7-Nitro-2,1,3-benzoxadiazole derivatives, a new class of suicide inhibitors for glutathione S-transferases. Mechanism of action of potential anticancer drugs.

Spectroscopic and rapid kinetic experiments were performed to detail the interaction of human glutathione S-transferases GSTA1-1, GSTM2-2, and GSTP1-1 with 6-(7-nitro-2,1,3-benzoxadiazol-4-ylthio)hexanol (NBDHEX). This compound is a representative molecule of a new class of 7-nitro-2,1,3-benzoxadiazole (NBD) derivatives (non-GSH peptidomimetic compounds) that have been designed both to give strong GST inhibition and to accumulate in tumor cells avoiding the extrusion mechanisms mediated by the multidrug resistance protein pumps. We have recently shown that submicromolar amounts of NBDHEX trigger apoptosis in several human tumor cell lines through the dissociation of the JNK.GSTP1-1 complex (Turella, P., Cerella, C., Filomeni, G., Bullo, A., De Maria, F., Ghibelli, L., Ciriolo, M. R., Cianfriglia, M., Mattei, M., Federici, G., Ricci, G., and Caccuri, A. M. (2005) Cancer Res. 65, 3751-3761). Results reported in the present study indicated that NBDHEX behaves like a suicide inhibitor for GSTs. It bound to the H-site and was conjugated with GSH forming a sigma complex at the C-4 of the benzoxadiazole ring. This complex was tightly stabilized in the active site of GSTP1-1 and GSTM2-2, whereas in GSTA1-1 the release of the 6-mercapto-1-hexanol from the sigma complex was the favored event. Docking studies demonstrated the likely localization of the sigma complex in the GST active sites and provide a structural explanation for its strong stabilization.

Glutathione S-transferases (GSTs, 1 EC 2.5.1.18) are a superfamily of dimeric enzymes present in human tissues and subdivided in at least eight gene-independent classes named Alpha, Pi, Mu, Theta, Zeta, Omega, Sigma, and Kappa (1-7). The design or discovery of efficient compounds that may bind to these enzymes modulating their biological activity has become one of the primary aims in cancer research because GST isoenzymes such as GSTP1-1 are overexpressed in many cancer cell lines (8) and induce drug resistance by inactivating many chemotherapeutic compounds via GSH conjugation (9). GSTP1-1 displays an additional antiapoptotic activity based on a protein-protein interaction with c-Jun N-terminal kinase (JNK), a key enzyme in the apoptotic cascade (10). Many efforts have been made in the last years to find tight inhibitors of these enzymes to reduce their protective role in vivo (9). Ethacrynic acid, an inhibitor that lacks class specificity for GSTs, represented a first attempt in this direction; however, its scarce affinity and the deleterious side effects have discouraged its use in clinical practice (9). Taking advantage of the strong specificity of the transferase G-site for GSH and GSH derivatives, a few GSH peptidomimetic compounds have been designed recently. For example, ␥-glutamyl-S-(benzyl)-cysteinyl-R(Ϫ)-phenylglycine diethyl ester (TER199) acts as proinhibitor for GSTs. It rapidly enters the cells and is activated by the intracellular esterases. In the active form, it selectively inhibits GSTP1-1, enhancing the effect of alkylating drugs in various cancer cell lines that overexpress this isoenzyme (11). This specific compound is under investigation for possible use in human therapy, but other GSH derivatives that are good inhibitors of GSTs may not be so efficient in cancer cells. In fact many GSH derivatives are actively extruded from the cell by specific export pumps, such as the multidrug resistance protein, thus avoiding their intracellular accumulation (12,13). Multidrug resistance protein activity represents a crucial problem for GST inhibition as the estimated concentration of GSTP1-1 in cancer cells may reach 0.05 ϫ 10 Ϫ3 M, and an equivalent or higher concentration of inhibitor must accumulate in the cell to modulate efficiently the GST activity. To overcome this problem, we have tried to design new molecules that bind efficiently to GSTs but are not GSH derivatives and display suitable lipophilic properties to be able to cross the cell membrane. Besides the G-site, which specifically recognizes GSH and its derivatives, a second binding subsite is present in all GSTs that is able to interact with a lot of different hydrophobic co-substrates. This subsite, termed H-site, is a hydrophobic cavity, located near the G-site, that displays different topography and substrate specificity in GSTA1-1, GSTM2-2, and GSTP1-1. As GSTs evolved to bind many different hydrophobic toxic species, the H-site normally displays moderate affinity for these compounds, i.e. K m values in the millimolar range. A few years ago, we found that 4-chloro-7-nitro-2,1,3-benzoxadiazole (NBD-Cl) is an unusual co-substrate for GSTs showing a K m value in the micromolar range (14). We therefore started from this evidence to design specific derivatives of NBD-Cl that are the object of the present study. We have recently shown that very low levels of NBD thioethers can induce tumor cell death through dissociation of the JNK⅐GSTP1-1 complex (15). The NBD derivatives represent an interesting class of GST inhibitors as they interact with the H-site with K D values in the micromolar range for GSTA1-1 and GSTP1-1 and in the nanomolar range for GSTM2-2. A detailed investigation of the binding mechanism of a representative molecule revealed that GSTs promote the formation in the active site of a complex between the inhibitor and GSH. Thus, these NBD derivatives act as suicide inhibitors and may represent a new type of potent antitumor agents.
Synthesis and Characterization of the NBD Derivatives-NBD-Cl (1 mmol) and 6-mercapto-1-hexanol (2 mmol) were incubated for 6 h in 20 ml of a 1:1 (v/v) mixture of ethanol and 0.1 M potassium phosphate buffer, pH 7.0, in a closed vessel at 25°C. The pH was continuously monitored and kept neutral by suitable addition of 1 M KOH. At the end of the reaction, the excess of 6-mercapto-1-hexanol was removed with 1.5 mmol of 3-bromopyruvate. The reaction product, 6-(NBD-4-ylthio-)hexanol (NBDHEX), is a dark yellow insoluble compound that was collected by filtration and washed twice with 15 ml of cold distilled water. The final product was 98% pure via HPLC analysis.
The mass spectrum of NBDHEX was obtained under electrospray conditions using an LCQ (Finnigan, San Josè, CA) instrument. m/z was 298 (M ϩ H) ϩ .
Cell Culture and Treatments-K562 (human myeloid leukemia), HepG2 (human hepatic carcinoma), and GLC4 (human small cell lung carcinoma) cell lines, kindly provided by Prof. M. Cianfriglia (Istituto Superiore di Sanità , Rome, Italy), were grown in RPMI 1640 medium supplemented with 5% fetal calf serum, 2 mM L-glutamine, 100 units/ml penicillin, and 100 mg/ml streptomycin in a humidified 5% CO 2 atmosphere at 37°C. NBDHEX stock solution (50 mM) was prepared in Me 2 SO. Just before use, the stock solution was diluted to the appropriate concentration in RPMI 1640 cell medium with a final Me 2 SO concentration not exceeding 0.05-0.1%. In all experiments control samples were exposed to the same Me 2 SO concentrations; 0.05-0.1% Me 2 SO had no cytotoxic effects. An evaluation of cell viability at different drug concentrations was determined by the sulforhodamine B assay (16). The cells were placed in 96-well microtiter plates at a density of 1.5 ϫ 10 4 cells/well in 100 l of medium. After 24 h, the cells were exposed to NBDHEX at the required concentration and allowed to incubate for 48 h. After incubation, the cell growth was evaluated by an in situ cell fixation procedure followed by a coloring procedure with sulforhodamine B. The dose-response profiles obtained gave the LC 50 value for NBDHEX (the concentration used to obtain 50% cell mortality).
The extent of apoptosis induced by NBDHEX was determined in GLC4, HepG2, and K562 cell lines after 24-h treatment with 10 ϫ 10 Ϫ6 M NBDHEX. Apoptosis was detected with a fluorescence microscope by analyzing the nuclear fragmentation after staining with the DNAspecific dye Hoechst 33342.
GST Activity-GST activity was assayed at 25°C as reported previ-ously (18). The standard assay mixture contained 1 ϫ 10 Ϫ3 M GSH, 1 ϫ 10 Ϫ3 M CDNB, and 0.1 ϫ 10 Ϫ3 M EDTA in 1 ml (final volume) of 0.1 M potassium phosphate buffer, pH 6.5. The activity was determined spectrophotometrically at 340 nm (where the product of reaction absorbs ⑀ ϭ 9,600 M Ϫ1 cm Ϫ1 ) within 10 s from addition of the substrates. One unit of GST is defined as the amount of enzyme that catalyzes the formation of 1 mol of product/min at 25°C. Inhibition of GSTs by NBD derivatives was studied using 44 ϫ 10 Ϫ9 M GSTA1-1 or GSTP1-1 or 10 ϫ 10 Ϫ9 M GSTM2-2. The IC 50 (inhibitor concentration that inhibits 50% of enzymatic activity) values were determined in the presence of various amounts of the selected NBD derivative. The inhibition mechanism with NBDHEX was determined from double reciprocal plots of the enzyme activity at different GSH and CDNB concentrations and in the presence of different amounts of inhibitor.
Binding of NBDHEX to GSTs-The affinities of NBDHEX for human GSTs were determined in the absence or presence of 1 ϫ 10 Ϫ3 M GSH. In a typical experiment, the quenching of the intrinsic fluorescence of the protein (excitation at 295 nm and emission at 340 nm) was measured in a single photon counting spectrofluorometer (Fluoromax, S. A. Instruments, Paris, France) at 25°C after the addition of variable amounts of NBDHEX to 4 ϫ 10 Ϫ6 M GSTs in 0.1 M potassium phosphate buffer, pH 6.5. Fluorescence data were corrected both for dilution and for inner filter effects and fitted to Equation 1, which yields the K D value for the NBDHEX bound to GSTs, where ⌬F is the protein fluorescence change observed in the presence of a given amount of NBDHEX, and ⌬F max is the maximum fluorescence change observed at saturating NBDHEX concentration; n H is the Hill coefficient.
Spectrophotometric and Fluorometric Analysis-The reaction between NBDHEX and GSH was followed in the absence of GSTs by recording at 25°C the spectrum of NBDHEX (50 ϫ 10 Ϫ6 M) in 0.1 M potassium phosphate buffer, pH 6.5, in the presence of different concentrations of GSH (from 5 ϫ 10 Ϫ3 to 80 ϫ 10 Ϫ3 M). The dissociation constant of the NBDHEX⅐GSH complex was calculated from the absorbance change (⌬A) at 432 nm following the addition of GSH. The experimental data were fitted to Equation 1 where ⌬F and ⌬F max are replaced by ⌬A and ⌬A max . Reversibility of the reaction was observed after acidification of the above mixtures with 1 M HCl. Moreover the UVvisible spectrum of NBDHEX (50 ϫ 10 Ϫ6 M) in 0.1 M potassium phosphate buffer, pH 6.5, containing 1 ϫ 10 Ϫ3 M GSH was recorded at 25°C before and after the addition of stoichiometric amounts of either GSTP1-1 or GSTM2-2, and the reversibility of the reaction was observed by recording UV-visible spectrum of this mixture after 5-min incubation with 7 M urea. With GSTA1-1, the UV-visible spectrum of NBDHEX (20 ϫ 10 Ϫ6 M) was recorded both at 8 and at 25°C in the presence of stoichiometric amounts of the enzyme in 0.1 M potassium phosphate buffer, pH 6.5, containing 1 ϫ 10 Ϫ3 M GSH. The fluorescence spectrum of 10 ϫ 10 Ϫ6 M NBDHEX (excitation at 430 nm) in 0.1 M potassium phosphate buffer, pH 6.5, containing 1 ϫ 10 Ϫ3 M GSH was recorded at 25°C before and after the addition of 20 ϫ 10 Ϫ6 M GSTA1-1, GSTM2-2, or GSTP1-1.
Stopped Flow Experiments-Rapid kinetic experiments were performed on an Applied Photophysics kinetic spectrometer stopped flow instrument (dead time ϭ 1 ms) equipped with a 1-cm light path observation chamber. In the absence of GSTs, variable amounts of GSH (from 200 ϫ 10 Ϫ3 to 15 ϫ 10 Ϫ3 M) were rapidly mixed with an identical volume of NBDHEX (0.1 ϫ 10 Ϫ3 M) dissolved in 0.1 M potassium phosphate buffer, pH 6.5 (25°C). Experiments in the presence of GSTs were performed by rapidly mixing the selected GST (about 20 ϫ 10 Ϫ6 M in 0.1 M potassium phosphate buffer, pH 6.5, containing 2 ϫ 10 Ϫ3 M GSH) with an identical volume of NBDHEX (from 40 ϫ 10 Ϫ6 to 2 ϫ 10 Ϫ6 M) dissolved in the same buffer (4°C).
1000 data points were collected on a logarithmic time base (from 1 ms to 10 s). Time-dependent spectra were reconstructed from single wavelength observations (between 490 and 310 nm) by repetitively changing the wavelength following different reagent mixing steps; a 10-nm increment step and 6-nm bandwidth were utilized. Optical deconvolution of time-dependent spectra sets was performed by means of the software MATLAB (MathWorks, South Natick, MA) running on an Intel Pentium-based personal computer by using singular value decomposition (SVD) in combination with curve fitting algorithms. The matrix of time-dependent spectra (A) is decomposed by SVD into the product of three matrices, A ϭ U ϫ S ϫ V T where the U columns are the basis spectra, and their time dependence is represented by the V columns.
The diagonal values of the S matrix yield the relative occupancies of the basis spectra in the data set. If a data set is composed of more than one optical transition, deconvolution of the optical components (provided they have different time courses) can be achieved by simultaneously fitting the chosen V column subset to the kinetic scheme. Such best fits of the experimental data to kinetic scheme, performed using the program GEPASI 3.30 (19 -21), were carried out by means of numerical integration, at variable steps, of the ordinary differential equations according to the selected kinetic scheme. The resulting matrix of the time dependences of the molar fraction of the intermediate species can be used to reconstruct (from the experimental matrix of time-dependent spectra) the spectra of the fully populated intermediate species (22)(23)(24).
Docking Simulations-Docking simulations were performed with the program Autodock 3.0.5 (25). For each isoenzyme, the ligand was removed from the Protein Data Bank structure file (Protein Data Bank codes 1k3l, 1hnc, and 6gss for GSTA1-1, GSTM2-2, and GSTP1-1, respectively), polar hydrogen atoms were added geometrically, and Kollman united atom charges (26) were assigned. Docking studies of the complex were performed with a flexible ligand, but the internal degrees of freedom of the GSH moiety were held fixed to the values obtained from the crystallographic conformation. Charges were added by the Gasteiger-Marsili method (27). Grids of molecular interactions were calculated in a cubic box: size, 25.125 Å; grid spacing, 0.375; centered on the oxygen atom of the tyrosine side chain participating in GSH activation (Tyr-9, Tyr-6, and Tyr-7 in the GSTA1-1, GSTM2-2, and GSTP1-1, respectively). Docking was performed 100 times using the Lamarckian genetic algorithm with random starting position and conformation, a population size of 50, standard parameters (25), 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 root mean square deviation. The MOLMOL program (28) was used for graphical interpretation and representation of results. Table I were synthesized as reported under "Experimental Procedures" and tested as possible inhibitors for GSTA1-1, GSTM2-2, and GSTP1-1, representative enzymes of human Alpha, Mu, and Pi classes, respectively. All the NBD thioether derivatives (compounds 1-4) were strong inhibitors of GSTs showing IC 50 values in the nanomolar range for GSTM2-2 and in the micromolar range for GSTP1-1 and GSTA1-1 (Table I). Interestingly the presence of the thioether bond seems to be an essential requirement for tight binding. In fact, all GSTs displayed much lower affinities for 6-(NBD-4-ylamino)hexanol (compound 5), which is a structural analogue of NBDHEX but with the sulfur atom replaced by an amino group (Table I).

Inhibition of GSTs by NBD Derivatives-The NBD derivatives listed in
Among all these compounds, NBDHEX showed the lowest IC 50 value toward GSTP1-1, the isoenzyme overexpressed in most cancer cell lines (8). This molecule was therefore utilized to detail the interaction of these compounds with human GSTs.
The inhibition mechanism of NBDHEX was studied with GSTP1-1 and GSTA1-1 by varying either GSH or CDNB concentration. An accurate analysis with GSTM2-2 was not possible due to its very high affinity toward NBDHEX. With GSTP1-1 and GSTA1-1 NBDHEX behaved like a competitive inhibitor in respect to CDNB indicating that this compound binds to the same hydrophobic pocket of the active site (H-site). In respect to GSH NBDHEX gave a mixed type inhibition, suggesting that NBDHEX interacts with the H-site of GSTs but that the adjacent G-site (which binds GSH) is also involved (data not shown).
Effect of NBDHEX on Tumor Cell Lines-The cytotoxic effect of NBDHEX was tested on leukemia and solid tumor cell lines. The LC 50 values obtained after 48-h treatment were 1.4 Ϯ 0.2 ϫ 10 Ϫ6 , 1.5 Ϯ 0.1 ϫ 10 Ϫ6 , and 2.9 Ϯ 0.3 ϫ 10 Ϫ6 M for GLC4, K562, and HepG2 cell lines, respectively (see Fig. 1, A-C). Evidence of apoptosis induction by NBDHEX was obtained by morphological analysis after nuclear staining with Hoechst 33342. Treatment of GLC4, HepG2, and K562 cells with NBD-HEX induced chromatin condensation and nuclear fragmentation in small rounded bodies (Fig. 1, D, F, and G), which represent the final steps of apoptosis (29). Interestingly the LC 50 values for these cell lines were of the same order of magnitude as the IC 50 value found for GSTP1-1 (0.8 ϫ 10 Ϫ6 M), which is overexpressed in these specific cells as well as in most tumor cell lines.
Isothermic Binding of NBDHEX to GSTs-The binding of NBDHEX to GSTP1-1, GSTA1-1, and GSTM2-2 was studied by following the quenching of the intrinsic fluorescence of the protein. Binding to GSTP1-1 in the presence of 1 ϫ 10 Ϫ3 M GSH ( Fig. 2A) followed an almost hyperbolic behavior (Hill coeffi-cient n H ϭ 0.9) with an apparent dissociation constant of 0.9 ϫ 10 Ϫ6 M, which is close to the IC 50 value obtained kinetically and reported in Table I. In the absence of GSH (Fig. 2, panel B), the affinity of NBDHEX toward GSTP1-1 strongly decreased (K D ϭ 86 ϫ 10 Ϫ6 M), and the binding displayed strong positive cooperativity (Hill coefficient n H ϭ 1.7).
Binding of NBDHEX to GSTA1-1 in the presence of 1 ϫ 10 Ϫ3 M GSH (Fig. 2C) followed a similar hyperbolic behavior with an apparent dissociation constant of 5.3 ϫ 10 Ϫ6 M. This value was about 5-fold lower than the IC 50 value obtained kinetically (see Table I). In the absence of GSH (Fig. 2D), the dissociation constant increased (K D ϭ 66 ϫ 10 Ϫ6 M), and a clear cooperative behavior was observed with an apparent Hill coefficient of 1.6. In the presence of 1 ϫ 10 Ϫ3 M GSH, GSTM2-2 showed so strong an affinity toward NBDHEX that was not possible to calculate the dissociation constant by using fluorescence binding experiments. An indication of the affinity between GSTM2-2 and the NBDHEX (IC 50 Յ 10 ϫ 10 Ϫ9 M) was obtained by following the inhibition of GST activity with both the enzyme and the inhibitor in the nanomolar range (Fig. 2E, and Table I). In the absence of GSH, the affinity of GSTM2-2 for NBDHEX decreased by about 3 orders of magnitude, and fluorescence binding experiments gave a K D value of 9 ϫ 10 Ϫ6 M. Spectrophotometric and Fluorometric Analysis at Equilibrium-The interaction of NBDHEX with GSTs was accomplished by remarkable spectral changes. When NBDHEX was incubated at 25°C in the presence of a stoichiometric amount of GSTP1-1 and 1 ϫ 10 Ϫ3 M GSH, the UV-visible spectrum of NBDHEX centered at 432 nm completely disappeared, and a new absorption band appeared at 348 nm (Fig. 3, panel A). No spectral changes were observed in the absence of enzyme or when GSH was omitted or replaced by 1 ϫ 10 Ϫ3 M S-methylglutathione (data not shown). Addition of urea (7 M) to the solution (after reaction of NBDHEX with enzyme and GSH) completely abolished the peak at 348 nm restoring the original NBDHEX spectrum. This indicates a reversible interaction between NBDHEX and GSH that is only possible in a folded active site. A similar behavior was observed for GSTM2-2 except that the new peak was centered at 353 nm and displayed a higher absorbance (Fig. 3, panel B). These new optical species probably represent a complex intermediate between NBDHEX and GSH stabilized in the active site of GSTP1-1 and GSTM2-2. In fact, it is well known that nucleophilic aromatic substitution reactions proceed through a complex intermediate (30,31) and that complexes of NBD have absorption bands between 300 and 400 nm (32,33). A different type of interaction was found in the presence of GSTA1-1. In this case, the NBDHEX spectrum at 432 nm did not disappears but was only shifted at 419 nm (Fig. 4A). Addition of urea (7 M) did not restore the original spectrum of NBDHEX.
When the interaction between NBDHEX, GSH, and GSTA1-1 was carried out at 8°C, a transient band centered at 343 nm was observed that was rapidly converted into the final spectrum observed at 25°C (419 nm) (Fig. 4B)  active site of GSTM2-2 and GSTP1-1, whereas the final species at 419 nm could be identified as the glutathione adduct with NBD that specifically absorbs at this wavelength (14). The formation of GS-NBD in the presence of GSTA1-1 was confirmed by HPLC analysis (data not shown) and by fluorometric analysis (see Fig. 5).
Interestingly in the absence of GSTs, no relevant spectral changes of NBDHEX could be observed at 1 ϫ 10 Ϫ3 M GSH. However, at higher GSH concentration (up to 80 ϫ 10 Ϫ3 M), a peak centered at 335 nm was formed with a simultaneous decrease of the spectral band centered at 432 nm (Fig. 6A); this spectral perturbation was similar but not identical to that observed in the presence of GSTs. Acidification of the solution to about pH 4.0 immediately restored the original NBDHEX spectrum, suggesting that the new species was formed via a reversible reaction that requires GSH in the ionized form. Spectral data obtained at variable GSH concentrations (from 5 ϫ 10 Ϫ3 to 80 ϫ 10 Ϫ3 M) gave a dissociation constant of 13 ϫ 10 Ϫ3 M for this reaction (Fig. 6, B and C). This means that, at 1 ϫ 10 Ϫ3 M GSH concentration, the complex was scarcely present as a free species in solution (about 1-2% of the total NBDHEX).
Stopped Flow Experiments in the Absence of GSTs-The interaction of NBDHEX with GSH in the absence of GSTs was detailed by means of a stopped flow apparatus. After rapid mixing of variable amounts of GSH with NBDHEX at pH 6.5 and 25°C time-dependent spectra were reconstructed from single wavelength observations (Fig. 7A). The presence of an isosbestic point in the differential spectra reported in Fig. 7D suggests that only a single new optical transition occurred during this reaction.
The whole data set obtained at different GSH concentrations was subjected to SVD analysis. This approach revealed that the main part of the optical transition was completed in less than 100 ms and accounts for more than 90% of the reaction, whereas 2-5% was due to a much slower phase detectable 1 s after mixing. Both the time dependence of basis spectra (from SVD analysis) and the experimental time courses recorded at 430 nm (where NBDHEX absorbs) and at about 320 nm (where the new spectral species absorbs) fit satisfactorily to a single step reaction mechanism (Fig. 7G). The apparent rate constants found by the fitting procedure were k 1 ϭ 12,500 M Ϫ1 s Ϫ1 and k Ϫ1 ϭ 152 s Ϫ1 . The overall dissociation constant (k Ϫ1 /k 1 ) of the new spectral species into GSH and NBDHEX was 12 ϫ 10 Ϫ3 M, a value that overlaps the one obtained from spectro-   (panel A, dotted line). Spectral data were also obtained at pH 6.5 and 25°C at variable GSH concentrations (from 5 ϫ 10 Ϫ3 to 80 ϫ 10 Ϫ3 M) (panel B). The dissociation constant of the NBDHEX⅐GSH complex was calculated from the absorbance change (⌬A) at 432 nm following the addition of GSH. The experimental data were fitted to Equation 1 where ⌬F and ⌬F max are replaced by ⌬A and ⌬A max (panel C).

TABLE II
Kinetic constants obtained from stopped flow experiments k 1 , k Ϫ1 , k 2 , and k Ϫ2 , were calculated from stopped flow experiments made as reported under "Experimental Procedures." The K D value was calculated from the k Ϫ1 /k 1 ratio for the reaction of NBDHEX with GSH and from the relation (k Ϫ1 /k 1 )(k Ϫ2 /k 2 ) for the reaction of NBDHEX with the GSH⅐GSTM2-2 and GSH⅐GSTA1-1 complexes.
Reaction of NBDHEX vs. photometric data at equilibrium (13 ϫ 10 Ϫ3 M). Reconstruction of the absolute spectra of the initial and final optical species are reported in Fig. 7 (panel L).
Stopped Flow Experiments in the Presence of GSTs-Stopped flow experiments revealed that a more complex interaction occurred when NBDHEX and GSH were mixed rapidly in the presence of GSTs. The reaction was studied at 4°C as at higher temperature it appeared too fast to be analyzed correctly. In the case of GSTM2-2 and GSTA1-1, time-dependent spectra were reconstructed from single wavelength observations by repetitively changing the wavelength following different reagent mixing steps (Fig. 7, B and C). Time-dependent spectra were not obtained for GSTP1-1 because most of the reaction was lost during the instrument dead time (1 ms). As shown in Fig. 7, differential spectra do not intersect in a single isosbestic point suggesting the involvement of at least two optical species during the reaction (Fig. 7, E and F). The whole data set obtained at different NBDHEX concentrations for each isoenzyme was subjected to SVD analysis. Both the time dependence of basis spectra and the experimental time courses recorded at 430 nm and at 350 nm (Fig. 7, H and I) fit well to the minimum two-step sequential reaction suggested in Scheme 1 that involves a first bimolecular reaction between NBDHEX and the GST⅐GSH complex to form a first intermediate that evolves toward a second optical species. The absolute spectra of the two species reconstructed from SVD analysis are shown in Fig. 7 (M and N). In both GSTM2-2 and GSTA1-1, the first transient species showed two peaks in the 300 -350 nm region and at 430 nm. This spectrum was similar to that obtained at high GSH concentrations in the absence of GSTs. The second optical species was characterized by a single peak at about 350 nm overlapping that of the complex observed at equilibrium after reaction of NBDHEX with GSH in the presence of either GSTM2-2 or GSTP1-1 (see Fig. 3) and with GSTA1-1 at low temperature (Fig. 4B). We emphasize that in GSTA1-1 the second optical species was not stabilized but was further converted into the more stable GS-NBD adduct. This event occurred very slowly (t1 ⁄2 ϭ 2 min at 8°C), and it could not be observed in the rapid kinetic conditions.
On the basis of previous reports (32,33) the first optical species could be identified as a complex with the sulfur atom of GSH bound at the C-6 of the benzoxadiazole ring. The second, more stable optical species may correspond to a complex with the sulfur atom of GSH bound at the C-4.
The kinetic constants for GSTA1-1 and GSTM2-2 found by the fitting procedure are listed in Table II. GSTM2-2 displayed FIG. 8. Docking simulations. Docking simulations of the complex bound to GSTA1-1 (panels A and D), to GSTP1-1 (panels B and E), and to GSTM2-2 (panels C and F) are shown. Analyses were performed as reported under "Experimental Procedure" starting from Protein Data Bank structure files (Protein Data Bank codes 1k3l, 1hnc, and 6gss for GSTA1-1, GSTM2-2, and GSTP1-1, respectively). In panels A, B, and C the active site is shown as a ribbon model, whereas the C-4 complex is shown as a "stick" representation. In addition, the solvent surface of the active site is represented in panels D, E, and F. The blue color indicates the positive electrostatic potential of the active site, whereas the red color indicates the negative potential. Black regions identify the protein surface clipped away for clarity. SCHEME 1 SCHEME 2 SCHEME 3. Determination of the dissociation constant (K A ) of the complex bound to GSTs at 25°C. K A was calculated by using the relation K A ϭ (K B ϫ K C )/K D where K B is the dissociation constant of NBDHEX from the GST⅐ complex obtained from binding experiments (5.3 ϫ 10 Ϫ6 , 0.9 ϫ 10 Ϫ6 , and 10 ϫ 10 Ϫ9 M for Alpha, Pi, and Mu GSTs, respectively); K C is the dissociation constant of GSH from the GST⅐GSH complex, and it was assumed to be 100 ϫ 10 Ϫ6 M for the Mu and Pi GSTs and about 70 ϫ 10 Ϫ6 M for the Alpha isoenzyme (34); and K D is the dissociation constant of GSH and NBDHEX from the complex formed spontaneously in solution (12 ϫ 10 Ϫ3 M). much lower k Ϫ1 and k Ϫ2 values when compared with GSTA1-1, whereas k 1 and k 2 were similar in these enzymes. This means that both the C-6 and the C-4 complexes were more firmly retained by the GSTM2-2 active site. The overall apparent K D value for the dissociation of the final C-4 complex from the active site, to give GST⅐GSH and NBDHEX (see Scheme 1), could be calculated from the relation K D ϭ (k Ϫ1 /k 1 )(k Ϫ2 /k 2 ). The values were 48 ϫ 10 Ϫ9 M for GSTM2-2 and 3.9 ϫ 10 Ϫ6 M for GSTA1-1. The K D value for GSTA1-1 corresponded to the one calculated by fluorometric experiments, whereas the value found for GSTM2-2 was about 10 times higher. This discrepancy may be due to the different temperatures used in these experiments (4 and 25°C). DISCUSSION The new NBD thioether derivatives described in this study are strong and selective inhibitors of GSTs, showing IC 50 values of about 10 ϫ 10 Ϫ9 M for GSTM2-2, 1 ϫ 10 Ϫ6 M for GSTP1-1, and 25 ϫ 10 Ϫ6 M for GSTA1-1, representative isoenzymes for the human Mu, Pi, and Alpha class GSTs. Moreover experiments performed on human cancer cell lines demonstrated a strong apoptogenic activity of NBDHEX, the most potent inhibitor of GSTP1-1 (see Fig. 1). The correspondence between the degree of GSTP1-1 inhibition (IC 50 ϭ 0.8 ϫ 10 Ϫ6 M) and of cytotoxicity on GLC4, K562, and HepG2 tumor cell lines (LC 50 ϭ 1.4 ϫ 10 Ϫ6 , 1.5 ϫ 10 Ϫ6 , and 2.9 ϫ 10 Ϫ6 M, respectively) suggests that NBDHEX recognizes GSTP1-1 as a preferential target even in living cells, and this interaction can be responsible for apoptosis. Recently we have published (15) a detailed investigation of the NBDHEX-mediated apoptosis on the leukemic cell lines showing that NBDHEX triggers tumor cell death through the dissociation of GSTP1-1 from the GSTP1-1⅐JNK complex. This event discloses interesting perspectives in cancer therapy and can be rationalized by assuming that NB-DHEX triggers a crucial perturbation in the GST structure. All fluorometric and kinetic data reported in the present study indicated that NBDHEX may perturb both the G-site and the H-site. It behaved like a suicide inhibitor: in fact, it was conjugated with GSH leading to a stable complex characterized by a single absorption band at about 350 nm. The formation of a complex or Meisenheimer complex has been observed previously in the active site of GSTs after reaction of 1,3,5-trinitrobenzene with GSH (30,31). The NBD derivatives are excellent electrophiles, readily forming complexes with many nucleophiles. A kinetically preferential attack of the nucleophile at the C-6 of the benzoxadiazole ring followed by a slow isomerization toward the more stable C-4 adduct (Scheme 2) also has been observed (32,33). Our spectroscopic and rapid kinetic data showed that GSTs strongly promoted the formation of the complex at the C-6 position that in turns isomerizes into the C-4 complex. Moreover in the active site of GSTA1-1, the C-4 complex rapidly evolved toward the GS-NBD product after release of 6-mercapto-1-hexanol. On the basis of all experimental data, we could calculate the dissociation constant (K A ), which measures the true affinity of each enzyme for the C-4 complex. By using the relation K A ϭ (K B ϫ K C )/K D (see the thermodynamic cycle reported in Scheme 3) we obtained K A values of 2.9 ϫ 10 Ϫ8 , 7.7 ϫ 10 Ϫ11 , and 6.9 ϫ 10 Ϫ9 M for GSTA1-1, GSTM2-2, and GSTP1-1, respectively. These values indicate that the complex between NBDHEX and GSH at the C-4 position represents one of the strongest inhibitors ever found for GSTs.
Docking simulation provided a rationale for this tight interaction. In fact, the C-4 complex bound to a similar binding site in all three isoenzymes (Fig. 8) with very favorable free energy (⌬G 0 between Ϫ6 and Ϫ14 kcal/mol). The GSH resides in the same G-site as found in the crystallographic structures of the GST⅐GSH complexes, whereas the benzoxadiazole ring fits tightly into a cavity limited by Tyr-108 (in GSTP1-1), Tyr-113 (in GSTM2-2), and Val-111 (in GSTA1-1) on one side and by the N terminus of helix 1 on the other. This cavity is present in all three proteins and is characterized by a positive electrostatic potential that favorably interacts with the net negative charge of the NO 2 Ϫ group. It is interesting to note that the negative nitro group is oriented differently in the three isoenzymes following the different localization of the positive potential region. In addition, the conformation of the hexyl chain is quite variable in each enzyme (see Fig. 8). The docking study did not explain the observed release of 6-mercapto-1-hexanol from the GSTA1-1 active site; this event is likely promoted by structural motions of the active site not predictable by our computational approach.
In conclusion, we believe that the design of non-GSH peptidomimetic molecules acting as suicide inhibitors for GSTs represents a novelty in the GST enzymology and opens interesting perspectives for cancer therapy.