The Ubiquinone-binding Site of the Saccharomyces cerevisiae Succinate-Ubiquinone Oxidoreductase Is a Source of Superoxide*

The mitochondrial succinate dehydrogenase (SDH) is a tetrameric iron-sulfur flavoprotein of the Krebs cycle and of the respiratory chain. A number of mutations in human SDH genes are responsible for the development of paragangliomas, cancers of the head and neck region. The mev-1 mutation in the Caenorhabditis elegans gene encoding the homolog of the SDHC subunit results in premature aging and hypersensitivity to oxidative stress. It also increases the production of superoxide radicals by the enzyme. In this work, we used the yeast succinate dehydrogenase to investigate the molecular and catalytic effects of paraganglioma- and mev-1-like mutations. We mutated Pro-190 of the yeast Sdh2p subunit to Gln (P190Q) and recreated the C. elegans mev-1 mutation by converting Ser-94 in the Sdh3p subunit into a glutamate residue (S94E). The P190Q and S94E mutants have reduced succinate-ubiquinone oxidoreductase activities and are hypersensitive to oxygen and paraquat. Although the mutant enzymes have lower turnover numbers for ubiquinol reduction, larger fractions of the remaining activities are diverted toward superoxide production. The P190Q and S94E mutations are located near the proximal ubiquinone-binding site, suggesting that the superoxide radicals may originate from a ubisemiquinone intermediate formed at this site during the catalytic cycle. We suggest that certain mutations in SDH can make it a significant source of superoxide production in mitochondria, which may contribute directly to disease progression. Our data also challenge the dogma that superoxide production by SDH is a flavin-mediated event rather than a quinone-mediated one.

The mitochondrial succinate dehydrogenase (SDH) is a tetrameric iron-sulfur flavoprotein of the Krebs cycle and of the respiratory chain. A number of mutations in human SDH genes are responsible for the development of paragangliomas, cancers of the head and neck region. The mev-1 mutation in the Caenorhabditis elegans gene encoding the homolog of the SDHC subunit results in premature aging and hypersensitivity to oxidative stress. It also increases the production of superoxide radicals by the enzyme. In this work, we used the yeast succinate dehydrogenase to investigate the molecular and catalytic effects of paraganglioma-and mev-1-like mutations. We mutated Pro-190 of the yeast Sdh2p subunit to Gln (P190Q) and recreated the C. elegans mev-1 mutation by converting Ser-94 in the Sdh3p subunit into a glutamate residue (S94E). The P190Q and S94E mutants have reduced succinate-ubiquinone oxidoreductase activities and are hypersensitive to oxygen and paraquat. Although the mutant enzymes have lower turnover numbers for ubiquinol reduction, larger fractions of the remaining activities are diverted toward superoxide production. The P190Q and S94E mutations are located near the proximal ubiquinone-binding site, suggesting that the superoxide radicals may originate from a ubisemiquinone intermediate formed at this site during the catalytic cycle. We suggest that certain mutations in SDH can make it a significant source of superoxide production in mitochondria, which may contribute directly to disease progression. Our data also challenge the dogma that superoxide production by SDH is a flavin-mediated event rather than a quinonemediated one.
Succinate dehydrogenase (SDH) 1 of the Krebs cycle and of the bacterial or the mitochondrial respiratory chain (MRC) is intriguing for its roles in energy generation and in mitochondria-related diseases (1)(2)(3). SDH is an iron-sulfur flavoprotein of the inner membrane that functions as a succinate-ubiquinone oxidoreductase, oxidizing succinate to fumarate and reducing ubiquinone to ubiquinol (4 -6). The yeast enzyme, like its mammalian counterpart, is encoded by four nuclear genes, SDH1-4 (4), and harbors a covalently attached flavin adenine dinucleotide (FAD), three iron-sulfur clusters, a b-type heme (7), and 2 quinone-binding sites termed the proximal and distal sites (Q P and Q D , respectively) (8,9). Electrons flow from succinate to the FAD through the iron-sulfur clusters before entering the membrane subunits where quinone reduction occurs.
The membrane subunits have the most diversity in subunit structure and cofactor content largely determining the specific catalytic properties of each enzyme. The structures of the Escherichia coli SDH and the closely related E. coli and Wollinella succinogenes fumarate reductases (FRD) have been determined (10 -13). All flavoprotein and iron-sulfur subunits are highly conserved and adopt very similar structures. In contrast, the membrane subunits are not. The E. coli SDH crystal structure revealed two membrane subunits with a single heme and a single bound ubiquinone (14); the E. coli FRD also contains a pair of membrane subunits with two menaquinone molecules but no heme (10); the W. succinogenes FRD has a single larger membrane subunit with two hemes, but quinone molecules were absent from the crystal (11).
Mutations in the human SDHA gene can result in Leigh syndrome, an infantile-onset progressive neurodegenerative disease (1,15). Mutations in the SDHB, SDHC, or SDHD genes can result in paraganglioma (benign vascularized tumors in the head and neck) or pheochromocytoma (catechol-secreting tumors commonly occurring in the adrenal medulla) (16,17). The carotid body, the most common tumor site for paraganglioma, senses oxygen levels in the blood (16). The striking phenotypic differences between SDHA and SDHB-D mutations suggest that SDH has additional functions, possibly in oxygen-sensing, besides its role as a succinate-ubiquinone oxidoreductase.
The Caenorhabditis elegans mev-1(kn1) mutation in the SDH3 homolog confers hypersensitivity to hyperoxia or to oxidative stress (18,19). The mutation, which affects the Q P site, leads to the overproduction of superoxide anions, the accumulation of aging markers, and a shortened life span. Iron-sulfur clusters are one-electron carriers, whereas quinones are fully reduced with two electrons. The order of the SDH electron carriers necessitates the formation of (at least transiently) a ubisemiquinone radical, which has been detected in SDH by electron paramagnetic resonance spectroscopy (20,21). The mev-1 mutation may alter the stability or accessibility of the ubisemiquinone intermediate.
In this work, we used the yeast enzyme to model molecular and catalytic effects of introducing paraganglioma-and mev-1like mutations. The human SDHB P198R mutation affects a conserved proline residue located near the [3Fe-4S] cluster and the Q P site (22). We mutated Pro-190 in the yeast SDH2 gene to Arg, Leu, or Gln (P190R, P190L, and P190Q, respectively). We also mutated Ser-94 in the SDH3 gene to Glu (S94E) to mimic the C. elegans mev-1 mutation. The SDH2 P190Q and SDH3 S94E mutants displayed reduced SDH activities, hypersensitivity to oxygen and paraquat, and a marked increase in superoxide production. Our results strongly suggest that the SDH Q P site, with its ubisemiquinone intermediate, can be a significant source of superoxide production.
Isolation of Mitochondria and Enzyme Assays-YPG-grown stationary phase cells were harvested by centrifugation and lysed in a Bead-Beater (BioSpec Products, Inc., Bartlesville, OK) with 0.5-mm acidwashed glass beads using five pulses of 1-min and five of 30-s duration interspersed with 3-min cooling periods. Complete protease inhibitor mixture tablets were added prior to cell lysis (Roche Diagnostics). Mitochondria were isolated by differential centrifugation and sucrose density centrifugation (27). Mitochondria were lysed by 10-fold dilution in sorbitol-free buffer and sonicated in ice water in a bath sonicator for 2 min. Succinate and glycerol phosphate dehydrogenase assays were performed as described in Ref. 8. Superoxide generation was assayed as the superoxide dismutase-sensitive reduction of cytochrome c in the presence of 1 mM potassium cyanide (Imlay 1991). Thirty units of superoxide dismutase/ml (from bovine erythrocytes, catalog number S-2515, Sigma) were added in a parallel reaction. SDH is normally isolated partially inhibited by oxaloacetate (28). To activate the enzyme, mitochondria were pre-incubated in 20 mM succinate for 15 min at 30°C immediately prior to analysis. When measuring the effect of succinate concentration on superoxide formation, activation was achieved with 100 mM potassium phosphate (29).
Miscellaneous Methods-Measurements of FAD and protein contents, yeast and E. coli transformation, and recombinant DNA methods have been described in Ref. 30.

RESULTS
Choice of Mutations-The C. elegans mev-1 mutation has directly implicated nematode SDH as a significant source of superoxide (19). We wished to determine whether the yeast enzyme could also be a source of free radicals. In an alignment of the yeast and nematode Sdh3p sequences, Gly-71 of the C. elegans protein aligns with Ser-94, which we mutated to a glutamate (S94E) to mimic the mev-1 mutation. Residues 92-99 of Sdh3p contain a sequence (LSSLHRIS) that closely resembles a motif associated with quinone-binding sites (LXXXHXXT) (bold letters indicate key amino acids) (31,32). Thus, Ser-94 is likely in the vicinity of a quinone-binding site.
We also mutated SDH2 Pro-190 to Arg, Leu, and Gln. The P198R mutation in the human SDHB gene, which affects this conserved proline, results in pheochromocytoma (22). The analogous residue in the E. coli FrdB subunit (Pro-159) has also been mutated (33). The P159Q phenotype was very unusual; aerobic but not anaerobic activity of the mutant enzyme led to radical-mediated auto-inactivation. The P159Q mutation was proposed to perturb the Q P site allowing the ubisemiquinone intermediate to participate in hydroxyl radical generation (22).
The mev-1 and the FrdB P159Q mutations strongly suggest that free radicals are generated near the Q P site. We have modeled the three-dimensional structure of the yeast SDH based on the high-resolution structures of the E. coli SDH, the E. coli FRD, and the W. succinogenes FRD. 2 From the modeled structure, we can see that Sdh2p Pro-190 is just over 4 Å from the ubiquinone occupying the Q P site, whereas the Sdh3p Ser-94 is less than 4 Å from the Q P site (Fig. 1).
Growth Phenotypes of the Mutants-The SDH2 and SDH3 mutations were transformed into SDH2 and SDH3 knockout strains and tested for their abilities to restore MRC-dependent growth using glycerol, a nonfermentable carbon source (Fig. 2). Transformants containing an empty vector were unable to form colonies (Fig. 2, A and C). The Sdh2p P190R and P190L mutations were unable to complement the SDH2 knockout strain for growth on glycerol, whereas the P190Q mutation supported growth, which was only slightly slower than that seen with the wild-type SDH2 gene ( Fig. 2A). The P190R and P190L mutations were not further studied. This result with the P190R mutation suggests that the human P198R mutation that results in pheochromocytoma is a severe allele that strongly impairs SDH activity. Similarly, we tested the ability of the Sdh3p S94E mutant enzyme to support MRC-dependent growth in an SDH3 knockout background, and we observed near wild-type growth (Fig. 2C).
The Mutant Enzymes Confer Hypersensitivity to Oxidative Stress-The C. elegans mev-1 mutant is hypersensitive to paraquat (methyl viologen) and hyperoxia (18). Paraquat, an herbicide, is believed to induce oxidative stress through the generation of superoxide ions either in the cytoplasm (34) or in the mitochondria (35). The SDH2 P190Q mutation severely im-2 K. Oyedotun and B. D. Lemire, submitted for publication. paired MRC-dependent growth in the presence of 1.0 mM paraquat (Fig. 2B). The SDH3 S94E mutation conferred mild sensitivity to this stress (Fig. 2D).
We tested whether the two mutant SDH enzymes also conferred sensitivity to hyperoxia in yeast. Wild-type SDH2 or SDH3 genes supported growth in 100% oxygen, although somewhat more slowly than in air (Fig. 3). The SDH2 P190Q mutation completely abolished growth in the oxygen atmosphere, whereas the SDH3 S94E mutation conferred mild sensitivity (Fig. 3B). Therefore, both mutations render cells less able to withstand exogenous oxidative stresses. We believe the mutant enzymes are themselves a source of oxidative stress that taxes the defense mechanisms of cells, which are then less able to cope with the additional stress of hyperoxia or paraquat. We cannot, however, rule out the possibility that the mutations indirectly impair the resistance mechanisms of cells.
The Mutant Enzymes Have Reduced Quinone Reductase Activities-We purified mitochondria from the SDH2 P190Q and the SDH3 S94E mutant strains and their corresponding wildtype counterparts. We measured the succinate-dependent phenazine methosulfate-mediated reduction of dichlorophenolindophenol; this activity reflects enzyme assembly because it is dependent on the Sdh1p and Sdh2p subunits only and does not require catalytically functional membrane subunits (9,30). We also determined the levels of covalent FAD in the mitochondria; SDH is the major covalent flavoprotein in Saccharomyces cerevisiae (36). The succinate-dependent reduction of decylubiquinone or of cytochrome c requires a fully functional SDH capable of transferring electrons to quinone acceptors. As shown in Table I, both the SDH2 P190Q and the SDH3 S94E enzymes are efficiently assembled; the FAD contents are greater than those of wild-type mitochondria, and the phenazine methosulfate-mediated reduction of dichlorophenolindophenol activities are diminished, although not severely. In contrast, the cytochrome c and the decylubiquinone reductase activities are significantly reduced indicating that both mutations impair quinone reductase activities. A specific loss of quinone reductase activity is observed when the quinone-binding sites are perturbed (8,9,37).
FIG. 3. MRC-dependent growth of wild-type and mutant strains under normoxic or hyperoxic conditions. SDH2 or SDH3 knockout strains transformed with empty vectors (vec) or with wildtype (WT) or mutant SDH2 or SDH3 genes were grown overnight at 30°C in YPG. Cultures were serially diluted with sterile water, and 10-l aliquots were spotted onto SG plates. Plates were incubated at 30°C in an air atmosphere and photographed after 6 days (A) or in a tank continuously flushed with 100% oxygen and photographed after 10 days (B). The SDH2 knockout strain transformed with the wild-type SDH2 gene is not shown, but its growth is not distinguishable from the SDH3 knockout transformed with the wild-type SDH3 gene.  A and B) or SDH3 (C and D) knockout strains transformed with empty vectors (vec) or with wild-type (WT) or mutant SDH2 or SDH3 genes were grown overnight at 30°C in YPG. Cultures were serially diluted with sterile water, and 10-l aliquots were spotted onto SG plates without (A and C) or with (B and D) 1.0 mM paraquat. Plates were incubated at 30°C in an air atmosphere and photographed after 6 days.

The Mutant Enzymes Are a Source of Superoxide Radicals-
Superoxide production was assayed (as the superoxide dismutase-sensitive cytochrome c reductase activity) in the presence of cyanide to inhibit cytochrome c oxidase (38). Two pathways of electron transfer can be used to reduce cytochrome c; in the first, electrons flow through SDH to ubiquinone through the ubiquinol-cytochrome c oxidoreductase (complex III) and to cytochrome c; the second pathway involves a superoxide anion, which can directly reduce cytochrome c. Superoxide dismutase will inhibit the second pathway by converting superoxide into hydrogen peroxide. As seen in Table II, superoxide dismutase inhibits the rate of cytochrome c reduction demonstrating that a fraction of the cytochrome c is directly reduced by superoxide radicals. The addition of excess superoxide dismutase did not further decrease activity (not shown). For the wild-type mitochondria, the superoxide dismutase-sensitive changes in the cytochrome c reductase activity are small (Ͻ5%) and not statistically significant. In contrast, the changes in cytochrome c reductase activities for the SDH2 P190Q and the SDH3 S94E mitochondria are larger and are highly statistically significant. The mutant enzymes, although less efficient catalysts, divert a larger fraction of their electron flow to the production of superoxide.
It has been suggested that flavin auto-oxidation is the source of oxygen-free radicals for the E. coli FRD and SDH, because excess substrate is able to suppress superoxide formation possibly by obstructing access of oxygen to the reduced flavin (39). For the E. coli SDH, an ϳ10-fold decrease in superoxide generation occurs between 0.05 and 20 mM succinate; superoxide production is almost completely suppressed by 2-3 mM succinate (39). We determined the substrate dependence of cytochrome c reductase activity and of superoxide formation of the wild-type and mutant yeast enzymes (Fig. 4). As expected, the turnover numbers for total succinate-cytochrome c reductase activities increased with substrate concentration (Fig. 4, A and  B). The substrate affinities of the mutant enzymes have not been significantly altered; the apparent k m values derived from the data in Fig. 4 for the SDH2, SDH2(P190Q), SDH3, and SDH3(S94E) enzymes are 0.40, 0.26, 0.44, and 0.57 mM, respectively. The similar affinities of the mutant and wild-type en-zymes are not unexpected because the SDH active site is quite distant from the Q P site where the mutations are located. The superoxide dismutase-sensitive activity also increases with the concentration of succinate for the wild-type enzyme (Fig. 4C); the concentration dependence of the mutant enzymes closely paralleled the profile of the wild-type enzyme (not shown). In conflict with the flavin auto-oxidation model, the rates of superoxide production for the mutant and wild-type enzymes are highest at saturating succinate concentrations. At a succinate concentration of 20 mM, the enzyme-active site should be saturated, and flavin auto-oxidation should have been abolished as it was for the E. coli SDH. The turnover numbers for superoxide production are associated with large standard deviations because they are calculated as the difference of two large numbers (total cytochrome c reductase activity minus the superoxide dismutase-insensitive component of that activity). Our data do not indicate any inhibition of superoxide formation at high succinate concentrations and are not consistent with the flavin auto-oxidation model. The dependence of superoxide generation on succinate concentration is strong evidence that SDH, rather than some other MRC component, is the site of superoxide production.
SDH and complex III have close structural and genetic interactions in yeast (40,41), and complex III is a known source of oxygen-free radicals. We measured the glycerol phosphatedependent reduction of cytochrome c in the presence and absence of superoxide dismutase to determine whether the SDH mutations affected complex III function. Glycerol phosphate is oxidized by the glycerol-phosphate dehydrogenase, an FADlinked enzyme that, like SDH, reduces ubiquinone to ubiquinol (42). From ubiquinol, the pathway the electrons follow to cytochrome c is identical to that for the succinate-dependent reaction. If complex III function had been altered by the SDH mutations, then we might expect a large superoxide dismutasesensitive component associated with glycerol-phosphate cytochrome c reductase activity. As shown in Table III, this is not the case; superoxide dismutase has no significant effect on this activity in wild-type or in mutant mitochondria. This supports our conclusion that SDH is directly responsible for the production of superoxide.

DISCUSSION
Mitochondrial SDH is not normally considered a significant source of oxygen-free radicals (43)(44)(45). Yet, there are theoretical considerations and experimental observations that suggest otherwise. A ubisemiquinone intermediate is an obligatory part of the SDH catalytic mechanism; single electrons are transferred from the [3Fe-4S] center in the Sdh2p subunit to ubiquinone bound at the Q P site. Until another electron arrives, the enzyme must stabilize the ubisemiquinone radical and prevent its escape or inappropriate re-oxidation. Mutations in the vi- cinity of the Q P site may compromise the ability of the enzyme to do so. The mev-1 mutation in the C. elegans SDH, upon which we modeled our SDH3 S94E mutation, is a Q P site mutation that results in the generation of superoxide (18,19). Similarly, the SDH2 P190Q mutation is modeled on the E. coli FrdB P159Q mutation, which was proposed to produce hydroxyl radicals via a Fenton-type reaction (33). The Fenton reaction forms hydroxyl radicals from hydrogen peroxide (generated by the superoxide dismutase reaction) when cuprous or ferrous ions are present (44). Mitochondria contain both iron and copper, although both are normally complexed with proteins to prevent their participation in Fenton chemistry. The FrdB P159Q mutation is located close to the [3Fe-4S] cluster and may allow access of superoxide to iron atoms of the cluster leading to hydroxyl radical formation and subsequent enzyme inactivation under aerobic conditions (33). We have not observed a similar inactivation of the yeast Sdh3p S94E enzyme possibly reflecting a different structure near the yeast [3Fe-4S] cluster that prevents hydroxyl radical formation.
Complex I is a known source of oxygen-free radicals as it also generates semiquinones during its reaction cycle (44,45), and reverse electron flow from SDH can provide the electrons for this to occur (46,47). Three points argue against this possibility accounting for our results. First, S. cerevisiae does not have a multisubunit complex I-like enzyme with numerous iron-sulfur centers (42,48). Rather, it has a single subunit NADH-ubiquinone oxidoreductase, Ndi1p (48). Purified Ndi1p apparently catalyzes a two-electron reaction, because semiquinone signals were not detected by electron paramagnetic resonance spectroscopy during a redox titration of the enzyme (49). Second, the mutant SDH enzymes are catalytically compromised for quinone reduction (Table I) and, thus, should also be inefficient in reverse electron flow. Finally, the production of superoxide radicals is proportional to the concentration of succinate (Fig.  4C) in the same way that the total succinate-cytochrome c reductase activity is proportional (Fig. 4, A and B). This would not be expected if the site of superoxide production was outside SDH itself.
The SDH mutations we introduced are localized near the Q P site and may increase superoxide production in one of two ways. They may allow an increased flow of electrons out of the enzyme to oxygen, or they may allow the premature release of the ubisemiquinone intermediate and its subsequent reaction to form superoxide. We cannot distinguish between these mechanisms, but we favor the latter possibility, because conserved residues in the E. coli SDH located near Sdh3p Ser-94 are required for efficient quinone binding. When the E. coli SdhC Ser-27 and Arg-31 are mutated, succinate-ubiquinone reductase activity is lost, the enzymes can no longer be labeled by a benzoquinone photoaffinity probe, and the apparent affinities (K m ) for ubiquinone-2 are too low to be determined (50). The E. coli Ser-27 and Arg-31 correspond to the S. cerevisiae Sdh3p Ser-93 and Arg-97, respectively.
The E. coli FRD and SDH enzymes are believed to produce superoxide through the reaction of oxygen with the reduced FAD cofactor (39) rather than through the loss of electrons from quinones or from iron-sulfur centers. For these enzymes, high concentrations of succinate can fully suppress superoxide formation; the substrate binds near the FAD and is believed to obstruct the access of oxygen to the reduced flavin (39). Our data do not support this interpretation for the yeast SDH. At a concentration of succinate of 20 mM, the wild-type yeast SDH demonstrates a more than 2-fold increase in the rate of superoxide formation over that at 0.1 mM (Fig. 4C); in contrast, superoxide production by the E. coli SDH is reduced by ϳ10fold at 20 mM succinate compared with its peak at 0.1 mM (39). Furthermore, when assayed under identical conditions, the fractions of superoxide dismutase-sensitive succinate-cytochrome c reductase activities of the mutant SDH enzymes are larger than their respective wild-types at all concentrations of succinate (not shown). The simplest explanation for our results is that the Q P site is a source of superoxide anions in the wild-type enzyme. The SDH2 P190Q and the SDH3 S94E mutations further increase superoxide production at the Q P site by perturbing its architecture.
In conclusion, our results demonstrate that mutated forms of a mitochondrial SDH can be significant sources of oxygen-free radicals. The mutations we investigated are located near the Q P site, strongly suggesting that this is the site of radical formation. Additional studies will be required to determine whether paraganglioma or other types of mutations can similarly increase the formation of superoxide radicals.