Cytoplasmic domain of the ubiquitous Na+/H+ exchanger NHE1 can confer Ca2+ responsiveness to the apical isoform NHE3.

The Na+/H+ exchanger isoforms NHE1 and NHE3 are regulated differently by various stimuli. Calcium has been recognized as one of the major second messengers in such exchanger regulation. We previously proposed that Ca2+-induced activation of NHE1 occurs via displacement of its autoinhibitory domain from the H+ modifier site due to direct binding of Ca2+/calmodulin. To further validate this hypothesis, the functional role of the cytoplasmic domain was studied in both wild-type and chimeric exchangers, i.e. NHE1, NHE3, NHE1 with the cytoplasmic domain of NHE3(N1N3), and NHE3 with the cytoplasmic domain of NHE1(N3N1). After expression in exchanger-deficient fibroblasts (PS120), early response (<80 s) to external stimuli was assessed as 5-(N-ethyl-N-isopropyl)amiloride-sensitive 22Na+ uptake. Among stimuli tested (ionomycin, α-thrombin, phorbol ester, hyperosmotic stress, and platelet-derived growth factor) that are all known to activate NHE1, only ionomycin and thrombin induced a significant intracellular Ca2+ mobilization and early activation of 22Na+ uptake, implying that Ca2+ is a main regulator of NHE1 in the early phase of the agonist response. However, all the stimuli did not activate NHE3 or N1N3. In contrast, a significant stimulation of 22Na+ uptake in response to ionomycin and thrombin was observed in N3N1, accompanied by an alkaline shift of pHi sensitivity (∼0.2 pH units). Deletion of the cytoplasmic calmodulin-binding domain within N3N1 resulted in a constitutive alkaline shift of pHisensitivity and abolished the activation by ionomycin and thrombin. Together, these data reinforce our concept of Ca2+-induced activation of NHE1. Furthermore, they provide evidence for a functional interaction of the autoinhibitory domain of NHE1 with the H+-modifier site of a different isoform, NHE3.

Calcium ion is an important second messenger in mammalian cells, regulating various cell functions including muscle contraction, secretion, cell cycle progression, and a large variety of nerve cell functions. In many of these processes, elevation of the intracellular Ca 2ϩ concentration ([Ca 2ϩ ] i ) and subsequent Ca 2ϩ -dependent activation of a ubiquitous regulator protein calmodulin (CaM) 1 have been recognized as a major mechanism of signal transduction in response to hormonal stimulation or membrane depolarization (1,2).
The electroneutral plasma membrane Na ϩ /H ϩ exchanger isoform 1 (NHE1) has been shown to be one of the targets regulated by intracellular Ca 2ϩ (3)(4)(5)(6)(7)(8). NHE1 (9) is a ubiquitous amiloride-sensitive transporter that regulates pH i and cell volume (10,11), and its structure-function relationship has been studied extensively (7,(12)(13)(14)(15)(16)(17). We have recently shown that NHE1 is a CaM-binding protein containing high and low affinity CaM-binding sites in the middle of the carboxyl-terminal cytoplasmic domain (17). Based on the analysis of function of NHE1 mutant molecules that do not bind CaM, we proposed that Ca 2ϩ -induced activation of NHE1 occurs via direct binding of Ca 2ϩ /CaM to the high affinity site that has an autoinhibitory function (7). However, further experiments were required to unambiguously confirm this hypothesis because of lack of evidence for the direct effect of Ca 2ϩ /CaM on the exchange activity.
When NHE1 is activated in response to various stimuli such as growth factors, calcium, and hyperosmotic stress, it is generally accepted that pH i sensitivity of Na ϩ /H ϩ exchange increases without an apparent change in V max (18 -20). This is thought to result from increased affinity of the allosteric modifier site of the exchanger for the intracellular H ϩ (21). However, recently cloned other exchanger isoforms (NHE2, NHE3, and NHE4) (22)(23)(24)(25) differ greatly from NHE1 in their regulation. Growth factors activate the epithelial isoforms NHE2 and NHE3 by increasing V max (26,27). Phorbol ester stimulates NHE1 and NHE2 but inhibits NHE3 (26,27). Hyperosmolarity stimulates NHE1, NHE2, and NHE4 but inhibits NHE3 (28 -30). These differences appear to be attributable to sequence divergence of the cytoplasmic domains of these NHE isoforms. The amiloride-resistant NHE3 that is expressed in the apical membrane of epithelial cells in kidney or intestine is the least related isoform among four mammalian NHEs. The NHE3 cytoplasmic domain shows very low sequence homology to that of NHE1. Particularly, the high affinity CaM-binding site sequence of NHE1 is very different from the corresponding sequence in NHE3. These findings prompted us to study differences in the Ca 2ϩ regulation between NHE1 and NHE3.
In this work, we studied the mechanism of Ca 2ϩ regulation in the early phase of agonist stimulation in fibroblastic cells expressing NHE1, NHE3, or their chimeras. Here we show that Ca 2ϩ -insensitive NHE3 becomes activatable in response to an ionomycin-or thrombin-induced increase in [Ca 2ϩ ] i by replacing the complete cytoplasmic domain of NHE3 with that of NHE1. Deletion mutant analysis revealed that this new function conferred by grafting the cytoplasmic domain of NHE1 was due to the transfer of its autoinhibitory CaM-binding site.

EXPERIMENTAL PROCEDURES
Materials-The amiloride derivative 5-(N-ethyl-N-isopropyl)amiloride (EIPA) was a gift from New Drug Research Laboratories of Kanebo, LTD (Japan). 22 NaCl and [7-14 C]benzoic acid were purchased from DuPont NEN. The NHE1-specific polyclonal antibody (RP-cd) was described previously (17). Mouse monoclonal P5D4 antibody directed against the VSVG tag (31) was a gift from Bruno Gould (Institut Pasteur, Paris). All other chemicals were of the highest purity available.
Cell Culture and Stable Expression of the Wild-type and Mutant Exchangers-The Na ϩ /H ϩ exchanger-deficient cell line PS120 (32) and the corresponding transfectants were maintained in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) containing 25 mM NaHCO 3 and supplemented with 7.5% (v/v) fetal calf serum, penicillin (50 units/ml), and streptomycin (50 g/ml). Cells were maintained at 37°C in presence of 5% CO 2 . Transfection of PS120 cells with plasmids and selection of cells expressing the exchangers were performed as described previously (15).

Construction of NHE Chimeras and Mutant
Molecules-The plasmid including cDNA coding for the NHE1 human isoform deleted of the 5Ј-untranslated region (pEAP-⌬5Ј) (15) and the plasmid (⌬637-656) deleted of the high affinity CaM-binding site sequence (17) were described previously. The cDNA coding for the rat NHE3 isoform was subcloned into HindIII/XbaI sites of the same pECE expression vector (plasmid designated pENHE3). The plasmid (pENHE3-VSVG) including the NHE3 cDNA tagged with a sequence coding the VSVG was also described previously (16). In the latter construct, the last COOH-terminal 47 amino acids of NHE3 were replaced by exogenous 22 amino acids encoding for the spacer and VSVG epitope. We constructed chimeras (pN3N1 and pN1N3-VSVG) by exchanging the cytoplasmic domains between NHE1 and the epitope-tagged NHE3 at an endogenous Eco47III restriction site (1549 base pairs from the first ATG) in human NHE1 cDNA and a created Eco47III site (1404 base pairs from the first ATG) in rat NHE3 cDNA. These restriction sites were located 15 amino acids apart from the last transmembrane spanning segments of both NHE1 and NHE3 (see Fig. 1). A more detailed description on the construction of these chimera will appear elsewhere. 2 We also constructed a chimera (pN1N3) containing the complete NHE3 cytoplasmic domain. For this, a DNA fragment corresponding to the COOH-terminal tail of NHE3 was produced by polymerase chain reaction using two specific primers, one of which contained the exogenous EcoRI site. After restriction enzyme digestion, the DNA fragment was inserted into XhoI/EcoRI sites of pN1N3-VSVG. The plasmid (pN3N1/⌬637-656) deleted of the N3N1 cDNA region corresponding to the high affinity CaM-binding site was constructed as described previously (17). Briefly, polymerase chain reaction was performed using an oligonucleotide sense primer containing the exogenous Bsu36I restriction site and the cDNA region corresponding to amino acids 657-662, and the antisense primer corresponding to the 3Ј-noncoding region. The polymerase chain reaction fragment digested with Bsu36I and KpnI was inserted into the N3N1 plasmid. The DNA inserts were confirmed by sequencing with a Sequenase (T7 DNA polymerase) sequencing kit (U. S. Biochemical Corp.).
Measurement of pH i Dependence of 22 Na ϩ Uptake-pH i dependence of Na ϩ uptake was measured as described previously (7) with slight modification. Stable transfectants grown to confluence in 24-well dishes were incubated for 5 h in a serum-free, bicarbonate-free H21 medium buffered with 20 mM HEPES, pH 7.4, to maintain the Na ϩ /H ϩ exchanger in the resting state. Cells were loaded with NH 4 Cl for 30 min at 37°C in NaCl standard solution (20 mM HEPES/Tris, pH 7.4, 120 mM NaCl, 5 mM KCl, 2 mM CaCl 2 , 1 mM MgCl 2 , and 5 mM glucose) containing 0 -30 mM NH 4 Cl. Cells were then rapidly washed once with prewarmed choline chloride standard solution (20 mM HEPES/Tris, pH 7.4, 120 mM choline chloride, 2 mM CaCl 2 , and 1 mM MgCl 2 ) by simultaneously adding the solution to each well through 24 syringes equipped in a handmade syringe-holder. Cells were incubated in the same medium for 40 s. 22 Na ϩ uptake was then started by simultaneously adding the choline chloride standard solution containing 22 NaCl (37 kBq/ml) (final concentration, 1 mM) and 1 mM ouabain. Forty s later, cells were rapidly washed four times with ice-cold phosphate-buffered saline to terminate 22 Na ϩ uptake. For some wells, the choline chloride solution additionally contained 0.1 mM EIPA (for NHE1, ⌬637-656, and N1N3 transfectants) or 0.5 mM EIPA (for NHE3, N3N1, and N3N1/⌬637-656 transfectants). When the effects of ionomycin, thrombin, PMA, PDGF-BB, and sucrose were measured, these agents were present in the choline chloride solution during 40 S incubation and 22 Na ϩ uptake. Cells were then solubilized with 0.1 N NaOH, and radioactivity was measured on a ␥-counter. pH i was estimated by measuring the distribution of [ 14 C]benzoic acid (74 kBq/ml) (33) under the same conditions as those used for 22 Na ϩ uptake measurement, except that the uptake medium contained [ 14 C]benzoic acid and nonradioactive NaCl.

Measurement of [Ca 2ϩ ] i -[Ca 2ϩ
] i was monitored using fura-2 as a fluorescent Ca 2ϩ indicator. Cells attached to glass coverslips were loaded with 4 M fura-2/acetoxymetylester for 15 min at 37°C in the culture medium. Loaded cells were washed twice with a medium containing 20 mM HEPES/Tris, pH 7.4, 140 mM NaCl, 1 mM NaHPO 4 , 5 mM KCl, 2 mM CaCl 2 , 1 mM MgCl 2 , and 5 mM glucose. The glass coverslip was fixed to a mount which was diagonally inserted into a cuvette filled with 2.2 ml of the above medium. The fluorescence signal was monitored at 510 nm with excitation wavelengths alternating between 345 and 375 nm using a SPEX Fluorolog-2 spectrofluorometer as described previously (34). [Ca 2ϩ ] i was calculated as described previously (35) after correction for fluorescence from extracellular fura-2 and autofluorescence.
Crude Membrane Preparation, Immunoblot Analysis, and Binding of Exchangers to CaM-Sepharose-Preparation of crude membranes from cells expressing various NHE variants and immunoblot analysis were performed as described previously (17). Assay of binding of expressed NHE variants to CaM-Sepharose was carried out essentially as described previously (17). Briefly, 30 g of detergent-solubilized membrane proteins were mixed with CaM-Sepharose beads in presence of 0.1 mM CaCl 2 or 1 mM EGTA at 4°C for 30 min under gentle agitation. After a brief centrifugation, the supernatant was subjected to polyacrylamide gel electrophoresis on 7.5% gels, and proteins were visualized by immunoblot analysis.

Characterization of Expressed NHE Variants-
We checked sensitivity to inhibition by the amiloride analogue EIPA of 22 Na ϩ uptake in cells expressing wild-type and chimeric molecules (Fig. 1C). When the transmembrane domain was derived from NHE1 (NHE1 and N1N3), 22 Na ϩ uptake was inhibited by very low concentrations of EIPA (K d ϭ 0.05 M). On the other hand, when this domain was derived from NHE3 (NHE3 and N3N1), high concentrations of EIPA (K d ϭ 20 M) were required for inhibition of 22 Na ϩ uptake. Thus, the transmembrane domain determines the sensitivity of different isoforms to inhibition by EIPA, consistent with the previous data that the fourth putative transmembrane spanning segment of NHE isoforms is important for inhibition by amiloride and its analogues (12). Fig. 2A shows immunoblots of NHE transfectants probed with a NHE1-specific antibody (RP-cd) and a monoclonal antibody (P4D5) recognizing the VSVG epitope. In this experiment, we used cells stably transfected with pNHE3-VSVG and pN1N3-VSVG in place of pNHE3 and pN1N3 for the availability of the antibody P4D5. Apparent molecular masses (ϳ80 kDa) of NHE3-VSVG and N3N1 were lower than those of NHE1 and N1N3-VSVG. The results were consistent with the finding that NHE3 is not glycosylated (13) and the observations obtained with kidney (36) or intestine (37) brush border membranes using a NHE3-specific antibody. The expression levels of NHE1 and N3N1 were higher than that of N3N1/⌬637-656, whereas the expression level of NHE3-VSVG was higher than that of N1N3-VSVG. These differences in expression agreed well with those in the V max values for 22 Na ϩ uptake (25 and 8 nmol/mg/min for pNHE3-VSVG and pN1N3-VSVG, respectively; see also Figs. 5 and 6). Although expression of N1N3-VSVG was low, two bands were detected ( Fig. 2A). The lower band may represent a non-glycosylated immature form of N1N3-VSVG.
CaM is known to bind to the NHE1 cytoplasmic domain in a Ca 2ϩ -dependent manner (17). Consistent with this, we found that the expressed N3N1 also bound to CaM-Sepharose (data not shown). In addition, deletion of the high affinity CaMbinding site (amino acids 637-656) from the cytoplasmic domain of NHE1 markedly reduced the ability of NHE1 and N3N1 to bind to CaM-Sepharose (data not shown, but see Fig.  6 of Ref. 17). Interestingly, we also found that the expressed NHE3-VSVG bound to CaM-Sepharose in a Ca 2ϩ -dependent manner (Fig. 2B). We thus conclude that like NHE1, NHE3 is also a CaM-binding protein. In this paper, however, we did not analyze the precise localization of the CaM-binding site and its binding affinity in detail.
Effects of Various Stimuli on [Ca 2ϩ ] i and Na ϩ /H ϩ Exchange Activity of NHE1, NHE3, and Their Chimeras-␣-Thrombin, phorbol esters, and hyperosmolarity are all known to induce cytoplasmic alkalinization in cells expressing NHE1 by activating the Na ϩ /H ϩ exchange (10, 11, 14, 17). Ca 2ϩ ionophores are also known to activate NHE1 (3-8), although cytoplasmic alkalinization is often undetectable because of a strong cytoplas-mic acidification induced by increased [Ca 2ϩ ] i . First, we tested whether these agents induce Ca 2ϩ mobilization in cells stably expressing various NHEs. As shown in Fig. 3A (Fig. 3A). These patterns in the [Ca 2ϩ ] i response were similar in cells expressing NHE3 and N3N1 (Fig. 3, B and C) and other NHE variants (data not shown), indicating that expression of NHE variants did not affect the machinery for Ca 2ϩ mobilization.
Next, we examined whether various agents activate the Na ϩ /H ϩ exchange in the early phase (ϳ1 min) of stimulation. In experiments shown in Fig. 4, we acidified cells by a classical NH 4 ϩ prepulse (5 mM NH 4 Cl for 30 min). This NH 4 ϩ concentration was chosen to produce a moderate acidification to "sensitize" the 22 Na ϩ uptake measurement and at the same time to minimize stimuli-induced pH i change that may affect the exchange activity. After the NH 4 ϩ removal, pH i changed to a steady level of 6.96 Ϯ 0.13 (n ϭ 17) in unstimulated cells. Under these conditions, thrombin, PMA, and PDGF-BB did not produce a significant change in this pH i value, whereas ionomycin and sucrose produced slight cytoplasmic acidification in some experiments (up to 0.1 pH unit). The extent of this cytoplasmic acidification was variable in experiments using different batches of cells. As shown in Fig. 4, short stimulation with ionomycin and thrombin significantly stimulated EIPA-sensitive 22 Na ϩ uptake in NHE1 transfectants, whereas PMA did not activate it. Although hyperosmolarity (sucrose) only slightly activated NHE1, the extent of its activation was much less than those for ionomycin and thrombin. PDGF-BB did not activate 22 Na ϩ uptake (data not shown). There is thus a good correlation between Ca 2ϩ mobilization and early activation of NHE1 in the responses to various stimuli.
In sharp contrast to NHE1, the epithelial isoform NHE3 was Na ϩ uptake by cells expressing NHE variants. This experiment were performed by the Li ϩ -loading method as described previously (12).

FIG. 2. Immunoblot analysis of expressed NHE variants (A) and binding of NHE3 to CaM-Sepharose (B).
A, crude membranes from cells expressing NHE variants (NHE1, N3N1, and N3N1/⌬637-656, 10 g each; NHE3 and N1N3, 20 g each) were subjected to immunoblot analysis as described under "Experimental Procedures." A polyclonal antibody RP-cd was used for NHE1, N3N1, and N3N1/⌬637-656, whereas a monoclonal antibody P5D4 was used for NHE3 and N1N3. In this experiment, plasmids pNHE3-VSVG and pN1N3-VSVG were used for availability of the antibody P5D4. B, detergent-solubilized membrane proteins from cells expressing NHE3 were incubated with CaM-Sepharose beads in the presence of 0.1 mM CaCl 2 or 1 mM EGTA as described under "Experimental Procedures." After brief centrifugation, an aliquot (20 l each) of supernatant was subjected to immunoblot analysis. not activated by ionomycin, thrombin, PMA, and sucrose (Fig.  4). PDGF-BB also was not effective (data not shown). A chimera N1N3 also did not respond to these agents. Interestingly, the reciprocal chimera N3N1 was markedly activated by ionomycin and thrombin. However, these activating effects of ionomycin and thrombin were abolished in cells expressing the deletion mutant of N3N1 (N3N1/⌬637-656) lacking the high affinity CaM-binding site (Fig. 4).
We measured pH i dependence of EIPA-sensitive 22 Na ϩ uptake by NHE variants. In cells expressing NHE3 and N1N3, short stimulation with ionomycin and thrombin did not affect the pH i dependence curve (Figs. 5 and 6). We found that thrombin did not induce a detectable increase in the V max value of NHE3 exchange activity (Fig. 6A). This finding is inconsistent with the recent reports (26, 27) that growth factors activate NHE3 by increasing its V max value. This discrepancy is likely to be due to a difference in exposure time to growth factors. Our exposure time (80 s) is much shorter than those used in these previous studies. In contrast to NHE3 and N1N3, in cells expressing N3N1, these agents clearly shifted the pH i dependence curve to an alkaline side by about 0.2 pH unit without a change in the maximal activity (V max ) (Figs. 5 and 6). The extent of this alkaline shift was almost the same as that in NHE1 (see Fig. 2 of Ref. 7). Such a shift in pH i dependence, however, did not occur in a deletion mutant of N3N1 (N3N1/ ⌬637-656) lacking the high affinity CaM-binding site. Thus, the cytoplasmic domain determines the notable difference in the response to ionomycin or thrombin between NHE1 and NHE3. In addition, the high affinity CaM-binding site of NHE1 is required for this response.
It is important to note that unstimulated cells expressing NHE1, NHE3, and their chimeras exhibited markedly different pH i dependences of 22 Na ϩ uptake activity. Fig. 7A shows pH i dependence curves normalized by the V max value of each NHE variant. NHE3 exhibited pH i dependence with a very high pK value (ϳ7.1). In contrast, N3N1 had pH i dependence with a relatively acidic pK value (ϳ6.6). This pK value is slightly lower than that of NHE1 (pK ϭ ϳ6.75) (7, 15). However, N1N3 Na ϩ uptake. Cells were loaded for 30 min with 5 mM NH 4 Cl, washed with nominally Na ϩ -free choline chloride solution, and placed for 40 s in the same medium. 22 Na ϩ uptake was then measured for 40 s in the same medium containing 1 mM 22 NaCl and 1 mM ouabain. Various agents described in the figure were continuously present for total 80 s in the choline chloride solution at the following concentrations: 2 units/ml for thrombin, 5 M for ionomycin, 200 nM for PMA, and 100 mM for sucrose. Percentage activation of the EIPA-sensitive 22 Na ϩ uptake activity was plotted in the ordinate. In the presence of thrombin, PMA, or sucrose, EIPA-insensitive 22 Na ϩ uptake was less than 15% of total 22 Na ϩ uptake in all transfectants, but it was slightly higher in the presence of ionomycin (less than 25% of total 22 Na ϩ uptake). The latter result may indicate the presence of Ca 2ϩ -dependent 22 Na ϩ uptake activity (probably Na ϩ /Ca 2ϩ exchange) in this cell line. Data are means Ϯ S.D. of at least three independent experiments. exhibited pH i dependence with an intermediate pK value (ϳ6.9). Thus, the order of pH i sensitivity was NHE3 Ͼ N1N3 Ͼ NHE1 Ͼ N3N1. Therefore, pH i sensitivity appears to be determined by a complex mechanism involving the interaction between the transmembrane and cytoplasmic domains of NHE isoforms. In addition, we found that deletion of the high affinity CaM-binding site induced a constitutive alkaline shift of pH i dependence in N3N1 (Fig. 7B).

DISCUSSION
In this work, we studied the mechanism of Ca 2ϩ regulation of Na ϩ /H ϩ exchange in fibroblastic cells expressing NHE1, NHE3, and their chimeras. We can summarize our observations as follows. First, ionomycin and thrombin but not PMA, hyperosmolarity, and PDGF-BB induced intracellular Ca 2ϩ mobilization as well as activation of NHE1 in the early phase of stimulation. It thus appears that intracellular Ca 2ϩ is a main regulator in the early phase of NHE1 activation. Second, ionomycin and thrombin activated neither NHE3 nor a chimera N1N3, but they activated another chimera N3N1. Thus, the cytoplasmic domain determines the difference in the Ca 2ϩinduced response between NHE1 and NHE3. Such regulatory role of the cytoplasmic domain is in accordance with a recent report that NHE1 becomes activable in response to cAMP by replacing its cytoplasmic domain with that of trout ␤NHE (38). Third, deletion of the high affinity CaM-binding site from N3N1 induced a constitutive alkaline shift of pH i dependence of 22 Na ϩ uptake and abolished the ionomycin-or thrombininduced activation of 22 Na ϩ uptake by this chimera. Thus, the CaM-binding site in the NHE1 cytoplasmic domain is also able to exert an inhibitory effect on the H ϩ modifier site of NHE3. All these data are consistent with our previous hypothesis (7) that early Ca 2ϩ -induced activation of NHE1 occurs via displacement by Ca 2ϩ /CaM of the autoinhibitory CaM-binding domain from the H ϩ modifier site.
Regulation of NHE1 is more complex in the late phase of stimulation (5-30 min). In the late phase, thrombin, PMA, PDGF-BB, and hyperosmolarity all activate NHE1 and induce cytoplasmic alkalinization. The role of Ca 2ϩ in NHE1 regula-tion in the late phase is not clear at present. After thrombin stimulation, [Ca 2ϩ ] i declined to base line in a relatively short time (see Fig. 3). In addition, there was no significant Ca 2ϩ mobilization induced by PMA, PDGF-BB, or hyperosmolarity under the conditions used in this study (see Fig. 3 and "Results"). On the other hand, previous studies showed that intracellular Ca 2ϩ depletion caused by extracellular or intracellular addition of calcium chelators reduced the extent of cytoplasmic alkalinization in response to long stimulation by growth factors such as thrombin and PDGF-BB (6, 39 -41). Furthermore, we previously showed that deletion or point mutation of CaMbinding domain in NHE1 reduced cytoplasmic alkalinization by ϳ50% in response to long stimulation with thrombin and ϳ80% in response to hyperosmolarity (17). These latter results raise the possibility that slight elevation of [Ca 2ϩ ] i in subplasmalemma space not detectable by ordinary fluorescence [Ca 2ϩ ] i measurements may partially be involved in the NHE1 regulation even in the relatively long time range. In our previous paper (17), we presented evidence suggesting that another mechanism not involving Ca 2ϩ /CaM or direct phosphorylation of NHE1 but involving unknown regulatory factor(s) that also activates NHE1 in the late phase of growth factor stimulation.
As described above, pH i sensitivity of chimera N3N1 is upregulated by Ca 2ϩ -mobilizing agents. Thus, the lack of Ca 2ϩ response in NHE3 or N1N3 is attributable to the structure of the NHE3 cytoplasmic domain. We found that NHE3 binds Ca 2ϩ /CaM, although at present we do not know the precise localization of its binding site (see Fig. 2B). However, the CaMbinding site in NHE3 does not appear to function as a Ca 2ϩ / CaM-sensitive autoinhibitory domain as in NHE1, because NHE3 does not respond to Ca 2ϩ at least in the conventional way it was tested. Comparison of the CaM-binding site sequence of rat NHE1 with the corresponding sequence of rat NHE3 reveals that in rat NHE3, 27 intervening amino acids are inserted between Ser-605 and Tyr-606, whereas 9 amino acids are conserved between NHE1 and NHE3 (see Fig. 1A). It is thus possible that insertion of the intervening sequence in this region of NHE3 resulted in disruption of the original autoinhibitory CaM-binding domain during the evolutional process. Recently, CaM antagonist W13 was reported to stimulate the NHE3 activity (42). At present, it is not clear how CaM-binding site(s) of NHE3 is related with the data obtained with this drug.
The pH i sensitivity of chimera N3N1 as well as NHE1 is constitutively alkaline-shifted by deletion of the CaM-binding site. This finding suggests that the autoinhibitory CaM-binding domain of NHE1 functionally interacts with an "acceptor" site(s) somewhere in the NHE sequences. This putative acceptor site(s), which could also be involved in the H ϩ binding to the exchanger, is most probably located in the NH 2 -terminal highly conserved regions of NHE1 and NHE3. Alternatively, it may be located in the homologous NH terminus of the cytoplasmic domain that was important for the maintenance of pH i sensitivity in NHE1 (14). 3 According to our recent study, the cytoplasmic tail (amino acids 659 -815) of NHE1 is not involved in the Ca 2ϩ regulation of NHE1 (7). 3 Identification of this acceptor site(s) would lead to a better understanding of the molecular entity of the H ϩ -modifier site.
In unstimulated cells expressing NHE1, NHE3, and their chimeras, we observed marked differences in the pH i dependence of 22 Na ϩ uptake activity (Fig. 7A). The pH i sensitivity of NHE3 is much higher than that of NHE1. This was confirmed by our finding that cells expressing NHE3 possess a much 3 T. Ikeda, unpublished observation. higher resting pH i as compared to cells expressing NHE1, 4 which is consistent with previous results (27). This is also consistent with the recent observation (26) obtained from analysis of pH i recovery rate, but not consistent with other reports (28,43). AP-1 cells were used in the latter studies, while PS120 cells were used in this study. Thus, there may be cell specific effects that influence the pH i sensitivity. It is important to note that pH i dependence was dramatically shifted to an acidic side, when the cytoplasmic domain of NHE3 was replaced by that of NHE1 to form N3N1 (Fig. 7A). Conversely, pH i dependence was significantly shifted to an alkaline side, when the cytoplasmic domain of NHE1 was replaced by that of NHE3 to form N1N3. Direction of the pH shift caused by COOH termini swaps can be explained by incorporation or removal of the autoinhibitory domain of NHE1. However, a large acidic pH shift observed in N3N1 is not explained only by incorporation of the inhibitory domain of NHE1. We found that in cells expressing NHE3-VSVG lacking the last COOH-terminal 47 amino acids of NHE3, pH i dependence was shifted to a much more acidic side (Ͼ0.4 pH unit) as compared to that in intact NHE3. 4 Therefore, the short COOH-terminal tail may function as an activator domain in NHE3. Thus, pH i sensitivity appears to be determined by a global mechanism involving interaction between the transmembrane domain and multiple cytoplasmic subdomains. More precise identification of the critical region of NHE3 would facilitate understanding of how pH i dependence of Na ϩ /H ϩ exchange is determined by the interaction between the transmembrane and cytoplasmic domains of NHE isoforms.
In summary, the present data indicate that the presence or absence of the autoinhibitory CaM binding region produces the remarkable difference in the Ca 2ϩ response between NHE1 and NHE3. Activation of CaM-dependent protein kinase II has been reported to inhibit Na ϩ /H ϩ exchange in kidney and intestine brush border membranes which express NHE3 (44 -46). Inhibition of the apical exchanger by CaM-dependent protein kinase II was also reported in proximal tubule cell line LLC-PK1 (47). Lack of early Ca 2ϩ activation in NHE3 would facilitate this CaM-dependent protein kinase II-mediated inhibition of the apical exchanger which lead to inhibition of Na ϩ reabsorption. On the other hand, activation of NHE1 by Ca 2ϩ /CaM binding would play an important role in a rapid and fine pH i regulation in various cells when [Ca 2ϩ ] i increases. Thus, the difference of Ca 2ϩ response appears to reflect distinct physiological requirements of these two exchangers.