Enzymatic Transition State Theory and Transition State Analogue Design

Abstract

The incredible catalytic rate enhancements caused by enzymes led Linus Pauling (1Pauling L. Chem. Eng. News. 1946; 24: 1375-1377Crossref Scopus (729) Google Scholar) to suggest that enzymes bind tightly to substrates distorted toward the transition state, thereby concentrating them and enforcing catalysis. Wolfenden (2Wolfenden R. Nature. 1969; 223: 704-705Crossref PubMed Scopus (338) Google Scholar) explained that chemically stable analogues that resemble the transition state would be expected to bind more tightly than substrate by factors resembling the rate enhancement imposed by enzymes. The theory for tight binding of transition state analogues was supported by natural product chemistry and synthetic approaches to mimics of proposed enzymatic transition states (3Wolfenden R. Acct. Chem. Res. 1972; 5: 10-18Crossref Scopus (544) Google Scholar, 4Lienhard G.E. Science. 1973; 180: 149-154Crossref PubMed Scopus (417) Google Scholar, 5Wolfenden R. Biophys. Chem. 2003; 105: 559-572Crossref PubMed Scopus (66) Google Scholar). The well documented tight binding of transition state analogues confirms the thermodynamic aspects of tight binding by mimics of enzymatic transition states. Recently, protein dynamic motion has been proposed to account for catalysis without the necessity of tight binding at the transition state, where the transition state is formed by the instantaneous and optimal alignment of functional groups at the catalytic site (6Schramm V. Arch. Biochem. Biophys. 2005; 433: 13-26Crossref PubMed Scopus (111) Google Scholar). Single molecule kinetics of enzymes supports the dynamic search mode of catalysis, with individual catalytic events showing a wide range of time intervals that average to the observed collective property of the enzyme (7English B. W. Min P. van Oijen A.M. Lee K.T. Luo G. Sun H. Cherayil B.J. Kou S.C. Xie X.S. Nat. Chem. Biol. 2006; 2: 87-94Crossref PubMed Scopus (629) Google Scholar). In the dynamic theory of catalysis, tight binding of a chemically stable transition state analogue arises from a conformational collapse of the protein around the inhibitor (8Pineda J.R. Schwartz S.D. Philos. Trans. R. Soc. Lond. B Biol. Sci. 2006; 361: 1433-1438Crossref PubMed Scopus (34) Google Scholar). The presence of a stable, attractive analogue causes a conformational convergence to the transition state geometry. Without catalysis the analogue forms a tightly bound complex. The dynamics of transition state formation is converted into static binding energy. Transition state analysis of enzymatic reactions and the use of transition state (TS) 2The abbreviations used are: TStransition stateKIEkinetic isotope effectMESPmolecular electrostatic potentialPNPpurine nucleoside phosphorylasehPNPhuman PNPbPNPbovine PNPMEPSmolecular electrostatic potential surfacesDADMe-Immucillin-H4′-deaza-1′-aza-2′-deoxy-1′,9-methylene derivative of Immucillin-HMTAmethylthioadenosineMTAPmethylthioadenosine phosphorylaseSAMS-adenosylmethionineMTANmethylthioadenosine nucleosidaseSAHS-adenosylhomocysteine. information to design transition state analogues requires: 1) a target enzyme with chemistry suited for kinetic isotope effect (KIE) analysis; 2) substrates with isotopic substitutions at the reaction center; 3) intrinsic KIEs (isotope effects from the chemical step); 4) a computed transition state matching the intrinsic KIEs; 5) a molecular electrostatic potential (MESP) map of the TS; 6) a stable compound to match the MESP map; and 7) testing of the TS analogue against the target enzyme. transition state kinetic isotope effect molecular electrostatic potential purine nucleoside phosphorylase human PNP bovine PNP molecular electrostatic potential surfaces 4′-deaza-1′-aza-2′-deoxy-1′,9-methylene derivative of Immucillin-H methylthioadenosine methylthioadenosine phosphorylase S-adenosylmethionine methylthioadenosine nucleosidase S-adenosylhomocysteine. This procedure has been developed gradually in parallel with the advances in KIE enzymology, computational chemistry, and synthetic organic chemistry with numerous laboratories making important contributions (9Bigeleisen J. Wolfsberg M. Adv. Chem. Phys. 1958; 1: 15-76Google Scholar, 10Sims L.B. Fry A. Netherton L.T. Wilson J.C. Reppond K.D. Crook S.W. J. Am. Chem. Soc. 1972; 94: 1364-1373Crossref Scopus (63) Google Scholar, 11Northrop D.B. Ann. Rev. Biochem. 1981; 50: 103-131Crossref PubMed Scopus (260) Google Scholar, 12Cleland W.W. Methods Enzymol. 1982; 87: 625-641Crossref PubMed Scopus (48) Google Scholar, 13Rodgers J. Femec D.A. Schowen R.L. J. Am. Chem. Soc. 1982; 104: 3263-3268Crossref Scopus (87) Google Scholar, 14Schramm V.L. Annu. Rev. Biochem. 1998; 67: 693-720Crossref PubMed Scopus (245) Google Scholar, 15Schramm V.L. Methods Enzymol. 1999; 308: 301-355Crossref PubMed Scopus (83) Google Scholar). Some examples of this newly developing field are provided here. Purine nucleoside phosphorylase has been a biochemical target for T-cell proliferation diseases since 1975 when Eloise Giblett (16Giblett E.R. Ammann A.J. Wara D.W. Sandman R. Diamond L.K. Lancet. 1975; 1: 1010-1013Abstract PubMed Scopus (626) Google Scholar) discovered that the human genetic deficiency of PNP led to T-cell immune deficiency, with other blood cells and tissues being normal. The metabolic defect results in the accumulation of 2′-deoxyguanosine in blood. Deoxyguanosine is phosphorylated to generate excess dGTP only in dividing T-cells (16Giblett E.R. Ammann A.J. Wara D.W. Sandman R. Diamond L.K. Lancet. 1975; 1: 1010-1013Abstract PubMed Scopus (626) Google Scholar). Bovine PNP (bPNP) was used as a surrogate for the human enzyme because at 87% amino acid sequence identity it was assumed that the transition state structures would be identical for bovine and human PNPs. The KIEs were measured for arsenolysis of inosine isotopically labeled in seven different positions, including 5′-14C as a remote label control (Fig. 1; Ref. 17Kline P.C. Schramm V.L. Biochemistry. 1993; 32: 13212-13219Crossref PubMed Scopus (205) Google Scholar). Binding of the TS analogue Immucillin-H is 739,000 tighter than substrate binding as judged by the Km value for inosine (18Miles R.W. Tyler P.C. Furneaux R.H. Bagdassarian C.K. Schramm V.L. Biochemistry. 1998; 37: 8615-8621Crossref PubMed Scopus (256) Google Scholar). Tight binding of TS analogues is dependent on both the geometry and charge of the analogue resembling the transition state more closely than the substrate (19Bagdassarian C.K. Schramm V.L. Schwartz S.D. J. Am. Chem. Soc. 1996; 118: 8825-8836Crossref Scopus (63) Google Scholar). Geometric and electrostatic similarity is apparent in the molecular electrostatic potential surfaces (MEPS) for the TS of bPNP compared with Immucillin-H, and both of these differ from the MEPS of the substrate (Fig. 2; Ref. 18Miles R.W. Tyler P.C. Furneaux R.H. Bagdassarian C.K. Schramm V.L. Biochemistry. 1998; 37: 8615-8621Crossref PubMed Scopus (256) Google Scholar). Immucillin-H was found to be a 56 pm inhibitor of human PNP (hPNP). It was a surprise to find that that binding of Immucillin-H differed for bPNP and hPNP given the 87% amino acid overall sequence identity and 100% conservation at the catalytic sites. Yet, the difference in TS analogue binding suggested different transition states. The TS structure of hPNP was established from KIEs and computational analysis (20Lewandowicz A. Schramm V.L. Biochemistry. 2004; 43: 1458-1468Crossref PubMed Scopus (143) Google Scholar). A comparison of the KIE values for bPNP (Fig. 1) with those for hPNP (Fig. 3) established different KIE values. Thus, the TS structures are different. The TS of hPNP is distinguished from that for bPNP by: 1) the increased distance between the ribosyl group and the hypoxanthine leaving group; 2) the increased cationic charge at C-1′ (the anomeric carbon), because unlike bPNP, electrons are not effectively shared across the 3-Å distance to the leaving group; and 3) tolerance for the 2′-deoxyribosyl analogue, because the physiological substrate for hPNP is 2′-deoxyguanosine. These features of the transition state for hPNP led to the design and synthesis of DADMe-Immucillin-H (21Lewandowicz A. Tyler P.C. Evans G.B. Furneaux R.H. Schramm V.L. J. Biol. Chem. 2003; 278: 31465-31468Abstract Full Text Full Text PDF PubMed Scopus (113) Google Scholar, 22Evans G.B. Furneaux R.H. Lewandowicz A. Schramm V.L. Tyler P.C. J. Med. Chem. 2003; 46: 5271-5276Crossref PubMed Scopus (99) Google Scholar). DADMe-Immucillin-H is a more powerful inhibitor for hPNP than Immucillin-H is for bPNP (compare Figs. 1 and 3). Moreover, DADMe-Immucillin-H shows 8-fold higher specificity for human than bovine PNP (Fig. 3). Thus, detailed transition state information provides a powerful tool for the design of tight binding and specific transition state analogues, even for closely related isozymes with high amino acid sequence homology. Attaining high specificity and high affinity with the synthesis of only a few molecules is unprecedented in drug design and establishes the value of transition state information. The difference between the transition states for human and bovine PNPs is also evident in their MEPS (Fig. 4). Comparison of the charge and geometry between inosine, transition state, and DADMe-Immucillin-H establishes the high similarity between the transition state and the inhibitor. Accordingly, DADMe-Immucillin-H binds 2,400,000 times tighter than substrate according to the Km/Kd ratio (21Lewandowicz A. Tyler P.C. Evans G.B. Furneaux R.H. Schramm V.L. J. Biol. Chem. 2003; 278: 31465-31468Abstract Full Text Full Text PDF PubMed Scopus (113) Google Scholar). Chemical synthesis of Immucillin-H requires the incorporation of four stereochemical centers and is inherently difficult (23Evans G.B. Furneaux R.H. Hutchison T.L. Kezar H.S. Morris P.E. Schramm Jr., V.L. Tyler P.C. J. Org. Chem. 2001; 66: 5723-5730Crossref PubMed Scopus (94) Google Scholar, 24Furneaux R.H. Tyler P.C. J. Org. Chem. 1999; 64: 8411-8412Crossref PubMed Scopus (64) Google Scholar). In contrast, DADMe-Immucillin-H has two stereocenters and can be made using the Mannich reaction, a three-way condensation of 9-deazahypoxanthine, 3-hydroxy-4-methoxy-pyrrolidine and formaldehyde under mild aqueous conditions (25Evans G.B. Furneaux R.H. Tyler P.C. Schramm V.L. Org. Lett. 2003; 5: 3639-3640Crossref PubMed Scopus (82) Google Scholar). Pharmacological applications of powerful inhibitors depend on their biological availability, specificity for their targets, and time of action. Oral administration of either Immucillin-H or DADMe-Immucillin-H causes rapid whole body inhibition of PNP in mice as indicated by the activity of the enzyme in the blood (21Lewandowicz A. Tyler P.C. Evans G.B. Furneaux R.H. Schramm V.L. J. Biol. Chem. 2003; 278: 31465-31468Abstract Full Text Full Text PDF PubMed Scopus (113) Google Scholar). A single oral dose of Immucillin-H gave complete inhibition of blood PNP. Recovery time for 50% of normal blood PNP activity was 4 days. Thus, the inhibitor has oral availability and a long lifetime on its target. A single oral dose of DADMe-Immucillin-H caused rapid inhibition of mouse blood PNP and return of blood PNP activity with a t½ of 11.5 days, the time for erythrocyte replacement by hematopoesis (21Lewandowicz A. Tyler P.C. Evans G.B. Furneaux R.H. Schramm V.L. J. Biol. Chem. 2003; 278: 31465-31468Abstract Full Text Full Text PDF PubMed Scopus (113) Google Scholar). Immucillin-H and DADMe-Immucillin-H have been tested for toxicity in animal studies, and both inhibitors have entered human clinical trials (26Balakrishnan K. Nimmanapalli R. Ravandi F. Keating M.J. Gandhi V. Blood. 2006; 108: 2392-2398Crossref PubMed Scopus (74) Google Scholar, 27Ravandi F. Gandhi V. Expert Opin. Investig. Drugs. 2006; 15: 1601-1613Crossref PubMed Scopus (43) Google Scholar, 28Gandhi V. Kilpatrick J.M. Plunkett W. Ayres M. Harman L. Du M. Bantia S. Davisson J. Wierda W.G. Faderl S. Kantarjian H. Thomas D. Blood. 2005; 106: 4253-4260Crossref PubMed Scopus (96) Google Scholar). Immucillin-H is in clinical trials for T-cell malignancies under the name of Fodosine™, and DADMe-Immucillin-H is in clinical trials under the name BCX-4208 (www.biocryst.com/index.htm). Human methylthioadenosine phosphorylase (MTAP) catalyzes the phosphorolysis of methylthioadenosine (MTA) to adenine and 5-methylthioribose 1-phosphate, an essential step in recycling MTA to S-adenosylmethionine (SAM) (30Grillo M.A. Colombatto S. Amino Acids (Vienna). 2007; (in press)Google Scholar, 31Tang B. Kadariya Y. Murphy M.E. Kruger W.D. Biochem. Pharmacol. 2006; 72: 806-815Crossref PubMed Scopus (19) Google Scholar). MTA is produced in the polyamine pathway, a target for cancer therapy (32Seiler N. Curr. Drug Targets. 2003; 4: 537-564Crossref PubMed Scopus (148) Google Scholar). SAM is a critical metabolite both as a precursor for polyamine synthesis and for methylation reactions that are essential to provide epigenetic control through methylation of histones and CpG islands in DNA. Kinetic isotope effects for arsenolysis of MTA by MTAP indicated full loss of the N-ribosidic bond and significant nucleophilic participation of the arsenate (33Singh V. Schramm V.L. J. Am. Chem. Soc. 2006; 128: 14691-14696Crossref PubMed Scopus (44) Google Scholar). In the phosphorolysis reaction of MTAP the products include fully protonated adenine and 5-methylthioribose 1-phosphate. Arsenate was used in the transition state determination to generate intrinsic kinetic isotope effects. With arsenate, the unstable arsenate intermediate rapidly hydrolyzes to make the reaction irreversible, and methylthioribose is the product (33Singh V. Schramm V.L. J. Am. Chem. Soc. 2006; 128: 14691-14696Crossref PubMed Scopus (44) Google Scholar). Unlike human and bovine PNPs, the transition state of human MTAP is late in the reaction coordinate. The leaving group bond to adenine is completely broken (3.0 Å), and the bond to the attacking arsenate nucleophile has begun to form (2.0 Å) (Fig. 5). Loss of the N-ribosidic bond is confirmed by the [9-15N]KIE of 1.037, near the upper limit for a 15N-labeled KIE. This limit can be reached only if the bond between the ribosyl group and adenine is fully broken and N-7 is unprotonated at the transition state (blue in Fig. 5). The nature of the late transition state is apparent in the MESP surfaces of reactant, transition state and product (Fig. 6). MT-DADMe-Immucillin-A was designed as a transition state analogue for MTAP and is an 86 pm slow-onset tight-binding inhibitor.FIGURE 6Reactant (MTA), transition state (TS), and products (5-methylthioribose 1-phosphate (MTR) and adenine) for the human MTAP reaction.View Large Image Figure ViewerDownload Hi-res image Download (PPT) In biological studies, MT-DADMe-Immucillin-A has been shown to block MTA metabolism in cultured cells and in mice. The combination of MT-DADMe-Immucillin-A together with MTA causes apoptosis in certain head and neck cancer cell lines, apparently from alterations in DNA methylation patterns of CpG islands (36Basu I. Cordovano G. Das I. Belbin T.J. Guha C. Schramm V.L. J. Biol. Chem. 2007; 282: 21477-21486Abstract Full Text Full Text PDF PubMed Scopus (65) Google Scholar). Thus, transition state analogues of MTAP may be interesting agents for some types of cancer. The transition states of PNPs and MTAP feature a cationic ribosyl C-1′ of ribose that migrates between the N-7-protonated adenine leaving group and the arsenate nucleophile in a mechanism called “nucleophilic displacement by electrophile migration.” In this novel mechanism the purine leaving group and the anionic nucleophiles are fixed in the catalytic site, and the flexible ribocation migrates between the fixed nucleophiles (34Schramm V.L. Shi W. Curr. Opin. Struct. Biol. 2001; 11: 657-665Crossref PubMed Scopus (60) Google Scholar, 35Fedorov A. Shi W. Kicska G. Fedorov E. Tyler P.C. Furneaux R.H. Hanson J.C. Gainsford G.J. Larese J.Z. Schramm V.L. Almo S.C. Biochemistry. 2001; 40: 853-860Crossref PubMed Scopus (202) Google Scholar). The transition states for bovine PNP, human PNP, and human MTAP all show this characteristic, proposed to be a general mechanism for many sugar transferases (34Schramm V.L. Shi W. Curr. Opin. Struct. Biol. 2001; 11: 657-665Crossref PubMed Scopus (60) Google Scholar). The TS for MTAP is reached when the ribosyl cation has left the adenine and has migrated more than half the distance to the arsenate nucleophile. MTAP does not use adenosine or inosine as substrates and the crystal structure shows the methylthio group surrounded by a hydrophobic pocket in the protein. Bacteria communicate with small molecules called autoinducers, first discovered in light generation by Vibrio harveyi, a marine luminescent bacterium (37Bassler B.L. Greenberg E.P. Stevens A.M. J. Bacteriol. 1997; 179: 4043-4045Crossref PubMed Google Scholar). More recently, it has been realized that some bacterial virulence factors, including biofilms, toxins, and adhesion molecules, are under the control of autoinducers (38Bjarnsholt T. Givskov M. Philos. Trans. R. Soc. Lond. B Biol. Sci. 2007; 362: 1213-1222Crossref PubMed Scopus (151) Google Scholar, 39Harraghy N. Kerdudou S. Herrmann M. Anal. Bioanal. Chem. 2007; 387: 437-444Crossref PubMed Scopus (38) Google Scholar). Autoinducers are generated from SAM, and the pathways for two types of autoinducers involve the action of methylthioadenosine nucleosidase (MTAN), expressed by the pfs locus in the Escherichia coli genome (29Singh V. Lee J.E. Nunez S. Howell P.L. Schramm V.L. Biochemistry. 2005; 44: 11647-11659Crossref PubMed Scopus (96) Google Scholar). In the production of the tetrahydrofuran autoinducer-2 (AI-2) molecules, MTAN is directly involved in converting S-adenosylhomocysteine (SAH) to ribosylhomocysteine, a direct precursor for AI-2 synthesis. The role of MTAN in homoserine lactones is to recycle MTA to SAM. Blocking the quorum-sensing pathways has been proposed as an approach to new antimicrobial agents (29Singh V. Lee J.E. Nunez S. Howell P.L. Schramm V.L. Biochemistry. 2005; 44: 11647-11659Crossref PubMed Scopus (96) Google Scholar, 37Bassler B.L. Greenberg E.P. Stevens A.M. J. Bacteriol. 1997; 179: 4043-4045Crossref PubMed Google Scholar). Targeting bacterial intracellular enzymes requires powerful inhibitors. The transition state of E. coli MTAN was solved as described above, and inhibitors were synthesized to match the electrostatic potential surface. Inhibitors to 47 fm were found for E. coli MTAN (Fig. 7). Kinetic isotope effects linked to computational chemistry provide sufficient information to develop geometric and electrostatic potential maps of enzymatic transition states. These are blueprints for the design of transition state analogues. Chemically stable molecules that resemble the transition states provide a wealth of inhibitors with dissociation constants in the picomolar to femtomolar range. Several of these have entered clinical trials, and others are in preclinical development. Determination of KIEs coupled to computational and synthetic chemistry can contribute significantly to the development of a broad spectrum of useful enzymatic inhibitors.