Electron paramagnetic resonance studies of the tungsten-containing formate dehydrogenase from Clostridiumthermoaceticum

Vol. 149, No. 2, 1987 December

BIOCHEMICAL

AND BIOPHYSICAL

16, 1987

RESEARCH COMMUNICATIONS Pages 424-430

Electron ParamagneticResonanceStudiesof the Tungsten-ContainingFormate Dehydrogenasefrom Clostridium thermoaceticum JosephC. Deatonaand Edward I. Solomon* Departmentof Chemistry, Stanford University Stanford, CA 94305 Gerald D. Watt* Batelle-Kettering ResearchLaboratories Yellow Springs,Ohio 45387 Phyllis J. Wetherbee and CharlesN. Durfor*b GTE Laboratories,Inc. Waltham, MA 02254 Received

October

16,

1987

Summary: The redox centers in the tungsten-containing formate dehydrogenase from Clostridium thermoaceticum were examined by potentiometric titration and electron paramagnetic resonancespectroscopy. At low temperature two overlapping iron-sulfur signalswhich correlated with enzymatic activity were observedwith formal potentialsnear -400 mV vs. SHE. Basedon their temperature dependences,one signal is assignedto a reduced Fe2S2 cluster and one to a reduced Fe4S4 cluster. Quantitation of signal intensity suggeststwo Fe2S2 and two Fe4S4 clusters per formate dehydrogenase molecule. Another signal (g= 2.101, 1.980, 1.950) present in low concentrations at more negative potentials was observableup to 200’ K and is not attributed to any iron-sulfur cluster. The possible orgin of this signal is analyzed using ligand field theory, and the redox behavior is consideredwith respectto possibleligation at the active site. 0 1987 Academic Press, 1°C.

Formate dehydrogenases(FDH) catalyze the reversible interconversion of formate and CO2 in numerousplants and bacteria (for a review see 1). Most are molybdenum enzymes, but a few require tungsten for activity and constitute the only known occurrence of a third transition series elementin metalloenzymes. For eachmolybdenum-containingFDH (MO-FDH) examinedby either an assayinvolving nit-l Neurosnoracrassamutants(2)or the fluorescenceof oxidatively degraded protein (3) evidence hasbeenfound for a non-nitrogenasetype molybdenum-containing cofactor, called “MO-CO”,in which a pterin is thought to coordinate to the MO. In the presenceof MoO42-, acidified samplesof the tungsten-containingFDH (W-FDH) from Clostridium thermoaceticumalso activated the u nitrate reductaseindicating the presenceof a tungstenanalogueof the MO-CO(~), which is consistentwith the fluorescenceresultsof Yamamoto et al. (5).

* To whom correspondenceshouldbe addressed a. Current Address: ResearchLaboratories,EastmanKodak Co., Rochester,NY 14650 b. Current Address: IGEN, Inc. Rockville, MD 20852. 0006-291X/87 $1.50 Copyright 0 1987 by Academic Press, Inc. All rights of reproduction in any form reserved.

424

Vol. 149, No. 2, 1987

BIOCHEMICAL

AND BIOPHYSICAL

RESEARCH COMMUNICATIONS

EPR and EXAFS analysesare beginningto yield significant insightsinto the active site structures of these MO and W enzymes. Potentiometric titration of the MO-FDH from Methanobacteriumformicium followed by EPR spectroscopy(6)revealedsignalsfrom onereduced iron-sulfur cluster and a MO(V) species. This MO(V) specieswas readily observable under reducing conditions, becausea large separationexists betweenMo(VI)/Mo(V)

and Mo(V)/Mo(IV)

reduction potentials. This MO(V) EPR signalwas atypical relative to xanthine oxidaseand related MO enzymesbecauseit exhibited a g,, > 2.0. Potentiometrictitration andEPR spectroscopyof the Clostridium uasteurianum MO-FDH revealed two distinct iron-sulfur centers and a free radical which was suggestedto arise from the pterin component of the MO-CO(~). However, no MO(V) EPR signal was observed. EXAFS studies on this protein indicated the presenceof three 0x0 groups, but no MO-S bonds(8). In addition, these studies suggestedthat the MO center is not reducedupon addition of formate or dithionite. The W-FDH from Clostridium thermoaceticum is a 340K dalton protein of a2P2 subunit structure which contains 2W and 2 Se per holo enzyme. The data on Fe content is lessclear, with between 20 to 40 Fe atomsper holo enzyme being observedin different samples(5,9). Recently, studiesinvolving the inhibition of enzymatic activity with butanedioneimplied that CO2 reduction and NADPH oxidation in this protein occur at separatebinding sites(l0). In contrast to the C. pasteurianumMO-FDH, the EXAFS of this FDH in the presenceof dithionite showedno evidence for W 0x0 bonds, but it did exhibit a coordination spherecontaining > 2 W-O or N and >2 W-S bonds(8). The present study probes the redox active sites of this W-FDH from C. thermoaceticum usingpotentiometric methodsand EPR spectroscopy. Materials and Methods: Formate dehydrogenasefrom C thermoaceticumwaspurified and assayed asdescribedpreviously(5,9), except that stepsthree (ammoniumsulfate, celite chromatography) and four (polyethylene glycol fractionation) were omitted. Instead, proteins in the heat treatment supernatantwith molecularweightsgreaterthan 1OOKdaltonswere separatedfrom the remainderof the cellular constituents by hollow fiber filtration (Amicon HP-100). This required filtration of approximately 500 ml of the heat treatedsupernatantagainst2.0 1of 50 mM triethanolamine-maleate buffer pH 7.5 containing 20 % glycerol, 10 mM NaN3 and 2.0 mM Na2S204. The resulting high molecular weight fraction was then precipitated with ammonium sulfate (45%) and chromatographed on a Bio-RAD A 1.5M molecular radius column as described in step five(5). While this use of hollow fiber filtration doesnot greatly improve the overall preparatory yield, it doesresult in a substantialtime savings.Enzymatic specific activity is reported in units of ymol of NADPH formed per min per mg of protein at 50°C in the presenceof 40 mM formate and 1 mM NADP. Anaerobic solutions of FDH were set to the desired potential as described previously (11). Solutions containing final concentrations of 25 yM indigo disulfonate, phenosafronine, benzyl viologen and methyl viologen were usedasmediators. After eachenzyme solution hadreachedthe desiredpotential, 200 11 sampleswere transferred anaerobically to an EPR tube and flame-sealed under vacuum at 77’K. EPR spectra and elemental analyses were performed as described previously(4,9). CuEDTA was usedas a standardfor spin quantitation with g-value correctionsof Aasa and Vanngard(12). ResultsandDiscussion FDH poised at a potential of -420 mV vs. SHE exhibited overlapping EPR signalsat 15°K (Fig. 1A). The g-valuesand the low temperaturesrequired for observing thesesignalsindicate that 425

Vol. 149, No. 2, 1987

BIOCHEMICAL

A.

AND BIOPHYSICAL

2.01 2Y31

196

1.69

i'a4r

15

0.

RESEARCH COMMUNICATIONS

2.93

K

I.64 I

Figure1.

(A) EPR spectrum of FDH (28mg!ml,400specificactivity) poisedat -420mV, T=lS’K. Conditions for EPR: 9.051 GHz, 10 mW, SG modulation, gain 2.5 x 104. (B) Same as (A) exceptT=3S°Kandgain= 5 x 104. (C) EPRof FDH (3 mghl, 310 specific activity) reduced in the absence of mediators with 10 mM sodium formate. Conditions for EPR: Same as in (A) except gain = 8 x 104. 4 denotes mediator signal (g = 2.0023); * denotes signalsnot correlatedwith activeenzyme.

they arise from reduced iron-sulfur centers. Resolution of the signalsis achieved above 35“K, where only an axial signal (gz = 2.028, gx,y = 1.940) is prominent up to 125°K (Fig. 1B). This temperaturedependenceis characteristicof Fe2S2centers(13) except for one of the Fe2S2 centers presentin xanthine oxidase which alsohasdifferent g-value characteristics(l4,15). The remaining major featuresin Fig. IA are prominent only below 2YK and on this basisthey can be assignedto a Fe4S4 center with gz= 2.01, gy= 1.958, g,= 1.891. The gz component can be more clearly observedwhen FDH is reducedby formate in the absenceof any mediator (Fig. 1C). Other minor features(indicated by asterisksin Figs. 1A and 1C) could not be examinedin any detail under most conditions becauseof their low intensity. However, thesedo not appearto correlatewith enzymatic activity . The EPR signal intensity from both the Fe2S2 and Fe4S4 centers displayed similar dependenceson applied potential with midpoint potentials at approximately -400 mV vs. SHE. Double integration of total EPR signal intensity from both iron-sulfur centers in several fully reduced samplesgave values between 2.5 to 4.5 spins per 340 K daltons with larger spin concentration values being obtained for higher activity samples.This variation in spin intensities 426

Vol. 149, No. 2, 1987

BIOCHEMICAL

AND BIOPHYSICAL

RESEARCH COMMUNICATIONS

results from the difficulty in purifying fully active, homogeneousFDH samples.However, these results suggestthat there are two Fe4S4 and two Fe2S2 centers per FDH molecule, which is consistent with the a2P2 structure of the enzyme and the presenceof two catalytic sites per molecule. FDH isolatedfrom C thermoaceticumcells grown on exclusively MO-containing media showedqualitatively the sameiron-sulfur signalsand redox behavior asthe W-FDH (not shown). However, the signalintensitieswere much lower, which correlateswith the decreasedFDH activity found in cells grown without tungsten(9). Two Fe4S4 and two Fe2S2 centersper molecule would account for a total of 12 Fe atoms, whereas22 Fe were determinedby elementalanalysesof the presentpreparation. Highly variable iron concentrationswere also observedin earlier studieson this enzyme(59). The variability and excessFe over that observedby EPR may be due to adventitiously bound Fe centerswhich are not involved in catalysis and/or other FeS clusters which are not detectable by EPR, such as those found in nitrogenase(16). Similarly, the Fe concentration determinedby analytical techniquesfor the FDH isolated from c pasteurianumwasalso substantially greater than that quantified by EPR spectroscopy(7) . A variability in the EPR signal intensity of the c thermoaceticum FDH in nominally 2.0 mM Na2S204 buffer was also observed previously(9). It is now clear that the signal intensity was low because the enzyme was only partially reduced by either gradual decompositionof Na2S204 and/or oxidation of the enzyme by bicarbonatein the buffer. Figures 2 A-C show variable temperatureEPR spectraof a W-FDH samplereducedto -450 mV. A new signal (g=2.101,1.980,1.950) may be observed up to 200’K with a temperature dependenceand g-value characteristicsunlike any reported for iron-sulfur centers. The signal is alsonot easily associatedwith any first seriestransition elementor a seleniumradical(l7). The high temperature signal was also weakly observed in the FDH samplepoised at -420 mV (at a higher EPR gain than Fig lA), but not at more positive potentials, indicating a formal potential more negative than -450 mV. Double integration of the -450 mV signal yielded 0.07 spinsper FDH molecule while 0.89 W and co.01 MO atomsper moleculewere determinedby elementalanalysis. Theseresultswould be consistentwith a smallconcentrationof an EPR detectableW(V) speciesin equilibrium with the EPR silent W(IV) and W(V1) speciesat this potential, which is analogousto the redox behavior of many MO enzymes. At this point one can useligand field theory to evaulate whether it is plausible for this high temperature signal to be from W(V). g, > 2 is not simply predicted for a one electron transition metal complex(l8). DeArmond et a1.(19) have examined g-values in group VIB oxohalides including the contribution of spin-orbit interactionson ligand nuclei which result upon admixture of ligand orbital characterinto the metal d-orbitals. Expressionsfor the antibondingorbitals, primarily of metal d character,which are pertinent to the presentdiscussionare:

bl” = Pl d ~2-~2 - Pl’(Pbl b.* = 02 d xy - P2”Pb2 where qbl and (Pb2 are symmetry adapted linear combinations of ligand orbitals of bl and b2 symmetry, respectively, and the p’s are mixing coefficients due to covalency. In the ground state 427

Vol. 149, No. 2, 1987

BIOCHEMICAL

AND BIOPHYSICAL

2.01 2C@

203

1.96 (ia"<

RESEARCH COMMUNICATIONS

189

194

8.

2.;1o C

1.98 1 ff

1.95

175K lr----

Figure2.

EPRof FDH (28mg/ml,400specificactivity) poisedat -450mV. (A) lYK, 10 mW, 5Gmodulation,gain= 2 x 104. (B) 1lOoK,50 mW, 5G modulation,gain= 5 x 104. (C) 175°C50mW, 5Gmodulation,gain1.25x 105. 4 and * asin Figure1.

of a W(V) complex, one electron is in the b2* orbital. A second-orderperturbation treatmentfor g, gives: g, - 2.0023 = E(bl’)&b2*)

(2P1P25~ - &‘P~‘CL) @Pip2 - Pi’P2’)

Equation 1

where CM and 5~ are one-electron spin-orbit coupling constants for the metal and ligands, respectively, and E(bl*) - E(b2*) is the energy difference betweenthe bl* and b2* levels. In the caseof W, 5~ (estimatedat 5000 cm- 1, Kon and Sharpless(20))is much larger than any possible 6~ (5s = 380 cm-l, cSe = 1870cm- 1, Blume and Watson(21)) making it unreasonableto have g, > 2.0 by this mechanismwith meaningful values of the mixing coefficients. However, Kon and Sharpless(21,22)have included an additional term involving a charge transfer excited statewhich coupleswith the b2* ground state. The bl bonding level of mainly ligand characterassociatedwith this chargetransfer transitionis: bl = Pl’Pbl + Pl’dx2 - y2 428

Vol. 149, No. 2, 1987

BIOCHEMICAL

AND BIOPHYSICAL

RESEARCH COMMUNICATIONS

and the expressionfor gz now alsoincludes: g, - 2.0023 = Eq. 1 + E(b1;-2E(b2*)

(2PilP2S~ + P1P25~) (2Pi’P2 + P1P2’) Equation 2

In the new term, the spin-orbit coupling on the metal and ligand both give positive contributions to the g-shift. Thus, it becomespossible for a W(V) complex with a low energy charge transfer excited state to have gz > 2.00. Hanson et a1.(23,24)have synthesized several such complexes including WO(SePh)4- (g = 2.086, 1.923; h,,,

= 588 nm) and WO(SPh)4- (g ;: 2.018, 1.903;

h max = 525 nm). Taking the oxothiolate complex as a reference point, we find through trial calculationsthat it is possibleto obtain g, = 2.1 with a rather small (< 5%) increasein admixture of d,2 _ y2 with ligand in the b1* orbital. This could readily result from removal of the stabilizing 0x0 group from the modelcomplex, which donateselectrondensity to the metal through the d,, anddyz orbitals. EXAFS has indicated that FDH is lacking a W=O bond in the presenceof dithionite(8). Substitution of Se for S ligation would also contribute to a more positive g-shift, but it shouldnot be concluded that a W(V) complex must have Se coordination to exhibit g-values such asthosein Figure 2C. In the caseof the inactive tungsten-substitutedderivative of sulfite oxidase (SO), dithionite reduction proceedsno farther than the W(V) oxidation state,and the g-values are all lessthen 2.0 (25). It may be inferred from EXAFS studieson the MO-SO that there are 0x0 groupsboundto the metal in the oxidized and reduced states(26). In view of the lack of any W=O bonds in the dithionite-reducedC thermoaceticumW-FDH, the W active siteis expectedto have a morepositive reduction potential than in W-SO. This would produceredox behavior similar to the & formicicum MO-FDH where reduction of MO from the VI to V and V to IV oxidation statesoccurs at -330 mV and -470 mV, respectively(6). Alternatively, for the c pasteurianumMO-FDH, three Mo=O bonds are presentand the MO apparently doesnot undergoreduction(g). In order to prove whether the high temperaturesignalarisesfrom the tungsten active site in the C, thermoaceticum W-FDH, additional studiesinvolving l83W (I= l/2) isotopic substitution and correlation with enzyme activity are required. The presentwork has limited the conditions under which a W(V) EPR signalmay be presentand consideredthe range of g-values to expect, in addition to characterizing the iron-sulfur redox centers.

Acknowledeement: This work wassupportedby GTE Laboratories,Incorporated.

1. Adams, M.W.W. & Mortenson, L.E. (1985) in Molybdenum Enzymes (Spiro, T.G. Ed.) pp. 519-594, Wiley, New York. 2. Nason, A., Lee, K.-Y., Pan, S.-S., Ketchum, P.A., Lamberti, A. & DeVries, J. (1971) proC. Natl. Acad. Sci. USA 68, 3242-3246. 3. Johnson, J.E. & Rajagopalan,K.V. (1982) Proc. Natl. Acad. Sci. USA 79,6856-6860. 4. Deaton, J.C., Solomon, E.I., Durfor, C.N., Wetherbee, P.J., Burgess, B.K. & Jacobs, D.B. (1984) Biochem. Biophys. Res. Comm. 121, 1042-1047. 5. Yamamoto, I., Saiki, T., Liu, S.-M. & Ljungdahl, L.G. (1983) ,I. Biol. Chem. 258, 1826-1832. 429

Vol. 149, No. 2,1987

6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16.

24.

BIOCHEMICAL

AND BIOPHYSICAL

RESEARCH COMMUNICATIONS

Barber, M.J., Siegel, L.M., Schauer, N.L., May, H.D. & Ferry, J.G. (1983) 1. m. Chem. u 10839-10845. Prince, R.C., Liu, C.-L., Morgan, V.T. & Mortenson, L.E. (1985) FEBS I.etters 189, 263-266. Cramer, S.P., Liu, C.-L.. Mortenson, L.E.. Snence. J.T.. Liu. S.-M.. Yamamoto. , I. &.~ Ljungdahl, L.G.. (1985) ,I. Inorg. B&hem. 23 119-124.’ ’ ’ Durfor, C.N.. Wetherbee, P.J., Deaton, J.C. & Solomon, E.I. (1983) Biochem. Biophys. I&s. Comm. 115,61-67. Wetherbee, P.J., Thornton, L.S. & Durfor, C.N. (1986) B&him. Biophys. &&!Q, 12-19. Watt, G.D. (1979) Anal. Biochem. !& 399-404. Aasa, R. & Vanngard, T. (1975) 1. &@n. Res. fi, 308-315. Or-me-Johnson, W.H. (1973) &Q. &. Biochem. 42, 159-204. Lowe, D.J., Lynden-Bell, R.M. & Bray, R.C. (1972) B&hem. 1. .QQ, 239-247. Wong, G.B., Kurtz, D.M., Holm, R.H., Mortenson, L.E. & Upchurch, R.C. (1979) 1. Am. Chem. Sot. 101, 3078-3090. Zimmerman, R., Munck, E., Brill, W.J., Shah, V.K., Henzl, M.T., Rawlings. J. & &me-Johnson, W.H. (1978) Biochim. Biophys. Acta 5=, 185-207. Jones, M.T., Jansen, S., Acampora, L.A. & Sandman, D.J. (1983) 1. Physia_%& C3-1159. Figgis, B.N. (1966) “Introduction to Ligand Fields,” Wiley, London. DeArmond, K., Garrett, B.B. & Gutowsky, H.S. (1965) 2. Chem. Phq. 4, 1019-1025. Kon, H. & Sharpless, N.E. (1966) L. m. Chem. 70, 105-111. Blume, M. & Watson, R.E. (1963) m. &y. Sot. London A271,565-578. Kon, H. & Sharpless, N.E. (1965) 1. m. Chem. 42, 906-909. Hanson, G.R., Brunette, A.A., McDonell, A.C., Murray, K.S. & Wedd, A.G. (1981) 2. Am. Chem. Sot. lm 1953-1959. Hanson, G.R., Wilson, G.L., Bailey, T.D., Pilbrow, J.R. & Wedd, A.G. (1987) J. Am. Chem. Sot. me, 2609-2616. Johnson, J.E. & Rajagopalan, K.V. (1976) ,I. Biol. Chem. 251, 5505-5511. Cramer, S.P., Wahl, R., Rajagopalan, K.V. (1981) ,I. Am. Chem. Sot. 103, 7721-7727.

430