NMR Reveals Double Occupancy of Quinone-type Ligands in the Catalytic Quinone Binding Site of the Na+-translocating NADH:Quinone Oxidoreductase from Vibrio cholerae*

Background: The Na+-NQR is a respiratory Na+ pump found in prokaryotes. Results: The NqrA subunit binds two quinone-type ligands in direct vicinity to each other. Conclusion: Simultaneous binding of two quinones enhances catalytic efficiency of the final electron transfer step. Significance: We provide the first direct experimental evidence of simultaneous quinone binding with relevance for the catalytic mechanism. The sodium ion-translocating NADH:quinone oxidoreductase (Na+-NQR) from the pathogen Vibrio cholerae exploits the free energy liberated during oxidation of NADH with ubiquinone to pump sodium ions across the cytoplasmic membrane. The Na+-NQR consists of four membrane-bound subunits NqrBCDE and the peripheral NqrF and NqrA subunits. NqrA binds ubiquinone-8 as well as quinones with shorter prenyl chains (ubiquinone-1 and ubiquinone-2). Here we show that the quinone derivative 2,5-dibromo-3-methyl-6-isopropyl-p-benzoquinone (DBMIB), a known inhibitor of the bc1 and b6f complexes found in mitochondria and chloroplasts, also inhibits quinone reduction by the Na+-NQR in a mixed inhibition mode. Tryptophan fluorescence quenching and saturation transfer difference NMR experiments in the presence of Na+-NQR inhibitor (DBMIB or 2-n-heptyl-4-hydroxyquinoline N-oxide) indicate that two quinone analog ligands are bound simultaneously by the NqrA subunit with very similar interaction constants as observed with the holoenzyme complex. We conclude that the catalytic site of quinone reduction is located on NqrA. The two ligands bind to an extended binding pocket in direct vicinity to each other as demonstrated by interligand Overhauser effects between ubiquinone-1 and DBMIB or 2-n-heptyl-4-hydroxyquinoline N-oxide, respectively. We propose that a similar spatially close arrangement of the native quinone substrates is also operational in vivo, enhancing the catalytic efficiency during the final electron transfer steps in the Na+-NQR.

year (World Health Organization, 2010). This pathogen, like many other bacteria, harbors a unique respiratory enzyme complex, namely the sodium ion-translocating NADH:quinone oxidoreductase (Na ϩ -NQR), 5 that is unrelated to the eukaryotic complex I on the level of primary structure yet serves a similar purpose in that it generates an electrochemical gradient across the cytoplasmic membrane that in turn drives many other cellular processes like H ϩ /Na ϩ -antiporters, solute uptake, and rotation of the flagellum (1).
Na ϩ -NQR is composed of six subunits NqrA-F and harbors at least five redox-active cofactors: a non-covalently bound FAD and a 2Fe-2S cluster in the NqrF subunit, two covalently bound FMNs in subunits NqrB und NqrC, and one non-covalently bound riboflavin in the subunit NqrB (2)(3)(4)(5). Upon oxidation of NADH, electrons are transferred from NADH via FAD and the 2Fe-2S cluster on NqrF to FMN on NqrC, to FMN on NqrB, and finally to riboflavin on NqrB (6,7). However, the final step of the reaction cycle, the reduction of the quinone substrate, and the coupling of redox chemistry to sodium ion translocation are still largely unclear.
Already in 1992 it was recognized that the resistance of Vibrio alginolyticus toward korormicin, a putative quinone analog, is brought about by two mutations in the NqrB subunit of its Na ϩ -NQR (8). Therefore, it was expected that the NqrB subunit would carry the active site for quinone binding and reduction. Instead, we have recently identified the NqrA subunit to bind ubiquinone-8 and to interact with short chain quinones in the context of the membrane-embedded/detergentsolubilized holo-Na ϩ -NQR enzyme complex as well as with the isolated, soluble NqrA subunit (9). On the other hand, Juárez et al. (10) have shown that the point mutations at glycine 140 and glycine 141 of the NqrB subunit affect Na ϩ -NQR reduction activity, leading them to conclude that NqrB would harbor the binding site for ubiquinone-1 (Q 1 ).
The quinone analog 2,5-dibromo-3-methyl-6-isopropyl-pbenzoquinone (DBMIB; see Fig. 1) is a potent inhibitor of the mitochondrial bc 1 complex and the cytochrome b 6 f complex of chloroplasts (11)(12)(13)(14)(15) but can also serve as an electron acceptor with a favorable potential (E 0 ) of ϩ180 mV (16). The binding of 2 eq of DBMIB into the quinol oxidase (Q o ) pocket of the b 6 f complex has been proposed. One equivalent binds with high affinity to a proximal niche, whereas the binding of a second equivalent with low affinity to a distal niche induces a rotation of the Rieske iron-sulfur protein domain of the complex (17). By EPR, it was shown that DBMIB attaches to and modifies the iron-sulfur center in the bc 1 complex with inhibition in the nanomolar range and interacts with cytochrome b. A mechanism was proposed whereby DBMIB does not just act as a simple competitor or redox mediator at the quinol oxidase site but also as an antagonist to ubiquinone, inducing a redox bypass of the respiratory chain (18).
Another quinone analog, 2-n-heptyl-4-hydroxyquinoline N-oxide (HQNO; Fig. 1), was already reported as an inhibitor of the Na ϩ -NQR from V. alginolyticus with a K I of 0.4 M (19). In our previous work, we also showed by STD NMR and surface plasmon resonance spectroscopy that HQNO binds to the NqrA subunit (9).
Here we show that DBMIB acts both as an inhibitor and as an alternative substrate of the Na ϩ -NQR of V. cholerae by a specific interaction with the NqrA subunit of the complex. Furthermore, NMR experiments with DBMIB and HQNO indicate that the NqrA subunit possesses an extended binding site for quinone analog ligands that can simultaneously accommodate two quinones involving a high affinity and a low affinity binding mode. Similar dual occupancy models have been proposed for other quinone-converting enzymes based on indirect experimental evidence (20 -22). Our findings provide important insight into mechanistic aspects of the final redox step catalyzed by the Na ϩ -NQR.

EXPERIMENTAL PROCEDURES
Purification of the Na ϩ -NQR and Subunit NqrA-Fulllength Na ϩ -NQR complex linked to an N-terminal hexahistidine tag on the subunit NqrA was produced and purified as described previously (23). Subunit NqrA encoded on plasmid pBR322 (9) was produced in Escherichia coli BL21(DE3). Perdeuterated NqrA was produced in labeled M9 medium according to Marley et al. (24). The cells were grown in unlabeled LB medium at 37°C with shaking at 150 rpm. At an A 600 of ϳ0.7, the cells were pelleted by centrifugation for 15 min at 5,000 ϫ g at room temperature. The cells were then washed once with M9 medium in D 2 O and pelleted again. Cells were resuspended in deuterated M9 medium that was supplemented with perdeuterated glucose and incubated for 1 h at 37°C and 150 rpm. Subsequently, protein expression was started by addition of isopropyl thio-␤-D-galactoside to a concentration of 1 mM. After 4 h of incubation, the cells were harvested. To purify His 6 -NqrA, washed cells (25 g) were suspended in 50 mM sodium phosphate, pH 8.0, 300 mM NaCl, 5% (v/v) glycerol. One spatula tip of MgCl 2 , DNase I (Roche Applied Science), and one tablet of protease inhibitor mixture (Roche Applied Science) were added to the cell suspension, which was passed twice through a French pressure cell at 7.58 megapascals. Cell lysate was centrifuged at 100,000 ϫ g for 60 min. The supernatant was filtrated through a syringe filter with a 0.2-m surfactant-free cellulose acetate membrane (Corning) and loaded onto a HisTrap 2-ml (GE Healthcare) column equilibrated with running buffer (50 mM sodium phosphate, pH 8.0, 300 mM NaCl, 5% (v/v) glycerol, 4 mM NaN 3 ) containing 20 mM imidazole. NqrA was eluted with running buffer containing 130 mM imidazole. All experiments were performed with monomeric NqrA, which was separated from NqrA aggregates on a Superdex 200 16/60 (GE Healthcare) column in 50 mM phosphate buffer, pH 8.0, 300 mM NaCl, 5% (v/v) glycerol, 4 mM NaN 3 (9).
Enzymatic Assays-NADH dehydrogenase activities of Na ϩ -NQR were determined with NADH (Applichem) (0.005-0.05 mM) and 0.1 mM Q 1 (MCAT GmbH) at pH 7.5 as described previously (3). Quinone reductase activities were determined with Q 1 (0 -0.1 mM) or Q 2 (0 -0.1 mM; Sigma-Aldrich) as electron acceptor at a fixed NADH concentration of 50 M. Rates of NADH oxidation were monitored at 340 nm, and quinol formation was determined from the difference in absorption at 248 and 268 nm (Tables 1 and 2). Kinetic measurements were performed under anaerobic conditions with magnetic stirring on an HP 8452A diode array spectrophotometer (Agilent) in reaction buffer (20 mM Tris-Cl, 50 mM Na 2 SO 4 , pH 7.5, 0.05 mg/ml BSA (Applichem), 0.05% DDM (Roche Applied Science)) at 25°C. K m and V max were calculated using the Michaelis-Menten equation. For enzyme inhibition studies, the Na ϩ -NQR was incubated at 4°C in reaction buffer containing DBMIB (Sigma-Aldrich) and Q 1 or Q 2 in Ͻ1% ethanol. The reaction was started by adding NADH.
K I for inhibition of NADH dehydrogenase activity by DBMIB was calculated assuming an uncompetitive mode of inhibition (25). The correlation coefficients and kinetic constants of specific quinone reductase activities are listed in Table 2. The rates were background-corrected by the changes in absorbance aris- Two Quinones Bind to the NqrA Subunit of Na ؉ -NQR ing from DBMIB reduction by the Na ϩ -NQR to enable kinetic analysis of ubiquinol formation only. The reductase activity of Na ϩ -NQR using only DBMIB as electron acceptor (16) was assayed under identical conditions as Q 1 reductase activity with DBMIB concentrations between 0 and 100 M using the same extinction coefficient as for Q 1 ( Table 2).
Binding of DBMIB Determined by Tryptophan Fluorescence Spectroscopy-NqrA was diluted in 50 mM phosphate buffer, 300 mM NaCl, pH 8.0, 5% (v/v) glycerol to a concentration of 1 M. DBMIB was added from an ethanol stock solution to final concentrations between 0 and 200 M (final ethanol concentration, 1%). Fluorescence was analyzed with a Fluorolog 3 spectrofluorometer (Horiba Scientific) using a temperaturecontrolled quartz cuvette. Measurements were performed in triplicate at 10°C. The intensity of tryptophan fluorescence emission was determined at 338 nm ( excitation ϭ 295 nm). The increase in quenching (⌬F) of the tryptophan emission was normalized to values between 0 and 1 and plotted against the concentration of DBMIB. Non-linear regression analyses were performed using the equations for one-site binding and two-site binding in Origin 6.1.
NMR Spectroscopy-NqrA was transferred to D 2 O containing 10 mM potassium phosphate, pH 8.0, 150 mM NaCl, 4 mM NaN 3 (PBS buffer) by repeated (at least 5-fold) ultrafiltration with Ultrafree 4 membranes (Millipore; molecular mass cutoff, 10 kDa). Na ϩ -NQR was transferred to the same buffer as used for NqrA with addition of 0.05% DDM; Ultrafree 4 membranes with molecular mass cutoff of 100 kDa were used in this case. To 10 M NqrA or Na ϩ -NQR, respectively, Q 1 was added from 20 mM stock solutions in DMSO-d 6 to a final concentration of 100 M. DBMIB and HQNO were added from 20, 4, or 1 mM stock solutions in DMSO-d 6 to obtain NqrA samples containing 0, 5, 10, 20, 40, 60, 80, 100, 125, or 150 M DBMIB or HQNO, respectively, and additionally 200 M in the case of HQNO. Additional DMSO-d 6 was added to keep its concentration constant within the series (1.5%, v/v). The STD NMR experiment with DBMIB alone was performed with 2.5 M NqrA in PBS buffer. DBMIB was added from a 20 mM stock solution to a concentration of 100 M. Controls were prepared in the same way without NqrA. The samples were transferred to 5-mm NMR tubes, and STD NMR experiments (26) were per-formed as described (9). The resonances of Q 1 were assigned as described (9). Water suppression was achieved by excitation sculpting (27). Resonances of NqrA and Na ϩ -NQR, respectively, were saturated by applying a train of low power Gaussian-shaped pulses at 0.2 ppm with a total saturation time of 2 s. Off-resonance irradiation was set to 33 ppm. Under these conditions, percent STD in control experiments was below 3% for the most upfield methyl group of DBMIB (H-8) and below 1% for all other protons. On-and off-resonance spectra were acquired in an interleaved manner and subtracted after processing. Up to 4096 transients were collected at a spectral width of 12 ppm. Percent STD effects were determined within the multiple display mode by scaling the off-resonance spectrum down to superimpose with the signal of interest in the difference (off Ϫ on) spectrum. The influence of DBMIB or HQNO, respectively, on Q 1 STD effects was plotted in Origin 8.1G and fitted to a one-site binding model equation: y ϭ ⌬I STD,max ϫ x/(k ϩ x). One value for k was obtained from globally fitting all Q 1 signals (Table 3).
Experiments for detecting interligand Overhauser effects (ILOEs) were performed in the same buffer as STD NMR experiments. To 25 M NqrA or perdeuterated NqrA (prepared as described above), respectively, 200 M Q 1 and 400 M DBMIB or HQNO were added from 20 mM stock solutions in DMSO-

TABLE 2 Analysis of the rates of quinol formation by Na ؉ -NQR at varying Q 1 or Q 2 concentrations in the absence or presence of DBMIB
The NADH concentration in the assays was 50 M. The rates of Q 1 and Q 2 reduction were background-corrected by the rate of quinol formation when using DBMIB as the sole electron acceptor as shown in the bottom section of the table. For calculation of K I and K I Ј, the equation for mixed inhibition was applied:

TABLE 3 Dependence of STD NMR effects of Q 1 on the concentration of DBMIB in the presence of NqrA and Na ؉ -NQR, respectively, and on the concentration of HQNO in the presence of NqrA
Changes of STD effects shown in Figs. 5-7 were globally fitted to a one-site binding model: ⌬I STD ϭ ⌬I STD,max ϫ c ligand /(k ϩ c ligand ).

TABLE 1 Inhibition of Na ؉ -NQR by DBMIB at varying NADH concentrations described by the Michaelis-Menten formalism
The concentration of Q 1 in the assays was 0.1 mM. Both NADH oxidation and quinol formation were recorded. Corresponding rates are presented in Fig. 2. For calculation of K I Ј, the equation for uncompetitive inhibition was applied .
Two Quinones Bind to the NqrA Subunit of Na ؉ -NQR OCTOBER 18, 2013 • VOLUME 288 • NUMBER 42 d 6 . DMSO-d 6 was added to a final concentration of 5% in the sample. Two samples for control experiments were prepared in the same manner but excluding NqrA or quinone analogs, respectively. The two-dimensional NOESY experiments were performed with water suppression achieved by low power presaturation of the water signal during relaxation delay and mixing time (28,29). The mixing time was set to 600 ms. Up to 1024 increments with 48 scans in each increment were collected at a spectral width of 10 ppm. Selective one-dimensional NOESY experiments based on the double pulsed field gradient spinecho method were performed using Bruker's pulse program SELNOGP (30). Resonances of interest were selectively excited with a low power Gaussian-shaped pulse. The mixing time was set to 600 ms. Up to 8192 scans were collected at a spectral width of 20 ppm.
All NMR experiments were acquired at 300 K on a Bruker AVANCE III 600-MHz spectrometer equipped with a cryogenic 5-mm TCI-H/C/N triple resonance probe with actively shielded z-gradient. Spectra were processed and analyzed with the software TopSpin (Bruker; v3.1).

RESULTS
Interaction of Na ϩ -NQR with Q 1 and DBMIB Monitored by the Electron Transfer Activities-As DBMIB is an analog and antagonist of ubiquinone, it was used in this work to obtain information on quinone binding by the Na ϩ -NQR. First, we questioned whether DBMIB has an influence on NADH oxidation by the Na ϩ -NQR. The effect of DBMIB on NADH oxidase activity was assayed by maintaining a constant concentration of the artificial electron acceptor Q 1 while varying the concentration of the substrate NADH. Both NADH oxidation and quinol formation rates were recorded. Determined V max and K m values for NADH oxidase activity were 147.5 Ϯ 1.7 mol min Ϫ1 mg Ϫ1 and 29.6 Ϯ 0.6 M, respectively, when quantifying NADH oxidation and 38.9 Ϯ 1.2 mol min Ϫ1 mg Ϫ1 and 24.5 Ϯ 1.5 M, respectively, when analyzing quinol formation ( Table 1). The difference between the two rates is caused by reaction of ubisemiquinone with oxygen as described (31). Preincubation of the enzyme with 10 M DBMIB led to a decrease in the NADH oxidation rate V max as well as in the Michaelis constant K m with similar manifestations on the NADH oxidation and quinol formation rates (Fig. 2).
Varying the ubiquinone-1 concentration of the buffer while starting the assay always by adding 50 M NADH allowed for characterization of the quinone reductase activity of the Na ϩ -NQR. Assays in the absence of DBMIB displayed a clearly defined Michaelis-Menten-like behavior of the enzyme with a V max value of 25.8 Ϯ 0.6 mol min Ϫ1 mg Ϫ1 and a K m value of 14.0 Ϯ 0.9 M for quinol formation ( Table 2). If quinone reductase activity was measured in the presence of 10 or 25 M DBMIB (Fig. 3, upper panel), a decrease of V max and an increase of K m values were observed (Table 2).
We also performed enzymatic assays with ubiquinone-2, decylubiquinone, and ubiquinone-10. With Q 2 , similar quinol formation rates were observed as with Q 1 ; however, the measured rates show a sigmoidal dependence on the concentration of Q 2 and did not yield meaningful kinetic parameters (Fig. 3, lower panel, and Table 2). This observation is in accordance with earlier reports (32) and is most likely caused by diffusion and desolvation processes involving the DDM micelle that become rate-limiting under these conditions. In the case of decylubiquinone and Q 10 , significantly lowered rates compared with Q 1 were observed (data not shown). This precludes inhibition studies with DBMIB because, under these conditions, DBMIB itself will be reduced much faster than decylubiquinone or Q 10 .
Binding of DBMIB to the NqrA Subunit-We have recently shown that subunit NqrA of the Na ϩ -NQR harbors a Q binding site (9). We now asked whether NqrA also interacts with DBMIB. NqrA contains three tryptophan residues. Measurement of tryptophan fluorescence during DBMIB titration  enabled us to detect changes in the microenvironment of the binding site. Incubation of NqrA with DBMIB led to quenching of the fluorescence. Saturation of the quenching was observed at concentrations Ͼ100 M DBMIB with 6% of the original fluorescence remaining, indicating that at least two of the three tryptophan residues of the protein are closely affected by DBMIB binding. The data were fitted to one-site and two-site binding models (Fig. 4). The goodness of fit was slightly higher with the two-site model, yielding an apparent K D value of 3.08 Ϯ 1.11 M and a second K D value of 33.24 Ϯ 34.10 M ( Table 4).
As a control, we performed the same titration with Q 1 (data not shown); however, Q 1 is apparently not as well suited for fluorescence quenching studies as DBMIB as documented by fluctuating data points at low concentrations. Therefore, we refer to the K D values reported previously (9).
STD NMR confirmed the binding of DBMIB to NqrA with, at a given ligand-to-protein-ratio, STD effects that were even stronger (maximum, 25%) than with Q 1 (maximum, 8.5%) (data not shown). Furthermore, signals in the proton NMR spectrum of DBMIB became broad and shifted upon binding to NqrA, which is indicative of a binding kinetic at the intermediate NMR time scale. To circumvent problems caused by exchange, broadening STD effects were quantified at a higher excess of DBMIB (40:1) as compared with experiments with Q 1 . The strongest STD signals originated from the isopropyl group of DBMIB (methyl groups, 25.0%; CH group, 24 -25%). The methyl group bound directly to the quinone ring showed an STD effect of 16.1%. This suggests that the isopropyl group of DBMIB is in closest contact to NqrA.
The Inhibitor DBMIB and Ubiquinone Q 1 Bind Simultaneously to the NqrA Subunit as Well as to Holo-Na ϩ -NQR-To investigate how DBMIB affects the binding of Q 1 to the Na ϩ -NQR, we studied the influence of DBMIB on the STD effects of Q 1 at constant Q 1 concentration with holo-Na ϩ -NQR and the NqrA subunit, respectively (Figs. 5 and 6). To this end, STD effects in the absence of DBMIB were set as a reference, and all other effects measured in the presence of the inhibitor were expressed as percent changes. Importantly, STD effects increased with higher concentrations of DBMIB in the presence of Na ϩ -NQR as well as NqrA. STD effects of Q 1 were affected by DBMIB in a concentration-dependent fashion. This effect FIGURE 5. Saturation transfer difference NMR spectroscopy of Q 1 and DBMIB in the presence of NqrA. A, reference (upper panel) and three STD NMR (three lower panels) spectra of Q 1 at increasing DBMIB concentration in the presence of NqrA. The reference spectrum was scaled to 37% of its original intensity. Signals of the same intensity in the STD spectrum correspond to an STD effect of 37%. The panel shows growing STD effects of Q 1 signals with increasing DBMIB concentration. Impurities and solvent signals are marked with asterisks. 1D, one-dimensional. B, changes of saturation transfer difference NMR effects of Q 1 depending on the concentration of DBMIB. Experimental data points were fitted to a one-site binding model (y ϭ I max ϫ x/(k ϩ x); solid lines); the constant k was fitted globally for all curves.

TABLE 4 Binding of DBMIB to the NqrA subunit according to tryptophan fluorescence quenching
Fluorescence data were fitted to a one-or two-site binding model.

One-site binding model:
y Two-site binding model:  Two Quinones Bind to the NqrA Subunit of Na ؉ -NQR OCTOBER 18, 2013 • VOLUME 288 • NUMBER 42 leveled off at DBMIB concentrations above 80 -100 M in both cases. STD changes of signals of Q 1 were fitted globally to a one-site binding model (y ϭ ⌬I STD,max ϫ x/(k ϩ x)). All five signals of Q 1 were used in the case of NqrA, and two signals of Q 1 could be quantified reliably in the case of Na ϩ -NQR. The other signals were occluded by DDM or affected by spectral artifacts. Resulting EC 50 values (k) were 28.5 Ϯ 11.2 M for the Na ϩ -NQR and 26.7 Ϯ 2.6 M for NqrA (Table 3). It should be noted that determining changes of STD effects cannot be performed with highest accuracy because of the intrinsically low signal-to-noise ratio. Therefore, the EC 50 values obtained by this technique should be considered as estimates with error margins in the range of 10 -30% in the present case. An identical experiment was also done with Q 1 and HQNO in the presence of NqrA. The influence of HQNO on STD effects of Q 1 was very similar to the influence of DBMIB (Fig. 7). STD changes of signals of Q 1 were fitted globally to a one-site binding model (y ϭ ⌬I STD,max ϫ x/(k ϩ x)), which provided an EC 50 value of 51.0 Ϯ 8.7 M ( Table 3).
The influence of DBMIB on STD effects of Q 1 was also analyzed with regard to the binding mode of Q 1 (Table 5). For this purpose, a ratio between the STD effect of Q 1 at maximal concentration of DBMIB and in the absence of DBMIB was calculated. These ratios were then normalized to the minimal ratio (for H-10) that was set to 1. Rather than increasing or decreasing all STD effects of Q 1 homogeneously, the quinone analog DBMIB affected the STD intensities of H-5Ј and H-7 more strongly than the remaining signals of Q 1 . Our analysis shows that STD effects of H-5Ј and H-7 increased disproportionately upon addition of DBMIB, suggesting that these protons are brought into closer contact to the binding site in the presence of the inhibitor.
The Inhibitors DBMIB and HQNO and Ubiquinone Q 1 Occupy an Extended Quinone Binding Site-The fact that, according to STD NMR, DBMIB and HQNO, respectively, do not displace Q 1 from the binding site but alter the binding mode of the native substrate led us to surmise whether both pairs, Q 1 /DBMIB and Q 1 /HQNO, interact with NqrA in direct vicinity to each other.
To test this, we recorded NOESY spectra of both pairs of ligands in the presence of NqrA (Figs. 8 and 9). Indeed, we observed ILOEs between Q 1 and DBMIB and between Q 1 and HQNO (33). These correlations were only observed in the presence of NqrA. A mixture of both ligands in the absence of NqrA did not show ILOEs (Figs. 8 and 9) nor did NqrA show crosspeaks at these positions in the absence of ligands. It is important to note that the observation of ILOEs critically depends on the presence of NqrA yet proved independent of whether protonated or perdeuterated NqrA was used during NOESY experiments (Fig. 8). This would indicate that the correlations observed originate from direct transfers between the ligands and not from protein-mediated effects.
Further evidence for a critical role of NqrA for observing ILOEs is that, in the absence of NqrA, Q 1 and DBMIB did not mutually perturb their chemical shifts. This indicates that both quinone analogs do not form complexes in solution.

DISCUSSION
The kinetic characterization of the Na ϩ -NQR complex suggests that quinone reduction is best described by the classic Michaelis-Menten formalism, indicating a single catalytically active quinone binding site. However, our recent surface plasmon resonance study of Q 1 binding to the isolated NqrA subunit provided hints for the presence of a second quinone binding site with lower affinity (9). Furthermore, numerous studies report uncompetitive inhibition of the Na ϩ -NQR by inhibitor molecules that are supposed to interact with the quinone binding site (8,34). It is also noteworthy that enzyme complexes catalyzing related chemical reactions (bc 1 complex and b 6 f complex) were indirectly shown to have spacious quinone binding sites that can accommodate more than 1 quinone eq at a time at least according to inhibitor studies by high resolution magic angle spinning NMR spectroscopy (21) and according to EPR experiments (20,22).
Most recently, crystal structures of the alternative NADH dehydrogenase Ndi1 from yeast were reported (35,36). Indeed,   Feng et al. (35) localized two ubiquinone-4 molecules in the binding site of Ndi1. One Q 4 is located close to FAD, and the other appears to be only loosely bound. This scenario might be very close to the situation of the Na ϩ -NQR characterized here. It is noteworthy, however, that almost simultaneously Iwata et al. (36) published a structure of Ndi1 with only one Q 2 mol-  C2, and C3) showing traces extracted from the NOESY spectrum of Q 1 and DBMIB. The ligand atom giving rise to the diagonal signal is indicated at the top left of each panel, and the diagonal signal is marked with an asterisk. In each panel, the blue trace originates from the NOESY spectrum measured in the presence of protonated NqrA, the green trace was measured in the presence of perdeuterated NqrA, and the red trace was measured in the absence of NqrA. ILOEs are indicated by arrowheads. The position of the diagonal signal is marked with an asterisk. Two Quinones Bind to the NqrA Subunit of Na ؉ -NQR OCTOBER 18, 2013 • VOLUME 288 • NUMBER 42 ecule in the binding site bound in a rather different way as compared with the study of Feng et al. (35). This controversy awaits clarification by independent experimental methods, e.g. NMR.
Here we show that the NqrA subunit of the Na ϩ -NQR binds two quinone-type ligands adjacent to each other in an extended binding site. We used the quinone analog and antagonist DBMIB, a well known inhibitor of electron transfer complexes, to serve as a second quinone-type ligand that can easily be distinguished from Q 1 by NMR spectroscopy. In addition, we found very similar results with another well known inhibitor, namely HQNO, which had been proposed to interact with Q-sites in earlier studies (19). Extension of our detailed NMR and kinetic studies to Q 2 as a ligand was unsuccessful probably because of unfavorable kinetics and interference caused by the detergent DDM (32).
In the current study, we provide (to our knowledge for the first time in the case of quinone-binding enzymes) direct experimental evidence that two quinone analog molecules are situated in immediate vicinity to each other in the binding pocket of NqrA. This result was obtained by measuring ILOEs between ubiquinone-1 and the inhibitors DBMIB and HQNO, respectively. NOEs between both molecules could in principle derive from several situations. If Q 1 and DBMIB or HQNO formed stable complexes in solution one would expect intermolecular NOEs. However, in the absence of NqrA, we did not detect intermolecular NOEs (see control experiment shown in Figs. 8 and 9). Alternatively, NOEs between Q 1 and inhibitors could have been mediated by protons of the binding site of NqrA. Such intermolecular NOEs, called INPHARMA effects, could build up even in a purely competitive binding situation (37). However, for INPHARMA effects to develop, it is relevant that the magnetization has to be "stored" on the receptor during the ligand exchange process. This transfer is not possible if the experiment is carried out with a perdeuterated receptor, and thus, INPHARMA effects should be absent or at least strongly reduced under these conditions. We performed the same experiment with protonated as well as perdeuterated NqrA ( 2 H labeling Ն95% according to mass spectrometry) and saw no significant reduction in the intensity of the ILOE cross-peaks (Fig. 8). The only plausible interpretation is simultaneous binding of both ligands directly adjacent to each other within an extended quinone binding site of NqrA.
In the case of DBMIB, the precise relative arrangement of Q 1 and DBMIB within the binding site cannot be determined to great accuracy from the ILOEs because of the generally low signal intensity and because essentially all protons of Q 1 show ILOEs to all protons of DBMIB with slightly varying intensities. Interestingly, in the case of HQNO, the strongest ILOEs toward protons of Q 1 originate from protons located at the terminus of the aliphatic chain, suggesting that the alkyl chain of HQNO is oriented toward Q 1 during simultaneous binding to NqrA (Fig.  9A). Furthermore, we could exploit the high sensitivity of selective one-dimensional double pulsed field gradient spin-echo NOESY experiments to show that H-3 is the only aromatic proton of HQNO that displays a weak ILOE effect to the methoxy groups of Q 1 (Fig. 9B). Taking all observed (and absent) ILOEs into account, we propose a plausible arrange-ment of the two ligands, Q 1 and HQNO, adopted during simultaneous binding to the NqrA subunit (Fig. 9C), noting, however, that this should not be considered as a unique solution.
The fact that DBMIB does not displace Q 1 from the binding site in a competitive manner is furthermore supported by an STD NMR titration. In a competitive binding situation, the STD effects of Q 1 should decrease upon titrating in DBMIB because DBMIB would displace Q 1 from the binding site, lowering the fraction of bound Q 1 and, thus, lowering the amount of saturation transferred from NqrA to Q 1 . However, we observed exactly the opposite. Increasing the concentration of DBMIB led to a markedly increasing STD effect of Q 1 . In addition, in the presence of DBMIB, the ratio of STD effects of individual protons differed from the situation without inhibitor. This suggests that DBMIB alters the binding mode of Q 1 , which in turn leads to a more efficient saturation transfer and therefore stronger STD effects. The EC 50 of this interaction was 27 M. A similar effect was observed in the case of HQNO with an EC 50 of 51 M. This finding, in addition to the presence of ILOEs between Q 1 and both DBMIB and HQNO, respectively, suggests that simultaneous binding of quinone analog ligands could be a general mechanism.
Our current results show that DBMIB, at the accessible concentration of 400 M, is not able to displace Q 1 from the binding site as we saw the highest STD effects of Q 1 at this concentration of DBMIB. Also the presence of substantial ILOEs under these conditions indicates that apparently the "mixed situation" with 1 eq of Q 1 and 1 eq of DBMIB residing at the same time in the binding site is favored over double occupancy by 2 eq of DBMIB. However, the two-dimensional NOESY experiments provide evidence for double occupancy of two Q 1 molecules within the extended binding site of NqrA. Rather strong NOE correlations were observed between methyl groups of the isoprenyl chain of Q 1 and the methoxy groups of Q 1 . Within the same Q 1 , these protons are separated by 8 -10 Å. Therefore, it is unlikely that these correlations originate from intramolecular dipolar interactions. Thus, we propose that these NOE contacts are indeed ILOEs between two individual Q 1 molecules. The fact that these cross-peaks showed up in the presence of protonated as well as deuterated NqrA further supports that these are true ILOEs and not due to relay or spin diffusion effects mediated by the protein.
Can these results, obtained with the isolated NqrA subunit, be transferred to the holo-Na ϩ -NQR complex? To gain insight into the physiological relevance, we performed the same STD NMR titration described above with Q 1 and DBMIB in the presence of holo-Na ϩ -NQR. This experiment provided an EC 50 value of 28.5 M, which is virtually identical to the EC 50 value for NqrA.
To further confirm this result, we performed an enzyme inhibition experiment with the entire enzyme complex. In the presence of DBMIB, the rate of NADH oxidation (V max ) was decreased, whereas K m was lowered. This is characteristic of an uncompetitive mode of inhibition typical for multisubstrate enzymes. That the catalytic site of NADH oxidation was not perturbed by DBMIB is indicated by the fact that the ratios of V max to K m in the presence or absence of DBMIB remain constant (Table 1). First, this confirms the notion that DBMIB dis-rupts the electron pathway further downstream, presumably at the quinone reductase site due to the structural resemblance to ubiquinone (Fig. 1). Second, this ascertains that NADH oxidase and quinone reductase sites of the Na ϩ -NQR are coupled in our experimental setup and no electrons short circuit to quinone at the level of the NADH oxidase. This needs to be considered because Q 1 can serve as an artificial electron acceptor when assaying the isolated NADH dehydrogenase domain (38). For quinone reduction, we observed a mixed mode of inhibition in the presence of DBMIB as documented by the rates of quinol formation.
Evaluating our experiments according to a Michaelis-Menten model yielded an average inhibition constant (K I Ј) of 53.2 M. This K I value is strikingly similar to the EC 50 values determined in the STD NMR experiments with the NqrA subunit and the holo-Na ϩ -NQR complex (26.7 Ϯ 2.6 and 28.5 Ϯ 11.2 M, respectively). In summary, this strongly indicates that the interaction of the inhibitor DBMIB with the extended binding site of the NqrA subunit is causing the inhibition of enzymatic turnover measured with the holoenzyme complex.
Recently, Juárez et al. (10) investigated in V. cholerae Na ϩ -NQR two mutations that had been discovered earlier in the Na ϩ -NQR of V. alginolyticus. These mutations in subunit NqrB confer resistance to the antibiotic korormicin in V. alginolyticus (8). Mutations of glycine residues Gly-140 and Gly-141, located in the NqrB subunit, to alanines were found to have a profound effect on the enzymatic activity of Na ϩ -NQR, leading the authors to propose that the NqrB subunit carries the site of quinone reduction (10). Our current study supports our earlier results indicating that the NqrA subunit carries the catalytically relevant quinone binding site of the Na ϩ -NQR. One should consider that long range structural perturbations induced by mutations on subunit NqrB might affect the quinone binding properties of subunit NqrA.
A plausible extension of our results shown here is to assume that holo-Na ϩ -NQR also binds 2 eq of the natural substrate, Q 8 , with their hydrophilic headgroups in vivo. We propose that these two quinones are central to the last electron transfer steps. In this model, two subsequent one-electron transfers, probably from a flavosemiquinone located on the NqrB subunit, would reduce ubiquinone to the ubisemiquinone and subsequently to ubiquinol, which is then released by the enzyme. A schematic view of the Na ϩ -NQR complex with the two binding sites for quinones and quinone analogs located on the NqrA subunit is shown in Fig. 10.
The presence of two quinones in the binding site could enhance the efficiency of catalysis in at least two ways. One quinone ligand could serve essentially as a cofactor that remains tightly bound to the enzyme and switches between the quinone and the semiquinone state, whereas the other quinone ligand is much less tightly bound, reduced in two steps to the quinol, and then released from the enzyme. Alternatively, the second, low affinity binding site could constitute a waiting position for the next ubiquinone substrate to enter the site of reduction, leading to a faster supply of fresh substrate and, thus, accelerated enzymatic turnover. Which of these mechanisms is finally operational in the Na ϩ -NQR will be addressed in future studies.