The crystal structure of a mu class glutathione S-transferase (EC 2.5.1.18) from rat liver (isoenzyme 3-3) in complex with the physiological substrate glutathione (GSH) has been solved at 2.2-angstrom resolution by multiple isomorphous replacement methods. The enzyme crystallized in the monoclinic space group C2 with unit cell dimensions of a = 87.98 angstrom, b = 69.41 angstrom, c = 81.34 angstrom, and beta = 106.07-degrees. Oligonucleotide-directed site-specific mutagenesis played an important role in the solution of the structure in that the cysteine mutants C86S, C114S, and C173S were used to help locate the positions of mercuric ion sites in nonisomorphous derivatives with ethylmercuric phosphate and to align the sequence with the model derived from MIR phases. A complete model for the protein was not obtained until part of the solvent structure was interpreted. The dimer in the asymmetric unit refined to a crystallographic R = 0.171 for 19 298 data and I greater-than-or-equal-to 1.5sigma(I). The final model consists of 4150 atoms, including all non-hydrogen atoms of 434 amino acid residues, two GSH molecules, and oxygen atoms of 474 water molecules. The dimeric enzyme is globular in shape with dimensions of 53 X 62 X 56 angstrom. Crystal contacts are primarily responsible for conformational differences between the two subunits which are related by a noncrystallographic 2-fold axis. The structure of the type 3 subunit can be divided into two domains separated by a short linker, a smaller alpha/beta domain (domain I, residues 1-82), and a larger alpha domain (domain II, residues 90-217). Domain I contains four beta-strands which form a central mixed beta-sheet and three alpha-helices which are arranged in a betaalphabetaalphabetabetaalpha motif. Domain II is composed of five alpha-helices. Domain I can be considered the glutathione binding domain, while domain II seems to be primarily responsible for xenobiotic substrate binding. The active site is located in a deep (19-angstrom) cavity which is composed of three relatively mobile structural elements: the long loop (residues 33-42) of domain I, the alpha4/alpha5 helix-turn-helix segment, and the C-terminal tail. GSH is bound at the active site in an extended conformation at one end of the beta-sheet of domain I with its backbone facing the cavity and the sulfur pointing toward the subunit to which it is bound. Fifteen hydrogen bond or salt-bridge contacts are involved in the interaction between GSH and the enzyme, among which the hydrogen bond (3.2 angstrom) between the sulfur of GSH and the side-chain hydroxyl group of Y6 is of special interest. This interaction appears to lower the pK(a) of the sulfhydryl group [Liu, S., Zhang, P., Ji, X., Johnson, W. W., Gilliland, G. L., & Armstrong, R. N. (I 992) J. Biol. Chem. 267, 4296-4299] and appears to help orient the thiolate anion in the active site such that half of its surface is shielded by hydrophobic residues and the other half is exposed for reaction with the electrophilic substrate. The nucleophilic reactivity of the GSH appears to be enhanced by both destabilization of the thiol and desolvation of the thiolate anion. The structure of the class mu isoenzyme is compared to that of a pi class isoenzyme which was recently reported at a preliminary stage of refinement [Reinemer, P., Dirr, H. W., Ladenstein, R., Schaffer, J., Gallay, O., & Huber, R. (1991) EMBO J. 10, 1997-2005].
