Construction and analysis of a semi-quantitative energy profile for the reaction catalyzed by the radical enzyme galactose oxidase

Biochimica et Biophysica Acta 1384 Ž1998. 43–54

Construction and analysis of a semi-quantitative energy profile for the reaction catalyzed by the radical enzyme galactose oxidase Rebekka M. Wachter, Bruce P. Branchaud

)

Department of Chemistry and Institute of Molecular Biology, UniÕersity of Oregon, Eugene, OR 97403, USA Received 5 August 1997; revised 3 December 1997; accepted 5 December 1997

Abstract Galactose oxidase ŽGOase. is a mononuclear type 2 copper enzyme which oxidizes primary alcohols to aldehydes using molecular oxygen ŽRCH 2 OH q O 2 s RCHOq H 2 O 2 .. An unusual crosslink between tyrosine 272 and cysteine 228 provides a modified tyrosine radical site which acts as a ligand for the active site copper and is believed to act as a one-electron redox center. The single active site copper is believed to act as a second one-electron redox center. The use of the tyrosine one-electron redox center and the copper one-electron redox center allows removal of two electrons from alcohol substrate for subsequent transfer to molecular oxygen. Previously, we and others have proposed a detailed step-by-step radical mechanism for the reaction catalyzed by galactose oxidase. The catalytic cycle can be divided into two half reactions. The first half reaction entails transfer of two electrons and two protons from the alcohol substrate to the enzyme to form aldehyde product and two-electron-reduced enzyme Žone electron at the tyrosine center and one at the copper center.. The second half reaction entails transfer of two electrons and two protons from the two-electron-reduced enzyme to O 2 to form H 2 O 2 product and regenerate fully oxidized catalytically active enzyme ready for another catalytic cycle. In this paper, we describe the construction of a semi-quantitative energy profile for this radical mechanism. Several significant points emerge from this analysis. One point is the prediction that galactose oxidase should have an unusually low redox potential for copper, to our knowledge lower than any other redox active copper protein. Another point is that the distorted or entatic copper site causes the unusually low redox potential. A final point is that crosslinking of tyrosine 272 and cysteine 228 alters the redox properties of the tyrosine center to enhance catalysis compared to what would be expected for a normal tyrosine. q 1998 Elsevier Science B.V. Keywords: Energy profile; Galactose oxidase; Radical mechanism; Enzymology

1. Introduction Many enzymes are now known to proceed through radical mechanisms w1–3x. Most of these reactions involve redox cofactors. New types of mechanisms have been proposed recently involving protein radi)

Corresponding author. Fax: q1-541-346-4645; E-mail: [email protected]

cals, mostly on amino acid side chains w4–6x, but even on the protein backbone as in pyruvate formate lyase w7x. H-atom abstraction from a substrate by a protein-derived radical is likely to occur in the mechanisms of several of these protein radical enzymes including the various types of ribonucleotide reductases Ž binuclear iron q tyrosyl radical, adenosylcobalamin ŽB 12 . q protein radical, binuclear manganeseq protein radical. w8x, several B 12 enzymes w9,10x, and

0167-4838r98r$19.00 q 1998 Elsevier Science B.V. All rights reserved. PII S 0 1 6 7 - 4 8 3 8 Ž 9 7 . 0 0 2 0 9 - 4

44

R.M. Wachter, B.P. Branchaudr Biochimica et Biophysica Acta 1384 (1998) 43–54

prostaglandin H-synthase Ž mononuclear iron q tyrosyl radical. in prostaglandin biosynthesis w11x. A hydrogen atom abstraction model has also recently been proposed for the function of YZ in photosynthetic oxygen evolution by Photosystem II w12x. Galactose oxidase Ž GOase. from the filamentous wheat-rot fungus Fusarium spp. w13x catalyzes the oxidation of primary alcohols with O 2 , producing aldehydes and H 2 O 2 Ž RCH 2 OH q O 2 s RCHOq H 2 O 2 . w14x. It is a single polypeptide with a molecular mass of 68,500 w15x. GOase contains two oneelectron redox centers, a mononuclear copper center and a tyrosine center covalently crosslinked Ž at the ortho position to the –OH. to a cysteine Ži.e., Tyr272 and Cys228 crosslink. w16,17x. The unusual Tyr272 is one of the equatorial ligands of the square–pyramidal copper center. GOase can exist in three distinct, stable oxidation states w18x. These can be assigned as highest oxidation state s CuŽII. and tyrosine radical, intermediate oxidation state s CuŽ II. and tyrosine Ž in equilibrium with CuŽ I. and tyrosine radical. , and lowest oxidation state s CuŽI. and tyrosine. Spectroscopic evidence strongly indicates that a tyrosine radical is the radical center, that the tyrosine radical is directly coordinated to the copper center, and that the highest oxidation state of GOase is the catalytically active form of the enzyme w16,18x. Before a structural model of GOase from X-ray crystallographic data revealed the unusual crosslink between tyrosine 272 and cysteine 228, Whittaker proposed a new type of radical mechanism utilizing the tyrosine-like protein radical that he had detected based on extensive spectroscopic evidence w18x. Taking advantage of the structural data from X-ray crystallography w17x and kinetic evidence with radicalprobing substrates w19x we proposed a more detailed mechanistic scheme w19x. Whittaker subsequently refined the mechanism further—we proposed that the alcohol is deprotonated by a nearby histidine acting as a base, but spectroscopic studies of anion binding to galactose oxidase lead Whittaker to propose that tyrosine 495 could act as the base w20x. Although there are some minor differences in the proposed mechanisms, the central feature of them all is that enzymic catalysis is proposed to proceed by a stepwise radical mechanism with a substrate derived ketyl radical as a key intermediate ŽScheme 1.. We have also considered the possibility of a concerted mecha-

nism for the two electron transfer from substrate to enzyme Ž Scheme 1. , where H-atom abstraction and electron transfer to copper occur simultaneously. Note that both the stepwise and concerted mechanisms are consistent with mechanistic evidence from kinetic studies including Ž1. an ordered binding mechanism with substrate bindings and product releases occurring in the order shown and Ž2. cleavage of the a C–H bond as at least a partially rate-determining step and probably the major rate-determining step Ž known from the large primary deuterium isotope effect when the a position is substituted with deuterium. w19,21x. The mechanism shown in Scheme 1 provides a plausible pathway to transfer electrons from substrate to enzyme then to O 2 . We became interested in analyzing the detailed energetics of this mechanism, with a focus on two main objectives. The first objective was to analyze whether the catalytic groups present at the GOase active site are capable of performing the fundamental reaction steps shown, i.e., could the proposed intermediates be generated and stabilized by the GOase active site? The second objective was to analyze whether the unusual structural features of the GOase active site, specifically the modified tyrosine center and the unusual ligand geometry at copper Žsee below., served to further enhance catalysis beyond what might be expected for a normal tyrosine and a normal type 2 copper center. There are very few enzymes for which it is possible to make direct experimental measurements to construct an energy profile for the reaction. Ideally all rate and equilibrium constants for each elementary step are determined experimentally. In practice, experimental techniques are limited for fast reaction steps. From steady-state kinetics, only lower bounds of individual rate constants are obtained w22x. Perhaps the best example of this is the detailed analysis of the energetics of the triose phosphate isomerase-catalyzed reaction w23x. In that work, measurements of initial velocities and isotopic content of reactants and products were used to obtain rate constants and fractionation factors for the individual steps, and the Gibbs free energies of the intermediates and transition states in the reaction were calculated from these data. Triose phosphate isomerase represents a relatively ideal system, in that there is only one intermediate apart from substrate and product complexes, all steps involve acid–base catalysis allowing for iso-

R.M. Wachter, B.P. Branchaudr Biochimica et Biophysica Acta 1384 (1998) 43–54

45

Scheme 1. Two possible radical mechanisms for galactose oxidase—stepwise catalysis via a ketyl radical intermediate vs. concerted E2-like oxidation of the alcohol, either mechanism using tyrosine 272 and copper as one-electron redox centers.

topic exchange measurements, and the reaction can be run forward and backward. Even so, activation energies were determined quantitatively only for the kinetically significant steps, the chemical conversion of bound substrate and the rate of product binding. Only a lower limit could be determined for the rate of product release which is faster than the rate of product binding. To measure individual rate constants for elementary steps that are not rate-limiting, rapid mixing techniques are usually employed. Using stopped-flow instrumentation, the practical upper limit for the observation of first-order rate constants is about 700 sy1 w22x. This poses a problem for GOase since the turnover number is approximately 800 sy1. The

cleavage of the alcohol C–H bond is at least a partially rate-determining step and probably the major rate-determining step. Much of the catalytic chemistry of interest occurs after the cleavage of the alcohol C–H bond, should be significantly faster than that step, and thus would not be directly observable. Relaxation methods, on the other hand, are useful only if appreciable concentrations of intermediates, reactants, andror products are present at equilibrium w22x. The GOase-catalyzed reaction is very exothermic and goes to completion rapidly with no equilibria to perturb. It is physically impossible to make enough direct experimental measurements to construct an energy profile for the galactose oxidase reaction. Neverthe-

46

R.M. Wachter, B.P. Branchaudr Biochimica et Biophysica Acta 1384 (1998) 43–54

less, it is possible to use the available data on GOase and on relevant model systems to construct a semiquantitative energy profile. Analysis of such a profile can provide otherwise unattainable insight into radical reaction mechanisms and how the unusual GOase active site is equipped to perform the catalysis of radical reactions.

2. Methods Following Scheme 1, the catalytic cycle can be divided into several elementary steps. The reduction half-reaction involves reduction of the enzyme with concomitant oxidation of alcohol to aldehyde, and comprises Steps 1–4. The oxidation half-reaction involves the re-oxidation the enzyme with concomitant reduction of dioxygen to hydrogen peroxide, and comprises Steps 5–9. In the calculations below, the numbers of the individual steps are shown in the headings, and also in Table 1 and Fig. 1. All potentials used are formal reduction potentials Ž mid-point potentials under the particular conditions of measurement, which are not standard conditions. , and are referenced relative to the NHE unless stated otherwise. 2.1. Substrate alcohol deprotonation (Step 2) 2.1.1. Estimate of kinetics of Step 2 Substrate binding is thought to be followed by proton transfer from the substrate to Tyr495, axially coordinated to copper w20x. Molecular modeling studies of enzyme-substrate interactions in GOase indicated that a hydrogen bond between O6 of galactose and Oh of Tyr495 should be formed upon substrate binding w24x. This is consistent with the observation that proton transfer frequently occurs along an existing hydrogen bond w25x. Proton transfer dynamics generally proceed at timescales considerably shorter than 10y9 s, and in the ideal case of a symmetric hydrogen bond as fast as 10y12 s w25x. The proton transfer step in GOase catalysis should be very rapid, in part due to the close matching of p K a’s Žsee below., and in part due to the optimized hydrogen bonding geometry found from molecular modeling w24x.

2.1.2. Estimate of thermodynamics of Step 2 Proton transfer from alcohol substrate to Tyr495 ŽStep 2. can be estimated to be nearly isothermic. The p K a of the groups participating in acid–base chemistry can be used to calculate the free energy change of proton transfer. Acidities of ligands coordinated to metal ions are dependent upon the stereochemistry of the complex and the charge and electronic configuration of the metal ion. Small, highly charged ions such as CuŽII. are best at increasing acidity w26x. The p K a of copper-coordinated water has been shown to be 7.3 for copper-hydrates where the overall charge of the metal complex changes from q2 to q1 upon deprotonation of copper-bound water w27x. Coordination of a water to the metal center in GOase would lower the p K a roughly 7 orders of magnitude, decreasing from a p K a of 14 in the non-complexed state to a p K a of 7 in the bound state. The p K a of the primary alcohol of galactose can be estimated to be approximately that of ethanol, which is 18 w28x. If the p K a drop upon coordination of alcohol to copper is similar to that of water, the p K a of bound galactose O6 would then be about 11. The p K a of Tyr495, the proton acceptor, is estimated to be at least 8.8 based on active site titrations in the presence of a substrate analog w20x, or around 10.2 based on the p K a of free tyrosine w29x. This p K a range represents unliganded Tyr495, since coordination to copper would result in a p K a drop. The measured p K a above 8.8 is consistent with evidence that the axial Tyr495 dissociates from copper upon substrate binding w20x. Using Eq. Ž 1. , deprotonation is calculated to be endothermic by roughly 1.1 kcalrmol. The proton transfer is nearly isothermic due to close p K a matching of the groups involved in acid–base chemistry. DG s yRT ln K aŽalcohol.rK aŽTyr495.

Ž1.

2.2. C–H Bond cleaÕage by H-atom transfer (Step 3) 2.2.1. Estimate of kinetics of Step 3 Breaking of the C–H bond in a primary alcohol by galactose oxidase is at least a partially rate-determining step and probably the major rate-determining step Žknown from the large primary deuterium isotope effect when the a position is substituted with deu-

R.M. Wachter, B.P. Branchaudr Biochimica et Biophysica Acta 1384 (1998) 43–54

terium. w19,21x. Evidence from radical-probing substrates indicates that the C–H bond cleavage is a hydrogen atom abstraction leading to the formation of a ketyl radical anion intermediate w19x. Thus, kinetic data can be used to estimate the activation barrier for the hydrogen atom transfer step. In the following calculation the assumption is made that the breaking of the C–H is fully rate-determining. Even if this assumption is not completely true, and breaking of the C–H bond is only partially but significantly rate-determining, the calculation will be changed by only a few kcalrmol; such a change will not alter the main point of this analysis. The free energy of activation, DG ‡ for a reaction may be calculated from rate constant data using Eq. Ž 2. , in which k is the rate constant, k is Boltzmann’s constant, h is Planck’s constant, R is the gas constant and T is the absolute temperature w30x. Turnover numbers of freshly purified protein tend to vary, with the highest k cat for galactose turnover obtained in our laboratory roughly 800 sy1. From this number, we estimate the free energy of activation DG ‡ to be about 14 kcalrmol using Eq. Ž2.. ks

kT h

eyDG

‡

r RT

Ž2.

2.2.2. Estimate of thermodynamics of Step 3 Bond energies can be used to calculate an approximate change in enthalpy for the C–H bond cleavage step in GOase. The calculated D H is assumed to be close to DG since the reaction is pseudo-unimolecular and D S should be small. The bond broken in Step 3 is a C a –H of a primary alcohol, with an energy of about 93 kcalrmol w31x. The bond formed is the phenolic O–H of the Tyr272rCys228 center. This center does not have a typical phenolic O–H bond and the bond energy of an appropriate model must be used. The best model compound available is 2-sulfobutyl-4-methyl-6-Ž1X-phenyl. ethyl phenol. This model has an SCH 2 CH 2 CH 2 CH 3 group in an ortho position to the OH Ž to mimic the ortho Cys228. and a methyl group in the para position to the OH Ž to mimic the para CH 2 group which attaches Tyr272 to the protein backbone. . The additional alkyl substituent in the other ortho position may slightly lower the bond energy further compared to the Tyr272rCys228 center but the main effect should be

47

the conjugation with the ortho sulfur substituent. The phenolic O–H bond strength of 2-sulfobutyl-4methyl-6-Ž1X-phenyl. ethyl phenol is 81 kcalrmol w32x, several kcalrmol weaker than that of an underivatized phenol. According to these bond energies Hatom abstraction from the protonated alcohol would be endothermic by 12 kcalrmol. Theoretical calculations have shown that deprotonation of methanol lowers the C a –H bond energy by 16.5 kcalrmol if no counter ions are present, and by 10 to 12 kcalrmol in the presence of monovalent cations w33x. Thus, alcohol deprotonation upon coordination to copper should dramatically lower the C a –H bond energy from 93 kcalrmol to about 81–83 kcalrmol, making the hydrogen atom transfer from the C a –H to the Tyr272rCys228 center only slightly endothermic and possibly nearly isothermic Ž DG ; 0.. 2.3. Electron transfer from a ketyl radical anion to the Cu(II) center (Step 4) Studies with radical probing substrates including quadricyclanes w19x, b-haloethanols w34x, para-halobenzyl alcohols 1 and a hypersensitive trans-phenylcyclopropanemethanol probe 1 are consistent with two possible mechanisms for substrate oxidation ŽSteps 3 and 4 here. . Either the reaction proceeds through a short-lived ketyl radical anion intermediate ŽStep 3 then Step 4. or it proceeds through a closely related concerted E 2 R mechanism Ž Steps 3 and 4 become one step and there is no ketyl radical anion intermediate.. 2.3.1. Estimate of kinetics of Step 4 If a ketyl radical anion intermediate is formed, it will be very short-lived. Thus, the activation energy for the electron transfer to copper should be very low, perhaps 1–2 kcalrmol at most, as shown in Fig. 1. 2.3.2. Estimate of thermodynamics of Step 4 The thermodynamics of electron transfer from the ketyl radical anion to CuŽ II. can be calculated using redox potentials. Unfortunately the GOase CuŽ II. – CuŽI. reduction potential is not known. We have

1

B.E. Turner, B.P. Branchaud, 1997, unpublished results.

48

R.M. Wachter, B.P. Branchaudr Biochimica et Biophysica Acta 1384 (1998) 43–54

unsuccessfully attempted to measure the half-wave potential directly by cyclic voltammetry Žsee Section 4.. Nevertheless it is easy to make a semi-quantitative estimate of the potential which is adequate for our purposes. Several pieces of evidence indicate that the GOase CuŽII. –CuŽ I. reduction potential is about y0.3 to y0.5 V and possibly even slightly more negative. First, potassium ferrocyanide, with a mid-point potential of 0.42 V at pH 7.0 w35x can reduce fully oxidized enzyme to the one-electron-reduced form but it cannot reduce the CuŽ II. in the one-electron-reduced form to CuŽI. w18x, indicating that the GOase CuŽ II. – CuŽI. reduction potential must be well below 0.42 V. Second, sodium dithionite, with a reduction potential at pH 7 of y0.42 V w36x can reduce fully oxidized or one-electron-reduced enzyme all the way to the two electron reduced form in which CuŽ II. has been reduced to CuŽI. w18x. The practical limit for dithionite reduction at pH 7 is approximately y0.55 V w37x. Thus, the GOase CuŽ II. –CuŽI. reduction potential can be no lower than approximately y0.55 V. Third, the two electron enzyme is only stable under anaerobic conditions and exposure to O 2 leads to rapid and complete oxidation back to the fully oxidized enzyme w18x. Since the reduction potential Em7 of the O 2rwO 2 xyP couple is y0.45 V at pH 7 w38x, the GOase CuŽII. –CuŽ I. reduction potential must be near y0.45 V or even more negative to allow such a facile electron transfer from CuŽI. to O 2 . All of these data indicate that the GOase CuŽ II. –CuŽ I. reduction potential is about y0.3 to y0.5 V. This is significantly lower than that of aqueous CuŽ II.rCuŽI., q0.158 V w31x. The thermodynamic reduction potential for the aldehyderketyl radical anion couple is not known exactly. A lower limit may be established from the midpoint potential Em for acetaldehyde reduction measured polarographically at pH 13 in 30% ethanol which is y2.03 V vs. SCE Žy1.79 V vs. NHE. w39x. The equilibrium potential for this compound may be slightly less negative since half-wave potentials are often measured under irreversible conditions. Generally, half-wave potentials for reduction of aliphatic aldehydes in aqueous solutions fall in the range of y1.5 to y1.8 V vs. SCE Žy1.3 to y1.6 vs. NHE., and show pH dependence w40x. Aqueous aliphatic ketones are yet more difficult to reduce than aldehydes, with typical potentials around y2.2 V vs. SCE

Žy2.0 V vs. NHE. w40x, and so constitute a lower Žmost negative. limit for the reduction potential of aldehydes. On the other hand, the reduction potentials of a , b-unsaturated carbonyl compounds should serve as an upper Ž least negative. limit for the potential of interest since in these compounds, the ketyl radical species is stabilized by overlap of the unpaired electron with the adjacent p-bond. For example, the mid-point potential of 1-cyclohexenecarbaldehyde was measured polarographically to be y2.03 V vs. SCE Žy1.79 V vs. NHE. in DMF w41x. Addition of water only gave slightly more positive potentials. Though experimental conditions for determination of the potentials quoted above vary to some extent Žprotic vs. aprotic solvents, pH. , it is clear that the potential determined for the aldehyderketyl radical couple, 1.79 vs. NHE w39x, falls within the correct range. The precision of this value is adequate for our purposes since the electron transfer to copperŽ II. is so highly exothermic Ž see below. that a small change in potential would not change the overall profile and high exothermicity of the reaction to any extent. Since the p K a of the reduced species Ž the ketyl radical. and the oxidized species Ž the aldehyde. differs by many orders of magnitude, the proton equilibrium must be considered. The p K a of the ketyl radical anion Žp K red . can be approximated by the p K a of PCHŽ OH.CH 3 , the radical of ethanol Ž protonated acetaldehyde ketyl. , which has been shown to have a p K a of 11.5 w42x. The p K a of the aldehyde product Žp K ox . can be approximated by the p K a of acetone, y0.66 w43x. The midpoint potential at the reference state of pH 7, Em7 , can then be calculated using Eq. Ž3. w36x. Here, the proton concentration wHqx is set to the reference state of 10y7 M, K ox and K red are as defined above, and EmŽbases. is equal to the measured potential at pH 13, y1.79 V. Using these data, Em7 was calculated to be y2.06 V. The difference in reduction potentials, E, of the fully reduced enzyme with CuŽI. and acetaldehyde adjusted to reflect proton equilibria can be calculated according to Eq. Ž4. using a value of y0.5 V for the GOase CuŽII. –CuŽ I. reduction potential and y2.06 V for the aldehyderketyl radical anion reduction potential. The calculated E is 1.56 V. The free energy change for the electron transfer from the ketyl radical to copper was then calculated to be y36 kcalrmol according to Eq. Ž 5., where n, the number of elec-

R.M. Wachter, B.P. Branchaudr Biochimica et Biophysica Acta 1384 (1998) 43–54

trons transferred, equals one, and F is the Faraday constant. Although the GOase CuŽ II. –CuŽ I. reduction potential and the aldehyde reduction potential are not known precisely, the actual values are probably not very far away from the estimates. The calculated free energy is so large that even a error of a few tenths of a Volt Ž0.1 V ; 2.3 kcalrmol. will not change the main conclusion that the transfer of an electron from the ketyl radical anion to the copper is a highly exergonic and irreversible step. q

Em7 s EmŽ bases. y 0.06 log  Ž 1 q H q

= Ž1 q H

rK ox .

49

mic by this unusually low redox potential of thioether-derivatized Tyr272, lowered by ; 0.6 V in relation to underivatized free tyrosine w45x. The midP to H 2 O 2 in point potential for reduction of Oy 2 aqueous solution has been determined to be q0.98 V, also at pH 7 w38x. The p K a of H 2 O 2 in water has been shown to be 11.62 w46x. To analyze the energetics of H-atom transfer, the following equilibria were taken into account. Tyr-OH™ Tyr-O P q Hqq ey

Em7 s

y0.425 V

Ž7.

P q y Oy 2 q 2H q e ™ H 2 O 2

E s EmCuŽII. y Em7Žacetaldehyde.

Ž3. Ž4.

DG s ynFE

Ž5.

P y Tyr-OHq Oy 2 ™ Tyr-O P q HO 2

rK red . 4

2.4. Electron transfer from the Cu(I) center to molecular oxygen (Step 6) 2.4.1. Estimate of thermodynamics of Step 6 The reduction potential Em7 of the O 2rwO 2 xyP couple is y0.45 V at pH 7 w38x. Using that value and a value of y0.5 V for the GOase CuŽ II. –CuŽ I. reduction potential in Eq. Ž6. , the difference in potentials, E, is shown to equal 0.05 V. The free energy change can be calculated using Eq. Ž 5. to be y1.15 kcalrmol, i.e., the electron transfer to molecular oxygen is exothermic by about 1.1 kcalrmol. E s Em7Ždioxygen. y EmŽCuII.

Ž6.

2.4.2. Estimate of kinetics of Step 6 This step should be very rapid. It is thermodynamically favorable Žslightly. and there should be little reorganization of the rigid coordination geometry for the copper site in GOase. It has been given a low barrier in the energy profile Ž Fig. 1. . 2.5. H-Atom transfer from the tyrosine phenolic OH to superoxide anion radical (Step 7) 2.5.1. Estimate of thermodynamics of Step 7 The formal reduction potential of Tyr272 at pH 7 has been measured by both kinetic techniques and EPR to be 0.41 V and 0.44 V, respectively w44x. Regeneration of the phenoxy radical is made exother-

q H 2 O 2 ™ HOy 2 qH

Em7 s q0.98 V p K a s 11.62

Ž8. Ž9.

The midpoint potentials of the equilibria in Eqs. Ž7. and Ž8. were converted to free energy changes using Eq. Ž5. . The p K a of the equilibrium in Eq. Ž9. was converted to a free energy change using Eq. Ž 10., where the proton concentration was set equal to the reference state pH 7. The individual DGs were then summed to give an overall free energy change of y6.5 kcalrmol for the H-atom transfer from the –OH of Tyr272 to superoxide anion radical. DG s yRT ln  K aŽ peroxide.r Hq

4

Ž 10.

2.5.2. Estimate of kinetics of Step 7 Phenols are efficient radical scavengers via H-atom transfers such as the one in this step w45x. In analogy, the H-atom transfer from the –OH of Tyr272 to superoxide anion radical should be rapid. It has been given a low barrier in the energy profile Ž Fig. 1. . 2.6. Protonation of HO2y (Step 8) 2.6.1. Estimate of thermodynamics of Step 8 The last step in the proposed mechanism Ž Scheme 1. is proton transfer from Tyr495 to HOy 2 , resulting in regeneration of the fully active enzyme species and hydrogen peroxide product. Aldehyde product release is also assumed to occur at this point. The p K a of H 2 O 2 in aqueous solution has been shown to be 11.6 w46x. If it is assumed that Tyr495 is still not coordinated to the copper then it might be reasonable to use

50

R.M. Wachter, B.P. Branchaudr Biochimica et Biophysica Acta 1384 (1998) 43–54

Table 1 Catalytic strategies employed by the GOase active site Step

Enzymic catalysis compared to non-enzymic catalysis

2 3

Alcohol deprotonation is facilitated by copper complexation Ž; 7 p K a units, 9.7 kcalrmol.. The substrate a C–H bond is weakened by alcohol deprotonationrcopper complexation Ž‘oxyanion effect’ ; 10 kcalrmol., making the H-atom transfer from the substrate a C–H to the Tyr 272 radical site nearly isothermic. The rigid, distorted square-planar coordination destabilizes CuŽI., shifting from a normal CuŽII. –CuŽI. reduction potential of q0.158 V to a GOase CuŽII. –CuŽI. reduction potential of approximately y0.3 to y0.5 V. This avoids the formation of a CuŽI. intermediate that might otherwise be too stable. The GOase CuŽII. –CuŽI. reduction potential of approximately y0.3 to y0.5 V, caused by stabilization of CuŽII. and destabilization of CuŽI. by a distorted square-planar geometry provides rapid electron transfer from CuŽI. to molecular oxygen. Regeneration of the phenoxy radical is made exothermic by the unusually low redox potential of thioether-derivatized Tyr 272 Žlowered by ; 0.6 V or ; 14 kcalrmol relative to underivatized free tyrosine.. Deprotonation of Tyr 495 is facilitated by copper complexation.

4

6 7 8

a p K a value of 10.2, that of free tyrosine w29x or better yet the experimentally determined value of 8.8 based on active site titrations in the presence of a substrate analog w20x. Using the value of 10.2 in Eq. Ž11. provides a value of y1.9 kcalrmol for the proton transfer. Using the value of 8.8 in Eq. Ž11. provides a value of y3.8 kcalrmol for the proton transfer. It is likely that copper complexation of the tyrosine will make the tyrosine OH even more acidic so that the actual value is at least a few kcalrmol more exergonic than the y3.8 kcalrmol estimate. The copper-complexed analysis is probably the most

valid because the enzyme has Tyr495 as a ligand again for the copper. DG s yRT ln K aŽ peroxide.rK aŽtyrosine.

Ž 11.

There will be a free energy associated with release of aldehyde product. That is difficult to estimate but aldehyde release should be facile since GOase does not exhibit product inhibition, indicating that the binding of product to the enzyme is weak. 2.6.2. Estimate of kinetics of Step 8 The proton transfer in Step 6 should be rapid, as are most proton transfers between heteroatoms. 2.7. Another plausible pathway for superoxide reduction to form hydrogen peroxide The order of Step 7 ŽH-atom transfer from Tyr272 to superoxide to form HOOy. then Step 8 Ž proton transfer from Tyr495 to HOOy to form H 2 O 2 . is reasonable because both steps can be estimated to be exothermic and rapid. An alternate two-step pathway would Ž 1. transfer a proton from Tyr495 to superoxide to generate HOO P then Ž2. transfer an H-atom from Tyr272 to HOO P to form H 2 O 2 .

Fig. 1. Free energy profile calculated for the reaction catalyzed by galactose oxidase ŽNote that Step 5, O 2 binding, is not shown..

2.7.1. Transfer of a proton from Tyr495 to superoxide to generate HOO P The p K a of the conjugate acid of superoxide ŽHOO P . is 4.7 w47x. The p K a of Tyr495 is estimated to be at least 8.8 based on active site titrations in the presence of a substrate analog w20x. Using these

R.M. Wachter, B.P. Branchaudr Biochimica et Biophysica Acta 1384 (1998) 43–54

values the free energy change on transfer of a proton from Tyr495 to superoxide should be roughly q5.7 kcalrmol Ža difference of 4.1 p K a units, 1 p K a unit ; 1.4 kcalrmol.. This estimate does not include the increase in the acidity of Tyr495 by complexation to the copper, which is likely to occur in this process and should make the reaction more favorable by a few kcalrmol or more, and perhaps make the reaction exergonic. 2.7.2. Transfer of an H-Atom from Tyr272 to HOO P to form H2 O2 The bond dissociation energy of the O–H in H 2 O 2 is 90 kcalrmol w48x Žsee also commentary in Ref. w49x.. As mentioned earlier, the phenolic O–H bond strength is 81 kcalrmol w32x for 2-sulfobutyl-4methyl-6-Ž1X-phenyl. ethyl phenol as a model for the phenolic O–H bond of Tyr272. From these data the transfer of an H-atom from Tyr272 to HOO P to form H 2 O 2 can be estimated to be exothermic by about 9 kcalrmol. Thus, it is not clear which pathway is most plausible for superoxide to get a proton from Tyr495 and an H-atom from Tyr272. Both may be occurring simultaneously. Regardless of the exact mechanism for this process, one can conclude that either pathway should be energetically favorable and rapid.

3. Results 3.1. The calculated free energy profile for the galactose oxidase catalytic cycle The calculated free energy profile for the galactose oxidase catalytic cycle is shown in Fig. 1. Several key features of the profile are highlighted in Section 4. 3.2. Catalytic strategies employed by the GOase actiÕe site to ensure a rapid and efficient catalytic cycle Unique structural features of the GOase active site provide several different strategies to enhance catalysis. These strategies are summarized in Table 1 and described in detail in Section 4.

51

4. Discussion 4.1. Prediction of an unusually low enzymatic Cu potential We predict that the copper potential of GOase lies somewhere between y0.3 V and y0.5 V. According to our calculations, a positive potential for the copper couple in GOase would render the electron transfer from CuŽI. to molecular oxygen ŽStep 4 in Fig. 1. rate-limiting due to a large endothermic free energy change. This would be inconsistent with the observed GOase kinetics where H-atom abstraction is at least a partially rate-determining step and probably the major rate-determining step. Small molecule copper chelate potentials vary over a broad range, from less than y1 V to well over q1 V w50,51x, underscoring the necessity to determine the protein metal potential accurately. We have attempted to measure the half-wave potential directly by cyclic voltammetry on a bare glassy carbon electrode, following published procedures w52x. Our results, though indicative of a potential around y0.3 V, were not reproducible with this technique. It might be possible to carry out a spectrochemical redox titration monitoring copper redox state by EPR according to Dutton w36x, but this type of study has not been performed on GOase. Is a negative potential reasonable for the GOase metal center? There is no known precedent for a protein copper with a negative potential. However, a rigid distorted square planar ligand field of the four protein ligands for copper in GOase Ž Tyr495, Tyr272, His496 and His581. should favor CuŽ II. over CuŽ I. , thus leading to a negative CuŽII.rCuŽI. redox potential. There is evidence that the GOase protein fold around the copper site is rigid and may not be able to respond to the geometric demands of CuŽ I. , which is known to prefer a tetrahedral geometry w53x. The protein matrix of the active site is thought to be rigid due to the high content of aromatic residues and the thioether crosslink. Few structural changes are observed in the apoprotein with the copper removed w15x. No other metals have been reported to bind to GOase w15x, suggesting that the copper binding site is a rigid cavity of fixed hole size and fixed stereochemistry w54x. We hypothesize that the protein rigidity

52

R.M. Wachter, B.P. Branchaudr Biochimica et Biophysica Acta 1384 (1998) 43–54

maintains a distorted square-planar geometry which is better for CuŽII. than CuŽ I. , thus lowering the metal potential. This explanation is a standard entatic state theory explanation of metalloenzyme control of metal redox potential. In other systems it is known that the nature and the extent of the coordination geometry distortion can perturb the redox potential over a broad range w55–57x. The predicted GOase CuŽ II. rCuŽI. redox potential of about y0.3 to y0.5 V is found to be quite plausible when the ligand field for copper at the GOase active site is compared with model compounds. The redox potentials of several four coordinate copper complexes with substituted sal and salen ligands have been measured w51,58x. These complexes are reasonable models for the copper in the GOase active site since the models have two phenolic oxygens as ligands to copper Ž analogous to Tyr495 and Tyr272 in GOase. and two imine nitrogen ligands Žanalogous to His496 and His581 in GOase.. Redox potentials measured in DMF Ž an environment somewhat similar to a protein active site. fall in the range of y0.42 to y0.97 V. Substituents in the ligand which distort the coordination from square planar raise the redox potential, with the y0.42 value for the most distorted w51,58x. The four protein ligands to copper in GOase Ž Tyr495, Tyr272, His496 and His581. provide an analogous distorted square planar ligand field for the copper. One might expect that other copper proteins with similar geometry would exhibit redox potentials similar to that of GOase, yet there is no known precedent for a negative-potential copper in proteins. Blue, type 1 proteins vary over a range of 0.6 V, with all potentials higher than that of the aquo ion, 0.15 V w59x. The redox properties of type 2 copper proteins have been less well characterized, though the potential of the distorted square–pyramidal copper center of superoxide dismutase has been shown to lie around 0.4 V, and the type 2 copper of the multicopper oxidase laccase has been shown to lie between 0.36 and 0.39 V w60x. These potentials are not nearly as low as that estimated for GOase copper, though it appears that geometry and active-site rigidity are similar. No simple relationship seems to exist between protein copper redox potentials and EPR parameters that probe for geometry w59x, indicating that the metal potential in proteins is modulated by a

complex set of conditions, not all of them well understood. It is likely that the thioether derivatization in the ortho position of the Tyr272 phenolate contributes to copper redox modulation. Remote substituents on aromatic ligands have been shown to exert a large influence on the electrochemical properties of copper. For example, in the case of bisŽ N X-arylpyrrole-2carboxaldiminato. copper chelates, it has been found that changing the inductive properties of the substituent in the para position of the aryl ring has a remarkable effect on redox potentials w59x. In particular, strongly electron-donating substituents lower the potential significantly, stabilizing CuŽ II. over CuŽ I. . The p K a of methyl thiocresol has been shown to be lowered by more than one order of magnitude in comparison to cresol w61x. Other factors, such as the dielectric constant of the protein matrix around the metal site, type of residues outside the first coordination shell, charge distribution and local polarizability w62x, are more difficult to quantitate.

5. Conclusions We have constructed a reaction energy profile for the complete catalytic cycle of the GOase reaction that is consistent with what is known about the enzyme energetics. Based on the energetics of the individual steps, we predict an unusually low copper redox potential for GOase, roughly between y0.3 and y0.5 V. There is no known precedent for an enzymatic copper potential below 0 V, and the prediction should be testable with the use of an appropriate technique such as an anaerobic redox titration by EPR.

Acknowledgements This work was supported by the National Science Foundation ŽGrant MCB-9311514. and the National Institutes of Health ŽGraduate Training in Molecular Biology and Biophysics 2T32GM07759. .

R.M. Wachter, B.P. Branchaudr Biochimica et Biophysica Acta 1384 (1998) 43–54

References w1x w2x w3x w4x w5x w6x

w7x w8x w9x w10x w11x w12x

w13x w14x

w15x w16x

w17x

w18x w19x w20x w21x w22x w23x w24x w25x w26x w27x w28x w29x

J.Z. Pederson, A. Finazzi-Agro, ` FEBS 325 Ž1993. 53–58. E.N.G. Marsh, BioEssays 17 Ž1995. 431–441. J.P. Klinman, Chem. Rev. 96 Ž1996. 2541–2560. J. Stubbe, Annu. Rev. Biochem. 58 Ž1989. 257–285. R.C. Prince, Trends Biochem. Sci. 13 Ž1988. 286–288. H. Sigel, A. Sigel ŽEds.., Metal Ions in Biological Systems, Metalloenzymes Involving Amino Acid-Residue and Related Radicals, Vol. 30, Marcel Dekker, New York, 1994. C.V. Parast, K.K. Wong, J.W. Kozarich, Biochemistry 34 Ž1995. 5712–5717. S. Licht, G.J. Gerfen, J. Stubbe, Science 271 Ž1996. 477– 481. T.T. Harkins, C.B. Grissom, Science 263 Ž1994. 958–960. Y. Zhao, A. Abend, M. Kunz, P. Such, J. Retey, Eur. J. ´ Biochem. 225 Ž1994. 891–896. A. Tsai, R.J. Kulmacz, G. Palmer, J. Biol. Chem. 270 Ž1995. 10503–10508. C.W. Hoganson, N. Lydakis-Simantiris, X.-S. Tang, C. Tommos, K. Warncke, G.T. Babcock, B.A. Diner, J. McCracken, S. Styring, Photosynthesis Res. 46 Ž1995. 177–184. ¨ Z.B. Ogel, D. Brayford, M.J. McPherson, Mycol. Res. 98 Ž1994. 474–480. D.J. Kosman, in: R. Lontie ŽEd.., Copper Proteins and Copper Enzymes, Vol. 2, CRC Press, Boca Raton, FL, 1984, pp. 1–26. N. Ito, S.E.V. Phillips, K.D.S. Yadav, P.F. Knowles, J. Mol. Biol. 238 Ž1994. 794–814. G.T. Babcock, M.K. El-Deeb, P.O. Sandusky, M.M. Whittaker, J.W. Whittaker, J. Am. Chem. Soc. 114 Ž1992. 3727– 3734. N. Ito, S.E.V. Phillips, C. Stevens, Z.B. Ogel, M.J. McPherson, J.N. Keen, K.D.S. Yadav, P.F. Knowles, Nature 350 Ž1991. 87–90. M.M. Whittaker, J.W. Whittaker, J. Biol. Chem. 263 Ž1988. 6074–6080. B.P. Branchaud, M.P. Montague-Smith, D.J. Kosman, F.R. McLaren, J. Am. Chem. Soc. 115 Ž1993. 798–800. M.M. Whittaker, J.W. Whittaker, Biophys. J. 64 Ž1993. 762–772. A. Maradufu, G.M. Cree, A.S. Perlin, Can. J. Chem. 49 Ž1971. 3429–3437. C.A. Fiercke, G.G. Hammes, Meth. Enzymol. 249 Ž1995. 3–37. W.J. Albery, J.R. Knowles, Biochemistry 15 Ž1976. 5627– 5631. R.M. Wachter, B.P. Branchaud, J. Am. Chem. Soc. 118 Ž1996. 2782–2789. G.J. Kearley, F. Fillauz, M.-H. Baron, S. Bennington, J. Tomkinson, Science 264 Ž1994. 1285–1289. R.P. Houghton, Metal Complexes in Organic Chemistry, Cambridge Univ. Press, Cambridge, UK, 1979, pp. 106–107. F. Achenza, Ann. Chim. 54 Ž1964. 240–245. W.K. McEwen, J. Am. Chem. Soc. 58 Ž1936. 1124–1129. G.D. Fasman ŽEd.., CRC Handbook of Biochemistry and

w30x

w31x

w32x w33x w34x w35x w36x w37x w38x w39x w40x

w41x w42x w43x

w44x

w45x w46x w47x w48x w49x w50x

w51x

w52x w53x

w54x w55x

53

Molecular Biology, 3rd edn., Vol. 1, CRC Press, Cleveland, OH, 1976, p. 305. J.I. Steinfeld, J.S. Francisco, W.L. Hase, Chemical Kinetics and Dynamics, Prentice-Hall, Englewood Cliffs, NJ, 1989, p. 322. Robert C. Weast ŽEd.., CRC Handbook of Chemistry and Physics, 55th edn., CRC Press, Cleveland, OH, 1974, p. D120 and p. F214. E.T. Denisov, Handbook of Antioxidants, CRC Press, New York, NY, 1995, p. 53. M.L. Steigerwald, W.A. Goddard III, D.A. Evans, J. Am. Chem. Soc. 101 Ž1979. 1994–1997. R.M. Wachter, M.P. Montague-Smith, B.P. Branchaud, J. Am. Chem. Soc. 119 Ž1997. 7743–7749. J. O’Reilly, Biochim. Biophys. Acta 292 Ž1973. 509–515. L.P. Dutton, Meth. Enzymol. 54 Ž1978. 411–435. F.A. Armstrong, J.N. Butt, A. Sucheta, Meth. Enzymol. 227 Ž1993. 479–500. R. Malkin, B.G. Malmstrom, ¨ Adv. Enzymol. 33 Ž1970. 177–244. A.M. Shur, Y.S. Lyalikov, M.M. Filimonova, J. Gen. Chem. ŽUSSR. 36 Ž1966. 417–420. C.K. Mann, K.K. Barnes, Electrochemical Reactions in Nonaqueous Systems, Marcel Dekker, New York, NY, 1970, pp. 177–178. H.O. House, L.E. Huber, M.J. Umen, J. Am. Chem. Soc. 94 Ž1972. 8471–8475. G.P. Laroff, R.W. Fessenden, J. Phys. Chem. 77 Ž1973. 1283–1288. ¨ Haldna, A.J. Talvik, in: Saul Patai ŽEd.., V.A. Palm, U.L. The Chemistry of the Carbonyl Group, Interscience Publishers, London, UK, 1966, pp. 421–460. G.A. Hamilton, P.K. Adolf, J. deJersey, G.C. DuBois, G.R. Dyrkacz, R.D. Libby, J. Am. Chem. Soc. 100 Ž1978. 1899– 1912. M.R. DeFelippis, C.P. Murthy, M. Faraggi, M.H. Klapper, Biochemistry 28 Ž1989. 4847–4853. W.H. Richardson, V.F. Hodge, J. Org. Chem. 35 Ž1970. 4012–4016. D.T. Sawyer, J.S. Valentine, Accts. Chem. Res. 14 Ž1981. 393–400. D.T. Sawyer, J. Phys. Chem. 93 Ž1989. 7977–7978. D.T. Sawyer, J. Phys. Chem. 94 Ž1990. 5412–5416. E.-I. Ochiai, in: D.H. Bush, H. Shull ŽEds.., Bioinorganic Chemistry An Introduction, Allyn and Bacon, Boston, MA, 1977, p. 236. S.J. Lippard, J.M. Berg, in: Aidan Kelly ŽEd.., Principles of Bioinorganic Chemistry, University Science Books, Mill Valley, CA, 1994, p. 27. W.R. Hagen, Eur. J. Biochem. 182 Ž1989. 523–530. G. Wilkinson, R.D. Gillard, J.A. McCleverty ŽEds.., Comprehensive Coordination Chemistry, Pergamon, Oxford, UK, 1987, pp. 538–539 and pp. 594–595. R.J.P. Williams, Eur. J. Biochem. 234 Ž1995. 363–381. B.L. Vallee, R.J.P. Williams, Proc. Natl. Acad. Sci. U.S.A. 59 Ž1968. 498–505.

54

R.M. Wachter, B.P. Branchaudr Biochimica et Biophysica Acta 1384 (1998) 43–54

w56x U. Ryde, M.H.M. Olsson, K. Pierloot, B.O. Roos, J. Mol. Biol. 261 Ž1996. 586–596. w57x P. Ghosh, D. Shabat, S. Kumar, S.C. Sinha, F. Grynszpan, J. Li, L. Noodleman, E. Keinan, Nature 382 Ž1996. 339–341. w58x G.S. Patterson, R.H. Holm, Bioinorg. Chem. 4 Ž1975. 257– 275. w59x A.W. Addison, in: K.D. Karlin, J. Zubieta ŽEds.., Copper Coordination Chemistry: Biochemical and Inorganic Per-

w60x w61x w62x

spectives, Adenine Press, New York, NY, 1982, pp. 109– 128. L. Morpurgo, L. Calabrese, A. Desideri, G. Rotilio, Biochem. J. 193 Ž1981. 639–642. M.M. Whittaker, Y.Y. Chuang, J.W. Whittaker, J. Am. Chem. Soc. 115 Ž1993. 10029–10035. G.W. Canters, G. Gilardi, FEBS Lett. 325 Ž1993. 39–48.