The Apical Na+/H+ Exchanger Isoform NHE3 Is Regulated by the Actin Cytoskeleton*

The epithelial isoform of the Na+/H+ exchanger, NHE3, associates with at least two related regulatory factors called NHERF1/EBP50 and NHERF2/TKA-1/E3KARP. These factors in addition interact with the cytoskeletal protein ezrin, which in turn binds to actin. The possible linkage of NHE3 with the cytoskeleton prompted us to test the effect of actin-modifying agents on NHE3 activity. Cytochalasins B and D and latrunculin B, which interfere with actin polymerization, induced a profound inhibition of NHE3 activity. The effect was isoform-specific inasmuch as the “housekeeping” exchanger NHE1 was virtually unaffected. Cytoskeletal disorganization was associated with a subcellular redistribution of NHE3, which accumulated at sites where actin aggregated, suggesting a physical interaction of exchangers with the cytoskeleton. An interaction was further suggested by the co-sedimentation of a detergent-insoluble fraction of NHE3 with the actin cytoskeleton. Inhibition of transport was not due to diminution in the number of transporters at the plasmalemma. Functional analyses of NHE1/NHE3 chimeras revealed that the cytoplasmic domain of NHE3 conferred sensitivity to cytochalasin B. Progressive carboxyl-terminal and internal deletions of the cytoplasmic region of NHE3 indicated that the region between residues 650 and 684 is critical for this response. This region overlaps with the domain reported to interact with NHERF and also contains a putative ezrin-binding site; hence, it likely plays a role in interactions with the cytoskeleton.

The epithelial isoform of the Na ؉ /H ؉ exchanger, NHE3, associates with at least two related regulatory factors called NHERF1/EBP50 and NHERF2/TKA-1/ E3KARP. These factors in addition interact with the cytoskeletal protein ezrin, which in turn binds to actin. The possible linkage of NHE3 with the cytoskeleton prompted us to test the effect of actin-modifying agents on NHE3 activity. Cytochalasins B and D and latrunculin B, which interfere with actin polymerization, induced a profound inhibition of NHE3 activity. The effect was isoform-specific inasmuch as the "housekeeping" exchanger NHE1 was virtually unaffected. Cytoskeletal disorganization was associated with a subcellular redistribution of NHE3, which accumulated at sites where actin aggregated, suggesting a physical interaction of exchangers with the cytoskeleton. An interaction was further suggested by the co-sedimentation of a detergent-insoluble fraction of NHE3 with the actin cytoskeleton. Inhibition of transport was not due to diminution in the number of transporters at the plasmalemma. Functional analyses of NHE1/NHE3 chimeras revealed that the cytoplasmic domain of NHE3 conferred sensitivity to cytochalasin B. Progressive carboxyl-terminal and internal deletions of the cytoplasmic region of NHE3 indicated that the region between residues 650 and 684 is critical for this response. This region overlaps with the domain reported to interact with NHERF and also contains a putative ezrin-binding site; hence, it likely plays a role in interactions with the cytoskeleton.
Na ϩ /H ϩ exchangers (NHE) 1 mediate electroneutral exchange of Na ϩ for H ϩ at the plasma membrane and across the membranes of some intracellular organelles. The NHE family consists of six known isoforms, all of which are integral membrane proteins with multiple transmembrane domains and a large cytosolic carboxyl-terminal domain (1). Some of these, including NHE1 and NHE6, are ubiquitous and are thought to fulfill housekeeping functions such as the regulation of cytosolic pH (1) and the control of ionic homeostasis in mitochondria (2), respectively. By contrast, other isoforms are expressed selectively in defined cell types and are therefore believed to perform more specialized functions. These include NHE3, which is confined to the apical membrane of polarized epithelial cells of the kidney, gastrointestinal tract, and gallbladder. This isoform plays a central role in the (re)absorption of Na ϩ and HCO 3 Ϫ across the epithelial layer and is therefore subjected to exquisite regulation. NHE3 activity is modulated by changes in blood pressure and osmolarity and is responsive to hormones that elevate intracellular second messengers, such as cAMP and diacylglycerol (3).
Little is known about the molecular mechanisms underlying the regulation of NHE3 activity. Electron microscopy and confocal immunofluorescence microscopy observations (4,5) indicated that a fraction of the exchangers resides in an intracellular vesicular pool. This distribution can be recapitulated when NHE3 is heterologously expressed in Chinese hamster ovary cells (5). It has therefore been postulated that, as in the case of other regulated transporters, modulation of the rate of transport could be accomplished by altering the distribution of NHE3 between endomembranes and the surface membrane (6). Accordingly, inhibition of exocytosis by wortmannin was found to be associated with a drastic decrease in NHE3 activity at the plasmalemma (7).
Regulation of NHE3 can also be exerted by stimulation of protein kinases, most notably protein kinase A (PKA) and protein kinase C (PKC), which induce a significant inhibition of the exchanger (8,9). Regarding the mechanism of action of PKA, there is evidence that direct phosphorylation of the exchanger is involved (8, 10 -12).
However, other studies have also invoked the participation of ancillary proteins that are seemingly required for the PKAmediated inhibition of NHE3 (13,14). These proteins, called NHERF1/EBP50 and NHERF2/TKA-1/E3KARP, bind directly to the exchanger (15) and are postulated to be substrates of PKA (16,17), although evidence for the latter is controversial. Interestingly, NHERF1/EBP50 was shown independently to bind ezrin, a cytoskeletal protein that is abundant at the apical membrane of epithelia (18). Ezrin itself is capable of binding filamentous actin, serving as a bridge between membraneassociated proteins and the cytoskeleton. These observations raise the possibility that the activity of NHE3 may be controlled in some manner by its state of association with the actin cytoskeleton.
The purpose of the present experiments was to assess whether the state of actin polymerization and/or its association with the membrane affects the rate of transport mediated by NHE3. To study the function of NHE3 in isolation, i.e. uncontaminated by other isoforms that can co-exist in native epithelial systems (19), we utilized antiport-deficient Chinese hamster ovary cells that were heterologously transfected with NHE3. This paradigm also enabled us to analyze and relate the structure and function of NHE3 by transfection of chimeric or truncated mutants of the exchanger. Our results indicate that NHE3 can associate with the cytoskeleton and that its exchange activity is profoundly affected by disruption of the Factin network.
Construction and Mammalian Cell Transfection of NHE cDNA Constructs-A Chinese hamster ovary cell line (AP-1 cells) that was functionally selected for its lack of endogenous NHE activity following chemical mutagenesis (20) was transfected with plasmids containing the wild type NHE1 and NHE3 and mutant NHE3 cDNA constructs.
Two full-length NHE3 constructs used in this study were tagged with an influenza virus hemagglutinin (HA) epitope YPYDVPDYAS as described previously (5,7). Briefly, the first construct contains a single HA epitope appended to the extreme cytoplasmic carboxyl terminus of NHE3 and is termed NHE3Ј-HA. The second construct contains a triple HA epitope inserted into the first extracellular loop of NHE3 between amino acids Arg 38 and Phe 39 and is named NHE3Ј-38HA3. The functional properties of both full-length HA-tagged NHE3 constructs were indistinguishable from untagged NHE3 (5,7). To analyze the structural basis of the actions of actin-modifying agents on NHE3 activity, a series of NHE1 and NHE3 chimeras and NHE3 deletion mutants were used. Two chimeras have previously been described. One was constructed from the transmembrane domain (residues 1-504) of NHE1 and the cytosolic tail (residues 456 -831) of NHE3 and is designated NHE1/3, whereas the second, reciprocal chimera, named NHE3/1, has the transmembrane domain of NHE3 (residues 1-455) and the cytosolic tail of NHE1 (residues 505-820) (5). The third chimera, called NHE3/1/3, is composed mainly of NHE3, but amino acids 650 -716 in the cytoplasmic tail were removed and substituted with the corresponding residues (666 -749) of NHE1 by polymerase chain reaction mutagenesis. The NHE3 deletion mutants were created by progressively truncating the cytosolic carboxyl-terminal tail at positions 684 (NHE3⌬684) and 638 (NHE3⌬638) as described previously (21). In addition, an internal deletion of residues 650 -716 of NHE3 was engineered and named NHE3⌬650 -716. All chimeras and deletion mutants were sequenced in the carboxyl-terminal region to ensure the fidelity of the constructions.
All NHE cDNA constructs were transfected into AP-1 cells by the calcium phosphate-DNA co-precipitation technique of Chen and Okayama (22) and stable clones were selected by the H ϩ suicide method of Pouysségur et al. (23). Clonal populations were maintained by routine H ϩ suicide challenge.
Measurement of Na ϩ /H ϩ Exchanger Activity-NHE activity was assessed as the rate of Na ϩ -induced recovery of cytosolic pH (pH i ) following an acid load imposed by preloading with NH 4 Cl, as described previously (5). Briefly, cells plated on coverslips were loaded with 2 g/ml acetoxymethyl ester precursor of BCECF at 37°C for 10 min, washed, and placed in Leiden CoverSlip holders. The cytosol was acidified by incubating the cells with an isotonic solution containing 50 mM NH 4 Cl for 10 min, followed by rapid washing with an isotonic Na ϩ -free solution. The cells were then bathed with isotonic Na ϩ -rich medium to induce Na ϩ -dependent recovery of pH i . The fluorescence of BCECF was measured and calibrated as described (5).
Detergent Extraction and Immunoblotting-Cytoskeleton-associated NHE3Ј-HA was extracted from transfected AP-1 cells cultured in 6-well plates (35 mm) according to the method of Goldman and Abramson (24). Briefly, the cells were rinsed 3 times with phosphate-buffered saline (PBS) and treated with 1 ml of 0.5% Triton X-100, 2 mM EDTA, 2 mM phenylmethylsulfonyl fluoride, 1 M pepstatin, 50 mM Tris-HCl, pH 6.8, for 15 min on ice. Lysates were centrifuged at 175,000 ϫ g at 4°C for 75 min. The resulting pellets were solubilized with Laemmli sample buffer. Samples were subjected to SDS-polyacrylamide gel electrophoresis in 7.5% polyacrylamide gels and transferred onto nitrocellulose filters. Blots were blocked with 0.2% gelatin and exposed to the primary antibody (monoclonal anti-HA antibody; 1:10,000 dilution). Horseradish peroxidase-conjugated goat anti-mouse secondary antibody was applied (1:5000), and immunoreactive bands were visualized using enhanced chemiluminescence (Amersham Pharmacia Biotech).
Immunofluorescence-AP-1 cells stably expressing NHE3Ј-HA were plated onto glass coverslips and grown to Ϸ70% confluence. The cells were washed 3 times with PBS and fixed for 15 min at room temperature using 4% paraformaldehyde in PBS. After fixation, the cells were washed 3-4 times with PBS and incubated with 100 mM glycine in PBS for 15 min. The coverslips were washed again and, where indicated, the cells were permeabilized with 0.1% Triton X-100 in PBS for 20 min at room temperature. Cells were blocked with 5% donkey serum in PBS for 1 h and then incubated with mouse anti-HA antibody (1:1000) dilution for 1 h. After this period, the coverslips were washed 4 -5 times with PBS and then incubated with Cy3-conjugated anti-mouse IgG (1:1000 dilution) for 1 h. The coverslips were washed 3-5 times over 15 min with PBS and mounted onto glass slides with DAKO fluorescence mounting medium (DAKO Corp., Carpinteria, CA).
Quantification of Surface NHE3-To quantify the amount of surface NHE3, we developed a modified enzyme linked immunosorbent assay (7). AP-1 cells stably expressing the externally tagged NHE3Ј-38HA3 were plated onto 12-well plates and grown to ϳ70% confluence. Next, they were incubated with the anti-HA antibody (1:1000 dilution) for 1 h at 4°C to prevent endocytosis. After washing the cells 6 times with PBS/␣-minimal essential medium (9:1 v/v) to remove excess unbound antibody, they were fixed for 10 min at room temperature using 4% paraformaldehyde in PBS. After fixation, the cells were washed 3-4 times with PBS and incubated with 100 mM glycine in PBS for 15 min. Cells were next blocked with 5% donkey serum in PBS for 20 min and then incubated with a peroxidase-conjugated donkey anti-mouse antibody (1:1000) for 1 h. The cells were washed 6 times with PBS and then incubated with 1 ml of Fast OPD reagent for 15 min at room tempera- The cells were then washed with an isotonic Na ϩ -free medium, and their pH i was monitored fluorimetrically, as described under "Experimental Procedures." Recording was started upon reintroduction of extracellular Na ϩ to induce Na ϩ /H ϩ exchange. The results are the means Ϯ S.E. of at least 10 determinations from 4 separate experiments.
ture. The reaction was stopped with 250 l of 3 M HCl. The supernatant was collected, and absorbance was measured at 492 nm (A 492 ) using a U-2000 spectrophotometer (Hitachi, Tokyo, Japan). In the range studied, A 492 varied linearly with the amount of peroxidase bound.

RESULTS
Effects of Cytochalasins on NHE3 and NHE1 Activity-Cytochalasins are fungal metabolites that induce gradual depolymerization of actin filaments by binding to their fast-growing barbed ends (25). They are membrane-permeant and can therefore be used to induce filament disruption in intact cells. We initially used cytochalasin B to investigate the relationship between NHE3 and the actin cytoskeleton. AP-1 cells stably expressing rat NHE3Ј-HA were incubated with or without 10 M cytochalasin B for 30 min. Na ϩ /H ϩ exchange activity was subsequently assessed fluorimetrically as the rate of Na ϩ -induced H ϩ (equivalent) extrusion. As reported earlier (5, 7), untreated cells expressing NHE3Ј-HA recovered rapidly from an acid load upon addition of Na ϩ (Fig. 1A). Pretreatment with cytochalasin B induced a profound inhibition of NHE3Ј-HA activity (Fig. 1A). Cells expressing the externally tagged NHE3Ј-38HA3 construct showed a similar inhibition of activity after pretreatment with cytochalasin B (data not shown), implying that epitope tagging of the carboxyl terminus was not responsible for imparting cytochalasin sensitivity to NHE3.
In addition to its cytoskeletal effects, cytochalasin B is an effective inhibitor of glucose uptake. Because NHE activity is sensitive to the intracellular concentration of ATP (26), it was conceivable that the observed inhibition of exchange was secondary to metabolic depletion of the cells. To evaluate this possibility we used cytochalasin D, which is a similarly effective cytoskeletal-disrupting agent as cytochalasin B yet has no effect on glucose transport. Pretreatment with 10 M cytochalasin D produced an inhibition of NHE3 that was indistinguishable from that induced by the B analog (not illustrated), ruling out indirect effects of metabolic depletion. Moreover, direct measurements of cellular ATP content using luciferase revealed that cytochalasin B did not effect depletion; under the conditions used in Fig. 1, the ATP content in cytochalasintreated cells was 103 Ϯ 6.8% of control (mean Ϯ S.E. of 3 determinations).
The effect of cytochalasin B was NHE isoform-specific. Unlike NHE3, the pH i recovery induced by addition of Na ϩ to acid-loaded NHE1-HA cells was essentially unaffected by pretreatment with cytochalasin B (Fig. 1B). As both NHE1 and NHE3 are sensitive to intracellular ATP (26), this observation confirms that the effect of cytochalasin B on the latter is not the result of metabolic depletion.
It is unlikely that cytochalasin inhibits transport by directly binding to NHE3. This tentative conclusion is supported by the finding that inhibition required several minutes to develop. As shown in Fig. 2A, maximal inhibition by 10 M cytochalasin B was attained only after 10 min. The time course of inhibition parallels the effect of cytochalasin on F-actin. As illustrated in Fig. 3F, cells treated for 10 min underwent a massive disruption of the actin cytoskeleton. The well defined stress fibers and actin accumulations near focal adhesions seen in control cells were no longer apparent after cytochalasin treatment. Instead, small punctate and large amorphous accumulations of F-actin were present throughout the cells.
Effects of Other Actin Cytoskeletal Modifiers-To confirm that disruption of the actin microfilament network was indeed responsible for the inhibition of NHE3, we examined the effects of latrunculin B. This compound is chemically unrelated to the cytochalasins and acts by forming complexes with monomeric actin, thus reducing the pool available for polymerization (27). As shown in Fig. 3G, treatment with latrunculin B altered the morphology of the cells and produced a drastic disruption of the actin network. Actin appeared concentrated near the cell borders. The effects of latrunculin B on F-actin were paralleled by inhibition of NHE3 (Fig. 2B), which resembled that produced by cytochalasins. These observations support the notion that optimal NHE3 activity requires the integrity of the cellular F-actin network.
Because decreasing F-actin inhibited NHE3, we wondered whether the opposite effect would be observed by increasing the F-actin content of the cells. This was accomplished using jasplakinolide, a cell-permeable monocyclic peptide, which binds and stabilizes F-actin (28). The binding of jasplakinolide to F-actin was indicated by the marked overall decrease in staining by phalloidin (Fig. 3H), which binds to the same site on F-actin as jasplakinolide (28). 2 Elevated F-actin content in jasplakinolide-treated cells was also suggested by the increase in the fraction of Triton-insoluble actin in cells treated with the cyclic peptide (see Fig. 5 and discussion below). As shown in Fig. 2B, jasplakinolide partially inhibited the activity of NHE3. This finding implies that NHE3 activity is not a simple function of the cellular F-actin content and suggests that an intact F-actin network is required for optimal NHE3 function.
Effect of Cytoskeletal Modifiers on the Subcellular Distribution of NHE3-The preceding data suggest that NHE3 may interact physically with the actin cytoskeleton. If this were the case, redistribution of the exchangers might be expected upon disruption of the actin filament network. To analyze this possibility, we compared the distribution of NHE3 before and after treatment with agents that influence actin polymerization. To visualize NHE3, cells transfected with epitope-tagged antiporters (NHE3Ј-HA) were used. Typical results are illustrated in Fig. 3. As described earlier (5) only a fraction of the cellular NHE3 resides at the plasma membrane; a sizable fraction is distributed in a punctate pattern that corresponds to sorting and recycling endosomes. NHE3 redistributes upon addition of cytochalasin B or latrunculin B, accumulating at or near the edges of the cell (Fig. 3, B and C). The sites of accumulation coincide with areas of actin aggregation (see the arrows in Fig.   3, B, C, F, and G), suggesting an interaction between NHE3 and the cytoskeleton. Jasplakinolide also altered the distribution of NHE3, but it was difficult to discern whether this was caused primarily by the shape change induced by the actin stabilizer (Fig. 3, D and H). Co-localization of NHE3 with actin in jasplakinolide-treated cells could not be analyzed due to competition of the compound with phalloidin. 2 The preceding results, obtained with cells expressing NHE3 epitope tagged at the carboxyl terminus and stained after permeabilization, cannot resolve whether disruption of the actin filament network redistributes NHE3 that is located intracellularly or at the plasmalemma. More importantly, such experiments cannot discern whether the inhibition of transport is associated with a reduction in the density of cell surface NHE3. To selectively visualize the surface exchangers, we used externally tagged NHE3 and immunostained without permeabilization. Typical results are illustrated in Fig. 4. The distribution of NHE3 at the surface of control cells is rather homogeneous, with faint punctation over a diffuse background (Fig. 4A). Cytochalasin B induced a redistribution of the surface exchangers, promoting their clustering (Fig. 4B) in areas underlied by aggregates of actin (not shown). Similar observations were made in cells treated with latrunculin (not illustrated). Jasplakinolide promoted the formation of smaller NHE3 clusters at the surface of the cells (Fig. 4C).
There was no striking difference in the overall amount of surface staining of control or cytochalasin-or jasplakinolidetreated cells. However, visual quantitation of plasmalemmal exchangers is limited by possible differences in focal plain and by heterogeneity in the population, which introduces consider- 2 We attempted to visualize actin by immunostaining. However, jasplakinolide was found to also mask the epitopes recognized by the monoclonal and polyclonal antibodies to actin that we tested. We were therefore unable to visualize actin in jasplakinolide-treated cells. able variance when comparing small numbers of cells by microscopy. To quantify the effects of the actin modifiers on the surface density of NHE3, we used a modified enzyme-linked immunosorbent assay, where the number of exchangers was measured by binding anti-HA antibody to the external epitope tag followed by a secondary, peroxidase-coupled antibody and assay of bound peroxidase activity (see "Experimental Procedures"). As summarized in Fig. 4D, at concentrations shown above to inhibit NHE3 activity, neither cytochalasin B nor jasplakinolide altered the number of exchangers at the cell surface.
Biochemical Assessment of the Interaction between NHE3 and the Cytoskeleton-Cytoskeletal proteins are reportedly insoluble in mild nonionic detergents (e.g. Ref. 24). To directly evaluate the possible interaction of NHE3 with the actin cytoskeleton, the relative fractions of NHE3 and actin that remain in the insoluble pellet after extraction with 0.5% Triton X-100 and centrifugation were quantified by immunoblotting. As shown in Fig. 5D, in untreated cells about 40% of the total cellular actin remains insoluble after extraction with Triton. Under these conditions, approximately 10% of the cellular NHE3 is in the insoluble fraction (not shown). This fraction resembles the percentage of total cellular NHE3 that is on the plasma membrane (7). Despite inducing a marked disorganization of the actin network (Fig. 3), cytochalasin B decreased the fraction of insoluble actin only moderately (Fig. 5, C and D). In parallel, the cytoskeleton-associated NHE3 decreased modestly (Ϸ20%) (Fig. 5, A and B). Latrunculin B induced a more profound solubilization of actin (Fig. 5, C and D) that was also accompanied by a modest decrease in cytoskeleton-associated NHE3. Last, jasplakinolide greatly elevated the fraction of detergent-insoluble actin (Fig. 5, C and D). Interestingly, despite the increased amount of precipitable actin, the fraction of insoluble NHE3 decreased significantly in cells treated with jasplakinolide, paralleling the effects of this drug on NHE3 activity. Although these observations are suggestive of an interaction between NHE3 and the actin microfilament network, the imperfect correlation implies an indirect association, possibly involving other actin-binding proteins.
Structural Analysis of the Regulation of NHE3 by Cytochalasin B-We next sought to determine the domain that confers cytochalasin sensitivity to NHE3. Because the sensitivity of NHE3 to actin modifiers is not shared by NHE1, the initial approach involved swapping the cytosolic tails of these two isoforms. One chimera consisted of the amino-terminal transmembrane domain of NHE1 (residues 1-504) and the carboxylterminal cytoplasmic domain of NHE3 (residues 456 -831) (termed NHE1/3), and the reciprocal chimera was comprised of residues 1-455 of NHE3 and 505-820 of NHE1 (called NHE3/1).
These were stably expressed in AP-1 cells, and their activity was analyzed measuring pH i fluorimetrically, as described above. The Na ϩ -induced recovery from an acid load was greatly inhibited in NHE1/3 cells by pretreatment with cytochalasin B (Fig. 6A). By contrast, the response of the NHE3/1 chimera was barely altered by cytochalasin B (Fig. 6B), implying that the cytoskeleton-associating element(s) of NHE3 reside in its carboxyl-terminal domain.
More precise, mapping of the region(s) that interacts with the cytoskeleton was accomplished using truncated mutants of NHE3. The NHE3⌬684 mutant, which lacked the carboxylterminal 147 residues, was markedly inhibited by cytochalasin B (Fig. 7A), resembling full-length NHE3. However, the inhibitory effect of cytochalasin B on a shorter truncated form lacking 193 residues, NHE3⌬638, was greatly reduced (Fig. 7B). This suggests that the region encompassing residues 638 -684 is essential for modulation of NHE3 by the cytoskeleton.
The effect of the truncations could be indirect, reflecting severe changes in the conformation of the remaining segment of the protein. This possibility was examined using a chimeric construct called NHE3/1/3, where residues 650 -716 of NHE3 were substituted by the corresponding residues of NHE1 (666 -749, based on sequence alignment). This chimera lacks the region in the middle of the tail of NHE3 that appears to be involved in cytoskeletal regulation yet would bear more resemblance to the original architecture of the protein than the truncated mutants. The transport activity of this chimera was insensitive to cytochalasin (not illustrated). Similar results were also obtained with an internal deletion mutant of NHE3 lacking residues 650 -716 (not shown). Jointly, these results indicate that the mid-portion of the cytosolic tail of NHE3 is largely responsible for the reported effects of cytoskeletal disruption.

DISCUSSION
The central findings of these studies are that NHE3 associates with the cytoskeleton and that its activity depends on the state of actin organization. Physical interaction was deduced from the ability of NHE3 to co-sediment with the cytoskeleton following Triton X-100 extraction, whereas functional association was inferred from the inhibitory effects of actin modifiers.
Disruption of microfilaments by capping their barbed ends or by scavenging away monomeric actin resulted in a marked inhibition of the exchanger. Unexpectedly, excess formation of F-actin also inhibited NHE3, although to a smaller extent. These observations indicate that the activity of NHE3 is not a simple function of the amount of polymerized actin and imply that optimal Na ϩ /H ϩ exchange requires an intact, native cytoskeletal organization. Alternatively, dynamic cycles of actin polymerization and depolymerization may be required for NHE3 function.
One possible mode of regulation by the cytoskeleton may involve NHE3 endocytosis. In both native (epithelial) and heterologous transfection systems, a sizable fraction of NHE3 is located in endosomes, from which it recycles to the membrane (5, 29), and it was recently shown that inhibition of NHE3 activity is observed when an imbalance between the rates of exo-and endocytosis is imposed (7). Unpublished observations from our laboratory indicate that endocytosis is mediated, at least in part, by clathrin-coated vesicles. In this regard, it is relevant that clathrin-mediated endocytosis at the apical membrane requires an intact actin cytoskeleton (30). One could therefore envisage a reduction in the density of plasmalemmal NHE3 upon impairment of recycling by cytochalasin and other actin modifiers, accounting for the observed inhibition. How- ever, this hypothesis is not tenable, based on the results of direct quantitation of plasmalemmal NHE3. Both immunofluorescence and enzyme-linked immunosorbent assay determinations showed little change in NHE3 exposure with either cytochalasin B or jasplakinolide (Fig. 4).
Because the number of transporters at the cell surface appears to be unaltered, inhibition of exchange must instead be due to changes in their intrinsic activity. The precise mechanism underlying this effect remains unknown, but possible effectors are suggested by the structural analyses summarized in Figs. 6 and 7 and in data not shown. These studies revealed that the region delimited by residues 650 and 684 plays an essential role in the cytoskeletal control of NHE3. This region overlaps with that defined by Yun et al. (15) to mediate the interaction of the exchanger with NHERF2 (residues 585-660). Thus, it is conceivable that NHERF modulates NHE3 activity in a manner that depends on the integrity of the actin network. Ezrin or other members of the ezrin/radixin/moesin family, which possess both NHERF-and actin-binding sites would be ideally suited to mediate the proposed interaction. Alternatively, ezrin may bind directly to the critical region of NHE3. Indeed, it is noteworthy that an RKRL sequence (residues 656 to 659) is present within this overlapping domain. Similar cationic/hydrophobic sequences were recently described by Yonemura et al. (31) in CD44, CD43, and ICAM-2 (KKL, KKRT and RRRT, respectively) to mediate ezrin/radixin/moesin-protein binding.
The means whereby the cytoskeleton controls the activity of NHE3 remain undefined, but a couple of possibilities can be considered. The interaction with NHERF, ezrin, and/or other proteins may maintain NHE3 in a state of optimal activity by sequestering an autoinhibitory domain. The presence of an equivalent autoinhibitory domain has been invoked for NHE1 (3). Alternatively, the conformational change resulting from interaction with the cytoskeleton may favor transport by altering the microenvironment surrounding the transport site, e.g. by increasing the pK of the putative H ϩ -binding site. These and other hypotheses need to be tested experimentally in the future.
The involvement of the actin cytoskeleton in modulating NHE3 activity is not without precedent in other transport systems. A number of channels, pumps, and symporters are affected by agents that modify the actin network. For instance, patch clamp analysis revealed that phalloidin can inhibit K ϩ and Na ϩ channels (32,33). Similarly, jasplakinolide prevents the activation of Cl Ϫ secretion induced by cAMP (34), which is believed to be mediated by a conductive pathway. Cytochalasins have been found to stimulate Na ϩ -K ϩ -2Cl Ϫ cotransport (34) and to activate Na ϩ /K ϩ -ATPase (35). In contrast, some K ϩ channels are inactivated by prior treatment with cytochalasin (36). To our knowledge, the detailed molecular mechanisms responsible for cytoskeletal control of ion transport in these systems remain largely unknown.
In addition to the cytoskeleton, NHE3 activity is also regulated by protein kinase A (1, 3). Although the inhibitory effect of the kinase is partly due to direct phosphorylation of NHE3, its effects on NHERF1 and NHERF2 or other heretofore unidentified substrates remains the subject of debate (see Refs. 10 -12 and 14). In this context, it is interesting to note that in various cell types elevation of cAMP can induce cytoskeletal alterations that include disruption of actin filaments and decreased phosphorylation of the light chain of myosin (37)(38)(39).
This raises the possibility that the effects of protein kinase A on NHE3 may also be indirect, through alteration of the cytoskeleton. In addition, the cellular content of F-actin is known to vary in response to osmotic challenges (40). Hence, the inhibition of NHE3 triggered by osmotic cell shrinkage (26) may be similarly mediated by reorganization of the actin cytoskeleton. These possibilities are currently being tested experimentally.