We have previously shown that HPNhaA (Helicobacter pylori Na+/H+ antiporter) forms an oligomer in a native membrane of Escherichia coli, and conformational changes of oligomer occur between monomers of the oligomer during ion transport. In the present study, we use Blue-native PAGE to show that HPNhaA forms a dimer. Cysteine-scanning mutagenesis of residues 55–61 in a putative β-sheet region of loop1 and subsequent functional analyses revealed that the Q58C mutation resulted in an intermolecular disulfide bond. G56C, I59C and G60C were found to be cross-linked by bifunctional cross-linkers. Furthermore, the Q58E mutant did not form a dimer, possibly due to electrostatic repulsion between monomers. These results imply that Gln-58 and the flanking sequence in the putative β-sheet of the monomer are located close to the identical residues in the dimer. The Q58C mutant of NhaA was almost inactive under non-reducing conditions, and activity was restored under reducing conditions. This result showed that cross-linking at the dimer interface reduces transporter activity by interfering with the flexible association between the monomers. A mutant HPNhaA protein with three amino acid substitutions at residues 57–59 did not form a dimer, and yet was active, indicating that the monomer is functional.
Na+/H+ antiporters are secondary transporters and ubiquitous membrane proteins in the cytoplasmic membranes of bacterial cells, as well as in the cytoplasmic and organelle membranes of higher plant and animal cells [1–8]. These antiporters play a central role in the homoeostasis of intracellular Na+, pH and cell volume, and are required for cell viability [1–8]. To date, distinct Na+/H+ antiporters have been identified in at least 15 bacterial species [5,9–22]. Among them, NhaA was the first Na+/H+ antiporter identified in Escherichia coli (ECNhaA) that contributes to Na+ and Li+ tolerance, utilizing the H+ gradient across the cell membrane. Thus this antiporter decreases concentrations of Na+ and Li+ . Among the four distinct Na+/H+ antiporters so far identified in E. coli [5,9–11], NhaA is the main member protein conferring salt resistance. The primary structure of NhaA, which is composed of hydrophobic 12 TM (transmembrane) domains, is highly conserved among various bacteria, including Helicobacter pylori , whereas small hydrophilic portions in the N-terminal, loop1 and loop8 regions are less well conserved. Mutagenic studies of ECNhaA and HPNhaA (H. pylori NhaA) proteins have revealed one important (Asp-141) and two essential (Asp-171 and Asp-172) aspartate residues for ion transport, as well as several other residues that contribute to antiporter activity [24–26]. These residues are clustered in TM domains 4, 5, 10 and 11, and are conserved in ECNhaA [24,27–29] and in HPNhaA . Cysteine-scanning mutagenesis has been performed in these TM domains of HPNhaA ([30,31] and A. Kuwabara, Y. Tsuboi, R. Kinoshita and H. Kanazawa, unpublished work). Several TM residues, including Asp-141, Asp-172 and Lys-347, reacted with NEM (N-ethylmaleimide), despite their predicted inner-membrane localizations. Thus these residues appear to be involved in a hydrophilic ion transport pathway ([30,31] and A. Kuwabara, Y. Tsuboi, R. Kinoshita and H. Kanazawa, unpublished work). Moreover, replacement of Asp-171 or Asp-172 residues of TM domain 5 with cysteine residues eliminated the antiporter activity of NhaA, confirming that these two aspartate residues are essential for the ion transport .
The crystal structure of ECNhaA has been solved at a resolution of 3.45 Å (where 1 Å=0.1 nm) , revealing the topological arrangement of 12 TM domains. TM domains 4, 5, 10 and 11 are closely associated and constitute an ion transport pathway. Thus the ion transport pathway and binding site predicted by mutagenic studies, including cysteine-scanning mutagenesis of HPNhaA and ECNhaA, correspond well with the residues identified within the crystal structure of ECNhaA. Previous studies have also shown that the transporter activity of ECNhaA is modulated by intracellular pH. The Vmax value of ECNhaA is 2000-fold higher at pH 8.5 than at pH 7.0, and no activity can be detected during growth in acidic conditions . In contrast, the antiporter activity of HPNhaA is not sensitive to acidic and neutral pH . We have shown that increased HPNhaA activity at alkaline pH is linked to a structure that involves loop7 and TM domain 8, whereas high activity at acidic pH is due to a structure formed by TM domains 4 and 10 .
Several independent approaches, including chemical cross-linking , BN-PAGE (Blue-native PAGE) , cryo-electron microscopy  and ESR (electron spin resonance) , have shown that ECNhaA forms a homodimer. Using FRET (fluorescence resonance energy transfer), we have shown previously that HPNhaA forms an oligomer, possibly a dimer . It has been reported that mutant ECNhaA that lacks the β-hairpin of loop1 does not form a dimer, but is fully functional, suggesting that each subunit consists of a functional monomer . Since loop1 of ECNhaA is poorly conserved in the primary sequences of various bacteria, including H. pylori, it still remains to be determined whether or not loop1 of HPNhaA is involved in dimer formation, as is the corresponding loop in ECNhaA. Direct evidence for the presence of an HPNhaA dimer form has not been found so far. In our study  demonstrating oligomer formation of HPNhaA using FRET, we detected conformational changes between monomers induced by ion binding between the monomers. In addition, pH-induced conformational changes have been demonstrated previously between [31,34] and within [39,40] ECNhaA monomers, and in Methanococcus jannaschii NhaP  and HPNhaA . A model of the pH-dependent conformational changes of NhaA has been created using computer simulation [i.e. MD (molecular dynamics) analysis] . However, the mechanistic details underlying these observed conformational changes remain unknown.
In the present study, we have analysed the subunit composition of HPNhaA by BN-PAGE and found that HPNhaA monomer is functional. The monomers form a dimer, and the putative loop1 is located in an interactive domain between the monomers. Furthermore, we found that a cysteine replacement at Gln-58 leads to the formation of an intermolecular disulfide bond and causes a marked reduction in antiporter activity. Taken together, these results suggest that conformational changes in NhaA monomers during antiporter activity require flexible interactions between the monomers at the dimer interface.
Bacterial strains and culture conditions
E. coli strain HITΔAB− (ΔlacY, ΔnhaA, nhaB−)  was used to express the HPNhaA mutants and GFP (green fluorescent protein)–HPNhaA variants. E. coli JM109  cells were used to construct plasmids. Cells were cultured in LB (Luria–Bertani) broth  containing 87 mM KCl instead of NaCl (i.e. LBK). For growth on solid plates, 1.5% (w/v) agar was added to the medium. Antibiotic-resistant transformants were selected after growth on medium containing the appropriate antibiotics. Plates and liquid cultures were incubated at 37°C for 36 h.
Construction of expression plasmids
Plasmids encoding artificial dimer genes
The HPNhaA gene was amplified from the pBR-HP  [i.e. WT (wild-type)] vector via PCR, using primer 2 (Table 1) as a forward primer, containing a SphI restriction site, and primer 23 (Table 1) as a reverse primer. The PCR products were digested with SphI and ligated into the SphI fragment of pBR-HP to create the pBR-WT-WT vectors containing the two WT HPNhaA genes.
|Number||Primer||Sequence (5′ →3′)|
|Number||Primer||Sequence (5′ →3′)|
Plasmids encoding the NhaA gene carrying a loop1 deletion or substitution
The N-terminal fragments of NhaA were amplified from an HPNhaA vector that did not contain cysteine residues [CL-NhaA (cysteine-less NhaA)] , using primer 1 (Table 1) and reverse primers (i.e. primers 39–43 in Table 1). The C-terminal fragments were amplified from the CL-HPNhaA vector , using the primer 23 and forward primers (i.e. primers 18–22 in Table 1). The amplified N- and C-terminal fragments were mixed and the total HPNhaA coding sequence (including the deletion) was amplified using primer 1 and primer 23 (Table 1). The PCR product was digested with EcoRI and SalI, and ligated into the EcoRI–SalI fragment of pBR-HP , creating pBR-loop1-deleted-FLAG vectors.
Plasmids encoding a single cysteine residue mutant gene
A site-directed point mutation was introduced into the NhaA gene via a two-step PCR amplification of the CL-HPNhaA vector , using primer 1 and reverse primers (i.e. primers 24–38 in Table 1) or primer 23 and forward primers (i.e. 3–17 in Table 1). The PCR products were digested with EcoRI and SalI, and ligated into the EcoRI–SalI fragment of pBR-HP , creating pBR-single-Cys-mutant-FLAG vectors.
Preparation of membrane vesicles and measurement of Na+/H+ antiporter activity
Membrane vesicles were prepared from E. coli cells transformed with various plasmids, as described previously . Briefly, E. coli cells were disrupted with a French press. Membrane vesicles (100 μg) were centrifuged and resuspended in 2 ml of assay buffer (i.e. 10 mM Tricine and 140 mM KCl, adjusted to the desired pH with KOH, as described previously ). To create nonreducing conditions, 2-mercaptoethanol was removed from all buffers. Proton flow was measured by monitoring reductions in the fluorescence of ACMA (9-amino-6-chloro-2-methoxyacridine) after the addition of potassium lactate (5 mM, pH 7.0) . Fluorescence de-quenching, as a measure of antiporter activity, was monitored with a fluorometer (FP-750; Jasco) after the addition of 5 mM NaCl or LiCl.
Membrane vesicles from E. coli transformants containing various fragments of NhaA were subjected to SDS/PAGE analysis, as described previously . To create non-reducing conditions, 2-mercaptoethanol was removed from the SDS sample buffer. DTT (dithiothreitol) was added to give a final concentration of 200 mM in the SDS sample buffer when reducing condition was provided. The separated proteins were blotted on to GVHP filters (Millipore) and probed with anti-FLAG antibody (Sigma–Aldrich). Immunoreactive bands were visualized by enhanced chemiluminescence (Amersham Biosciences), as described previously .
Membrane vesicles derived from cells with HPNhaA, HPNhaA-linked dimer or Gln-58 mutants (0.5 mg of protein) were suspended in 300 μl of 20 mM sodium phosphate buffer (pH 7.4) containing 0.1 M NaCl, 0.1% DDM (n-dodecyl-β-D-maltoside) or 1% Triton X-100 and 10% glycerol. The suspensions were centrifuged at 100000 g for 1 h to remove insoluble materials, and the resultant supernatants were subjected to BN-PAGE (NativePAGE™ Novex® Bis-Tris gel system; Invitrogen). The separated proteins were blotted on to GVHP filters (Millipore) and probed with anti-FLAG antibodies. Immunoreactive bands were visualized by enhanced chemiluminescence (Amersham Biosciences), as described above.
Cross-linking between cysteine residues of HPNhaA
Membrane vesicles from E. coli HITΔAB− cells expressing single cysteine residue mutants were prepared as described above under non-reducing conditions. Aliquots of the membranes (100 μg of each) were reacted with 5 mM MTS-2-MTS (1,2-ethanediyl bismethanethiosulfonate) (Toronto Research Chemicals) or o-PDM (N,N′-o-phenylenediamaleimide) (Sigma–Aldrich). After 30 min of cross-linking at 25°C, the membranes were subjected to SDS/PAGE and Western blot analysis using anti-FLAG antibodies.
Fluorescence measurements and microscopic observations
Fluorescence measurements were performed using an FP-750 fluorometer (Jasco). FRET measurements of the everted membrane vesicles from cells expressing fusion proteins were performed in Tricine/KOH buffer (10 mM Tricine/140 mM KCl, pH 8.5). Membranes (100 μg of each) were irradiated in a 2 ml volume at 433 nm to excite CFP (cyan fluorescent protein). Fluorescence emission was recorded between 450 nm and 600 nm. Venus excitation was performed by illuminating membranes at 473 nm and fluorescence emission was recorded between 500 nm and 600 nm. Results of the FRET analysis were determined by subtracting the control emission spectra from the FRET emission spectrum. The control emission spectra were obtained via the excitation of two types of control cells or membranes, one expressing the CFP-tagged protein and the other expressing the Venus-tagged protein [38,48]. Venus gene was provided by Dr Atsushi Miyawaki (Riken Brain Science Institute, Wako, Japan). The apparent efficiency of FRET was calculated by dividing the maximum value of FRET fluorescence intensity by the maximum value of fluorescence intensity obtained upon direct excitation of Venus .
Gene manipulation and DNA sequencing
Preparation of plasmids, digestion of DNA with restriction endonucleases, ligation with T4 DNA ligase and other techniques were performed according to standard protocols . The nucleotide sequences of DNA fragments cloned into various expression plasmids were verified using an automated sequencer (PE Biosystems). Restriction endonucleases, T4 DNA ligase, Taq and KOD DNA polymerases were purchased from Toyobo. Oligonucleotides were synthesized by Invitrogen. Other reagents and materials were of the highest grade commercially available.
Stoichiometry of HPNhaA oligomers
We previously demonstrated that HPNhaA forms an oligomer . In the present study, we analysed the monomer stoichiometry in the HPNhaA oligomers using BN-PAGE. The mass of the membrane protein is often unpredictable in BN-PAGE analyses. Therefore we constructed a covalently bonded dimer of HPNhaA (linked WT–WT), as shown in Figure 1(A), for use as a mass marker. As expected, WT–WT was detected as double the size of the WT by SDS/PAGE (Figure 1B). Next, membrane vesicles containing WT or linked WT–WT molecules were solubilized in Triton X-100 and subjected to BN-PAGE. Under this condition, two bands were observed for the WT (Figure 1C, lane 3). Indeed, the apparent sizes of HPNhaA in BN-PAGE (160 or 320 kDa) (Figure 1C) and SDS/PAGE (40 kDa) (Figure 1B) were different. Since we observed previously that treatment with 1% Triton X-100 caused the HPNhaA oligomer to dissociate almost completely , it is probable that the lower band of WT should be a monomer, whereas the upper should be a dimer. Next, we repeated this experiment using a low concentration (0.1%) of DDM instead of Triton X-100. Under this condition, we observed a single band, corresponding to the linked dimer (Figure 1C, lane 5), and the WT–WT dimer separated at a larger mass (approx. 480 kDa) than it did when using Triton X-100, suggesting that WT–WT forms a tetramer, possibly via the introduced linker (Figure 1C, lane 4).
Detection of HPNhaA dimers by BN-PAGE
Identification of the dimer interface in HPNhaA
The results of a previous 2D (two-dimensional) electron crystallography analysis , in conjunction with information of the 3D (three-dimensional) crystal structure  of ECNhaA, indicate that the β-hairpin of loop1 is located at the interface between two ECNhaA monomers. An ECNhaA mutant lacking the β-hairpin of loop1 has been shown to be a functional monomer and cross-linking was observed between the monomers in the dimers by MTS cross-linkers [35,50]. This finding suggests that the β-hairpin in loop1 is important for dimerization, but not for antiporter activity . Therefore, to examine whether or not loop1 of HPNhaA is essential for dimerization, we constructed a mutant NhaA (βL1-FLAG) in which loop1 (41–67 amino acids) was completely replaced with a flexible linker peptide (GGSGG). This loop1 mutant could not form a dimer (results not shown), suggesting that loop1 in HPNhaA is important for dimerization as in ECNhaA.
Next, we constructed a series of single cysteine residue mutants within loop1, on the basis of CL-HPNhaA (C226S and C348S) (Figure 2A) , to identify the dimer interface more precisely. Among the cysteine residue mutants only the Q58C mutant was detected in dimer size (Figure 2B). The Q58C dimer disappeared after the membrane vesicles were treated with a reducing reagent (200 mM DTT), indicating that Q58C was cross-linked by a disulfide bond at the cysteine residue. These results suggest that the Gln-58 residue is located at the dimer interface. To further examine the dimer interface, we constructed a series of single cysteine residue mutants surrounding Gln-58. With the exception of Q58C, none of the single cysteine residue mutants appeared to form a dimer (Figure 3A). However, when membranes containing a single cysteine residue mutant were treated with cross-linkers (MTS-2-MTS, 5.2 Å; or o-PDM containing a spacer arm, 7.7–10.5 Å), the G56C, G60C and I59C mutants were found to form dimers (Figure 3B). These results confirm that the region surrounding Gln-58, including Gly-56, Ile-59 and Gly-60, is located at the dimer interface.
Cysteine residue substitution of loop1 residues and effects on intermolecular cross-linking
Intermolecular cross-linking of HPNhaA mutation in loop1, using bifunctional cross-linkers (i.e. MTS-2-MTS and o-PDM)
Amino acid substitutions of residues 57–59 affect dimer formation, but not antiporter activity
To examine whether the region surrounding Gln-58 is important for dimerization, we replaced the consecutive Phe-57, Gln-58 and Ile-59 residues with glycine, serine and glycine residues respectively, as shown in Figure 4(A), and named the resulting mutant L1βGL. Since amino acids harbouring a bulky side chain are known to favour β-sheet formation , we expected that these substitutions would break the β-sheet domain, and thereby prevent this mutant from adopting the dimer form. As a result, we could not detect a significant FRET signal between WT–CFP and L1βGL–Venus (results not shown). Furthermore, the L1βGL mutant was detected only in the monomer size by BN-PAGE (Figure 5). Thus these data confirm that the residues around Gln-58 are important for dimerization. The monomeric mutant created by these three substitutions was expressed at a level that was approx. one-eighth that of the WT (Figure 4B), implying that dimerization is important for either the stability and/or insertion of HPNhaA in the membrane. The pH-dependent antiporter activity of this monomeric mutant was as high as that of the WT NhaA (Figures 4C and 4D). These results show that, similarly to the monomeric ECNhaA mutant, monomeric HPNhaA is functional.
Construction and detection of L1βGL mutant NhaA and its antiporter activity
L1βGL does not form a dimer in detergent solution
Effects of intermolecular cross-linking on antiporter activity
To analyse the functional significance of loop1 interactions within the HPNhaA dimer, we further examined the effects of single amino-acid replacements on antiporter activity. The antiporter activities of each cysteine residue mutant at pH 7.5 are shown in Figure 6. The Q58C mutant exhibited significantly less Na+/H+ and Li+/H+ antiporter activity than did the control or other mutants (Figure 6), whereas the activities of the other mutants were similar to that of CL-HPNhaA. The mutants were expressed at a similar level in the membrane (Figures 2B and 3A), indicating that the reduction in the antiporter activity in the Q58C mutant of NhaA is not a result of reduced expression in the membrane. To determine whether cross-linking at the Q58C residue (Figure 3) caused a decrease in antiporter activity, we prepared membrane vesicles containing the Q58C mutant of NhaA under reducing conditions, in the presence of 1.5 mM 2-mercaptoethanol, and confirmed that very low levels of dimer were present in the membrane (Figure 7D). Surprisingly, these conditions restored Na+/H+ antiporter activity of the mutant (Figures 7A and 7B) to 60% of the activity of the positive control. As levels of 2-mercaptoethanol in excess of 10 mM inhibit the respiratory chain, we were not able to increase the concentration of the reducing reagent in the assay. Thus it is conceivable that the lower concentration of reducing reagent used in this assay reduced the disulfide bond, but did not restore the activity fully to WT levels. In contrast with Na+/H+ antiporter activity, the activity of the Li+/H+ antiporter was not completely eliminated in the Q58C mutant of NhaA.
Antiporter activity of single cysteine residue mutants
Effects of intermolecular cross-linking and point mutations at Cys-58 on transport activity
To characterize the nature of the interactions at the Gln-58 residue, we replaced Gln-58 with a serine, glutamic acid or lysine residue. We analysed the NhaA mutants by BN-PAGE (Figure 7C). The Q58E mutant was mostly present as a monomer, suggesting that replacement of the glutamine residue with the negatively charged glutamic acid residue caused dissociation of the HPNhaA dimer by repulsion of charge. This result strongly supports the notion that Gln-58 is located at the dimer interface. Other mutants were found to form dimers (Figure 7C). The antiporter activity of these mutants was at the WT level (Figures 7A and 7B), suggesting that Gln-58 itself is not essential for antiporter activity, although flexible interactions at the dimer interface involving this residue are important for activity.
In the present study, we applied BN-PAGE analysis and established that HPNhaA forms an oligomer, possibly a dimer, in detergent solution. We detected the WT form as two bands in a solution of 1% Triton X-100 solubulization buffer (Figure 1C, lane 3). The lower band was major and may be a monomer, because we observed previously that 1% Triton X-100 causes extensive dissociation of the HPNhaA oligomer . Under a milder solubilization conditions, using 0.1% DDM, almost all of the WT molecules were detected as a single band of approx. 300 kDa, corresponding to the linked dimer position in the 1% Triton X-100 treatment (WT–WT, Figure 1C, lane 2), suggesting that HPNhaA forms a dimer (Figure 1C, lane 5). The linked WT–WT appeared to be twice as large as the WT molecules in the presence of 0.1% DDM (Figure 1C, lane 4), suggesting that the linked WT–WT molecules associate with each other to form a tetramer. This might be caused by the artificial linking between the monomers, because no bands larger than the dimer were detected for the WT form. In a FRET analysis, we demonstrated previously that HPNhaA forms an oligomer  in the native membrane. Thus it is conceivable that HPNhaA forms dimers in the membranes.
ECNhaA has been shown to occur as a dimer both in detergent solution and in the native membranes . On the basis of the 3D structure of ECNhaA , cross-linking  and ESR study , it was shown that a β-hairpin in the loop1 structure participates in the dimer interaction. Rimon et al.  prepared a mutant ECNhaA that lacks the β-hairpin, and showed that this mutant of ECNhaA exists as a monomer in both the membrane and detergent micelles and retains antiporter activity . Furthermore, this region was cross-linked by bifunctional cross-linker between ECNhaA monomers, confirming that this region is located at the dimer interface of ECNhaA . Since the crystal structure of HPNhaA has not yet been determined, and the primary sequence of loop1 is not conserved between HPNhaA and ECNhaA, we first analysed the effect of replacing loop1 of HPNhaA with a short linker on dimer formation. Since this mutant of HPNhaA did not form a dimer in the membrane (results not shown), we further tried to identify the interface between the monomers within HPNhaA dimer, and asked whether or not the monomers of ECNhaA and HPNhaA interact in a similar way. For this, we conducted cysteine-scanning mutagenesis of the HPNhaA in loop1 and performed cross-linking experiments. These approaches revealed that the Q58C mutation alone caused a disulfide bond bridge between the monomers, whereas other cysteine residue replacements failed to form S–S cross-links. This result strongly suggests that the Gln-58 residues of the monomers are located in close proximity to each other in the dimer. As the G56C, I59C and G60C mutations were also able to connect the monomers by bifunctional cross-linkers, we concluded that these residues, in addition to Gln-58, participate in the interface between monomers within HPNhaA dimer. Thus we propose that similar monomer–monomer interactions take place for both HPNhaA and ECNhaA at loop1, and that loop1 of HPNhaA most probably forms a β-sheet (Figure 2).
We observed that replacing the Gln-58 residue with a cysteine residue resulted in disulfide bond formation and extensive loss of Na+/H+ exchange activity, whereas the activity was restored when the bond was reduced (Figures 7A and 7B). Although less extensive, a significant decreases in the Li+/H+ exchange was also observed. We established previously that Li+/H+ transport causes a bigger conformational change during the transport cycle than does Na+/H+ transport [30,38]. Therefore it is conceivable that the disulfide bridge between the monomer may not completely inhibit the conformational change of HPNhaA required for Li+ ion transport. Since the replacement of Gln-58 with three other amino acids (i.e. lysine, glutamic acid and serine residues) did not reduce the antiporter activity, local conformational changes resulting from these replacements do not affect antiporter activity. Taken together, we suggest that the conformational change induced by the cross-linking itself affects the antiporter activity. Figure 7(D) shows that significant amounts of monomer were present during non-reducing conditions. Since membranes prepared under non-reducing conditions did not exhibit antiporter activity, and reducing conditions restored activity, it is probable that the monomers originated during electrophoresis. The Q58E mutant did not form a dimer in detergent solution, possibly due to electric repulsion between the monomers because of the minus charge of the glutamic acid residue, and yet it had the antiporter activity. It should be noted that this mutant may be useful in future studies, such as a FRET experiment using fluorescence-labelled HPNhaA monomer, to detect intermolecular conformational changes.
In line with results obtained with ECNhaA , we have shown previously that HPNhaA activity was recovered by simultaneous expression of two independent inactive forms of HPNhaA in the same cell, indicating functional complementation of the inactive monomers . Furthermore, we showed that Li+ binding induced changes in FRET activity between two monomers tagged with a donor CFP or an acceptor Venus tag . These two observations suggest that conformational changes occur in each monomer during ion transport, and that this change is transmitted to the other monomer. Our present findings support this idea by showing that a change in the conformation at the dimer interface affects transport. It is possible that a flexible interaction, particularly at the Gln-58 residue, which is required for the antiporter activity, is present in the dimer of HPNhaA. We have also shown that the monomer form is functional, because the HPNhaA mutant that replaced three residues between residues 57 and 59 by Gly-Gly-Gly (L1βGL) still retained the activity. This observation led us to question the significance of dimer formation. We observed that only approx. 20% of the WT level of NhaA was recovered in the membrane fraction for L1βGL, suggesting that the monomer is unstable (results not shown). This assumption was supported by that the expression level of monomeric mutant HPNhaA (Q58E) was also lower than that of the other HPNhaA mutants (Figure 7C). This evidence may support an idea that dimer formation is required for the formation of a structurally stable molecule in the native membrane. The importance of dimer formation for stability of ECNhaA has been reported during preparation of the present paper .
Inhibition of transporter activity by monomeric cross-linking within a trimeric structure has been demonstrated for the multi-drug transporter AcrB . A drug-binding site exists in the monomer of AcrB . On the basis of the asymmetrical structures of AcrB monomers in the presence or absence of drug, it is thought that monomers may engage in some form of co-operation in transport cycle . In the H+-transporting F-ATPases that employ motor mechanisms, cross-linking between the catalytic β-subunit and the rotating γ-subunit is known to eliminate activity of the pump . These two examples of dynamic conformational changes of the transporter protein during the transport process might encourage our assumption that the transmission of conformational changes between two NhaA monomers affects the ability of the entire dimer to transport ions. However, in both ECNhaA and HPNhaA the monomer is a functional unit, so the effects caused by changes in the dimer interface are not essential for functionality, but may affect it indirectly.
Although the crystal structure of ECNhaA has been solved at 3.5 Å , a detailed structure of the dimer has not been reported, but a structure model has been obtained  and supported experimentally . In the present study, we show that the HPNhaA dimer might be very similar to the ECNhaA dimer. The HPNhaA dimer model will facilitate future studies that aim to characterize the functional significance of the dimer in more detail.
cyan fluorescent protein
electron spin resonance
fluorescence resonance energy transfer
green fluorescent protein
Na+/H+ antiporter A
Escherichia coli NhaA
Helicobacter pylori NhaA
Akira Karasawa conducted most of the experiments and wrote the original draft of the manuscript. Keiji Mitsui conducted some parts of the experiments and advised on the completion of the manuscript. Masafumi Matsushita provided suggestions and discussion for the completeion of the manuscript. Hiroshi Kanazaza suggested the original direction of the research, edited the manuscript and supported the project financially.
We thank Dr Atsushi Miyawaki (Riken Brain Science Institute, Wako, Japan) for providing the GFP variant (i.e. Venus). We also thank Dr Shushi Nagamori for critical comments and encouragement for the present study.
This study was supported by a Grant-in-Aid from the Japanese Ministry of Education, Science, Sports, Technology and Culture.