Tyrosine 115 Participates Both in Chemical and Physical Steps of the Catalytic Mechanism of a Glutathione S-Transferase*

The participation of the hydroxyl group of tyrosine 115 in the catalytic mechanism of isoenzyme 3-3 of rat glutathione (GSH) S-transferase is implicated by x-ray crystallographic analysis of a product complex and confirmed by comparison of the catalytic properties of the native enzyme and the Y115F mutant. Tyrosine 115 is located in domain I1 of the protein (the xenobiotic substrate binding domain) and is the first residue in this domain to be shown to play a direct role in catalysis. The 1.8-bi structure of isoenzyme 3-3 in complex with (SS,lOS)-9-(S-glutathionyl)-lO-hydroxy-9,10-dihydrophenanthrene, one of the diastereomeric products of the reaction of GSH with phenanthrene 9,10-oxide, indicates that the hydroxyl group of Tyr115 is within hydrogen-bonding distance of the 10-hydroxyl group of the bound product and, by implication, is proximal to the oxirane oxygen of the substrate in the Michaelis complex. Site-specific replacement of Qr115 with phenylalanine has pro-foundly different effects on catalysis depending on the type of reaction and whether the rate-limiting step in catalysis is a chemical step or a physical step. Stopped flow measurements of the rate constants

The participation of the hydroxyl group of tyrosine 115 in the catalytic mechanism of isoenzyme 3-3 of rat glutathione (GSH) S-transferase is implicated by x-ray crystallographic analysis of a product complex and confirmed by comparison of the catalytic properties of the native enzyme and the Y115F mutant. Tyrosine 115 is located in domain I1 of the protein (the xenobiotic substrate binding domain) and is the first residue in this domain to be shown to play a direct role in catalysis. The 1.8-bi structure of isoenzyme 3-3 in complex with (SS,lOS)-9-(S-glutathionyl)-lO-hydroxy-9,10-dihydrophenanthrene, one of the diastereomeric products of the reaction of GSH with phenanthrene 9,10-oxide, indicates that the hydroxyl group of Tyr115 is within hydrogen-bonding distance of the 10-hydroxyl group of the bound product and, by implication, is proximal to the oxirane oxygen of the substrate in the Michaelis complex. Site-specific replacement of Qr115 with phenylalanine has profoundly different effects on catalysis depending on the type of reaction and whether the rate-limiting step in catalysis is a chemical step or a physical step. Stopped flow measurements of the rate constants for product release and viscosity effects on the steadystate kinetics establish that the rate-limiting step in catalysis with phenanthrene 9,lO-oxide (kc,, = 0.4 s-') is probably a chemical one, whereas the physical step of product dissociation ( k , d is rate-limiting in the reaction of l-chlor0-2,4-dinitrobenzene (kc,, = 20 s-'). The Y115F mutant is severely impaired in catalyzing the addition of GSH to phenanthrene 9,lO-oxide (kcat = 0.0044 s-l), evidence that the -OH of Tyr116 provides electrophilic assistance in the epoxide ring opening. In contrast, the Y115F mutant is a better catalyst toward l-chloro-2,4-dinitrobenzene (k,,, = 72 s-') than is the native enzyme. The enhanced rates of product release in the mutant are ascribed to the loss of hydrogen bonds between the -OH of %11' and the side chain -OH and main chain NH of serine 209, interac-* This work was supported by Grant GM 30910 from the National Institute of General Medical Sciences. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby markec! "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.  11 To whom correspondence should be addressed. Tel.: 301-405-1812; tions that block the channel to the active site or inhibit the segmental motion of the protein.
The glutathione S-transferases (EC 2.5.1.18) catalyze the addition of glutathione (GSH) to endogenous and xenobiotic compounds bearing electrophilic functional groups and are arguably the single most important group of enzymes involved in the metabolism of potentially toxic alkylating agents. As a group the enzymes exhibit a remarkable degree of catalytic diversity, which is due, in part, to a large number of isoenzymes with somewhat different preferences for substrate topology and reaction type. For a n historical perspective and recent reviews the reader is referred to Jakoby (1978), Mannervik (1985), Mannervik and Danielson (1988), Pickett and Lu (1989), and Armstrong (1991). Recent crystallographic studies indicate that the polypeptide is organized into two domains, a GSH binding domain (domain I) at the N terminus and a xenobiotic substrate binding domain (domain 11) at the C terminus (Reinemer et al., 1991;J i et al., 1992). Crystallographic, spectroscopic, and kinetic investigations of native and mutant enzymes have established that tyrosine 6 of domain I participates directly in catalysis by lowering the pK, of bound GSH (Liu et al., 1992;Kolm et al., 1992;Widersten et al., 1992).
It is reasonable to anticipate that other residues, perhaps contributed by the xenobiotic substrate binding domain, might participate directly in catalysis by helping to stabilize particular transition states such as the oxyanion in epoxide ring openings (Scheme I), the enol (enolate) in Michael additions, or the Meisenheimer complex in nucleophilic aromatic substitutions (Scheme I). For example, the enzyme might provide electrophilic assistance in epoxide ring openings by donating a hydrogen bond to the developing charge on the oxirane oxygen (Cobb et al., 1983). The effectiveness of this participation in any given reaction type, which helps define the catalytic diversity of the various isoenzymes, depends on the stereoelectronic demands of the transition state and the nature of the rate-limiting step. Although the construction of chimeric isoenzymes has implicated residues in the C-terminal tail as participating in catalysis (Zhang and Armstrong, 1990;, no evidence has been forthcoming defining the direct participation of any individual residue from domain I1 in the catalytic mechanism. In this paper we present evidence that the hydroxyl group of tyrosine 115 (Tyr115) from the xenobiotic substrate binding domain participates in both chemical and physical steps of the catalytic mechanism of isoenzyme 3-3 and that the net effect of that participation is highly dependent on the nature of the transition state and the rate-limiting step for the reaction in question. Katusz and Colman (1991) provided the first evidence that Tyr115 is located near the active site of mu class isoenzymes by showing that this residue in isoenzyme 4-4 was modified by the active site-directed reagent l-(S"glutathionyl)-2,3dioxo-4-bromobutane. Subsequent crystallographic studies of the isoenzyme 3-3.GSH complex located Tyr115 in domain I1 on the face of the a4-helix that forms one wall of the xenobiotic substrate binding site, the hydroxyl oxygen being about 7.5 a from the sulfur of GSH . In this paper we provide crystallographic and kinetic evidence that the hydroxyl group of Tyr115 provides electrophilic assistance in the addition of GSH to an arene oxide substrate, a reaction in which the chem-l3r1I5 and the Mechanism of Glutathione Dunsferase 11509 ical step is rate-limiting. The same hydroxyl group is also shown to inhibit egress of the products from the active site. Thus removal of the hydroxyl group as in the Y115F mutant has quite different consequences for reactions in which the physical step of product release, rather than chemistry, is ratelimiting.

EXPERIMENTAL PROCEDURES
General Materials and Methods-The substrates CDNBl and PhenO were obtained as previously described (Cobb et al., 1983). The products (S,S)and (RJ)-GSPhen and GSDNB were synthesized by the method described by Chunget al. (1987). X-ray crystal structure determinations were carried out by the general methods described by J i et al. (1992) and will be described in detail elsewhere.2 Reaction kinetics were determined as previously described (Liu et al., 1992) at pH 7.0 and 25 "C. Kinetic constants were derived by fitting initial velocity data to a hyperbola with the program HYPER (Cleland, 1979). DNA sequencing was performed with Sequenase version 2 from U.
S. Biochemical Corp. with the enzyme described by Tabor and Richardson (1989).
Preparation of the Y115FMutant--An expression vector encoding the Y115F mutant was prepared by using overlap extension for sitedirected mutagenesis via the polymerase chain reaction (Higuchi, 1991). The two "inside" primers were mismatched a t single bases in the target sequence to change the codon a t position 115 of the expression plasmid pGT33MX (Zhang et al., 1991) from TAC to TTC to introduce theY115F mutation. The two "outside" primers flanked the unique SphI and BstEII restriction sites located on either side of position 115. The two primary polymerase chain reaction products of 56 and 145 base the 3'-end with Taq polymerase. The resulting 185-base pair fragment pairs were purified, mixed, denatured, reannealed, and extended from was amplified using the two "outside" primers. The product was purified and digested with SphI and BstEII to yield a 130-base pair frag-pGT33MX which had been digested with the same two restriction en-ment containing the mutant codon. This cassette was introduced into zymes. The relevant region of the resultant plasmid pGT33MXY115F was sequenced to ensure that the mutant codon and no other unwanted mutation had been introduced. Expression and isolation of the mutant protein was accomplished as described by Zhang et al. (1991).
Stopped Flow Fluorescence Measurements-The rate constants for the dissociation of products from the E.P complexes were measured a t pH 7 using a KinTek Instruments stopped flow spectrometer thermostated at 25 "C in the fluorescence mode with excitation at 290 nm and emission monitored through a 340-nm filter (Johnson, 1986). The binary enzyme-product complex (20 PM) was rapidly mixed with an equal volume of a 200 p~ solution of glutathione sulfonate (GSO,), a concentration sufficient to ensure that k,,,, >> kOff. The dissociation of either GSDNB, (R,R)-GSPhen, or (S,S)-GSPhen was followed by the increase in fluorescence with time.
Typically, four to eight such traces were averaged and the rate constant obtained by fitting the data to a single assay of the enzyme in the presence of a viscogen. The microviscosity of the solvent was varied by the addition of sucrose (0 to approximately 30 weight %) to the buffer for the reactions. This resulted in a range of relative viscosities (q/qo) between 1.0 and 2.7, where qo is the viscosity of the buffer in the absence of viscogen and q is viscosity of the reaction mixture. The relative viscosities were measured in an Ostwald viscometer at 25 "C.
RESULTS AND DISCUSSION X-ray Crystallographic Evidence for the Participation of !l?~r"~-The 1.8-A crystal structure of isoenzyme 3-3 in complex with product (S,S)-GSPhen (Scheme I) reveals a number of contacts between the xenobiotic portion of the product and the enzyme as illustrated in Fig. 1. Close contacts with residues in domain I (valine 9) and domain I1 (isoleucine 111 and serine 209) and the phenanthryl portion of the product are evident. Of particular interest is the fact that the hydroxyl group of Tyr115 is located very near the 10-hydroxyl group of the product. In fact, in one subunit ofthe dimer (shown in Fig. 1) the Tyr"'-OH is within hydrogen-bonding distance (3.3 A) of the product hydroxyl while in the other subunit it is slightly further away (3.9 A). The difference appears to be due to a crystal packing effect.
These observations are highly suggestive that the side chain of Tyr115 may also be quite close to the oxirane oxygen in the Michaelis complex with the substrate, PhenO. If the hydroxyl group of Tyr115 were to act as a hydrogen bond donor to the oxirane oxygen of the substrate, then it could provide electrophilic assistance in the transition state for oxirane ring opening. Kinetic Properties of the Y115F Mutant-The role posited for T y r 1 I 5 from the crystal structure can be tested by site-specific mutagenesis. Removal of the hydroxyl group from the T y r 1 1 5 has a rather dramatic effect on the enzyme-catalyzed addition of GSH to PhenO, reducing kcat and kcat/KmPheno by 2 orders of magnitude (Table I). There is little difference in the stereoselectivity of the Y115F mutant (51% (S,S)-GSPhen) and the native enzyme (43% (S,S)-GSPhen) toward PhenO. Thus, the severely impaired catalytic efficiency of the mutant is consistent with the removal of an interaction crucial for the chemical step of the reaction and not a deleterious change in the conformation of the active site. Given the proximity of the hydroxyl group of !@r115 to the hydroxyl group of the product in the crystal structure, it seems likely that the side chain of Tyr115 acts as a hydrogen bond donor to the oxirane oxygen which stabilizes, with roughly equal facility, the two diastereomeric transition states for opening of the oxirane ring. The magnitude of the transition state stabilization provided by the hydroxyl group, evaluated from the differences in kc,, and kcat/ 1 u from the 1.8-A structure of isoenzyme 3-3 in complex with FIG. 1. View of the 2 F , -F, electron density map contoured at (S,S)-GSPhen. Electron density for the product and residues that participate in van der Waals or hydrogen-bonding contacts with the 10hydroxy-9.10-dihydrophenanthryl portion of the product is shown.

TABLE I
Kinetic constants for the enzyme-catalyzed additions of GSH to PhenO and CDNB and rate constants for dissociation of the corresponding products All measurements were carried out a t pH 7.0 and 25 "C. KmPhen0 (Table I), is between 2.7 and 3.2 kcaymol a t 298 K.
Removal of the hydroxyl group of Tyr115 has quite the opposite effect on the addition of GSH to CDNB as shown in Table  I. The Y115F mutant is actually more efficient than the native enzyme both with respect to kcat and kCaJKmCDNB. Although the changes are modest, it is obvious that the hydroxyl group of Tyr115 in the native enzyme provides no catalytic benefit in this reaction. In fact, just the opposite is true. The rather striking difference in the response of the catalytic properties of the mutant enzyme derives from the different stereoelectronic demands of the transition states for the two reactions and the different rate-limiting steps for each reaction.
Catalysis and the Nature of the Rate-limiting Step-The rate of an enzyme-catalyzed reaction can be limited by a chemical step or a number of different physical steps in the mechanism including the formation of the encounter complex, the dissociation of product, or a conformational isomerization of the protein. In order to fully analyze the consequence of any mutation on catalysis, it is necessary to understand the mechanism of catalysis and to define the nature of the rate-limiting step. The second-order rate constants (kCat/KmS) for the formation of the encounter complexes ofE.GS-with either CDNB or PhenO (see Table I) are significantly less than estimates of the diffusioncontrolled limit (lOa--lO1o M -~ s-') for enzyme-catalyzed reactions. However, indirect evidence has suggested that the turnover of isoenzyme 3-3 in nucleophilic aromatic substitution reactions with reactive substrates may be limited by product release (Jakobson et al., 1977;Liu et al., 1992;Zhang et al., 1992). If this is true, the unimolecular rate constant for dissociation of the product from the E.P complex should be equivalent to kcat (e.g. koff = kcat). Direct measurement of k ,~ for GSDNB by stopped flow fluorescence shows that k,ff = kcat for isoenzyme 3-3 (Table I), a strong indication that the physical step of product dissociation limits turnover of the enzyme with CDNB. Similar experiments with the two GSPhen diastereomers give a different result, with rate constants for dissociation (2.6 and 4.4 s-l) that are some 5-10 times larger than the turnover number for PhenO, suggesting that the chemical step is rate-limiting with this substrate. It is interesting to note that the koa for all three products is increased about 2-fold for the Y115F m~t a n t .~ The frequency of diffusional encounter (or its microscopic reverse, separation) of two molecules is inversely proportional to the viscosity, q, of the medium as anticipated by the Stokes-Einstein relationship (Garners, 1940). If the turnover number of a n enzyme is limited by the rate of product release, then it is The fact that koR for GSDNB is less than k,,, for the Y115F mutant suggests that the observed fluorescence change may be the result of a conformational change of the protein, subsequent to binding of the trapping agent, that occurs with a rate constant of 32 s-l. Thus direct measurements of kOn > 32 s-l would not possible by this method.  (Cleland, 1979). T h e two lines represent the expected effect of the relative viscosity where the rate-limiting step is a diffusional process (slope = 1.0) and where diffusion is not rate-limiting (slope = 0).
expected that kcat will decrease as q increases. Thus, kca2 observed at a reference viscosity qo will be related to keat observed at some higher viscosity, q, in the presence of a viscogen by kCato/kcat = n/qo. A plot of the inverse relative rate constant versus the relative viscosity (q/qo) should be linear with a unit slope when the release of product is limited by a strictly diffusional barrier or should have a slope approaching zero if chemistry or another non-diffusional barrier is rate-limiting.4 The inverse relative kcat for the enzyme-catalyzed reaction with CDNB shows the expected linear dependence on the relative viscosity with a slope (1.05 2 0.08) very close to unity as illustrated in Fig. 2. In contrast, the reaction with PhenO shows no detectable dependence on viscosity. Moreover, the viscosity dependences of the Y115F-catalyzed reactions are the same as those catalyzed by the native enzyme, indicating that the mutation does not change the identity of the rate-limiting step in either reaction. Both sets of results support the conclusions that a chemical barrier limits the turnover of PhenO whereas the turnover of CDNB is limited by a diffusional barrier. 5r1I5 and the Diffusional Barrier-A possible contribution of the hydroxyl group of Tyr115 to the physical barrier for egress of the product from the active site in the native enzyme is suggested by the three-dimensional structure of the protein.

*115,
which is located on the a4-helix, is within hydrogenbonding distance of the side chain hydroxyl group of SerZo9 and the main chain amide nitrogen of the same residue as shown in Figs. 1 and 3. These interactions appear to tie together two of the major structural elements (the a4/a5 helix-turn-helix and the C-terminal tail) that help form the 19-A-deep channel to the active site. Thus the hydroxyl group of Tyr115 may restrain the segmental motion of structural elements that constitute the approach to the active site and impede the departure of products bound at the active site. In this sense Tyr115 acts as a cap on the channel to the active site. The cap is apparently loosened a bit on loss of the hydrogen-bonding interactions in the Y115F mutant.
Conclusions-Tyrosine 115 from the xenobiotic substrate binding domain of isoenzyme 3-3 of GSH transferase has been Intermediate values of the slope (0 < slope < 1) could be indicative of a viscosity-dependent segmental motion or conformational isomerization of the protein in which the solvent viscosity effect is damped by the "internal friction" of the protein motion (Ansari et al., 1992;Sampson and Knowles, 1992) or to a nonspecific effect of the viscogen on the reaction.
to-tyrosine 6 is also illustrated as a dashed line.
identified by x-ray crystallography and by comparison of the kinetic properties of the Y115F mutant to the native enzyme, as a residue that directly participates in catalysis. Two substrates, which have distinctly different requirements for transition state stabilization and different rate-limiting steps, were used to dissect the contribution of the hydroxyl group of Tyr115 to both chemical and physical (product release) steps in catalysis. Although the hydroxyl group appears to provide substantial electrophilic assistance in the addition of GSH to epoxides, accelerating the formation of product, it also seems to slow the egress of products from the active site. The net result of the Y115F mutation is to increase the substrate selectivity ratio (kcat/KmCDNBIkcat/Kmpheno) from about 70 in the native enzyme to close to 26,000 in the mutant.