The Epithelial Na+/H+ Exchanger, NHE3, Is Internalized through a Clathrin-mediated Pathway*

Trafficking of the Na+/H+ exchanger isoform 3 (NHE3) between sub-apical vesicles and apical membrane of epithelial cells is a suggested mechanism of regulation of NHE3 activity. When epitope-tagged NHE3 was stably expressed in NHE-deficient Chinese hamster ovary cells, a sizable fraction was found in recycling endosomes. This system was used to analyze the mechanism of endocytosis of NHE3. Immunofluorescence and radiolabeling experiments showed that inhibition of clathrin-mediated endocytosis using hypertonicity, acid treatment, or K+ depletion inhibited internalization of NHE3. Moreover, transient transfection of an inhibitory mutant of dynamin (DynS45N) blocked the clathrin-mediated uptake of transferrin, as well as the endocytosis of NHE3. In ileal villus cells, endogenous NHE3 was also found to co-purify with isolated clathrin-coated vesicles, thereby confirming their association in native tissues. The role of COP-I subunits in the intracellular traffic of NHE3 was evaluated usingldlF cells, which bear a temperature-sensitive mutation in the ε-COP subunit. At the permissive temperature, NHE3 distributed normally, whereas at the restrictive temperature, which induces rapid degradation of ε-COP, NHE3 was still internalized, but its subcellular distribution was altered. These results indicate that endocytosis of NHE3 occurs primarily via clathrin-coated pits and vesicles and that normal intracellular trafficking of NHE3 involves an ε-COP-dependent step.

The Na ϩ /H ϩ exchangers (NHE) 1 are a family of integral membrane proteins that translocate Na ϩ in exchange for H ϩ across the cell surface and mitochondrial membranes. Na ϩ /H ϩ exchange is electroneutral and is therefore driven by the combined chemical gradients of Na ϩ and H ϩ . NHE plays an important role in the maintenance of intracellular pH (pH i ) and in the regulation of cell volume. In epithelial cells, where the exchangers are particularly abundant, they are essential for the vectorial transport of salt and water. To date, six mammalian NHE isoforms have been identified (reviewed in Ref. 1). NHE1 is widely expressed and is generally considered to be the "housekeeping" isoform primarily responsible for pH i homeostasis. NHE2-5 have a more restricted tissue distribution, whereas NHE6 is ubiquitous and present in mitochondria. Of these, NHE2 and NHE3 are predominantly epithelial isoforms, which are highly expressed in the gastrointestinal tract and kidney (2)(3)(4), where they are thought to be important for Na ϩ , bicarbonate and water (re)absorption (5)(6)(7).
Multiple NHE isoforms can co-exist in epithelial cells, where they are segregated to distinct subcellular domains. NHE1 is located almost entirely in the basolateral side of the cell (8,9), whereas NHE2 has been reported to be present in both apical and basolateral membranes (10 -12). By contrast, NHE3 is exclusively targeted to the apical domain of the cells (13)(14)(15). Interestingly, unlike NHE1 and NHE2 which are expressed primarily at the plasma membrane, immunolocalization experiments have revealed a significant intracellular pool of NHE3. In native epithelia (16) as well as in cultured epithelial cells (11)(12)(13), NHE3 was detectable not only at the cell surface, but also in a population of subapical vesicles. These direct immunolocalization studies confirmed earlier suggestions that NHE was also present and functional in endomembranes. Fractionation studies in conjunction with either transport (17)(18)(19) or immunological assays (20,21) had previously found that NHE was detectable in membranes other than the plasmalemma. Na ϩ /H ϩ exchange in these endomembranes was reported to be amiloride-resistant (19), a characteristic of the NHE3 isoform.
An intracellular pool of antiporters may provide a source for the rapid mobilization of transporters to the plasma membrane in response to stimulation, thereby affording the cell a powerful and versatile mechanism for regulating NHE3 activity. Indeed, chronic acidosis, prolonged hyperosmolarity, and responsiveness to angiotensin II have been shown to increase plasmalemmal NHE3 activity, with minimal increase in mRNA expression (22)(23)(24). Shuttling of NHE3 between the plasmalemmal and endosomal pools may well be responsible for this effect. Despite the apparent importance of NHE3 internalization and recycling, little is known about the underlying mechanisms. This is due in large part to the difficulties involved in discerning the internal and superficial pools of NHE3 in epithelial systems, which are also rather refractory to transfection.
The partition of NHE3 in an intracellular vesicular compartment was replicated when NHE3 was heterologously transfected in Chinese hamster ovary (CHO) cells (25). While exchangers were present and active at the plasma membrane, a sizable fraction was found in a juxtanuclear vesicular complex that co-localized with internalized transferrin and with cellubrevin, hallmarks of the recycling endosomal compartment. By transfecting a modified NHE3 bearing an external epitope, Kurashima et al. (26) were able to quantify the fraction of exchangers in endomembranes and to monitor their traffic. They found that NHE3 is internalized continuously and that recycling to the surface is responsible for maintaining a constant number of plasmalemmal transporters.
Several mechanisms are utilized to internalize plasma membrane proteins, including clathrin-mediated coated vesicle formation (27), generation of caveolae (28,29), or other uncoated vesicles (28,30), and actin-dependent macropinocytosis. The pathway(s) utilized by NHE3 have not been defined. The purpose of the current study was to identify the mechanism(s) that mediate NHE3 internalization and to analyze the factors required for recycling of the internalized exchangers to the plasma membrane. To this end, we used cells heterologously transfected with rat NHE3 that was externally tagged with a triple HA epitope. Pharmacological and transfection experiments indicate that NHE3 is largely internalized via clathrincoated pits and vesicles. The presence of NHE3 in coated vesicles was confirmed in intestinal cells from rabbits, implying that clathrin is also involved in NHE3 internalization in native epithelia.

EXPERIMENTAL PROCEDURES
Materials and Media-Nigericin, rhodamine-conjugated human transferrin (Tfn), and the acetoxymethyl ester of 2Ј7Ј-bis-(2-carboxyethyl)-5(and 6)-carboxyfluorescein (BCECF) were purchased from Molecular Probes, Inc. (Eugene, OR). The plasmid containing the enhanced green fluorescent protein (pEGFP) was purchased from CLONTECH (Palo Alto, CA). Monoclonal antibodies to clathrin, adaptor proteins 1 and 2 (AP-1 and AP-2), and all other chemicals were purchased from Sigma. Monoclonal antibodies recognizing the heavy chain of clathrin and the influenza virus HA epitope (YPYDVPDYAS) were from ICN and Babco (Richmond, CA), respectively. The anti-NHE3 antibody (Clone 2B9) was a kind gift from Dr. D. Biemesderfer (Yale University School of Medicine, New Haven, CT). Goat anti-mouse and anti-rabbit Cy3-labeled IgG were purchased from Jackson Laboratories and goat anti-mouse I 125 -labeled IgG from ICN (Costa Mesa, CA). Ficoll 400 was from Amersham Pharmacia Biotech, and sucrose was from Fisher.
Cell Lines-AP-1 is a cell line derived from wild-type CHO cells that are devoid of endogenous Na ϩ /H ϩ exchanger activity (31). AP1/ NHE3Ј 38HA3 cells are AP-1 cells stably transfected with rat NHE3 containing a triple HA tag in the first extracellular loop following residue 38 (26). ldlF cells, a gift from Dr. M. Krieger (Massachusetts Institute of Technology, Cambridge, MA), are CHO cells that have a temperature-sensitive mutation in ⑀-COP that is expressed only at the restrictive temperature (Ն39°C) but not at permissive temperatures (Յ34°C) (32)(33)(34).
AP1/NHE3Ј 38HA3 cells and ldlF cells were grown in ␣-minimal essential medium (Ontario Cancer Institute, Toronto, Ontario, Canada) containing 25 mM NaHCO 3 and supplemented with 10% fetal calf serum, 100 units/ml penicillin and 100 g/ml streptomycin (Life Technologies, Inc.). AP-1 cells were incubated in a humidified environment containing 95% air and 5% CO 2 at 37°C. ldlF cells were grown at 34°C. To induce the mutant ⑀-COP phenotype, these cells were incubated at 39°C for 8 -12 h prior to assay. Cultures were re-established from frozen stocks regularly and cells from passages 3 to 20 were used for the experiments.
Transfection-The cDNAs encoding wild-type dynamin I and the S45N mutant (DynS45N) subcloned into the pcDNA3 vector were a gift of Dr. S. Schmid (Scripps Institute, La Jolla, CA). To facilitate detection by immunostaining, both dynamin constructs were HA-tagged at the amino terminus. Rat NHE3Ј 38HA3 subcloned into the pCMV vector has been described previously (26). All vectors were transformed into DH5␣ E. coli, grown in Luria broth with the appropriate selection antibiotics, and purified using commercially available kits (Qiagen Maxiprep, Qiagen, Mississauga, Ontario, Canada). For transient expression experiments, cells were grown to 30 -40% confluence on 25-mm glass coverslips. The vectors of interest were co-transfected with pEGFP at a 10:1 ratio, which enabled us to positively identify the transfectants. All cells were transfected at 37°C, with the exception of the ldlF cells, which where maintained at the permissive temperature of 34°C throughout the transfection protocol. Immunofluorescence and microfluorimetry were performed 24 -48 h post-transfection.
Inhibition of Clathrin-mediated Endocytosis-Three different procedures were used to prevent clathrin-mediated endocytosis. Hypertonic challenge was performed according to Hansen et al. (35). Briefly, the normal growth medium was replaced with HEPES-buffered medium supplemented with 0.45 M sucrose for 15 min at 37°C, prior to estimation of NHE3 distribution. A second procedure, involving depletion of intracellular K ϩ , was modified from Hansen et al. (35,36). Cells were washed three times with K ϩ -free medium (140 mM NaCl, 20 mM HEPES, 1 mM CaCl 2, 1 mM MgCl 2 and 1 g/liter D-glucose), then shocked for 5 min with hypotonic K ϩ -free solution (1:1 K ϩ -free medium:H 2 O) at room temperature. Cells were washed 3-fold with K ϩ -free medium and allowed to incubate in the same medium for an additional 15 min at 37°C. Acid treatment was also used to inhibit clathrin-mediated endocytosis (35). In this case the cells were treated with Na ϩ medium supplemented with 5 mM acetic acid, pH 5.0, for 10 min at room temperature prior to the assay, which was carried out in the same medium.
Immunofluorescence-To localize total NHE3 or to verify the expression of dynamin by immunofluorescence, the cells were washed three times with PBS prior to fixation with 4% paraformaldehyde. The cells were then incubated with 100 mM glycine in PBS for 15 min and permeabilized with 0.1% Triton X-100 in PBS supplemented with 10% fetal bovine serum for 20 min. Incubation with anti-HA antibody (1/ 1000) for 1 h at room temperature was followed by extensive washes with PBS, blocking for 1 h with 10% goat serum, and then incubating for 1 h with 1/1000 goat Cy3 anti-mouse antibody. The coverslips were mounted using Dako Fluorescence Mounting Medium.
To assess internalization of the antiporter, intact NHE3Ј 38HA3 -expressing cells were incubated for 45-60 min at 37°C with anti-HA antibody (1/1000) in HEPES-buffered medium supplemented with 10% goat serum. Unbound antibody was removed by extensive washes with cold PBS, and the cells were then fixed, permeabilized, blocked, and labeled with Cy3-conjugated secondary antibody as above. Where specified, incubation with the primary anti-HA antibody was performed at 4°C, followed by varying times of internalization in the absence of added antibody.
To assess recycling of Tfn receptors, cells grown on 25-mm coverslips were first incubated in serum-free HEPES-buffered medium for at least 30 min to deplete endogenous Tfn. Next, the cells were incubated with 20 g/ml rhodamine-conjugated human Tfn for 45-60 min at 37°C. Unbound Tfn was washed extensively with PBS and the cells were fixed for 45 min with 4% paraformaldehyde and mounted as described above. Cells were visualized using the 100ϫ objective of a Leica DM1RB fluorescence microscope (Heidelberg, Germany) equipped with a Micromax cooled CCD camera (Princeton Instruments, Trenton, NJ), operated from a Dell computer using Winview software (Princeton Instruments). Digitized images were cropped using Adobe Photoshop (Adobe Systems, Inc.). All images are representative of at least three separate experiments.
Quantitation of NHE Internalization Using Radiolabeled IgG-Quantitation of recycling of NHE3Ј 38HA3 was performed using I 125labeled anti-mouse IgG. AP1/NHE3Ј 38HA3 cells were grown to 80 -90% confluence on six-well plastic dishes. The cells were then incubated at 4°C for 1 h or at 37°C for the time periods indicated with monoclonal anti-HA antibody (1/1000) in medium supplemented with 10% goat serum. Internalization was terminated by extensive washes with icecold PBS wash buffer (PBS supplemented with 1 mM CaCl 2 and 1 mM MgCl 2 ), and the preparation was blocked by incubation with wash buffer supplemented with 10% goat serum for 1 h at 4°C. The cells were then labeled with mouse I 125 -IgG (0.8 C i /ml) for 1 h at 4°C, followed by removal of unbound I 125 -IgG by extensive washing with ice cold wash buffer. Last, the cells were eluted from the culture dish with 1 ml of 2 M formic acid. Cells exposed only to I 125 -IgG without prior incubation with anti-HA antibody were used as background controls. Radioactivity was counted using a 1282 Compugamma LKB ␥ counter. Data were analyzed using Microcal Origin™ and are presented as means Ϯ S.E. Significance was assessed using Student's t test.
Isolation of Clathrin-coated Vesicles-Clathrin-coated vesicles (CCV) were isolated from rabbit liver and from ileal villus cells by differential centrifugation using a protocol adapted from Campbell et al. (37,38). This method initially yields crude clathrin-coated vesicles that are ideally suited for further purification using a sucrose gradient. Following this protocol, highly enriched CCV (80 -90% pure) are obtained (38). The major contaminants are ferritin and some filamentous material (39). Ileal villus cells were isolated by a modification of the method of Weiser (40) as reported previously (41). All procedures for the purification of CCV were done at 4°C. The isolated cells were homogenized in 20 volumes of buffer containing 0.1 M MES, 0.5 mM MgCl 2 , 1 mM EGTA and 0.02% NaN 3 , pH 6.5 (buffer A), with protease inhibitors, using a Dounce homogenizer. For isolation of CCV from liver, the tissue was homogenized in the same buffer using a Polytron homogenizer. The homogenate was centrifuged at 8000 ϫ g for 50 min at 4°C and the supernatant saved. The supernatant was sedimented at 186,000 ϫ g for 60 min. The pellet was homogenized with buffer A in a Potter-Elvehjem apparatus and mixed with an equal volume of Ficoll-sucrose solution (both at a final concentration of 12.5% in buffer A, pH 6.5). The redissolved pellet was centrifuged at 40,000 ϫ g for 40 min and the supernatant saved. The pellet from this fraction was again resuspended in Ficoll-sucrose as described above, centrifuged, and the supernatant saved. The supernatants were pooled and mixed with 3-4 volumes of buffer A and centrifuged at 186,000 ϫ g for 60 min. The pellet represented the CCV preparation. The pellet was resuspended in 1-2 volumes of buffer A layered on a discontinuous sucrose gradient (from the bottom: 60%, 50%, 40%, 30%, 20%, 10%, 5%) and centrifuged at 50,000 ϫ g for 2 h at 4°C. Fractions were collected and analyzed by SDS-polyacrylamide gel electrophoresis followed by Coomassie blue staining and by immunoblotting. CCV were enriched in the 10 -20% fraction of the gradient.
Measurement of Intracellular pH-Cytosolic pH was measured in BCECF-loaded cells by either microfluorimetry or by ratio imaging, essentially as described (25,42). All measurements of pH i were performed at 37°C. Calibration of the fluorescence intensity to pH i was performed in the presence of 5 M nigericin in high potassium medium (140 mM KCl, 20 mM HEPES, 1 mM MgCl 2 , and 5 mM glucose) as described previously (43). Each coverslip was calibrated at the end of the experiment using at least three pH values.

Use of Anti-HA Antibodies to Probe NHE3 Distribution and
Function-To facilitate the assessment of NHE3 traffic between the plasmalemmal and intracellular pools, we inserted an HA epitope tag in the predicted first extracellular loop of the protein (between residues 38 and 39). An extracellular epitope was essential to detect NHE3 in intact (nonpermeabilized) cells and thereby quantify its rate and extent of internalization. Using heterologous expression in antiport-deficient AP-1 cells, we have shown that the insertion of the HA tag in this position does not measurably alter the expression or ion exchange properties of NHE3 (26). When expressed in AP1 cells, the externally tagged exchanger, named NHE3Ј 38HA3 , was found to accumulate in a pericentriolar compartment (Fig. 1A). This coincides with the distribution of the carboxyl-terminal HAtagged form of NHE3, which was described earlier to reside in recycling endosomes (25). The similar localization of the two forms of the exchanger implies that the placement of epitope tags has little effect on NHE3 targeting.
Binding of anti-HA antibody to the external epitope was used next to monitor the distribution of the exchangers. It was essential to ascertain that the binding of antibody to NHE3 would not by itself alter the behavior of the exchangers. The effect of the antibody on NHE3 distribution was assessed first. NHE3Ј 38HA3 transfectants were incubated with the antibody and the cells were then incubated at 37°C for 45 min to allow internalization to proceed. The cells were then fixed, permeabilized, and labeled with secondary antibodies. As illustrated in Fig. 1B, the distribution of NHE3 resembles that observed in untreated cells which were fixed and permeabilized prior to immunostaining (cf. Fig. 1, A and B). Thus, binding of monoclonal antibody to the extracellular epitope did not alter the subcellular distribution of NHE3Ј 38HA3 , implying normal traffic within the cell.
We also determined whether the antibody alters Na ϩ /H ϩ exchange across the plasma membrane. Cells were initially incubated in the cold in the presence or absence of a concentration of antibody determined earlier to saturate all the available extracellular epitope sites. Na ϩ /H ϩ exchange activity was then estimated by microfluorimetric measurement of pH i using the pH-sensitive dye BCECF. Following an acute acid load imposed by a 50 mM NH 4 Cl prepulse, we measured the pH i recovery induced by superfusion with Na ϩ . Comparison of traces 1 and 2 in Fig. 1C illustrates that engagement of the external epitope by the antibody had no discernible effect on the rate of ion exchange mediated by NHE3Ј 38HA3 .
While binding of antibody to the exchangers exposed at the surface had little effect on their transport activity, it was conceivable that prolonged exposure at physiological temperature might induce cross-linking and redistribution of the exchangers, possibly altering their number at the plasmalemma. To rule out this possibility, AP1/NHE3Ј 38HA3 cells were incubated with anti-HA antibody for 30 min at 37°C, thereby allowing endocytosis to proceed. The activity of NHE3 was measured subsequently, as described above. This pretreatment had no significant effect on the rate of Na ϩ /H ϩ exchange (Fig. 1C, trace  3), suggesting that the overall distribution of NHE3 was unaltered, a conclusion consistent with the morphological determinations reported in Fig. 1, A and B. In summary, our data imply that neither the introduction of the external HA-tag nor binding of monoclonal antibody to the external epitope had any effect on the subcellular distribution or activity of NHE3.
Time Course of Internalization of NHE3-To study the route FIG. 1. Binding of antibody to the HA epitope has no effect on NHE3 distribution or activity. AP1/NHE3Ј 38HA3 cells grown to subconfluence on 25-mm glass coverslips were used for immunofluorescence or pH i determinations, as described under "Experimental Procedures." A, the cells were fixed, permeabilized, and then incubated with 1/1000 anti-HA antibody, followed by Cy3-conjugated anti-mouse IgG (1/1000). B, live cells were incubated with 1/1000 anti-HA antibody, and endocytosis was allowed to proceed by incubation at 37°C for 45 min. Unbound antibody was removed by extensive washing with ice-cold PBS, and the cells were next fixed, permeabilized, and labeled with secondary antibody as in A. C, the Na ϩ /H ϩ exchange activity of NHE3 was estimated by microfluorimetric measurement of pH i using BCECF after an acute NH 4 Cl-induced acid load. Otherwise, untreated cells (trace 1 and diagram 1) were compared with cells that had been incubated with anti-HA antibody (1/500) for either 30 min at 4°C, resulting in binding to surface NHE3 (trace 2 and diagram 2) or with the same concentration of antibody for 30 min at 37°C, to induce internalization of the antibody-NHE3 complex (trace 3 and diagram 3). Images and traces are representative of at least three individual experiments of each kind. of internalization used by NHE3, it was important to define its normal time course of endocytosis. Initial determinations were made by immunofluorescence methods that involved labeling the surface-exposed exchangers in the cold using anti-HA antibody and then allowing internalization upon rewarming. After varying times, the cells were fixed, permeabilized, and stained with labeled secondary antibodies. Typical results are shown in Fig. 2, A-C. As expected, only superficial NHE3 molecules were labeled in the cold, whereas progressive internalization and juxtanuclear accumulation occurred upon warming. The process was clearly detectable by 20 min and neared completion by 40 min.
To more quantitatively evaluate the process of internalization, cells were labeled in the cold as above and, after warming to 37°C for the indicated periods, the fraction of antibody remaining at the surface was quantified using I 125 -labeled secondary antibody. As summarized in Fig. 2D, over 50% of the HA antibody bound initially had disappeared after 20 min at 37°C, and over 80% was missing after 40 min. Disappearance of the antibody was due to internalization, rather than dissociation or degradation. This was demonstrated by fixing and permeabilizing the cells prior to addition of the secondary antibody (not shown). Jointly, these experiments show that NHE3 becomes internalized with a half-life of under 20 min.
Effects of Hypertonicity, Intracellular Acidification, and K ϩ Depletion on Internalization of NHE3-The possible role of clathrin-mediated endocytosis (CME) in NHE3 internalization was evaluated next. As an initial approach, we used several maneuvers reported earlier to effectively impair clathrin-mediated endocytosis. These included cell shrinkage by prolonged elevation of the medium osmolarity (35,44), acidification of the cytosol (35), and depletion of intracellular K ϩ (35, 36). Fig. 3 shows experiments comparing the extent of NHE3 internalization in isotonic (Iso) and hypertonic (Hyp) media. In accordance with the results above, the surface antiporters are largely internalized after 45 min at 37°C under physiological isotonic conditions (A and B in Fig. 3). By contrast, most of the exchangers remained at the cell surface when the cells were rewarmed in hypertonic medium (Fig. 3C). As before, we could quantify these effects using radiolabeled secondary antibody. The results of nine experiments are summarized in D of Fig. 3. Only 35 Ϯ 5% of the antibody initially bound to NHE3 was found at the surface after 45 min under isotonic conditions, whereas virtually all of the original NHE3 remained at the plasma membrane in the hypertonically treated cells (Fig. 3D).
Similar results were obtained in cells subjected to two other treatments that block CME, namely cytosolic acidification by incubation in acetate-rich medium of low pH and K ϩ depletion, accomplished by a combination of hypotonic stress and incubation in K ϩ -free medium (data not shown). The cumulative evidence indicates that inhibition of CME by physicochemical means precludes internalization of NHE3.
Inhibition of Endocytosis Using a Dynamin Mutant, Effect on NHE3Ј 38HA3 Internalization-Because the physical treatments used above are comparatively harsh, they are likely to have additional effects that are unrelated to inhibition of clathrin coat formation. In order to study the involvement of clathrin in NHE3 internalization more precisely, we used a more selective means to inhibit CME. Specifically, we prevented the fission of clathrin-coated pits from the plasma membrane using an inhibitory form of dynamin. Dynamin is a 100-kDa GTPase that has been found to be essential for CME. After being recruited to FIG. 2. Time course of internalization of NHE3. A-C, AP1/ NHE3Ј 38HA3 cells were cooled to 4°C by washing in ice-cold PBS and incubated with anti-HA antibody (1/1000) for 1 h at 4°C. Unbound antibody was removed by washing with ice-cold PBS. The cells were then incubated for the indicated periods of time at 37°C, fixed, and permeabilized. Finally, the cells were stained with Cy3-conjugated secondary antibody before visualization by epifluorescence. D, for quantitation of surface NHE3, AP1/NHE3Ј 38HA3 cells grown on six-well plates were labeled with anti-HA antibody as in A-C. Internalization was allowed to occur by incubation for the specified periods at 37°C and terminated by extensive washes with ice-cold PBS supplemented with 1 mM CaCl 2 and 1 mM MgCl 2 . After blocking for 1 h at 4°C with buffer supplemented with 10% goat serum, the cells were labeled with 0.8 C i /ml labeled 125 I-sheep anti-mouse IgG for 1 h at 4°C. Unbound 125 I-IgG was removed by extensive washing, the cells were eluted from the plate with 2 M formic acid, and radioactivity was counted. Cells exposed to the 125 I-IgG without preincubation with anti-HA antibody were used to determine nonspecific binding, which was subtracted from all determinations. Data are representative of three separate experiments, each with triplicate determinations. Normalized to the initial binding and presented as mean Ϯ S.E.

FIG. 3. Effect of hypertonicity on internalization of NHE3. A-C,
immunofluorescence of AP1/NHE3Ј 38HA3 cells grown on glass coverslips. Anti-HA antibody was bound to the cell surface in the cold as in Fig. 2, followed by incubation at 37°C in isotonic (Iso) medium for 0 or 45 min (A and B, respectively) or for 45 min in hypertonic (Hyp) solution (normal medium supplemented with 0.45 M sucrose; C). The cells were then fixed, permeabilized, and subjected to immunostaining with Cy3conjugated anti-mouse antibody as in Fig. 2. Images are representative of at least three similar experiments. D, cells were allowed to bind anti-HA antibody in the cold and were then incubated in isotonic or hypertonic solution as in A-C. Last, surface-exposed anti-HA antibody was quantified using I 125 -labeled sheep anti-mouse IgG as in Fig. 2D. Data are representative of three separate experiments, each with triplicate determinations. Normalized to the initial binding and presented as mean Ϯ S.E. clathrin-coated pits by interactions involving its COOH-terminal proline-rich region, multiple dynamin monomers arrange into a helical structure at the neck of the coated pits. Upon hydrolysis of GTP, dynamin severs the pits from the plasma membrane, releasing clathrin-coated vesicles (see Refs. 45 and 46 for reviews). Dominant-negative mutant forms of dynamin I, which lack GTPase activity, can bind to the coated pits yet fail to effect fission of coated vesicles and thereby block CME (47,48).
To verify the role of clathrin in NHE3 internalization, we used a mutant form of dynamin I where Ser-45 has been mutated to Asn (DynS45N) and which was shown earlier to exert a dominant-negative effect (47,48). To verify the inhibitory effect of DynS45N, we tested whether transient transfection of this dominant-negative mutant was able to prevent the internalization of Tfn receptors, which are exclusively internalized via CME (28,30). Cells transfected with DynS45N were identified by co-transfection with enhanced-GFP (EGFP), while the internalization of the receptor was assessed using rhodamine-conjugated human Tfn. As illustrated in Fig. 4B, whereas Tfn was readily accumulated in pericentriolar recycling endosomes in nontransfected cells, little Tfn was internalized by cells expressing DynS45N (indicated by an arrow). Importantly, the mutant dynamin also impaired the internalization of NHE3 (Fig. 4, C and D).
Taken together, the experiments using hypertonicity, acidity, or cytosolic K ϩ depletion and those using transfection of dominant-negative constructs of dynamin indicate that endocytosis of NHE3 occurs predominantly through clathrin-mediated vesiculation.
Colocalization of NHE3 with Clathrin Coat Components in Epithelial Cells-While the preceding experiments strongly implicate clathrin-dependent endocytosis in the internalization of NHE3 in AP-1 cells, it is unclear whether these findings in heterologous transfectants can be extrapolated to native epithelial cells. We therefore undertook fractionation studies in an attempt to define whether NHE3 is localized in clathrin-coated vesicles in intestinal epithelial cells.
Purification and identification of CCV from several tissues using a method of differential centrifugation and cell fractionation is well documented (38). We used this procedure to isolate CCV from absorptive villus cells of the rabbit ileum, which have previously been shown to express high levels of NHE3 (11). To validate the effectiveness of the isolation procedure, we also isolated liver CCV, which are readily purified by this method (38). CCV isolated from the liver and the ileum were separated by SDS-polyacrylamide gel electrophoresis and the gel stained with Coomassie Blue. As shown in Fig. 5A, the major proteins detected in both preparations were similar and had sizes that were compatible with those of the clathrin heavy chain, the clathrin light chain, and the adaptor proteins. This protein pattern is similar to that of CCV isolated from adipocytes, which also contain the glucose transporter, . To confirm the presence of the coat proteins in our preparations, CCV from the liver and from isolated ileal cells were separated by SDS-polyacrylamide gel electrophoresis, transferred to nitrocellulose, and immunoblotted with monoclonal antibodies to the clathrin heavy chain and to the adaptor proteins 1 and 2 (AP-1 and AP-2, respectively). As shown in Fig.  5B, all of these coat proteins were found in the CCV isolated from both the liver and the intestine. To determine whether they also contained NHE3, CCV from intestinal cells were subjected to immunoblotting using a monoclonal antibody to NHE3. A protein extract of isolated brush border membranes from rabbit ileum was used to confirm the effectiveness of the antibody. As shown in Fig. 5C, NHE3 was present not only in the brush border membranes, but in the ileal CCV as well. We also found that NHE3 and the CCV markers became enriched in parallel during the purification procedure (not illustrated). These data indicate that NHE3 co-purifies with CCV in native tissues where it is endogenously expressed, complementing the observations in heterologous transfectants.
Role of COP-I in Intracellular Traffic of NHE3-Once internalized, the contents of early endosomes are sorted to different compartments. Some components proceed to late endosomes Routing of endosomal constituents to their appropriate destinations depends, at least in part, on members of the COP family of coat proteins (50,51). A distinct subset of COP-I proteins, which includes the ␣, ␤, ␤Ј, and ⑀ subunits, was found to associate with early endosomes (50), where it plays a role in endosomal maturation. For example, cellular microinjection of an antibody to ␤-COP interfered with delivery of cargo from early to late endosomes (50). Additional evidence of a role of COP-I in endosomal traffic was obtained using ldlF cells, a line derived from CHO cells which bears a temperature-sensitive mutation in ⑀-COP (32-34, 52, 53). At permissive temperatures (Յ34°C), ldlF cells express reduced levels of ⑀-COP, which nevertheless suffice for normal COP-I function. As the temperature is raised to nonpermissive levels (Ն39°C), ⑀-COP becomes unstable and is rapidly degraded (33). As a result, endosomal traffic is affected: recycling is partially inhibited and delivery of carrier vesicles from early to late endosomes is blocked (33,53).
We took advantage of the inducible nature of the mutation in ldlF cells to define the role of ⑀-COP in the uptake and intracellular traffic of NHE3. NHE3Ј 38HA3 was transiently transfected into ldlF cells grown at 34°C, a permissive temperature. As shown in Fig. 6C, the distribution of the exchangers expressed transiently in cells maintained at 34°C was similar to that of the wild-type stable transfectants used above. After allowing transient expression at 34°C, some of the cells were warmed to 39°C for 8 -12 h and the distribution of NHE3 was evaluated again. Elimination of ⑀-COP, which was verified by immunoblotting (not shown), did not prevent internalization of NHE3 (Fig. 6D). However, the intracellular distribution of NHE3 seemed to be somewhat affected, with reduced juxtanuclear concentration. In most cells, comparatively large vesicles containing NHE3 were scattered throughout the cell. This pattern resembles the distribution of Tfn receptors, which were similarly internalized normally at 34°C, but scattered into larger structures at 39°C (not illustrated). These findings confirm that, in CHO cells, NHE3 is distributed in a compartment which strongly resembles that occupied by Tfn receptors. Moreover, they indicate that whereas ⑀-COP is not essential for internalization of the antiporters, this COP-I component affects the normal intracellular traffic of NHE3. DISCUSSION The present results confirm our earlier observation that, in AP-1-derived CHO cells, NHE3 accumulates in recycling endosomes. Recycling endosomes of nonepithelial cells were recently suggested to be analogous to apical recycling endosomes in polarized cells (54), where endogenous NHE3 is expressed (16). This conclusion was based on studies of Rab17, a small GTPase that is highly localized to apical recycling endosomes in epithelial cells. When ectopically expressed in nonpolarized fibroblasts, Rab17 was found to be localized in the pericentriolar recycling endosomes (54), where NHE3 also accumulates. These observations validate the use of the nonpolarized CHO cells as an adequate model to study the internalization of the epithelial NHE3.
A central role of CME in the internalization and recycling of NHE3 is suggested by the inhibitory effects of DynS45N. This dominant-negative form of dynamin effectively prevented the endocytosis of NHE3. While elegant and specific, the transfection experiments require comparatively long periods of incubation. Under these conditions, compensatory mechanisms such as induction of non-clathrin-coated vesicles may be up-regulated (47,48), in order to ensure cell survival. Moreover, dynamin is also involved in other endocytic processes. Internalization of some glycosylphosphoinositol-linked membrane proteins and bacterial toxins occurs mainly through caveolae, small flask-shaped non-clathrin-coated plasmalemmal vesicles (55)(56)(57). Like clathrin-coated pits, fission of caveolae from the membrane to form free endocytic vesicles requires GTP hydrolysis and was recently found to involve dynamin (29,58). Ultrastructural and biochemical analyses have revealed dynamin to associate directly with caveolae where, as found in clathrincoated pits, it assembles at the neck. Functional data with inhibitory antibodies and with dominant-negative dynamin mutants revealed that the GTPase activity of dynamin is crucial for budding of caveolae (29,58). For these reasons, transfection with DynS45N is not conclusive evidence of involvement of CME in NHE3 internalization.
In this context, physicochemical maneuvers, such as depletion of intracellular K ϩ or prolonged treatment of the cells with hypertonic or acidic solutions, provide useful complementary information. Although less specific than transfection, these procedures are not believed to affect caveolae or other noncoated internalization pathways and have been used extensively to inhibit CME in a variety of cell types (e.g. Refs. 35, 36, and 59). In CHO cells, they effectively prevented the uptake of Tfn, a hallmark of CME, and they similarly inhibited the internalization of NHE3. Jointly, these approaches support the notion that CME is the predominant pathway for NHE3 internalization, at least in heterologous transfectants.
Importantly, subcellular fractionation studies suggested that native NHE3 is also internalized via CME in epithelial cells. As summarized in Fig. 5, a fraction of the NHE3 isolated from intestinal epithelial cells was found to co-purify with clathrin-coated vesicles. Together with the molecular and physicochemical approaches described above, these observations argue in favor of CME as an important route for NHE3 endocytosis.
Internalization of plasmalemmal transporters is important FIG. 6. Effect of ⑀-COP mutants on NHE3 distribution. ldlF cells grown at 34°C were transiently transfected with vectors encoding NHE3Ј 38HA3 and EGFP at a 10:1 ratio. Twenty-four hours after transfection the cells were incubated for a further 8 -12 h at either a permissive temperature (34°C) or at a restrictive temperature (39°C). Next, the cells were cooled to 4°C by washing in ice-cold PBS and incubated with anti-HA antibody (1/1000) for 1 h at 4°C. Unbound antibody was removed by washing with ice-cold PBS. The cells were then incubated for the indicated periods of time at 37°C, fixed, and permeabilized. Finally, the cells were stained with Cy3-conjugated secondary antibody before visualization by epifluorescence. A and B, fluorescence of EGFP. C and D, immunostaining of the HA epitope, indicating the distribution of NHE3. Representative of three separate experiments.
not only for their catabolism, but in some instances also as a regulator of the number of active transporters at the membrane. Constitutive internalization of the epithelial sodium channel by CME has recently been shown to be important for regulating its activity both in vivo and in cell culture (60). In this system, overexpression of a dominant-negative dynamin resulted in stimulation of channel activity, mimicking the functional phenotype of Liddle's syndrome (60). Internalization of transporters is also important in the termination of the stimulatory effects of insulin on glucose transport (61) and of acid secretion in the stomach (62).
By analogy, modulation of CME may play a role in regulation of NHE3 activity. This hypothesis is lent credence by the observations that NHE3 activity increases in response to prolonged hyperosmolarity and acidity, both of which are potent inhibitors of CME (44,63,64). The stimulation of transport is associated with increased number of exchangers, with little increase in mRNA, ruling out de novo NHE3 synthesis (22,23,65). It is tempting to speculate that reduced internalization leads to accumulation of plasmalemmal exchangers and contributes to the stimulation of transport. Conversely, earlier subcellular fractionation studies concluded that a net displacement of antiporters from the plasma membrane to the endosomal pool of renal cells mediates the inhibitory effects of parathyroid hormone and hypertension on NHE3 (20,21,66). Likewise, the acute inhibition of NHE3 activity induced by protein kinase C in colonic epithelial cells was attributed to a redistribution of exchanger molecules from the brush border into a subapical cytoplasmic compartment (67).
Contrary to the involvement of clathrin-coated pits and vesicles, ⑀-COP was seemingly not required for NHE3 internalization, suggesting that COP-I-coated vesicles were not essential. However, genetic ablation of ⑀-COP in the temperature-sensitive ldlF mutants modified the appearance of the intracellular compartment where NHE3 resides. These findings are compatible with those of those of Daro et al. (53) who showed redistribution of endomembranes and partial inhibition of Tfn receptor recycling in ldlF cells grown at the restrictive temperature. 2 We suggest that normal traffic of NHE3 along the endocytic pathway requires COP-I. Future experiments will be required to ascertain whether the normal complement of NHE3 at the surface membrane is affected in ⑀-COP-deficient cells. Quantitation of the number of transporters or their activity was not performed in the present experiments in view of the wide variability of expression in transient transfectants.
In summary, we have shown that CME contributes to the physiological internalization of NHE3 and that intracellular traffic of NHE3 involves components of the COP-I co-atomer. Ongoing studies aimed at identifying the motifs that target NHE3 to clathrin-coated pits will shed additional light on the role of recycling in the function and regulation of this isoform of the Na ϩ /H ϩ exchanger.