Sodium ion cycling mediates energy coupling between complex I and ATP synthase

Abstract

We show here sodium ion cycling between complex I from Klebsiella pneumoniae and the F₁F₀ ATP synthase from Ilyobacter tartaricus in a reconstituted proteoliposome system. In the course of NADH oxidation by complex I, an electrochemical sodium ion gradient was established and served as a driving force for the synthesis of ATP from ADP and phosphate. In the opposite direction, the Δμ̃_Na⁺ generated by ATP hydrolysis could be coupled to NADH formation by reversed electron transfer from ubiquinol to NAD. For reverse electron transfer, a transmembrane voltage larger than 30 mV was obligatory. No NADH-driven proton transport into the lumen of proteoliposomes was detected. We conclude that Na⁺ is used as the exclusive coupling ion by the enterobacterial complex I.

Every living cell establishes transmembrane electrochemical gradients with the help of primary ion pumps. These power plants of the cell provide or use energy-rich metabolites like ATP or NADH. The chemiosmotic theory (1) describes how the exergonic oxidation of NADH by the NADH:quinone oxidoreductase (complex I) generates an electrochemical potential that drives the endergonic synthesis of ATP by ATP synthase. Homologues of these two enzyme complexes are found in the inner membrane of mitochondria, in the thylakoid membrane of chloroplasts, and in the cytoplasmic membrane of bacteria. Complex I is a large lipoprotein complex (500 or 1,000 kDa in bacteria or eukaryotes) that couples the oxidation of NADH with quinones to the transport of protons (2) across the membrane. The redox cofactors of complex I (one FMN and up to nine Fe/S clusters) are located in the promontory arm of the L-shaped complex extending into the cytoplasm, whereas the coupling cations (H⁺ or Na⁺) must pass through the membranous part of the complex (Fig. 1). Diminished complex I activity is associated with Parkinson's disease (3) and aging (4) and represents the most frequently encountered inherited defect of the oxidative phosphorylation (OXPHOS) system (5). Despite considerable knowledge of primary sequences (6), cofactors (7), and assembly (8), the mechanism of redox-driven proton transport by complex I and the subunit(s) that guide the proton through its membranous part is unknown. A promising approach is to study bacterial counterparts of complex I that are smaller but possess all central subunits required for redox-driven H⁺ (or Na⁺) transport (9). In particular, an Na⁺-translocating complex I found in enterobacteria like Escherichia coli (10) or Klebsiella pneumoniae (11) is a useful model to trace the pathway of the coupling cation, as exemplified by the Na⁺-translocating F₁F₀ ATP synthase (12). The analysis of the cation-translocating step is facilitated by selectively removing or adding Na⁺, which does not affect the stability of the protein. Variation of the proton concentration is far more restricted, however, because the physiologically active proteins usually tolerate proton concentrations of 10⁻⁸ to 10⁻⁶ M. By using Na⁺- rather than H⁺-translocating complexes, we demonstrate that complex I and the ATP synthase reconstituted into proteoliposomes are fully coupled via the sodium cycle. Both NADH-dependent ATP synthesis and ATP-driven NAD reduction strictly depend on the electrochemical sodium gradient (Fig. 2). These results will be fundamental in elucidating the mechanism of Na⁺ or H⁺ translocation by complex I.

Figure 1.

Schematic model of K. pneumoniae complex I. The model is based on the L-shaped gross structure of E. coli complex I that exhibits 90% sequence identity to the K. pneumoniae complex I. Complex I from E. coli consists of two arms, which may be arranged perpendicularly to each other or side by side (31). Subunits NuoB and NuoI and the fused NuoCD subunit connect a peripheral, NADH-oxidizing subcomplex (NuoE-G) with the membranous NuoA, NuoH, and NuoJ–N subunits. NADH oxidation is catalyzed by the FMN located on subunit NuoF (51 kDa) at the cytoplasmic (matrix, or N-) side of the membrane. Electrons are transferred via the Fe/S clusters to the redox-active head group of a substrate quinone (indicated in white) that exchanges with the quinone pool in the membrane (32). This quinone is located at the N-side in the vicinity of the NuoH (ND1), NuoB (PSST), and NuoD (49 kDa) subunits (bovine nomenclature in parentheses; ref. 33). Two distinct Q radical species participate in electron transport in bovine complex I (32). One (or both) might represent a tightly bound redox cofactor of complex I (indicated in black) that mediates electron transfer from the high-potential Fe/S cluster N2, presumably located on NuoB (PSST; ref. 34) to the substrate quinone. During NADH oxidation, two sodium ions are transported from the cytoplasmic (N-) side to the periplasmic (P-) side of the membrane (11).

Figure 2.

Sodium cycle mediates energy coupling between complex I and ATP synthase. The two primary sodium pumps, complex I from K. pneumoniae and ATP synthase from I. tartaricus, were coreconstituted into proteoliposomes and coupled via an electrochemical sodium gradient. The direction of Na⁺ transport is indicated by dotted arrows. (A) NADH-driven ATP synthesis (forward electron transfer) was measured with coreconstituted quinones serving as electron acceptors. (B) ATP-driven NADH synthesis (reverse electron transfer) was measured with coreconstituted quinols serving as electron donors.

Materials and Methods

Reconstitution.

Complex I from K. pneumoniae (11), the ATP synthase from Ilyobacter tartaricus (13), or the ATP synthase from E. coli (14) was purified as described. Reconstitution of the complexes into liposomes was achieved by a dilution method established for complex I from K. pneumoniae under anoxic conditions in the presence of Q2, Q6, Q8 (Qn, the ubiquinone side chain homologue with n isoprene units in the polyprenyl side chain), and E. coli lipids (11). To follow NAD reduction by complex I, the corresponding quinols obtained by reduction of Q2 (200 nmol), Q6 (300 nmol), and Q8 (300 nmol) with LiAlH₄ were added. LiAlH₄ selectively reacts with carbonyls but not with NAD and was removed from the organic phase by mixing with water. The ethereal layer containing the quinols was recovered and added to the lipids before reconstitution.

Complex I from K. pneumoniae (2–3 mg) and the ATP synthase from I. tartaricus (2–3 mg) were reconstituted individually or together. Complex I was also coreconstituted with the ATP synthase from E. coli (6 mg). The dilution buffers required for the formation of proteoliposomes were chosen according to subsequent experiments. Proteoliposomes used for Na⁺ transport measurements were prepared in 10 mM Tris⋅HCl (pH 7.5), 50 mM KCl, 10 mM MgCl₂, and 10% glycerol. KCl was omitted in the reconstitution buffer if K⁺ diffusion potentials (inside positive) were established by adding KCl and valinomycin to the outside reservoir. Proteoliposomes used for acetate uptake experiments were prepared in 2 mM Tris⋅HCl, pH 7.0. The proteoliposomes were collected by ultracentrifugation and resuspended in dilution buffer to a final concentration of 2–4 mg·ml⁻¹ protein.

Formation of NADH.

ATP-dependent reduction of NAD by proteoliposomes containing complex I from K. pneumoniae, the ATP synthase from I. tartaricus or E. coli, and quinols was followed by dual wavelength spectroscopy at 340–370 nm (ɛ₃₄₀ = 6.22 mM⁻¹·cm⁻¹). Experiments were performed in sealed cuvettes under exclusion of dioxygen at 25°C. The reaction mixture (1 ml) contained 100 μl of proteoliposomes in buffer (10 mM Tris⋅HCl, pH 7.5/50 mM KCl/10 mM MgCl₂/10% glycerol) and 3 mM phosphoenolpyruvate and pyruvate kinase (10 units) for the regeneration of ATP from ADP. The reaction samples were degassed and flushed with N₂ eight times to remove dihydrogen from the anaerobe chamber that could act as a substrate for the NAD(P) reducing hydrogenase from K. pneumoniae (15). NAD (0.1 mM) was added first, and after 90 s, reverse electron transfer was started by the addition of 2.5 mM ATP. The Na⁺-specific ionophore monensin (20 μM) or the K⁺-specific ionophore valinomycin (20–50 μM; the final ratio of lipid:valinomycin was 58:1) was added as indicated.

Other Methods.

NADH oxidation and quinone reduction (11) and ATP hydrolysis (16) were performed as described.

Synthesis of ATP by proteoliposomes containing complex I and the ATP synthase was performed in reaction mixtures (1 ml) containing 10 mM Tris⋅HCl (pH 7.5), 50 mM KCl, 10 mM MgCl₂, 10 mM potassium phosphate, 10% glycerol, 300 μl of proteoliposomes, 2.5 mM ADP, and 1 or 5 mM NaCl. ATP synthesis was initiated by the addition of NADH (0.1 mM) and followed by the luciferin–luciferase assay (17).

Na⁺ transport into proteoliposomes was followed by atomic absorption spectroscopy in 10 mM Tris⋅HCl (pH 7.5), 50 mM KCl, 10 mM MgCl₂, 10% glycerol, and 5 mM NaCl (11). Membrane potentials were determined from the accumulation of [¹⁴C]thiocyanate (18). Proteoliposomes containing complex I and the ATP synthase maintained a membrane potential of 76 mV (inside positive) imposed by K⁺/valinomycin.

Proton uptake into proteoliposomes was measured by the quenching of 9-amino-6-chloro-2-methoxyacridine at 25°C (19).

The alkalization of proteoliposomes in the course of NADH oxidation by complex I was determined by the accumulation of [1-¹⁴C]acetate (20). Protein was determined with the bicinchoninic acid method by using reagent obtained from Pierce.

Results

We showed previously that complex I from K. pneumoniae translocates 2 Na⁺ per NADH oxidized and quinone reduced (11). This translocation was shown to be a primary transport event that did not depend on the proton motive force. However, the formation of a transmembrane voltage due to Na⁺ transport was not demonstrated, and pumping of protons by complex I in addition to the observed Na⁺ transport could not be excluded. Here, we address these questions by coupling the Na⁺-translocating complex I from K. pneumoniae to another primary Na⁺ pump, the ATP synthase from I. tartaricus (13), via the electrochemical sodium gradient (Fig. 2).

Complex I and the ATP Synthase Establish a Sodium Cycle.

Complex I from K. pneumoniae and the ATP synthase from I. tartaricus were coreconstituted into liposomes by a dilution procedure established for complex I, and NADH- or ATP-dependent Na⁺ translocation was performed. If either NADH or ATP was omitted, no transport of Na⁺ into the proteoliposomes occurred. In the presence of 0.1 mM NADH or 2.5 mM ATP, the initial rates of Na⁺ transport were 4.0 μmol·min⁻¹·mg⁻¹ protein for complex I or 0.2 μmol·min⁻¹·mg⁻¹ protein for the ATP synthase, respectively, demonstrating that the procedure suitable for complex I also allows the reconstitution of the ATP synthase in an active state. Assuming an average volume of the proteoliposomes of 7 μl·mg⁻¹ phospholipids (21), an internal sodium ion concentration of 25 mM established with NADH or 15 mM established with ATP, at an external sodium concentration of 5 mM, clearly demonstrates active Na⁺ transport by complex I or the ATP synthase into the proteoliposomes. We then asked whether a sodium motive force established by complex I during NADH oxidation (forward electron transfer) could drive the synthesis of ATP from ADP and phosphate in vesicles containing both complex I and the ATP synthase (Fig. 2). Addition of ADP (2.5 μmol) to the assay mixture containing proteoliposomes (0.6 mg of protein), NADH (100 nmol), and phosphate (10 μmol) resulted in the formation of 640 pmol of ATP during the first 10 s. It is concluded that complex I and the ATP synthase are coupled via the electrochemical sodium gradient during NADH → Q electron transfer. Lowering the NaCl concentration in the assay did not affect the total amount of ATP synthesized after 10 min, but the initial rate decreased from 640 pmol of ATP (5 mM Na⁺) to 30 pmol of ATP (1 mM Na⁺) formed during the first 10 s of the reaction (Fig. 3). Note that both complex I from K. pneumoniae (11) and the ATP synthase from I. tartaricus (13) require ≈0.3 mM Na⁺ for half-maximal activity. After 1 min, the amount of ATP formed in the presence of 1 mM Na⁺ reached the value observed with 5 mM Na⁺ (≈700 pmol of ATP). Hereby, 90 nmol of NADH was oxidized to NAD. Further incubation of the proteoliposomes resulted in only slightly higher amounts of ATP formed (850 pmol after 10 min). The formation of ATP by the Na⁺-dependent ATP synthase is an endergonic process (ΔG₀′ = 30 kJ·mol⁻¹) that obligatorily requires a transmembrane voltage (18). We conclude that the NADH-driven Na⁺ transport by complex I from K. pneumoniae (11) results in the buildup of a transmembrane voltage (ΔΨ). This was independently confirmed by measuring ΔΨ with the help of a membrane-permeable anion, [¹⁴C]thiocyanate. NADH oxidation by reconstituted complex I resulted in the formation of a transmembrane potential of 80 mV (inside positive), compared with 5 mV of residual potential observed in the absence of NADH.

Figure 3.

NADH-driven ATP synthesis. Complex I from K. pneumoniae and ATP synthase from I. tartaricus were coreconstituted into proteoliposomes (0.6 mg of protein) and added to a reaction mixture containing 100 nmol of NADH and 10 μmol of phosphate. ATP synthesis was started by the addition of 2.5 μmol of ADP in the presence of either 1 mM NaCl (●) or 5 mM NaCl (○). Error bars indicate results from four measurements.

The observed coupling of complex I and the ATP synthase during NADH:Q oxidoreduction prompted us to study the reverse reaction: the sodium motive force established by the ATP synthase during ATP hydrolysis should drive the reduction of NAD by complex I (Fig. 2). This reaction requires quinols as electron donors that were added to the lipids before the reconstitution of complex I and the ATP synthase. No NADH formation was observed with proteoliposomes containing complex I and the ATP synthase in the presence of quinols and NAD. Upon addition of ATP, an increase in absorbance at 340 nm was observed, indicating the formation of NADH (Fig. 4). The presence of NADH in the assay mixture was confirmed by its oxidation with lactate dehydrogenase (15 units) and pyruvate (3 mM). It is concluded that the sodium motive force generated by the ATP synthase drives the endergonic reduction of NAD by complex I with a more positive electron donor, quinol. The anaerobically purified and reconstituted complex I from K. pneumoniae is fully functional in both forward (NADH → Q) and reverse (QH₂ → NAD) electron transfer. The initial rate of reverse electron transfer by purified and reconstituted complex I from K. pneumoniae was 0.03 μmol·min⁻¹·mg⁻¹, compared with 0.1 μmol· min⁻¹·mg⁻¹ (22) or 0.26 μmol·min⁻¹·mg⁻¹ (23) observed with complex I in inside-out vesicles from Paracoccus denitrificans or in submitochondrial particles, respectively. To follow reverse complex I activity in submitochondrial particles, a high phosphorylation potential was maintained by an excess of ATP (4.5 mM; ref. 23). In our assay, the addition of pyruvate kinase and phosphoenolpyruvate ensured a high ATP-to-ADP ratio. From 100 nmol of NAD present at the start of the reaction, 16 nmol of NADH was formed after 9 min. Because ATP was continuously regenerated, the concentrations of NAD and NADH observed at the end of the reaction do not represent educt and product concentrations under conditions of true equilibrium. The initial rate of NAD reduction (0.03 μmol·min⁻¹·mg⁻¹ protein or 2.0 nmol·min⁻¹·mg⁻¹ lipid) during reverse electron transfer by complex I coreconstituted with the ATP synthase (Fig. 4) was comparable with the rate of ATP synthesis (0.04 μmol·min⁻¹·mg⁻¹ protein or 0.4 nmol·min⁻¹·mg⁻¹ lipid) during forward electron transfer in the same system (Fig. 3). Both NADH oxidation and ATP hydrolysis occur at the external side of the liposomes (corresponding to the cytoplasmic or matrix aspect of the membrane), because these substrates cannot diffuse through the lipid bilayer. Proteoliposomes containing complex I and the ATP synthase in opposite orientations with respect to their NADH- or ATP-binding sites do not contribute to the observed NADH-driven synthesis of ATP or the ATP-driven reduction of NAD.

Figure 4.

ATP-driven reverse electron transfer by complex I. Complex I from K. pneumoniae was coreconstituted with the Na⁺-translocating ATP synthase from I. tartaricus or with the H⁺-pumping ATP synthase from E. coli. The assay contained proteoliposomes, quinols, and NaCl (5 mM). Reduction of NAD was observed only after the addition of ATP (2.5 mM). The presence of NADH in the assay mixture was confirmed by addition of 15 units of lactate dehydrogenase (LDH) and 3 mM pyruvate. Traces from top to bottom: Na⁺-coupled ATPase, no ionophore added; H⁺-coupled ATPase, no ionophore added; Na⁺-coupled ATPase, 20 μM monensin; Na⁺-coupled ATPase, valinomycin (ratio of lipid:valinomycin = 58:1).

Reverse electron transfer by complex I from K. pneumoniae was also studied with the help of the H⁺-pumping ATP synthase from E. coli. Rates of NAD reduction dropped from 0.03 μmol·min⁻¹·mg⁻¹ protein (Na⁺-coupled ATPase) to 0.003 μmol· min⁻¹·mg⁻¹ protein (H⁺-coupled ATPase; Fig. 4). Hence, the electrochemical sodium gradient established by the Na⁺-coupled ATPase resulted in 10-fold higher rates of NAD reduction by complex I, but reverse complex I activity was also observed with the H⁺-coupled ATPase. Apparently, the transmembrane potential generated by the H⁺-coupled ATPase was sufficient for reverse electron transfer by complex I in the presence of Na⁺ (5 mM on the inside). The electrochemical Na⁺-gradient composed of ΔpNa⁺ and ΔΨ, as established by the Na⁺-coupled ATPase, resulted in higher rates of NAD reduction by complex I. This was further analyzed with the help of specific ionophores (see below).

Driving Forces for Reverse (QH₂ → NAD) Electron Transfer by Complex I.

Reverse electron transfer by complex I from K. pneumoniae depends on ATP hydrolysis by the Na⁺-translocating ATP synthase from I. tartaricus that generates both a chemical sodium gradient (ΔpNa⁺) and a transmembrane voltage (ΔΨ). We now asked whether ΔΨ, ΔpNa⁺, or both driving forces are required for NAD reduction by complex I. The components of the electrochemical sodium gradient established during ATP hydrolysis can be selectively destroyed in the presence of valinomycin and K⁺ (ΔΨ) or monensin (ΔpNa⁺). The minimum concentrations of these ionophores required for complete dissipation of ΔΨ or ΔpNa⁺ were determined in control experiments with proteoliposomes containing the reconstituted ATP synthase from I. tartaricus. The accumulation of Na⁺ was completely abolished in the presence of 20 μM monensin, indicating that the Na⁺ concentration gradient established by the ATP synthase was rapidly collapsed because of compensating fluxes of Na⁺ complexed to monensin. Likewise, the effect of K⁺/valinomycin on the transmembrane voltage (inside positive) generated by the ATP synthase during ATP hydrolysis was analyzed (17). The transmembrane voltage decreased from 100 mV (without addition) to a residual value of 30 mV in the presence of 20 μM valinomycin (ratio of lipid:valinomycin = 58:1). No further decrease in membrane potential was achieved by increasing the valinomycin concentration.

As a next step, the effect of these ionophores on ATP-driven NAD reduction by complex I was analyzed. The formation of NADH was initiated by adding proteoliposomes containing complex I from K. pneumoniae, the ATP synthase from I. tartaricus, and quinol to anoxic buffer in the presence of NAD and ATP (Fig. 5). Upon addition of monensin (20 μM final concentration, in ethanol), the reverse complex I activity dropped by 85%. An equal amount of ethanol had no effect on NAD reduction by complex I. The initial rates observed before the addition of ethanol with or without ionophores varied by 3% (±0.001 μmol·min⁻¹·mg⁻¹ protein). By adding valinomycin at a lipid:valinomycin ratio of 58:1, NADH formation was completely inhibited (Fig. 5). As expected, the residual transmembrane potential of 30 mV established by the ATPase despite saturating concentrations of valinomycin was not sufficient to bridge the redox potential gap between ubiquinol and NAD. Similar results were obtained if the ionophores were mixed with the proteoliposomes before the addition of NAD and ATP (Fig. 4). In the presence of valinomycin, NAD reduction was completely prevented (Fig. 4, lowest trace), but some residual reverse complex I activity (6%) was observed in the presence of monensin (Fig. 4, second trace from bottom). We conclude that the chemical Na⁺ concentration gradient established by the ATP synthase accelerates NAD reduction by complex I but is dispensable, whereas a transmembrane voltage larger than 30 mV is obligatorily required for QH₂:NAD oxidoreduction (reverse electron transfer) by complex I.

Figure 5.

Inhibition of reverse electron transfer by complex I with ionophores. Assays contained NAD (100 μM), ATP (2.5 mM), and NaCl (1 mM). Formation of NADH was initiated by the addition of proteoliposomes coreconstituted with complex I from K. pneumoniae and the ATP synthase from I. tartaricus and supplied with quinols. Arrows indicate the addition of monensin (A, 20 μM, gray trace) or valinomycin (B, 50 μM, gray trace; ratio of lipid:valinomycin = 58:1). The ionophores were added from stock solutions in ethanol (20 μl). The addition of 20 μl of ethanol had no effect on NADH formation (black traces).

Complex I from K. pneumoniae Does Not Transport H⁺ in Addition to Na⁺.

The coupling of the Na⁺-translocating complex I to the Na⁺-translocating ATP synthase from I. tartaricus via a sodium cycle strongly suggests that Na⁺ is the exclusive coupling ion used by complex I from K. pneumoniae. However, the transport of protons by complex I in addition to Na⁺ ions should be considered. Another option is that under low Na⁺ concentrations, protons are transported by complex I instead of Na⁺, as observed with the Na⁺-translocating ATP synthase (16, 24). To address these questions, complex I from K. pneumoniae was reconstituted together with the proton-translocating ATP synthase from E. coli. H⁺ transport into the proteoliposomes was followed by the quenching of 9-amino-6-chloro-2-methoxyacridine fluorescence. As expected, the addition of ATP to the proteoliposomes resulted in proton transport by the E. coli ATP synthase, whereas no protons were transported by the subsequent addition of NADH (not shown). The Na⁺ concentration in the assays was 50 μM, which is well below the 0.3 mM Na⁺ required for half-maximal activity of complex I from K. pneumoniae (11). Similar results were obtained if NADH was added before ATP. No H⁺ transport was observed in the course of NADH oxidation by complex I from K. pneumoniae, but the subsequent addition of ATP resulted in H⁺ translocation by the ATP synthase from E. coli. The experiments were repeated under saturating Na⁺ concentrations (5 mM NaCl), but again, no NADH-induced proton transport was observed (data not shown). Note that the reconstituted complex I was active and catalyzed NADH-driven Na⁺ transport, as confirmed in parallel control experiments. It is concluded that complex I from K. pneumoniae does not transport protons at low (50 μM) or saturating (5 mM) Na⁺ concentrations. This was further corroborated by measuring the change in internal pH of the proteoliposomes on oxidation of NADH by reconstituted complex I. Na⁺ transport into vesicles by a primary Na⁺ pump results in alkalization of the proteoliposomes if the uptake of the positively charged Na⁺ ions is compensated by the efflux of protons by the protonophore. In contrast, primary H⁺ transport results in an acidification of the liposomes. The alkalization of vesicles can be followed by the accumulation of radioactively labeled [1-¹⁴C]acetate. Acetic acid traverses the membrane but is trapped inside the vesicles in its anionic form at alkaline pH. Addition of NADH to proteoliposomes containing complex I in the presence of carbonylcyanide m-chlorophenylhydrazone (CCCP) resulted in the uptake of 0.7 nmol acetate per mg phospholipid. In control experiments without NADH, 0.2 nmol acetate per mg phospholipid was accumulated. These results demonstrate that NADH oxidation by complex I results in alkalization of the proteoliposomes and further support our notion that complex I from K. pneumoniae does not translocate protons instead of, or in addition to, Na⁺.

Discussion

Complex I from K. pneumoniae exclusively translocates Na⁺ and can be coupled to an Na⁺-translocating ATP synthase from I. tartaricus via the electrochemical sodium gradient in vitro (Fig. 2). Under physiological conditions, the electrochemical sodium gradient established by complex I will not directly drive ATP synthesis catalyzed by the H⁺-coupled ATP synthase from K. pneumoniae. However, the conversion of the sodium motive force into a proton motive force and vice versa is readily achieved by Na⁺/H⁺ antiporters located in the inner bacterial membrane (25). Therefore, the combination of a primary respiratory Na⁺ pump and a H⁺-coupled ATP synthase is not detrimental to the bacterial cell. For example, Vibrio alginolyticus likewise contains an Na⁺-translocating NADH:quinone oxidoreductase (9) and an H⁺-dependent ATP synthase (26). An E. coli mutant strain containing an Na⁺-coupled ATP synthase is capable of using the proton motive force established during succinate respiration for ATP synthesis (19). Enterobacteria like E. coli or K. pneumoniae contain an alternative NADH:Q oxidoreductase (NDH II) that, in contrast to complex I, does not act as a primary redox pump. The two NADH dehydrogenases are synthesized in varying amounts depending on environmental and growth phase-dependent signals (27). The electrochemical Na⁺ gradient generated by complex I in addition to the proton motive force established by quinol oxidases may be of advantage to enterobacteria during growth on nutrients that have to be cotransported with Na⁺ or under conditions of alkaline pH (pH_in < pH_out) or high osmolarity.

In complex I, electrons derived from the oxidation of NADH are transferred via FMN to several low-potential iron–sulfur clusters, the high-potential Fe₄/S₄ cluster N2, and enzyme-bound semiquinone radical(s) to the electron acceptor quinone (ref. 28; Fig. 1). Cluster N2 is assumed to be located on the NuoB subunit (PSST in bovine complex I) and exhibits a pH-dependent redox potential, which is considered an indication of proton-coupled electron transfer events. Current models of H⁺ pumping by complex I state that the reduced cluster N2 is oxidized by a tightly bound semiquinone (29) or by quinone drawn from the Q pool in the membrane (30). Protonation of a reduced quinone species with protons coming from the negative side of the membrane yields quinol. The subsequent deprotonation of the quinol is directed toward the positive side of the membrane. Both models rely on the reduction of carbonyl oxygen to a hydroxyl group under covalent binding of the coupling cation (H⁺). This is not feasible in the case of Na⁺ acting as coupling cation for complex I. Because we found no evidence for proton transport in addition to Na⁺ translocation by complex I from K. pneumoniae, an H⁺ transport mechanism that involves the formation of quinol can be excluded for the enterobacterial complex I.

There seems to be a discrepancy between the amounts of ATP or NADH produced by complex I and the ATP synthase during forward or reverse electron transfer. Less than 0.1% of the ADP but 16% of the NAD present at the start of the reaction were converted to the corresponding product during NADH-driven ATP synthesis or ATP-driven NAD reduction, respectively. This difference is due to the use of an ATP-regenerating system that, during reverse electron transfer by complex I, ensured the detection of NADH by visible spectroscopy. In contrast, no regenerating system was required to demonstrate NADH-driven ATP synthesis, as the luciferin–luciferase assay is specific for ATP and very sensitive. Forward and reverse electron transfer by reconstituted complex I was studied in the presence of chloride, which partially dissipates the membrane potential because of compensatory fluxes of chloride. This favors the accumulation of Na⁺ in the proteoliposomes, which enabled us to follow Na⁺ translocation by complex I or the ATP synthase. In contrast, proteoliposomes prepared in potassium phosphate buffer are tightly coupled and exhibit no steady-state oxidation of NADH and measurable accumulation of Na⁺. Under these conditions, secondary ion fluxes are lacking, and the rapid generation of a stable membrane potential by complex I prevents further transport of Na⁺ into the proteoliposomes (11).

During forward electron transfer, complex I generates a membrane potential of 80 mV that drives the synthesis of ATP. In the reverse reaction, the chemical gradient of Na⁺ ions, [Na⁺]_P > [Na⁺]_N, accelerates the reduction rates of NAD by complex I but is not indispensable, whereas the membrane potential is obligatorily required for NAD reduction by complex I. This suggests a voltage-dependent step in the reaction cycle of complex I. The redox cofactors and substrate binding sites of complex I that have been identified so far are located toward the negative side of the membrane (Fig. 1). Up to now, there is no experimental evidence for transmembrane electron transfer in complex I that could be driven by the membrane potential. As a working hypothesis, we assume that the transport of Na⁺ ions through the hydrophobic part of the pump represents an electrogenic step in the reaction cycle of complex I.

Abbreviation

Qn

the ubiquinone side chain homologue with n isoprene units in the polyprenyl side chain