Transmutation of Human Glutathione Transferase A2-2 with Peroxidase Activity into an Efficient Steroid Isomerase

Abstract

A major goal in protein engineering is the tailor-making of enzymes for specified chemical reactions. Successful attempts have frequently been based on directed molecular evolution involving libraries of random mutants in which variants with desired properties were identified. For the engineering of enzymes with novel functions, it would be of great value if the necessary changes of the active site could be predicted and implemented. Such attempts based on the comparison of similar structures with different substrate selectivities have previously met with limited success. However, the present work shows that the knowledge-based redesign restricted to substrate-binding residues in human glutathione transferase A2-2 can introduce high steroid double-bond isomerase activity into the enzyme originally characterized by glutathione peroxidase activity. Both the catalytic center activity (kcat) and catalytic efficiency (kcat/Km) match the values of the naturally evolved glutathione transferase A3-3, the most active steroid isomerase known in human tissues. The substrate selectivity of the mutated glutathione transferase was changed 7000-fold by five point mutations. This example demonstrates the functional plasticity of the glutathione transferase scaffold as well as the potential of rational active-site directed mutagenesis as a complement to DNA shuffling and other stochastic methods for the redesign of proteins with novel functions. A major goal in protein engineering is the tailor-making of enzymes for specified chemical reactions. Successful attempts have frequently been based on directed molecular evolution involving libraries of random mutants in which variants with desired properties were identified. For the engineering of enzymes with novel functions, it would be of great value if the necessary changes of the active site could be predicted and implemented. Such attempts based on the comparison of similar structures with different substrate selectivities have previously met with limited success. However, the present work shows that the knowledge-based redesign restricted to substrate-binding residues in human glutathione transferase A2-2 can introduce high steroid double-bond isomerase activity into the enzyme originally characterized by glutathione peroxidase activity. Both the catalytic center activity (kcat) and catalytic efficiency (kcat/Km) match the values of the naturally evolved glutathione transferase A3-3, the most active steroid isomerase known in human tissues. The substrate selectivity of the mutated glutathione transferase was changed 7000-fold by five point mutations. This example demonstrates the functional plasticity of the glutathione transferase scaffold as well as the potential of rational active-site directed mutagenesis as a complement to DNA shuffling and other stochastic methods for the redesign of proteins with novel functions. glutathione transferase Δ5-androstene-3,17-dione glutathione The engineering of a catalytically active protein involves the acquisition of determinants for substrate recognition, installment of a catalytic apparatus, and provision of a structural scaffold with the dynamic properties necessary for supporting catalysis. Accepting the view that only a finite number (a few thousand) of protein folds exist (1Thornton J.M. Orengo C.A. Todd A.E. Pearl F.M.G. J. Mol. Biol. 1999; 293: 333-342Crossref PubMed Scopus (137) Google Scholar, 2Wolf Y.I. Grishin N.V. Koonin E.V. J. Mol. Biol. 2000; 299: 897-905Crossref PubMed Scopus (155) Google Scholar), we conclude that the vast majority of the naturally occurring proteins have evolved by recruitment and redesign of existing structural modules. For the engineering of a new enzyme, it would consequently be feasible to select a suitable polypeptide scaffold and then insert new building blocks in the form of essential amino acid residues, secondary structure elements, or even new domains. Given that the fold is maintained, the question is whether it is sufficient to build the active-site cavity with the limited number of residues contacting the substrate or whether additional more remote residues have to be included in the design (Fig. 1). In the first alternative, it would appear possible to make a rational design based on the computational algorithms optimizing the choice of functional groups, active-site topography, various anisotropic parameters representing charge density, and hydrophobicity among other variables. For the second alternative, it is not obvious that the design is computationally tractable on the basis of available knowledge. Despite numerous investigations, it is still not clear to what extent it is possible to rationally change the substrate specificity and obtain an enzyme with high activity by mutating amino acid residues that determine substrate selectivity in the active site (4Penning T.M. Jez J.M. Chem. Rev. 2001; 101: 3027-3046Crossref PubMed Scopus (114) Google Scholar). For example, aspartate transaminase was made more efficient with the cationic amino acids Lys and Arg than with its natural anionic substrates but at the expense of a dramatic loss of catalytic efficiency (5Cronin C.N. Malcolm B.A. Kirsch J.F. J. Am. Chem. Soc. 1987; 109: 2222-2223Crossref Scopus (39) Google Scholar). Rational modifications of active-site residues in a phosphotriesterase elevated the activity (kcat) with the alternative substrate diisopropylfluorophosphate up to 10-fold, but the catalytic efficiency (kcat/Km) was not increased >2-fold because of increased Km values in all mutants (6Watkins L.M. Mahoney H.J. McCulloch J.K. Raushel F.M. J. Biol. Chem. 1997; 272: 25596-25601Abstract Full Text Full Text PDF PubMed Scopus (57) Google Scholar). Several attempts at redesigning enzymes suggest that the mutations distant from the active site may be necessary complements to obvious active-site modifications (4Penning T.M. Jez J.M. Chem. Rev. 2001; 101: 3027-3046Crossref PubMed Scopus (114) Google Scholar, 7Oue S. Okamoto A. Yano T. Kagamiyama H. J. Biol. Chem. 1999; 274: 2344-2349Abstract Full Text Full Text PDF PubMed Scopus (157) Google Scholar). The conversion of trypsin into a mutant with chymotrypsin-like activity required surface loops to be modified in addition to the substrate specificity pocket to approach the catalytic properties of the naturally evolved chymotrypsin (8Hedstrom L. Szilagyi L. Rutter W.J. Science. 1992; 255: 1249-1253Crossref PubMed Scopus (452) Google Scholar). Major efforts are currently focused on computational approaches to the tailoring of proteins for new functions (9Voigt C.A. Mayo S.L. Arnold F.H. Wang Z.-G. Proc. Natl. Acad. Sci. U. S. A. 2001; 98: 3778-3783Crossref PubMed Scopus (146) Google Scholar, 10Bolon D.N. Mayo S.L. Proc. Natl. Acad. Sci. U. S. A. 2001; 98: 14274-14279Crossref PubMed Scopus (336) Google Scholar, 11Kraemer-Pecore C.M. Wollacott A.M. Desjarlais J.R. Curr. Opin. Chem. Biol. 2001; 5: 690-695Crossref PubMed Scopus (50) Google Scholar, 12Looger L.L. Hellinga H.W. J. Mol. Biol. 2001; 307: 429-445Crossref PubMed Scopus (160) Google Scholar). The molecular recognition of the substrate and the complementary binding of transition states may involve contacts with ≤10–15 active-site residues (Fig. 1) that govern substrate specificity and catalytic efficiency of the enzyme. Successful examples of redesign by rational modification of active-site residues leaving the remainder of the protein scaffold intact will be consequential to protein engineering and clarify options in the natural evolution of enzyme function. Tailor-making enzymes for novel functions is largely based on directed molecular evolution of parental structures (13Arnold F.H. Nature. 2001; 409: 253-257Crossref PubMed Scopus (344) Google Scholar, 14Minshull J. Stemmer W.P.C. Curr. Opin. Chem. Biol. 1999; 3: 284-290Crossref PubMed Scopus (99) Google Scholar). The redesign would be more powerful if necessary active-site modifications could be predicted and implemented. We tested the feasibility of rational redesign based on homology by exchanging active-site residues in glutathione transferase (GST)1 A2-2 to install the high 3-ketosteroid isomerase activity of the homologous GST A3-3. Five amino acid substitutions afforded a 3500-fold increase in catalytic efficiency with Δ5-androstene-3,17-dione (AD). The substrate selectivity of GST A2-2 changed 7000-fold from cumene hydroperoxide to AD, demonstrating functional plasticity and a notable switch from peroxidase into steroid isomerase activity. PCR with the proof-reading Pfu DNA polymerase and mutagenic oligonucleotides introduced the point mutations in the GSTA2 cDNA sequence in pGEM (Promega Co., Madison, WI). All of the nucleotide sequences were confirmed subsequent to PCR. For protein expression, wild-type and mutant GST cDNA were subcloned into the pET-21 vector (Novagen, Inc., Madison, WI). Expression and purification of wild-type GSTs A2-2, A3-3, and mutants were made as described previously (15Pettersson P.L. Mannervik B. J. Biol. Chem. 2001; 276: 11698-11704Abstract Full Text Full Text PDF PubMed Scopus (44) Google Scholar, 16Gustafsson A. Mannervik B. J. Mol. Biol. 1999; 288: 787-800Crossref PubMed Scopus (31) Google Scholar, 17Johansson A.-S. Mannervik B. J. Biol. Chem. 2002; 277: 16648-16654Abstract Full Text Full Text PDF PubMed Scopus (42) Google Scholar) using a HiTrap SP cation exchange column (AmershamBiosciences). The activity with AD was measured in 50 mm Tris-HCl at pH 8.0 (15Pettersson P.L. Mannervik B. J. Biol. Chem. 2001; 276: 11698-11704Abstract Full Text Full Text PDF PubMed Scopus (44) Google Scholar), and the activities with 1-chloro-2,4-dinitrobenzene and cumene hydroperoxide were assayed in 100 mm sodium phosphate at pH 6.5 (18Mannervik B. Jemth P. Maines M.D. Costa L.G. Reed D.J. Sassa S. Sipes I.G. Current Protocols in Toxicology. John Wiley & Sons, Inc., New York1991: 6.4.1-6.4.10Google Scholar). Specific activities with cumene hydroperoxide were determined in the presence of 2 mm glutathione (GSH) to approach saturation. All of the measurements were made at 30 °C. The ionization of the thiol group of GSH was monitored by measuring the absorbance of the thiolate at 239 nm (Δε239 = 5200 m−1cm−1 (19Gustafsson A. Etahadieh M. Jemth P. Mannervik B. Biochemistry. 1999; 38: 16268-16275Crossref PubMed Scopus (47) Google Scholar)) as a function of pH. The spectra of GSH bound to enzyme were obtained by subtracting the spectra of free GSH at different pH values and of free enzyme at pH 5.8 from the spectrum of the sample. The enzyme (subunit) and GSH concentrations were 15 and 500 μm, respectively. The pKa value of Tyr-9 in the absence of GSH was determined by subtracting the spectrum at pH 5.5 from the spectra obtained at higher pH values and calculated from the absorbance of the tyrosinate at 253 nm (Δε253= 11000 m−1 cm−1 (19Gustafsson A. Etahadieh M. Jemth P. Mannervik B. Biochemistry. 1999; 38: 16268-16275Crossref PubMed Scopus (47) Google Scholar)). The buffers were 100 mm sodium phosphate, pH 5.8–7.9, and ethanolamine-HCl, pH 8.2–9.6. To test a redesign restricted to active-site residues, we decided to attempt the conversion of human GST A2-2 with non-selenium-dependent glutathione peroxidase activity into a variant GST with steroid double bound isomerase activity. Wild-type A2-2 is the human GST with the highest peroxidase activity (20Zhao T. Singhal S.S. Piper J.T. Cheng J. Pandya U. Clark-Wronski J. Awasthi S. Awasthi Y.C. Arch. Biochem. Biophys. 1999; 367: 216-224Crossref PubMed Scopus (108) Google Scholar), but as a steroid isomerase, it is a very poor catalyst in comparison with the homologous human GST A3-3 (Table I). The GST-catalyzed isomerization of AD involves acid-base catalysis promoted by Tyr-9 with the thiolate of glutathione serving as a catalytic cofactor (15Pettersson P.L. Mannervik B. J. Biol. Chem. 2001; 276: 11698-11704Abstract Full Text Full Text PDF PubMed Scopus (44) Google Scholar). In the peroxidase reaction, the same thiolate assisted by Tyr-9 makes a nucleophilic attack on oxygen in the hydroperoxide substrate. The reduction of the hydroperoxide proceeds in two steps via an unstable glutathione sulfenic acid conjugate and leads to the formation of glutathione disulfide (Fig. 2). The catalytic residues are common to the homologous Alpha class GSTs, but still the catalytic efficiency of GST A3-3 with AD is 5000-fold higher than that of GST A2-2 (Table II). The protein sequences of 221 residues are 89% identical, and of the 25 variant amino acids, five (Fig. 1) are located in the binding site of the hydrophobic substrate (H-site). Apparently, the limited structural differences between GST A2-2 and GST A3-3 govern the differential utilization of the intrinsic reactivity of the glutathione sulfur atom.Table ISpecific activities with the substrates AD, CDNB and CuOOH for wild-type human GST A2–2, GST A3–3, and GST A2–2 mutantsGSTTerminologySpecific activities (μmol/min−1/mg−1)ADCDNBCuOOHA2–2, wild-typeA2–20.155821A2–2/S10F/I12GN-Double530NDA2–2/M208A/S216AC-Double337NDA2–2/S10F/I12G/M208A/S216AQuad1058NDA2–2/F111LF111L36023A2–2/S10F/I12G/F111LN-Triple23755.1A2–2/F111L/M208A/S216AC-Triple38276.2A2–2/S10F/I12G/F111L/M208A/S216APenta230643.6A2–2/S10F/112G/D42G/F111L/M208/S216AHexa22073NDA3–3, wild-typeA3–3360303.7 Open table in a new tab Table IICatalytic constants kcat and specificity constants kcat/Km for wild-type human GST A2–2, GST A3–3, and GST A2–2 mutants with the substrates AD and CuOOH.GSTADCuOOHkcatkcat/Kmkcatkcat/Kms−1mM−1s−1s−1mM−1s−1A2–20.28 ± 0.021.0 ± 0.133 ± 329 ± 0.3F111L6 ± 111 ± 134 ± 839 ± 2N-triple34 ± 2140 ± 1010 ± 125 ± 1C-triple44 ± 2300 ± 1728 ± 2011.6 ± 0.1Penta195 ± 113500 ± 4506.0 ± 0.514.6 ± 0.6A3–3228 ± 95000 ± 3505.9 ± 0.79.9 ± 0.5 Open table in a new tab The five variant H-site residues in GST A2-2 were mutated into the residues of GST A3-3 with the goal of installing increased steroid isomerase activity. The mutations were introduced sequentially and in different combinations, and the highest activity was obtained when all of the five amino acid substitutions were present in GST A2-2 mutant Penta (Table I). The H-site is formed by three regions of the primary structure: the α1-β1 loop (residues 9–15), the C-terminal portion of α4 (residues 104–111), and α9 with flanking amino acids (residues 208–222). Table I shows that all of the three regions contribute to the steroid isomerase activity. A sixth residue distinguishing GST A2-2 from GST A3-3 (Asp-42 into Gly) is located in helix α2 in which Pi class GSTs have been shown to be highly mobile and linked to catalysis (21Stella L., Di Iorio E.E. Nicotra M. Ricci G. Proteins Struct. Funct. Genet. 1999; 37: 10-19Crossref PubMed Scopus (21) Google Scholar, 22Hitchens T.K. Mannervik B. Rule G.S. Biochemistry. 2001; 40: 11660-11669Crossref PubMed Scopus (25) Google Scholar). However, this residue is not part of the H-site, and the mutation of Asp-42 to give GST A2-2 mutant Hexa did not add to the isomerase activity (Table I). The mutation of a seventh variant residue, S18P, did not markedly alter the specific activity with any of the substrates used (AD, 0.09 μmol min−1mg−1; 1-chloro-2,4-dinitrobenzene, 38 μmol min−1mg−1). Residue 18 is located in helix α1 leading away from the active site, and Pro is potentially a helix breaker. The gain of isomerase activity afforded in the GST A2-2 mutants is primarily akcat effect (Fig.3). The Km values are only moderately changed. It is noteworthy that the mutations did not markedly alter the kinetic parameters for the standard GST substrate 1-chloro-2,4-dinitrobenzene (cf. Table I). Furthermore, the high glutathione peroxidase activity of wild-type GST A2-2 was only moderately decreased by the mutations (Table II). Thus, the main result of the mutagenesis was a highly selective installment of a 3500-fold increased catalytic efficiency in the steroid isomerase reaction. Expressed in terms of relative substrate specificity (ratio betweenkcat/Km values), mutant Penta has acquired enhanced substrate selectivity for steroid isomerization over hydroperoxide reduction by a factor of 7000-fold. GST A3-3 is characterized by an active-site Tyr-9 of unusual acidity (pKa 7.9). The significantly lower acidity of the corresponding Tyr-9 in GST A2-2 (pKa 9.2) is increased in mutant Penta (pKa 8.2) approaching that of GST A3-3. Thus, a low pKa value of the phenolic group is reflected in an enhanced catalytic activity in the steroid isomerization. In contrast, no such correlation is apparent with the other substrates tested (Table I). The pKa value of the thiol group of enzyme-bound glutathione is also low compared with the value for glutathione in aqueous solution (pKa9.2). However, in this case, the differences among GST A2-2 (pKa 6.8), GST A3-3 (pKa 6.1), and the mutant Penta (pKa 6.2) are less pronounced. It is noteworthy that the H-site mutations of uncharged residues distant from the ionizable groups of Tyr-9 and glutathione significantly increase their acidities and selectively enhance the isomerization activity. The isomerization of the Δ5- to the Δ4-keto steroid involves an intermediate dienolate with an extended planar π-orbital system (23Thornburg L.D. Goldfeder Y.R. Wilde T.C. Pollack R.M. J. Am. Chem. Soc. 2001; 123: 9912-9913Crossref PubMed Scopus (14) Google Scholar). The stabilization of this planar structure relative to the puckered conformation of the substrate AD is expected to provide transition state stabilization in the catalyzed reaction. The 3500-fold increasedkcat/Km value in mutant Penta compared with wild-type GST A2-2 corresponds to an incremental stabilization of the transition state (ΔΔG =R T ln3500, where G is the Gibbs-free energy, R is the gas constant, T is the absolute temperature, and ln is natural logarithm) of 21 kJ/mol. On the basis of the homologous GST A1-1 structure (Fig. 1) (3Cameron A.D. Sinning I. L'Hermite G. Olin B. Board P.G. Mannervik B. Jones T.A. Structure. 1995; 3: 717-727Abstract Full Text Full Text PDF PubMed Scopus (170) Google Scholar, 24Sinning I. Kleywegt G.J. Cowan S.W. Reinemer P. Dirr H.W. Huber R. Gilliland G.L. Armstrong R.N., Ji, X. Board P.G. Olin B. Mannervik B. Jones T.A. J. Mol. Biol. 1993; 232: 192-212Crossref PubMed Scopus (415) Google Scholar), we conclude that the H-site residue 111 is in a sterically critical position for this transition state stabilization to occur. In the low activity enzyme GST A2-2, this residue is Phe, whereas in GST A3-3, the residue is Leu (Table III). Apparently, the bulky and conformationally restricted benzyl group of Phe-111 in GST A2-2 is unfavorable for catalytically productive interactions with the steroid structure. However, for maximal catalytic efficiency, the H-site has to be further adjusted by mutations in addition to F111L (Tables I and II). This finding is in accord with the notion that the fine tuning of substrate-enzyme interactions is critical for optimized catalytic efficiency (25Koshland D.E., Jr. Nat. Med. 1998; 4: 1112-1114Crossref PubMed Scopus (91) Google Scholar).Table IIIActive-site residues and catalytic efficiencies with AD distinguishing human GST A1–1, GST A2–2, and GST A3–3GST (wild-type)Amino acid positionCatalytic efficiency1012111208216kcat/KmmM−1 s−1A2–2SerIlePheMetSer1.0 ± 0.1A1–1PheAlaValMetAla500 ± 403-aFrom Ref. ± Ref. 12Looger L.L. Hellinga H.W. J. Mol. Biol. 2001; 307: 429-445Crossref PubMed Scopus (160) Google Open table in a new tab GST which has a catalytic efficiency with AD in between that of GST A2-2 and GST A3-3, is more similar in the H-site to the most efficient GST A3-3 (Table III). Residue 111 in GST A1-1 is similar to Leu in GST A3-3. In residue is in GST A1-1 and GST A3-3. In two of the variant H-site residues, GST A1-1 and GST A3-3, are The most is residue in GST A3-3, and the more bulky in GST Thus, the H-site residues in GST A1-1 a for with AD, which only moderately different from that of GST A3-3 that is reflected in the in transition state stabilization of (ΔΔG = R T It has been that the of the catalyzed is a in the evolution of enzymes with new activities J. Biol. Chem. 1997; 272: Full Text Full Text PDF PubMed Scopus Google Scholar, J.M. M. J. 2001; Full Text Full Text PDF PubMed Scopus Google Scholar). The glutathione peroxidase activity of GSTs is based on nucleophilic attack of the thiolate of enzyme-bound and the steroid isomerization involves the same thiolate as a The common is the sulfur serving as a in with a acid in the form of oxygen of the hydroperoxide and in the second with an (pKa of the Despite a major is still that the involves the of glutathione as a whereas the isomerization not glutathione but the thiol as a catalytic cofactor in the active site of the enzyme (Fig. 2). The present shows that active can be in protein by a rational of amino acid residues that contribute to interactions with the desired substrate (Fig. 1). Penta of GST A2-2 has acquired not only the and kcat/Km values AD as a substrate for the naturally evolved GST A3-3 but also by two of the catalytic efficiency of the isomerase as the double-bond isomerization in steroid J. Biochem. Mol. Biol. Scopus Google Scholar). The acquisition of efficient steroid double-bond isomerase activity by a number of mutations in the enzyme GST A2-2 to the notion of GSTs as a of enzymes with great functional plasticity A. Mannervik B. Proc. Natl. Acad. Sci. U. S. A. 2000; PubMed Scopus Google Scholar, Proc. Natl. Acad. Sci. U. S. A. 2000; PubMed Scopus Google Scholar).