Inhibitor Probes of the Quinone Binding Sites of Mammalian Complex II and Escherichia coli Fumarate Reductase*

The structural and catalytic properties of beef heart succinate dehydrogenase (succinate-ubiquinone oxi-doreductase, complex II) and Escherichia coli fumarate reductase are remarkably similar. One exception is that whereas electron exchange between the mammalian en- zyme and its quinone pool is inhibited by thenoyltrifluoroacetone and carboxanilides, the enzyme from E. coli is not sensitive to these inhibitors. The lack of good inhibitors has seriously hampered the elucidation of the mechanism of quinone oxidation/reduction in the E. coli enzyme. We have previously reported (Tan, A. K., Ram- say, R. R., Singer, T. P., and Miyoshi, H. (1993) J. Biol. Chem. 268, 19328–19333) that 2-alkyl-4,6-dinitrophenols inhibit mammalian complexes I, II, and III, but with different potencies and kinetic characteristics. Based on these studies we have selected a series of 2-alkyl-4,6-dinitrophenols which proved to be very effective non- competitive inhibitors of mammalian complex II, particularly when acting in the direction of quinone reduction, the physiological event. These compounds turned out to be even more potent inhibitors of E. coli fumarate reductase, particularly when acting in the direction of quinol

Beef heart succinate dehydrogenase (succinate-ubiquinone oxidoreductase; complex II) and Escherichia coli fumarate reductase (menaquinol-fumarate oxidoreductase) exhibit remarkably similar catalytic and structural properties. Each complex consists of four subunits, including a flavoprotein sub-unit containing an 8a-histidyl-FAD moiety, and an iron-sulfur subunit with three distinct Fe-S clusters that transport electrons between the FAD and bound quinones, and two transmembrane polypeptides which also contain the quinone binding sites (1). The anchor subunits show very little sequence homology among species, which probably accounts for the fact that thenoyltrifluoroacetone and carboxanilides, two types of highly potent inhibitors of electron transfer to quinone in the mammalian succinate dehydrogenase complex (2)(3)(4), do not inhibit E. coli fumarate reductase (1). The lack of good inhibitors for the E. coli fumarate reductase has hampered attempts to discern a common mechanism of quinone interaction and the identification of binding sites utilizing site-directed mutagenesis.
Saitoh et al. (5) reported that a series of 2-alkyl-4,6-dinitrophenols inhibit both photosystem II and the mammalian bc 1 complex by interaction at the quinone sites. Further examination showed that the potency of these inhibitors with mammalian complexes I (NADH-ubiquinone oxidoreductase) and III (ubiquinol-cytochrome c oxidoreductase) was comparable, but that complex II (succinate dehydrogenase) was much less sensitive (6). These observations suggested that tailored 2-substituted 4,6-dinitrophenols may provide the means to probe selectively the structural and functional differences displayed at the quinone binding sites on respiratory chain dehydrogenases.
This work reports studies where derivatives of 4,6-dinitrophenol have now been identified that bring about complete inhibition of succinate dehydrogenase. Importantly, these analogues proved to be even more effective inhibitors of E. coli fumarate reductase. This report also compares the structurepotency relationships of several 2-alkyl-4,6-dinitrophenols as inhibitors of quinone reduction and quinol oxidation by beef heart complex II and by wild-type and mutant forms of E. coli fumarate reductase impaired by single amino acid substitutions at the putative Q A 1 and Q B binding sites (7).

EXPERIMENTAL PROCEDURES
Materials-Beef heart electron transport particles (ETP, an inverted submitochondrial particle preparation) and complex II were isolated as in previous work (6). Membrane preparations from wild-type and mutant strains of E. coli were isolated as described elsewhere (7,8). Construction and nomenclature of the mutant strains were described previously by Westenberg et al. (7).
Assays-Succinate oxidation was determined at 38°C with 2,3-dimethoxy-5-methyl-6-decyl-1,4-benzo-quinone utilizing 2,6-dichlorophenolindophenol as a terminal electron acceptor (E mM ϭ 19.1), as in previous work (6). Samples of enzyme were preactivated in assay buffer (50 mM Tris, 0.1 mM EDTA, pH 7.6) containing 20 mM succinate and 1 mM KCN (38°C for 7 min). Fumarate reduction was measured under strictly anaerobic conditions with 2,3-dimethyl-1,4-naphthoquinol as electron * This work was supported by the Department of Veterans Affairs, a grant from the Senate of the University of California San Francisco, National Institutes of Health Grant HL16251, and National Science Foundation Grant MCB 9104297. 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.
¶ To whom correspondence should be addressed: Center for Biomolecular Sciences, University of St. Andrews, Irvine Bldg., North St., St. Andrews, KY16 9AL, Scotland.
donor as described previously (7). Kinetic patterns were identified from computer-fitted plots of the data using Enzfitter software (Biosoft, Cambridge, UK).  Table I were tested for their effects on quinone reduction and quinol oxidation by the enzyme. With all five inhibitors, complete inhibition was obtained in both assay of quinol oxidation and quinone reduction. In contrast to the incomplete inhibition noted in our earlier study (6), with the more potent inhibitors and the extended range of concentrations used in the present work, complete inhibition of membrane-bound succinate dehydrogenase was attained. In confirmation of our previous studies (6), the inhibition of quinone reduction by the 4,6-dinitrophenol derivatives was noncompetitive. Dixon plots of the inhibition of quinone reduction were curved as shown in Fig. 1. One explanation for this behavior is inhibitor binding at two sites. Accordingly, the data were analyzed using a simplified version of the rate equation for noncompetitive inhibition at two sites:

Inhibition of Mammalian
where I denotes the inhibitor concentration, K 1 and K 2 are the dissociation constants for a 4,6-dinitrophenol derivative bound at site 1 or 2, respectively, and ␤ is a constant, proportional to the fraction of the electron flux inhibited at site 1. Table II presents the K i values for inhibitor binding at sites 1 and 2 (K i 1 and K i 2 ) and the ␤ values calculated for the five inhibitors listed in Table I. For quinone reduction, the inhibition at site 1 is nearly 2 orders of magnitude more effective than that at site 2. The values (0.3-0.4) calculated for ␤, the fractional activity inhibited at site 2, are similar for all five inhibitors. The satisfactory agreement supports the 2-site model. Thus, the high affinity site (low K i(app) ) would account for about 65% of the observed inhibition. The remaining 35% inhibition occurs via site 2, which requires a much higher concentration of inhibitor for saturation.
Table II also presents the K i values obtained for inhibition of quinol oxidation, the reverse reaction for succinate dehydrogenase. In this case, the inhibition was also noncompetitive but the Dixon plots were linear, i.e. no evidence for two inhibition sites with different K i values was found.
Inhibition of E. coli Fumarate Reductase by 2-Substituted 4,6-Dinitrophenols-As found with mammalian succinate de-hydrogenase, the fumarate reductase of E. coli was inhibited noncompetitively by all five of the 4,6-dinitrophenols tested (data not shown), and complete inhibition could be achieved in both directions of catalysis ( Fig. 2). Fig. 3A shows that Dixon plots for the inhibition of quinol oxidation are linear, whereas Fig. 3B shows that reduction of quinone in the reverse reaction is characterized by a biphasic Dixon plot, indicative of dual binding sites, as found for the mammalian succinate dehydrogenase. The inhibition patterns for both enzymes are thus shown to be qualitatively similar in that linear patterns are found for quinol oxidation but biphasic curves are observed for quinone reduction. Table III summarizes the K i values for the inhibition of E. coli quinol oxidation by the 5 compounds listed in Table I. These inhibitors are clearly much more effective against E. coli fumarate reductase than the mammalian counterpart by at least an order of magnitude. In general, the potency exhibited by the 4,6-dinitrophenol derivatives in inhibiting quinol oxidation (the physiological direction of catalysis in the cell) was considerably greater than that observed for quinone reduction. The same principle thus applies as with succinate dehydrogenase, namely, that the greater sensitivity to inhibition is shown in the direction of catalysis providing the highest electron flux. In the case of E. coli fumarate reductase this is quinol oxidation, but quinone reduction for beef heart complex II. a The numbering of the compounds follows the system used by Saitoh et al. (5).
b The hydrophobicity index (⌸ value) for the substituents of compounds 17, 19, and 20 was estimated from the partition coefficient (n-octanol/water system) according to Hansch and Leo (9). The ⌸ value for the substituent of compound 17 was calculated by adding the p(CH 3 ) (0.54) to that of compound 15. The ⌸ value for 19 was estimated by adding ⌸(pClC 6    Inhibition of Fumarate Reductase Mutants-Westenberg et al. (7) concluded from their studies of point mutations in the anchor polypeptides (C and D subunits) of E. coli fumarate reductase that two quinone binding sites (Q A and Q B ) were present and functioned similarly to those in photoreaction centers (10,11). In this mechanism, the quinone acting as primary electron acceptor at Q A is nonexchangeable and functions between the oxidized and semiquinone states in passing single electrons to Q B . Once fully reduced and protonated, the qui-none at Q B exchanges with an oxidized quinone of the coenzyme Q pool. Quinol oxidation is assumed to proceed by a reversal of these steps (7). We selected four different mutants for the present study, two with an amino acid substitution in the putative Q A site and two with a residue change in Q B . All four mutations had been shown to partially inhibit catalysis with quinones. That one of the amino acids substituted might also be part of the inhibitor binding site was tested by screening each mutation for its effect on inhibitor potency (see Table IV).
Table IV presents the apparent K i values obtained for the 4,6-dinitrophenol derivatives acting on quinone reduction by each of the four mutants. Interestingly, and in contrast to what was observed with the wild-type enzyme, the inhibition of mutant enzymes was incomplete (approximately 80%). The data were analyzed in terms of one inhibition site and a residual, noninhibitable activity (r) using the equation, Excellent fit of the data to this equation strongly indicates the operation of a single inhibition site giving the respective K i values listed in Table IV. Note that the value for r with all inhibitors possessing an alkyl side chain was 0.2. It is not presently known whether any significance may be attached to the fact that this value of r approximates that for ␤ (0.3), the fraction of electron flux inhibited at the weaker inhibitor site (K i 2 ) in the wild-type enzyme (Table III). Note also that this noninhibitable pathway of electron flux arising in the mutants is detectable only in the direction of quinone reduction, not quinol oxidation (see below).
The inhibition of quinol oxidation by the mutants was noncompetitive in all cases as illustrated for the inhibition of fumarate reductase of mutant E29D by compound 19 in Fig. 4. The interpretation of these data in terms of the type of amino acid replacement and the putative functioning of sites Q A and Q B will be presented under "Discussion." DISCUSSION Continuing our previous studies to develop selective and effective probes of the quinone chemistry in the respiratory chain, we investigated four 2-alkyl-4,6-dinitrophenol derivatives as inhibitors of quinone reduction and oxidation in beef heart succinate dehydrogenase and E. coli fumarate reductase. The 2-alkyl substituents vary in shape, size and hydrophobicity and confer a slightly higher pK a (4.51) than that of the unsubstituted 4,6-dinitrophenol (4.09). Compounds 17 and 20 appeared to be potent inhibitors of the mammalian and E. coli enzymes. The common structural property of the two compounds is the existence of branching structure at ␣-position of the 2-substituent from the benzene ring. Molecular orbital calculations indicated that ␣-branching structure makes the alkyl chain almost perpendicular to the plane of the benzene ring, resulting in a configuration similar to isoprenoid side  chain of ubiquinone (or ubiquinol) (5). This kind of sterochemical similarity between the inhibitor molecules and ubiquinone may be the reason for the potent inhibition. Although compound 19 also possesses ␣-branching structure, conformation of the 2-substituent of this compound somewhat differs from that of compound 17 (or 20) because of significant steric hindrance due to the existence of bulky chlorophenyl group close to the benzene ring.
In accord with the previous conclusion that hydrophobicity was important for the inhibition of quinone reduction by complex II, the four 2-alkyl derivatives used here proved effective blocking agents of electron transport to and from quinones by the mammalian enzyme. The compounds were even more effective inhibitors of E. coli fumarate reductase, particularly when catalysis is in the direction of quinol oxidation (Table III). The discovery of such a potent class of inhibitors represents a significant step for the investigation of quinone interactions in these enzymes, especially those occurring in E. coli fumarate reductase for which very few alternatives are available.
Further, these compounds completely inhibit beef heart complex II and wild-type E. coli fumarate reductase irrespective of whether quinone reduction or quinol oxidation is monitored. However, these reactions are not equally sensitive. With beef heart complex II, quinone reduction is much more sensitive than quinol oxidation, while the opposite is true for the reac-tions catalyzed by E. coli fumarate reductase. The direction of electron flow in the enzyme, i.e. the rate-limiting step and redox status of the enzyme, are important considerations when using these inhibitors.
The present kinetic analysis of the inhibition of mammalian succinate dehydrogenase by the 4,6-dinitrophenol derivatives listed in Table I confirmed the previous finding (6) that the inhibition is noncompetitive. It further revealed that Dixon plots for these compounds as inhibitors of quinone reduction all show a downward curvature at high inhibitor concentrations ( Figs. 1 and 3B), whereas the corresponding plots for inhibition of quinol oxidation are linear (data not shown). E. coli fumarate reductase behaves in an identical manner to beef heart complex II, yielding linear Dixon plots for quinol oxidation (Fig. 3A) but curved ones for quinone reduction (Fig. 3B). Thus for quinol oxidation by both enzymes, inhibition by the 2-alkyl-4,6-dinitrophenol derivatives is complete and consistent with one inhibition site. For quinone reduction, again in both enzymes, the curvature of the Dixon plots suggests two nonequivalent inhibitory interactions.
In the absence of detailed structural information, these observations are rationalized in terms of the mechanism of coenzyme Q reduction occurring in photoreaction centers (10, 11). As described above, the model visualizes two binding sites for quinones: the Q A site containing a quinone that is the primary acceptor of electrons from the Fe-S clusters and passes electrons singly to a secondary quinone at the Q B site. Once fully reduced, the secondary quinone exchanges with the coenzyme Q pool. To account for the biphasic inhibition curves observed for quinone reduction, it is proposed either that the two inhibitor sites (K i 1 and K i 2 ) are located in the span between the Fe-S cluster(s) donating electrons and the Q B site OR that blocking of the normal pathway between Q A and Q B produces leakage of electrons to Q B via an alternate route, which can only be prevented by high concentrations of inhibitor (low affinity site). Alternatively, the appearance of this second site may depend on the redox status of the enzyme. Lack of radiolabeled inhibitor has prevented verification of this. In contrast, inhibition does not open up alternate routes during quinol oxidation, electron flow is by the main path only and is completely blocked by a single inhibitory site.
Since the inhibition is noncompetitive with respect to the externally added quinone acceptor, a possible explanation is that the inhibitor sterically blocks electron exchange between Q A and Q B . This environment is likely to be hydrophobic which would explain why the inhibition by different 2-alkyl-4,6-dinitrophenols depends on the nature of the alkyl substituent. Because all the pK a values (4.51) are the same, it seems unlikely that the different potencies of the inhibitors can be solely ascribed to a difference in the degree of ionization unless the pK a changes differently for each inhibitor when bound to the  protein.
It is seen in Tables II and III that inhibitor 1, the parent dinitrophenol without a 2-alkyl substituent, has a K i for mammalian succinate dehydrogenase 5 times lower in the oxidative direction than in the reverse direction (Table II). This might reflect a similar behavior to thenoyltrifluoroacetone and the carboxanilides, the classical inhibitors of beef heart complex II, which have 2 to 3-fold higher affinity for the oxidized than reduced form of the enzyme (12). In E. coli fumarate reductase the reverse may be true: the K i is nearly 15 times lower in the reductive than in the oxidative direction (Table III). The same patterns of inhibition are exhibited by derivatives of dinitrophenol with 2-alkyl substituents, although the potency varies widely. With both mammalian and the E. coli enzymes, the ␣-branching in the side chain of compound 17 leads to increased inhibition, as compared with ␥-branching (compound 15). As mentioned above, ␣-branching forces the 2-substituent to extend almost perpendicular to the ring plane, thus mimicking the favored conformation for the side chain of ubiquinone (5). The presence of a bulky group (compound 19) next to the ␣-branching greatly diminishes this difference, but not if the bulky group is significantly distant from the point of branching (compound 20). In summary, inhibitory potency is enhanced by increasing the hydrophobicity of the derivative but the shape of the substituent must fit the confines of the binding site.
Interestingly, in contrast to completely inhibiting the quinone reductase activity of the wild-type E. coli fumarate reductase, the inhibitors blocked only 80% of the activity in the mutant forms of the enzyme. In addition, the biphasic nature of the inhibition seen with unmodified enzyme no longer is present. It seems that the fraction of noninhibitable activity approximates the fractional activity blocked by the low affinity inhibitor site in the wild-type enzyme. This might possibly indicate that it is this site that is lost or rendered ineffective in the mutants. The mutations did not impair the ability of the inhibitors to block quinol oxidase activity. The inhibition of quinol oxidation remained complete, indicative of one inhibitor site.
It is evident from the K i values presented in Table III that among the derivatives compound 17 is consistently the best inhibitor of quinol oxidation by the wild-type enzyme, as well as by those mutant forms of E. coli fumarate reductase that were tested (data not shown). This inhibitor, which is effective at submicromolar concentrations, has the structure most closely resembling that of ubiquinone.
In the reverse direction, based on the K i values obtained from the complementary series of experiments using quinone reduction as the test assay, which are presented in Table IV, it is clear that substitution of Glu-29 with Asp increased the potency of all of the 2-alkyl-substituted dinitrophenol derivatives some 14 -20-fold. In this substitution, glutamate is replaced by a smaller amino acid but the carboxylate group and, hence, ability to exchange protons, is maintained. Since other studies have indicated a role for Glu-29 in proton exchange reactions at the Q B site, it is quite possible that these 2-alkyl 4,6-dinitrophenols inhibit these reactions, either by preventing directly proton donation/abstraction or by interrupting the "bucket brigade" of amino acids carrying protons to and from Q B .