Thermodynamic and electron paramagnetic resonance characterization of flavin in succinate dehydrogenase.

Thermodynamic parameters of succinate dehydrogenase flavin were determined potentiometrically from the analysis of free radical signal levels as a function of the oxidation-reduction potential. Midpoint redox potentials of consecutive 1-electron transfer steps are -127 and -31 mV at pH 7.0. This corresponds to a stability constant of intermediate stability, 2.5 x 10(-2), which suggests flavin itself may be a converter from n = 2 to n = 1 electron transfer steps. The pK values of the free radical (FlH . in equilibrium Fl . -) and the fully reduced form (FlH2 in equilibrium FlH-) were estimated as 8.0 +/- 0.2 and 7.7 +/- 0.2, respectively. Succinate dehydrogenase flavosemiquinone elicits an EPR spectrum at g = 2.00 with a peak to peak width of 1.2 mT even in the protonated form, suggesting the delocalization in the unpaired electron density. A close proximity of succinate dehydrogenase flavin and iron-sulfur cluster S-1 was demonstrated based on the enhancement of flavin spin relaxation by Center S-1.

It is generally accepted that the succinate dehydrogenase molecule consists of two subunits: one flavo iron-sulfur subunit with M, of approximately 70,000 and one iron-sulfur subunit of 27,000. The flavo iron-sulfur subunit contains 4 Fe, 4 S,' and one covalently bound FAD; the iron-sulfur subunit contains 4 Fe and 4 S (1-4). We have proposed the presence of two binuclear iron-sulfur clusters (Center S-1 and S-2) in the flavo iron-sulfur subunit and one tetranuclear cluster (Center S-3) in the iron-sulfur subunit based on the EPR and thermodynamic characteristics of these centers and on the correlation between enzymic activities toward various artificial electron acceptors and functional redox components which were studied with various types of soluble succinate dehydrogenase preparations, such as reconstitutively active BS-SDH,2 reconstitutively inactive B-SDH, and AA-SDH (2, 5, 6). * This work was supported by National Institutes of Health Grants GM12202, GM16767, GM25052, and HL12576, and by National Science Foundation, Grant PCM-78-16779. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ' Preliminary results of this study were presented at the International Symposium on Frontiers of Bioenergetics (7).
In all the succinate dehydrogenase preparations we have so far reported, Center S-2 is EPR-detectable and its spin relaxation behavior seems to be highly sensitive to the molecular environment around the cluster (2, 5); a weak dipole-dipole interaction between Centers S-1 and S-2 was observed which was manifested as the relaxation enhancement of S-1 spins by S-2 or as either splitting or broadening of low temperature EPR spectra of fully reduced enzymes. The spatial relationship between Center s-1 and s-2 in different succinate dehydrogenase preparations has been discussed in detail in a preceding paper (5).
In this paper, we report thermodynamic parameters and EPR characteristics of the succinate dehydrogenase flavin and discuss a possible mechanism for flavin to convert the n = 2 electron transfer step to that of n = 1 in a mitochondrial primary dehydrogenase. A relatively short Center S-1 to flavin distance is implied by the spin-spin interaction we report, supporting our earlier assignment of cluster S-1 and S-2 in the flavo iron-sulfur subunit.

EXPERIMENTAL PROCEDURES
Potentiometric titration of the flavin-free radical of succinate dehydrogenase was conducted anaerobically as described by Dutton (8).
The following redox-mediating dyes weremed selectively to minimize interference from the dye g = 2.00 signals; 1-4-naphthoquinone, duroquinone, indigo disulfonate, indigo tetrasulfonate, 2-OH-1,4naphthoquinone, phenosafranine, and safranine T; each at a final concentration of 50 PM. These dyes undergo redox changes in close to n = 2 processes and variations of the dye concentration from 10 to 500 p~ did not affect the size of the measured g = 2.00 signal. In addition, dye signals can be distinguished from that of the flavin free radical associated with succinate dehydrogenase based on their slower spin relaxation rates similar to flavodoxin g = 2.00 signals as will be described later. Spin quantitation of the free radical signals at titration peaks were conducted by double integration of the g = 2.005 signal, using the flavodoxin free radical from Peptostreptococcus elsdenni as a standard (9). EPR samples were rapidly frozen in a 1:5 mixture of methylcyclohexane and isopentane at 81 K. The sample temperature was controlled by a JEOL liquid nitrogen flow system and the temperature was monitored by a thermocouple (Cromel-gold at 0.07% iron). Flavin concentration of the succinate dehydrogenase preparation was determined fluorometrically according to the method of Wilson and King (10). Protein concentration was determined as described previously (2). Redox titration data were simulated using the University of Pennsylvania Medical School DEC-system 10 Computer Facility. Fig. 1 presents a potentiometric titration of the g = 2.005 signal of flavin free radical in the soluble succinate dehydrogenase preparation (BS-SDH) at pH 7.0. Peak to peak amplitude of the free radical signal was plotted as a function of Eh (redox potential referred to the potential of a standard hydrogen electrode) of the enzyme solution. The bell-shaped titra- Duroquinone, 2-OH-1,4naphthoquinone, 1,4-naphthoquinone, indigodisulfonate, indigotetrasulfonate, phenosafranine, and safranine T. EPR conditions: microwave power, 1 milliwatt; modulation amplitude, 6.3 G; microwave frequency, 9.1 GHz; time constant, 0.25 s; sample temperature, 173 K. -, experimental; ---, theoretical curve. The maximum spin concentration of flavin free radical was obtained from the spin quantitation of g = 2.00 signal as described under "Experimental Procedures." tion curve gives maximum signal intensity at an Eh of -79 mV at pH 7.0 which corresponds to the E, value of the overall 2-electron transfer reaction (F1 t* FlH2). We designate midpoint potentials of the fist and second electron transfer processes as EmI and Em2, respectively (F1 c* FlH./Fl; c* FlH,/FlH-). Spin quantitation of the free radical signal showed that 7.3% of the total flavin (measured fluorometrically as described in Ref. 8) is found maximally in the intermediate redox state in the titration at pH 7.0. At the E,, giving rise to the peak of the titration curve, the concentrations of flavin in the oxidized and fully reduced states are equal; thus, the free radical formation constant ( K ) is readily calculated as 2.5 X The E,, and E,, values are calculated from these two parameters, namely K and E,, based on the following two equations: (i) E,, -Em2 = 60 log K and (ii) E,,,] + Em2 = 2 E,,,.

RESULTS
The Eml and Em2 values at pH 7.0 obtained are -127 mV and -31 mV, respectively. The titration gives an experimental curve (solid line) which fits relatively well to a theoretical curve (broken line) except that the experimental points at both ends of the titration are more broadly distributed.
In order to analyze the protonation state of reduced forms of flavin in the physiological pH range, the pK values of the (FlHZ c* F1H-). As seen in Fig. 2, the ~K R value can be obtained from the pH dependence of the midpoint potential (E,) for the overall 2-electron transfer, namely from the oxidized to the fully reduced form of the flavin. These E , values are obtained empirically from the E* value of the peak position of the titration curves a t different pH. The line in the figure is a theoretical curve with ~K R 7.7. The straight lines with a slope of -60 mV and -30 mV/pH, respectively, which fit the curve in the lower and higher pH ranges, intersect at the ~K H .
The pK.9 does not affect the pH dependence of E, values. Fig. 3 gives the maximum per cent free radical concentration of the potentiometric titrations as a function of pH. Below pH 7 the maximum free radical concentration remains unchanged and raising the pH above pH 7 leads to increasing maximum free radical concentration as seen in gradually diminished; the free radical state is more stabilized. Fitting the theoretical curves gave optimal values of ~K K and pKs as 7.7 f 0.2 and 8.0 f 0.2, respectively. Thus, at pH 7.0, about 90% of the free radical is expected to be in the neutral (protonated) form (FlH.), while at pH 9.0 about 90% is in the anionic (deprotonated) form (Fl:). Fig. 5 presents the flavinfree radical in succinate dehydrogenase (BS-SDH) at pH 6.5 and 9.0 which were obtained by poising the enzyme at redox potentials of -16 mV and -150 mV, respectively. The radical gives a g 2.00 signal with a linewidth of 1.15 mT at both pH 6.5 and 9.0 where the free radicals are mostly in the protonated and deprotonated form, respectively. Both spectra are typical of flavins with wings both on the lower and higher field side (11,12).
Previously, Beinert et al. (13) reported a high level of the free radical state (about 80% at pH 7.8) in reconstitutively inactive succinate dehydrogenase preparations upon reduction with succinate. When reconstitutively active BS-SDH was reduced with succinate at different pH values we obtained much lower maximum free radical concentrations, namely, about 10% below pH 7 and 20% even at pH 9 ( Fig. 6). In order to understand the widely different amounts obtained of the flavin free radicals in the succinate-reduced dehydrogenase preparations, we analyzed possible redox states of succinate dehydrogenase components when the dehydrogenase was poised with a succinate and fumarate couple which delivers 2 electrons a t a time. The high potential iron-sulfur protein type iron-sulfur cluster, Center S-3 becomes extremely labile toward oxidants in the soluble state and is EPR-detectable only in the reconstitutively active form (6,14,15). Scheme 1 gives possible redox states of succinate dehydrogenase components upon 2-electron reduction in two different succinate dehydrogenase systems: System A retains active Center S-3; System B contains only inactive Center S-3. Intermolecular redox equilibration is assumed to be much slower than the intramolecular equilibration. In case A, two different redox states (S-1 reduced, S-3 oxidized, flavin free radical) and (S-1 oxidized, S-3 reduced, flavin free radical) are the states which give rise to free radical signals among a total of six different redox states. In case B, (S-1 reduced and flavin free radical) is the only state which gives rise to the free radical signal among three different redox states obtainable. The maximum free radical concentration a t different pH values can be calculated for these two cases using E,,,, and Em2 values of the flavin as reported above and E,,, values of 0 mV and 60 mV for Center S-1 and S-3, respectively. This is plotted in Fig. 6 with a dashed line for case B and dotted-dashed line for case A. High free radical concentrations reported for the reconstitutively inactive succinate dehydrogenase preparations (13) fits very well with case B. Data on BS-SDH fit rather well to the calculated curve in the pH range below 8, but gives slightly higher radical concentrations above pH 8 (Fig. 6). This is explicable from the partial inactivation of Center S-3 even in the reconstitutively active BS-SDH. It is worth pointing out that if we titrate Center S-1 and flavin in succinate dehydrogenase preparations where no Center S-3 is reactive, the titration with a succinate/fumarate couple gives the same redox titration curve for these two components, an intermediate n value between 1 and 2 below the midpoint potential, and approximately n = 1 curve in the higher potential range of the midpoint.
In order to study spatial relationships between flavin and Center S-1 in succinate dehydrogenase, we compared the power saturation behavior of the g = 2.00 signals of the flavin free radical of succinate dehydrogenase with that of flavodoxin a t two different temperatures. In flavodoxin, no transition metal ion is present in the vicinity of the flavin free radical (7). A saturation parameter, was determined according to the quantitative procedure reported by Blum and Ohnishi (16). P 1 / 2 is the input microwave power level at which the saturation condition is satisfied as discussed in Ref. 16. The Plln value of succinate dehydrogenase flavin is significantly higher than that of flavodoxin both at 193 and at 233 K. It is interesting to see that the spin relaxation of the flavodoxin free radical is insensitive to the temperature difference of 40 K in the above temperature range, whereas the PI/* value of the succinate dehydrogenase flavin free radical signal is doubled by raising the temperature from 193 K to 233 K. This indicates that in the succinate dehydrogenase molecule, spin relaxation of the flavin free radical signal is enhanced by nearby Center S-1 spins. In order to test this S-1 effect, flavin spin relaxation was examined after complete destruction of Center S-1 by lowering the pH of the enzyme to 4 and then returning to neutral pH. As seen in Table I

TABLE I
Saturation parameter (P1,2) ofthe flavin free radical in the presence and absence of Center S-1 spins Flavodoxin free radical was produced by illumination in the presence of EDTA as described under "Experimental Procedures." Succinate dehydrogenase free radicals were obtained potentiometrically in the case of B-SDH, and by succinate addition for B-SDH. To destroy cluster S-1, the pH of the enzyme was lowered to pH 4.0 by the addition of sodium acetate and then brought back to neutral pH. EPR conditions are the same as in Fig. 1  one of the binuclear iron-sulfur clusters, namely Center S-1. Center S-2 is not paramagnetic under the experimental conditions required for samples shown in Table I.

DISCUSSION
The present study shows that succinate dehydrogenase flavin has a free radical formation constant of K = 2.5 X at neutral pH which is a far more stable intermediate redox state than the typical n = 2 redox components, such as a free ubiquinone/ubiquinol couple in a hydrophobic milieu ( K = 10"') (17), or the NAD+/NADH couple ( K = lo-") (18). This indicates that succinate dehydrogenase flavin itself may function as a good converter from an n = 2 to an n = 1 electron transfer process. The flavin accepts electrons from succinate in a 2-electron step: the flavosemiquinone is stable enough to permit the reduction of iron-sulfur clusters of the dehydrogenase in sequential 1-electron steps, although the detailed mechanism of the latter steps is not yet completely known. It is therefore unnecessary to postulate simultaneous reduction of two iron-sulfur clusters by flavin as an n = 2 to n = 1 stepdown mechanism (19).
Recently, Albracht proposed that succinate dehydrogenase contains only one binuclear (Center S-1) and one tetranuclear iron-sulfur cluster (Center S-3) per molecule based mostly on the nondetectability of the Center S-2 EPR signal in a succinate-cytochrome c reductase preparation (20) as well as in succinate dehydrogenase which was isolated from Complex I1 and retains almost full reconstitutive activity in a highly pure form (4,21). We have c o n f i i e d these experimental observat i o n~.~ Even in these two systems, spin relaxation of Center S-1 is dramatically enhanced by fully reducing the enzyme with dithionite. Albracht (20) interpreted this spin relaxation enhancement as being caused by a protein conformational change induced by the reduction of the flavin to the fully reduced state. The enhancement of the S-1 spin relaxation is caused by an n = 1 redox component (Center S-2) with midpoint potential of approximately -400 mV in soluble succinate dehydrogenase preparations (2,6).
Analysis of the titration data presented in Fig. 1 indicates that a midpoint potential (E,) of -81 mV would be obtained for a titration monitoring the fully reduced form of flavin, and -77 mV for a titration of the oxidized form, both with n value close to 2. Neither midpoint potential nor n value for the flavin reduction support the hypothesis proposed by Albracht (20) and the enhancement of S-1 spin relaxation seems to be consistent with the cross-relaxation of S-1 via S-2 spins (2, 5, 6). As we reported previously, spin relaxation of Center S-2 greatly depends on the microenvironment of the active center. We interpret the nondetectability of Center S-2 in these intact systems as being due to either (i) an extremely short relaxation time for S-2 spins or (ii) the spin coupling between S-1 and S-2 is such that we do not see signals for more than one spin per molecule in the fully reduced enzyme of these intact preparations.
The presence of two binuclear and one tetranuclear ironsulfur clusters in the succinate dehydrogenase molecule has also been demonstrated by an independent procedure, namely, iron-sulfur core extrusion and core displacement as reported by Coles et al. (4).
Ackrell et al. (22) conducted a potentiometric titration of the activation process of the enzyme in the presence of excess oxalacetate; oxalacetate binds tightly to succinate dehydrogenase in the oxidized state, inactivating the dehydrogenase (22). In this method an oxidation-reduction component was titrated with midpoint potential Em7.0 value between -60 and -90 mV in an n = 2 process, suggesting the titration of the T. Ohnishi, H. Blum, C. A. Yu, and L. Yu, unpublished data. succinate dehydrogenase flavin. This is in good agreement with the titration of the fully reduced form of flavin.
In this paper we have demonstrated the close proximity between succinate dehydrogenase flavin and Center s-1 from the enhancement of flavin spin relaxation by S-1 spins (Fig. 7 and Table I). We previously reported that the distance between the S-l and S-2 clusters is approximately 10 A (2, 5 ) ; a high potential iron-sulfur protein-type cluster (Center s-3) and a ubiquinone pair (23,24) which is the specific electron acceptor of succinate dehydrogenase are also considered to be located within about 15 A of each other (24,25). As shown in Fig. 7 in this paper, the saturation parameter ( P I / P ) of succinate dehydrogenase flavin is 2.0 milliwatts at 233 K, indicating that there is an iron-sulfur cluster in the near vicinity of the flavin. This is a binuclear rather than tetranuclear cluster, since intact Center S-3 is present in BS-SDH and absent in B-SDH and destruction of S-1 and S-2 causes the flavin relaxation to slow markedly. The very short relaxation time (Pl12 > 100 milliwatts at 233 K) observed with the g = 2.00 signal of ubisemiquinone is consistent with its location close to the tetranuclear cluster, Center S-3 (24)(25)(26). These results further strengthen our earlier topographical assignment of two binuclear clusters S-1 and S-2 in flavin-iron-sulfur subunit and Center S-3 tetranuclear iron-sulfur cluster in the ironsulfur subunit.
Using the previously reported lifetime broadening of the Center S-1 EPR spectrum (27), we can estimate the TI of Center S-1 as about s a t 230 K. The flavin power saturation curve in the presence of reduced Center S-1 has a of about 2 milliwatts, which corresponds to a cross-relaxation time of the order of 1O"j s. Since no broadening or splitting of the iron-sulfur EPR spectrum due to the flavin semiquinone can be observed at low temperature, we can place an upper limit of t 5 G on any coupling between the two. We thus expect a flavin-iron-sulfur distance of at least 12 A.
In the absence of spin-spin interaction, a system containing two S = ?h species in a static magnetic field fi,, will  The energies of the four states are thus L.'L(gl + gdPHC1, %(gl -gdPHo, -Wgl -gdPHcI and -%(gI + gz)pHu If S I has a rapid relaxation rate but S2 relaxation is slow, the aa ++ pa and ap t , pp transitions will be difficult to saturate while aa t , ap and pa t , pp will be easily saturable.
The dipolar interaction between two spins S I and S I can be represented by the Hamiltonian than those connecting aa and pa in the uncoupled system. The transition probability depends on the square of the matrix element connecting two states. Thus, we expect the predominantly SP transition to acquire a new relaxation mechanism with a characteristic time The terms in SI,Szz in the dipolar and exchange Hamiltonians give rise to the first order splittings; clearly, these must be within the linewidth. The terms in S1+S2-and Sl-S2' connect the up and pa states. This contributes to the observed splitting when the magnitude of the off diagonal terms is not much smaller than the separation of the basis states.
Since the g tensors of both the iron-sulfur cluster and flavin semiquinone are not very anisotropic and overlap, rather small dipolar or exchange couplings would be effective in mixing the flavin and iron-sulfur transitions through the SI'Szand SI-Sn' terms. The anisotropy of the iron-sulfur center, while not large in proportion to its magnetic moment, will lead to an angular dependent & = gl -g2. At some orientations, gl = g 2 and the iron-sulfur and flavin semiquinone transitions will be completely mixed. It will be difficult to saturate the "flavin" transition of molecules oriented in Ho so that gl and g2 are nearly equal.
The majority of molecules will not have these special orientations. The onset of saturation will be determined by molecules which have (g,g z [ nearly maximal or have the value of the B' term near zero because of the (1-3 cos%) dipolar angular dependence; the overall saturation behavior will be more characteristic of the average separation and average B' value. A full description would require numerical integration over all orientations a t each power. This would depend on the orientation of the g tensors with respect to each other and the f as well as the magnitudes of r and J. We can get some idea of the coupling strength necessary to account for the observed relief of saturation by considering molecules oriented so that glgz is close to the average.
The average separation of the two middle states ("ap" and where B' is the magnitude of the SI'S?and SI-&' terms in gauss: B' z 3 G is sufficient to account for the observed relaxation through this mechanism. If B' is due to dipolar terms, the flavin-iron-sulfur distance would be -15 8, using a point dipole approximation. The unpaired electron should be highly delocalized; the 15 A tion (Capaldi, R. A., ed) pp. 1-87, Marcel Dekker, Inc., New distance would lie between the center to center and edge to edge distances because of the r -3 weighting imposed by dipolar coupling. B' can also have contributions from exchange coupling, but it is unlikely that substantial coupling would occur a t much larger distances than this, since exchange interactions are believed to fall off roughly an order of magnitude per bond length. The separation is probably thus between 12 and 18 A, but more sophisticated techniques will be necessary to sort out the system thoroughly.
From the comparison of EPR and optical properties of various flavo proteins, Massey and Palmer (9) demonstrated that neutral flavin free radical species, protonated at N(5), generally exhibit an EPR spectrum with a linewidth of 1.9 m T and anionic species a 1.5 mT linewidth. The exceptionally narrow linewidth of succinate dehydrogenase flavin (Fig. 5) was suggested to be due to the anionic form further narrowed due to covalent linkage to the apoprotein polypeptide at the 8a carbon position which eliminates the hyperfine coupling contribution from the 8-CHs group (28). The present study points out that both neutral and anionic flavo semiquinone of succinate dehydrogenase exhibit the same narrow linewidths of 1.2 mT, suggesting that the unpaired spin density distribution in succinate dehydrogenase flavo semiquinone is probably more delocalized than in an isolated succinate dehydrogenase flavin peptide preparation (29, 30) or in other flavin species (31,32). On the other hand, equilibrium titration data do not exclude the possibility that the pH-dependent E , value for the flavin/flavin semiquinone couple is due to protonation of a neighboring amino acid residue of the succinate dehydrogenase protein concurrent with electron transfer to the flavin (see Ref. 33).