Probing substrate binding site of the Escherichia coli quinol oxidases using synthetic ubiquinol analogues.

Substrate binding sites of the Escherichia coli bo- and bd-type quinol oxidases were probed with systematically synthesized ubiquinol analogues. The apparent Km values of ubiquinol-2 derivatives to the bo-type enzyme were much lower than that of the corresponding 6-n-decyl derivatives. The isoprenoid structure is less hydrophobic than the saturated n-alkyl group with the same carbon number; therefore, the native isoprenoid side chain appears to play a specific role in quinol binding besides simply increasing hydrophobicity of the molecule. The Vmax values of 2-methoxy-3-ethoxy analogues were greater than that of 2-ethoxy-3-methoxy analogues irrespective of the side chain structure. This result indicates not only that a methoxy group in the 2-position is recognized more strictly than the 3-position by the binding site but also that the side chain structure does not affect binding of the quinol ring moiety. Systematic analysis of the electron-donating activities of the analogues with different substituents in the 5-position revealed that the 5-methyl group is important for the activity. In the parallel studies with the bd-type enzyme, we obtained similar observations except that almost all quinol analogues, but not ubiquinol-1, elicited a remarkable substrate inhibition at higher concentrations. These results indicate that the two structurally unrelated terminal oxidases share common structural properties for the quinol-oxidation site.

In the aerobic respiratory chain of Escherichia coli, bo-type quinol oxidase is predominantly expressed under high oxygen tension while bd-type quinol oxidase is the primary terminal enzyme under microaerobic conditions (1). Both enzymes catalyze the 2-electron oxidation of ubiquinol-8 (Q 8 -H 2 ) 1 at the periplasmic side of the cytoplasmic membrane and the 4-electron reduction of dioxygen at the cytoplasmic side and establish a transmembrane electrochemical gradient of protons via scalar protolytic reactions (2,3). The former enzyme belongs to the heme-copper terminal oxidase superfamily and functions as a redox-coupled proton pump (3)(4)(5). Site-directed mutagenesis studies on the E. coli bo-type oxidase have identified the axial ligands of three redox metal centers in subunit I, low spin heme b, high spin heme o, and Cu B , and have provided a clue for the understanding of energy coupling and proton translocation (6,7). Recent x-ray crystallographic studies on cytochrome c oxidase of Paracoccus denitrificans (8) and bovine (9) confirmed a structure model for the metal centers of the heme-copper terminal oxidases based on molecular biological studies.
In contrast, the substrate oxidation site of bacterial quinol oxidases is poorly characterized. Photoaffinity cross-linking studies on the bo-type enzyme showed that a Q-binding site resides in subunit II (10). Therefore, electrons seem to transfer from subunit II to the metal centers in subunit I as in cytochrome c oxidase. We have carried out systematic screening of the potent Q-H 2 -oxidation site inhibitors among benzoquinones and substituted phenols and have identified the new compounds such as 2,6-dimethyl-1,4-benzoquinone and 2,6-dichloro-4-dicyanovinylphenol (11). Similarly, Meunier et al. (12) identified aurachin C as the potent inhibitor for both the boand bd-type enzymes, aurachin D and a tridecyl derivative of stigmatellin as the specific inhibitors for the bo-and bd-type enzymes, respectively. Subsequently, we demonstrated the presence and functional importance of a novel high affinity Q-binding site (Q H ) in the bo-type enzyme that is distinct from a low affinity Q-H 2 -oxidation site (Q L ) (13). Potentiometric studies showed that the bound Q 8 at the Q H site can be reduced to Q 8 -H 2 through a ubisemiquinone radical (14,15). These results indicate that the Q H site mediates the intramolecular electron transfer from Q-H 2 in the Q L site to low spin heme b (13,14). Determination of the mutation points for the analogue resistance will facilitate the elucidation of the Q-binding site in terminal quinol oxidases (16).
Gennis and co-workers (17)(18)(19)(20)(21) studied the Q-H 2 -oxidation site of the E. coli bd-type enzyme using protein chemical and immunochemical techniques. Photoaffinity cross-linking with an azidoubiquinone derivative identified subunit I as the substrate binding site (17). Anti-subunits I and II antibodies effectively inhibited the oxidation of quinols and N,N,NЈ,NЈ-tetramethyl-p-phenylenediamine, respectively (18). The epitope of monoclonal antibodies which binds to subunit I and specifically blocks the quinol oxidation was mapped to a region having a single 11-residue stretch of the large periplasmic loop V2VI (Q-loop) of subunit I (19). Alternatively, limited proteolysis demonstrated that the cleavage of the Q-loop caused a loss of the quinol oxidase activity (20,21). These results suggest that the Q-H 2 -oxidation site is localized in subunit I of the bd-type enzyme. Structure-function studies with quinone analogues have demonstrated that the E. coli bo-and bd-type quinol oxidases share a similar Q-binding site although they are structurally unrelated (11). Further, the bd-type enzyme was also shown to be able to stabilize ubisemiquinone (22).
The structural requirements for Q-H 2 in the oxidation reaction should be tightly related to the structural features of the Q-H 2 -oxidation site and the reaction mechanism. Thus, a systematic set of ubiquinone analogues was used to probe the Q-mediated electron transfer systems such as mitochondrial succinate-cytochrome c (23,24) and NADH-Q (25,26) oxidoreductases. In contrast to mammalian mitochondrial respiratory enzymes (23)(24)(25)(26)(27)(28), bacterial terminal quinol oxidases were poorly characterized in this respect. Yu and co-workers (10) synthesized the 6-alkyl derivatives of 2,3-dimethoxy-5methyl-1,4-benzoquinone and examined the effects of a length of the n-alkyl side chain on the oxidase activity of the E. coli bo-type enzyme. The substitution pattern of the quinol ring was fixed as for that of native Q 8 -H 2 . Therefore, the effects of substituents at all positions in the quinol ring upon the electron transfer efficiency remain obscure. On the other hand, saturated alkyl chains such as n-pentyl and n-decyl groups have been claimed to mimic the native isoprenoid side chain in mammalian mitochondrial enzymes (26 -28). Further, the electron-accepting and -donating activities of Q 8 -H 2 and DB-H 2 were found to be identical in mitochondrial succinate-cytochrome c oxidoreductase (27). These results indicate that the side chain contributes primarily to an increase of hydrophobicity of the Q molecule to fit into the hydrophobic Q-binding site. To the contrary, the native isoprenoid tail structure was claimed to play a specific role in Q-binding at the Q A and Q B sites of the reaction center of Rhodobactor sphaeroides (29) although it is unclear whether or not the above premise holds in general for the Q-mediated electron transfer systems. Thus, it is worthwhile to examine the role of the side chains of Q-H 2 in the oxidation by bacterial quinol oxidases.
To probe the structural features of the substrate binding site based upon structure-activity studies, we examined the effects on the enzyme activity of replacement of one of the alkyl groups (or moieties) on the quinol ring by another alkyl group. Such a structural modification inevitably alters the molecular shape while minimizing changes in the redox property of the molecule because the electronic nature of the alkyl groups is almost identical (23,24). In this study, we synthesized a wide variety of alkyl analogues of Q-H 2 , including the compounds containing an isoprenoid side chain, and attempted to examine the effects of the side chain structures on recognition of the substituents at all positions in the quinol ring by the Q-H 2 -oxidation site. In addition, the structural properties of the substrate oxidation site of the E. coli bo-and bd-type quinol oxidases were compared in terms of the kinetic and potentiometric properties of ubiquinol analogues.

EXPERIMENTAL PROCEDURES
Materials-Compounds 8 to 11 and 14 to 17 ( Fig. 1) were the same samples as described previously (25). Ubiquinone-1 (Q 1 ) and -2 (Q 2 ) were generous gifts from Eisai Co. Ltd., Tokyo. Other chemicals were commercial products of analytical grade.

Role of the 2-and 3-Methoxy Groups of Quinol Ring in Re-
action with the bo-type Enzyme-Native Q 8 -H 2 has two methoxy groups on both the 2-and 3-positions of the quinol ring. Nonempirical molecular orbital calculation demonstrates that the conformation of these methoxy groups affects electrical potentials of an oxidized form of quinone or semiquinone radical through conformer interconversion (35). Therefore, a binding manner of the methoxy groups to the protein environment is expected to impose a significant influence on the Q redox reaction. To probe the structural features of the Q-binding site for the 2- DB by means of different synthetic routes. The introduction of a bulky monoethoxy or diethoxy group was shown to be useful in discriminating these two positions by the mitochondrial Q-mediated electron transfer system (24,25). Here we applied the synthetic methods to preparations of the monoethoxy and diethoxy analogues that have a native isoprenoid side chain (i.e. compounds 2, 3, and 4) (Fig. 1).
Taking Q 2 -H 2 as a reference compound, we first compared the electron-donating activity of compounds 2, 3, and 4 ( Table  I). The activity of compound 2, in terms of apparent K m and V max values, was almost identical to that of Q 2 -H 2 . In contrast, the V max value of compound 3 decreased to 18% of the reference while K m was doubled. A change in the electron-donating activity of the diethoxy analogue (compound 4) relative to Q 2 -H 2 was the same as that of compound 3. Our synthetic strategy now revealed that a methoxy group at the 2-position is recognized more strictly than the 3-methoxy group by the Q L site.
Role of Isoprenoid Side Chain in Reaction with the bo-type Enzyme-Subsequently, we examined the effects of side chain structures on the oxidase activity using the ethoxy analogues with two isoprenoid units (compounds 1-4) or an n-decyl group (compounds 8 -11) and found that the effects of 2-and 3-substituents on the electron-donating activity were similar irrespective of the side chain structures (Table I). The order of the V max values were 2,3-dimethoxy Ն 2-methoxy-3-ethoxy Ͼ 2-ethoxy-3-methoxy Ն 2,3-diethoxy derivatives. This finding indicates that binding of the quinol ring moiety to the Q L site is not significantly affected by the side chain structure.
It is, however, notable that the apparent K m values of Q 2 -H 2 and its 3-monoethoxy analogues (compound 2) were 10-fold lower than that of corresponding n-decyl analogues (DB-H 2 and compound 9, respectively) while the total number of carbon atoms in these side chains is identical. In general, the hydrophobicity of the hydrocarbon tail differs significantly depending upon its structure. Based on water to hexane solventtransfer free energies calculated from the partition coefficient in the hexane-water system, Warncke et al. (29) estimated the hydrophobicity of Q analogues with various side chains. Values for Q 2 and DB are found to be Ϫ7.34 and Ϫ9.79 kcal/mol, respectively (29). Therefore, the isoprenoid structure is rather less hydrophobic than the saturated n-alkyl group of the same number of carbon atoms (also see Ref. 26). Accordingly, a lower K m value of Q 2 -H 2 (or compound 2) cannot be attributed to its hydrophobicity. The native isoprenoid tail structure (two isoprenoid units in this case) may increase the binding affinity through a specific interaction, as claimed for Q-binding at the Q A and Q B sites of the reaction center of R. sphaeroides (29). Our observation that saturation of the isoprenoid side chain of Q 2 -H 2 resulted in a marked increase in the K m value (Q 2 -H 2 versus compound 7) supports this possibility.
It is still unclear whether the increase in the K m value of saturated alkyl analogues is related to a lack of -electron systems or due to unfavorable conformational energy in the protein-bound state. Warncke et al. (29) examined the native isoprenoid tail structure of the bound quinones at the Q A and Q B sites of the R. sphaeroides reaction center (36) and ruled out the contribution of favorable enthalpic interactions of the -electron systems in the isoprenoid double bonds with the protein. An x-ray crystallographic study of the Q-mediated respiratory enzymes will provide a clue for understanding this effect (37).
To examine the role of isoprenoid tail structure in the electron transfer reaction, we also compared the kinetic parameters between Q 1 -H 2 and PB-H 2 (5-carbon atom analogues). In contrast to 10-carbon atom analogues (Q 2 -H 2 and DB-H 2 ), activities of Q 1 -H 2 and PB-H 2 may be taken to be almost identical (Table I). This finding suggests that the second isoprenoid unit of Q 2 -H 2 plays a specific role in the binding interactions.  as an inhibitor at higher concentrations, whereas Q 1 and PB are comparable electron acceptors. However, electron transfer reactions of mitochondrial succinate-cytochrome c oxidoreductase with Q 2 and DB are identical (27), suggesting that the molecular recognition of the 6-isoprenyl group by the Q-binding site differs in the respiratory redox proteins. To elucidate the roles of the isoprenoid side chain, Q 2 -H 2 analogues in which a double bond of the first or second isoprenoid unit is selectively saturated should be useful. Synthetic strategy that enables the selective saturation of the two double bonds is now being examined.

Role of the 5-Methyl Group of Quinol Ring in Reaction with the bo-type
Enzyme-Effects of a methyl group at the 5-position on the physicochemical properties of the Q molecule are complicated. Substitutions of the 5-methyl group alter not only the shape and redox potential of the molecule but also the conformational property of the vicinal (6-position) alkyl side chain by steric hindrance. Molecular orbital calculation suggests that the Q molecule is stable when the side chain nearby the Q ring extends almost perpendicular to a plane of the Q ring due to a neighboring 5-methyl group (25,38). Therefore, the effects of the 5-methyl group on the electron-donating activity should be considered separately from the steric effect against conformational property of the vicinal side chain.
We found that the V max value of compound 5, which lacks the 5-methyl group, markedly decreases with respect to Q 2 -H 2 (Table I). Similar observations were obtained with a pair of the analogues with different side chain structures (DB-H 2 versus compound 12 and PB-H 2 versus compound 15). However, these results are not enough to conclude whether the change in the activity is due to a lack of the 5-methyl group itself or due to the secondary effect on enhanced flexibility of the alkyl side chain as discussed above.
The conformational properties of the side chain of compounds 16 and 17 lacking the 5-methyl group are similar to that of PB-H 2 because of steric congestion arising from a branched structure at the ␣-position of the side chain (25). Therefore, a comparison of the electron-donating efficiencies between PB-H 2 and compound 16 or 17 is useful for probing a specific role of the 5-methyl group. As shown in Table I, the electron-donating activities of these two compounds are remarkably poorer than that of PB-H 2 , indicating that the 5-methyl group itself is important for the electron donation. In addition, we found that replacement of the 5-methyl group by an ethyl group results in a slight but significant decrease in the activity for Q 2 -H 2 and DB-H 2 derivatives (i.e. Q 2 -H 2 versus compound 6 and DB-H 2 versus compound 13, respectively). This indicates that the 5-alkyl substituents larger than the methyl group are unfavorable to fit into the Q L site due to steric restriction arising from the protein environment. It is notewor-thy that similar changes in the electron-donating activity against the structural modifications at the 5-position were again observed irrespective of the side chain structure. This result also supports the above notion that a binding manner of the quinol ring moiety into the Q L site is not significantly affected by the side chain structure.
Electron-donating Activity of Q-H 2 Analogues in Reaction with the bd-type Enzyme-The above set of Q-H 2 analogues were used to probe the Q-H 2 -oxidation site of the bd-type quinol oxidase. Recent kinetic analysis showed that the reaction with Q 1 -H 2 proceeds by a mixed ping-pong/sequential reaction involving a ternary complex (39). Fig. 2 shows the concentration dependence of the electron-donating activity of four reference compounds (Q 1 -H 2 , Q 2 -H 2 , DB-H 2 , and PB-H 2 ). A remarkable decrease in the efficiency was observed at higher concentrations of Q 2 -H 2 , DB-H 2 , and PB-H 2 but not with Q 1 -H 2 . The apparent K m and V max values for Q 1 -H 2 were 205 M and 32 mol/min/nmol of enzyme, respectively, comparable with those previously reported (2,39). Similar unusual kinetics have been reported for the electron-donating activity of compound 12 with isolated mitochondrial ubiquinol-cytochrome c oxidoreductase (23).
To characterize the unusual kinetics for Q 2 -H 2 , DB-H 2 , and PB-H 2 , in the presence or absence of 200 M Q 1 -H 2 , the concentration dependence of the electron-donating activity of Q 2 -H 2 was further examined as a representative case (Fig. 3). The Q 2 -H 2 oxidase activity in the presence of Q 1 -H 2 increased slightly to a plateau level and then decreased gradually. The presence of Q 1 -H 2 shifted maximal activity to a lower concentration of Q 2 -H 2 . Above ϳ60 M Q 2 -H 2 , the activity was about the same with or without Q 1 -H 2 . These observations indicate that Q 2 -H 2 works as both an electron donor and an inhibitor. Since Q 2 -H 2 (or Q 1 -H 2 ) oxidase activity was not completely reduced in the presence of various concentrations of the oxidized form of Q 2 (data not shown), the substrate inhibition cannot be attributed to competition for the Q-H 2 -oxidation site between the substrate (Q 2 -H 2 ) and the product (Q 2 ) molecules. The precise mechanism of the inhibition remains to be elucidated.
Besides the above reference compounds, almost all Q-H 2 analogues studied here elicited the substrate inhibition while the extent of inhibition varied depending upon the structures. Therefore, their kinetic parameters (K m and V max ) were unable to be deduced from a double reciprocal plot analysis. In order to qualitatively analyze the substituent effects on the activity between the bo-and bd-type terminal oxidases, we focused on examining a concentration-dependent electron-donating activity in low concentration ranges. Fig. 4 shows the concentration-dependent activities of Q 2 -H 2 and its monoethoxy derivatives (compounds 2 and 3). The Q 2 -H 2 and compound 2 elicited similar activities while the activity of compound 3 was significantly lower than the others. Unexpectedly, a diethoxy derivative (compound 4) showed greater activity compared with compound 3 (data not shown). This may be due to weaker substrate inhibition by compound 4 since the residual enzyme activity at its higher concentrations was significantly higher than that of the other three compounds. The structural factors of the Q-H 2 molecule required for electron donation or substrate inhibition appear to be specific for the bd-type quinol oxidase.
To examine the effects of side chain structure on binding properties of the quinol ring moiety, the electron-donating activity was compared among the monoethoxy derivatives with the n-decyl group (compounds 8 -10). The order of the activity of the n-decyl analogues (i.e. 2,3-dimethoxy Ն 2-methoxy-3ethoxy Ͼ 2-ethoxy-3-methoxy) was similar to that of Q 2 -H 2 analogues (data not shown). Thus, the effects of the substituents at the 2-and 3-positions on the electron transfer activity appear to be identical irrespective of the side chain structure. These findings indicate that a methoxy group at the 2-position of the ubiquinol ring is recognized more strictly than that at the 3-position, as observed for the bo-type enzyme. Furthermore, although a quantitative analysis is quite difficult for the bdtype enzyme, Q 2 -H 2 gives an apparent maximal activity at a significantly lower concentration than DB-H 2 (Fig. 2). This tendency is also shared with the bo-and bd-type quinol oxidases.
Next, the electron-donating activities of DB-H 2 and compounds 12 and 13 to the bd-type enzyme were compared to examine the role of the 5-methyl group on the quinol ring (Fig.  5). Compound 12 gave the maximal activity at a lower concentration than DB-H 2 , indicating that the binding affinity of the former is higher than that of the latter. The electron-donating activity of compound 13 was significantly weaker than that of DB-H 2 . Relative changes in the electron-donating activities of DB-H 2 analogues (Fig. 5) and Q 2 -H 2 analogues (Q 2 -H 2 and compounds 5 and 6, data not shown) are dependent on the structure of substituents at the 5-position and are similar to those observed for the bo-type enzyme (Table I). These findings also strongly suggest that the two terminal quinol oxidases share common structural properties for the Q-H 2 -oxidation site although it is difficult to identify any Q/Q-H 2 -binding motif in the primary sequences (4,40,41). The present study is in good agreement with our previous structure-activity studies of a series of specific inhibitors that the two terminal quinol oxidases share the common structural features of the Q-H 2 -oxidation site (11). Furthermore, the recent work reporting the existence of common, very potent inhibitors, like aurachin C analogues (12), of the Q-H 2 -oxidation sites provides further support.
Conclusion-In the present study, we synthesized a systematic set of short-chain Q-H 2 analogues. The native isoprenoid tail structure increases the binding affinity of Q-H 2 molecules for the bo-type quinol oxidase. The effects of substituents at all positions of the quinol ring upon the electron-donating activity are similar between the bo-and bd-type enzymes. In addition, recognition of the quinol ring moiety by the two enzymes is not affected by the side chain structures, indicating that the quinol ring and the side chain moieties contribute independently to the Q-H 2 oxidation reaction. Our results indicate that the E. coli quinol oxidases, though they are structurally unrelated, share common structural properties for the Q-H 2 -oxidation site.