Unconventional Chemiosmotic Coupling of NHA2, a Mammalian Na+/H+ Antiporter, to a Plasma Membrane H+ Gradient*

Background: Inwardly directed Na+ gradients have failed to reveal transport activity of NHA2 in mammalian cells, although it functionally complements salt tolerance in yeast. Results: NHA2 mediates sodium-lithium countertransport, and cation efflux using an inwardly directed H+ gradient. Conclusion: NHA2 is chemiosmotically coupled to the V-ATPase. Significance: H+-driven salt and pH homeostasis in the kidney may be important in hypertension. Human NHA2, a newly discovered cation proton antiporter, is implicated in essential hypertension by gene linkage analysis. We show that NHA2 mediates phloretin-sensitive Na+-Li+ counter-transport (SLC) activity, an established marker for hypertension. In contrast to bacteria and fungi where H+ gradients drive uptake of metabolites, secondary transport at the plasma membrane of mammalian cells is coupled to the Na+ electrochemical gradient. Our findings challenge this paradigm by showing coupling of NHA2 and V-type H+-ATPase at the plasma membrane of kidney-derived MDCK cells, resulting in a virtual Na+ efflux pump. Thus, NHA2 functionally recapitulates an ancient shared evolutionary origin with bacterial NhaA. Although plasma membrane H+ gradients have been observed in some specialized mammalian cells, the ubiquitous tissue distribution of NHA2 suggests that H+-coupled transport is more widespread. The coexistence of Na+ and H+-driven chemiosmotic circuits has implications for salt and pH regulation in the kidney.

Recently, we identified a novel human gene, NHA2 (NHEDC2, SLC9B2), belonging to the cation/proton antiporter (CPA) 2 superfamily, located within a chromosomal region (4q24) associated with hypertension and Na ϩ -Li ϩ countertransport (SLC) activity in numerous linkage studies (1). Hypertension is a major disorder affecting 1 in 4 adults in the United States, with increased renal sodium reabsorption as an established risk factor (2). SLC activity is an established clinical marker for essential hypertension and a highly heritable trait. Since elevated SLC activity in erythrocytes from hypertensive patients correlated with increased renal reabsorption of sodium (3), it presents a potential therapeutic target in the management of hypertension. It is likely that SLC represents an alternate mode of Na ϩ -H ϩ exchange (3), however, the molecular identity of the transporter is still unknown. Of note, a role for the ubiquitous and well-characterized Na ϩ /H ϩ exchanger NHE1 has been ruled out by linkage and inhibitor studies (4).
Heterologous expression of NHA2 in yeast complemented salt-sensitive growth phenotypes and revealed characteristic features of SLC including cation selectivity for Na ϩ and Li ϩ (but not K ϩ ), insensitivity to amiloride (a hallmark of plasma membrane NHE) and sensitivity to phloretin and quinidine (1,5,6). In mammals, NHA2 is expressed in erythrocytes and a wide range of tissues, with high levels detected in the distal tubules of the kidney nephron (7). Although the distal segments of the nephron reabsorb less than 10% of the filtered load of sodium, this region responds to dietary sodium intake and is subject to hormonal control, with subsequent up-or downregulation of Na ϩ transporters and channels to fine tune renal Na ϩ reabsorption (8). Taken together, the evidence points to a potential critical role for NHA2 in salt homeostasis.
Despite the promising complementation studies in yeast, functional evidence of NHA2 transport activity in mammalian cells was lacking. Moe and co-workers (7) introduced NHA2 into mammalian cell lines devoid of endogenous NHE1 and reportedly failed to detect Na ϩ -driven H ϩ efflux, the conventional mode expected for mammalian secondary transport. Therefore, we reasoned that there must be fundamental mechanistic differences between NHA2 and NHE antiport. Phylogenetic analysis places metazoan NHA2 in a clade distinct from the NHE cluster, and more closely related in sequence to bacterial NhaA (1). In contrast to mammalian NHE1 that is powered by the inwardly directed Na ϩ gradient to extrude H ϩ from the cell, bacterial NhaA is driven by a primary H ϩ gradient to function in the direction of Na ϩ efflux (9). We noted that this is similar to prevailing chemiosmotic circuits in yeast, where we observed H ϩ -driven cation extrusion by NHA2. Furthermore, whereas NHE mediates electroneutral Na ϩ /H ϩ exchange, NhaA is electrogenic, with a 2H ϩ /Na ϩ stoichiometry (10). Finally, although both NHE and NhaA antiporters are highly pH-sensitive, their regulation is in opposite directions: thus, NHE is activated by acid pH, NhaA from Escherichia coli is "acid-locked" or inactive below pH 6.5, and demonstrates an increase in rate by three orders of magnitude between pH 7 and 8.5 (11). These differences could contribute to the failure to detect NHA2 activity in mammalian cells.
Here, we provide evidence supporting unusual functional coupling of NHA2 to a proton motive force generated by the V-type H ϩ -ATPase at the plasma membrane of mammalian cells. Thus, in contrast to the vast majority of sodium-coupled transporters, mammalian NHA2 recapitulates its phylogenetic origins with bacterial orthologs. This unique chemiosmotic coupling results in a virtual cation-H ϩ efflux pump with physiological implications for Na ϩ and pH homeostasis in the kidney.

Generating Stable MDCK Line Overexpressing GFP-HsNHA2
On day 1, MDCK cells were plated in a 6-well plate at ϳ30% confluence. After 24 h, the cells were transfected with pEGFPC2 vector carrying human NHA2 tagged to GFP at the N terminus. The following day, the transfected cells were split into 10-cm plates at different dilutions. G418 (final concentration of 200 g/ml) was added to the plates after 24 h to select for transfected cells. The concentration of the antibiotic was increased to 400 g/ml the next day. Antibiotic containing media was changed every 2 days for a week; well-isolated clones were selected and then transferred to a 24-well plate. Following selection (400 g/ml G418) for 72 h, isolated clones were moved to a 6-well plate and high expressing clone was selected by testing the GFP expression on a Western blot and also by microscopy.

Immunofluorescence
Cells were grown in 6-well plates containing glass coverslips (22 ϫ 22 mm) and probed with modifications from a previously published staining protocol (12). A monolayer of MDCK cells were pre-extracted with a solution of PHEM buffer (60 mM PIPES, 25 mM HEPES, 10 mM EGTA, and 2 mM MgCl 2 , pH 6.8) containing 0.025% saponin for 2 min. The coverslips were washed twice for 2 min with a solution of PHEM buffer containing 0.025% saponin and 8% sucrose. The cells were fixed with a solution of 4% PFA and 8% sucrose in PBS for 30 min at room temperature. Coverslips were rinsed three times quickly to rehydrate, and then three times for 5 min each with PBS. A solution of 1% BSA and 0.025% saponin in PBS was used to block for 1 h. The coverslips were probed with the following primary antibodies in the blocking solution for 1 h: A) E-cadherin (Transduction Laboratories/BD Biosciences) at 1:50 dilu-tion or B) ZO-1 1:500 (Transduction Laboratories/BD Biosciences) or C) V-ATPase a1 (Santa Cruz Biotechnology, INC.) at a dilution of 1:100 After incubation with primary antibodies, coverslips were washed three times for 5 min with 0.2% BSA in PBS. Anti-mouse and anti-rabbit Alexa Fluor 568 at dilutions of 1:1000 (Invitrogen, Carlsbad, CA) were used as secondary antibodies. For immunostaining with NHA2 antibody (1:100 dilution) raised in rabbit against a 15-aa peptide of HsNHA2 (1), a similar protocol was used. The coverslips were incubated in DAPI/PBS (5 mg/ml) for 10 s and rinsed with H 2 O and subsequently mounted with Dako Cytomation Fluorescent Mounting Medium (Invitrogen). Cells were imaged on a Zeiss 710NLO Meta confocal microscope.

RT-PCR
mRNA was isolated from control and MDCK stable cells using RNeasy Mini kit from Qiagen following manufacturer's instructions. RNA was treated with DNase I (1 unit for 1 g RNA; Roche) following which the DNA was inactivated by 25 mM EDTA (final concentration of 3 mM) at 65°C for 10 min. RNA was reverse transcribed using 0.6 g/l random primers (Invitrogen) and superscript III reverse transcriptase (200 units/l; Invitrogen) in the presence of RNaseOUT RNase inhibitor (40 units/l; Invitrogen). The reaction was inactivated by heating at 70°C for 15 min. A 450-bp region of NHA2 was amplified to detect expression and GAPDH (250 bp) was used as loading control.

Li ؉ Sensitivity Assays
A) The effect of LiCl on the growth of MDCK cell line was evaluated by MTS cell proliferation assay. Equal numbers of control and NHA2-GFP cells were plated in a 96-well plate in MEM medium, supplemented with 10% FBS and 5% antibioticantimycotic solution (referred to as complete media henceforth). Twenty-four hours after the plating, the growth medium was removed, and cells received freshly prepared complete medium (untreated controls), media containing 40 or 100 mM LiCl with no drug, 200 M amiloride or 200 M amiloride and 200 M phloretin. After 48 h of exposure, growth was monitored using a Celltiter 96 Aqueous One Solution cell proliferation assay kit (Promega, Madison, WI) according to the manufacturer's instructions. B) Control and NHA2-GFP expressing MDCK cells were plated in a 9:1 ratio on a 6-well plate in complete MEM medium. Twenty-four hours after the plating, the growth medium was removed, and cells received freshly prepared complete medium (untreated controls) or media containing 90 mM LiCl for 48 h. The cells were then washed with PBS, trypsinized, resuspended in fresh MEM media and divided into two aliquots. Flow cytometry and Western analysis were used to analyze GFP expression with the aliquots. C) Control MDCK cells were treated with 100 nM aldosterone, 1 M vasopressin or vehicle control for 4 h after which 90 mM LiCl was added to the culture media. The media was replaced by complete media without the hormones and LiCl after 24 h. The effect of LiCl on the growth of MDCK cell line was evaluated by MTS cell proliferation assay after 48 h.

Intracellular Li ؉ Measurement
Control and NHA2-GFP-expressing cells were plated in equal numbers on a 6-well plate. Culture media was replaced with media containing 90 mM LiCl. After 2.5 h, the cells were washed quickly with PBS (twice), lysed in 0.1% Nonidet P-40 lysis buffer (EDTA-free), and sonicated. The lysate was centrifuged at 14,000 rpm for 15 min to remove cell debris. Part of the supernatant was diluted in 35% nitric acid, and the lithium levels were measured by atomic absorption spectroscopy. Remaining lysate was used to measure total cellular protein concentrations (Thermo Scientific Pierce BCA Protein Assay Kit). Intracellular lithium levels were expressed as parts per million per microgram of protein. In a similar experiment, NHA2-GFP-expressing MDCK cells were first pretreated with 200 M phloretin for 10 min and then exposed to 90 mM LiCl for 2.5 h (with 200 M phloretin). Intracellular lithium levels were determined as above.

SLC Assay
The method to measure sodium stimulated lithium efflux was based on an adaptation to the technique described by Zerbini et al. and Canessa et al. in fibroblasts and erythrocytes respectively (13,14). NHA2-GFP-expressing MDCK cells in a 6-well dish were then loaded with Li ϩ by incubation in complete MEM media containing 90 mM LiCl, for 2.5 h. During the last 5 min of incubation, ouabain (100 M) and bumetanide (100 M) were added to anticipate the inhibition of the Na ϩ pump and of the Na ϩ -K ϩ cotransport, respectively. After washing in cold, choline chloride wash buffer (150 mM choline chloride, 1 mM MgCl 2 , 10 mM Tris, MOPS, pH 7.4 at 4°C), cells were incubated in efflux buffer containing 10 mM Tris-MOPS buffer (pH 7.4 at 37"C), 1 mM MgCl, 10 mM glucose, 100 M ouabain, 100 M bumetanide, 200 M DIDS (to block the activity of the anion exchanger), and either 150 mM NaCl or 150 mM choline chloride for 5 or 10 min. Parallel experiments were performed using media with amiloride and/phloretin at final concentrations of 200 M. To minimize between-assay variability, the number of culture wells was always sufficient to perform the individual comparisons between inhibitors in a single experimental set. After incubation, efflux solutions were removed and transferred to tubes to measure the Li ϩ concentration. Li ϩ was measured by atomic absorption against Li standards in double distilled water. Cells were washed two times with cold choline chloride wash buffer and lysed using 0.1% Nonidet P-40 lysis buffer to measure total cellular protein. Lithium efflux was finally expressed in parts per million per milligram of protein.

Measurement of Na ؉ (Li ؉ )/H ؉ Exchange Activity
A. Na ϩ -dependent H ϩ Efflux-The rate of Na ϩ -dependent recovery of pHi from an acid load, equivalent to the rate of Na ϩ /H ϩ exchange, was determined at the cellular level by measuring the fluorescence emission of BCECF (a pH-sensitive dye) using dual-wavelength excitation as described previously (15,16). Briefly, MDCK cells (control and NHA2-GFP cells) were grown to a confluence of ϳ80% on glass coverslips. The cells were loaded with 4.2 mM BCECF acetoxymethyl ester (BCECF/AM) for 20 min at 37°C and mounted in a specially designed perfusion chamber. The chamber was placed on the stage of fluorescence spectrophotometer (Photon Technology Quantamaster). Cells were constantly perfused using a gravitydriven system. BCECF fluorescence emission (collected at 530 nm) was measured in response to sequential excitation at two excitation wavelengths, 440 nm and 490 nm. Fluorescence data were recorded every 3.4 s by irradiating the cells for 7 ms at each of the excitation wavelengths. Data were processed and corrected for autofluorescence (measured from cells not loaded with BCECF). Calibration of the BCECF fluorescence ratio versus pH i was performed in the presence of 0.5 mM of the K ϩ /H ϩ ionophore nigericin (in "K-clamp media" of varying pH values, see below) according to a protocol described previously (15)(16)(17). For study of Na ϩ /H ϩ exchange activity cells were pulsed with NH 4 Cl (40 mM NH 4 Cl in TMA-medium, for 20 min). Removal of NH 4 Cl and perfusion with "TMA-medium" resulted in acidification of the cells. Acid-loaded cells were then exposed to Na ϩ ("Na medium" see later) and the initial rate of pHi recovery was used for quantification of Na ϩ /H ϩ exchange activity. NHE agonist, HOE 694, was introduced during perfusion with Na-medium and the initial rate of pH i recovery was compared with control experiments without the agonist. Concentration of agonist (50 mM) was chosen to be the same, as determined previously to elicit maximal or near-maximal responses. It has been previously established that MDCK cells tolerated at least three consecutive NH 4 Cl pulses without a major fatigue effect on plasma membrane Na ϩ /H ϩ exchange activity (16). Calibration of the BCECF fluorescence ratio versus pH i was performed and the rate of pH i recovery were calculated as described above in (A).

Acid Suicide
The method to measure lithium-stimulated proton influx was based on an adaptation to the technique described by Pouyssegur et al. (18). MDCK cells were first incubated for 2 h in 90 mM LiCI buffer (pH 7.4) at 37°C in the absence of CO 2 (18). These Li ϩ -loaded cells were then washed twice with 90 mM choline chloride (pH 7.4), and transferred at 37°C to 90 mM choline chloride acid solution (pH 5.5) for 1 h. For acid suicide study in the presence of NHE and NHA2 agonists, a similar protocol was adopted, but the cells were pretreated with 200 M amiloride or phloretin treated for 10 min. The cells were then loaded with 90 mM Li and moved choline chloride solution as before. Amiloride or phloretin were also present during the lithium loading and in the choline chloride solution. Cells surviving acid suicide were then incubated in complete MEM media overnight at 37°C (with 5% CO 2 ), and cell survival was measured by MTS assay.

Bafilomycin Sensitivity
Equal numbers of control and NHA2-GFP cells were plated in a 96-well plate in complete MEM media. Twenty-four hours after the plating, the growth medium was removed, and cells received freshly prepared complete medium (untreated controls) or media containing 100 mM LiCl with no drug or 5 nM bafilomycin. After 48 h of exposure, growth was monitored using MTS assay.

Coimmunoprecipitation
Coimmunoprecipitation in stable MDCK cells was performed 72 h after seeding cells. Cells were lysed in Nonidet P-40 lysis buffer. Cell lysate was incubated 4 h with GammaBind Plus Sepharose (GE Healthcare, Waukesha, WI) for preclearance and 3 h with anti-NHA2 and anti-VATPase a1 antibodies (2 g/0.5 ml lysate) at 4°C. IgG, nonspecific antibody was added to the negative control samples. Beads were washed using lysis buffer containing 1% Nonidet P-40 before SDS-PAGE and immunoblotting. The blots were probed with anti-NHA2, anti-VATPase a1, or anti-VATPase B1/B2 antibodies used at 1:100 dilution. Clean-blot IP detection reagent (Thermoscientific) was used to detect the primary antibodies on the Western blots without interference from denatured IgG.

Statistical Analysis
All plotted data were replicated at least three times independently and the mean Ϯ S.D. is shown. Statistical significance was assessed using the Student's t test, and p Ͻ 0.05 considered significant.
Using heterologous expression in yeast, we had previously shown that NHA2 conferred robust growth resistance to Li ϩ (and Na ϩ ) toxicity (1). Here, we asked whether NHA2 conferred salt tolerance in mammalian cells as well. We used Li ϩ as a surrogate for Na ϩ ions because it is toxic at lower concentrations, thus avoiding secondary osmotic effects of high salt. We showed that NHA2-GFP increased cell survival in media supplemented with 40 or 100 mM LiCl, relative to control MDCK cells ( Fig. 2A). Growth was insensitive to amiloride (Fig. 2B), a well-known inhibitor of the NHE family. Previously, we had shown that salt tolerance conferred by NHA2 in yeast was sensitive to phloretin, an inhibitor of sodium lithium countertransport. Similarly, phloretin reversed the growth advantage conferred by NHA2 in 40 mM LiCl (Fig. 2C). Although phloretin was only mildly toxic to cells in medium lacking LiCl, it was lethal in the presence of 100 mM LiCl, suggesting a requirement for NHA2 function under these conditions (Fig. 2C).
These results were surprising because conventional NHElike cation/H ϩ exchange activity was expected to equilibrate Li ϩ across the membrane, whereas growth advantage is indicative of active cation efflux. To confirm these findings, we followed the growth of mixed cultures of NHA2-GFP-expressing cells and control cells seeded in a 1:9 ratio. After 48 h, these ratios remained relatively similar in standard medium, but in the presence of 90 mM LiCl, NHA2-GFP cells were enriched by nearly 3-fold at the expense of control cells, as determined by flow cytometry (Fig. 2D) and Western analysis (inset). Furthermore, mean fluorescence intensity of GFP-positive cells increased (from 22,613 Ϯ 1,205 to 31,955 Ϯ 781 a.u.) upon LiCl treatment (Fig. 2D). Consistent with these findings, NHA2-GFP cells were found to retain correspondingly less (ϳ3.3-fold) intracellular Li ϩ relative to control cells following 2 h exposure to LiCl (Fig. 2E). Again, treatment with phloretin reversed this difference (Fig. 2F), as expected for NHA2 inhibition. Taken together, these results show that NHA2 functions in effect as a phloretin-sensitive Li ϩ efflux pump.
NHA2 Has SLC Activity-Elevated SLC activity is a secondary metabolic indication of an underlying cell membrane abnormality associated with essential hypertension (21). The exchange of intracellular Li ϩ for Na ϩ has been extensively documented in erythrocytes and fibroblasts (13,14). We moni-

NHA2 Is Coupled to Plasma Membrane V-ATPase
tored Li ϩ efflux in cells stably transfected with NHA2-GFP and preloaded with Li ϩ for 2.5h, as described earlier (14). We show that time-dependent Li ϩ efflux is 3 times higher in the presence of external Na ϩ , compared with equivalent concentrations of choline chloride (Fig. 3A). Thus, NHA2 mediates Li efflux ϩ / Na influx ϩ exchange as illustrated in Fig. 3D. Li ϩ efflux in Na ϩ  OCTOBER 19, 2012 • VOLUME 287 • NUMBER 43

NHA2 Is Coupled to Plasma Membrane V-ATPase
containing medium was not significantly affected by the addition of amiloride but was inhibited by phloretin (Fig. 3, B-D). Thus, NHA2 mediates SLC activity (Fig. 3D) with inhibitor characteristics previously reported in red blood cells, as a marker of hypertension. NHA2 Is Inactivated by Intracellular Acidification and Does Not Mediate Cation-driven H ϩ Efflux-Plasma membrane NHE1 transports H ϩ out of the cell, in exchange for Na ϩ influx (9,22). Indeed, this is the conventional direction for active secondary transport in all metazoan cells, powered by an inwardly directed Na ϩ gradient. We attempted to measure similar transport by NHA2-GFP in MDCK cells loaded with H ϩ and in the presence of external Na ϩ . Following acute acid load using NH 4 Cl prepulse, Na ϩ -dependent intracellular pH (pH i ) recovery was monitored fluorometrically using the pH-sensitive dye BCECF. Despite the favorable ion gradients in both directions (H efflux ϩ /Na influx ϩ ), there was no significant difference in rates of pH i recovery (i.e. H ϩ efflux) between control and NHA2-GFP-overexpressing cells (Fig. 4, A and C; 0.14 Ϯ 0.0312 and 0.16 Ϯ 0.0282 ⌬pH units/minute, respectively). HOE694, a known inhibitor of NHE1, inhibited pH i recovery completely in both control and NHA2-GFP-overexpressing cells, indicating that pH i recovery was solely due to NHE activity (Fig. 4, B and  D). We considered the possibility that NHA2 was electrogenic, exchanging 2H ϩ /Na ϩ like E. coli NhaA (11). However, there was no significant pH i recovery upon addition of K ϩ /valinomycin to relieve membrane potential differences in HOE694treated cells (Fig. 4, E and F). In parallel experiments, we monitored Li ϩ influx under similar acid-loaded conditions (H ϩ efflux /Li ϩ influx ) and failed to see a significant increase in Li ϩ entry in NHA2-GFP cells, relative to control (Fig. 3E). Thus, NHA2 does not show NHE1-like transport activity. Furthermore, these results suggest a model in which NHA2 is inhibited by cytoplasmic acidification and operates at a pH range distinct from NHE (Fig. 3F).
NHA2 Mediates Cation Efflux with an Inwardly Directed pH Gradient-Next, we investigated H ϩ -dependent cation efflux, which is the physiologically reverse mode of plasma membrane NHE. Following preloading with Li ϩ at pH 7.4, NHA2-GFPexpressing cells were placed in Na ϩ -free buffer at pH 7.4 or pH 5.5. We show that Li ϩ efflux within the first 10 min was ϳ4-fold higher with an inwardly directed H ϩ gradient, and was sensitive to phloretin (Fig. 3G). It is worth noting that this mode of antiport (Li efflux ϩ /H influx ϩ ; Fig. 3H) is similar to the physiological mode of bacterial NhaA orthologs and recapitulates our previous results from heterologous expression of NHA2 in yeast (1). In addition, H ϩ -driven Li ϩ efflux could potentially explain the gain in Li ϩ tolerant growth conferred by NHA2-GFP demonstrated in Fig. 2, A and D. NHA2 Confers Tolerance to Acid Suicide-To investigate the difference between NHA2 and NHE further we subjected control and NHA2-GFP cells to an acid suicide protocol, originally described by Pouyssegur et al. to isolate cells lacking plasma membrane NHE function (18). In this scenario, cells are loaded with Li ϩ and placed in a low pH environment (Fig. 5A). Cell death would be consistent with Li efflux ϩ /H influx ϩ exchange activity. Surprisingly, NHA2 conferred tolerance to acid suicide. When Li ϩ -loaded cells were placed in pH 5.5 buffer fol-lowed by outgrowth in normal medium, cell survival was 2-fold greater in the stable line expressing NHA2-GFP, relative to control MDCK cells (Fig. 5B). Given the results from Fig. 2E, we concluded that reduced accumulation of Li ϩ in NHA2-GFP cells conferred protection against acid suicide. Addition of amiloride prior to external acidification also protected both cell lines, consistent with a role for NHE1 in mediating acid suicide as described by Pouyssegur et al. (Fig. 5C). In contrast, phloretin was lethal to both cell lines under these conditions (Fig. 5C). These experiments suggest that NHA2 protects against cytosolic acidification whereas NHE1 promotes it. Thus two subtypes of cation/proton antiporter within the same superfamily exhibit strikingly opposite behavior.
NHA2 Is Coupled with the V-ATPase to Function as a Virtual Cation Efflux Pump-Our findings can be explained by a model in which active cation efflux by NHA2 is coupled to a primary H ϩ gradient at the plasma membrane. A candidate H ϩ pump in mammalian cells is the lyso/endosomal V-type H ϩ -ATPase that has been observed on the plasma membrane in cells specialized to secrete protons, including osteoclasts and the intercalated cells of the kidney collecting duct (23,24). To test this hypothesis, we investigated the effect of bafilomycin, a specific V-ATPase inhibitor, on the Li ϩ tolerant phenotype conferred  Ouabain, bumetanide, DIDS, and amiloride were also added as described under "Experimental Procedures" during the last 10 min of lithium loading. The cells were then perfused with choline chloride wash buffer for 200 s following which the buffer was switched to an acidic efflux buffer (pH 5.5). Calibration of the BCECF fluorescence ratio versus pH i was performed, and ⌬pH (red dotted line) was calculated as described under "Experimental Procedures." D, same as A except that Bafilomycin was added along with ouabain, bumetanide, DIDS, and amiloride. E, MDCK cells were loaded with 90 mM Li ϩ for 2.5 h. BCECF (4.2 mM) was added in the last 20 min of the loading. Bafilomycin, ouabain, bumetanide, DIDS, and amiloride were also added as described under "Experimental Procedures" during the last 10 min of lithium loading. The cells were then perfused with choline chloride wash buffer for 200 s following which the buffer was switched to an acidic efflux buffer (pH 5.5). Calibration of the BCECF fluorescence ratio versus pH i was performed, and ⌬pH was calculated as described under "Experimental Procedures." F, same as E except that the cells were also treated with 200 M phloretin, beginning during the last 10 min of Li ϩ loading. Mean Ϯ S.D. is shown for all data. OCTOBER 19, 2012 • VOLUME 287 • NUMBER 43 by NHA2-GFP overexpression. Bafilomycin, at low concentrations, was not significantly toxic to MDCK cells in the absence of LiCl. However, bafilomycin completely abolished Li ϩ tolerance conferred by NHA2-GFP overexpression, suggesting a critical role for the V-ATPase in NHA2 function (Fig. 6A). Therefore, we proposed a model in which the proton motive force generated by the V-ATPase drives cation efflux by NHA2 resulting in a combined H ϩ -cation efflux mechanism (Fig. 6B). Consistent with this model, bafilomycin lowered cytoplasmic pH and increased ⌬pH (from 0.42 Ϯ 0.02 to 0.57 Ϯ 0.03; Fig. 6, C and D) in Li ϩ -loaded NHA2 cells subjected to an inwardly directed H ϩ gradient. Also as expected, in MDCK cells that do not overexpress NHA2 and therefore accumulate higher levels of Li ϩ upon loading, even larger ⌬pH was observed upon extracellular acidification, in the presence of bafilomycin (0.93 Ϯ 0.24). Treatment with phloretin decreased cytoplasmic acidifi- cation and ⌬pH, consistent with inhibition of NHA2-mediated Li ϩ efflux /H ϩ influx exchange (Fig. 6, E and F). In further support of this model of H ϩ driven cation efflux, we show expression and plasma membrane localization of the V-ATPase in MDCK cells (Fig. 7, A-C; a1 subunit). Significantly, V-ATPase was found to co-localize with NHA2-GFP on lateral domains of these cells (Fig. 7, A and C), supporting a functional interaction. Finally, we explored the possibility of association between NHA2 and V-ATPase, at a molecular level. In stably-transfected MDCK cells, V-ATPase robustly coimmunoprecipitated with NHA2 and vice versa (Fig. 7, D and E), consistent with a stable physical interaction between the H ϩ pump and NHA2 antiporter.

DISCUSSION
There exists a remarkable multiplicity of genes encoding cation/proton antiporters in mammals. In addition to nine NHE (SLC9A1-9) isoforms belonging to the CPA1 clade, two NHA isoforms (SLC9B1-2) were recently added to the CPA2 clade (25). Of note, all plasma membrane NHE isoforms (NHE1-5) mediate proton efflux powered by an inwardly directed Na ϩ gradient established by the ubiquitous Na ϩ /K ϩ -ATPase. However, CPA2 family members, represented in metazoan genomes by two genes (NHA1-2), have remained poorly characterized due to lack of functional evidence for transport activity in mammalian cells. Specific efforts by our group and others to uncover Na ϩ -driven H ϩ efflux from acid-loaded cells failed to reveal transport activity of NHA2 (7). These failures can now be explained by our findings showing that, (i) NHA2 is acid-inhibited, in contrast to the NHE antiporters known to be acid-activated, and (ii) NHA2 is chemiosmotically coupled to H ϩ rather than Na ϩ electrochemical gradients. Thus, NHA2 recapitulates shared phylogenetic origins with distant bacterial NhaA orthologs even in the native mammalian environment. This is consistent with our previous results from mutational analysis in the yeast model, suggesting that common transport and regulation characteristics are preserved within sequence similarities (26).
Our findings are significant and paradigm shifting because it is widely accepted that the vast majority of secondary transport at the plasma membrane of mammalian cells is coupled to the Na ϩ electrochemical gradient. This is in contrast to bacteria, fungi, protozoans, and plants, where H ϩ gradients drive secondary uptake of solutes and metabolites. H ϩ gradients have thus far only been characterized in some specialized cells such as phagocytes, osteoclasts, and kidney-intercalated cells where the V-type H ϩ -ATPase has been shown to be recruited to the plasma membrane (24). Given the ubiquitous tissue distribution of NHA2, our findings suggest that H ϩ -coupled transport may be more widespread in the plasma membranes of metazoan cells. The V-ATPase is also a ubiquitously expressed pump, and although best known for lysosomal and endosomal acidification, our results implicate an additional general role in driving secondary transport at the plasma membrane of mammalian cells. In the kidney, under physiological conditions, microdomains of pH gradients may exist near the transporter. For example, acidic microclimate in the small intestine drives the H ϩ -coupled/pH-dependent transporters like PepT1, PCFT, PAT1, MCT1, and OATP2B1 (27). This is also seen in renal tubules with PepT2 and MATE1 transporters (27). In addition, functional coupling with other transporters is also known to provide pH gradients for H ϩ -coupled transport both in the small intestine and kidney (27). Diseased conditions such as metabolic or respiratory acidosis may also cause global pH changes that may provide the gradient necessary to drive H ϩ -coupled transport.
It is noteworthy that in insect epithelia, the electrogenic V-ATPase establishes a membrane potential that drives sensory signaling, fluid secretion, and alkalization (24). In the goblet cell apical membrane of the insect Manduca sexta, there is precedence for a hybrid of the V-ATPase and a K ϩ /2H ϩ antiporter functioning as a K ϩ pump (28). We suggest that functional coupling of NHA2 and V-ATPase establishes a mechanistically novel Na ϩ (Li ϩ ) efflux pump in mammalian cells. Cultured MDCK cells retain properties of intercalated cells of the distal nephron (29), where V-ATPase localization is known to alternate between apical and basolateral membranes. It is likely that NHA2 localization is also coupled with that of V-ATPase resulting in a virtual cation-H ϩ efflux mechanism to regulate Na ϩ homeostasis, volume regulation and consequently, hypertension.