The carrier-mediated, electroneutral exchange of Na+ for H+ across the plasma membrane does not directly consume metabolic energy. Nevertheless, acute depletion of cellular ATP markedly decreases transport. We analyzed the possible involvement of polyphosphoinositides in the metabolic regulation of NHE1, the ubiquitous isoform of the Na+/H+ exchanger. Depletion of ATP was accompanied by a marked reduction of plasmalemmal phosphatidylinositol 4,5-bisphosphate (PIP2) content. Moreover, sequestration or hydrolysis of plasmalemmal PIP2, in the absence of ATP depletion, was associated with profound inhibition of NHE1 activity. Examination of the primary structure of the COOH-terminal domain of NHE1 revealed two potential PIP2-binding motifs. Fusion proteins encoding these motifs bound PIP2 in vitro. When transfected into antiport-deficient cells, mutant forms of NHE1 lacking the putative PIP2-binding domains had greatly reduced transport capability, implying that association with PIP2 is required for optimal activity. These findings suggest that NHE1 activity is modulated by phosphoinositides and that the inhibitory effect of ATP depletion may be attributable, at least in part, to the accompanying net dephosphorylation of PIP2.
Na+/H+ exchangers (NHEs) are a family of electroneutral antiporters that play an essential role in the regulation of the intracellular pH (pHi) and cell volume, and also mediate transepithelial Na+ and HCO3− absorption (for reviews see Orlowski and Grinstein 1997; Aharonovitz and Grinstein 1999). Fluxes through the NHE are driven solely by the combined chemical gradients of Na+ and H+ and, hence, do not directly consume metabolic energy (Kinsella and Aronson 1980). Nevertheless, the presence of physiological levels of ATP is required for optimal Na+/H+ exchange in all cases studied to date (Cassel et al. 1986; Brown et al. 1991; Kapus et al. 1994; Levine et al. 1993). Procedures that reduce intracellular ATP drastically inhibit Na+/H+ exchange in a variety of native systems (Cassel et al. 1986; Brown et al. 1991), as well as in antiport-deficient cells transfected with either NHE1, 2, or 3 (Levine et al. 1993; Kapus et al. 1994). Metabolic depletion depresses the rate of transport, at least in some instances, by reducing the affinity of the exchangers for intracellular H+, without altering the number of plasmalemmal transporters.
The pronounced inhibition of exchange activity induced by metabolic depletion is not accompanied by detectable alterations in the phosphorylation of the antiporter (Goss et al. 1994). Moreover, in the case of the ubiquitous isoform NHE1, the sensitivity to ATP persists after elimination of virtually all the putative phosphorylation sites by mutagenesis (Goss et al. 1994; Wakabayashi et al. 1994). Comparable studies have not been reported for other isoforms, but NHE3 remains sensitive to ATP even after truncation of a large part of its cytosolic domain, where most of the phosphorylation sites reside (Cabado et al. 1996). In view of these findings, direct phosphorylation appears unlikely to account for the ATP dependence of the exchangers and alternative mechanisms must be considered.
Associated proteins or lipids are likely to play a role in the control of NHE activity. Such ancillary components may themselves be subject to phosphorylation, perhaps accounting for the observed ATP dependence of Na+/H+ exchange. Indeed, a variety of proteins have been proposed to bind to NHE1, including calmodulin (Bertrand et al. 1994), HSP70 (Silva et al. 1995), and CHP, a calcineurin homologue (Lin and Barber 1996). Similarly, NHERF-1 and 2 have been shown to interact directly with NHE3 (Weinman et al. 1993; Yun et al. 1997). While direct phosphorylation of NHERF by protein kinase A was initially postulated to account for its effects on the activity of NHE3 (Weinman et al. 1993), this conclusion was subsequently revised (Zizak et al. 1999). Therefore, to date, there is no conclusive evidence that protein phosphorylation is responsible for the ATP sensitivity of the NHE isoforms.
Polyphosphoinositides are ubiquitous constituents of animal plasma membranes, where they have been found to exert modulatory effects on the activity of several ion transporters. Thus, optimal activity of K+ channels (Hilgemann and Ball 1996; Baukrowitz et al. 1998) and Na+/Ca2+ antiporters (Hilgemann and Ball 1996) was found to require the presence of phosphatidylinositol-4,5-bisphosphate (PIP2). Because phosphorylation of polyphosphoinositides is in a dynamic equilibrium, depletion of cellular ATP is anticipated to favor net dephosphorylation of inositol phospholipids, reducing the concentration of the most highly phosphorylated species, particularly PIP2. Therefore, the ATP sensitivity of transporters, including NHE, could be attributable to alterations in the cellular content of PIP2 (Hilgemann and Ball 1996; Baukrowitz et al. 1998).
Binding of channels and exchangers to PIP2 is thought to be mediated, at least in some cases, by a characteristic motif comprised of cationic and hydrophobic amino acids (Huang et al. 1998). This linear sequence was initially identified in actin-binding proteins such as gelsolin and profilin, which are themselves regulated by polyphosphoinositides (Yu et al. 1992). Interestingly, two related motifs can be discerned in the cytosolic domain of NHE1 (in rats, residues 513–520 and 556–564). This observation raised the possibility that NHE1 may interact physically and functionally with phosphoinositides. The purpose of the experiments described in this report was threefold: (1) to determine if alterations in the cellular content of PIP2 are associated with the inhibition of NHE1 observed upon ATP depletion; (2) to assess whether the cytosolic domain of NHE1 can specifically bind to PIP2; and (3) to define whether such binding is required for optimal NHE1 activity.
Materials and Methods
Materials and Solutions
Nigericin, 2′,7′-bis(2-carboxyethyl)-5(6)-carboxyfluorescein (BCECF) acetoxymethylester (AM), Fura-2-AM, and rhodamine-phalloidin were obtained from Molecular Probes, Inc. Chymotrypsin, N-tosyl-l-phenylalanine chloromethyl ketone (TPCK), 2-deoxy-d-glucose, antimycin A, sulfinpyrazone, PIP2, and the substrate for detection of peroxidase activity (Sigma Fast OPD) were from Sigma Chemical Co. Paraformaldehyde was purchased from Canemco Inc., and Triton X-100 was from Bio-Rad Laboratories. Mouse mAbs to PIP2 were obtained from PerSeptive Biosystems, Inc. Mouse mAbs to influenza virus hemagglutinin (HA) peptide were obtained from BAbCo. HRP-coupled goat anti–mouse antibodies were from Jackson ImmunoResearch Laboratories. Enhanced chemiluminescence reagents were from Amersham International. Fatty acid–free BSA and the ATP assay kit were purchased from Calbiochem. A peptide corresponding to amino acids 550–564 of NHE1 was synthesized by Sheldon Biotechnology Center (McGill University, Montreal, Quebec). All other chemicals were of analytical grade and were obtained from Aldrich Chemical Co.
The isotonic Na+-rich medium contained (in mM): 140 NaCl, 5 KCl, 1 CaCl2, 1 MgCl2, 10 glucose, and 10 Hepes-Na+, pH 7.4. Isotonic Na+-free medium contained (in mM): 140 KCl, 1 CaCl2, 1 MgSO4, 5.5 glucose, and 25 N-methyl-d-glucammonium (NMG)-Hepes, pH 7.4. PBS contained (in mM): 150 NaCl, 10 KCl, 8 sodium phosphate, and 2 potassium phosphate, pH 7.4. TBS contained (in mM): 150 NaCl and 20 Tris-HCl, pH 7.4.
cDNA Construction and Transfection
A vector for expression of a chimeric protein consisting of the pleckstrin homology (PH) domain of phospholipase Cδ (PLCδ and enhanced green fluorescent protein [GFP]), termed PHPLCδ-GFP, was constructed as reported earlier (Varnai and Balla 1998). PHPLCδ-GFP cDNA was transfected into AP-1 cells stably expressing wild-type rat NHE1′HA (AP-1/NHE1′HA; described below) by the calcium phosphate coprecipitation method of Chen and Okayama 1988. A vector for expression of the myristoylated and palmitoylated form of Inp54p, a yeast PIP2-specific 5′-phosphatase, fused to GFP and, termed PM-5′-phosphatase-GFP, was constructed as reported earlier (Raucher et al. 2000). PM-5′-phosphatase-GFP and PHPLCδ-GFP cDNA was transfected into COS-1 cells using FuGENE™ 6 transfection reagent (Roche Diagnostics Corp., Indianapolis, IN).
The rat NHE1 cDNA, engineered to contain a series of unique restriction endonuclease sites for subcloning purposes, was inserted into a mammalian expression vector under the control of the enhancer/promoter region from the immediate early gene of human cytomegalovirus (plasmid called pNHE1′), as previously described (Orlowski and Kandasamy 1996). To facilitate immunological detection of the protein, the influenza virus HA epitope YPYDVPDYAS, preceded by a single G amino acid linker (added to create peptide flexibility), was inserted at the very COOH-terminal end of NHE1′ using the PCR. In control experiments, these modifications had no obvious effect on the basal activity or functional properties of NHE1 (called NHE1′HA) when expressed in AP-1 cells (described below). Individual and combined mutations of two putative PIP2-binding motifs (termed M1 and M2) in NHE1′HA were accomplished using a commercially available, PCR mutagenesis procedure (QuikChange site-directed mutagenesis kit, Stratagene). Clusters of positively charged residues in the M1 (513KKKQETKR520) and the M2 (556RFNKKYVKK564) motifs were substituted with alanine, i.e., 513AAAQETAA520 and 556AFNAAYVAA564, respectively. The cDNAs were sequenced to confirm the presence of the mutations and to ensure that other random mutations were not introduced.
COS-1 cells were obtained from American Type Culture Collection. WT5 is a subline of wild-type CHO cells. AP-1, a cell line devoid of endogenous Na+/H+ exchange activity, was isolated from WT5 cells as previously described (Rotin and Grinstein 1989). AP-1 cells were transfected with plasmids containing the wild-type and mutant NHE1′HA constructs by the calcium phosphate–DNA coprecipitation technique of Chen and Okayama 1988. Starting 48 h after transfection, the AP-1 cells were selected for survival in response to repeated (5–6 times over a 2-wk period) acute NH4Cl-induced acid loads (Orlowski 1993) to discriminate between NHE-positive and negative transfectants.
Na+-induced changes of pHi were measured fluorimetrically using BCECF, essentially as described (Grinstein et al. 1992). In brief, cells grown to 60–70% confluence on glass coverslips were loaded with 2 μg/ml of BCECF-AM for 10 min at 37°C in Na+-rich medium. Two methods were used to measure the fluorescence of BCECF. For population measurements, the coverslip was placed in a thermostated Leiden holder on the stage of a Nikon TMD-Diaphot microscope equipped with a Nikon Fluor oil immersion objective and Hoffman modulation optics. A chopping mirror was used to direct the excitation light alternately to two excitation filters (490BP10 and 440BP10 nm) in front of a xenon lamp. To minimize dye bleaching and photodynamic damage, neutral density filters were used to reduce the intensity of the excitation light reaching the cells. The excitation light was directed to the cells via a 510-nm dichroic mirror, and fluorescence emission was collected through a 535BP25 nm filter. Photometric data were acquired at 10 Hz using Oscar software (Photon Technologies Inc.).
The fluorescence of transiently transfected single cells was measured using a ratio imaging system controlled by the Metafluor software (Universal Imaging), essentially as previously described (Gan et al. 1998). Transfected cells were identified by detecting GFP fluorescence before loading with BCECF. A neutral density filter was interposed in the excitation pathway, to decrease the signal emanating from GFP, and the cells were loaded with BCECF while on the microscope stage. The fluorescence of BCECF greatly exceeded that of GFP and was readily visible in the presence of the neutral density filter.
An acute acid load was imposed by preincubating the cells with 20 mM (NH4)2SO4 for 10 min, followed by five rapid washes in NH4+- and Na+-free solution. The rate of pHi recovery was measured upon reintroduction of isotonic Na+-rich medium. Calibration of the fluorescence ratio versus the pH was performed for each experiment using nigericin- and K+-rich solutions of varying pH, as before (Kapus et al. 1994).
Cytosolic Free Calcium ([Ca2+]i) Measurements
To measure [Ca2+]i, AP-1/NHE1′HA cells were loaded with 1 μg/ml of Fura-2-AM for 15 min at 37°C in Na+-rich medium containing 500 μM sulfinpyrazone. The fluorescence of Fura-2 was measured essentially as described above for BCECF, but using dual excitation with 340BP10- and 380BP10-nm filters, a 400-nm dichroic mirror and a 510BP25-nm emission filter. Unless indicated otherwise, all measurements of [Ca2+]i were made in Na+-rich medium without calcium.
To assess the formation of stress fibers, AP-1/NHE1′HA cells were plated onto coverslips and grown to 60–70% confluence. After washing three times with PBS, the cells were fixed for 40 min at room temperature using 4% paraformaldehyde in cold PBS containing 1 mM MgCl2. After fixation, cells were washed twice with PBS and incubated with 100 mM glycine in PBS for 15 min. The coverslips were washed twice again, and the cells were permeabilized with 0.1% Triton X-100 in PBS for 15 min at room temperature. After washing three more times with PBS, the cells were incubated with a 1:500 dilution of rhodamine-phalloidin for 30 min. The coverslips were finally washed and mounted onto glass slides using DAKO mounting medium.
To estimate the amount of PIP2 at the plasma membrane, AP-1/NHE1′HA cells were transfected with the PHPLCδ-GFP plasmid. After 48 h, the cells were treated as specified, immediately fixed and mounted as above. The localization of the fluorescent probe was examined with an LSM510 laser confocal microscope (Zeiss Inc.) using a 100× oil immersion objective. Digital images were analyzed using NIH Image software. To quantify fluorescence, regions of the cytosol or plasma membrane were enclosed by boxes and, after background subtraction, the intensity per unit area was determined. The ratio of plasmalemmal to cytosolic fluorescence is presented. In those cases where no defined plasmalemmal staining was discernible, the outermost layer of cytosolic fluorescence was used for the calculation, yielding a minimum ratio of ∼1.
22Na+ Influx Measurements
The cells were grown to confluence in 24-well plates. To examine the activity of the NHE as a function of the intracellular H+ concentration, the pHi was clamped at different concentrations over the range of 5.4–7.4 by suspending the cells in media of varying K+ concentrations containing the K+/H+ exchange ionophore nigericin (Thomas et al. 1979). The confluent monolayers were washed with isotonic NMG-chloride solution (140 mM NMG-chloride, 1 mM MgCl2, 2 mM CaCl2, 5 mM glucose, 10 mM Hepes, pH 7.4), and incubated for 4 min at 22°C in NMG-balanced salt solutions specific for each pHi (pH-clamp solutions). All solutions contained 2 mM NaCl, 1 mM MgCl2, 10 mM Hepes, pH 7.4, varying concentrations of K+, and 10 μM nigericin. Additional components were included for the specified pHi values as follows (in mM): for pHi 7.4, 14 KCl, 126 potassium glutamate, 24 NMG-methane sulfonate; for pHi 7.0, 14 KCl, 42 potassium glutamate, 108 NMG-methane sulfonate; for pHi 6.6, 14 KCl, 8 potassium glutamate, 142 NMG-methane sulfonate; for pHi 6.2, 8.8 KCl, 155 NMG-methane sulfonate; for pHi 5.8, 3.6 KCl, 160 NMG-methane sulfonate; and for pHi 5.4, 1.4 KCl, 163 NMG-methane sulfonate. Because the rate of ATP consumption is likely to vary with pH, the ATP depletion procedure was initiated before acid loading. 22Na+ influx measurements were performed in the same K+-nigericin solutions supplemented with 22Na+ (1 μCi/ml), 1 mM ouabain, and 0.1 mM bumetanide in the absence or presence of 2 mM amiloride. 22Na+ influx was linear with time for at least 10 min under these experimental conditions. Measurements of 22Na+ influx specific to the NHE were determined as the difference between the initial rates of H+i-activated 22Na+ influx in the absence and presence of 2 mM amiloride and expressed as amiloride-inhibitable 22Na+ influx.
The influx of 22Na+ was terminated by rapidly washing the cells three times with 4 vol of ice-cold NaCl stop solution (130 mM NaCl, 1 mM MgCl2, 2 mM CaCl2, 20 mM Hepes, pH 7.4). To extract the isotope, the monolayers were solubilized with 0.25 ml of 0.5 N NaOH, and the wells were washed with 0.25 ml of 0.5 N HCl. Both the NaOH cell extract and the HCl wash solution were combined in 5 ml scintillation fluid and transferred to scintillation vials. The radioactivity was assayed by liquid scintillation spectroscopy. Protein content was determined using the Bio-Rad DC protein assay kit according to the manufacturer's protocol. Each data point represents the average of at least three experiments, each performed in quadruplicate.
Depletion and Determination of ATP
Depletion of cellular ATP was carried out as described by Goss et al. 1994. In brief, AP-1/NHE1′HA cells were incubated for 10 min in a medium containing 100 mM K+-glutamate, 30 mM KCl, 10 mM NaCl, 1 mM MgCl2, 10 mM Hepes, pH 7.4, 5 mM 2-deoxy-d-glucose and 1 μg/ml antimycin A. This high K+, low Na+, and nominally Ca2+-free medium was used to prevent Na+ and/or Ca2+ loading of the cells upon inhibition of the Na+/K+ or Ca2+ pumps. Substitution of most of the Cl− by glutamate− was intended to minimize cell swelling. Cellular ATP content was determined using the Calbiochem assay kit. Cells (∼5 × 105) were extracted with 0.4 ml of 8% perchloric acid and placed on ice. The extract was neutralized with 1 M K2CO3, debris were sedimented, and aliquots of the supernatant (5 μl) were mixed with the buffer and luciferin–luciferase mixture provided by the kit's manufacturer. Sample luminescence was determined using a Beckman LS7000 counter and compared with ATP standards.
Quantification of Surface NHE1
We took advantage of the proteolytic susceptibility of NHE1 to quantify the fraction of the antiporters exposed at the cell surface, based on the earlier observations of Shrode et al. 1998. In brief, chymotryptic cleavage sites have been identified in the first and/or second putative extracellular loops of NHE1. When the protease is added to intact cells, cleavage at these sites can only be accomplished when NHE1 is inserted in the plasmalemma. This results in the disappearance of the fully glycosylated ∼105-kD band, corresponding to NHE1 in immunoblots. By contrast, the core-glycosylated immature form of NHE1 (∼80 kD), which resides in endomembranes, is unaffected by the protease, since under the conditions used chymotrypsin does not enter the cells. For these experiments AP-1/NHE1′HA cells, which were grown on 6-well plates, were incubated for 5 min at 37°C in the presence or absence of chymotrypsin (100 U/ml). Next, the cells were scraped off the wells with a rubber policeman and washed twice using Na+-rich medium containing 100 μM TPCK. Finally, the cells were resuspended in Laemmli sample buffer and boiled for 5 min. Samples were subjected to electrophoresis in 7.5% polyacrylamide gels and transferred to nitrocellulose. Blots were blocked with 5% nonfat dried milk and exposed to a 1:5,000 dilution of mouse mAbs against HA. The secondary antibody, goat anti–mouse coupled to HRP, was used at 1:5,000 dilution. Immunoreactive bands were visualized using enhanced chemiluminescence. The amount of chymotrypsin-sensitive immunoreactive NHE1 was used to quantify the amount of each one of the mutants expressed at the cell surface.
PIP2 Content Measurements
Two methods were used to quantify the PIP2 content of AP-1/NHE1′HA cells. Radiolabeled PIP2 was quantified by TLC, essentially as described by Okada et al. 1994. Cells were incubated at 37°C for 1 h in a phosphate-free medium, followed by incubation for 1 h with [32P]orthophosphate (150 μCi/ml). Next, the cells were washed twice with PBS and, where indicated, depletion of cellular ATP was induced. Cells were washed twice more with ice-cold PBS, scraped off the wells with a rubber policeman, and sedimented. For lipid extraction ∼106 cells were resuspended in 620 μl of chloroform/methanol/8% HCIO4 (50:100:5). The mixture was vortexed and 200 μl of chloroform and 200 μl of 8% HCIO4 were added to separate the organic phase, which was washed with chloroform-saturated 1% HCIO4 before being dried under N2. The dried lipids were dissolved in 15 μl of 95:5 chloroform/methanol to be spotted on a silica gel H plate (Analtech) that had been preactivated by heating at 80°C for 90 min before spotting. The plate was developed in chloroform/acetone/methane/acetic acid/water (80:30:26:24:14) and dried. Radioactivity was quantified with a Molecular Dynamics Storm 840 PhosphorImager.
PIP2 content was also assessed by immunoblotting. Approximately 106 AP-1/NHE1′HA cells were washed three times with large volumes of ice-cold PBS, scraped off the wells with a rubber policeman, and sedimented. The cells were resuspended in ∼15 vol of methanol/chloroform/concentrated HCl (100:50:1). The mixture was vortexed and 1 vol (equal to the volume of the original cell pellet) of 100 mM EDTA was added. Chloroform (5 vol) and water (5 vol) were added, the cloudy mixture was swirled and centrifuged at 400 g for 5 min to induce phase separation. The bottom (organic) phase was transferred to glass tubes and dried under a stream of N2. The dried lipid extract was redissolved in 10 μl of 1:1 chloroform/methanol (containing 0.1% HCl) and spotted onto polyvinylidene difluoride membranes (Millipore). After drying, blots were blocked overnight at 4°C in TBS containing 3% BSA. Blots were exposed to a 1:500 dilution of mouse mAbs against PIP2. The secondary antibody, goat anti–mouse coupled to HRP, was used at 1:5,000 dilution. Immunoreactive bands were visualized by enhanced chemiluminescence.
Construction and Expression of Glutathione-S-transferase (GST) Fusion Proteins
GST-NHE1 fusion proteins were constructed by amplifying cDNA regions encoding the COOH-terminal amino acids 506–576 of wild-type and mutant (M1, M2, and M1 + 2) NHE1′HA by PCR using the appropriate 5′ and 3′ primers containing unique BamHI and EcoRI restriction sites, respectively, at their 5′-ends. The BamHI-EcoRI DNA fragments were subcloned into the corresponding sites of pGEX-2T and sequenced to confirm the fidelity of the fusion constructs. The plasmid encoding a fusion between GST and residues 548–813 of NHE1 was provided by Dr. L. Fliegel (University of Edmonton, Alberta, Canada). GST fusion proteins were expressed in competent DH5α bacteria and purified using glutathione-agarose as described (Frangioni and Neel 1993).
PIP2 Binding Determinations
To assay their ability to bind lipids, GST fusion proteins were allowed to adhere overnight at 4°C onto 96-well plates (1 μg protein in 50 mM sodium bicarbonate buffer, pH 9.6 per well). After washing three times with the sodium bicarbonate buffer, the wells were overlaid for 2 h at room temperature with PIP2 (1 μg/well). After three washes to remove unbound lipid, the samples were blocked with 5% nonfat dried milk and exposed overnight at 4°C to a 1:500 dilution of mouse mAb to PIP2. The secondary antibody, goat anti–mouse coupled to HRP, was used at 1:2,000 dilution in the presence of 2.5% nonfat dried milk. Sigma Fast OPD tablets were used as a substrate for the detection of peroxidase activity. Absorbance at 450 nm was quantified on a microtiter plate reader.
Effect of ATP Depletion on NHE1 Activity and PIP2 Content
To test the involvement of phosphoinositides in the regulation of NHE1 by ATP, the nucleotide was depleted by incubation for 10 min in a medium containing inhibitors of both glycolysis (2-deoxy-d-glucose) and mitochondrial respiration (antimycin A). As shown in Fig. 1 A, such treatment resulted in the consistent depletion of ≥90% of the cellular ATP. Depletion was performed in K+-rich solution to preclude Na+ loading and dissipation of the inward Na+ gradient, which drives Na+/H+ exchange due to the loss of Na+/K+-ATPase activity. Despite these precautions, depletion of ATP caused a pronounced inhibition of NHE1, in agreement with earlier observations (Cassel et al. 1986; Brown et al. 1991; Levine et al. 1993; Kapus et al. 1994; Aharonovitz et al. 1999). The robust Na+-induced recovery from a mild acid load (pHi ∼6.4) observed in untreated cells transfected with NHE1′HA was inhibited by ≥90% after metabolic depletion (Fig. 1 B).
We next analyzed whether metabolic depletion was accompanied by a reduction in the content of cellular PIP2. This phospholipid is the primary substrate of phosphoinositide-specific PLC, which upon activation by agonists, releases diacylglycerol and inositol 1,4,5,-trisphosphate (IP3). The latter induces the release of calcium stored in the ER, promoting an increase in [Ca2+]i. Depletion of plasmalemmal PIP2, the substrate of the phospholipase, would be expected to blunt the [Ca2+]i response to agonists. Therefore, we measured [Ca2+]i in AP-1/NHE1′HA cells subjected to purinergic stimulation. In otherwise untreated cells, purinergic activation led to a rapid and transient elevation in [Ca2+]i (Fig. 1 C). Because the cells were suspended in Ca2+-free medium, this response reflected exclusive mobilization of intracellular stores by IP3. By comparison, the transient was virtually eliminated in ATP-depleted cells. Failure to respond was not attributable to complete depletion of calcium stored in endomembrane compartments, since addition of ionomycin, a Ca2+ ionophore, elicited a sizable [Ca2+]i increase (not illustrated). These findings are consistent with depletion of PIP2, but other mechanisms may also have contributed to this effect.
The cellular content of PIP2 was analyzed more directly by radiolabeling phospholipids in situ with [32P]orthophosphate, followed by their separation by thin layer chromatography. Using this approach, the amount of 32P-labeled PIP2 decreased by >75% when AP-1/NHE1′HA cells were subjected to the ATP-depletion protocol for 10 min (Fig. 1 D).
Because radiolabeling with [32P]orthophosphate may not have attained equilibrium, it is conceivable that the labeled pool of PIP2 is not representative of the total content of the phosphoinositide. Therefore, the total content of PIP2 was analyzed using a novel immunoblotting assay. Total cellular lipids were extracted in acidified chloroform/methanol, dried, and spotted onto polyvinylidene difluoride. The amount of PIP2 was estimated by blotting with a PIP2-specific antibody, which, in turn, was detected by enhanced chemiluminescence. To validate this novel procedure, we compared the PIP2 content of control and ionomycin-treated cells. By elevating [Ca2+]i, this ionophore is known to activate PLC, thereby reducing the cellular content of PIP2 (Rhee and Bae 1997). Exposure of the cells to 10 μM ionomycin for 5 min reduced the PIP2 content by 53% (not shown). As shown in Fig. 1 D, depletion of ATP similarly reduced the PIP2 content by >50%, consistent with the chromatographic data.
Effect of ATP Depletion on Plasmalemmal PIP2 Content
Regardless of the method used for PIP2 determination, depletion of the phosphoinositide in metabolically inhibited cells was incomplete after 10 min. This may reflect the existence of multiple subcellular pools of PIP2 with varying susceptibility to ATP depletion. Because NHE1 is likely to be affected exclusively by plasmalemmal PIP2, we sought methods to analyze this subcompartment more specifically. To this end, we took advantage of the recent observation that the PH domain of PLCδ binds with high affinity and selectivity to PIP2 (Ferguson et al. 1995; Lemmon et al. 1995). Therefore, we constructed a cDNA encoding a chimeric protein encompassing this PH domain and enhanced GFP (PHPLCδ-GFP), to monitor the subcellular distribution of PIP2 before and after depletion of ATP. As reported for other cell types (Stauffer et al. 1998; Varnai and Balla 1998), the chimeric protein is largely associated with the plasma membrane of otherwise untreated AP-1/NHE1′HA cells (Fig. 2 A). Activation of PLC by addition of ionomycin induced the rapid translocation of PHPLCδ-GFP to the cytosol, an indication of PIP2 hydrolysis (Fig. 2 B), as reported previously (Varnai and Balla 1998). Similarly, the chimeric protein was displaced from the membrane by incubation with neomycin (Fig. 2 C), a cell-permeant cationic antibiotic known to bind tightly to the headgroup of phosphoinositides (Schacht 1976). Importantly, ATP depletion also resulted in extensive translocation of PHPLCδ-GFP from the membrane to the cytosol (Fig. 2 D). The displacement was almost complete, comparable in extent to that induced by ionomycin and neomycin (Fig. 2 E). Jointly, these results indicate that the metabolic depletion protocol used for inhibition of NHE1 concomitantly causes depletion of plasmalemmal PIP2.
Effect of Neomycin on NHE1 Activity
The correlation between the depletion of ATP and the reduction in PIP2 content suggests, but does not prove, that binding of the phosphoinositide to NHE1 modulates the rate of Na+/H+ exchange. To further evaluate this hypothesis, we sought to modify the amount of available PIP2 without simultaneously depleting the intracellular ATP.
Neomycin has been shown to bind tightly to PIP2, restricting the availability of its headgroup to proteins including PLC (Downes and Michell 1981). Therefore, we tested whether neomycin would similarly interfere with the putative interaction between NHE1 and PIP2. Cells expressing wild-type NHE1 were preincubated with neomycin to allow entry of the antibiotic and sequestration of PIP2. As shown above, this resulted in effective displacement of PHPLCδ-GFP from the inner surface of the plasma membrane (Fig. 2, compare A and C). Because several actin-binding proteins are modulated by PIP2 (Yu et al. 1992), we also evaluated the effects of neomycin in cells stained with rhodamine-phalloidin. As illustrated in Fig. 3, the stress fibers, which are routinely observable in AP-1/NHE1 cells, were largely eliminated by treatment with neomycin. The antibiotic also increased the number of binucleated cells (Fig. 3 B), a phenotype that was associated earlier with hydrolysis of PIP2 by synaptojanin (Sakisaka et al. 1997). These observations suggest that neomycin effectively entered the cells, where it interacted with the headgroup of PIP2. Fig. 3 C illustrates the effect of neomycin on NHE1 activity. On average, the antibiotic depressed the initial rate of recovery from an acid load by 62% (n = 4). These observations are consistent with the notion that PIP2 regulates the activity of NHE1.
Effect of Calcium-induced PIP2 Depletion
Elevation of [Ca2+]i can activate endogenous PLC, thereby inducing hydrolysis of PIP2. The effectiveness of this maneuver was documented earlier in cells transfected with PHPLCδ-GFP (Fig. 2). Therefore, we used this approach as an alternative means of decreasing the content of plasmalemmal PIP2, to assess its effects on NHE1. However, it is noteworthy that calcium can also alter the activity of NHE1 by mechanisms that do not involve PIP2. Specifically, NHE1 possesses sites that interact with calcium–calmodulin, resulting in stimulation of antiport activity (Bertrand et al. 1994). Therefore, it was important to dissociate the direct effects of calcium on the exchanger from its indirect action mediated by PLC. This was accomplished by using cells transfected with NHE1Δ582, a truncated form of the exchanger that lacks the calmodulin-binding sites (residues 636–656 and 657–700).
As shown in Fig. 4 A, this truncated version of NHE1 actively exchanges Na+ for H+ and, more importantly, displays marked sensitivity to ATP. In five experiments, the rate of pH recovery in NHE1Δ582 transfectants, measured at pHi 6.4, was inhibited by 87% (Fig. 4 B) upon metabolic depletion. Next, we tested the effects of ionomycin on H+ extrusion. The contribution of Ca2+/H+ exchange mediated by the ionophore to the rate of pHi change was minimized by measuring the pHi recovery in a Ca2+-free medium. Treatment with either 2 or 10 μM ionomycin greatly inhibited NHE1 activity (Fig. 4A and Fig. B). Importantly, we ascertained that treatment with the Ca2+ ionophore did not affect the cellular ATP content (e.g., in cells treated with 2 μM ionomycin ATP was 93% of the control). Together with the neomycin experiments, these results indicate that conditions that block or deplete plasmalemmal PIP2 markedly inhibit the activity of NHE1.
Effect of Overexpression of PIP2-binding PH Domains on NHE Activity
As described above, the PH domain of PHPLCδ, which binds preferentially to the headgroup of PIP2, can be used to detect the location of the phosphoinositide. If expressed in large quantity, however, PHPLCδ will effectively compete with endogenous ligands for binding to the finite amount of plasmalemmal PIP2. To assess the dependence of NHE1 activity on PIP2, we expressed PHPLCδ-GFP in COS-1 cells. These cells express the T antigen and, therefore, allow the replication of vectors containing the SV40 origin, with consequent amplification of the amount of protein expressed. The presence of the SV40 origin of replication in the pEGFP vector enabled us to test the effect of overexpression of PHPLCδ-GFP on NHE activity in COS-1 cells. The results of these experiments are illustrated in Fig. 5. Sequestration of PIP2 by PHPLCδ-GFP greatly reduced the rate of Na+-induced pH recovery in acid-loaded cells (Fig. 5 A). In five separate experiments, the initial rate of alkalinization was inhibited by ≥80% (Fig. 5 B).
Effect of a PIP2 Phosphatase on NHE Activity
The putative role of PIP2 in regulating NHE1 was also tested using a 5′-specific PIP2 phosphatase. To selectively reduce plasmalemmal PIP2, cells were transfected with a yeast 5′-specific phosphatase, Inp54p, targeted to the plasma membrane by fusion with a myristoylation/palmitoylation sequence from the NH2 terminus of Lyn (Raucher et al. 2000). GFP was also included in this chimeric construct, termed PM-5′-phosphatase-GFP, to facilitate identification of the transfected cells. Inp54p was shown recently to function as an effective 5′-phosphatase towards PIP2 both in vitro and in vivo (Raucher et al. 2000). When expressed in COS-1 cells, PM-5′-phosphatase-GFP markedly inhibited Na+/H+ exchange (Fig. 5 B). The accumulated evidence strongly suggests that normal levels of plasmalemmal PIP2 are essential for optimal NHE1 function.
Binding of PIP2 to NHE1
The mechanism underlying the inhibition of NHE activity upon depletion of plasmalemmal PIP2 was investigated next. As suggested for other transporters (Huang et al. 1998), it is conceivable that PIP2 binds directly to the exchanger, thereby modulating its activity. Indeed, perusal of the primary structure of NHE1 revealed the presence of two sequences that resemble the PIP2-binding motifs of a variety of proteins, including gelsolin and profilin (Yu et al. 1992). These putative PIP2-binding motifs (residues 513–520, called hereafter site 1, and 556–564, site 2) are located in the cytosolic tail, near the predicted point of emergence of the last putative transmembrane domain (residue 504). The sequences of these motifs, which alternate cationic and hydrophobic residues and a schematic representation of their location within the predicted topology of NHE1, are illustrated in Fig. 6.
Recombinant fusion proteins and synthetic peptides encompassing one or both of these motifs were generated to test their ability to interact with PIP2. A summary of the results is presented in the Fig. 6 (bottom left). In brief, a fusion of GST with residues 548–813 bound PIP2 with much greater efficiency than GST alone (such excess binding is defined hereafter as “specific binding”). This suggests that residues encompassing the site 2 motif may be involved in phosphoinositide binding. Accordingly, preincubation of the lipid with a synthetic 550–564 peptide precluded specific binding of PIP2 to the GST (548–813) fusion protein. To more precisely evaluate the role of the juxtamembrane domain of NHE1 in PIP2 binding, we analyzed GST fusions encompassing residues 506–576 and several mutants where the cationic residues were simultaneously replaced by alanine. As shown in Fig. 6, GST-NHE1(506–576) bound PIP2 to an extent comparable to that found for the 548–813 construct. Elimination of the NH2-terminal cationic motif (513KKKQETKR520 to 513AAAQETAA520; residues involved in mutagenesis are underlined; M1 mutant in Fig. 6) had little effect on phosphoinositide binding. Unexpectedly, mutation of the more COOH-terminal motif (556RFNKKYVKK564 to 556AFNAAYVAA564; M2 mutant in Fig. 6) was equally ineffectual. However the combined mutation of both motifs (Fig. 6, M1 + 2) largely eliminated specific PIP2 binding. These findings suggest that either motif is capable of binding the phospholipid, and that only one site can be occupied at any one time. Because of the comparatively large size of PIP2 micelles (∼90 kD), binding of one such micelle to either motif likely precludes binding of a second one to the other cationic sequence. Indeed, both sites may interact with the same micelle simultaneously.
Mutation of Putative PIP2-binding Motifs: Functional Consequences
Having identified two PIP2-binding motifs in the cytosolic domain of NHE1, we proceeded to assess their functional role. Stable lines were generated by transfecting antiport-deficient AP-1 cells with either full-length wild-type NHE1, or with mutated forms containing substitutions of the cationic residues in either one or both putative PIP2-binding motifs with alanines, identical to those engineered in the GST-NHE1(506–576) fusions. All constructs were epitope-tagged to facilitate their immunological detection. As illustrated in Fig. 7 A, the two singly mutated constructs of NHE1 (also called M1 and M2, by analogy with Fig. 6), as well as the double mutant (M1 + 2) generated full-length proteins that were at least partly expressed at the plasma membrane. Two lines of evidence indicate that the mutant proteins reach the plasma membrane. First, transfected AP-1 cell lines were selected by their ability to survive an acid challenge, indicating that the mutant exchangers were functional and, therefore, most likely at the cell surface. Second, as found earlier for wild-type NHE1 (Shrode et al. 1998), two distinct species of the mutant proteins were detected by immunoblotting: a faster migrating band that approximates the molecular mass predicted from the primary sequence, and a form that is larger and more heterogeneous as a result of complex glycosylation (Fig. 7 A). In the case of wild-type NHE1, the former was shown to be an immature intracellular species, whereas the fully glycosylated form reaches the plasma membrane. The similarity in the expression patterns suggests that the mutants are also fractionally targeted to the surface membrane. More direct evidence was obtained from analyzing the susceptibility of the proteins to chymotrypsin. When added to the external medium, this protease cleaves the glycosylated wild-type NHE1, yielding a membrane-bound form of increased mobility that lacks carbohydrate (Shrode et al. 1998). By contrast, the intracellular species is refractory to the protease, as anticipated. An identical proteolysis pattern was observed for the three mutant forms of NHE1, implying that they are properly processed and inserted in the plasma membrane.
Having ascertained that the mutants were appropriately expressed and targeted, we proceeded to evaluate the effect of the mutations on the basal rate of transport and on the ATP dependence of this process. The expression level of the exchangers varied among the transfected lines, and meaningful comparison of their rates of transport required normalization with respect to the number of plasmalemmal exchangers. This was estimated from the relative intensities of the chymotrypsin-sensitive bands in immunoblots, as shown in Fig. 7 A. Using this procedure, we compared the rates of Na+/H+ exchange in acid-loaded wild-type and mutant cells by measuring the pHi recovery induced by Na+. It is noteworthy that in all cases, the rate of alkalinization was negligible in the absence of Na+ (data not shown). A comparison of the basal rates of transport is shown in Fig. 7 B. Mutation of the juxtamembrane cationic residues (mutant M1) drastically reduced the efficiency of transport. At a pHi of 6.4, exchange was inhibited by ∼80%. An even greater inhibition was noted when the more COOH-terminal cationic cluster was mutated (mutant M2), and transport was negligible when both mutations were present simultaneously (mutant M1 + 2). A similar pattern was noted when NHE activity was assessed independently by measuring 22Na+ influx in cells clamped at different pHi ranging from 5.4 to 7.4 (Fig. 8, A–D). In these experiments, the radioisotope uptake at pHi 5.4, when normalized for cell-surface NHE1′HA protein expression, was decreased by 78, 90, and 94% in M1, M2, and M1 + 2 cells, respectively, compared with wild-type NHE1′HA. These results indicate that the motifs capable of binding PIP2 are essential for optimal Na+/H+ exchange by NHE1.
We next compared the effect of ATP depletion on the activity of the wild-type and mutated forms of NHE1. As shown earlier, metabolic depletion induced a marked depression in the rate of Na+/H+ exchange by native NHE1′HA, measured either as pHi recovery from a mild acid load (pHi 6.4; Fig. 7 C) or as 22Na+ influx over a broader pHi range (Fig. 8 A). With respect to 22Na+ influx, the inhibition was most noticeable at pHi 6.6, but was partially relieved at more acidic pHi levels (i.e., activity restored to 70–80% of wild-type values at pHi 5.4), in agreement with an earlier report (Ikeda et al. 1997). Despite their comparatively low rates of transport in the presence of ATP, the M1 and M2 mutants were further inhibited by depletion of the nucleotide (Fig. 7 C, and Fig. 8B and Fig. C). At pHi 6.4, the basal rate of transport in the dual mutant M1 + 2 was so low that no significant diminution could be detected after ATP depletion by measurements of pHi (Fig. 7 C). However, using the more sensitive isotopic method (Fig. 8 D), the small, residual activity measurable in the double mutant retained some sensitivity to metabolic depletion. Unlike the flux in wild-type cells, however, the inhibition caused by ATP depletion could not be reversed by lowering the pHi. These findings suggest that ATP depletion exerts a dual effect on NHE1: one component of the inhibitory response is mimicked by elimination of the phosphoinositide-binding sites 1 and 2 and can be counteracted by lowering pHi, whereas a second, smaller component is independent of sites 1 and 2, and is not reversed by acidification.
Binding of PIP2 to NHE1
It is well established that PIP2 plays a central role in signal transduction as a precursor to IP3 and diacylglycerol (Rhee and Bae 1997). However, it was only recently appreciated that PIP2 is essential for the regulation of several processes by acting as a direct ligand or cofactor of a variety of proteins. Physical interaction of PIP2 with proteins containing specific PH domains serves to target them to the vicinity of their substrates (Lemmon et al. 1997). Moreover, association with PIP2 can promote assembly of cytoskeletal proteins (Sakisaka et al. 1997), and alters the functional activity of enzymes and transporters (Hilgemann and Ball 1996; Baukrowitz et al. 1998; Huang et al. 1998). NHE1 can now be added to the growing list of proteins that bind to, and are regulated by, PIP2. The evidence supporting this contention is as follows. First, the COOH-terminal domain of NHE1 includes two sequences that resemble PIP2-binding motifs identified in other proteins. Second, the recombinant COOH terminus of NHE1 can bind PIP2 in vitro and a synthetic peptide containing one of the putative PIP2-interacting motifs effectively competes for such binding. Third, mutation of the cationic residues within the putative PIP2-binding motifs has profound effects on the activity of NHE1. Fourth, sequestration of PIP2 with neomycin or PHPLCδ-GFP reduced NHE activity; and fifth, depletion of PIP2 with ionomycin or PM-5′ phosphatase induced a marked inhibition of NHE1.
Mutation of either site 1 or 2 greatly depressed ion exchange (Fig. 7 and Fig. 8), whereas elimination of both sites was required to eliminate PIP2 binding (Fig. 6). This apparent discrepancy can be readily explained by assuming that, while both sites 1 and 2 can bind PIP2 separately, simultaneous binding at both sites is required for optimal transport. Indeed, the in vitro findings suggest that either motif is capable of binding the phospholipid. Because of the large size of the PIP2 micelle, it is likely that only one micelle can bind to each fusion protein. This explains why binding is not greater when the protein has two cationic domains than when it contains only one.
The mechanism whereby PIP2 alters the activity of the exchanger remains to be defined. In the case of the Na+/Ca2+ exchanger, an analogous PIP2-binding motif is thought to be autoinhibitory, since cytosolic perfusion with a synthetic peptide of similar sequence antagonizes exchange (DiPolo and Beauge 1994). Binding of the motif to PIP2 is believed to sequester it away from the transport moiety of the protein, precluding its autoinhibitory effect (Shannon et al. 1994). In principle, a similar mechanism could be envisaged for NHE1. However, deletion and truncation experiments are inconsistent with this model. Unlike the Na+/Ca2+ exchanger, which remains functional after removal of the PIP2-binding motif, NHE1 becomes greatly inhibited when the region encompassing sites 1 and 2 is truncated (Ikeda et al. 1997; Orlowski, J., and S. Grinstein, unpublished observations) or mutated (Fig. 7 and Fig. 8). We feel it is more likely, instead, that the optimal transport configuration of NHE1 requires the tight apposition of sites 1 and/or 2 with the inner surface of the plasma membrane. Departures from this configuration, induced either by truncation, mutation, or depletion of the PIP2 required to maintain the protein in place, result in inhibition of transport.
While we have shown that NHE1 can interact with PIP2 in vitro, the existence of such an interaction in situ remains inferential, based exclusively on the functional effects of phosphoinositide depletion. It is therefore possible that such functional effects may be indirect. In this regard, it is noteworthy that ezrin was recently reported to interact with NHE1 (Denker et al. 1998). Members of the ezrin/radixin/moesin (ERM) family are themselves capable of binding PIP2, which in turn modulates the ability of ERM to interact with other proteins (Hirao et al. 1996). Therefore, one could envisage a model wherein the availability of PIP2 dictates the extent of association of ERM proteins with NHE1. The latter interaction may be responsible for modulation of NHE1 activity, though this premise has not yet been tested experimentally. This model would be compatible with most of our observations, but would not account for the observed direct binding of PIP2 to sites 1 and 2 of NHE1. Finally, it is conceivable that PIP3 rather than PIP2 is required for NHE activity, and that the effects of depletion of the latter are indirect. This appears unlikely, in that inhibition of phosphatidylinositol 3′-kinase with wortmannin has no effect on NHE1 activity.
Role of PIP2 in the ATP Sensitivity of NHE1
The exquisite dependence on PIP2 could account, at least in part, for the well established ATP sensitivity of NHE1. Several lines of evidence support this notion. First, metabolic depletion is accompanied by a parallel decrease in the total (Fig. 1) and particularly in the plasmalemmal content of PIP2 (Fig. 2). Second, the extent of NHE inhibition induced by depletion of ATP is comparable to that obtained by extensive hydrolysis of PIP2 (Fig. 4); and third, elimination of the putative PIP2-binding motifs greatly reduces the magnitude of the ATP-dependent component of exchange (Fig. 7 and Fig. 8). It is also noteworthy that the PIP2-binding sequences identified in this report are within the region mapped earlier to confer ATP dependence to NHE1 (Ikeda et al. 1997). The hypothesis that PIP2 mediates the effects of ATP would explain why changes in the phosphorylation of NHE1 itself were not found to correlate with its inhibition in metabolically depleted cells (Goss et al. 1994) and why mutants such as NHE1Δ582, which lack the phosphorylation sites identified in NHE1, retain their sensitivity to ATP.
While mutation of the putative PIP2-binding sites 1 and 2 profoundly reduced the ATP-sensitive fraction of Na+/H+ exchange, a measurable nucleotide-sensitive component of transport remained (Fig. 8). This implies that an additional, phosphoinositide-independent mechanism contributes to the effect of ATP on NHE1. The existence of two or more sites of action of ATP in the regulation of NHE was suggested previously by the findings of Demaurex et al. 1997, who found that nonhydrolyzable analogues of ATP could partially restore NHE activity in ATP-depleted cells. The target of such nonhydrolyzable nucleotides, which are unable to phosphorylate phosphoinositides, remains undefined. Finally, though the ubiquitous isoform NHE1 was used in this study, the interaction with PIP2 may extend to other isoforms. Not only are all isoforms, thus far, tested exquisitely sensitive to depletion of ATP, but motifs similar to those postulated to bind PIP2 in NHE1 exist also in NHE2-5. This conservation of sequence argues in favor of an important role for these motifs in the regulation of Na+/H+ exchange.
This work was supported by the Medical Research Council of Canada (MRC) and the National Institutes of Health (grant RO1-HL 55672). O. Aharonovitz was supported by the Canadian Cystic Fibrosis Foundation and the Arthritis Society of Canada. J. Orlowski is supported by a Scientist award from the MRC. S. Grinstein is supported by a Distinguished Scientist award from the MRC and is cross-appointed to the Department of Biochemistry of the University of Toronto. S. Grinstein is an International Scholar of the Howard Hughes Medical Institute, and is the current holder of the Pitblado Chair in Cell Biology.
Abbreviations used in this paper: AM, acetoxymethylester; BCECF, 2′,7′-bis(2-carboxyethyl)-5(6)-carboxyfluorescein; Ca2+i, cytosolic free calcium; ERM, ezrin/radixin/moesin; GFP, green fluorescent protein; GST, glutathione-S-transferase; IP3, 1,4,5,-trisphosphate; HA, hemagglutinin; NHE, Na+/H+ exchanger; NMG, N-methyl-d-glucammonium; pHi, intracellular pH; PH, pleckstrin homology; PIP2, phosphatidylinositol-4,5-bisphosphate; PLC, phospholipase C; TPCK, N-tosyl-l-phenylalanine chloromethyl ketone.