Structures of class pi glutathione S-transferase from human placenta in complex with substrate, transition-state analogue and inhibitor

Research Article

1287

Structures of class pi glutathione S-transferase from human placenta in complex with substrate, transition-state analogue and inhibitor Lars Prade1*, Robert Huber1, T Herbert Manoharan2, William E Fahl2 and Wolfgang Reuter1 Background: Glutathione S-transferases (GSTs) are detoxification enzymes, found in all aerobic organisms, which catalyse the conjugation of glutathione with a wide range of hydrophobic electrophilic substrates, thereby protecting the cell from serious damage caused by electrophilic compounds. GSTs are classified into five distinct classes (alpha, mu, pi, sigma and theta) by their substrate specificity and primary structure. Human GSTs are of interest because tumour cells show increased levels of expression of single classes of GSTs, which leads to drug resistance. Structural differences between classes of GST can therefore be utilised to develop new anti-cancer drugs. Many mutational and structural studies have been carried out on the mu and alpha classes of GST to elucidate the reaction mechanism, whereas knowledge about the pi class is still limited. Results: We have solved the structures of the pi class GST hP1-1 in complex with its substrate, glutathione, a transition-state complex, the Meisenheimer complex, and an inhibitor, S-(p-bromobenzyl)-glutathione, and refined them to resolutions of 1.8 Å, 2.0 Å and 1.9 Å, respectively. All ligand molecules are welldefined in the electron density. In all three structures, an additionally bound N-morpholino-ethansulfonic acid molecule from the buffer solution was found.

Addresses: 1Max-Planck-Institut für Biochemie, Abt. Strukturforschung, Am Klopferspitz 18a, D-82152 Martinsried, Germany and 2McArdle Laboratory for Cancer Research, University of the Winconsin, Madison 53706-1599, USA. *Corresponding author. E-mail: [email protected] Key words: 3D structure, glutathione S-transferase, ligandin, reaction mechanism Received: 23 June 1997 Revisions requested: 7 August 1997 Revisions received: 19 August 1997 Accepted: 20 August 1997 Structure 15 October 1997, 5:1287–1295 http://biomednet.com/elecref/0969212600501287 © Current Biology Ltd ISSN 0969-2126

Conclusions: In the structure of the GST–glutathione complex, two conserved water molecules are observed, one of which hydrogen bonds directly to the sulphur atom of glutathione and the other forms hydrogen bonds with residues around the glutathione-binding site. These water molecules are absent from the structure of the Meisenheimer complex bound to GST, implicating that deprotonation of the cysteine occurs during formation of the ternary complex which involves expulsion of the inner bound water molecule. The comparison of our structures with known mu class GST structures show differences in the location of the electrophile-binding site (H-site), explaining the different substrate specificities of the two classes. Fluorescence measurements are in agreement with the position of the N-morpholino-ethansulfonic acid, close to Trp28, identifying a possible ligandin-substrate binding site.

Introduction The glutathione S-transferases (GSTs, EC 2.5.1.18) catalyse the conjugation of glutathione (GSH, γ-Glu–Cys–Gly) with a broad variety of electrophilic substrates. They act as phase II detoxification enzymes in mammalian cells. The products of this conjugation are excreted via the urinary tract. Cytosolic GSTs are classified into five distinct classes (alpha, mu, pi, sigma and theta) by their substrate specificity and primary structure [1–3]. They occur as homo- or heterodimers with a molecular mass of about 50 kDa. Every subunit has its own active site consisting of a GSH-binding site (G-site) and a xenobiotic binding site (H-site). The G-site is very specific for GSH, whereas

the H-site has only low specificity to allow the acceptance of a wide range of electrophiles. For the pi, mu, alpha [4] and sigma [5] classes of GST, a tyrosine residue at the N terminus was identified as the catalytic residue, whereas for the theta class it is a serine residue [6,7]. The tyrosine residue helps to stabilize the thiolate anion of glutathione. Intensive structural and mutational analyses of the reaction mechanism have been carried out for the alpha and the mu class GSTs (recently reviewed in [8]). It was proposed that the alpha and mu classes have different mechanisms of deprotonation of glutathione. So far, for the pi class of GST, not enough data are available to give a clear view of its deprotonation mechanism.

1288

Structure 1997, Vol 5 No 10

Spectroscopic studies have shown that GSTs with a higher specific activity toward 1-chloro-2,4-dinitrobenzene (CDNB) stabilize the Meisenheimer complex, a complex of glutathione and trinitrobenzene, better than GSTs with a lower specific activity for CDNB [9]. Structures of the product complex 1-(S-glutathionyl)-2,4-dinitrobenzene and of the transition state analogue complex 1-(S-glutathionyl)-2,4,6-trinitrocyclohexadienate with class mu GST led to the proposal that the aromatic ring has to move from the ‘in’ to the ‘out’ position for the conjugation reaction to occur [10]. No transition state analogue structure for other GST classes have been determined so far. For reviews see [4,8,11]. In addition to their enzymatic function, GSTs act as socalled ligandins — they bind a wide range of hydrophobic compounds without displaying any enzymatic activity. Despite numerous attempts to elucidate the binding site of these hydrophobic compounds and the physiological role of ligandins, both remained unknown [12]. Recently, the structures of GST from Schistosoma japonica in complex with the antischistosomal drug praziquantel [13] and of class sigma GST in complex with S-(p-iodobenzyl)-glutathione [14] were solved. They show a binding site at the dimer interface, which was referred to as a potential ligandin-substrate binding site. But so far, no structure in complex with a typical ligandin substrate, such as haemin, bilirubin and cholic acid, has been solved. We have solved the structures of GST hP1-1 (class pi formed by two monomers numbered 1), in complex with its substrate, glutathione, a transition-state complex, the Meisenheimer complex, and an inhibitor, S-(p-bromobenzyl)-glutathione, and refined them to resolutions of 1.8 Å, 2.0 Å and 1.9 Å, respectively. These results presented here allow a reaction mechanism for pi class GSTs to be proposed and they explain differences in substrate specificity of the different classes. Further, it is demonstrated by fluorescence measurements that the N-morpholino-ethansulfonic (MES) acid binding site is the ligandin-substrate binding site.

Results and discussion From our high resolution structure of human glutathione S-transferase (GST hP1-1) with glutathione bound at pH 6.5 presented here, a clear picture of the activated state of the enzyme emerges. This structure represents the protein at the pH at which most kinetic, fluorescence and stability measurements have been done. The high resolution allows determination of the water molecule positions with high accuracy. The formation of a complex between GST and the Meisenheimer complex, a transition-state complex shows that the enzyme is active in the crystal. The structures of GST in the glutathione-bound state and in the transition state, represented by the Meisenheimer complex bound to GST, allow a detailed

reaction mechanism of the nucleophilic substitution at an aromatic ring catalysed by class pi GST to be proposed. Two crystal forms were obtained under identical conditions, with space groups P21 and C2. Both forms are closely related. There are no changes in their secondary structure, and the root mean square deviations (rmsd) of the Cα atoms between the structures are within the range of the deviations seen between each monomer of one structure. If the second dimer in the P21 form is rotated 180° around the x axes with a small translational rearrangement, the packing of the C2 form is obtained. The bound ligand does not affect the crystal packing, as it is not involved in any crystal contacts. Similar effects with two closely related crystal forms growing under the same condition were also observed for alpha class GST [15]. GST–GSH complex

The glutathione molecule is well-ordered and the density is unambiguously interpretable. In comparison to the GST–S-hexylglutathione complex [16], no changes in the positions of residues participating in the G-site are seen in the GST–GSH complex (Figure 1). Only the position of the glutathione is slightly changed. The position of the γ-glutamyl residue is identical, whereas the positions of the cysteine and the glycine have slightly moved. These changes, however, have no influence on the hydrogen-bonding network, which is responsible for the binding of glutathione, and they are not observed in the high resolution structure of porcine GST pP1-1 [17] which has glutathionesulfonate bound. One interesting difference in our structure is that the sulfur atom now points into the direction of Tyr7. With regard to their number and their positions, the network of water molecules has changed, including their hydrogen bonds to the catalytic Tyr7 and to the sulphur atom of the glutathione. In addition to the hydrogen bond between the sulphur atom of the glutathione and the OH oxygen of Tyr7, there is a water molecule that forms hydrogen bonds to these two atoms. This water molecule has two additional hydrogen bonds to a second water molecule and to the mainchain nitrogen of Arg13. The second water molecule is hydrogen bonded to the OH oxygen of Tyr108, which has further hydrogen contacts to the mainchain nitrogen of Gly205 and to the ND2 nitrogen of Asn204. The OD2 oxygen of Asn204 is hydrogen bonded to the OH oxygen of Tyr103. All these hydrogen bonds have distances in the range of 3.0–3.5 Å, and they are seen in all four monomers. The otherwise lack of an appropriate countercharge suggests that the sulphur atom of the glutathione is protonated when bound to the pi class GSTs. In contrast to mu and alpha class GSTs, in class pi the network of hydrogen bonds may be responsible for the activation of the sulphur of glutathione. For the alpha class, it was suggested that the carboxyl groups of the glutathione are involved in the deprotonation of

Research Article Glutathione S-transferase Prade et al.

1289

Figure 1 Stereo representation of the active site of glutathione S-transferase (GST) class pi from human placenta in complex with glutathione (purple). All residues (with atoms in standard colours) from one monomer comprising the active site are shown and labelled. (a) View from the α2 helix into the H-site, showing two water molecules (blue spheres) with their hydrogen bonds (red dotted lines). (b) View from the other monomer to the complete active site. Overlay with the S-hexylglutathione (green) from the structure of GST in complex with S-hexylglutathione [16]. The final |2Fo–Fc| map from the structure of GST in complex with glutathione is contoured at 1σ and shown in blue.

the sulphur [18]. In class mu, a hydrogen bond between the NH amide of Leu12 and the OG oxygen of Tyr6 and a second sphere electrostatic interaction between the OH oxygen from Thr13 and the π-electron cloud of Tyr6 are thought to be the main interactions responsible for the activation of the sulfur of glutathione [19]. The NH amide of Arg13 is also hydrogen bonded to Tyr7 in pi class, but instead of the Thr13, Cys14 points into the direction of the ring of Tyr7. GST–S-(p-bromobenzyl)-glutathione complex

The glutathione moiety of the S-(p-bromobenzyl)-glutathione in complex with GST (Figure 2) occupies an identical position as in the GST–glutathione complex. The p-bromobenzyl part is sandwiched between the aromatic rings of Phe8 and Tyr108. The distances between these aromatic rings are in the range of 3.5–4 Å. The p-bromobenzyl moiety is held in place by Val10 and Val35. Val10 is in van der Waals contact, in the range of 3.6–4.0 Å, with the carbons of the benzyl aromatic rings, whereas Val35 contacts the bromine atom at a distance of 3.9 Å. These contacts within the H-site coincide with those found in the structure of mouse pi class GST in

complex with S-(p-nitrobenzyl)-glutathione [20] and with those in the structure of human pi class GST in complex with ethacrynic acid and its glutathione conjugate [21]. Out of all the residues that participate in the G-site and H-site, Val35, the residue with the weakest interaction, is the only one that is not conserved throughout pi class GSTs from different mammalians. Meisenheimer complex bound to GST

The complex of GSH with 1,3,5-trinitrobenzene (the Meisenheimer complex) represents a valid model for the transition state σ-complex of the nucleophilic substitution at the aromatic ring. This reaction is typically catalysed by GSTs. Figure 3 shows the binding of the Meisenheimer complex within the protein. The aromatic ring is sandwiched between the aromatic rings of Phe8 and Tyr108. Val10 and Val35 are localized at the sides as in the complex of GST with S-(p-bromobenzyl)-glutathione. Ile104 sits at the other end at a distance of 5 Å from the trinitrobenzene moiety. The nitro groups form several hydrogen bonds with the protein. One ortho-nitro group is hydrogen

1290

Structure 1997, Vol 5 No 10

Figure 2 Stereo representation of the active site of GST hP1-1 in complex with S-(pbromobenzyl)-glutathione (cyan). The |2Fo–Fc| map is derived from this structure and contoured at 1.0σ (shown in blue). All residues from one monomer participating in the active site are shown, but only the residues from the H-site are labelled. The S-(p-nitrobenzyl)-glutathione moiety, from the structure of GST mP1-1 in complex with it [20], is superimposed and shown in green.

bonded to Tyr7 and Tyr108. The geometry for this group was derived from the small molecule structure of 1-methoxy-2,4,6-trinitrocyclohexadienate, in which the nitro groups are in plane. The nitro groups may twist to improve contacts with the protein, as it has been observed in the structure of mu class GST [10]. In addition, there are some contacts from Gly205 to the para-nitro group. The second ortho-nitro group, directed towards the dimer interface, only has hydrogen bond contacts with some water molecules and is within hydrogen-bond distance of the oxygen of the γ-glutamylgroup of the glutathione. The structure of GST bound to the Meisenheimer complex differs largely from the known structure of mu class GST [10]. As seen in Figure 3, the aromatic rings stretch out in opposite directions. Although the ortho-nitro groups of the ligand make similar contacts with the protein, all other protein–ligand contacts are totally different between mu and pi classes. The H-subsite is placed at

different positions in pi and mu class, explaining the different substrate specificity of these two classes. MES-binding site

The binding of MES, as seen in both P21 and C2 crystal forms, is illustrated in Figure 4. There are van der Waals contacts to Ala22, Trp28, Glu30 and Phe192 and a salt bridge between the carboxyl group of Glu197 and the nitrogen of MES. Hydrogen bonds, mediated by a water molecule, are built between MES and the mainchain atoms of residues 188 and 192. The proximity of MES to the crystal contacts is remarkable, although these contacts are not consistent for all sites. At two sites of the P21 form Lys188 is involved in crystal contacts, whereas it is Lys29 that mediates all other contacts. Due to the close relation of the two crystal forms, all contacts of the C2 form are also seen in the P21 form. The MES molecule has in all cases some hydrogen bonds to other dimers in the crystal

Figure 3 Stereo representation of the Meisenheimer complex in the active site of GST hP1-1 shown in red. The final |2Fo–Fc| map is contoured at 0.7σ, shown in blue. The Meisenheimer complex as bound in a µ class GST [10] is superimposed onto the atoms of the cysteine residue and the γ-glutamyl moiety of the glutathione and shown in green. All residues in the active site of one monomer are shown.

Research Article Glutathione S-transferase Prade et al.

1291

Figure 4 View on the MES-binding site as a stereo representation. All important residues from one monomer are shown and labelled. As seen in all six binding sites of the two crystal forms, the MES molecule is shown in red and a conserved water molecule in cyan. The |2Fo–Fc| map is shown in blue and contoured at 1.0σ.

The covalent labelling of pi class GST with fatty acid conjugated with Woodward’s reagent K shows a labelling at the sequence between residues 141 and 188 [25]. At

residue 185, the helix αH, which is involved in the MES binding, starts. Analysis of porcine pi class GST in complex with ANS and bromosulfophthalein revealed that both these non-substrate ligands bind to the same site [23]. Trp28, Trp38, Cys45 or a group of residues throughout the protein molecule, positioned close to these tyrosines, were suggested as the non-substrate binding site. Furthermore, a microstructural change caused by the binding was noted that influences the structure of active site. As illustrated in Figure 6, the binding of bigger substrates Figure 5

Relative fluorescence

mediated by some water molecules. All these interactions are not consistently seen in the six sites of crystal contacts, found in both crystal forms. They were therefore considered as not biologically important. The observation that buffer molecules bind to protein in crystals, but do not bind proteins in normal solution, is not unusual. Nevertheless, the proximity of Trp28 to the MES molecule made it possible to detect binding of the MES molecule in solution by fluorescence measurements. As seen in Figure 5, binding of MES induces a shift to UV in the fluorescence spectra. There are two tryptophan residues in class pi GST, Trp38, which is involved in glutathione binding, and Trp28, which is localized at the beginning of β strand 2 and is close to the first part of the C-terminal helix αH. Earlier fluorescence studies on ligandin binding demonstrated that 8-anilino-1-naphthalene sulfonate (ANS) binds to the ligandin site. ANS induces an UV shift in phosphate buffer [22] but not in MES buffer [23], indicating that MES influences the binding of ANS. As seen in Figure 5, the fluorescence is influenced by cholic acid only in the phosphate buffer, but not in the presence of MES, implicating that MES also influences the binding of cholic acid. MES has no influence on the activity of GST with CDNB as substrate as seen in kinetic tests (data not shown). As cholic acid is a noncompetitive inhibitor of GST [24], MES should have an influence on the inhibition by cholic acid. Unfortunately, this could not be tested as cholic acid, MES, CDNB and glutathione seem to react. Attempts to soak crystals to exchange the MES against cholic acid led to the destruction of the crystals possibly caused by a rupture of the crystal contacts by binding of the cholic acid in the MESbinding site.

300

320

340

360

Wavelength

380

400 Structure

Fluorescence emission spectra for GST hP1-1, in purple is the spectra for the native, in red with 10 mM cholic acid, in green with 100 mM MES and in black with 10 mM cholic acid and 100mM MES. All measurements were done at 77K with a KM 25 fluorescence spectrometer from Kontron (Germany). Ecitation and emission band passes were set to 2 nm. Ecitation wavelength was set to 280 nm.

1292

Structure 1997, Vol 5 No 10

Figure 6

Surface representation of one monomer of GST hP1-1. The C-terminal helix and it’s surface is shown in red. The MES molecule is in yellow, the trinitrophenyl moiety of the Meisenheimer complex is in green.

like cholic acid at the MES site may induce a structural change at the C-terminal helix leading to a disturbance in the active site architecture, where the C terminus is involved, resulting in a loss of activity. This hypothesis has to be examined further. Comparison of the three GST complexes

In Figure 7, an overlay of all three structures described above is shown. There are no significant differences in the protein. The glutathione molecule is in an identical position. The p-bromobenzyl group and the trinitrophenyl group are in the same plane between Tyr108 and Phe8, but their orientations are different; this is caused by the

ortho-nitro group, which is hydrogen bonded to Tyr108 and Tyr7, such that the trinitrobenzene group is turned to the outside. Hence, the contacts of the para-nitro group are different to the contacts of the para-bromo group. This emphasises the importance of the ortho-nitro group for a correct positioning of the substrate. It also explains the tenfold higher specific activity for CDNB over that for 3,4dichloro-1-nitrobenzene [26]. The differences in binding of the p-bromobenzyl group and the trinitrophenyl group show the limitation in information for substrate specificity arising from inhibitor structures. The H-site can now be described as a pocket with the side walls formed by Phe8 and Tyr108. The other two sides are formed by Val35 and Ile104. The bottom is built by Val10 and Gly205. The entrance of this pocket points towards the wide cleft between the two monomers. This H-site is conserved throughout all mammalian pi class GSTs. Only Val35 and Ile104 show conservative changes. These two residues are farthest away from the substrate; nevertheless, their importance for substrate recognition is shown by the changed enzymatic properties of a naturally occurring isoform that has valine in position 104 rather than isoleucine [27]. Reaction mechanism

As seen in the structure of the GST–glutathione complex, the sulphur atom of the glutathione is hydrogen bonded to Tyr7 OH and to a water molecule. In the structure of GST bound to the Meisenheimer complex, the trinitrobenzene moiety occupies the place of this water molecule, implicating that the second hydrophobic substrate pushes the water molecule out of the active site. The sulphur is no longer solvent accessible, thus increasing its reactivity to attack the second substrate, which subsequently forms a covalent bond after proton transfer to either Tyr7 or the water molecule acting as the general base. The water molecule seems the more appropriate

Figure 7 Overlay of the active site of all three structures described in this paper. View is from the outer solvent area. Shown in green is the GST–GSH complex, in blue the GST–S(p-bromobenzyl)-glutathione complex and in red the Meisenheimer complex bound to GST.

Research Article Glutathione S-transferase Prade et al.

base as Tyr7 OH acts as hydrogen acceptor of NH Arg13. Tyr7 could thus stabilize, through a hydrogen bond, the cysteine anion. From the water, the proton may be transferred to further partners via the hydrogen-bonding network described above. Thus, the final acceptor of this proton would be the hydrogen-bond network spanning from the C terminus to the domain II. The contribution of every single residue of this network cannot be estimated from the structure. A network instead of one single residue functioning as a proton acceptor may provide a wider range of pH for GST activity.

Biological implications Glutathione S-transferases (GSTs) catalyse the conjugation of glutathione with a broad range of electrophiles, and so are involved in the protection of the cell against serious damage caused by electrophilic compounds. In addition to their enzymatic functions, GSTs are ligandins, that is they bind a wide range of hydrophobic compounds without displaying enzymatic activity. GSTs bind a broad variety of electrophilic substrates, with the different classes of GSTs showing different substrate specificities. As the GST classes are differently expressed depending on the tissue type, differences in substrate specificity between the classes have important physiological relevance. Single classes of GSTs are shown to be expressed at raised levels in different types of tumors, leading to resistance towards various anti-cancer drugs [28–30]. Differences in the structures of the specific classes of GST can therefore be used for the development of new anti-cancer drugs.

1293

We have shown that the S-benzylglutathione inhibitor binds in the same way as the transition-state analogue in contrast to S-alkylglutathione inhibitors. This explains the better inhibition constants (smaller Ki) for the S-benzylglutathione inhibitors compared to the S-alkylglutathione inhibitors [26]. The elucidation of the binding of MES at a distinct site close to Trp28 suggests a possible ligandin binding site. The proximity to Trp28 makes it possible to use MES as a fluorescence probe for this binding site. This opens a wide field of possibilities for biochemical investigations on the ligandin binding site.

Materials and methods S-(p-bromobenzyl)-glutathione was synthesised according to the method already described [31]. All other chemicals were standard commercial products used without further purification.

Protein purification and crystallisation The recombinant protein was expressed and purified as previously described [32]. Crystals were grown by vapour diffusion using PEG 8000 as precipitant. 4 µl of a 20 mg/ml protein solution were mixed with 2 µl of reservoir liquor containing 30% PEG, 100 mM MES-NaOH, pH 6.5, and 200 mM NaAc. The ligand was added to the protein solution to give a final concentration of 5 mM. Crystals grew in one to two weeks at 20°C. All fluorescence measurements were performed in solutions buffered with 100 mM sodium phosphate, pH 6.5, containing 2 µM protein and were done at 77K with a KM25 Spectrometer from KONTRON (Germany). Ecitation and emission band passes were set to 2 nm. The ecitation wavelength was 280 nm and detection was done for wavelengths from 390 to 280 nm.

Data collection

We have solved the structures of the human pi class GST in binary complex with its substrate, glutathione, and in ternary complex with a transition-state complex, the Meisenheimer complex, and an inhibitor, S-(p-bromobenzyl)-glutathione. The structures described here showed that the differences in the substrate specificities between mu and pi class are not only caused by different residues in the electrophile-binding site (H-site), but also by a different location of the H-site. A number of investigations of the degree of deprotonation of the sulphur of glutathione in the binary complex between glutathione and GST were undertaken. The structure of the binary complex at pH 6.5 shows a protonated state of the sulphur. We suggest that the deprotonation of the cysteine in the G-site occurs upon entry of the second substrate and expulsion of the inner bound water. This may explain the low pKa derived from kinetic data of the cysteine in the complex compared with that of the free protein in solution. In the ternary complex, the sulphur is in a hydrophobic environment with no access of the solvent; it therefore has increased reactivity to attack the second substrate.

Diffraction data were collected using an imaging plate scanner (MARresearch). Graphite monochromatised CuKα radiation from a RU200 rotating anode generator (Rigaku, Tokyo, Japan) operating at 5.4 kW was used. The glutathione (GSH) and S-(p-bromobenzyl)-glutathione containing crystals were measured at 16°C. For the Meisenheimer complex formation, the GSH crystals were soaked in the crystallisation buffer containing additional 10% glycerol and 10 mM trinitrobenzene. When the crystal became red, it was frozen to –170°C with a cryosystem (KGW, Karlsruhe, Germany) and measured. Reflection data were processed using the MOSFLM package [33] and subsequently the structure factors were scaled, reduced and truncated to intensities using the CCP4 program suite [34].

Refinement As the initial model, the structure of hGST P1-1 in complex with Shexylglutathione [16] was taken from Brookhaven database. Four monomers were positioned with the program AMoRe [35] for the GST–GSH complex structure. After one round of simulated annealing and several steps of restrained least-squares refinement on positional parameters, B factors without any noncrystallographic symmetry (ncs) constraints using the program X-PLOR [36] and the parameters developed by Engh and Huber [37], the R factor dropped below 30%. In the |Fo–Fc| map, a clear density for the glutathione moieties appeared. The glutathione and several water molecules were added. Model building was performed on Silicon Graphics workstations using the program O [38]. Four MES molecules were identified and added in further steps of refinement. The refinement was stopped when the further addition of solvent molecules brought no further drop in the free R factor.

1294

Structure 1997, Vol 5 No 10

Table 1 Data processing statistics. Ligand

GSH

Meisenheimer complex

p-bromo-benzyl-glutathione

Space group

P21

P21

C2

90.29, 74.67, 69.50

89.59, 72.57, 69.34

80.66, 90.07, 69.57

β = 90.07°

β = 90.11°

β = 99.19°

Resolution range (Å)

18.602–1.800

30.269–2.030

12.875–1.939

No. of observations

164,829

100,891

91,365

No. of unique reflections

80,112

52,986

34,247

93.7

90.9

94.3

Cell constants (a, b, c) (Å)

Completeness (%) Rmerge* (%)

7.5

6.7

7.3

Percentage reflections in highest resolution shell (%)

38.8

19.5

24.2

1.85–1.80

2.08–2.03

1.99–1.94

Bin resolution (Å) *Rmerge = Σ | Iobs – < I > | / Σ Iobs.

Due to the shrinking of the cell constants caused by the freezing of the crystals, it was necessary to restart for the structure of the Meisenheimer complex with the positioning of four monomers with the program AMoRe [35]. As a starting model, the GST structure in complex with GSH, without water, GSH and MES molecules, was used. After an initial refinement, clear electron density for the glutathione appeared, but not for the trinitrobenzene group. Glutathione, MES and several water molecules were added. Refinement was done analogous to that used for the GST–GSH structure. After another

refinement, no clear electron density appeared for the trinitrobenzene. The |2Fo–Fc| map was then fourfold averaged using the program AVE [39]. The averaged map clearly showed the trinitrobenzene group, which was built in and the refinement was stopped when the free R factor was at a minimum. As a starting model for the GST–S-(p-bromobenzyl)-glutathione complex structure, the GST–GSH complex structure without water, MES and GSH molecules was used. It was placed with AMoRe [35]

Table 2. Final refinement statistics. Ligand Resolution range (Å) Used reflections

GSH

Meisenheimer complex

p-bromo-benzyl-glutathione

8.00–1.80

7.00–2.00

7.00–1.94

79,175

52,602

33,587

Total number of atoms (excluding H)

7031

7262

3603

Number of active protein atoms

6552

6524

3276

Number of inhibitor atoms

80

140

56

Number of MES atoms

48

48

24

Number of solvent atoms

351

550

247

R factor of final model

20.6%

20.1%

19.4%

Rfree of final model*

26.1%

28.3%

25.6%

Standard deviation from ideal values Bond length (Å) Bond angles (°)

0.009 1.39

0.009 1.58

0.011 1.67

Standard deviation in B factors for bonded atoms (Ų)

2.53

2.18

2.52

Temperature factors (Ų) All atoms Mainchain atoms Sidechain atoms Inhibitor atoms MES atoms Solvent molecules

22.5 20.5 23.7 27.0 42.3 28.6

15.9 15.4 14.9 31.6 48.5 17.7

21.3 18.8 21.6 22.4 43.7 33.7

*5% of reflections were used to calculate Rfree.

Research Article Glutathione S-transferase Prade et al.

and then refined with X-PLOR [36]. The inhibitor and the MES molecules were seen in the difference map and built in. The rest of the refinement was handled as before. The data processing statistics are shown in Table 1 and final statistics in Table 2.

Accession numbers The coordinates have been deposited with the Brookhaven databank with the entry numbers 1aqv for S-(p-bromobenzyl)-glutathione, 1aqw for glutathione, and 1aqx for the Meisenheimer complex, all bound to hGST P1-1.

Acknowledgement We thank MA Schräml for his help with the protein crystallisation.

References 1. Mannervik, B. et al., & Jörnvall, H. (1985). Identification of three classes of cytosolic glutathione S-transferases common to several mammalian species: correlation between structural data and enzymatic properties. Proc. Natl. Acad. Sci. 82, 7202–7206. 2. Meyer, D.J., Coles, B., Pemble, S.E., Gilmore, K.S., Frazer, G.M. & Ketterer, B. (1991). Theta, a new class of glutathione S-transferase purified from rat and man. Biochem. J. 274, 409–414. 3. Buetler, T.M. & Eaton, D.L. (1992). Glutathione S-transferases: amino acid sequence comparison, classification and phylogenetic relationship. Environ. Carcino. Ecotox. Rev. C10, 181–203. 4. Dirr, H., Reinemer, P. & Huber, R. (1994). X-ray crystal structures of cytosolic glutathione S-transferases. Eur. J. Biochem. 220, 645–661. 5. Ji, X., et al., & Gilliland, G. (1995). 3-Dimensional structure, catalytic properties, and evolution of a sigma-class glutathione transferase from squid, a progenitor of the lens S-crystallins of cephalopods. Biochemistry 34, 5317–5328. 6. Board, P., Coggan, M., Wilce, M. & Parker, M. (1995). Evidence for an essential serine residue in the active-site of the theta-class glutathione transferases. Biochem. J. 311, 247–250. 7. Reinemer, P., et al., & Bieseler, B. (1996). Three-dimensional structure of glutathione S-transferase from Arabidopsis thaliana at 2.2 Å resolution: structural characterization of herbicide-conjugating plant glutathione S-transferases and a novel active site architecture. J. Mol. Biol. 255, 289–309. 8. Armstrong, R.N. (1997). Structure, catalytic mechanism, and evolution of the glutathione transferases. Chem. Res. Toxicol. 10, 2–18. 9. Bico, P., Chen, C., Jones, M., Erhardt, J. & Dirr, H. (1994). Class-pi glutathione-S-transferase – Meisenheimer complex formation. Biochem. Mol. Biol. Int. 33, 887–892. 10. Ji, X., Armstrong, R. & Gilliland, G. (1993). Snapshots along the reaction coordinate of an s(n)ar reaction catalyzed by glutathione transferase. Biochemistry 32, 12949–12954. 11. Wilce, M.C.J. & Parker, M.W. (1994). Structure and function of glutathione S-transferases. Biochim. Biophys. Acta 1205, 1–18. 12. Litowski, I. (1993) Glutathione S-transferase: intracellular binding, detoxification and adaptive response. In Hepatic Transport and Bile Secretion: Physiology and Pathophysiology. (Tavolini, N. & Beck, P.D., eds), pp. 397–405, Raven Press, New York. 13. McTigue, M., Williams, D. & Tainer, J. (1995). Crystal structures of a schistosomal drug and vaccine target — glutathione S-transferase from Schistosoma japonica and its complex with the leading antischistosomal drug praziquantel. J. Mol. Biol. 246, 21–27. 14. Ji, X., von Rosenvinge, E.C., Johnsson, W.W., Armstrong, R.N. & Gilliland, G.L. (1996). Location of a potential transport binding site in a sigma class glutathione transferase by x-ray crystallography. Proc. Natl. Acad. Sci. 93, 8208–8213. 15. Cameron, A., et al., & Jones, T. (1995). Structural analysis of human alpha-class glutathione transferase a1-1 in the apo-form and in complexes with ethacrynic acid and its glutathione conjugate. Structure 3, 717–727. 16. Reinemer, P., et al., & Parker, M.W. (1992). The three-dimensional structure of class pi glutathione S-transferase from human placenta in complex with S-hexylglutathione at 2.8 Å resolution. J. Mol. Biol. 227, 214–226. 17. Dirr, H., Reinemer, P. & Huber, R. (1994). Refined crystal structure of porcine class-pi glutathione-s-transferase (pgst p1-1) at 2.1 angstrom resolution. J. Mol. Biol. 243, 72–92. 18. Widersten, M., Björnestedt, R. & Mannervik, B. (1996). Involvement of the carboxyl groups of glutathione in the catalytic mechanism of human glutathione transferase A1-1. Biochemistry 35, 7731–7742.

1295

19. Xiao, G., et al., & Armstrong, R.N. (1996). First-sphere and secondsphere electrostatic effects in the active site of a class mu glutathione transferase. Biochemistry 35, 4753–4765. 20. Garciasaez, I., Parraga, A., Phillips, M., Mantle, T. & Coll, M. (1994). Molecular structure at 1.8-angstrom of mouse liver class pi glutathione S-transferase complexed with S-(p-nitrobenzyl) glutathione and other inhibitors. J. Mol. Biol. 237, 298–314. 21. Oakley, A.J., Rossjohn, J., Lo Bello, M., Caccuri, A.M., Federici, G. & Parker, M.W. (1997). The three-dimensional structure of human pi class glutathione transferase P1-1 in complex with the inhibitor ethacrynic acid and its glutathione conjugate. Biochemistry 36, 576–585. 22. Nishihira, J., Ishibashi, T., Sakai, M., Nishi, S., Kumazaki, T. & Hatanaka, Y. (1992). Circular dichroic evidence for regulation of enzymatic activity by nonsubstrate hydrophobic ligand on glutathione Stransferase P. Biochem. Biophys. Res. Comm. 189, 1243–1251. 23. Bico, P., Erhardt, J., Kaplan, W. & Dirr, H. (1995). Porcine class-pi glutathione S-transferase – anionic ligand-binding and conformationalanalysis. Biochem. Biophys. Acta 1247, 225–230. 24. Vessey, D.A. & Zakim, D. (1981). Inhibition of glutathione Stransferase by bile acids. Biochem. J. 197, 321–325. 25. Nishihira, J., Ishibashi, T., Sakai, M., Nishi, S., Kondo, H.A. & Makita, A. (1992). Identification of the fatty acid binding site on glutathione Stransferase. Biochem. Biophys. Res. Comm. 189, 197–205. 26. Mannervik, B. & Danielson, U.H. (1988). Glutathione transferases — structure and catalytic activity. CRC Critic. Rev. Biochem. 23, 283–337. 27. Zimniak, P., et al., & Awasthi, Y. (1994). Naturally occurring human glutathione S-transferase gstp1-1 isoforms with isoleucine and valine in position-104 differ in enzymatic properties. Eur. J. Biochem. 224, 893–899. 28. Coles, B. & Ketterer, B. (1990). The role of glutathione and glutathione transferases in chemical carcinogenesis. Critic. Rev. Biochem. Mol. Biol. 25, 47–70. 29. Tsuchida, S. & Sato, K. (1992). Glutathione transferases and cancer. Critic. Rev. Biochem. Mol. Biol. 27, 337–384. 30. Hayes, J.D. & Pulford, D.J. (1995). The glutathione S-transferase supergene family: regulation of GST and the contribution of the isoenzymes to cancer chemoprotection and drug resistance. Critic. Rev. Biochem. Mol. Biol. 30, 445–600. 31. Vince, R., Daluge, S. & Wadd, W.B. (1971). Studies on the inhibition of glyoxalase I by S-substituted glutathiones. J. Med. Chem. 14, 402–404. 32. Manoharan, T.H., Gulick, A.M., Puchalski, R.B., Servais, A.L. & Fahl, W.E. (1992). Structural studies on human glutathione S-transferase Pi substitution mutations to determine amino acids necessary for binding glutathione. J. Biol. Chem. 267, 18940–18945. 33. Leslie, A.G.W. (1994). Mosflm user guide. MRC Labotary of Molecular Biology, Cambridge. 34. Collaborative Computational Project No. 4 (1994). The CCP4 suite: programs for protein crystallography. Acta Cryst. D 50, 760–763. 35. Navaza, J. (1994). AMoRe: an automated package for molecular replacement. Acta Cryst. A 50, 157–163. 36. Brünger, A.T. (1992). X-PLOR Manual (Version 3.1). Yale University Press, New Haven. 37. Engh, R.A. & Huber, R. (1991). Accurate bond and angle parameters for X-ray protein structure refinement. Acta Cryst. A47, 392–400. 38. Jones, T.A., Zou, J.Y. & Kjeelgaard, M. (1991). Improved methods for building protein models in electron-density maps and location of errors in these models. Acta Cryst. A 47, 110–119. 39. Kleywegt, G.J. & Jones, T.A. (1993). Masks made easy. ESF/CCP4 Newsletter 28, 56–59.