A Slow pH-dependent Conformational Transition Underlies a Novel Mode of Activation of the Epithelial Na+/H+ Exchanger-3 Isoform*

Allosteric control of Na+/H+ exchange by intracellular protons ensures rapid and accurate regulation of the intracellular pH. Although this allosteric effect was heretofore thought to occur almost instantaneously, we report here the occurrence of a slower secondary activation of the epithelial Na+/H+ exchanger (NHE)-3 isoform. This slow activation mode developed over the course of minutes and was unique to NHE3 and the closely related isoform NHE5, but was not observed in NHE1 or NHE2. Activation of NHE3 was not due to increased density of exchangers at the cell surface, nor was it accompanied by detectable changes in phosphorylation. The association of NHE3 with the cytoskeleton, assessed by its retention in the detergent-insoluble fraction, was similarly unaffected by acidification. In contrast to the slow progressive activation elicited by acidification, deactivation occurred very rapidly upon restoration of the physiological pH. We propose that NHE3 undergoes a slow pH-dependent transition from a less active to a more active state, likely by changing its conformation or state of association.

Na ϩ /H ϩ exchangers (NHEs) 1 catalyze the electroneutral counter-transport of Na ϩ for H ϩ across biological membranes (for review, see Refs. [1][2][3]. To date, seven different NHE isoforms have been cloned, all of which are integral membrane proteins with multiple transmembrane domains and an extensive cytosolic carboxyl-terminal domain. NHE1 and NHE3 are by far the most widely studied isoforms. The former is expressed ubiquitously and, by mediating Na ϩ /H ϩ exchange across the plasma membrane, contributes to the regulation of cytosolic pH and cell volume (2). NHE3 is found in the brush border of epithelial cells in the kidney and gastrointestinal tract and is involved in the transepithelial (re)absorption of NaCl and, indirectly, HCO 3 Ϫ and water (4).
Because Na ϩ /H ϩ exchange is electroneutral with a 1:1 stoichiometry, thermodynamic equilibrium is predicted to be attained when [H ϩ where K e is the equilibrium constant and the subscripts i and o refer to intraand extracellular, respectively (5). In almost all cells, a steep inwardly directed Na ϩ gradient is maintained across the plasma membrane due to extrusion of Na ϩ by the Na ϩ /K ϩ -ATPase. Based on the concentrations recorded in most cases ([Na ϩ ] o being at least 10 times higher than [Na ϩ ] i ), it is apparent that NHE could drive the intracellular pH (pH i ) at least 1 unit above the external pH. Nevertheless, pH i rarely exceeds and is usually lower than the extracellular pH because NHE becomes virtually quiescent at pH i Ն7.2. This behavior has been attributed to the existence of an allosteric pH-sensitive site on the cytosolic aspect of NHE. The set point at which quiescence is dictated by this allosteric or "modifier" site is presumably intended to stabilize pH i near the physiological optimum level and to preclude deleterious alkalinization (6,7).
Although little is known about the molecular identity of the modifier site, the pH dependence of activation of NHE suggests that the rate of transport is modulated by protonation of one or more side chains of the protein. It has been postulated that the resulting change in surface potential may alter the local concentration of H ϩ equivalents and thus their availability for the exchange reaction. Because protonation/deprotonation of side chains is expected to occur almost immediately after pH is altered, the effect of the modifier site has been tacitly assumed to occur instantaneously, providing rapid feedback and thereby accurate regulation of pH i .
In the course of experiments using isolated brush-border membranes at subphysiological temperatures, we detected significant hysteresis in the inactivation of NHE3 (preliminary results are described in Ref. 8). This slow responsiveness, which developed over the course of minutes, cannot be easily reconciled with the simple side chain protonation model described above. We therefore decided to investigate whether slow activation and/or inactivation transitions are observed in intact cell systems and to examine the underlying mechanism. To this end, we used porcine kidney LLC-PK 1 cells, which are known to express endogenous NHE3 (9). In addition, we compared the behavior of several isoforms of NHE expressed heterologously in antiport-deficient Chinese hamster cells. Our results revealed a novel mode of regulation that is unique to NHE3 and the related isoform NHE5 and that likely involves large conformational changes and/or interaction with other molecules.
Cells-LLC-PK 1 cells were obtained from the American Type Culture Collection. For surface detection and immunoprecipitation experiments, LLC-PK 1 cells were transfected with wild-type NHE3 containing three tandem copies of the influenza virus HA epitope (YPYDVPD-YAS) in the first extracellular loop (between Arg 38 and Phe 39 ) as described previously (13). LLC-PK 1 cells were transfected using Fu-GENE 6 (Roche Molecular Biochemicals) following the manufacturer's instructions. For selection of stable lines, cells were cotransfected with the pCMV plasmid (which contains the aminoglycoside phosphotransferase gene that confers resistance to G418), selected by limiting dilution cloning in the presence of 500 g/ml G418, and screened by immunofluorescence for expression of HA-tagged NHE3 (NHE3Ј 38HA3 ). Cells were maintained in 1:1 Dulbecco's modified Eagle's medium/nutrient mixture F-12 with 10% fetal bovine serum in an atmosphere containing 5% CO 2 .
AP-1 is a cell line devoid of endogenous Na ϩ /H ϩ exchange activity that was isolated from WT5 Chinese hamster ovary cells as previously described (10). NHE cDNA constructs were transfected into AP-1 cells by the calcium phosphate/cDNA coprecipitation technique, and stable clones were selected for survival by imposing acute NH 4 Cl-induced acid loads (11,12). Cells were maintained in ␣-minimal essential medium with 10% fetal bovine serum in an atmosphere containing 5% CO 2 .
Measurement of Na ϩ /H ϩ Activity-pH i was measured by microphotometry of the fluorescence emission of BCECF using dual-wavelength excitation. The activity of NHE was determined as the rate of Na ϩinduced pH i recovery after acid loading with NH 4 Cl. LLC-PK 1 cells grown to confluence on 25-mm glass coverslips were incubated with 2 g/ml BCECF/acetoxymethyl ester plus 15 mM NH 4 Cl; and after 10 min, the cells were washed with Na ϩ -free solution. Na ϩ /H ϩ exchange was then initiated by reintroduction of extracellular Na ϩ at the indicated times. Clusters of the cells adhering to the coverslip were selected for pH i measurements, whereas cells that formed domes were avoided because, in these, the Na ϩ -induced pH i recovery likely occurred via basolateral NHE. To minimize the consequences of dye leakage from the cells, the chamber was continuously perfused at Ϸ1.0 ml/min using a gravity-driven system. The fluorescence of BCECF was measured and calibrated as described previously (13). The cellular buffering capacity was determined from the pH i changes observed in response to NH 4 Cl pulses imposed at various pH i levels. H ϩ (equivalent) flux rates were calculated by multiplying the rate of pH i recovery by the buffering capacity at the corresponding pH values. All procedures were performed at room temperature.
Intracellular Na ϩ Content Manipulation and Determination-LLC-PK 1 cells were plated onto six-well plates and grown to subconfluence. We did not use confluent cells because of potential trapping of extracellular fluid under domes. Intracellular Na ϩ depletion was accomplished by incubating the cells in Na ϩ -free solution. Intracellular Na ϩ content was determined by flame photometry (Model 443 photometer, Instrumentation Laboratory, Lexington, MA) using Li ϩ as an internal standard. After intracellular Na ϩ depletion, the cells were washed with ice-cold medium containing 300 mM sucrose and 10 mM HEPES/Tris (pH 7.4). The cells were scraped off with a rubber policeman into Li ϩ standard solution (Instrumentation Laboratory); the samples were frozen and thawed; insoluble material was sedimented; and the supernatant was used for measurement. Intracellular Na ϩ content was normalized by the number of cells used. Cells were counted electronically with a Coulter Counter (Coulter Electronics, Hialeah, FL).
Immunofluorescence-LLC-PK 1 cells stably expressing NHE3Ј 38HA3 were plated onto glass coverslips and grown to confluence. The cells were rinsed with phosphate-buffered saline (PBS) and, where indicated, acidified for 5 min with NH 4 Cl as described above before fixation for 20 min using 8% paraformaldehyde in ice-cold PBS. Following fixation, the cells were washed three times with PBS and incubated with 100 mM glycine in PBS for 10 min. The cells were next pre-blocked with 5% skimmed milk in PBS for 30 min and then incubated with anti-HA monoclonal antibody (1:1000 dilution) for 1 h. After washing three times to remove unbound antibody, the cells were incubated with Alexa 488-conjugated donkey anti-mouse antibody (1:1000 dilution) for 1 h, and the coverslips were mounted onto glass slides with mounting me-dium (Dako Corp., Carpinteria, CA).
Detergent Extraction and Immunoblotting-The detergent solubility of NHE3Ј 38HA3 was assessed in transfected LLC-PK 1 cells cultured in 35-mm dishes and grown to confluence. Where indicated, the cells were first acid-loaded to pH 6.4 as described above. The cells were then rinsed twice with ice-cold medium and extracted with 1 ml of ice-cold lysis medium containing 150 mM NaCl, 25 mM HEPES, 25 mM MES, 0.5 mM EGTA, and 8% protease inhibitor mixture (Complete, Roche Molecular Biochemical) plus 2 M pepstatin and either 0.1% Triton X-100 or 1% digitonin. The pH of the medium was adjusted to either 6.4 or 7.4 as indicated. After swirling on ice for 6 min on an orbital shaker, the supernatant was removed and saved. The adherent material was washed with ice-cold PBS and scraped off with a rubber policeman into 1 ml of ice-cold PBS containing 0.1% SDS and 8% protease inhibitor mixture plus 2 M pepstatin. Protein concentration was measured by the method of Ghosh et al. (14) and quantification by densitometry (AlphaImager, AlphaInnotech, Cannock, United Kingdom). Following addition of 5ϫ concentrated Laemmli sample buffer, samples were subjected to SDS-PAGE and transferred onto nitrocellulose filters. Blots were blocked with 5% milk and exposed to primary antibody (anti-HA (1:5000) or anti-actin (1:250) antibody). Horseradish peroxidase-conjugated secondary antibody was applied (1:5000 dilution), and immunoreactive bands were visualized using enhanced chemiluminescence (ECL, Amersham Biosciences, Inc.) and quantified by densitometry.
Quantification of Surface NHE3-To quantify surface NHE3 expression, LLC-PK 1 cells stably expressing NHE3Ј 38HA3 were plated onto six-well plates and grown to confluence. The cells were rinsed with PBS and, where specified, acidified for 5 min with NH 4 Cl as described above. The cells were next washed with ice-cold PBS, pre-blocked with 10% goat serum in PBS for 30 min on ice, and then incubated for 45 min on ice with anti-HA antibody (1:500 dilution) in medium with 10% goat serum. After washing four times to remove unbound antibody, the cells were incubated with 125 I-labeled goat anti-mouse IgG (0.4 Ci/well) in medium with 10% goat serum on ice. After 60 min, cells were washed four times with ice-cold PBS to remove unbound radioactivity. The radiolabeled antibodies were eluted with 1 ml of 2 M formic acid, and radioactivity was counted using a ␥-counter (1282 Compu Gamma, Amersham Biosciences, Inc., Turku, Finland). Nonspecific binding of the radiolabeled antibodies, assessed by omitting primary antibodies, was subtracted.
Assessment of NHE3 Phosphorylation-LLC-PK 1 cells stably expressing NHE3Ј 38HA3 grown to confluence on 15-cm dishes were labeled for 4 h by incubation with 4 ml of phosphate-free Dulbecco's modified Eagle's medium containing 0.4 mCi of [ 32 P]orthophosphate. After labeling, the cells were washed three times with PBS and, where specified, acidified with NH 4 Cl for 5 min as described above. To enrich the preparation in NHE3 prior to immunoprecipitation, we prepared brushborder membrane vesicles from the labeled LLC-PK 1 cells by the method of Brown et al. (15), with minor modifications. The cells were scraped off with a rubber policeman into 1 ml of ice-cold medium containing 300 mM mannitol, 5 mM NaF, 2 mM Na 3 VO 4 , 1% protease inhibitor (Sigma), and 10 mM HEPES/Tris (pH 7.4) and homogenized with a bath sonicator (Fisher Model 300, setting 60) for 5 min. Next, MgCl 2 was added to a final concentration of 10 mM, followed by incubation for 30 min on ice and centrifugation for 15 min at 2300 ϫ g. The supernatant was collected and subjected to centrifugation at 22,200 ϫ g. The resulting pellet containing the brush-border vesicles was resuspended in 1 ml of radioimmune precipitation assay buffer (150 mM NaCl, 0.1% SDS, 1% Triton X-100, 50 mM NaF, 2 mM Na 3 VO 4 , 1% protease inhibitor, and 20 mM HEPES/Tris at pH 7.4 or 6.4) by drawing through a 26-gauge needle and shaken on a rotating rocker for 30 min at 4°C. After centrifugation at 88,800 ϫ g for 30 min to remove insoluble cellular debris, the supernatant was added to protein G-Sepharose beads (Pierce) that were pre-conjugated with anti-HA monoclonal antibody, and the suspension was shaken for 2 h at 4°C. The beads were spun down and washed eight times with radioimmune precipitation assay buffer. Immunoprecipitated proteins were eluted by incubation in 50 l of Laemmli sample buffer at 65°C for 15 min. An aliquot of 40 l of each sample was analyzed by SDS-PAGE and electrophoretically transferred to a Trans-Blot membrane (Bio-Rad). The relative amounts of NHE3 in the samples were determined by ECL and quantified using the AlphaImager. Once the enhanced chemiluminescence decayed, 32 P signals from the same membranes were subsequently measured using a PhosphorImager and quantified using ImageQuant (Molecular Dynamics, Inc.).
Statistical Analysis-Experimental values are given as the means Ϯ S.E. of the indicated number of determinations. Comparisons between two groups were made by either unpaired or paired Student's t tests, as appropriate.

RESULTS
Time Dependence of NHE3 Activation-The time dependence of the activation of NHE3 was studied in LLC-PK 1 cells by monitoring the Na ϩ -induced recovery of pH i following an acid load. Controlled cytosolic acidification was imposed by the NH 4 Cl (15 mM) prepulse technique (16), and Na ϩ /H ϩ exchange was initiated when desired by addition of 140 mM Na ϩ to the medium. It is noteworthy that (i) in the absence of Na ϩ , NHEindependent pH i recovery was negligible (0.003 pH units/min); and (ii) the acid loading was essentially complete at the time of addition of Na ϩ (Fig. 1A), so the rate of pH recovery was uncontaminated by an opposing tendency to acidify due to residual intracellular ammonium. Only in the case of the measurements made 1 min after removal of ammonium was a small acidification still apparent. This rate of acidification was measured in parallel experiments (0.011 mM/s) and was added to the determinations of NHE based on the net rate of alkalinization.
Two measures were taken to ensure that the activity of NHE3 was assessed without contribution from NHE1, which is expressed in the basolateral membrane of LLC-PK 1 cells (17). First, the cells were grown to confluence, and Na ϩ was added only apically. Second, 100 M amiloride was added to selectively inhibit NHE1 (18).
Under the conditions used (pH i 6.53 Ϯ 0.03, n ϭ 25), the Na ϩ -induced pH i recovery was slow when the cells were acidified for only 1 min. However, the rate of alkalinization increased gradually as the period of acidification was lengthened (Fig. 1A). The open squares in Fig. 1C summarize the results of six experiments: the rate of Na ϩ /H ϩ exchange increased progressively up to 5 min and decreased slowly thereafter. The maximal rate observed was 2-fold greater than the rate measured at 1 min. Shorter periods were not analyzed because the acid loading procedure was still ongoing before 1 min (see above).
A limited number of experiments were also performed using opossum kidney cells, with similar results (data not shown). These findings indicate that the slow activation of Na ϩ /H ϩ exchange is a widespread phenomenon and suggest that the observations in LLC-PK 1 cells were attributable to NHE3 because opossum kidney cells express NHE3, but not NHE1 (19,20).
The faster rate observed after prolonged acidification is likely to reflect increased activity of NHE3. However, it is conceivable that the cytosolic buffering power diminishes over time after acidification, resulting in a faster recovery at constant NHE activity. The latter possibility was explored using an exogenous ion exchanger. As shown in Fig. 1B, the pH i recovery induced by nigericin and high K ϩ solution was similar whether the ionophore was added 1 or 5 min after acidification was imposed. The nigericin-induced recovery averaged 0.05 Ϯ 0.01 and 0.04 Ϯ 0.01 mM/s at 1 and 5 min, respectively (Fig. 1C, closed squares). These observations rule out a reduction in buffering power and also other possible time-dependent artifacts in the recording system. Intracellular Na ϩ Depletion Does Not Contribute to an Increase in NHE3 Activity-Imposition of an acid load of variable duration required incubation in Na ϩ -free medium for different periods of time. This may have resulted in variable intracellular Na ϩ contents at the time of assaying NHE. The observed time-dependent increase in activity may therefore have resulted from an increased driving force for forward Na ϩ /H ϩ exchange or from stimulation of a Na ϩ -sensitive G protein, as has been invoked for the regulation of NHE1 (21). To analyze this possibility, cells were initially preincubated in Na ϩ -free medium for 10 min and then subjected to acid loading and challenged with extracellular Na ϩ . As illustrated in Fig. 2A, the rate of recovery after 1 min of acidification was not significantly different from that recorded in control cells in Fig. 1 (0.035 Ϯ 0.003 versus 0.034 Ϯ 0.003 mM/s, respectively; p ϭ 0.87). More importantly, the time-dependent acceleration of NHE3 activity persisted in the Na ϩ -depleted cells: the rate of pH recovery was nearly 3-fold higher after 5 min than after 1 min (Fig. 2, A and B).
To confirm that depletion of Na ϩ was not the mechanism underlying stimulation, we measured the intracellular content of this cation by flame photometry under all the conditions used. Omission of extracellular Na ϩ prior to the acid load reduced intracellular Na ϩ by 45% (from 44 to 20 mM). The magnitude of the depletion was not significantly different (52 Ϯ 5 versus 45 Ϯ 8%; p ϭ 0.61) when Na ϩ removal was extended for 15 min, indicating that no substantive further depletion occurred during the acid loading period in Fig. 2. These results indicate that intracellular Na ϩ depletion does not contribute to the time-dependent activation of NHE3.
pH i Dependence of the Activation of NHE3-To analyze the effect of pH i on the kinetics of NHE3 activation (Fig. 3), various concentrations of NH 4 Cl were used during the prepulse. The time dependence of the activation was recorded at all the pH i levels that we measured (between 6.5 and 7.5), and in all cases, the differential between 1 and 5 min was Ն2-fold. Interestingly, when plotting efflux rate versus [H ϩ ] i (Fig. 3C), the data conformed to Michaelis-Menten kinetics when the acidification was imposed for 5 min (R 2 ϭ 0.781), but the fit was poorer after 1 min (R 2 ϭ 0.453). This observation suggests that a more complex process involving the time-dependent activation of NHE3 is ongoing at 1 min. If activation is complete by 5 min, only the expected pseudo first-order kinetics is noted.
The Number of Surface NHE3 Molecules Does Not Change during Acidification-In native as well as in heterologous expression systems, NHE3 is present in at least two subcellular compartments: the plasma membrane and an endomembrane vesicular compartment that includes subapical/recycling endosomes (13,22,23). In opossum kidney cells, it was shown that chronic exposure to acidic solutions increases the activity of NHE3 and that such an increase is accompanied by mobilization of endomembrane NHE3 to the cell surface (24). It was therefore conceivable that the number of surface NHE3 molecules may change also during acute intracellular acidification. To assess this possibility, the abundance of cell-surface NHE3 was quantified by two methods. First, we generated a stable clone of LLC-PK 1 cells expressing an epitope-tagged form of NHE3. The epitope was placed in an extracellular loop to enable detection of surface-exposed NHE3 in intact cells (see Ref. 25 for details). Immunostaining and fluorescence imaging analysis of intact (non-permeabilized) cells before and after acidification revealed no visible differences in the density or distribution of NHE3 (Fig. 4A). Second, to more accurately quantify surface NHE3, we used 125 I-labeled antibodies. As shown in Fig. 4B, the surface density of NHE3 did not change following acidification. These results imply that the intrinsic activity, and not the number of exchangers, was increased when acid loading was sustained for 2-5 min.
Phosphorylation of NHE3 Does Not Change during Acidification-It is well established that NHE3 can become phosphorylated on Ser residues and that phosphorylation is accompa-nied by changes in transport activity (26 -29). In addition, Tyr phosphorylation may play a role in NHE3 regulation because stimulation of exchange by chronic acidification is associated with activation of c-Src (30). To define whether the more acute acid-induced stimulation reported here was similarly attributable to phosphorylation, cells were preincubated with [ 32 P]orthophosphate and subjected to acidification. The cells were then immediately lysed in the cold, and NHE3 was immunoprecipitated. Quantitation of the radioactivity following analysis by SDS-PAGE revealed that the extent of phosphorylation of NHE3 was essentially identical before and after acidification for 5 min (Fig. 4B). Therefore, a mechanism other than direct phosphorylation must account for the observed time-dependent activation of NHE3.
Biochemical Assessment of the Interaction between NHE3 and the Actin Cytoskeleton-We previously demonstrated a functional interaction between NHE3 and the actin cytoskeleton in AP-1 cells (31). In that study, the activity of the antiporter was modulated by the extent of actin polymerization, raising the possibility that the cytoskeleton may contribute to

FIG. 3. Effects of pH i and length of acidification on the activation of NHE3.
A, Na ϩ -induced pH i recovery after 1 min of acidification. LLC-PK 1 cells were stained with BCECF. To clamp pH i at the desired level, the cells were incubated with varying concentrations of NH 4 Cl, determined empirically in preliminary experiments. Thirty seconds after initiation of the trace, the cells were transferred to Na ϩ -free solution devoid of NH 4 Cl, resulting in cytosolic acidification. Where indicated by the arrow, Na ϩ was reintroduced. Three different representative traces are superimposed. Amiloride (100 M) was present throughout to minimize the contribution of NHE1. B, Na ϩ -induced pH i recovery after 5 min of acidification. The cells were acidified for 5 min, and recovery was induced by reintroduction of Na ϩ (arrow) as described for A. Three different representative traces are superimposed. C, pH i dependence of Na ϩ -induced H ϩ (equivalent) efflux following 1-min (Ⅺ) and 5-min (f) acidification periods. H ϩ fluxes were calculated by multiplying the rate of pH i change by the cytosolic buffering power, measured independently as described under "Experimental Procedures." The hyperbolic curves were calculated by fitting data from 15 and 17 experiments for 1-and 5-min acidification periods, respectively, to the following equation: flux ϭ V max ⅐[H ϩ ]/(K m ϩ [H ϩ ]). R 2 ϭ 0.781 and 0.453 for the 5-and 1-min curves, respectively. the pH-dependent effects reported here. To test this possibility, we examined the association of NHE3 with the detergentinsoluble fraction of LLC-PK 1 cells, which contains (part of) the actin cytoskeleton. Cells were left untreated or were acidloaded to pH 6.4 as described for the determinations of exchange activity. Next, the cells were extracted in detergent at pH 6.4 or 7.4, and the soluble and insoluble fractions of NHE3 were quantified by immunoblotting (Fig. 5A). For reference, the total amounts of protein and actin solubilized were also quantified in the extracts. When using Triton X-100, only a fraction of NHE3 was solubilized, regardless of the pH of the extraction solution (27 Ϯ 2 versus 31 Ϯ 1%, pH 7.4 and 6.4, respectively). This implies that the majority of the exchangers are constitutively associated with the cytoskeleton. Under the conditions used, the Triton X-100-insoluble fraction may also include glycolipid-enriched microdomains, known as rafts (32). It was therefore conceivable that the lipid partitioning of NHE3, as opposed to its cytoskeletal interactions, dictated the insolubility of the exchanger. This alternative explanation was tested using digitonin, another nondenaturing detergent that, unlike Triton X-100, preferentially interacts with and extracts cholesterol from biological membranes. Digitonin solubilized a slightly larger fraction of NHE3 (42 Ϯ 8%) than did Triton X-100, favoring the notion that the exchanger is rendered insoluble primarily through its interaction with the cytoskeleton.
Importantly, the fraction of the exchangers that were extractable by Triton X-100 was not significantly altered when the cells were acidified before and/or during the extraction procedure (Fig. 5, A and B). The amounts of protein and actin solubilized in both instances were also comparable (Fig. 5B), and similar results were obtained with digitonin (data not shown).
Jointly, these results suggest that the interaction of NHE3 with the cytoskeleton is not grossly altered by the acid loading procedure. It therefore appears unlikely that the cytoskeleton plays a role in the time-and pH-dependent activation of the exchanger. This concept was further supported by observations made in cells transfected with NHE3⌬638, a truncated version of NHE3 that was shown earlier to be insensitive to cytoskeletal alterations (31). As shown in Fig. 5C, the rate of Na ϩ /H ϩ exchange in these transfectants was accelerated when the acidification period was extended, as shown above for the endogenous (full-length) exchanger of epithelial cells. The rate of recovery was nearly 3-fold greater after 3 min than after 1 min (Fig. 5D). These observations strongly suggest that the cy- Kinetics of NHE3 Inactivation following Slow Activation-We next assessed the rate at which NHE3 inactivates following slow activation by a sustained acid load. The cells were first acidified for 5 min to activate NHE3 fully, omitting Na ϩ from the medium to preclude pH recovery during this period. Next, the cytosolic pH was increased as fast as possible by reintroducing an amount of NH 4 Cl found empirically to elevate pH to near base-line (pre-acid loading) levels. Finally, Na ϩ was added to the bathing medium to initiate Na ϩ /H ϩ exchange. Reintroduction of Na ϩ failed to elevate pH beyond the resting levels ( Fig. 6), implying that NHE was not more active than it was prior to acid loading. This indicates that NHE3 inactivated rapidly (within 1 min) after restoration of neutral pH, so inactivation appeared to be much faster than the slow activation process, which developed over 5 min.
Is Slow Activation an Intrinsic Property of NHE3?-It could be argued that the slow activation of NHE3 reported above is a property of the LLC-PK 1 cells in which its activity was analyzed, rather than being characteristic of this isoform. This concern was addressed by studying the behavior of NHE3 expressed heterologously in AP-1 cells. AP-1 cells, a subline of Chinese hamster ovary cells, are devoid of endogenous NHE activity (10) and are therefore well suited to analyze the activity of heterologously expressed exchangers. As shown in Fig. 7B, a clear time-dependent activation was noted also in NHE3-transfected AP-1 cells. This finding implies that the slow pH-dependent activation of NHE3 is an intrinsic property of this isoform that is manifested in both epithelial and non-epithelial contexts.
Time Course of Activation of Various NHE Isoforms-It was of interest to establish whether, like NHE3, other NHE isoforms similarly undergo a slow activation following acidification. To this end, we analyzed the behavior of AP-1 cells transfected with defined isoforms of the exchanger. As described above, AP-1 cells enabled us to unambiguously attribute the exchange activity to the heterologously transfected NHE isoform of interest. Typical results are shown in Fig. 7A, illustrating the recovery mediated by NHE1 following acid loading for either 1 or 3 min. As summarized in Fig. 7B, the rate of Na ϩ /H ϩ exchange was not affected by the duration of the acid load. Similar results were obtained in AP-1 cells transfected with NHE2. NHE5, which among the known NHE isoforms shares the greatest homology with NHE3 (33), displayed a significantly greater activity after 3 min than after only 1 min of acidification, although the difference was less striking than in the case of NHE3. Thus, the time-and pH-dependent activation of NHE is isoform-specific, occurring only in NHE3 and the closely related isoform NHE5, but not in NHE1 and NHE2.

DISCUSSION
Our results revealed the existence of a slowly developing activation of NHE3, which required 4 -5 min to attain completion. This comparatively slow transition is unlikely to reflect the rate of protonation of dissociable groups of the side chains of NHE3 and may instead result from sizable conformational changes of the exchanger and/or of ancillary proteins that regulate its activity. Such structural changes could either induce or stabilize an active conformation of the antiporter. As is widely accepted in the case of ion channels, NHE is proposed to exist in an inactive (or poorly active) form (I) and one or more active configurations (A). Protonation would facilitate this transition either by increasing the rate of conversion (k 1 ) or by inhibiting its rate of reversal (k Ϫ1 ). To the extent that other isoforms show a pH-dependent activation, but not the slow transition displayed by NHE3 and NHE5, we suggest that two activated states (A 1 and A 2 ) exist for NHE3 and NHE5 (Equation 1), but only one for the other isoforms.
FIG. 6. Rapid inactivation of NHE3 following prolonged acidification. LLC-PK 1 cells were loaded with BCECF and used for pH i determination as described for Fig. 1. Where indicated, the concentration of NaCl or NH 4 Cl in the perfusate was altered. Note that 6 mM NH 4 Cl restored pH to near normal values after acid loading and that subsequent reintroduction of Na ϩ had little effect. The trace is representative of four experiments.

FIG. 7. Comparison of the effect of varying time of acidification on the activity of NHE isoforms.
A, representative traces of Na ϩ -induced pH i recovery. Chinese hamster ovary cells expressing NHE1 cells were stained with BCECF, incubated with 15 mM NH 4 Cl, and used for measurement of pH i as described under "Experimental Procedures." Ten seconds after initiation of the trace, the cells were transferred to Na ϩ -free solution devoid of NH 4 Cl, resulting in cytosolic acidification. Where indicated by the arrows, Na ϩ was reintroduced. Two different representative traces are superimposed. B, comparison of the responsiveness of various NHE isoforms to sustained acidification. AP-1 cells transfected stably with the NHE isoform indicated were acid-loaded for either 1 (open bars) or 3 min (closed bars), and the rate of Na ϩ -induced recovery of pH i was measured fluorometrically as described under "Experimental Procedures." To facilitate comparison between isoforms, the activity was normalized in each case to that observed after 1 min of acidification. Data are means Ϯ S.E. of at least four experiments of each type. *, p Ͻ 0.05 compared with 1 min of acidification.
The uniquely slow activation from A 1 to A 2 would be dictated by k Ϫ2 , which is presumably lower than k 1 and therefore limiting to the overall activation process. Note that only k 1 (and/or k Ϫ1 ) need to be pH-sensitive, which would agree with our observation that the course of the slow transition was not markedly different when the extent of acid loading was varied.
The molecular nature of the event mediating the slow activation of NHE3 remains to be completely defined. Kinsella et al. (8) explored earlier the possibility that large multimeric complexes of exchangers may form upon acidification. However, cross-linking experiments failed to show evidence that higher order aggregates are formed. Alternatively, when the cytosol is acidified, NHE3 may partition preferentially into detergent-insoluble lipid rafts, as has been suggested for apically targeted proteins in the trans-Golgi network (32). This explanation was rendered unlikely by the observation that the solubility of NHE3 in digitonin, a cholesterol-extracting detergent that effectively disrupts rafts, was similar in control and acid-loaded cells.
A functional relationship between NHE3 and the actin cytoskeleton had been suggested before (31) based on the inhibitory effects of cytoskeletal disruption. Our data showing that most of NHE3 in LLC-PK 1 cells was not solubilized by either Triton X-100 or digitonin (Fig. 5) are consistent with association of the exchangers with the actin cytoskeleton. Accordingly, pretreatment of the cells with Clostridium difficile toxin B, which inhibits Rho family GTPases and causes microvillar retraction (34,35), markedly increased the fraction of NHE3 that is solubilized by Triton X-100. 2 However, the interaction of the antiporter with the cytoskeleton was not noticeably altered by cytosolic acidification (Fig. 5, A and B). Moreover, a truncated form of NHE3 lacking modulation by the cytoskeleton nevertheless displayed the time-and pH-dependent stimulation (Fig. 5, C and D). We therefore believe that the activation of the exchanger is not caused by drastic alterations in its cytoskeletal anchorage.
We tentatively favor the notion that the A 1 3 A 2 transition reflects a change in the conformation of the exchanger. The comparatively slow kinetics of the transition implies that the event requires a large activation energy, suggesting that the conformational change is drastic. Future structural studies should be able to test this prediction.
The observation that NHE3 undergoes a slow activation transition adds to the complexity of regulation of this isoform by H ϩ . In addition to the rapid activation noticeable in all isoforms within seconds of acid loading (likely reflecting the I 3 A 1 transition) and the slower phase described herein (postulated to be the A 1 3 A 2 transition), a much slower stimulation of NHE3 by chronic acidification was described earlier (36 -38). This acceleration of transport differs from the one described herein in that it requires several hours to develop, involves de novo synthesis of mRNA and proteins (38), and is accompanied by increased exocytosis of intracellular NHE3 molecules.
Why are so many forms of regulation of NHE3 by acid required? Rapid responses occurring within seconds are necessary for the acute and fine regulation of the intracellular pH. Those developing over minutes may be intended to compensate for acute variations in systemic pH and bicarbonate concentration. More severe and chronic challenges likely require further and more sustained increases in the rate of transport. This is accomplished by biosynthetic means, at the expense of reduced response time. In this regard, it is noteworthy that the second-ary activation described in this report appears to revert rapidly upon restoration of the normal cytosolic pH (Fig. 6), conferring to the system effective feedback properties. In contrast, the insertion into the brush border of the extra transporters synthesized and mobilized during chronic acidosis would show greater hysteresis, leading to a potentially dangerous overshoot in the rate of acid extrusion when the normal pH is restored. Thus, the various levels of regulation are not redundant, but complementary and individually well suited for fast or sustained challenges to the cellular and systemic pH and bicarbonate homeostasis.