A conserved N-capping motif contributes significantly to the stabilization and dynamics of the C-terminal region of class Alpha glutathione S-transferases.

Helix 9, the major structural element in the C-terminal region of class Alpha glutathione transferases, forms part of the active site of these enzymes where its dynamic properties modulate both catalytic and ligandin functions. A conserved aspartic acid N-capping motif for helix 9 was identified by sequence alignments of the C-terminal regions of class Alpha glutathione S-transferases (GSTs) and an analysis by the helix-coil algorithm AGADIR. The contribution of the N-capping motif to the stability and dynamics of the region was investigated by replacing the N-cap residue Asp-209 with a glycine in human glutathione S-transferase A1-1 (hGST A1-1) and in a peptide corresponding to its C-terminal region. Far-UV circular dichroism and AGADIR analyses indicate that, in the absence of tertiary interactions, the wild-type peptide displays a low intrinsic tendency to form a helix and that this tendency is reduced significantly by the Asp-to-Gly mutation. Disruption of the N-capping motif of helix 9 in hGST A1-1 alters the conformational dynamics of the C-terminal region and, consequently, the features of the H-site to which hydrophobic substrates (e.g. 1-chloro-2,4-dinitrobenzene (CDNB)) and nonsubstrates (e.g. 8-anilino-1-naphthalene sulfonate (ANS)) bind. Isothermal calorimetric and fluorescence data for complex formation between ANS and protein suggest that the D209G-induced perturbation in the C-terminal region prevents normal ligand-induced localization of the region at the active site, resulting in a less hydrophobic and more solvent-exposed H-site. Therefore, the catalytic efficiency of the enzyme with CDNB is diminished due to a lowered affinity for the electrophilic substrate and a lower stabilization of the transition state.

control of gene expression (1). These proteins, ubiquitous in aerobes, form a superfamily of species-independent classes that, except for the Kappa GST (2), share a common fold. Typically, the canonical GSTs are soluble, dimeric proteins with each subunit possessing a thioredoxin-like domain 1 fused to an all-␣-helical domain 2 (3). The Kappa GST, however, is more closely related to the protein disulfide bond isomerase, dsbA, in that domain 2 is inserted in domain 1 (2). Nevertheless, the active site of GSTs consists of two adjacent regions, a G-site on the thioredoxin-like domain that supports GSH as the thiol substrate or cofactor, and a nonpolar H-site on both domains that contributes to the binding of hydrophobic substrates. Although the molecular recognition of GSH is conserved among GSTs, these enzymes employ different strategies such as the nature of the residue (Tyr, Ser, or Cys) in contact with the thiol group of GSH and the structural contribution by domain 2 to the active site to realize their diverse functionalities (1,4). 2 The class Alpha GSTs possess, like many other GST classes, an active site tyrosine residue for the activation of GSH but, unlike other GST classes, they possess an extended C-terminal region that forms an integral part of the active site (5)(6)(7). The major structural element in the C-terminal region is the amphipathic helix 9 that, although not directly involved in the chemical mechanism of catalysis, is an important determinant of substrate selectivity (8 -11), the binding of substrates (12), rate-determining steps (13), desolvation of the active site (14), and the pK a of the catalytic tyrosine residue (12,15). In addition to its contributions to catalysis, the C-terminal region is an important determinant of ligandin function, i.e. the binding of nonsubstrate ligands (16 -18). Given these contributions of the C-terminal region to the functions of class Alpha GSTs, the conformational dynamics of the region and their impact on enzyme function have received much attention.
Although the C-terminal region in hGST A1-1 is highly dynamic and not observed in the crystal structure of the apo enzyme (5,7), it is observed in the apo structure of the homologous hGST A4-4 enzyme (8). Rather than being completely disordered, the C-terminal region in apo-hGST A1-1 assumes helix-like conformations close to the surface of the protein (19), where it experiences tertiary contacts that facilitate its folding (16,20). It is only when the enzyme binds G-site and/or H-site ligands that the C-terminal region becomes fully immobilized on the protein (5)(6)(7)19). However, the structural determinants that maintain the conformational stability of the region in the apo and complex forms of hGST A1-1 (and other class Alpha GSTs) are not clear. Recently, we demonstrated that a bulky, * This work was supported by the University of the Witwatersrand, the South African National Research Foundation (Grant 205359), and the Wellcome Trust (Grant 060799). 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.
hydrophobic residue at position 219 contributes significantly toward the stability of the region in apo and complexed hGST A1-1 (17).
The type of amino acid residue found at the N-terminal end of a helix has a major effect on the stability of the entire helix and is referred to as the N-capping residue (21). All GSTs have a conserved N-capping motif for helix 6 in domain 2 (22) shown to play an important role in the folding and stability of GSTs (23,24). In this study, we identified a conserved aspartate N-capping motif for helix 9 in all class Alpha GSTs and, given the importance of N-capping motifs in the folding and stability of ␣-helices (25,26), investigated its contribution to the dynamics of the C-terminal region of hGST A1-1.

EXPERIMENTAL PROCEDURES
Chemicals-GSH was from ICE Biomedical Inc. (Aurora, OH). TFE (Ͼ99% grade), 8-aniline-1-naphthalene sulfonate, ethacrynic acid, glutathione sulfonate, p-bromobenzyl GSH, and 1-chloro-2,4-dinitrobenzene (CDNB) were purchased from Sigma-Aldrin. All other reagents were of analytical grade. Peptides corresponding to the wild-type and D209G forms of the C-terminal region of hGST A1-1 were synthesized by Alpha Diagnostic Inc. (San Antonio, TX), and their molecular masses and purity were confirmed by matrix-assisted laser desorption ionization time-of-flight mass spectrometry. The peptides have the following sequences: Ac-Tyr-Gly-Met-Asp/Gly-Glu-Lys-Ser-Leu-Glu-Glu-Ala-Arg-Lys-Ile-Phe-Arg-Phe. Residues Met 3 to Phe 17 correspond to the C-terminal sequence in hGST A1 (i.e. Met 208 to Phe 222 ). The underlined residues correspond to those residues located in helix 9 in the crystal structures of hGST A1-1 complexed with active site ligands (5)(6)(7). The presence of a tyrosine at the N terminus of the peptide, used to quantitate the peptide, should not induce a significant aromatic band in their CD spectra because of the flexible helix-breaking glycine spacer separating the tyrosine from the rest of the peptide (27). Peptide stock solutions were made in water and their concentrations determined at 275 nm using an extinction coefficient of 1450 M Ϫ1 ⅐cm Ϫ1 (28).
Mutagenesis, Protein Expression, and Purification-The plasmid (pKHA1) used for the expression of hGST A1-1 (29) was a generous gift from Bengt Mannervik (Department of Biochemistry, University of Uppsala, Sweden). Oligonucleotide site-directed mutagenesis, based on the QuikChange method (Stratagene, La Jolly, CA), was performed to generate the D209G mutant of hGST A1-1 using the following primers: 5Ј-GCCTCCATGGGTGAGAAATCTCTAGAAGAAGC and 5Ј-GCTTCT-TCTAGAGATTTCTCACCCATGGGAGGC. The mutations introduced for generating D209G hGST A1-1 are underlined. The presence of the mutation in the plasmid was confirmed by DNA sequencing. Wild-type and D209G hGST A1-1 were expressed in BL21(DE3) pLysS Escherichia coli cells and purified by CM-Sepharose chromatography as described previously (20). The concentration of the dimeric proteins was determined at 280 nm using the molar extinction coefficient of 38,200 M Ϫ1 ⅐cm Ϫ1 .
Spectroscopic Methods-Far-UV CD measurements were done at 20°C in a Jasco model 810 CD spectropolarimeter. Averaged CD signals, corrected for solvent, were converted to mean residue ellipticity [⌰], where C is the peptide concentration in millimolar, ⌰ is the measured ellipticity in millidegree, n is the number of residues (17 in this case), and l is the path length (cm). The percentage helical content was calculated from the mean residue ellipticity at 222 nm, using the equation, where [⌰] helix and [⌰] coil represent the mean residue ellipticity of a helix [Ϫ42,500(1 Ϫ (3/n)] and random coil (ϩ640), respectively (30). The units of mean residue ellipticity are deg⅐cm 2 ⅐dmol Ϫ1 . Fluorescence measurements were performed at 20°C in a Hitachi model 850 fluorescence spectrofluorometer. The intrinsic fluorescence of hGST A1-1 was monitored by selectively exciting a single tryptophan residue (Trp-2 1 ) per subunit at 295 nm. The binding of the anionic dye, ANS, to hGST A1-1 was monitored by fluorescence enhancement using an excitation wavelength of 390 nm (31).
Steady-state Kinetics-Kinetic parameters were determined at pH 6.5 for the S-conjugation reaction between GSH and CDNB (32). Reaction rates were measured in triplicate at 20°C and corrected for non-enzymatic rates. Conditions for determining K m GSH were 0.05-6.5 mM GSH and 1.5 mM CDNB, and for K m CDNB they were 0.15-1.5 mM CDNB and 5 mM GSH. For k cat /K m GSH determinations, CDNB was kept at 1.5 mM, whereas GSH was varied from 0.05 to 0.15 mM. For k cat /K m CDNB determinations, GSH was kept at 5 mM, whereas CDNB was varied from 0.05 to 0.15 mM. The data were fitted by nonlinear regression analysis using Origin5 MicroCal Inc. software.
Active Site Tyrosinate Formation-The ionization of the catalytic Tyr 9 in the active site of hGST A1-1 was determined over the pH range 5.5-9.0 by UV absorbance difference spectroscopy as described (17). The concentration of tyrosinate ion formation was calculated using a molar extinction coefficient of 2350 M Ϫ1 ⅐cm Ϫ1 and the pK a of Tyr 9 determined as described (15,33).
Isothermal Titration Calorimetry-Isothermal titration calorimetry with a VP-ITC calorimeter from MicroCal (Northampton, MA) was employed to determine the thermodynamic parameters of ANS⅐hGST A1-1 complex formation in 20 mM phosphate buffer, pH 6.5, as described (17,18). Total heats were obtained by titrating 0.04 mM D209G hGST A1-1 (subunit concentration) with 5-l aliquots of 4.1 mM ANS. Heats of dilution, determined by titrating ANS into buffer alone, were subtracted from the total observed heats, and the corrected data analyzed by nonlinear regression with Origin5 (MicroCal).
Molecular Docking-Two molecular docking programs, Cerius 2 (Accelrys Inc., San Diego, CA) and LIGIN (34), were used to predict the ANS binding site in hGST A1-1. The 1K3Y high resolution structure of wild-type hGST A1-1 without the bound ligand was used as the host apo molecule. Cerius 2 treated ANS as a flexible, guest ligand and the protein host as a rigid body. Various ligand conformations and binding locations were ranked according to conformational energy minima, with the lowest being the most probable binding complex. Energy minimization was carried out with GROMOS (35) in the WHATIF version 2.0 package (36). LIGIN, which uses surface complementarity between guest and host as the guiding principle for predicting binding sites, was implemented as described (34). Docking between ANS and hGST A1-1 was also done with the binary GSH-protein (PDB file 1GSE without the ethacrynic acid moiety) and GSO 3 -protein (coordinates for GSO 3 Ϫ obtained from PDB file 1EV4) complexes.

RESULTS AND DISCUSSION
Asp-209, the N-cap Residue of Helix 9 -The AGADIR algorithm, based on helix-coil transition theory that explicitly considers specific interactions occurring in helices devoid of tertiary interactions (37), was used to identify N-capping motifs for helices in the class Alpha hGST A1-1. Two such motifs were identified; one for helix 6 and the other for helix 9, both of which are conserved in class Alpha GSTs. The N-capping motif of helix 6 (22) is a class 5 Ser motif (26), which has been shown to play an important role in the folding and stability of GSTs (23,24). The motif of helix 9 is a class 3 Asp N-capping motif (26) and is conserved in all class Alpha GSTs (Table I). The motif together with its hydrogen-bonding pattern with the main chain of helix 9 is shown in Fig. 1.
Helical Content of C-terminal Region Peptides-Far-UV CD spectra, used to determine the helical content of two peptides corresponding to the sequences of the wild-type and D209G mutant C-terminal regions (WT-pep and DG-pep, respectively) in water, are shown in Fig. 2. The spectra, displaying ellipticity minima at 222 and 201 nm and ellipticity maxima at 218 and 194 nm, are characteristic of a mixture of helical and random coil conformations. The low helical content of WT-pep and DG-pep in water (10 and 4%, respectively) was substantiated by AGADIR (i.e. 20 and 8% for the wild-type and mutant sequences, respectively). AGADIR also predicted that the length of the helix in the wild-type sequence is longer (residues Asp-210 to Ile-219 with individual propensities of 21.2-37.4%) than that in the D209G mutant sequence (residues Leu-213 to Ile-219 with individual propensities of 11.4 -18.6%). The residues predicted to form a helix in the wild-type sequence are in agreement with those that form helix 9 in the protein structure (6). Capping motifs are known to increase helical propensity by initiating and/or stabilizing helix propagating tendencies at their ends (38 -40). Thus, the reduced CD signal of the DG-pep can also be interpreted as an increase in fraying of the helix at its N-terminal end. TFE, a co-solvent known to enhance the helical content of peptides with intrinsic helical properties (30,41,42), increases the helix content of both peptides, as shown in Fig. 2. The TFE-induced changes of the CD spectra exhibit an isodichroic point at 202-204 nm indicative of a two-state, helix-coil transition (43). The helix content of both peptides reaches a maximum level of about 35% in the presence of TFE (inset in the bottom panel of Fig. 2).
Although the sequence of the C-terminal region displays a preference to be disordered in aqueous solution, the flexible region in apo-GST A1-1 assumes an ensemble of native helixlike structures close to the surface of the protein where it samples multiple environments (19). It is only when the active site becomes occupied by ligand that the C-terminal region becomes stabilized and localized on the surface of the protein (5-7, 16, 19, 44). Tertiary interactions at the interfaces between the C-terminal region, protein, and bound ligand, therefore, play a major role in stabilizing the region (16,17), consistent with the helix-stabilizing effect of TFE. Disrupting the conserved N-capping motif of helix 9 by the D209G mutation substantially reduces the intrinsic ability of its sequence to form a helix. Consequently, this alters the dynamics of the C-terminal region and, because it forms part of the active site, the function of the enzyme (see below).
Characteristics of D209G hGST A1-1-Helix 9 (residues 210 -220) is the major structural feature in the C-terminal region (residues 210 -222) of GST A1-1 (6). As expected, replacing the N-cap residue, Asp-209, with a glycine residue had little impact on the global structural features of hGST A1-1 in that its secondary and dimeric structures were unchanged (data not shown). The tertiary environment of the lone tryptophan, Trp-21, is, however, different in the mutant, as indicated by a lower fluorescence intensity, although the maximum emission wavelength was unaltered (data not shown). Trp-21 is located at the domain-domain interface of each subunit where its indole ring is surrounded by helix 1 in domain 1 and helices 6 and 8 in domain 2. Although the Asp-210 N-capping motif of helix 9 is about 18 Å away from Trp-21, it is connected to helix 8 via a 10-amino acid linker. Not only is the fluorescence of Trp-21 sensitive to changes occurring at the domain-domain interface (45), it is also sensitive to changes in the conformational dynamics or deletion of the C-terminal region of GST A1-1 (16,20,46,47). Therefore, disrupting the stabilizing N-capping motif in helix 9 appears to alter the conformational dynamics of the C-terminal region. The Asp N-capping motif most likely plays an important role in the proper folding and stabilization of the C-terminal region of hGST A1-1 allowing the region to associate with the assembled and folded dimeric protein (20). Aspartate is one of the most stabilizing residues commonly found at the N-cap position in ␣-helices of proteins, substituting a Gly residue at this position has been found to significantly destabilize proteins (48,49). Therefore it is expected that the substitution of Asp-209 with a glycine would compromise the stability of helix 9. Tyr-9, the catalytic residue involved in the activation of GSH, displays an unusually low pK a of 8.1 (15) and is proposed to control the dynamics of the C-terminal region in GST A1-1, in that the region becomes more dynamic when Tyr-9 becomes ionized (50,51). This is unlikely a reason for the conformational change in the C-terminal region of D209G GST A1-1 at pH 6.5 because the pK a of Try-9 is not perturbed by the mutation (pK a is 8.2 for both wild-type and mutant; data not shown).
GSH Binding and Steady-state Kinetics-The binding affinity of GSH, determined by tryptophan fluorescence quenching, is unaltered by the D209G mutation, because the K d values for the wild-type and D209G proteins are similar; 0.16 Ϯ 0.01 mM and 0.18 Ϯ 0.01 mM, respectively. This is consistent with the binding of GSH not being significantly influenced by the Cterminal region (12,16,17,52,53).
Although the C-terminal region does not contribute directly to the chemical mechanism of catalysis, it does form an integral part of the active site, the H-site specifically. Consequently, the catalytic function of GST A1-1 is highly sensitive to perturbations in the conformation and dynamics of the C-terminal region (12, 13, 16, 17, 53). The K m values of GSH (0.21 mM for wild-type and 0.29 mM for D209G) and CDNB (0.31 mM for The conserved N-cap Asp residue is in boldface, and the underlined residues correspond to those located in ␣-helix 9 of the crystal structures. wild-type and 0.35 mM for D209G) are not adversely affected by the mutation. Perturbations in or the deletion of the C-terminal region have also been shown not to affect the K m of GSH, whereas the impact on the K m of CDNB is dependent on the nature of the perturbation (9,13,17,54). For example, deletion of the entire region increases the K m of CDNB significantly, whereas smaller perturbations in the region result in no or smaller changes in the K m CDNB . The catalytic efficiency of the D209G enzyme with CDNB, k cat /K m CDNB , is reduced by 15% (86 mM Ϫ1 ⅐s Ϫ1 and 73 mM Ϫ1 ⅐s Ϫ1 for wild-type and D209, respectively). The difference in the free energy change for the formation of the transition states in the mutant and wild-type enzymes (⌬⌬G), as calculated from Equation 3 (55), is 0.4 kJ⅐mol Ϫ1 at 20°C, indicative of a lower stabilization of the transition state on disruption of the conserved N-capping motif in helix 9. The C-terminal region of GST A1-1 plays an important role in guiding the substrates into the transition state (13). The diminished stabilization effect observed for the D209G mutant is, however, not as pronounced as that observed when Met-208, the hydrophobic staple motif at the N terminus of helix 9, is altered (53). It would appear that the reduction in catalytic efficiency is not due to an increase in K m CDNB but rather a decrease in k cat CDNB (not determined). A decrease in k cat CDNB , in turn, would not be due to a decreased rate of product release, because mobility-enhancing perturbations in the Cterminal region generally increase the rate of product release (12,51,53). Rather, it is most likely that the affinity of the H-site for CDNB is reduced, which is consistent with the link between productive binding of CDNB and the dynamics of the C-terminal region (9). According to the structure of the complex between hGST A1-1 and S-benzylglutathione (1GUH (6)), the benzene ring of CDNB should be within van der Waals distance of Met-208, Phe-220, and Phe-222 at the H-site. The latter two residues are located in the C-terminal region and changes in their positioning at the H-site would compromise the binding of CDNB (17) and the stabilization of the transition state (13). Both phenyl rings of Phe-220 and Phe-222 contribute to the hydrophobic environment of the H-site (6), whereas the phenyl ring of Phe-222 also reduces the exposure of the site to bulk solvent (17). The calorimetric and fluorescence data presented below suggest that the D209G-induced perturbation in the C-terminal region reduces the hydrophobicity of the active site and increases its accessibility to solvent.
ANS as a Probe of the Active Site-Although there are no experimental structures available of an ANS⅐GST complex, molecular docking and ligand-displacement studies indicate the anionic dye to bind the H-site of hGST A1-1. Both Cerius 2 (Accelrys Inc.) and LIGIN (34) docking software predict ANS to bind the H-site albeit in different binding modes in the absence and presence of GSH, as shown in Fig. 3. Ethacrynic acid, another H-site ligand, has also been shown to bind hGST A1-1 in different modes (5). It should be noted that the docking experiments were done on the assumption that the conformation of the C-terminal region of the protein in the hGST A1-1⅐ANS complex is the same as that observed in other ligand complexes of the protein. The assumption is supported by the finding that ANS binding, like that for ethacrynic acid, induces the localization of the C-terminal region (see below).
The hydrophobic anilino and naphthyl rings of ANS occupy the nonpolar H-site and are within van der Waals distance of Leu-107, Leu-108, Val-111, Met-208, Leu-213, Ala-216, Phe-220, and Phe-222, whereas the negatively charged sulfonate group is located at the interface between the G-site and H-site and is within van der Waals distance of Tyr-9, Phe-10, Arg-15, and Val-55. These amino acid residues are also found within van der Waals distance of various other ligands bound to the H-site of hGST A1-1 (5-7). The predominantly nonpolar character of the binding site for the aromatic moieties of ANS is consistent with those of other ANS-binding proteins (57)(58)(59). Of particular note, is that no binding site for ANS could be detected when GSO 3 Ϫ was present at the active site. Although GSO 3 Ϫ binds the G-site in the same way that GSH does, its negatively charged sulfonate group occupies a similar position to that of ANS bound to the H-site. ANS is strongly displaced by ligands that bind the H-site, as shown in Fig. 4, consistent with the H-site being the binding site of ANS in both wild-type and D209G hGST A1-1. The displacement of ANS by GSO 3 Ϫ is explained by their competitive binding due to overlapping bind- ing sites for their sulfonate groups (see above). GSH, on the other hand, does not displace ANS, because there is no competition for binding space, as suggested in Fig. 3.
That ANS binds the H-site is consistent with the position located for ANS by fluorescence-resonance energy transfer (31) and by the finding that ANS inhibits GST A1-1 competitively with respect to CDNB (60). ANS also competes with AFB 1 (61), whose exo-8,9-epoxide form binds the H-site of class Alpha GSTs where it is conjugated with GSH (62)(63)(64). A high affinitybinding site for ANS at or near the promiscuous H-site of hGST P1-1 has also been reported (65,66), in agreement with the competitive binding between ANS and BSP (67). The crystal structure of the BSP⅐GST P1-1 complex demonstrates the anionic ligand to bind the H-site (68). The inhibition of hGST A1-1 by BSP, however, is the result of BSP binding to its low affinity site, namely the H-site (69).
Recently, we demonstrated that the energetics of ANS binding can be used to report on changes at the active site of hGST A1-1, including the C-terminal region of the protein, because it forms an integral part of the H-site (17,70). The calorimetric titration isotherm for the binding of ANS to D209G hGST A1-1, shown in Fig. 5, indicates that complex formation is exothermic.
The binding curve fits well to a model describing one binding site per subunit with a stoichiometry of one ANS molecule per site consistent with that reported for the wild-type protein (18). The affinity of hGST A1-1 for ANS is not significantly affected by the D209G mutation (K d ϭ 52 M) when compared with that of the wild-type protein (K d ϭ 65 M) (18) and the I219A mutant (K d ϭ 47 M) (17). ANS does, however, bind more tightly to these enzymes forms than to the Phe-222 deletion mutant (K d ϭ 105 M) (18), indicating the importance of this residue in the binding of ligands at the H-site.
The temperature dependence of the thermodynamic parameters of the D209G⅐ANS complex formation is shown in Fig. 6. Across the experimental temperature range (5-25°C), the free energy of complex formation is defined by a negative enthalpy change (⌬H obs ) and a positive entropy change (T⌬S obs ). The free energy remains constant throughout the temperature range due to enthalpy-entropy compensation (71), a process characteristic of the weak interactions found between proteins and their ligands (72). This compensatory effect is also observed between wild-type and mutant complex formation. Although their ⌬G values are very similar, their ⌬H obs and T⌬S vary greatly. The linear temperature dependence of enthalpy (Fig.  6) indicates that the heat capacity change of D209G⅐ANS complex formation (⌬C p ϭ 0.3 kJ⅐mol Ϫ1 ⅐K Ϫ1 ) is not coupled to other structural equilibria with significant enthalpies.
Given the amphipathic nature of ANS, its interaction with proteins cannot be strictly hydrophobic, as often assumed, but will involve van der Waals and electrostatic forces. The importance of van der Waals interactions has been demonstrated for the enthalpically driven formation of the wild-type complex (18), consistent with the putative van der Waals contacts between the aromatic ring systems of ANS and the H-site (Fig. 3), and for the formation of other protein⅐ANS complexes (57,73). The role of electrostatic interactions in ANS binding to hGST A1-1, however, was previously not clear. Considering the close proximity (within 4 Å) of the negatively charged sulfonate group in ANS to the positively charged guanidinium group of Arg-15 at the interface between the H-site and G-site (see Fig.  3), electrostatic interactions between these moieties could very well be important contributors to complex formation (74). This would certainly apply to addressing the energetically unfavorable desolvation of the highly solvated, negatively charged sulfonate group of free ANS that becomes buried in a predominantly nonpolar environment. Electrostatic interactions between the sulfonate group of ANS and Arg-15 would, at least in part, compensate for this unfavorable energy of desolvation by their large exothermic contributions to enthalpy (74). Furthermore, compared with that of bulk solvent, the lower dielectric environment of the ANS binding site, a consequence of the nonpolar site and anilinonaphthalene moiety of ANS, would enhance the strength of these electrostatic interactions. Conversely, the increased solvent exposure of ANS bound to the D209G mutant (see below) would attenuate the exothermic contribution to enthalpy from the ANS⅐Arg-15 electrostatic effect resulting in a less favorable ⌬H obs . Arg-15 is conserved at the active site of all class Alpha GSTs, providing electrostatic stabilization of the thiolate anion of glutathione (15). The presence of electrostatic interactions between the sulfonate group of ANS and the guanidinium group of an arginine residue is also demonstrated in the structures of other protein⅐ANS complexes (57)(58)(59)75), supporting the important role that the ion pair plays in determining binding affinity. Because most native proteins do not bind ANS, or at least appear not to because the many ANS molecules bound to surface cationic residues of proteins by electrostatic forces only do not necessarily fluoresce and, thus, go undetected by fluorescence techniques (74), binding specificity appears to be determined largely by the anilinonaphthalene moiety of ANS. Each of the ANS-binding proteins mentioned here possesses a specific nonpolar site for ANS, and binding of the anion occurs with a stoichiometry of one molecule per monomer of protein.
The positive T⌬S of the D209G mutant reflects a net favorable entropy change of solvation in that restrained water molecules solvating, at least, the nonpolar surface of the aromatic ring systems and negatively charged sulfonate group of free ANS are released to bulk solvent upon complex formation (76). The T⌬S term will, however, be reduced by a smaller, unfavorable change in conformational entropy due to a loss of degrees of freedom at the interface between ANS and protein (18). Free ANS is relatively constrained, because only two angles, one a dihedral angle and the other a torsion angle, define its conformation (57). The positive enthalpy associated with desolvation events will attenuate the favorable enthalpy of binding. Only at temperatures below 17°C does entropy contribute favorably toward ANS binding to wild-type hGST A1-1 (18), implying that at higher temperatures, desolvation of the interacting surfaces is inadequate to compensate for losses in conformational entropy. These losses will include those associated with ANS and the protein groups involved in binding, including the C-terminal region. Although the binding of another H-site ligand, ethacrynic acid, induces localization of the C-terminal region of the protein (5), the situation with respect to the binding of ANS is not clear. Nevertheless, given that the putative interactions that ANS undertakes with the region (Fig. 3) are similar to those involved in binding ethacrynic acid (5), it is feasible that complex formation with ANS also induces the localization of the C-terminal region in the wild-type protein, thus contributing unfavorably to entropy. Furthermore, localization of the C-terminal region would contribute favorably to enthalpy due to the additional interactions at the interfaces between the region, protein, and ANS. On the other hand, the positive entropy of ANS⅐D209G complex formation suggests that ANS binding to the mutant does not induce localization of the region resulting in a less favorable enthalpy. Further, ANS bound to the mutant is also more exposed to solvent compared with the wild-type complex (see below). Complex formation between ANS and two other ANS-binding proteins, ALBP and IFABP, display T⌬S values of 1.62 kJ⅐mol Ϫ1 at 25°C (57) and Ϫ7.5 kJ⅐mol Ϫ1 at 22°C (73), respectively. The favorable entropy of the ANS⅐ALBP complex is likely due to the release of ordered water molecules from the ANS binding site, because binding does not induce major changes in the conformation of the protein. On the other hand, the unfavorable entropy of the ANS⅐IFABP appears to result from the ordering of ANS (73) and a water molecule at the binding cavity (77). The waters displaced by ANS at this site are short-lived waters that are not highly ordered. Ligand binding also does not significantly affect the structure of IFABP (78).
The heat capacity change of complex formation with the D209G mutant is positive (0.3 kJ⅐mol Ϫ1 ⅐K Ϫ1 ), whereas it is negative (Ϫ0.84 kJ⅐mol Ϫ1 ⅐K Ϫ1 ) for the wild-type protein (18), indicative of significant differences in the type and amount of solvent-exposed surface areas buried on binding. Because the change in heat capacity originates primarily from the changes in the hydration of nonpolar and polar molecular surfaces areas (79), the positive change observed for the mutant suggests the burial of a smaller nonpolar surface area and the burial of a larger polar surface area on complex formation (80). Because electrostatic interactions generate positive heat capacity changes (76, 81), the interactions between the negatively charged sulfonate group of ANS and the positively charged guanidinium group of Arg-15 at the H-site offers an explanation for the positive ⌬C p values of ANS complex formation with the D209G mutant (this study) and the ⌬Phe-222 mutant (18). Electrostatic effects, therefore, play a larger role in ANS binding to the D209G mutant than in binding to the wild-type protein in which a localized C-terminal region is an important contributor to the burial of nonpolar surfaces and the formation of van der Waals contacts. The affinity displayed by ANS toward wild-type and D209G hGST A1-1 is similar. However, the D209G mutation renders the formation of the ANS⅐hGST A1-1 complex enthalpically less favorable by 29.2 kJ⅐mol Ϫ1 and, due to the compensatory effect, entropically more favorable by 29.8 kJ⅐mol Ϫ1 at 25°C. This is consistent with fewer van der Waals interactions and higher conformational entropy in the mutant complex as a result of a more dynamic C-terminal region.
Like that for wild-type hGST A1-1, the binding of ANS to I-FABP (73) and A-LABP (57) is also associated with similar, negative values of ⌬C p (Ϫ1.18 kJ⅐mol Ϫ1 ⅐K Ϫ1 and Ϫ0.92 kJ⅐mol Ϫ1 ⅐K Ϫ1 , respectively). Without empirical structural data of the GST⅐ANS complex, it is not possible to calculate the heat capacity changes based on changes in nonpolar and polar surfaces on complex formation (82,56). This can, however, be done for the ANS⅐ALBP complex (PDB code 2ANS (57)). Because ANS binding does not induce any major conformational changes in ALBP, the apo form of ALBP was generated by removing the coordinates of ANS from those of the ANS⅐ALBP complex. The solvent-accessible surface areas of ANS, apo-ALBP, and the ANS⅐ALBP complex, as well as the surface area changes associated with complex formation were calculated as described (18)  where ⌬C p,calc is the calculated change in heat capacity on complex formation, ⌬ASA is the change in accessible surface area on complex formation, the subscripts ali, arm, bb, and pol represent aliphatic, aromatic, backbone and polar surfaces, and the four numerical coefficients are expressed in kJ⅐mol Ϫ1 ⅐K Ϫ1 per Å 2 . A ⌬C p,calc of Ϫ0.72 kJ⅐mol Ϫ1 ⅐K Ϫ1 accounts for 78% of the experimental ⌬C p indicating that the majority of ⌬C p is a consequence of the formation of the ALBP⅐ANS interface with its associated changes in solvation. Although the proteins are different, it is probable that this also applies to the formation of the wild-type ANS⅐hGST A1-1 given a ⌬C p,calc value of Ϫ0.9 kJ⅐mol Ϫ1 ⅐K Ϫ1 determined for the ANS⅐protein complexes generated by docking programs (see above). Furthermore, in a recent study of the binding of S-hexylglutathione to wild-type hGST A1-1, the experimental ⌬C p of the interaction could be successfully predicted from the changes in solvent-accessible surface areas on complex formation. 2 In light of the thermodynamic data and the low helix propensity of the sequence corresponding to the C-terminal region, it is not unreasonable to assume that disruption of the stabilizing N-capping motif of helix 9 increases the mobility of the C-terminal region preventing it to become properly localized on binding of ANS. This is supported by ANS fluorescence data. ANS experiences a blue shift in its maximum emission wavelength from 545 to 474 nm on complex formation, demonstrating that the ANS binding site is nonpolar. However, because water quenches ANS fluorescence (57,73), a 30% reduction in the quantum yield of ANS bound to the D209G mutant (data not shown) indicates that the dye is more exposed to solvent than when the dye is bound to wild-type hGST A1-1. The quantum yield of ANS is not affected by the conformation of its bound form (57). ANS complexes with the ⌬Phe-222 and I219A mutants of hGST A1-1 also displayed lower quantum yields of ANS fluorescence (17,18). These studies demonstrated the important role of the phenyl side chain of Phe-222 in reducing the solvent accessibility of the active site to solvent. In fact, a crystal structure of the I219A mutant indicates that the side chain of Phe-222 no longer occupies a position over the H-site, thus exposing it to solvent. 2