Quantifying energetic contributions to ground state destabilization

Quantifying energetic contributions to ground state destabilization

ABB Archives of Biochemistry and Biophysics 433 (2005) 27–33 www.elsevier.com/locate/yabbi Minireview Quantifying energetic contributions to ground ...

840KB Sizes 0 Downloads 46 Views

ABB Archives of Biochemistry and Biophysics 433 (2005) 27–33 www.elsevier.com/locate/yabbi

Minireview

Quantifying energetic contributions to ground state destabilization Vernon E. Anderson* Departments of Biochemistry and Chemistry, Case Western Reserve University, 10900 Euclid Avenue, Cleveland, OH 44106-4935, United States Received 2 August 2004, and in revised form 22 September 2004

Abstract Vibrational spectroscopy has identified that in many cases, substrate association with enzyme active sites results in significant bond polarization. This bond polarization can be attributed to a combination of desolvation, conformational restriction, and true polarization by the local electric field. Quantum chemical calculations permit the extent of polarization to be quantified both in terms of partial charge and energy. The changes in vibrational frequency that occur during the binding process necessarily result in equilibrium isotope effects. The equilibrium isotope effect on association is one feature that differentiates isotope effects on kcat and kcat/Km. An improved chemical understanding of the changes that occur on substrate binding will help elucidate the role of substrate activation in enzyme catalysis Ó 2004 Elsevier Inc. All rights reserved. Keywords: Vibrational spectroscopy; Binding isotope effect; Electronic polarization; Kinetic isotope effect; Nucleophile activation; Substrate destabilization

Isotope effects on binding: why might they be of interest? The difference between isotope effects on kcat and kcat/Km Isotope effects have proven to be a powerful technique to characterize enzyme reaction mechanisms. The contributions of kinetic complexity to the analysis of isotope effects on enzyme reaction have proven to be a constant experimental concern, but also have proven to be useful in characterizing the partitioning of intermediates. Northrop [1,2] provided the algebraic formulation and terminology for discussing isotope effects on enzyme reactions. The partitioning of the intermediates that immediately precede and follow a bond cleavage step is of central importance. The ratio of rate constants for the forward chemical conversion of the preceding intermediate and for its dissociation from the enzyme became *

Fax: +1 216 368 3419. E-mail addresses: [email protected], [email protected].

0003-9861/$ - see front matter Ó 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.abb.2004.09.026

defined as the ‘‘forward commitment to catalysis,’’ cf, while the related terms for the intermediate following the chemical conversion defined the ‘‘reverse commitment,’’ cr. This concept of commitment was sufficient to permit the experimentally measurable kinetic isotope effects on kcat/Km and kcat (indicated by a leading superscript) to be expressed in terms of the intrinsic isotope effect on the bond cleavage/formation step. D

D

D

k þ cf þ D K eq cr 1 þ cf þ cr D k þ cVf þ D K eq cr ¼ : 1 þ cVf þ cr

ðk cat =K m Þ ¼ k cat

ð1Þ

Fortunately, isotope effects on kcat/Km can be measured by competitive methods, allowing them to be determined with sufficient precision to evaluate heavy atom kinetic isotope effects whose maximum values are <1.1. Further, the analysis indicated that the kinetic isotope effect compared the quantum chemical partition function for the substrate free in solution to that of the highest free energy transition state along the reaction coordinate at 0 product concentration.

28

V.E. Anderson / Archives of Biochemistry and Biophysics 433 (2005) 27–33

Isotope effects on kcat are expressed with a different algebraic commitment factor, cVf, (or alternatively the ‘‘ratio of catalysis,’’ Rf (2)) along with the same reverse commitment factor defined for analysis of kcat/Km (Eq. (1)). The utility in measuring isotope effects on kcat effectively would determine the difference between cf and cVf. Analysis of cVf revealed that there were two contributions to this term, one contribution came from the potential for rate determining steps arising after the dissociation of one product (or more formally the first irreversible step). This step could either be the release of a second product or a unimolecular step that occurs in isomechanisms. If either of these processes limits kcat, it will reduce the observed isotope effect on kcat since there is presumably at most a minimal kinetic isotope effect on product release or enzyme isomerizations resulting in Dkcat < D(kcat/Km). A different scenario could result in cVf being less than cf when the second-order substrate association limited kcat/Km (or more correctly, slow substrate dissociation leads to a significant cf but does not contribute to cVf). Under these conditions, larger observed isotope effects on kcat, i.e., Dkcat > D(kcat/ Km), would be of interest as it would reflect differences in the quantum partition functions for the substrate in the Michaelis complex and in the transition state. This recognition led to our conception that the most interesting comparison of isotope effects on kcat and kcat/Km will come when both cf and cVF were determined to be negligible. Under these two conditions, any difference between Dkcat and D(kcat/Km) would necessarily be attributable to changes that occur to the substrate on formation of the Michaelis complex; an inherently interesting proposition. Although it had been routinely assumed that there should be no effect of isotopic substitution on this process. Many fruitless attempts were made in my laboratory to devise a method of measuring isotope effects on kcat with sufficient precision to distinguish differences between the very precise effects on kcat/Km determined by competitive methods with those to be determined on kcat. Eventually such a method was devised by Phil Huskey based on constant flow stirred tank reactor concepts [3]. However, further reflection suggested that since our primary interest in comparing kinetic isotope effects on kcat and kcat/Km was to characterize differences in the different ground states, i.e., free in solution and bound at the active site. This measurement might be more appropriately done by a comparison between these two states, i.e., to directly measure the isotope effect on binding. Evidence for substrate destabilization The effect of a substrate binding to the active site of an enzyme has been of fundamental interest for decades. Before it had been established that most metabolic en-

zymes were proteins, kinetics had identified complex formation as a key principle of enzyme catalysis. The selectivity of enzymes was envisioned as arising from interactions in analogy to ‘‘lock and key’’ complementarity [4]. The catalytic activity is not encompassed within this analogy, and perhaps because of less familiarity with transition state concepts, the focus was on substrate activation. Of specific note, Bayliss [5] expressed the following idea in 1925, ‘‘The physical view holds that the increased rate of reaction is due to an increase of active mass. . .the act of condensation may be accompanied by the intervention of molecular forces which result in a rise in the chemical potential of the reacting substances.’’ The specific concept that the rise in chemical potential could be attributed to electronic interactions was also introduced early by Quastel [6]. ‘‘Associated with certain molecules or groups of [the enzyme] there will exist electric fields whose nature will be dependent on the nature of the groups which give rise to them. Some of these fields will be very powerful due to. . .the juxtaposition or association of certain molecules or groups. . .these locally intense electric fields being a function of the ÔgeographyÕ of [the enzyme]. . .It is from a consideration of the effects induced in [the substrate] by the application of an external field that I would suggest that an interpretation of the mechanism of activation is forthcoming.’’ The electronic strain-induced mechanism of substrate activation (or destabilization) was eclipsed by the more graphic ‘‘rack mechanism’’ of substrate activation [7]. However, with the availability of spectroscopic quantities of enzyme the investigation of the interaction of substrates with proteins advanced. Carey et al. in studies of thiol proteases [8,9] and Belasco and Knowles [10,11] in studies of triosephosphate isomerase and aldolase identified changes in vibrational spectra of enzyme bound substrates and rcognized these changes as being consistent with an ‘‘enzyme-induced distortion of the substrate’’ that resulted in a mechanistically significant strain. A necessary corollary derived from the change in vibrational spectrum of the substrate is that there must be an isotope effect on the equilibrium binding constant, the only question being if it would be large enough to be measurable. Thus, the rationale for measuring isotope effects on the equilibrium binding of substrates to enzymes is clear, it provides a mechanism for quantifying changes in the vibrational modes of the substrate that are a direct function of the electronic state of the substrate. In lactate dehydrogenase, the on-enzyme equilibrium constant is near unity while the solution equilibrium constant favors the formation of NAD+ and lactate, suggesting that NAD+ has become a stronger oxidant on association with the enzyme. Indeed, we successfully determined the equilibrium binding isotope

V.E. Anderson / Archives of Biochemistry and Biophysics 433 (2005) 27–33

effect of [4-3H]NAD+ and [4-2H]NAD+ to lactate dehydrogenase [12]. This observation established that significant binding isotope effects could exist, contrary to some assumptions [13]. However, it also pointed out the difficulties in interpretation. Binding equilibrium isotope effects can be calculated from changes in the normal vibrational modes [14], but attributing an experimentally observed isotope effect to changes in a specific vibrational mode is not possible. This assignment can only be made directly by vibrational spectroscopy. The recent observation in the crystal structure of phosphoglucoisomerase of a ‘‘high-energy’’ pentacovalent phosphorane intermediate [15] underscores that the same fundamental concepts form the basis of transition state stabilization and ground state destabilization. In the chemical reaction catalyzed by this enzyme, a phosphoryl group is transferred from the C1(O) to an active site nucleophile. The electron density is interpreted to show that the electrons of the PAO(C1) bond have been polarized resulting in a lengthening of this bond and rehybridization of the phosphoryl group. This structure can be viewed as a destabilized substrate as readily as a stabilized transition state. The fact that a more prosaic enzyme–substrate Michaelis complex was not observed suggests that a classical glucose-1-phosphate complex is of higher energy than the strained structure observed in the crystal. Quantitation of electronic strain Many subsequent studies, including prominent contributions from the Callender [16–19] and Carey [20– 23] groups, identified changes in substratesÕ vibrational spectra when bound at enzyme active sites. Correlating the experimentally observed changes in spectra with an energetic contribution to catalysis is more problematic. An empirical approach is to model the spectral changes with a series of solution complexes whose interaction enthalpy can be experimentally determined. This approach yielded a correlation of 0.5 kJ/mol for each cm1 decrease in the carbonyl stretching frequency when an H-bond is formed [24]. A necessary consideration is that the observed electronic strain must contribute to the catalytic action (a macroscopic analogy is that twisting an ankle would contribute little to breaking an arm). In the case of proteases, polarization of the carbonyl directly enhances the electrophilicity hence the reactivity of the carbonyl carbon. The Callender group has made circumstantial arguments based on the Badger–Bauer rule that correlates changes in vibrational frequency with changes in bond order and energy [18]. The assumptions that bond order, energy, and vibrational frequencies are linearly correlated limit this approach to providing only an estimate of the substrate destabilization.

29

Recent developments in both computational hardware and software have advanced quantum chemical calculations to the point that isotope effects, both kinetic and equilibrium, on enzyme reactions can be calculated [25–28]. Consistency between the calculated and observed isotope effects provides an experimental validation to the computational model. A comparison of the quantum chemical model of the substrate bound at the active site and free in solution permits the electron polarization to be modeled [29]. The electronic strain can then be quantified by calculating the energetic change required to polarize the electrons from the distribution that exists when the substrate is present in aqueous solution and when it is bound to the enzyme [30].

Application to enoyl-CoA hydratase Enoyl-CoA hydratase proved to be a fortuitous enzyme to characterize spectroscopic changes that occur with substrate association. Extending the conjugation of the a,b-unsaturated thiolester of the substrate with a phenyl group results in a substrate that is thermodynamically stabilized so that equilibrium favors the substrate so that its spectroscopic properties in an active enzyme–substrate complex can be determined as shown in Eq. (2). Furthermore, there are substantial changes in the UV spectra of the bound substrates. The largest increase in kmax of 90 nm occurs with para-dimethylamino-cinnamoyl-CoA (DAC-CoA). This increase corresponds to a decrease in the energy of excitation of 13 kcal/mol [31]. If this amount could be demonstrated to come from polarization of the ground state of the substrate, it could account for most of the catalytic power of the enzyme. The polarization would be catalytically relevant, as enhancing the electrophilicity of the b-carbon would facilitate the hydration reaction.

ð2Þ The polarization of the substrate is confirmed by both C NMR of the bound substrates and by red shifts in the stretching frequency of the carbonyl and carbon– carbon double bond (C@C). These spectroscopic changes are tabulated in Table 1. 13

30

V.E. Anderson / Archives of Biochemistry and Biophysics 433 (2005) 27–33

Table 1 Spectroscopic changes in DAC-CoA that occur on association with ECH Method and observation

Solution value

ECH-DAC-CoA complex value

UV–Vis DAC-thiolester 13 C NMR carbonyl carbon 13 C NMR a-carbon 13 C NMR b-carbonyl carbon Raman carbonyl stretch Raman Ca@Cb stretch

404 nm 195.7 ppm 121.7 ppm 144.9 ppm 1660 cm1 1573 cm1

494 nm 198.5 118.7 147.7 ppm 1614 cm1 1499 cm1

In general, such spectroscopic changes can be attributed to three possible changes in the substrate: (i) the substrate is desolvated, (ii) the ensemble of conformations present in solution become restricted to a small subset of conformations that are bound at the enzyme active site, and (iii) the substrate may be electronically strained. An in-depth analysis of the binding isotope effects for glucose and hexokinase refers to the first two processes as prebinding [32]. The process of DACCoA binding to the ECH active site is conceived to occur roughly, in this order as shown in Fig. 1. Spectroscopic approaches permit a focused examination of the changes that occur. In DAC-CoA the interesting portion of the molecule is the a,b-unsaturated thiolester moiety that becomes hydrated in the reaction with the normal substrate. The dimethylamino-cinnamoyl moiety absorbs in the UV and the shift in the kmax is strong evidence for an interaction with the active site. The shift cannot be attributed to desolvation as the concomitant decrease in dielectric constant gives rise to a modest blue shift in model compounds. The conformation of the dimethylamino-cinnamoyl group is constrained by conjugation and symmetry. Conjugation maintains the near planarity of the a,b-unsaturated carbonyl with the phenyl ring, leaving only the s-cis and s-trans conformations (see structures IV and V in Fig. 1, respectively). Characterization of 1H–1H NOEs [33] and subsequently determination of the crystal structure established the unique conformation of DAC-CoA present at the active site [34]. While the rack mechanism of inducing substrate strain has fallen into some disfavor [35], careful analysis of many substrates bound at active sites finds their conformations distorted from the lowest energy conformer present in solution. This has been extensively characterized in lysozyme [36], and in ECH, the phenyl ring is twisted from planarity with the a,b-carbon–carbon double bond by between 12° and 29°. Once bound in a constrained conformation, the interaction of the DAC-CoA with the electrostatic field generated by the functional groups at the active site results in a redistribution of the electrons. By definition, this is an electronic strain as there is a motion in response to an external force (or stress). The DAC-CoA case is of

unusual interest because the red-shift observed in the UV spectrum is so large. The calculation of this electronic polarization is most easily interpretable if done at a QM/MM level with the boundary drawn between the substrate and the enzyme atoms [29]. The inclusion of any of the enzyme atoms within the QM system results in some charge transfer, and the electrons ‘‘belonging to the substrate’’ cannot be identified.1 Determining the charge on each substrate atom is problematic and we resorted to displaying a difference map of the calculated electron densities as shown in Fig. 2. The qualitative analysis of the QM/MM calculation of polarization was gratifying, it highlighted the polarization of the p-electrons of the O@CAC@C moiety. Further, it demonstrated how focused the polarization effect could be since the conjugated phenyl system was minimally polarized which is consistent with the absence of polarization of the c,d-C@C observed in the Raman spectrum of hexadienoyl-CoA [37]. This emphasizes that electric fields in enzyme active sites can be highly localized. However, in spite of the large spectral changes, the change in electron density was small. The variation was ˚ 3, except in the immediate vicinity of the car<0.01 e/A bonyl O and the changes in the Mullikan point charges were <0.07 e. The strain can be quantified as the difference in electronic energy of the substrate in the presence and absence of the electric field generated by the enzyme active site without allowing the conformation to relax [29,30]. This calculation for the electronic strain in the bound DAC-CoA was 3.2 kcal/mol, a significant contribution to catalysis but only a modest fraction of the catalytic power of the enzyme. We have considered two additional sources of catalytic power: stabilization of the transition state by the H-bonds to the carbonyl and activation of the nucleophilic water. H-bonding to the thiolester carbonyl is a hallmark of the ECH family of enzymes [38,39]. Two backbone amides interact with the thiolester carbonyl in every crystal structure examined. The orientation of these amides may significantly contribute to the preferential interaction with the transition state. The preferred orientation of N-methylformamide with S-methylthiolacetate is in the plane of the thiolester as shown in Fig. 3, consistent with the H-bond interaction being with an sp2 lone pair. When the H-bonded complex is optimized with an enolate, the preferred orientation is orthogonal to the thiolester (the improper dihedral SACAOAN 60°).2

1 This is an example of how any attempt to dissect a specific interaction energy necessarily involves limiting approximations. 2 The exact dihedrals depend on the level of the calculation. The scheme is derived from B3LYP/HF6-31+G.

V.E. Anderson / Archives of Biochemistry and Biophysics 433 (2005) 27–33

31

Fig. 1. Three conceptual processes that contribute to changes in substrate spectroscopic properties and fractionation factors. Solvated ensemble of conformations (A) are desolvated (B) and in adapting to the enzyme active site are constrained to a reduced number of conformations (C) which in the final conformation are polarized by interactions at the active site (D). In concept, these processes can be subdivided while physically, desolvation and conformational selection occur early in the binding process and continued conformation restriction is associated with enhanced interaction with the active site.

Nucleophile activation Nucleophile activation has been less well spectroscopically characterized than electrophilic activation. General base catalysis is often invoked as a mechanism of promoting enzyme catalysis [40,41]. In ECH, two Glu residues are bridged by the nucleophilic water molecule. This water molecule can be said to be electronically strained: it has been completely desolvated as there are no apparent Hbond donors to the two electron pairs and the H-bonds to the carboxylates are potentially stronger than the average H-bond to solvent water. These stronger H-bonds will be reflected in a polarization of the OAH sigma bonding electrons, and hence an electronic strain. Because of the lack of spectroscopic probes for interactions with water, we sought for a spectroscopic probe of H-bond donation by substrate nucleophiles. Monitoring the OAH stretching frequency, mOAH, would be attractive, but the solvent background prevents this sim-

ple approach. Quantum calculations identified a correlation of the CAH or CAD stretching frequencies, mCAH, at C2 of [1-2H]ethanol with H-bond formation [42], which has been confirmed experimentally [43]. A close analysis of this phenomenon identified an increase in H-bond strength correlated with a decrease the HAC(OH) bond strength due to an increase in negative hyperconjugation. Importantly, this decrease in the bond strength can be detected spectroscopically by selective D incorporation and monitoring mCAD or by measuring equilibrium isotope effects on complex formation. From the change in vibrational modes that occur on changes in H-bonding, it was predicted that equilibrium isotope effects on association should be observed when the H-bond donor forms H-bonds with an enzyme active site. These effects can potentially be quite large, e.g., the ionization of trifluoroethanol results in an equilibrium isotope effect 1.065 [42]. Similar conclusions have been reached by Lewis and Schramm [32,44,45]

32

V.E. Anderson / Archives of Biochemistry and Biophysics 433 (2005) 27–33

˚ 3 and color coded as shown on the scale at Fig. 2. Polarization of DAC-CoA represented in the plane of the thiolester with contours of 0.001 e/A right.

Fig. 3. B3LYP/HF6-31G optimized H-bonds between a neutral thiolester (A) or the enethiolate (B) and N-methylformamide. The NAH bond lies in the plane of the thiolester but is 60° out of plane in the enethiolate.

by examining isotope effects on ionization and binding of glucose to hexokinase. Notable in these studies, the 3 H binding isotope effect at C6 of 1.065 supports the formation of an active site H-bond that activates the nucleophilic geminal hydroxyl [46].

Summary Changes that occur on the association of substrates or inhibitors when they associate with enzyme active

sites will continue to be of interest as the details of enzyme reaction mechanism are probed both experimentally and computationally. Vibrational spectroscopy offers the capability of identifying electronic polarization that can be quantified either by empirical or computational correlations. A virtue of the spectroscopic approach is that the effect on specific vibrational modes, and hence the electronic distribution in specific bonds can be identified. Equilibrium isotope effects on binding provide a probe where detection of the appropriate individual vibrational modes is not feasible. The binding iso-

V.E. Anderson / Archives of Biochemistry and Biophysics 433 (2005) 27–33

tope effects also provide necessary information to determine the extent that desolvation, conformational selection, and substrate activation in the Michaelis complex contribute to measured kinetic isotope effects.

Acknowledgments This work has been supported by the NIH (GM 36562) and NSF (MCB 0110610). Many co-workers have made significant contributions to the work, but collaborations with Dr. Paul Carey, Dr. Peter Tonge, and Dr. Piotr Paneth have been fruitful and contributed to both the experimental details and intellectual direction of the science described. Thanks to Drs. Lewis and Schramm (Einstein College of Medicine) for providing a preprint of their work on binding isotope effects.

References [1] D. Northrop, in: W.W. Cleland, M.H. OÕLeary, D.B. Northrop (Eds.), Isotope Effects on Enzyme-Catalyzed Reactions, University Park Press, Baltimore, MD, 1977, pp. 122–152. [2] D.B. Northrop, in: P.F. Cook (Ed.), Enzyme Mechanism from Isotope Effects, CRC Press, Boca Raton, FL, 1991, pp. 181–202. [3] H. Xue, X. Wu, W.P. Huskey, J. Am. Chem. Soc. 118 (1996) 5804–5805. [4] E. Fischer, Ber. Dt. Chem. Ges. 27 (1894) 2985–2993. [5] Sir W.M. Bayliss, The Nature of Enzyme Action, fifth ed., Longmans Green, London, 1925. [6] J.H. Quastel, Biochem. J. 20 (1926) 166. [7] H. Eyring, R. Lumry, J.D. Spikes, in: W.D. McElroy, B. Glass (Eds.), The Mechanism of Enzyme Action, Johns Hopkins Press, Baltimore, MD, 1954, pp. 123–140. [8] P.R. Carey, R.G. Carriere, D.J. Phelps, H. Schneider, Biochemistry 17 (1978) 1081–1087. [9] P.R. Carey, R.G. Carriere, K.R. Lynn, H. Schneider, Biochemistry 15 (1976) 2387–2393. [10] J.G. Belasco, J.R. Knowles, Biochemistry 19 (1980) 472–477. [11] J.G. Belasco, J.R. Knowles, Biochemistry 22 (1983) 122–129. [12] R.D. LaReau, W. Wan, V.E. Anderson, Biochemistry 28 (1989) 3619–3624. [13] W.W. Cleland, Adv. Enzymol. Relat. Areas Mol. Biol. 45 (1977) 273–387. [14] L. Melander, Isotope Effects on Reaction Rates, Ronald Press, New York, 1960. [15] S.D. Lahiri, G. Zhang, D. Dunaway-Mariano, K.N. Allen, Science 299 (2003) 2067–2071. [16] H. Deng, J. Zheng, J. Burgner, R. Callender, Proc. Natl. Acad. Sci. USA 86 (1989) 4484–4488.

33

[17] H. Deng, W.J. Ray Jr., J.W. Burgner 2nd, R. Callender, Biochemistry 32 (1993) 12984–12992. [18] H. Deng, J. Zheng, A. Clarke, J.J. Holbrook, R. Callender, J.W. Burgner 2nd, Biochemistry 33 (1994) 2297–2305. [19] H. Deng, R. Callender, Methods Enzymol. 308 (1999) 176–201. [20] P.J. Tonge, P.R. Carey, Biochemistry 29 (1990) 10723–10727. [21] K. Taylor, R. Liu, P. Liang, J. Price, D. Dunaway-Mariano, P. Tonge, J. Clarkson, P. Carey, Biochemistry 34 (1995) 13881– 13888. [22] D. Dinakarpandian, B.C. Shenoy, D. Hilvert, D.E. McRee, M. McTigue, P.R. Carey, Biochemistry 38 (1999) 6659–6667. [23] J. Dong, A.C. Drohat, J.T. Stivers, K.W. Pankiewicz, P.R. Carey, Biochemistry 39 (2000) 13241–13250. [24] P.J. Tonge, P.R. Carey, Biochemistry 31 (1992) 9122–9125. [25] S. Marti, V. Moliner, I. Tunon, I.H. Williams, Org. Biomol. Chem. 1 (2003) 483–487. [26] L.S. Devi-Kesavan, J. Gao, J. Am. Chem. Soc. 125 (2003) 1532– 1540. [27] C. Alhambra, J. Corchado, M.L. Sanchez, M. Garcia-Viloca, J. Gao, D.G. Truhlar, J. Phys. Chem. B 105 (2001) 11326–11340. [28] N. Diaz, D. Suarez, T.L. Sordo, K.M. Merz Jr., J. Phys. Chem. B 105 (2001) 11302–11313. [29] R.L. DÕOrdine, J. Pawlak, B.J. Bahnson, V.E. Anderson, Biochemistry 41 (2002) 2630–2640. [30] V.E. Anderson, Encyclopedia of Life Sciences, Macmillan, London, 1999. [31] R.L. DÕOrdine, P.J. Tonge, P.R. Carey, V.E. Anderson, Biochemistry 33 (1994) 12635–12643. [32] B.E. Lewis, V.L. Schramm, in: A. Kohen, H.-H. Limbach (Eds.), Isotope Effects in Chemistry and Biology, 2004. [33] W.J. Wu, V.E. Anderson, D.P. Raleigh, P.J. Tonge, Biochemistry 36 (1997) 2211–2220. [34] B.J. Bahnson, V.E. Anderson, G.A. Petsko, Biochemistry 41 (2002) 2621–2629. [35] A. Fersht, Structure and Mechanism in Protein Science, first ed., W.H. Freeman, New York, 1998. [36] N.C. Strynadka, M.N. James, J. Mol. Biol. 220 (1991) 401–424. [37] P.J. Tonge, V.E. Anderson, R. Fausto, M. Kim, M. PusztaiCarey, P.R. Carey, J. Biol. Spectrosc. 1 (1995) 387–394. [38] H.M. Holden, M.M. Benning, T. Haller, J.A. Gerlt, Acc. Chem. Res. 34 (2001) 145–157. [39] J.A. Gerlt, P.C. Babbitt, Curr. Opin. Chem. Biol. 2 (1998) 607– 612. [40] K.B. Schowen, H.H. Limbach, G.S. Denisov, R.L. Schowen, Biochim. Biophys. Acta 1458 (2000) 43–62. [41] P.C. Bevilacqua, Biochemistry 42 (2003) 2259–2265. [42] E. Gawlita, M. Lantz, P. Paneth, A. Bell, P. Tonge, V.E. Anderson, J. Am. Chem. Soc. 122 (2000) 11660–11669. [43] N.C. Maiti, P.R. Carey, V.E. Anderson, J. Phys. Chem. A 107 (2003) 9910–9917. [44] B.E. Lewis, V.L. Schramm, J. Am. Chem. Soc. 125 (2003) 7872– 7877. [45] B.E. Lewis, V.L. Schramm, J. Am. Chem. Soc. 123 (2001) 1327– 1336. [46] B.E. Lewis, V.L. Schramm, J. Am. Chem. Soc. 125 (2003) 4785– 4798.