Phosphoinositide Binding Differentially Regulates NHE1 Na+/H+ Exchanger-dependent Proximal Tubule Cell Survival*

Background: Chronic kidney disease is perpetuated by tubular epithelial cell apoptosis, and the NHE1 Na+/H+ exchanger defends against apoptosis in response to undefined regulatory mechanisms. Results: Phosphatidylinositol 4,5-bisphosphate (PI(4,5)P2) and phosphatidylinositol 3,4,5-trisphosphate (PI(3,4,5)P3) bind and differentially regulate NHE1 through weak electrostatic and pH-dependent interactions. Conclusion: NHE1-phospholipid binding regulates NHE1 activities. Significance: NHE1-dependent cell survival is mediated through toggling between interactions with PI(4,5)P2 and PI(3,4,5)P3. Tubular atrophy predicts chronic kidney disease progression, and is caused by proximal tubular epithelial cell (PTC) apoptosis. The normally quiescent Na+/H+ exchanger-1 (NHE1) defends against PTC apoptosis, and is regulated by PI(4,5)P2 binding. Because of the vast array of plasma membrane lipids, we hypothesized that NHE1-mediated cell survival is dynamically regulated by multiple anionic inner leaflet phospholipids. In membrane overlay and surface plasmon resonance assays, the NHE1 C terminus bound phospholipids with low affinity and according to valence (PIP3 > PIP2 > PIP = PA > PS). NHE1-phosphoinositide binding was enhanced by acidic pH, and abolished by NHE1 Arg/Lys to Ala mutations within two juxtamembrane domains, consistent with electrostatic interactions. PI(4,5)P2-incorporated vesicles were distributed to apical and lateral PTC domains, increased NHE1-regulated Na+/H+ exchange, and blunted apoptosis, whereas NHE1 activity was decreased in cells enriched with PI(3,4,5)P3, which localized to basolateral membranes. Divergent PI(4,5)P2 and PI(3,4,5)P3 effects on NHE1-dependent Na+/H+ exchange and apoptosis were confirmed by selective phosphoinositide sequestration with pleckstrin homology domain-containing phospholipase Cδ and Akt peptides, PI 3-kinase, and Akt inhibition in wild-type and NHE1-null PTCs. The results reveal an on-off switch model, whereby NHE1 toggles between weak interactions with PI(4,5)P2 and PI(3,4,5)P3. In response to apoptotic stress, NHE1 is stimulated by PI(4,5)P2, which leads to PI 3-kinase activation, and PI(4,5)P2 phosphorylation. The resulting PI(3,4,5)P3 dually stimulates sustained, downstream Akt survival signaling, and dampens NHE1 activity through competitive inhibition and depletion of PI(4,5)P2.

Chronic kidney diseases affect more than 26 million people in the United States (1), and over 500,000 people require renal replacement therapy, at an annual Medicare cost of $24 billion (2). Tubular atrophy is a hallmark of chronic kidney disease progression, and is superior to glomerular histopathology as a predictor of clinical outcomes (3)(4)(5)(6). Proximal renal tubular epithelial cell (PTC) 2 apoptosis is a prominent feature in human renal biopsies and mouse models of chronic, progressive kidney diseases, and an important mechanism of tubular atrophy (7)(8)(9)(10)(11)(12).
Apoptosis is characterized by cell volume shrinkage and cytosol acidification (13), which favor activation of pro-apoptotic caspases (14,15), Bax (16), and Smac/DIABLO (17), consistent with a pH threshold for triggering cell death pathways. The NHE1 Na ϩ /H ϩ exchanger is expressed on the basolateral plasma membrane of epithelial cells (18), and is normally quiescent (19,20). NHE1 counteracts apoptosis in the proximal tubule and other tissues (21), in part through Na ϩ influx and H ϩ extrusion, resulting in restoration of cell volume and cytosolic pH, respectively. In addition, the NHE1 C-terminal tail serves as a scaffold for assembly of signaling modules (22,23). Multiple NHE1 stimuli have been identified, including growth factors, integrins, hypertonicity, and apoptotic stress, but the mechanisms regulating NHE1 activation and inactivation in response to apoptotic stress have not been described.
Regulation of NHE1 is complex, involving multiple protein binding partners. Because NHE1 is a transmembrane protein there is also potential for association and regulation by membrane lipids. However, interaction between PI(4,5)P 2 and the NHE1 cytosolic tail is the sole example of a phospholipid binding with the exchanger. Other plausible candidates include PI(3,4,5)P 3 , which carries a relatively greater negative charge (Ϫ6 versus Ϫ4 for PI(4,5)P 2 ), and phosphatidylserine (PS), which is more abundant than PI(4,5)P 2 , and perpetuates apoptosis by translocating from the inner to outer plasma membrane leaflet (40). In this report, we show that NHE1 is regulated by toggling between low affinity interactions with PI(4,5)P 2 and PI(3,4,5)P 3 .
Cell Culture-LLC-PK1 cells were purchased from ATCC (Manassas, VA) and maintained in DMEM (Invitrogen) plus 10% fetal bovine serum (HyClone, Logan, UT). C57BL/6 wildtype and NHE1-null C57BL/6 Swe/Swe (42) proximal tubule cells were derived from mice, which were purchased from Jackson Laboratories. Proximal tubules were isolated by Percoll gradient centrifugation (43), maintained in primary culture in DMEM/F-12 (Invitrogen) plus 10% fetal bovine serum (HyClone), and then immortalized by infection with temperature-sensitive SV40. Cell lines were propagated at 33°C, and then studied under differentiating conditions after 24 h at 37°C. In some experiments, cells were cultured on permeable supports (Costar Corning, Lowell, MA) to generate polarized monolayers. For experiments to assess NHE1 and phosphoinositide membrane domain sorting, we used 24-mm diameter, 0.4-m pore, polyester membrane supports (Costar number 3450). For experiments to assess single cell NHE1 Na ϩ /H ϩ exchange activity, 12-mm 0.4-m pore supports (Costar number 3801) were used. Typically, cells achieved confluence after 5-6 days, and were studied 2-3 days later.
Phospholipid Overlay Assays-The peptide corresponding to the entire NHE1 cytosolic tail (residues 501-815 (44); cNHE1) was PCR cloned from rat kidney (forward primer: 5Ј-GATG-GGGATTCGCCCCTGGTAGACCTGTTGGCT-3Ј, reverse primer: 5Ј-GGGGAAGCTTCTCGAGTTCTCGAGTTCTA-CTGCCCTTTGGGGATGAA). The primers contained Xho/ HindIII and BamHI sites, respectively, which permitted subcloning into a plasmid that added a His 6 tag to the N terminus. The peptide was purified to homogeneity by passage over Ni 2ϩ columns and sequential dialysis to remove urea. The His 6 -cNHE1 peptide was suspended in renaturing buffer containing 10 mM HEPES, pH 7, 150 mM NaCl, 5% glycerol, and 2 mM DDT. cNHE1 fusion protein (1 g/ml) was then incubated for 1 h with membrane phospholipids spotted on nitrocellulose membranes (PIP strips, Echelon, Salt Lake City, UT), and probed for binding with either anti-NHE1 (45) or anti-His antibodies. Quantitative densitometric data were generated using a Storm phosphorimager and normalized to PI(4,5)P 2 intensity.
Surface Plasmon Resonance-Binding parameters for the interaction of phosphoinositides with the NHE1 cytosolic domain were measured using a Biacore 3000 SPR-based biosensor (Biacore AB, Uppsala, Sweden). cNHE1 peptide was prepared as previously described, and immobilized on the CM5 chip (Biacore) at a density of Ͼ1000 response units according to the manufacturer's protocol. Phospholipid analytes included 70% PC, 30% PS, which served as a control for inner leaflet phospholipids, and experimental groups comprised of C16and diC 8 -PI(4,5)P 2 or PI(3,4,5)P 3 (3% (w/w) in 70% PC, 30% PS), all of which were purchased from Echelon. Phospholipids were dispersed by sonication and passage through an extruder (Avanti Polar Lipids, Alabaster, AL). Different concentrations of phospholipid suspensions in Biacore HBS-P buffer (10 mM HEPES, pH 7.4, 150 mM NaCl, 0.005% surfactant P20) were flowed at 5 l/min for 4 min to determine association rate constants, followed by a 4-min perfusion with buffer only to determine the dissociation rate constants. All data were corrected for the response obtained using a blank reference flow cell blocked with de-lipidated albumin (0.1 mg/ml). The chip surface was regenerated between experiments with 50 mM NaOH. Data were analyzed using the BIAevaluation 3.1 program as previously described (46).
Phosphoinositide Assays-Methods have previously been described (47,48). LLC-PK1 cells were plated to obtain a density of 1.5 ϫ 10 6 cells/100-mm culture dish in DMEM/F-12 medium supplemented with 10% FBS and 1% penicillin/streptomycin. The medium was then replaced with phosphate-free DMEM supplemented with 0.2 mCi of 32 P i (PerkinElmer Life Sciences NEX063) for 2 h at 37°C (48,49). Medium was replaced with buffer (110 mM KH 2 PO 4 ϩ 30 mM NaCl) containing 0.2 mCi of 32 P i and 10 M nigericin at experimental pH values (6.5, 7.0, or 7.5) for 5 or 15 min. Cells were lysed with 1 M HCl containing 5 mM glacial acetic acid (250 l). Lipids were extracted in 665 l of chloroform and vortexed to uniformity. Aqueous and chloroform fractions were separated by centrifugation (1000 ϫ g for 5 min), and 50 l of the chloroform fraction was loaded as described below. As a positive control for PI(3,4,5)P 3 , an in vitro lipid kinase assay was conducted (48). Briefly, 20 g of sonicated PI(4,5)P 2 (Echelon) substrate was added to the reaction assay buffer (10 mM Tris-Cl, pH 7.4, 150 mM NaCl, 5 mM EDTA, 100 M sodium orthovandate). The reaction was started by adding 500 g of SF-9 purified PI 3-kinase-␥ enzyme, 10 l of 440 M ATP, and 10 Ci of [␥-32 P]ATP (25°C, 10 min) with continuous agitation, and stopped with 20 l of 6 N HCl. Lipids were extracted in 160 l of chloroform/ methanol (1:1) followed by vortexing and centrifugation at 1000 ϫ g to separate the phases. Thirty l of the organic phase was spotted on 200-m silica-coated flexi-TLC plates (Selectoflexible; Fischer Scientific) precoated with 1% potassium oxalate. Spots were dried and resolved chromatographically with 2 N glacial acetic acid, 1-propanol (1:1.87). Plates were dried after resolution, and exposed for autoradiography. PI(4,5)P 2 and PI(3,4,5)P 3 were quantified by densitometry and liquid scintillation counting.
Phospholipid Loading-Cells were cultured on permeable supports to generate polarized monolayers. To enhance specific phospholipid content, PI(4,5)P 2 or PI(3,4,5)P 3 vesicles, which have been chemically modified to resist phosphatase activity (Echelon), were incorporated into plasma membranes according to published methods (50,51). For PI(4,5)P 2 analogs the phosphodiester bond is modified by addition of an ␣-fluorophosphonate moiety. For PI(3,4,5) 3 analogs the 3-phosphate bond is rendered metabolically stable as a phosphorothioate. Phosphoinositide insertion into membrane domains was verified by immunodetection of GFP-tagged PLC␦ or Akt pleckstrin homology (PH) domain peptides (see below).
Total internal reflection fluorescence (TIRF) microscopy and fluorescence resonance energy transfer (FRET) measurements were performed using an Olympus IX71 Inverted Microscope configured for FRET (52) and equipped with a two-line Olympus TIRFM system and ϫ60 TIRFM objective lens. TIRF BODIPY-PI(4,5)P 2 fluorescence was acquired using a multiargon laser with 488 nm excitation and 510 nm emission filters. NHE1-PI(4,5)P 2 interactions within the plasma membrane were determined by combined TIRF/FRET using BODIPY-PI(4,5)P 2 (Avanti Polar Lipids) and RFP-NHE1 FRET donor/ acceptor pairs. Images were acquired using a Hamamatsu ORCA-ER charge-coupled device (12 bit) controlled by SLIDE-BOOK software (Intelligent Imaging Innovations, Denver, CO). Image analyses were performed using ImageJ and involved subtraction of background autofluorescence and blurred light and quantification of fluorescence intensity. FRET was measured using the relative fluorescence intensity of the donor (BODIPY-PI(4,5)P 2 , peak excitation/emission ϭ 503/525), according to established methods (53). Calculation of FRET efficiency was performed using the equation, E ϭ 1 Ϫ F DA /F D (54) , where F DA is the fluorescence intensity of the donor BODIPY-PI(4,5)P 2 in cells expressing the acceptor RFP-NHE1, and F D is the fluorescence intensity of the donor BODIPY-PI(4,5)P 2 in the absence of the acceptor RFP-NHE1. In control experiments with PI(4,5)P 2 -BODIPY vesicles or cells expressing only RFP-NHE1, fluorescence values were unchanged over the pH 6.5-7.5 range, indicating that neither fluorophore was pH-sensitive.
Cytosolic pH Measurements-Spectrofluorimetric cytosolic pH assays in cell suspensions using BCECF-AM have been described previously (22,45). Single cell pH assays were conducted in LLC-PK1 cells maintained on glass coverslips according to methods adapted from Bachmann et al. (55). Cells were transiently transfected with GFP-tagged PLC␦-PH or Akt-PH constructs. Fields containing cells with a plasma membrane fluorescence pattern, as well as GFP-transfected control cells, were isolated for examination at ϫ40 magnification with a Leica DM IRE2 microscope. Cells were loaded in situ with BCECF-AM (Molecular Probes, Eugene, OR; 2.5 M, 15 min) Ϯ N-ethyl-N-isopropyl amiloride (EIPA, Sigma; 1 M), then incubated with NH 4 Cl (30 mM in sodium-free Ringer solution, 20 min), acidified following a 90-s washout with sodium-free Ringer buffer, which was then replaced with Na ϩ -containing Ringer buffer to allow Na ϩ /H ϩ exchange. Cells were then observed for pH recovery, which was monitored by BCECF fluorescence (emission ϭ 530 nm) following an 0.08-s excitation at ϭ 495 and 440 nm every 20 s. Excitation pulses Յ0.05 s yielded unacceptably low signal to background ratios, whereas Ն0.12 s duration or more frequent pulses caused significant photobleaching. Calibration curves were generated with pH 6.0, 7.0, and 8.0 buffers in high K ϩ , nigericin-permeabilized cells at the end of each experiment. Epifluorescence was recorded using a SPOT-RT camera (Diagnostic Instruments, Sterling Heights, MI) and images were acquired and analyzed using SimplePCI imaging software (Compix Inc., Cranberry Township, PA). Cytosolic pH was determined by calculating BCECF fluorescence ratios within a subcellular area that excluded plasma membrane and nuclei using SimplePCI software.
Apoptosis Assays-DNA degradation was employed as an index of apoptosis, and was assessed in fixed cells by nuclear morphology with DAPI labeling and TUNEL assays, according to previously described methods (7).
Statistics-All data are representative of a minimum of three experiments per condition. Quantitative results are presented as mean Ϯ S.E., unless otherwise indicated. Comparisons between multiple groups were made by one-way analysis of variance with the Student-Newman-Keuls, Kruskal-Wallis, or Dunnett tests for parametric and nonparametric data. Comparison between two groups was made by a paired t test. Statistical significance is defined as p Ͻ 0.05.

RESULTS
NHE1 Binds Multiple Membrane Phospholipids-PI(4,5)P 2 has previously been shown to bind NHE1 and regulate Na ϩ /H ϩ exchange. Although PI(4,5)P 2 is relatively abundant, multiple other inner leaflet plasma membrane lipids could also potentially bind and influence NHE1 function. To test this possibility, membrane overlay assays were conducted with a His 6 -tagged polypeptide corresponding to the entire cNHE1. The strips were then probed with antibodies immunoreactive to either a cNHE1 peptide or the His 6 tag, revealing NHE1 binding most prominently to PI(3,4,5)P 3 , and slightly less to the phosphatidylinositol (PI) bisphosphates (Fig. 1B). More prolonged exposure of blots to film revealed weaker cNHE1 binding to other membrane phospholipids, such as anionic PI monophosphates, phosphatidic acid, and rarely to PS ( Fig. 1C and supplemental Fig. S1). Incubation with a His 6 peptide resulted in no binding, and His 6 peptide preincubation did not inhibit cNHE1 binding (not shown), indicating that the cNHE1-lipid interaction was not mediated by the His 6 tag.
Binding affinity constants were calculated using surface plasmon resonance and soluble C8-phosphoinositides, which permitted more accurate determination of phospholipid concen- trations. As seen in Table 1, PI(4,5)P 2 and PI(3,4,5)P 3 each bound with low affinity to cNHE1. PI(3,4,5)P 3 affinity was relatively greater compared with PI(4,5)P 2 , in agreement with the overlay and C16-phosphoinositide surface plasmon resonance studies. As a negative control, cNHE1 did not bind PI.
NHE1 Binds Phosphoinositides through Electrostatic Interactions-To test the mechanism of NHE1-phosphoinositide interaction, wild-type and cNHE1 mutant peptide ( Fig. 2A) binding was tested by two different methods. Fig. 2B demonstrates complete abrogation of membrane phospholipid binding to the cNHE1 peptide containing combined M1 ϩ M2 mutations, by surface plasmon resonance, consistent with a prior report that both polybasic NHE1 domains are required for PI(4,5)P 2 binding (29). Absence of M1 ϩ M2 peptide association with PI(4,5)P 2 or PI(3,4,5)P 3 was confirmed by lipid overlay assays (Fig. 2C). To determine the relative contribution of each polybasic NHE1 domain to phosphoinositide binding, membrane overlay assays were conducted with 513-520 (M1) or 556 -564 (M2) NHE1 mutant peptides. These experiments demonstrated equivalent phospholipid binding between both mutants (Fig. 2C), which was not different compared with wildtype NHE1 binding. These results indicate that positively charged amino acids within each domain are required for phosphoinositide binding.
Effect of pH on NHE1-Phosphoinositide Binding-Multiple stimuli cause rapid Na ϩ /H ϩ exchange, which alters pH in the microenvironment of the NHE1 cytosolic tail. To determine whether local pH modulation affects the NHE1-phosphoinositide interaction, surface plasmon resonance experiments were conducted at varying pH values within the physiologic range. Fig. 3, A and B, show enhanced NHE1 association with PI(4,5)P 2 and PI(3,4,5)P 3 with progressive acidification of binding buffer in vitro, suggesting that intracellular acidosis may amplify NHE1 activity by promoting interactions between the exchanger and nearby membrane phosphoinositides.
To verify the effect of cytosolic pH on NHE1-PI(4,5)P 2 binding in live cells, plasma membrane interactions were determined by combined TIRF/FRET using BODIPY-PI(4,5)P 2 and RFP-NHE1 FRET donor/acceptor pairs. Plasma membrane NHE1 and PI(4,5)P 2 are shown by TIRF microscopy (Fig. 3C). At pH 6.5, the fluorescence intensity of BODIPY-PI(4,5)P 2 was decreased in cells expressing RFP-NHE1 (Fig. 3C) due to a FRET signal between the donor and acceptor, confirming their interaction in the plasma membrane (Fig. 3D). FRET was abrogated by increasing pH (Fig. 3D) as seen by no change in the BODIPY-PI(4,5)P 2 fluorescence in cells expressing RFP-NHE1 (Fig. 3C). The data suggest that the plasma membrane PI(4,5)P 2 -NHE1 interaction occurs under acidic conditions, whereas negligible FRET at higher pH is consistent with the quiescent state of NHE1 under ambient conditions. Alteration of cytosolic pH over the physiologic range and duration of NHE1 activity experiments (see Fig. 5) had no effect on cellular PI(4,5)P 2 or PI(3,4,5)P 3 content (supplemental Fig. S2).

FIGURE 2. NHE1 binds phosphoinositides through electrostatic interactions.
Association of diC 8 -PI(4,5)P 2 and diC 8 -PI(3,4,5)P 3 with wild-type and mutant cNHE1 peptides (mutated residues are shown in A), as measured by surface plasmon resonance (B). Mutant cNHE1 binding to phospholipids, by membrane overlay assays (C). In control experiments binding to phosphatidylinositol monophosphates in the absence of cNHE1 peptides was occasionally observed (not shown), indicating a lack of binding specificity to these sites.
Interpretation of the PI(3,4,5)P 3 data is complicated because PI(3,4,5)P 3 simultaneously inhibits anti-apoptotic NHE1 activity, and serves as a docking site for the pro-survival kinase, Akt. To clarify the link between PI(3,4,5)P 3 and NHE1, apoptosis was measured in wild-type mouse PTC in the presence of PI 3-kinase and Akt inhibitors. As expected, inhibition of Akt enhanced apoptosis in wild-type RTC (Fig.   7B). However, combined Akt plus PI 3-kinase inhibition did not increase apoptosis as significantly. Preincubation with the amiloride derivative EIPA, at a concentration specific for NHE1 inhibition (1 M), significantly enhanced apoptosis (Fig. 7B), highlighting the importance of NHE1 function in cell survival.
To confirm the role of phosphoinositides upon NHE1-regulated cell survival, experiments were repeated in NHE1-deficient Swe/Swe cells. Staurosporine-induced apoptosis was modestly enhanced in Swe/Swe (Fig. 7C) compared with wildtype (Fig. 7B) cells. In contrast to wild-type cells, no difference in apoptosis was noted in Swe/Swe cells preincubated with Akt versus Akt plus PI 3-kinase inhibitors (Fig. 7C), indicating that the PI(3,4,5)P 3 interaction with NHE1 is nearly as potent for stimulating apoptosis as the absence of NHE1. As a negative control, no additional apoptosis was observed following EIPA co-incubation.
The findings with PI 3-kinase and Akt inhibitors were corroborated by experiments in wild-type and Swe/Swe mouse PTC loaded with phosphoinositides, which were then induced to undergo apoptosis by staurosporine (Fig. 7, D-G) or cisplatin (supplemental Fig. S5). Apoptosis was measured by TUNEL and activated caspase-3 assays. Importantly, enhanced apoptosis observed in Swe/Swe cells exposed to either staurosporine or cisplatin was unaffected by PI(4,5)P 2 or PI(3,4,5)P 3 supplementation, suggesting that the interaction between NHE1 and phosphoinositides regulates cell survival.

DISCUSSION
We previously demonstrated that the NHE1 Na ϩ /H ϩ exchanger is instrumental in defending against PTC apoptotic stress, but the regulatory mechanisms had not been elucidated. We now describe molecular interactions between the NHE1 cytosolic tail and multiple membrane phospholipids, most notably polyvalent phosphoinositides. These data are largely consistent with previous studies showing that PI(4,5)P 2 binds to NHE1 and stimulates Na ϩ /H ϩ exchange (29).
The 25-50 M binding affinity between PI(4,5)P 2 or PI(3,4,5)P 3 and NHE1 is in agreement with estimated membrane PI(4,5)P 2 concentrations of at least 10 -15 M (36 -38). PI(3,4,5)P 3 content is estimated to be an order of magnitude less that PI(4,5)P 2 under basal conditions, but can increase by 15-fold in response to agonist and PI 3-kinase stimulation (59). Such low affinity binding is purported to be physiologically relevant, by facilitating rapid on-off modulation of target proteins, in contrast to higher affinity interactions, which tend to be irreversible (35). Further evidence of significance can be extrapolated from a recent screen for inhibitors of PH domain binding to PI(3,4,5)P 3 , which yielded peptides with IC 50 values of 30 -40 M and in vivo activity (60). Unlike PH domain-containing proteins, which typically bind phosphoinositides with 1:1 stoichiometry, proteins with polybasic motifs, like NHE1, bind in a multivalent manner to regions of high phosphoinositide density (61,62). We propose that focal, enhanced concentrations of NHE1 and phosphoinositides promote discrete, low affinity electrostatic interactions, which regulate exchanger activity.
The mechanism of NHE1-phosphoinositide association appears to be through electrostatic interaction, as evidenced by the absence of phospholipid binding to NHE1 constructs containing KR/A mutations within two juxtamembrane sites. We and others (29) 3 have shown that the M1 ϩ M2 mutant exhibits diminished Na ϩ /H ϩ exchange activity. Basic residues within each domain were required for phosphoinositide binding, confirming a prior report for NHE1 (29), and consistent with recent Kir2 channel-PI(4,5)P 2 data (63,64). Anionic regions of other proteins, such as ezrin/radixin/moesin also interact with these same cationic NHE1 residues (22,65), highlighting the notion that multiple proteins and lipids could cooperatively regulate NHE1.
Because NHE1 mediates H ϩ extrusion, we tested the effect of pH on the NHE1-phosphoinositide interaction. Binding was optimum at pH 6.5, and lessened with progressive alkalinity, consistent with observations that both acidification and PI(4,5)P 2 binding facilitate Na ϩ /H ϩ exchange, and PI(4,5)P 2 binding to other proteins is amplified at low pH (66). However, pH changes of this magnitude are unlikely to significantly alter binding to KR residues, which have very high pK a , and charge should therefore remain unchanged in the physiologic pH range. On the other hand, histidine has a pK a of 6.5-7.5, and a 544 HYGHHH 549 motif between the KR-rich binding sites has been implicated in NHE1 regulation (67). Histidine protonation could therefore be responsible for minor modulation in observed phosphoinositide binding, either through direct interaction or by altering the polypeptide structure to expose distant binding sites. This is particularly plausible because curve fitting programs indicated that NHE1-phosphoinositide dissociation was not entirely logarithmic (not shown), and suggested a two-component process.
The possibility of NHE1 binding to PS was explored because PS is negatively charged, and PS flipping from the inner to outer plasma membrane leaflet during apoptosis would result in loss of interaction with the NHE1 cytosolic tail. However, significant binding was not observed between NHE1 and PS. A recent report demonstrated that a different C-terminal NHE1 peptide lipid-interacting domain (residues 542-598) bound multiple membrane phospholipids, including PS, using a different assay (31). Our data indicate that the N-terminal charged residues, some of which were deleted from the lipid-interacting domain, may be important for discriminating NHE1-phospholipid binding. However, because vesicles used for our phosphoinositide binding experiments contained 30% PS we cannot exclude the possibility that PS and phosphoinositides may cooperatively bind and regulate NHE1 function, as recently demonstrated for Akt (68).
Because of homology between NHE isoforms, similar regulation by phospholipids might be anticipated. However, in contrast to NHE1, interaction between the NHE3 C terminus and PI(3,4,5)P 3 enhanced Na ϩ /H ϩ exchange (28,30,69), whereas conflicting effects of PI(4,5)P 2 on NHE3 were described (28,30). Moreover, opposite NHE1 and NHE3 activities were observed in response to cell volume perturbations (70 -73), which may be a relevant stimulus with apoptosis, and suggest that phosphoinositide interactions may regulate Na ϩ /H ϩ exchange in an isoform-specific manner.
Based upon our current and previously published data (21,22,45) we propose a model in which PI(4,5)P 2 and PI(3,4,5)P 3 , the major phosphoinositides within the plasma membrane inner leaflet, form an on-off switch to regulate NHE1 (Fig. 8).
The PI(4,5)P 2 -NHE1 interaction promotes NHE1 activity, and is consistent with anti-apoptotic roles of both molecules. The relatively greater magnitude of the PI(4,5)P 2 effect on apoptosis, compared with NHE1-regulated Na ϩ /H ϩ exchange, suggests that PI(4,5)P 2 -dependent cytoprotection is mediated by additional, NHE1-independent mechanisms. We have previously shown that NHE1 activation stimulates PI 3-kinase (22), which leads to Akt activation. Two groups have shown that NHE1 Ser 648 is an Akt substrate (74,75), although the ramifications are unclear, because opposite effects of Ser 648 phosphorylation on NHE1 activity were observed.
In contrast to the straightforward interpretation of the NHE1-PI(4,5)P 2 interaction on cell survival, the role of PI(3,4,5)P 3 is complex, because it potentially promotes apoptosis through NHE1 inhibition, as well as anti-apoptosis by docking PDK1 and Akt. This dichotomy is best illustrated by the different consequences of PI(3,4,5)P 3 modulation in wild-type versus NHE1-null cell lines (Fig. 7). PI 3-kinase inhibition lessened apoptosis in wild-type, but not in NHE1-null cells. Conversely, loading cell membranes with PI(3,4,5)P 3 enhanced apoptosis in wild-type, but not NHE1-null cells, indicating that the isolated effect of PI(3,4,5)P 3 binding to NHE1 is to facilitate apoptosis. However, the kinetics of PI(3,4,5)P 3 -effector interactions differ. NHE1 activity is induced within seconds, which stimulates downstream PI 3-kinase activity (22). NHE1 activity recedes by 3-5 min, which coincides with Akt release to the cytosol (76), where Akt activity peaks at 30 -60 min (22). Therefore, we propose that initially, concomitant NHE1 and PI(3,4,5)P 3 -mediated Akt activation support cell survival. After Akt release, PI(3,4,5)P 3 would become available to bind and modulate NHE1, ensuring that the exchanger returns to its quiescent state.