The C-terminally truncated NuoL subunit (ND5 homologue) of the Na+-dependent complex I from Escherichia coli transports Na+.

The NADH:quinone oxidoreductase (complex I) from Escherichia coli acts as a primary Na+ pump. Expression of a C-terminally truncated version of the hydrophobic NuoL subunit (ND5 homologue) from E. coli complex I resulted in Na+-dependent growth inhibition of the E. coli host cells. Membrane vesicles containing the truncated NuoL subunit (NuoLN) exhibited 2-4-fold higher Na+ uptake activity than control vesicles without NuoLN. Respiratory proton transport into inverted vesicles containing NuoLN decreased upon addition of Na+, but was not affected by K+, indicating a Na+-dependent increase of proton permeability of membranes in the presence of NuoLN. The His-tagged NuoLN protein was solubilized, enriched by affinity chromatography, and reconstituted into proteoliposomes. Reconstituted His6-NuoLN facilitated the uptake of Na+ into the proteoliposomes along a concentration gradient. This Na+ uptake was prevented by EIPA (5-(N-ethyl-N-isopropyl)-amiloride), which acts as inhibitor against Na+/H+ antiporters.

Mitochondrial complex I (NADH:quinone oxidoreductase) is the largest multiprotein complex of the oxidative phosphorylation (OXPHOS) system. Diminished complex I activity is associated with Parkinson's disease (1) and aging (2) and represents the most frequently encountered inherited defect of the OXPHOS system (3). Despite the considerable knowledge on primary sequences, cofactors and assembly (4 -8), the mechanism of redox-driven proton transport by complex I and the subunit(s) that guide the proton through its membranous part are unknown. A promising approach is to study bacterial counterparts of complex I that are smaller but possess all subunits required for redox-driven H ϩ (or Na ϩ ) transport (9). In particular, a Na ϩ -translocating complex I found in enterobacteria like Escherichia coli (10) or Klebsiella pneumoniae (11,12) is a useful model to trace the pathway of the coupling cation, as exemplified by the Na ϩ -translocating F1F0 ATP synthase (13).
The L-shaped complex I is composed of a peripheral arm extending into the bacterial cytoplasm (or the mitochondrial matrix) and a hydrophobic arm that is embedded in the membrane (14). In addition to the L-shaped conformation, a "horseshoe-conformation" was observed for E. coli complex I, with two arms arranged side by side (15). Upon purification, the peripheral and the membrane arm tend to dissociate (16 -18). Se-quence comparisons indicate that the membrane-embedded part of complex I consists of seven conserved hydrophobic subunits with homologues found on bacterial or mitochondrial genomes and additional nuclear-encoded subunits in the case of eukaryotic complex I. One or several of these seven conserved subunits (NuoA, H, J, K, L, M, N, in the Na ϩ -translocating complex I from E. coli, or the homologues ND3, 1, 6, 4L, 5, 4, 2 in the H ϩ -translocating mitochondrial complex I) participate in Na ϩ (or H ϩ ) transport through complex I. So far, the cation-translocating subunit(s) of complex I have not been identified. A likely candidate is the NuoL subunit (ND5 according to bovine nomenclature), which is related to subunit A of bacterial multicomponent Na ϩ (or K ϩ ):H ϩ antiporters (19 -22). Here evidence is presented that the C-terminally truncated NuoL subunit of E. coli complex I transports Na ϩ .

EXPERIMENTAL PROCEDURES
Strains and Media-E. coli K-12 strain DH5␣ (MBI Fermentas) was used as host for clonings. Expression of the truncated nuoL gene was performed using E. coli DH5␣, E. coli strain CP875 (23), or E. coli strain EP432 (24). E. coli EP432 is resistant to kanamycin and lacks the Na ϩ /H ϩ antiporters NhaA and NhaB. These antiporters are not related to subunits of the multicomponent antiporters belonging to the CPA3 family (22). Cells were grown aerobically in Luria-Bertani medium without added NaCl in the presence of 10 mM glucose and 50 mM potassium phosphate (pH 7.6). E. coli EP432 was also grown aerobically in mineral medium (25) at pH 8.0 with glucose or glucose plus Larabinose as indicated. If necessary, 100 g ml Ϫ1 ampicillin or 50 g ml Ϫ1 kanamycin was added. Cells were harvested, washed once in 10 mM Tris-HCl, pH 8.0, 200 mM KCl, 0.5 mM dithiothreitol and suspended in buffer. Inside-out membrane vesicles were obtained by French press cell rupture (26). Washed membrane vesicles used for Na ϩ transport measurements were prepared in the anaerobic chamber under an atmosphere of N 2 /H 2 (95/5%) (11) in 10 mM Tris-HCl, pH 8.0, 200 mM KCl, 0.5 mM dithiothreitol, 0.25 M sucrose (24).
Construction of Plasmids-All plasmids are pISC1 derivatives, which contain a ColE1 origin of replication and the araC gene encoding the activator protein, which induces transcription in the presence of arabinose (27). PISC vectors confer ampicillin resistance. Plasmid pISC2 contains a unique NdeI site downstream of the p ara promoter. Vector pISC3 is a pISC2 derivative containing an XbaI/NdeI linker derived from vector pET-14b (Novagen), which replaces the unique NdeI site of pISC2 and introduces six histidine residues to the N terminus of the target protein (28). A 1.1-kb DNA fragment encoding for the truncated NuoL subunit (amino acids 1-369; NuoL N ) of complex I from E. coli was obtained by PCR using Pfu-DNA-polymerase (New England Biolabs). The primers used were 5Ј-TTCATATGAACATGCT-TGCCTTAACC3-Ј (forward) and 5ЈGCGAATTCTTAACGCAGACCGC-CCATCTTG-3Ј (reverse). Coding sequences of the nuoL gene are underlined. NdeI and EcoRI sites (bold letters) and a stop codon (italics) were introduced by PCR. The PCR product was digested with NdeI and EcoRI and cloned into pISC2, resulting in vector pECL3. An NdeI/EcoRI fragment containing the truncated nuoL gene was obtained from pECL3 and cloned into pISC3, resulting in vector pECL5. Vector pECL5 was subjected to double-strand sequencing (Microsynth, Balgach, Switzerland). The sequence obtained was identical to the nuoL sequence (accession number NP-416781) found on the E. coli genome (29).
Solubilization and Reconstitution-Washed membrane vesicles (40 mg protein, 1 ml) from E. coli DH5␣ transformed with pECL5 (NuoL N ) or pISC3 (control) grown on Luria Bertani medium in the presence of 10 mM glucose were mixed with 9 ml buffer (20 mM Tris-HCl, pH. 8.0, 300 mM KCl, 2 mM NaCl, 9% glycerol) containing 0.1 g dodecylmaltoside (Glykon) and stirred on ice (30 min). After ultracentrifugation, the supernatant containing the solubilized membrane proteins was added to 2.5 ml nickel-nitrilotriacetic acid agarose resin (Qiagen) equilibrated with buffer. The slurry was gently shaken overnight at 4°C. The resin was centrifuged and washed with 50 ml buffer containing 0.05% dodecylmaltoside. The resin was packed in a column, and bound NuoL N was eluted with 5 ml buffer containing 0.05% dodecylmaltoside and 50 mM imidazole. The enriched NuoL N protein (10 mg) was reconstituted into liposomes by a dilution method established for complex I from K. pneumoniae (11) using E. coli lipids (150 mg, Avanti). The lipids were mixed with NuoL N (4.5 ml eluate from the affinity column in 0.05% dodecylmaltoside), and the formation of proteoliposomes was achieved by dropwise addition of 90 ml buffer (10 mM Tris-HCl, pH 8.0, 300 mM KCl). The proteoliposomes were collected by ultracentrifugation and resuspended in buffer to a final concentration of 60 -120 mg ml Ϫ1 lipid.
Na ϩ Transport-Na ϩ transport was followed under air at room temperature as described previously (11). Incubation mixtures contained in 0.31 ml the reconstituted proteoliposomes (4.5 mg lipid) and 20 M valinomycin (Sigma) in 10 mM Tris-HCl, pH 8.0, 300 mM KCl. If indicated, 0.1 mM EIPA 1 (5-(N-ethyl-N-isopropyl)-amiloride (Sigma)) was added to the assay. The reaction was started by the addition of 5 mM NaCl (final concentration). At different times, samples of 70 l (1.1 mg lipid) were applied to a 1-ml plastic syringe containing 0.6 ml Dowex 50 (K ϩ ). The proteoliposomes were immediately eluted with 0.8 ml deionized water. The eluate containing Na ϩ entrapped in the vesicles was analyzed by atomic absorption spectroscopy. In controls performed with 5 mM NaCl in buffer in the absence of vesicles, Na ϩ was completely absorbed to the Dowex columns. Blank values obtained with vesicles in buffer without NaCl added corresponded to the internal Na ϩ content of the liposomes (0.5-2.4 nmol Na ϩ per sample after passage through the Dowex column). Na ϩ transport into native membrane vesicles was followed in 0.3 ml 10 mM Tris-Mes, pH 8.0, 200 mM KCl, 2.5 mM MgCl 2 (24). The incubation mixture contained 2 mg of protein and 20 M valinomycin. The internal Na ϩ content of the vesicles was 4 nmol Na ϩ mg Ϫ1 protein. The reaction was started by adding 5 mM NaCl, and at different times, aliquots of 70 l were applied to the Dowex columns as described above.
Other Methods-Oxidation of NADH or deaminoNADH (nicotinamide hypoxanthine dinucleotide) by membrane vesicles with O 2 as electron acceptor was followed at 340 nm (⑀ 340 ϭ 6.22 mM Ϫ1 cm Ϫ1 ). NADH-induced proton uptake into native membrane vesicles was measured by the quenching of ACMA (9-amino-6-chloro-2-methoxyacridin) in 1.5 ml of 10 mM Tris-HCl, pH 8.0, 200 mM KCl at 25°C (30). The residual Na ϩ concentration in the buffer was 0.13 mM. DeaminoNADH, NADH (potassium salt), and ACMA were obtained from Sigma. Protein was determined by the bicinchoninic acid method using the reagent obtained from Pierce. SDS-PAGE was performed with 10% polyacrylamide according to (31). The polypeptides were blotted onto a nitrocellulose membrane (Hybond-C Extra, Amersham Biosciences), and His-tagged proteins were identified by immunostaining using anti-His (4) antibodies (Qiagen).

RESULTS
The C-terminally Truncated NuoL Subunit Increases the Na ϩ Permeability of E. coli Membranes-Complex I (NDH I) from E. coli couples the oxidation of NADH with quinone to the translocation of Na ϩ (10). An important goal is to identify the membranous complex I subunit(s) that participate in cation transport. A prime candidate is the hydrophobic NuoL subunit of complex I from E. coli (ND5 in Bos taurus, or Nqo12 in Paracoccus denitrificans) that exhibits striking sequence similarity to multicomponent Na ϩ /H ϩ antiporters. A sequence alignment of the NuoL subunit from E. coli complex I with the MnhA subunit of the Na ϩ /H ϩ antiporter from Staphylococcus aureus reveals 133 conserved amino acids in NuoL encompassing a total of 613 amino acid residues. Fig. 1 shows the putative topology of the NuoL subunit of E. coli complex I that is based on phylogenetic analyses and experimental data obtained for NuoL of Rhodobacter capsulatus complex I (32). The central region of the NuoL subunit encompassing transmembrane helices III to XIII exhibits the highest sequence similarity to Na ϩ /H ϩ antiporters. The region from amino acid 1 to 369 encompasses 107 residues of the 133 amino acids that are fully conserved in the complete E. coli NuoL subunit and in the FIG. 1. Putative topology of the NuoL subunit (ND5 homologue) of complex I from E. coli and relationship with subunit A of multicomponent Na ؉ /H ؉ antiporters. The topology of the E. coli NuoL subunit is based on data obtained for NuoL from R. capsulatus complex I (32). A total of sixteen hydrophobic helices are predicted for NuoL, and its N and C termini are located in the periplasm. The beginning of a putative helix is indicated by its first N-terminal amino acid residue (E. coli numbering). Shaded helices contain four or more residues that are conserved in NuoL from E. coli complex I and in the MnhA antiporter subunit from S. aureus. The amphipathic helices (X) and (XI) probably do not traverse the membrane (32). Truncation of NuoL after arginine 369 results in the shortened NuoL N polypeptide that includes helices I-(XI) and the conserved parts of the cytoplasmic loop between helices (XI) and XII. Circles indicate the positions of mutations in helices XII and XV that result in diminished function of complex I or the related Na ϩ /H ϩ antiporter (19,44).
MnhA antiporter subunit from S. aureus. This portion of the NuoL subunit contains nine putative transmembrane helices (I-IX) and the amphipathic helices (X) and (XI) that are not assumed to traverse the membrane (32). Truncation of NuoL at R369 (Fig. 1) results in a shortened NuoL subunit that includes the conserved parts of the cytoplasmic loop between helices (XI) and XII. The gene fragment encoding for amino acid residues 1-369 was amplified by PCR and cloned into a low-copy expression vector that is repressed by glucose and derepressed by L-arabinose, resulting in vector pECL3. The truncated NuoL subunit (NuoL N ) has a calculated molecular mass of 40,054 Da, compared with 66,440 Da of the NuoL subunit. In addition, a vector encoding for NuoL N encompassing an N-terminal Histag (His 6 -NuoL N ) was constructed (pECL5). Detection of His 6 -NuoL N by immunostaining revealed that the protein was predominantly found in membranes of E. coli (Fig. 2). No signal was detected in cell-free extracts from E. coli in the absence of plasmid (not shown). Due to its hydrophobic nature, the truncated NuoL subunit showed an apparent molecular mass of only 33 kDa on SDS-PAGE.
The effect of plasmid pECL5 encoding His 6 -NuoL N on the growth of E. coli DH5␣ or E. coli EP432 was investigated. Cells were grown aerobically at 37°C in phosphate-buffered Luria Bertani medium (pH 7.6) containing 10 mM glucose and 20, 110, or 140 mM KCl or NaCl. The optical density (OD) was followed at 600 nm for 24 h, and synthesis of the truncated NuoL subunit was induced by addition of 8 mM arabinose at A 0.5 . A general trend observed with both E. coli strains was a decrease in final growth yield in pECL5 cells compared with the control cells at increasing salt concentrations (Table I). Both KCl and NaCl diminished the growth yield in the presence of His 6 -NuoL N , but NaCl showed a more drastic effect than KCl. In E. coli DH5␣, growth in the presence of His 6 -NuoL N was impaired at 140 mM NaCl (final A 1.9 , compared with A 8.8 with the control vector). This inhibition of growth was not accompanied by an apparent change in morphology. Growth of E. coli EP432 transformed with pECL5 encoding for His 6 -NuoL N was already inhibited at 110 mM NaCl (final A 1.0 , compared with A 2.8 observed with the control vector). The increased toxicity of Na ϩ even in the absence of NuoL N observed with E. coli EP432 compared with E. coli DH5␣ is due to the lack of two Na ϩ /H ϩ antiporters in strain EP432. In contrast to E. coli DH5␣, Na ϩ -induced growth inhibition of E. coli EP432 in the presence of His 6 -NuoL N was accompanied with an elongation and aggregation of cells. These morphological changes were not observed with E. coli EP432 growing in 140 mM KCl in the presence of NuoL N . A similar association of growth inhibition with cell elongation was observed with E. coli strain DK8 cells overproducing the truncated, membrane-bound subunit a of the proton-translocating F1F0 ATPase (33).
It is concluded that the NuoL N protein promotes ion leakage through the E. coli membranes, hereby dissipating electrochemical gradients across the cell membrane and diminishing the growth yield. The data indicate an increased permeability of the membranes for Na ϩ induced by the NuoL N protein. This was further investigated by following the Na ϩ influx into membrane vesicles from E. coli EP432 transformed with either pECL5 or pISC3. In the presence of NuoL N , the initial rate of Na ϩ uptake was 18 nmol mg Ϫ1 protein during the first 10 s, compared with 9 nmol Na ϩ mg Ϫ1 observed with the control membrane vesicles (Fig. 3). The addition of ubiquinone-1 (0.49 mM, oxidized form) to the assay did not stimulate the rate of Na ϩ transport by membrane vesicles containing the truncated NuoL N subunit. In six independent experiments, the uptake of Na ϩ was 2-4-fold higher in membrane vesicles containing NuoL N compared with the control (50 -240 nmol Na ϩ min Ϫ1 mg Ϫ1 , corrected for initial Na ϩ uptake by control vesicles), demonstrating that NuoL N enhances Na ϩ transport through native membranes with initial rates that are comparable with secondary Na ϩ transporters. For example, the purified, reconstituted Na ϩ /H ϩ antiporter NhaA from E. coli has a specific activity of 600 nmol Na ϩ min Ϫ1 mg Ϫ1 (34).
An important question is whether the presence of the truncated NuoL N subunit in the cytoplasmic membrane of E. coli affects the assembly of complex I. The observed Na ϩ transport might be catalyzed by misassembled complex I containing the truncated NuoL subunit. The specific complex I activity of membrane vesicles was estimated from the oxidation of deami-noNADH with O 2 as terminal electron acceptor. In contrast to NADH, which is oxidized both by complex I and the alternative, non-electrogenic NADH dehydrogenase (NDH II), deamino-NADH is preferentially oxidized by complex I and is used to distinguish between complex I and NDH II in E. coli membranes (35). The rates of deaminoNADH oxidation by control membrane vesicles from E. coli EP432 (0.06 mol min Ϫ1 mg Ϫ1 ) were comparable with the rates observed with membrane vesicles containing the truncated NuoL subunit (0.08 mol min Ϫ1 mg Ϫ1 ). The rates of NADH oxidation by control membrane vesicles (0.36 mol min Ϫ1 mg Ϫ1 ) were essentially identical to the rates observed with membrane vesicles containing NuoL N (0.38 mol min Ϫ1 mg Ϫ1 ). The NuoL N protein did not diminish the activity of complex I in E. coli membranes, suggesting functional assembly of complex I despite the presence of its hydrophobic, truncated NuoL subunit.  Na ϩ -dependent Increase of Proton Permeability in Membranes Containing the C-terminally Truncated NuoL Subunit-Respiratory proton transport activity was studied in membrane vesicles from E. coli CP875 containing the His 6 -NuoL N protein that were energized by the addition of NADH as outlined above. E. coli contains two respiratory NADH:quinone oxidoreductases, the Na ϩ -translocating NDH I or complex I (10), and the non-electrogenic NDH II. During aerobic growth, NADH is mainly oxidized by the NDH II (36), and the quinol formed is oxidized by terminal oxidases under formation of an electrochemical proton gradient. E. coli membrane vesicles obtained by French press cell rupture are predominantly oriented inside-out (26), and the addition of NADH in the presence of oxygen results in the build-up of a proton motive force (inside positive). The NADH-driven proton transport into vesicles is followed by the quenching of ACMA fluorescence. Vesicles containing His 6 -NuoL N showed a strong, NADH-induced quench in the presence of KCl (20 mM), whereas the addition of NaCl (20 mM) prior to NADH diminished the quench signal by 30% (Fig. 4, upper panel). In contrast, the quenching of ACMA fluorescence was hardly affected by NaCl in membrane vesicles prepared from cells transformed with the control vector (Fig. 4,  lower panel). The slight increase in proton permeability of the control membranes in the presence of Na ϩ was due to the activity of endogenous Na ϩ /H ϩ antiporters from E. coli (24). Similar results were obtained with E. coli DH5␣ transformed with pECL3 encoding for the NuoL N protein (not shown). Note that the rates of NADH oxidation by membrane vesicles with O 2 as terminal electron acceptor were not affected by NuoL N (0.34 mol min Ϫ1 mg Ϫ1 in vesicles containing NuoL N , compared with 0.36 mol min Ϫ1 mg Ϫ1 in control vesicles). The diminished proton gradient established during NADH oxidation by vesicles in the presence of NuoL N and Na ϩ ions provides a rationale for the Na ϩ -dependent decrease in growth yield of E. coli cells producing the truncated NuoL subunit.
Na ϩ Transport by Reconstituted NuoL N and Inhibition by EIPA-The truncated, His-tagged NuoL subunit from E. coli complex I was solubilized, enriched by affinity chromatography (Fig. 2), and reconstituted into proteoliposomes. Addition of proteoliposomes containing His 6 -NuoL N to buffer in the presence of 5 mM NaCl resulted in a transient accumulation of Na ϩ inside the liposomes that was ϳ6-fold higher compared with the control liposomes (Fig. 5). Internal Na ϩ concentrations were calculated assuming a liposome volume of 3-10 l per mg lipid (37). At the start of the reaction, the Na ϩ concentration in the lumen of the liposomes was 100 -200 M, corresponding to an initial chemical Na ϩ concentration gradient (⌬pNa ϩ ) of 80 -100 mV (inside negative). The experiments were performed without respiratory substrates added, and the build-up of a transmembrane voltage, ⌬⌿, was prevented by valinomycin/ K ϩ . The uptake of Na ϩ by reconstituted His 6 -NuoL N was driven by the chemical Na ϩ concentration gradient (⌬pNa ϩ ). After 30 s, the maximum Na ϩ concentration in the interior was 6 -20 mM. This is in the range of the external Na ϩ concentration in the buffer and thus in accord with passive downhill transport of Na ϩ into the proteoliposomes mediated by His 6 -NuoL N . The initial uptake during the first 30 s was followed by an efflux of Na ϩ to the values observed with the control proteoliposomes (Fig. 5). Similar results were obtained with different preparations of NuoL N and different concentrations of proteoliposomes in the Na ϩ transport assay (not shown).
The inhibition of the truncated NuoL N subunit by EIPA was investigated. EIPA is an amiloride analogue that inhibits eukaryotic Na ϩ /H ϩ antiporters but not NhaA, the main Na ϩ /H ϩ antiporter in E. coli (38). Na ϩ uptake by His 6 -NuoL N was drastically diminished in the presence of 100 M EIPA, whereas the residual uptake of Na ϩ by the control proteoliposomes was not affected by EIPA (Fig. 5). This result suggests that the truncated NuoL subunit of complex I from E. coli shares features with Na ϩ /H ϩ antiporters.

DISCUSSION
Complex I is the largest redox-driven H ϩ pump in the respiratory chain. It is composed of a peripheral arm harboring one FMN and up to nine Fe/S clusters (5), and a membrane arm containing up to three ubiquinones (11,39,40). Although the peripheral, NADH-oxidizing fragment of complex I has been studied intensively, the membrane part is a terra incognita that awaits its functional characterization. The dissection of the membranous fragment of complex I into functional units will be a prerequisite to identify the subunits and amino acid residues that contribute to H ϩ (or Na ϩ ) transport. The Na ϩdependent complex I found in E. coli (10) and the related enterobacterium K. pneumoniae (11,12) serves as a model for the mitochondrial, H ϩ -pumping enzyme. The high degree of similarity of the NuoL subunit (ND5 homologue) from complex I with subunit A from multicomponent cation/H ϩ antiporters has led to the speculation that NuoL provides (part of) a proton channel in complex I (19). Unfortunately, the observed sequence similarity contributes very little to the understanding of H ϩ (or Na ϩ ) transport by complex I, because the multicomponent cation/H ϩ antiporters are only poorly characterized with respect to function and catalytic activity.
Evidence for Na ϩ transport by the C-terminally truncated NuoL subunit of complex I is presented here. The truncated NuoL subunit comprises most of the regions conserved between NuoL and the A subunit of the multicomponent antiporters belonging to the CPA3 family (22). Because the truncated NuoL subunit was produced in E. coli in the presence of endogenous complex I, one cannot completely exclude that the observed Na ϩ transport was due to other complex I subunits copurified with His 6 -NuoL N on the affinity column. On the other hand, the solubilization of E. coli membranes with dodecylmaltoside under the applied conditions of pH and salt concentration cleaves and inactivates complex I (17). It should also be noted that the specific complex I activity in native membranes was not affected by the presence of the truncated NuoL subunit. It is therefore considered very unlikely that the Na ϩ transport observed with reconstituted His 6 -NuoL N is catalyzed by misassembled complex I. The translocation of Na ϩ by the C-terminally truncated NuoL subunit of complex I from E. coli provides strong evidence for the participation of this subunit in cation transport through the membranous fragment of complex I.
The Na ϩ -translocating complex I from K. pneumoniae that exhibits high sequence similarity to E. coli complex I exclusively pumps Na ϩ at a ratio of 2 Na ϩ /2 electrons (11). Na ϩ transport by native complex I generates a transmembrane voltage that is not compensated by the counter-flow of protons (12). Therefore, one would expect unidirectional transport of Na ϩ by NuoL N . On the other hand, the results presented herein suggest a Na ϩ -dependent increase of proton permeability of membrane vesicles in the presence of NuoL N , which can be interpreted in terms of Na ϩ /H ϩ antiport. Na ϩ uptake by reconstituted His 6 -NuoL N was followed by the extrusion of Na ϩ , similar to results obtained with the reconstituted NhaA antiporter from E. coli (34). Na ϩ transport by His-NuoL N was drastically diminished in the presence of EIPA, an amiloride derivative that acts as inhibitor against Na ϩ /H ϩ antiporters. Note that EIPA also inhibits NADH oxidation by bovine complex I (41). It should be considered that the presence of NuoL N in E. coli host cells could result in an increase of the intracellular Na ϩ concentration, which in turn leads to the up-regulation of endogenous Na ϩ /H ϩ antiporters (42). As a consequence, the Na ϩ /H ϩ antiporter activity in membrane vesicles or enriched protein fractions could be higher in E. coli cells producing NuoL N compared with the control cells. Studies aimed at the determination of Na ϩ (or H ϩ ) transport stoichiometries using purified, reconstituted NuoL N are currently underway in this laboratory.
Biochemical (18) and structural (43) studies with bovine complex I suggest that the large ND5 (NuoL) subunit might be situated at the distal end of the membrane-embedded arm of the L-shaped molecule together with the ND4 (NuoM) subunit. In complex I from E. coli, the N-terminal part of the NuoL subunit (helices I-XI) may be part of the Na ϩ -translocating machinery, but electrogenic Na ϩ transport by the holo-complex probably requires the C-terminal NuoL domain (Fig. 6). Studies on site-directed mutants support a functional role of the C-terminal part of NuoL (helices XII-XVI) (Fig. 1). In human complex I, a substitution of glycine 465 (Gly-468 in E. coli complex I) in helix XIV with glutamate represents a primary mutation that causes Leber's hereditary optic neuropathy (44). In the related Na ϩ /H ϩ antiporter from Bacillus C-125, mutation of glycine 382 in helix XII to arginine results in an alkalisensitive phenotype, indicating diminished activity of the Na ϩ /H ϩ antiporter that confers alkali resistance in the wild type (19).
A central question is how the endergonic transport of Na ϩ (or H ϩ ) by complex I is driven by the exergonic oxidation of NADH with quinone (Fig. 6). Two quinones (Q6 and Q8) were found in complex I from K. pneumoniae (11). A quinone-binding site was identified in the vicinity of the NuoH (ND1), NuoB (PSST), and NuoD (49 kDa) subunits (45) (bovine nomenclature in brackets). Based on comparisons of primary sequences and known high-resolution structures of Q-binding proteins, the binding of Q to the NuoL (ND5) subunit was postulated from a putative Q binding motif located in helix (XI) and the loop between helices XII and XIII (Fig. 1) (46). Recently, Yagi and co-workers (41) reported on the labeling of the ND5 subunit (NuoL homologue) of bovine complex I with a photoaffinity analogue of fenpyroximate, a specific inhibitor of complex I. Fenpyroximate is assumed to bind at or close to a quinone binding site. The labeling of complex I by the photoaffinity analogue of fenpyroximate was diminished by various complex I inhibitors including amiloride derivatives like EIPA. This result compares favorably with the observed inhibition of Na ϩ transport by the truncated NuoL subunit (ND5 homologue) of E. coli complex I in the presence of EIPA. Passive transport of Na ϩ by the truncated NuoL subunit was not stimulated by the addition of water-soluble ubiquinone-1 (oxidized form). It will now be important to analyze whether there is an endogenous quinone bound to the NuoL subunit of E. coli complex I, and whether it participates in redox-driven Na ϩ transport. . Electron transport starts with the oxidation of NADH by FMN located on the NuoF subunit in the peripheral arm located in the cytoplasm (or in the mitochondrial matrix). Electrons are transferred via several FeS clusters to the high-potential [4Fe-4S] cluster N2 that is located on NuoB (PSST) (5) or on NuoI (TYKY) (47). The reduced cluster N2 is oxidized by quinone (Q) bound in close vicinity of the B and CD subunits and the membranous H subunit (45) under formation of quinol. This exergonic reaction provides the driving force for the endergonic transport of Na ϩ by the N-terminal part of NuoL (L N ) submitted via the C-terminal NuoL domain (L C ) (shaded arrow). Per NADH oxidized, two electrons are transferred to quinone, and two sodium ions are transported from the negatively to the positively charged aspect of the membrane (11).