinetic experiments were done in 1 ml (final volume) of 0.1 M sodium acetate buffer, pH 5.0, containing 5-10 g of GST P1-1 and variable amounts of substrates. The reaction rates were measured at 0.1-s intervals for a total period of 12 s. Initial rates were determined by linear regression and corrected for the spontaneous reaction. Kinetic data with NBD-Cl were collected by varying NBD-Cl from 10 to 200 M and GSH from 12 to 500 M over a matrix of * This work was partially supported by a grant of Ministero dell'Università e della Ricerca Scientifica e Tecnologica (fund 40%) and by a grant from National Research Council, Progetto Finalizzato ACRO. 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.
Kinetic data with CDNB were obtained at pH 5.0 by varying CDNB and GSH from 50 M to 1 mM over a matrix of 36 substrate concentrations.
Fluorometric Measurements-Binding of GSH, EtS-NBD, and NBD-Cl to the enzyme was followed in 0.1 M sodium acetate buffer, pH 5.0, by the intrinsic fluorescence quenching of the enzyme after ligand addition with the procedure previously described (8). Fluorescence measurements were done with a Perkin-Elmer LS-5 fluorometer with a sample holder at 25°C. Excitation was at 280 nm and the emission was at 340 nm.
Data Analysis-The nomenclature of kinetic parameters was essentially that of Segel (7). The dissociation constant (K) for substrates and the substrate analogue EtS-NBD was obtained by fluorescence measurements by fitting the data to the simplest model by assuming a single and noncooperative binding site per subunit as shown in Equation 1: where Y corresponds to the fractional saturation, K is the dissociation constant, and [L] is the ligand concentration. The dissociation constant values were used for the analysis of kinetic data that were fitted to a rapid equilibrium random sequential bi-bi model, according to Scheme 1. The parameters to be varied in the fitting procedure were only limited to V max and the coupling factor ␣, according to Equation 2: where v is the initial velocity, [A] and [B] are GSH and NBD-Cl (CDNB) concentrations, and K A and K B their respective dissociation constants (without the second substrate) obtained by fluorescence experiments. The dissociation constant by fluorometry for CDNB is uncertain due to its high absorbance at 280 nm and its low affinity for GST; K B CDNB was then left free to vary in the kinetic fits. ␣ is the coupling factor. V max is expressed as [product]⅐s Ϫ1 ⅐[GST] Ϫ1 . Parameters derived from binding data (K A and K B

NBD-Cl
) and from the fit of kinetic data (V max , K B CDNB , and ␣) were used as follows: (i) to calculate the theoretical curves of log K A app versus log [B], i.e. to correlate the variation of the affinity of GST P1-1 for a substrate to the concentration of the second substrate; data were fitted according to Equation 3. The fluorescence data for binding of the unreactive substrate analogue EtS-NBD in the presence of GSH were analyzed by using Equation 3. The theoretical curve is calculated by using values of K A and

K B
EtS-NBD obtained by fluorometry and ␣ by kinetic fit. Binding and kinetic data were analyzed with a software package WIN-MATLAB (Math Works-South Natick, MA).
Stopped Flow Analysis-Stopped-flow measurements were performed on a stopped-flow spectrophotometer consisting of a High-Tech SHU-51 rapid mixing device thermostatted at 25°C equipped with a Jasco J600 spectropolarimeter. Light path of the cell was 0.2 cm. Deadtime of the instrument was Յ5 ms. Presteady-state analysis was performed by rapid mixing of GST P1-1 (about 100 M active sites) in 0.1 M sodium acetate buffer, pH 5.0, containing 4 mM GSH with an identical volume of 0.4 mM NBD-Cl in the same buffer. Blank was done in the same conditions without GST. The increase of absorbance at 419 nm was monitored every 1 ms. Spectrum of the species which accumulates in the dead time (Յ5 ms) was constructed point by point from the absorbance values measured at different wavelengths (from 300 to 600 nm) after rapid mixing 100 M GST active site in 0.1 M sodium acetate buffer, pH 5.0, containing 500 M GSH with an identical volume of 100 M NBD-Cl in the same buffer. At each wavelength, the absorbance value at t ϱ was subtracted from the absorbance value at t°, and this value was then added to the absorbance of the product at the same wavelength (9).
Cooperativity-The cooperativity of C47S and C47S/C101S mutants toward GSH was assayed by varying GSH concentration (from 5 M to 1 mM) at 0.2 mM NBD-Cl constant concentration (in 0.1 M sodium acetate buffer, pH 5.0, and 25°C). Similar experiments at fixed GSH concentration (0.5 mM) and variable NBD-Cl (from 2.5 M to 0.2 mM) were used to check the cooperativity toward NBD-Cl. Kinetic data were analyzed by the KaleidaGraph (version 2.0.2) (Abelbeck Software) computer program and fitted to a rate equation expressing positive cooperativity as previously reported (8).
Viscosity Dependence of Kinetic Parameters-The effect of viscosity on kinetic parameters and on cooperativity was assayed by using 0.1 M sodium acetate buffer, pH 5.0, containing variable glycerol concentrations. Viscosity values () at 25°C were calculated as described (10) and randomly checked with an Ostwald viscometer. Viscosities are reported relative to 0.1 M sodium acetate buffer, pH 5.0.
Molecular Modeling-Modeling was performed on a Silicon Graphics Indigo 2 workstation using the software modeling package INSIGHT II (Biosym Technologies Inc.). Molecular dynamics simulations and energy minimization calculations were performed using DISCOVER (Biosym Technologies Inc.). A model of NBD-Cl was built and energyminimized to tidy up its geometry. The ligand was roughly positioned in the H-site of the human GST P1-1 so it was bound in a productive mode (i.e. the chlorine atom was positioned close to where the thiol group of GSH would be located when bound). Molecular dynamics simulation was performed on the complex, followed by 200 steps of conjugate gradient energy minimization to relieve any residual strain and to optimize any potential electrostatic and van der Waals contacts. This approach was applied to NBD-Cl, CDNB, and their glutathione conjugates.

RESULTS
Steady-state Kinetics-Steady-state kinetics with NBD-Cl at pH 5.0 was analyzed and compared with that obtained with CDNB under the same conditions. Thermodynamic constants obtained by fluorometry (Table I) were used for fitting the SCHEME 1 TABLE I Dissociation constants and kinetic parameters at pH 5.0 and pH 6.5 ␣ is the coupling factor; K A is the dissociation constant for the GST-GSH complex; K B is the dissociation constant for the GST-co-substrate complex. V max is coincident to k cat .
Although k cat NBD-Cl Х k cat CDNB at pH 5.0, NBD-Cl must be considered a poor substrate for GST Pl-1 as this pH value is the optimal one for the NBD-Cl reaction, while k cat CDNB is 76 s Ϫ1 at pH 6.5. b Data from Ref. 12 kinetic data as reported under the "Experimental Procedures." As already found with CDNB as co-substrate at pH 6.5 (3), the rapid equilibrium random sequential bi-bi model (Scheme 1) or the mathematically equivalent steady-state ordered bi-bi mechanism was a minimum model that describes satisfactorily the experimental data for both co-substrates at pH 5.0 (Fig. 1). A diagnostic procedure to distinguish between these two mechanisms is based on product inhibition studies (7). The pattern of inhibition at pH 5.0 by GS-NBD (Fig. 2) is consistent with a rapid equilibrium random mechanism (7). Data presented in the form of a double-reciprocal plot (Fig. 1) provide evidence of the convergence of the lines above the abscissa axis, showing a coupling factor ␣ ϭ 0.036 with NBD-Cl and ␣ ϭ 0.38 with CDNB. These fractional numbers indicate that the affinity for GSH increases in the presence of NBD-Cl about 30 times (and vice versa), but only 3 times in the presence of CDNB (and vice versa) (7). The agreement between fitting procedures and experimental data is evident in Fig. 3 which shows the plots obtained by using Equations 3-5. Furthermore, the results obtained from fluorescence quenching measurements for the binding of a nonreactive analogue of the second substrate (EtS-NBD), in the presence of variable GSH concentrations, indicate that the kinetic synergism between the G-site and H-site must be related to a mutual modulation of the dissociation constants (Fig. 4).
Presteady-state Kinetics-A careful investigation of the presteady-state kinetics of GST P1-1 with NBD-Cl as co-substrate does not confirm our preliminary evidence for an accumulation of GS-NBD product before the attainment of the steady state (12). No evident burst phase is now observed at 419 nm where GS-NBD absorbs strongly (2) (Fig. 5, inset). Some technical problems were responsible for that earlier report; the major problem was an inadequate washing volume for the cell (based on advice by manufacturer). Furthermore, the spectral image of the accumulating species, obtained as described under "Experimental Procedures," is very similar to that of NBD-Cl with a little red shift of about 3 nm (Fig. 5). This spectrum has been assigned to the NBD-Cl complexed enzyme, since a very different spectrum is expected for the Meisenheimer (or ) complex which occurs as an intermediate in similar aromatic substitution reactions (13). On the basis of the present data and of the previously reported k cat independence of the nature of the leaving group (12), we suggest that the ratelimiting step is a physical event, and it must occur between the ternary complex formation and the chemical step, as found for the CDNB/GSH-catalyzed reaction (1). Thermodynamic data also indicate different rate-determining events for the catalyzed and uncatalyzed reactions. In the latter the -complex formation was found to be the rate-determining step (12). Calculations from the Arrhenius plots shown in Fig. 6 give an activation energy for the enzymatic NBD-Cl/GSH reaction of about 20 kJ/mol higher than that for the spontaneous one. This may be a paradoxical result if a coincident rate-limiting step occurs in both reactions.
Cys-47 Mutants Display Cooperativity toward Both GSH and NBD-Cl-When CDNB is the co-substrate, the substitution of Cys-47 by Ala or Ser does not change the k cat value but lowers the affinity of the G-site for GSH (about 9-fold for the C47S mutant and 14-fold for the C47S/C101S double mutant) and triggers a marked positive cooperativity toward GSH (8). An increased flexibility of helix 2, due to the lack of a peculiar electrostatic bond between Cys-47 and Lys-54, may be the basis of this structural interaction between subunits (4, 8). Conversely, no homotropic effect by CDNB is observed (8). A more complex behavior is found with NBD-Cl. As shown in Table II the affinity toward both GSH and NBD-Cl decreases 4 -10-fold in the mutants while k cat values are quite unchanged. The homotropic behavior by GSH is again recovered both for C47S mutant and C47S/C101S double mutant (Table II) with Hill coefficients very similar to that obtained in the presence of CDNB (8). Surprisingly, a positive cooperativity is observed toward NBD-Cl with Hill coefficient values of 1.56 and 1.40 for the single and double mutants, respectively (Table II). This reveals a structural intersubunit communication between the two H-sites of the dimeric enzyme which is absent or not detectable in the wild-type and in Cys-47 mutants with CDNB as co-substrate.
Viscosity Effect on Wild Type-In the preceding paper (1) we demonstrated a viscosity dependence both of k cat and K m GSH values for the GST-catalyzed reaction of CDNB with GSH. This has been interpreted assuming that kinetic parameters depend on some diffusion-controlled motions of the protein. In particular, helix 2 flexibility seems to be involved in determining the G-site affinity, although it is unclear whether it plays a role in the k cat modulation. By using NBD-Cl as co-substrate, k cat , K m GSH , and K m NBD-Cl remain almost unchanged even at a high viscosity value (Table II). These data may be interpreted by means of the x-ray structure of the enzyme (see "Discussion"); moreover, they are a good control to show that the viscosity dependence of k cat and K m GSH previously observed in the presence of CDNB (1) is not due to a nonviscosity solvent effect (14).
Viscosity Effect on C47S and C47S/C101S Mutants-As ob-  (Table II). On the contrary the positive cooperativity toward GSH, as expressed by the Hill coefficient, is markedly lowered by increasing the viscosity (Table II). Thus it appears that the flexibility of helix 2, responsible for the GSH cooperativity in these mutants (1,8), is still diffusion-controlled in the presence of NBD-Cl as co-substrate. Conversely, the homotropic behavior of NBD-Cl is viscosity-independent (Table II). This new intersubunit interaction between H-sites should involve protein regions different from helix 2.

DISCUSSION
Steady-state kinetic data at pH 5.0 with NBD-Cl (or CDNB) as co-substrate are consistent with a rapid equilibrium random sequential bi-bi model (Figs. 1-3). The effects of change of the leaving group (12) and of temperature (Fig. 6) on k cat and presteady-state kinetic data (Fig. 5) suggest that the ratelimiting step in the NBD-Cl/GSH system is a physical event that occurs before the -complex formation, i.e. a structural transition of the ternary complex as occurs in the CDNB/GSH system (1). However, with NBD-Cl a strong kinetic and binding synergism between H-and G-site is observed (Figs. 1 and 4 and Table I) which implies the existence of a mutual communication between these binding sites. Only a little positive modulation appears with CDNB at the same pH value ( Fig. 1 and Table I). Unlike CDNB, NBD-Cl gives a very low and viscosityindependent k cat value (Table II). These findings may be tentatively interpreted on the basis of x-ray structure of GST P1-1 (15). A computer modeling shows the rings of NBD-Cl packed between the two aromatic rings of Phe-8 and Tyr-108. Furthermore, the oxygen atom of the 5-membered ring of NBD-Cl (for structure, see Fig. 5) is in hydrogen-bonding distance of the hydroxyl group of Tyr-108 (3.1 Å) which resides in helix 4 (residues 83-109) (Fig. 7). This hydrogen bond, missed in the CDNB complex (distance from the oxygen atoms of the nitro groups Х5 Å), may in part account for the high affinity of GST toward NBD-Cl (␣K B ϭ 3.3 M and K B ϭ 93 M), and it probably causes an increased rigidity of the active site.
There is evidence that helix 4 displays diffusion-controlled motions in absence of ligands (1). If these motions are assumed to be related to the conformational transition of the ternary complex (rate-limiting), the observed low k cat value and its viscosity independence (in the presence of NBD-Cl) are the logical consequence of a less flexible helix 4. In fact, the energy barrier for a rate-limiting structural transition depends on the relative contributions of internal and external friction (14); thus, increased intramolecular constraints could minimize the effect of the external friction due to solvent viscosity. 2 In addi- 2 During the revision of this paper, according to a reviewer's suggestion, we checked the effect of nondenaturing urea concentrations (up to 2 M) on k cat values. With NBD-Cl (in 0.1 M sodium acetate buffer, pH 5.0), k cat increases up to 200% of the basal value at 2 M urea. No change of k cat has been observed with CDNB as co-substrate (in 0.1 M potassium phosphate buffer, pH 6.5). By assuming that nondenaturing urea concentrations only lower the internal friction of a protein, its different effect on k cat values suggests that the internal constraints play a crucial role in determining the structural transition energy barrier only in the presence of NBD-Cl as co-substrate. These findings agree well with the conclusions of this paper.  tion, helix 4 provides a number of important residues close to or in contact with the bound GSH such as Asp-94, Glu-97, and Asp-98 (15). Hence, a structural perturbation due to the interaction of NBD-Cl with Tyr-108, localized on the H-site, may be transmitted to the G-site via helix 4 yielding the observed synergistic effect between the H-and G-sites responsible for the increased affinity toward substrates. As concerns K m GSH , its viscosity-independent value (Table II) seems no more related to diffusion-controlled motions of helix 2 that, on the contrary, are relevant in determining K m GSH in the CDNB/GSH system (1). It is possible that the increased rigidity of the active site due to NBD-Cl binding may be transmitted to helix 2. Alternatively, helix 2 motions may be always diffusiondependent, but the K m GSH value is mainly determined by GSH interaction with helix 4 promoted by NBD-Cl. The viscosity effect on the Hill coefficients of Cys-47 mutants (Table II) suggests that helix 2 (whose flexibility modulates this parameter (1)) is still diffusion-controlled even in the presence of NBD-Cl.
Preliminary evidence obtained with a mutated GST P1-1, where Tyr-108 is replaced by Phe, confirms the influence of the hydroxyl group of Tyr-108 on k cat and K m NBD-Cl as these kinetic parameters increase remarkably and become viscosity-dependent in this mutant. 3 Comparison of the crystallographic data and of the catalytic properties toward NBD-Cl in Alpha, Pi, and Mu GST isoenzymes indirectly confirms the above suggestions. Both Pi and Mu class isoenzymes, characterized by low k cat and K m NBD-Cl values (2), have equivalently located hydroxyl groups (Tyr-108 and Tyr-115, respectively) which may form a hydrogen bond with the heterocyclic portion of NBD-Cl. On the contrary, Alpha class GST, which displays higher k cat and K m NBD-Cl values (2), lacks this hydrogen bond as a Val residue replaces the Tyr residue.