Proline-rich Motifs of the Na+/H+Exchanger 2 Isoform

The NHE2 isoform of the Na+/H+ exchanger (NHE) displays two proline-rich sequences in its C-terminal region that resemble SH3 (Src homology 3)-binding domains. We investigated whether these regions (743PPSVTPAP750, termed Pro-1, and786VPPKPPP792, termed Pro-2) can bind to SH3 domains and whether they are essential for NHE2 function and targeting. A fusion protein containing the Pro-1 region showed promiscuous binding to SH3 domains of several proteins in vitro, whereas a Pro-2 fusion bound preferentially to domains derived from kinases. In contrast, cytoplasmic regions of NHE1, NHE3, or NHE4 failed to interact. When expressed in antiporter-deficient cells, truncated NHE2 lacking both Pro-rich regions catalyzed Na+/H+exchange, retained sensitivity to intracellular ATP, and was activated by hyperosmolarity, resembling full-length NHE2. The role of the Pro-rich regions in subcellular targeting was examined by transfection of epitope-tagged forms of NHE2 in porcine renal epithelial LLC-PK1 cells. Both full-length and Pro-2-truncated NHE2 localized almost exclusively to the apical membrane. By contrast, a mutant devoid of both Pro-1 and Pro-2 was preferentially sorted to the basolateral surface but also accumulated intracellularly. These observations indicate that the region encompassing Pro-1 is essential for appropriate subcellular targeting of NHE2.

The Na ϩ /H ϩ exchangers (NHEs) 1 are a family of proteins found in virtually all mammalian cells, where they catalyze the electroneutral exchange of intracellular H ϩ for external Na ϩ with a 1:1 stoichiometry (see Refs. 1 and 2 for review). In most cells, NHE activity is important for intracellular pH (pH i ) homeostasis and also for maintenance of normal cellular volume (3). In addition, in epithelia of the kidney, gastrointestinal tract, and other organs, NHE plays a central role in the absorption of NaCl, bicarbonate, and water. Six distinct NHE isoforms have been identified to date in mammalian cells (1). Although the primary sequence homology between the isoforms is limited (20 -60%), they all share the same predicted membrane topology, consisting of 12 membrane-spanning segments at the N terminus and a hydrophilic C-terminal tail thought to extend into the cytoplasm (Ref. 1, but see Ref. 4 for alternative view). Functional characterization of deletion mutants of NHE1 and NHE3 has localized the site of Na ϩ /H ϩ exchange to the N-terminal (transmembrane) region of the protein (1). The C-terminal region is believed to be responsible for modulation of transport by such diverse agents as growth promoters, hormones, and changes in medium osmolarity (1). The cytosolic domain is also thought to encompass the site that confers ATP dependence to Na ϩ /H ϩ exchange, a hallmark of NHE that is common to all the isoforms studied to date (1,2).
The pattern of expression of individual NHE isoforms varies among tissues. NHE1 and NHE6 are expressed in virtually all cells, whereas other isoforms have a more restricted tissue distribution (1). NHE2 is preferentially expressed in the gastrointestinal tract and, to a lesser extent, in the kidney, uterus, and brain (5,6). Despite its discovery several years ago, comparatively little is known about the physiology of NHE2. Its selective expression pattern initially suggested a role for this isoform in fluid and electrolyte balance, similar to the function fulfilled by NHE3 in epithelial cells. However, creation of a null mutation of NHE2 in mice caused no significant perturbations of organismal acid-base or salt homeostasis (7). Instead, the only apparent defect was a significant loss of net acid secretion in the stomach due to a severe reduction in the viability of parietal cells of the gastric mucosa. Although the underlying mechanism is unclear, it was proposed that NHE2 is part of the basolateral transport system that maintains parietal cell volume during high acid secretion at its apical surface and that disruption of NHE2 function causes a chronic state of volume depletion, leading to cellular necrosis (7).
Whereas an attractive hypothesis, the subcellular distribution of NHE2 in stomach has yet to be determined and, indeed, remains a controversial issue in other tissues examined. On one hand, NHE2 activity was reported to be restricted to the basolateral membrane of an inner medullary collecting duct cell line, mIMCD-3 (8). In contrast, others detected NHE2 activity predominantly on the apical membranes of ileal tissue (9) and in a cortical collecting duct cell line (10). Similar observations were made when NHE2 was transiently expressed in a colonic carcinoma cell line (11), and apical localization was also documented immunologically in the medullary thick ascending limb (12). Thus, NHE2 is mainly sorted to the apical surface of most polarized epithelial cells examined, as reported for NHE3. * This work was supported by the Medical Research Council of Canada and the Kidney Foundation of Canada. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
ʈ By contrast, the functional properties of NHE2 differ considerably from NHE3 and more closely resemble those of NHE1 when assessed in transfected mammalian cells: it is activated by agonists of the protein kinase A and C pathways as well as by hyperosmotic-induced cell shrinkage, and its pH i sensitivity is moderately reduced by cellular depletion of ATP (3,(13)(14)(15).
The structural determinants that confer this unique behavior have not been identified. Perusal of the primary structure of NHE2 revealed that the C-terminal region, which is most divergent between isoforms, contains two proline-rich motifs: residues 743-750 (PPSVTPAP), which will be called hereafter Pro-1, and residues 786 -792 (VPPKPPP), designated Pro-2. The latter conforms to the consensus sequence of proteins capable of binding Src homology 3 (SH3) domains, PpXP, where is a hydrophobic residue, X is any amino acid, and p tends to be (but is not always) Pro (16 -18). Pro-1 approximates but does not match perfectly the SH3 consensus structure. These proline-rich regions are unique to NHE2 and may be important in defining the distinctive behavior of this isoform.
SH3 domains are sequences of 50 -75 amino acids found in diverse proteins that include cytoskeletal components, such as spectrin (19), adaptor and signaling molecules, such as Grb2 (20,21) and Ras-GAP (22), and protein and lipid kinases, including Src, Abl, and the p85 subunit of phosphatidylinositol 3Ј-kinase (23,24). Coupling of proteins via their SH3 domains has been implicated in a variety of functions, including regulation of cell growth and proliferation (25,26), endocytosis (27,28), and activation of the respiratory burst (29,30). The importance of SH3 domains in other systems prompted us to analyze whether the Pro-rich motifs of NHE2 are capable of interacting with SH3-containing proteins and to assess the functional role of this interaction. To this end, we studied the ability of fusion proteins encoding the Pro-1 and Pro-2 regions of NHE2 to bind SH3 domains in vitro. In addition, we used transfection of epitope-tagged constructs to compare the functional properties and subcellular distribution of the full-length NHE2 with those of mutants lacking either Pro-1 or Pro-2.
Monoclonal antibodies to a peptide derived from the influenza virus hemagglutinin (HA) were purchased from BAbCo. Goat anti-mouse Cy3-conjugated IgG and goat peroxidase-coupled anti-mouse secondary antibody were purchased from Jackson Immunoresearch Laboratories. Horseradish peroxidase coupled to avidin was purchased from Cappel.
Phosphate-buffered saline (PBS) contained 140 mM NaCl, 10 mM KCl, 8 mM sodium phosphate, 2 mM potassium phosphate, pH 7.4. The sodium chloride solution contained 117 mM NaCl, 1.66 mM MgSO 4 , 1.36 mM CaCl 2 , 5.36 KCl mM, 25 mM Na-HEPES, 5.55 mM glucose, pH 7.4. Sodium-free solutions were made with equimolar substitution of Nmethyl-D-glucammonium solutions of the appropriate salts. Unless otherwise indicated, all solutions were nominally bicarbonate-free and were adjusted to 290 Ϯ 10 mosM with the major salt. Hypertonic NaCl solution was prepared by addition of sufficient NaCl to raise the osmolarity to 600 Ϯ 10 mosM.
Preparation of GST Fusion Proteins-GST fusion proteins of the Pro-rich regions of NHE2 were prepared by PCR amplification of the appropriate regions of rat NHE2 cDNA (31). The PCR product encompassing the more N-terminal Pro-rich region, which contains amino acids 743 PPSVTPAP 750 (Pro-1) plus 19 and 16 flanking residues at the 5Ј-and 3Ј-ends, respectively, was amplified using the primers 5Ј-ccggatcctcaccagcctac-3Ј and 5Ј-ggaattcgggctgcctcag-3Ј. The primers used to amplify the more C-terminal proline-rich region containing amino acids 786 VPPKPPP 792 (Pro-2) plus 12 and 21 flanking residues at the 5Ј-and 3Ј-ends, respectively, were 5Ј-ccggatccggccggggcagg-3Ј and 5Ј-ggaattcaccacaagtctgc-3Ј. The PCR products were purified following electrophoresis in 1.8% agarose (Nu-Sieve Agarose, Biocan) using the Qiaex DNA extraction kit, according to the manufacturer's directions. The Pro-1 and Pro-2 sequences were subsequently subcloned into the BamHI and EcoRI sites of pGEX-2T (Amersham Pharmacia Biotech). HB101 E. coli transformed with the resulting plasmid were grown at 37°C in Luria broth supplemented with 50 g/ml ampicillin to logphase, and then expression of the fusion protein was induced with 1 mM isopropyl-1-thio-␤-D-galactopyranoside for 5 h at 37°C. Bacteria were lysed in the presence of protease inhibitors (10 g/ml phenylmethylsulfonyl fluoride, 1 g/ml leupeptin, 1 g/ml pepstatin, 1 g/ml aprotinin, and 5 mM EDTA), and the fusion proteins were purified from the lysate using glutathione-agarose beads. GST fusion protein encompassing amino acids 560 -690 of NHE3 and 667-717 of NHE4 were prepared similarly. A GST fusion protein containing the last 178 residues of the cytosolic tail of rabbit NHE-1 was the kind gift of Dr. L. Fliegel (University of Alberta, Edmonton, Alberta, Canada).
Fusion proteins of the SH3 domain of c-Abl, p85, Ras-GAP, N-Src, Src, and ␣-spectrin were obtained and generated as described previously (32). GST-SH3 domain fusions were subsequently biotinylated using NHS-LC biotin (Pierce), as described (19). ϾIn Vitro Far Western Binding Assays-Fusion proteins (8 g) of the indicated regions of the exchangers were separated by SDS-PAGE (12% acrylamide), transferred to nitrocellulose, and incubated overnight in blocking buffer (0.25% gelatin, 10% ethanolamine, 0.1 M Tris, pH 9.0) supplemented with 5% powdered milk. The blots were next incubated with 2 g of the appropriate biotinylated SH3 fusion protein in 10 ml of blocking buffer for 2 h. The blots were then washed and incubated for 2 h at room temperature with streptavidin-peroxidase (12.5 g/ml) in a buffer containing 0.25% gelatin, 0.05% Nonidet P-40, 0.15 M NaCl, 5 mM EDTA, 50 mM Tris, pH 7.5, followed by washing and detection using ECL.
cDNA Constructs-The rat NHE2 cDNA was engineered to contain a series of unique restriction endonuclease sites that did not alter the primary structure, but created convenient DNA cassettes for mutagenesis. The modified NHE2 cDNA was inserted into the pCMV mammalian expression vector (plasmid renamed pNHE2) as described previously (6,33). To allow for immunological detection of the protein, the influenza virus HA epitope YPYDVPDYAS, preceded by a single G amino acid linker (inserted to create peptide flexibility), was inserted at the very C terminus of NHE2 (after amino acid 813) using PCR amplification of a C-terminal cDNA fragment. Similarly, the NHE2⌬777 HA and NHE2⌬731 HA deletion mutants were engineered by inserting the HA epitope plus a stop codon immediately following positions 777 and 731 of the wild type transporter. The PCR fragments were sequenced prior to substitution into NHE2 to confirm the presence of the mutations and to ensure that other random mutations were not introduced.
Cells-AP1 is a cell line devoid of endogenous Na ϩ /H ϩ exchange activity that was derived from wild type Chinese hamster ovary (CHO) cells by the "H ϩ -suicide" technique (34). These cells and the transfectants derived thereof were grown in ␣-minimal essential medium (Ontario Cancer Institute, Toronto, Ontario, Canada). HEK-T is a subclone of the transformed human embryonic kidney (HEK) cell line containing the large T antigen. LLC-PK 1 are epithelial cells derived from porcine kidney proximal tubule. HEK-T and LLC-PK 1 cells were grown in Dulbecco's minimal essential medium (Ontario Cancer Institute, Toronto, ON). Both ␣-minimal essential medium and Dulbecco's minimal essential medium contained 25 mM NaHCO 3 and were supplemented with 10% fetal calf serum, 100 units/ml penicillin, and 100 g/ml streptomycin (Life Technologies, Inc.). Cells were incubated in a humidified environment containing 95% air and 5% CO 2 at 37°C. Cultures were re-established from frozen stocks regularly, and cells from passages 3-20 were used for the experiments.
Where indicated, intracellular ATP was depleted by incubating the cells for 10 min in glucose-free medium with 5 mM 2-deoxy-D-glucose and 1 g/ml antimycin A, to inhibit both glycolysis and oxidative phosphorylation. This protocol has previously been shown to deplete Ͼ90% of the ATP in CHO cells within 10 min (35). Subsequent fluorescence measurements were performed in glucose-free medium containing 5.5 mM 2-deoxy-D-glucose.
Transfection and Selection-AP-1, LLC-PK 1 , and HEK-T cells were transfected with plasmids containing the epitope-tagged wild type or truncated NHE2 constructs by the calcium phosphate-DNA co-precipitation technique of Chen and Okayama (36). For selection of stable lines, the AP-1 cells were selected for survival by repeated acute acid loads (5-6 times over a 2-week period), in order to discriminate between Na ϩ /H ϩ exchanger-positive and negative-transfectants, starting 48 h after transfection (33,34). For stable transfection of LLC-PK 1 cells, which express endogenous NHE, we subcloned NHE2 HA , NHE2⌬777 HA , and NHE2⌬731 HA into the pBK vector (Stratagene), which contains the neomycin resistance gene. Stably expressing cells were selected by incubation with 500 g/ml G418 and then screened by immunofluorescence for expression of the wild type or mutant NHEs.
Immunofluorescence-Cells were grown to subconfluence on sterile 18 mm glass coverslips (Thomas Scientific, Swedesboro, NJ), washed with PBS, and fixed for 20 min with 4% paraformaldehyde, and the excess formaldehyde was quenched with 100 mM glycine for 15 min. Where indicated, the fixed cells were stained with a mixture of fluorescein isothiocyanate-labeled lectins (peanut, orange, wheat germ, and pea, 2 g/ml each) for 45 min at 4°C. The cells were then permeabilized with 0.1% Triton X-100 supplemented with 5% bovine serum albumin and incubated with anti-HA antibody (1:1000) at room temperature for 45 min. The cells were washed again with PBS and incubated with Cy3-conjugated anti-mouse antibody (1:5000) for 45 min. The coverslips were mounted using Dako mounting reagent and visualized using a Leica fluorescence microscope. Images were obtained using the Winview program and processed using Adobe Photoshop.
Measurements of Cytosolic pH-The pH i of small groups of cells was determined by microphotometry of the fluorescence emission of BCECF using dual wavelength excitation. Cells grown to confluence on 25-mm glass coverslips (Thomas Scientific, Swedesboro, NJ) were loaded with BCECF by incubation with 2 g/ml of the precursor acetoxy-methylester form for 10 min at 37°C. The coverslips were then mounted in a Leiden coverslip dish (Medical System Corp., Greenvale, NY) and placed into a thermostatted holding chamber heated to 37°C (open perfusion microincubator, Medical Systems Corp., Greenvale, NY) attached to the stage of a Nikon Diaphot TMD inverted microscope (Nikon Canada, Toronto, Ontario, Canada). Cells were visualized using a Nikon Fluor ϫ 40/1.3 numerical aperture oil immersion objective and a Hoffman modulation contrast video system with an angled condenser (Modulation Optics) through a CCD-72 video camera and control unit (Dage-MTI, Michigan City, IN) connected to a Panasonic monitor. Clusters of 6 -12 cells from the confluent culture were selected for analysis with an adjustable diaphragm.
The chamber was continuously perfused at ϳ 0.5 ml/min to allow for complete exchange of the bath solution once every minute using a gravity-driven system and a Leiden aspirator. When rapid solution changes were required, three aliquots of 1 ml of the new medium were quickly pipetted (Ͻ15 s) into the chamber and perfusion was continued using the new medium (37). Fluorescence measurements were made using an M Series dual wavelength illumination system from Photon Technologies, Inc. (South Brunswick, NJ) in a dual excitation/single emission configuration. Excitation light provided by a Xenon lamp was alternately selected using 495 Ϯ 10 and 445 Ϯ 10 nm filters (Omega Optical, Brattleboro, VT) at a rate of 50 Hz and then reflected to the cells by a 510 nm dichroic mirror. Emitted light was first selected by a 520 nm long-pass filter and then separated from the red light used for Hoffman imaging by a 550 nm dichroic mirror and directed to the photometer through a 530 Ϯ 30 nm band-pass filter. This optical system allowed for continuous visualization of cells without interfering with fluorescence measurements. Photometric data was acquired at 10 Hz using a 12 bit A/D board (Labmaster, National Instruments, Austin, TX) interfaced to a Dell 486 computer and analyzed with the Felix software (Photon Technologies Inc., South Brunswick, NJ). Calibration of the fluorescence intensity to pH i was performed in the presence of 5 mM nigericin in high potassium medium (140 mM KCl, 20 mM HEPES, 1 mM MgCl 2 , and 5 mM glucose) as detailed previously (38). Each coverslip was calibrated at the end of the experiment using at least three pH values. Quantification of cell-associated fluorescence was performed using the Felix software package (Photon Technologies, Inc., South Brunswick, NJ). The rate of pH i change was derived by linear regression of the pH i versus time curve over 4 s intervals using the Origin software (MicroCal Software Inc., Northampton, MA).

Interaction of Proline-rich Regions with SH3 Domains in
Vitro-To assess whether SH3 domains are capable of binding specifically to the Pro-rich motifs of the cytosolic domain of NHE2 we initially tested the ability of the Src-SH3 domain to bind to this region in an in vitro overlay assay. A GST fusion protein encompassing residues 724 -766 and including the Pro-1 region ( 743 PPSVTPAP 750 ) of NHE2 was subjected to SDS-PAGE, transferred to nitrocellulose, and overlaid with the biotinylated SH3 domain of Src, and binding was detected with streptavidin-horseradish peroxidase and ECL. In order to assess the specificity of the interaction, GST alone and GST fusion proteins prepared from the cytoplasmic segments of other NHE isoforms, which have no Pro-rich domains, were analyzed simultaneously. As illustrated in Fig. 1, the Src SH3 domain bound to the GST-NHE2-Pro-1 fusion protein but failed to interact with either GST alone or with the cytoplasmic regions of NHE1, NHE3, or NHE4. This observation suggests that SH3 domains bind to the cytosolic segment of the exchangers only when Pro-rich sequences are present.
We next compared the ability of the Pro-1 and Pro-2 regions of NHE2 to associate with a variety of SH3 domains, i.e. those of Abl, Ras-GAP, p85 of phosphatidylinositol 3Ј-kinase, ␣-spectrin, Src, or the neuronal (N-Src). As before, GST alone was included as a measure of the specificity of the interaction. Our results show that none of the indicated SH3 domains interacted measurably with GST (Fig. 2). By contrast, all of these bound to the Pro-1 fusion protein, with Ras-GAP interacting most strongly, whereas N-Src bound only marginally. The Pro-2 domain was less promiscuous, interacting strongly with the SH3 domains of p85 and Src, only moderately with N-Src and Abl, but not measurably with either Ras-GAP or ␣-spectrin. It therefore appears that both Pro-rich motifs of NHE2 can bind Functional Role of Pro-rich Regions of NHE2-To assess the functional role of the Pro-rich regions of NHE2, we compared the behavior of the full-length protein with that of truncated mutants lacking one or both Pro-rich regions. To this end, vectors encoding epitope-tagged intact or truncated constructs were transfected into mammalian cells and their expression and transport properties were evaluated. As a preliminary experiment to evaluate the immunodetectability of the epitopetagged constructs, the cDNAs were transiently transfected into HEK-T cells. As illustrated in Fig. 3, transfection of the fulllength NHE2 (NHE2 HA ) construct resulted in expression of a 95-105-kDa polypeptide that was recognized by an antibody directed to the HA epitope, which was attached at the C terminus of the protein. Transfection of constructs truncated at position 777 (NHE2⌬777 HA ), lacking the C-terminal Pro-2, or at position 731 (NHE2⌬731 HA ), lacking both Pro-1 and Pro-2, induced the expression of polypeptides of ϳ90 -95 and ϳ85-92 kDa. The levels of expression of the full-length and truncated molecules were comparable and the observed differences in molecular mass are consistent with the deletion of 36 and 82 residues, respectively.
The functional properties of the wild type and mutant NHE2s were studied next. To quantify the ion exchange mediated by the intact and truncated forms of NHE2, it was imperative to dissociate their activity from that of endogenous Na ϩ /H ϩ exchangers present in the cells used for transfection. This was accomplished by stably expressing the cDNAs in AP-1 cells, a mutant line of CHO cells devoid of NHE activity (34). Transport activity was estimated as the Na ϩ -dependent changes in pH i , measured fluorometrically using BCECF. In all experiments, pH i was measured after overnight serum deple-tion to minimize confounding effects of growth factors in the regulation of NHE2 activity.
The activity of the full-length and truncated forms of NHE2 was initially compared in cells that were acutely acid-loaded by preincubation with 25 mM NH 4 Cl for 10 min, followed by rapid removal of the weak base. When bathed in Na ϩ -free, N-methyl-D-glucammonium ϩ -rich medium, the cells remained acidic, indicating that Na ϩ -independent systems contribute negligibly to pH i restoration in the time frame studied. As reported earlier, re-addition of Na ϩ elicited a rapid cytosolic alkalinization in cells expressing the original unmodified full-length NHE2, and a comparable pH change was obtained when NHE2 was epitope-tagged, indicating that addition of the C-terminal HA nonapeptide did not interfere with the basic function of the antiporter (Fig. 4). Similar recoveries from the acid load were obtained in cells transfected with NHE2⌬777 HA or with NHE2⌬731 HA . Comparison of the relative rates of recovery from the acid load indicates that the pH i dependence of the exchange process was not markedly affected by truncation of either one or both Pro-rich domains. The full-length NHE2, as well as the truncated versions, became minimally active between pH i 7.0 and 7.5, and their activity increased steeply at more acidic pH i (Figs. 4 and 5). NHE2⌬777 HA appeared to become quiescent at a somewhat more acidic pH than the other constructs. We did not quantify the absolute rates of transport because this parameter is not informative when comparing expression levels in different stable transfectants.
Although the basal operation of NHE2 is not grossly affected by deletion of the Pro-rich regions, it is possible that its regulation may be altered. Intracellular ATP is an important regulator of NHE2 activity (39). Although the precise mechanism of action of ATP is not well defined, depletion of the nucleotide is known to depress the activity of NHE2, shifting its activation threshold to more acidic pH values. We analyzed whether the Pro-rich motifs influenced this regulation. As illustrated in Fig.  5A, insertion of the HA epitope at the C terminus of the fulllength NHE2 did not modify the responsiveness of the exchanger to metabolic depletion: the rate of recovery from an acid load was greatly reduced by preincubation for 10 min in glucose-free medium with 5 mM 2-deoxy-D-glucose and 1 g/ml antimycin A, conditions shown earlier to deplete ATP in CHO cells by Ͼ90% within 10 min (35). A comparable inhibition of exchange was noted in NHE2⌬777 HA and in NHE2⌬731 HA transfected cells (Fig. 5, C and D). Therefore, the presence of the Pro-rich sequences is not essential for regulation of NHE2 by ATP. This result is not entirely unexpected because other NHE isoforms are also regulated by ATP, albeit to differing extents, but lack obvious SH3-binding domains.
Osmotic Activation of NHE2-In addition to its important role in maintaining pH i homeostasis, the NHE family of proteins plays an important role in volume regulation and H 2 O transport. In response to hypertonic stress, NHE2 is activated in transfected CHO and porcine proximal tubule LLC-PK 1 cells (Refs. 3 and 15, but see Ref. 40 for a differing view). The mechanism by which hypertonicity modulates antiport activity is not known, but cytoskeletal involvement is suspected. Because several cytoskeletal components contain SH3 domains, we considered the possibility that the cognate Pro-rich motifs may play a role in volume regulation. For these experiments, cells incubated in isotonic medium were challenged with a hypertonic (600 Ϯ 20 mosM) Na ϩ -rich solution. As illustrated in the representative traces shown in Fig. 6, the hypertonic response described earlier for the full-length NHE2 persists in the epitope-tagged construct. A clear osmotic stimulation was also noted in cells transfected with NHE2⌬731 HA . By comparison, the response recorded in cells expressing NHE2⌬777 HA was considerably smaller. This depressed response is not likely due to a reduced number of copies of the exchanger, because the rate of recovery from an acid load was comparable to that of the other forms. Instead, it may be a consequence of the acidic shift in the pH i dependence of this truncated form. Alternatively, removal of Pro-2 may have allowed interaction of an inhibitory SH3 domain with Pro-1. The inhibition caused by this interaction would be relieved upon truncation of Pro-1. In any event, because the hypertonic response persists in NHE2⌬731 HA cells, which lack both Pro-1 and Pro-2, it is clear that the SH3binding domains are not essential for the osmotic activation of NHE2.
Localization of Full-length NHE2 and Its Truncation Mutants in LLC-PK 1 Cells-Because the Pro-rich regions of NHE2 were seemingly not required for either basal Na ϩ /H ϩ exchange activity or for its regulation by ATP or cell volume, we considered the possibility that they may be involved in subcellular targeting. As mentioned in the introduction, NHE2 is abundant in epithelial cells of the kidney and gastrointestinal tract. In epithelial cells, where proper targeting of transporters is needed for vectorial movement of electrolytes and fluid (41,42), the distribution of individual NHE isoforms is polarized. NHE3 is exclusively apical, whereas NHE1 is almost entirely basolateral (43)(44)(45)(46). More uncertainty exists concerning NHE2, which has been reported by some to concentrate apically (9 -11), whereas others have found that it functions at the basolateral FIG. 4. Na ؉ -induced acid extrusion from cells transfected with full-length and truncated NHE2 . Antiport-deficient Chinese hamster ovary cells (AP-1) were stably transfected with full-length NHE2, with or without a C-terminal HA tag (NHE2 HA and NHE2, respectively), or with epitope-tagged truncated forms lacking either Pro-2 only (⌬777 HA ) or both Pro-1 and Pro-2 (⌬731 HA ). Cells were grown to subconfluence on glass coverslips and loaded with BCECF as described under "Experimental Procedures." The cells were acid-loaded by prepulsing with 25 mM NH 4 Cl, and pH i determination was initiated upon perfusion with Na ϩ -free solution. Where indicated, the solution was replaced by Na ϩ -rich medium. pH i was determined by ratio microfluorometry and calibrated using nigericin/K ϩ . Samples that were acidloaded to slightly different levels are illustrated, to offset the traces and avoid overlap. Shown are representatives of at least four determinations of each type. membrane (8).
To re-assess the localization of full-length NHE2 and to define the role of the Pro-rich motifs in the establishment of polarity, we transfected HA-tagged constructs into LLC-PK 1 cells. Stably transfected cells were obtained by subcloning the constructs into the pBK plasmid that encodes for neomycin resistance, and selected using G418. Stable transfectants were grown to confluence and analyzed by immunofluorescence and confocal laser scanning microscopy using anti-HA antibodies followed by Cy3-conjugated secondary antibodies. Prior to fixation, the apical membrane was counter-stained by overlaying the cells with a mixture of fluoresceinated lectins (see under "Experimental Procedures"). As illustrated in Fig. 7, both the full-length (NHE2 HA ) and the NHE2⌬777 HA , which lacks the C-terminal Pro-rich domain (Pro-2), displayed a punctate appearance that followed the contour of the apical membrane. Accordingly, the distribution of these proteins matched closely the distribution of apically bound lectins (cf. Fig. 7, B and C). There was little evidence of intracellular retention of either the full-length NHE2 HA or NHE2⌬777 HA , despite wide variation in the level of expression among individual clones.
By contrast, a large fraction of the cells expressing NHE2⌬731 HA appeared to retain the construct in the endoplasmic reticulum and/or Golgi complex (Fig. 7D). This was not due to clonal variation or differences in the level of expression, because similar results were obtained in multiple clones with varying degrees of expression. The fraction of cells that successfully exported NHE2⌬731 HA to the plasma membrane displayed a distinct distribution pattern. When viewed en face, the truncated exchangers appeared to delineate the cell borders, a pattern characteristic of basolateral proteins. Accordingly, this distribution differed from that of the apical lectins (cf. Fig. 7, E and F) and was instead similar to that of ZO-1, a tight-junctional protein (not illustrated). Only a small fraction of NHE2⌬731 HA appeared to be present at the apical membrane.
A quantitative summary of these observations, obtained from 10 separate observations made in polyclonal populations, is presented in Fig. 8. The data confirm that whereas NHE2 HA and NHE2⌬777 HA are predominantly apical, a large fraction of NHE2⌬731 HA is retained intracellularly, and those truncated exchangers that reach the membrane localize predominantly to the basolateral side of the cell. We conclude that the Pro-1 motif is important for effective processing and targeting of NHE2 to the apical membrane of epithelial cells. DISCUSSION The observations presented in this report provide evidence that the cytoplasmically oriented Pro-rich regions of NHE2 can bind to SH3 domains. Pro-2 bound preferentially to SH3 domains derived from tyrosine kinases and from the regulatory domain of phosphatidylinositol 3Ј-kinase, whereas Pro-1 was more promiscuous, interacting with a wider variety of SH3 domains. Although promiscuous, this interaction was nevertheless specific to SH3 domains, inasmuch as it was not observed for GST alone, nor for fusion proteins derived from the cytosolic domains of other NHE isoforms, which lack Pro-rich sequences.
The functional significance of these interactions was evaluated by comparing the behavior of intact and truncated forms of NHE2, using heterologous expression systems. Truncation of the C-terminal Pro-2 region (NHE2⌬777 HA ) had subtle yet reproducible functional consequences. The pH i dependence of FIG. 6. Osmotic responsiveness of full-length and truncated NHE2. pH i was measured fluorometrically in AP-1 cells transfected with either full-length or truncated forms of epitope-tagged NHE2, as in Fig. 4. The cells were initially suspended in isotonic, Na ϩ -rich medium (Iso). Where specified, the osmolarity of the medium was increased to 600 Ϯ 10 mosM (Hyper). Traces are representative of at least four experiments.
FIG. 7. Localization of full-length and truncated NHE2 in LLC-PK 1 cells. LLC-PK 1 cells were stably transfected with NHE2 HA (A), NHE2⌬777 HA (B and C), or NHE2⌬731 HA (D-F) using the calcium phosphate method and selected with G418 as described under "Experimental Procedures." The cells were grown to confluence on coverslips, washed with PBS prior to fixation with 4% paraformaldehyde, and subsequently labeled with fluorescein isothiocyanate-labeled lectins. After extensive washing, the cells were permeabilized with 0.1% Triton X-100 in 5% bovine serum albumin and labeled with 1:1000 anti-HA antibody, followed by secondary labeling was with 1:1000 Cy3 labeled anti-mouse IgG (A, C, D, and F). Fluorescence microscopy was used to detect the distribution of the lectin (B and E) or of the HA epitope, indicative of NHE2. Images are representative of at least five experiments. Bar in D is 10 m, and all panels are the same magnification. the rate of exchange was modestly shifted to more acidic values, and the osmotic activation was depressed. However, an even more profound truncation (NHE2⌬731 HA ) that eliminated both Pro-1 and Pro-2 restored both the normal pH i dependence and osmotic responsiveness. This biphasic behavior is not unprecedented. Moderate truncation of the C-terminal domain of NHE1 can produce an alkaline shift in its pH i dependence, with loss of responsiveness to a variety of stimuli (47,49). The shift and loss of responsiveness are reversed by more profound truncations, which can in fact induce an acidic shift in the pH i versus activity relationship (50,51). Thus, the exchangers can seemingly exist in several conformations that are exquisitely dependent on subdomains of the cytosolic tail.
Despite the subtle effects noted upon deletion of Pro-2, it is clear that neither Pro-1 nor Pro-2 is absolutely essential for the ability of NHE2 to exchange Na ϩ for H ϩ , nor for its dependence on ATP or sensitivity to osmotic activation. This conclusion stems from the observations made using the NHE2⌬731 HA construct, which displayed near-normal behavior under these conditions. We therefore considered the possible role of the Pro-rich regions in protein processing and subcellular targeting. This initially required confirmation of the subcellular distribution of NHE2 in epithelial cells, which has been the source of controversy. By transfecting the full-length, epitope-tagged NHE2 into LLC-PK 1 cells, we found that this isoform accumulates in the apical membrane, as reported by the majority of studies (9 -12). The source of the discrepancies with the results of Soleimani et al. (8), who detected basolateral NHE2, remains unclear. Neither tissue nor species differences appear to account for the inconsistency, because both apical and basolateral locations have been reported for NHE2 in renal medullary cells of rodents (8,12).
Despite these uncertainties, comparison of the full-length and truncated constructs in a single expression system provided useful information concerning the role of the Pro-rich regions in targeting. Briefly, we found that whereas deletion of Pro-2 was ineffectual, the additional removal of Pro-1 greatly impaired the ability of NHE2 to reach the apical membrane. A sizable fraction of the truncated exchanger was retained in the secretory pathway, and the exchanger that was exported to the plasmalemma was primarily found in the basolateral membrane. It is tempting to speculate that the existence of splice variants or posttranslational truncation of NHE2 may account for the reports in which this isoform was detected basolaterally.
These observations suggest that the cytosolic region between residues 732 and 777, encompassing Pro-1, is necessary for appropriate targeting of NHE2 to the apical surface. In this context, Pro-1 was found to bind to the SH3 domain of ␣-spectrin, a protein that has previously been reported to participate in the apical targeting of epithelial Na ϩ channels (ENaCs) (19). Like NHE2, ENaC interacts with the SH3 domain of ␣-spectrin in vitro. Moreover, microinjection of a fusion protein containing the Pro-rich region of ENaC revealed targeting to the apical membrane of epithelial cells (19). Based on these observations, it was suggested that ␣-spectrin may function to retain ENaC apically, a mechanism that may also apply in the case of NHE2. Alternatively, interaction with spectrin present in the Golgi (48) may be essential for export of the exchangers and channels to the plasmalemma. Clearly, other SH3-containing proteins may also function in apical delivery and retention of NHE2. A complete understanding of the role of Pro-rich domains awaits identification of all the proteins that interact with the exchanger in vivo. FIG. 8. Quantitative estimation of the subcellular distribution of full-length and truncated NHE2. LLC-PK1 cells transfected with full-length or truncated forms of NHE2 were stained and analyzed by epifluorescence microscopy using a ϫ 100 objective as in Fig. 7. The subcellular distribution of each one of the tagged exchangers was assessed in 10 random samples. Data summarize the aggregate observations.