Identification of sites in the second exomembrane loop and ninth transmembrane helix of the mammalian Na+/H+ exchanger important for drug recognition and cation translocation.

Mammalian Na(+)/H(+) exchanger (NHE) isoforms are differentially sensitive to inhibition by several distinct classes of pharmacological agents, including amiloride- and benzoyl guanidinium-based derivatives. The determinants of drug sensitivity, however, are only partially understood. Earlier studies of the drug-sensitive NHE1 isoform have shown that residues within the fourth membrane-spanning helix (M4) (Phe(165), Phe(166), Leu(167), and Gly(178)) and a 66-amino acid segment encompassing M9 contribute significantly to drug recognition. In this report, we have identified two residues within M9, one highly conserved (Glu(350)) and the other non-conserved (Gly(356)), that are major determinants of drug sensitivity. In addition, residues in the second exomembrane loop between M3 and M4 (Gly(152), Phe(157), and Pro(158)) were also found to modestly influence drug sensitivity. A double substitution of crucial sites within M4 and M9 of NHE1 with the corresponding residues present in the drug-resistant NHE3 isoform (i.e. L167F/G356A) greatly reduced drug sensitivity in a cooperative manner to levels nearing that of wild type NHE3. The above mutations did not appreciably affect Na(o)(+) affinity but did markedly decrease the catalytic turnover of the transporter. These data suggest that specific sites encompassing M4 and M9 are critical determinants of both drug recognition and cation translocation.

Na ϩ /H ϩ exchangers (NHE) 1 are present at the cell surface and various organellar compartments of mammalian cells and mediate the electroneutral exchange of Na ϩ for H ϩ , a process driven by the relative concentration gradients of the respective cations. To date seven distinct isoforms (NHE1 to NHE7) have been isolated that share ϳ20 -70% amino acid identity (calculated M r ranging from ϳ74,000 to 93,000) and exhibit similar membrane topologies, with 12 predicted N-terminal membrane-spanning (M) ␣-helices and a large C-terminal cytoplasmic region (1)(2)(3)(4)(5). They show considerable differences in their patterns of tissue expression, membrane localization, kinetic properties, sensitivity to pharmacological antagonists, and responsiveness to various signaling pathways. Consistent with their molecular diversity, the exchangers participate in a broad spectrum of physiological processes, including the regulation of intracellular pH (pH i ), maintenance of cell volume, and transepithelial transport of electrolytes. In addition, activation of certain exchangers appear to facilitate cellular growth and proliferation in response to numerous growth factors and other mitogens and are associated with events leading to apoptosis (6 -8).
The NHE is a known target for inhibition by the diuretic compound amiloride and its analogues (9). Amiloride analogues containing hydrophobic substituents on the 5-amino group of the pyrazine ring, such as 5-(N-ethyl-N-isopropyl) amiloride (EIPA), have higher affinity and specificity for NHE relative to other ion transporters. Comparison of the NHE isoforms in heterologous expression systems show that they have varying affinities for amiloride and its analogues that span more than 2 orders of magnitude, with the following order of sensitivity: NHE1 Ն NHE2 Ͼ NHE5 Ͼ NHE3 (10 -12). NHE4 also has an apparent low affinity for many of these antagonists, but its activity in transfected fibroblasts can only be detected under specialized experimental conditions (13,14) that preclude direct comparisons with other ectopically expressed isoforms. Recently, novel benzoyl guanidinium compounds (e.g. HOE694, HOE642 or cariporide, and EMD85131) have been developed that inhibit the NHE isoforms with a similar rank order but over a larger concentration range (3-4 orders of magnitude) (15)(16)(17)(18). The more selective binding properties of these compounds for NHE1 have been exploited therapeutically as effective agents in the treatment of cardiac ischemia and reperfusion injuries (17)(18)(19)(20)(21)(22) and may prove beneficial in the prevention of diabetes-induced vascular hypertrophy (23). More recently, a preferential antagonist (S3226) of NHE3 has been synthesized that may also facilitate functional studies of this isoform in renal and intestinal epithelia (24,25).
Biochemical analyses indicate that inhibition by amiloride compounds (26) and HOE694 (15) is reduced by high external Na ϩ (Na o ϩ ). This competitive inhibition suggests they bind near the Na o ϩ transport site and may also share a common site. However, under chloride-free buffer conditions, amiloride and its derivatives also inhibit transport noncompetitively, suggesting that the Na o ϩ and amiloride binding sites may not be identical (27,28). Furthermore, the Na o ϩ -and amiloride-binding sites can be altered independently of each other using genetic selection techniques (29). Taken together, these data indicate that amiloride and other antagonists probably interact with multiple sites on the exchanger. Consistent with the above notion, mutational analyses have identified residues in the predicted fourth membrane-spanning helix (M4) of NHE1 (Phe 161 , Leu 163 , and Gly 174 of human NHE1) that confer sensitivity to amiloride and its analogues but do not seemingly affect Na o ϩ affinity (30,31). By contrast, mutation of a neighboring residue, Phe 162 , was found to decrease affinities for both Na o ϩ and HOE642 (32). In addition, pharmacological analyses of chimeras of rat NHE1 and NHE3 defined a 66-amino acid segment encompassing M9 and its adjacent loops (residues 327-392 of rat NHE1) as a major determinant of the differential drug sensitivity between these two isoforms (16). Homologous substitution of this region between NHE1 and NHE3 caused a reciprocal change in their drug sensitivities by 1-3 orders of magnitude, with the greatest effects observed for the more NHE-selective drugs, EIPA and HOE694. A role for this region in drug recognition is further supported by observations of Wang et al. (33), who serendipitously found that mutation of His 349 in the putative M9 domain of human NHE1 produced either a modest 2-fold increase (H349Y or H349F) or 2-fold decrease (H349G or H349L) in amiloride sensitivity, although other amino acid substitutions had no effect. Although this site is unlikely to account for the large changes observed for the NHE1/3 chimeras, it does implicate the involvement of other nearby sites as critical determinants of drug sensitivity.
The objective of the present study was to define specific residues that may contribute to the large difference in drug sensitivity between NHE1 and NHE3. The results show that sites within transmembrane ␣-helix M9 as well as between helices M3 and M4 are significant determinants of drug recognition. These sites were also found to be important for optimal cation translocation. These data provide additional insight in NHE drug interactions and may aid in the design of more potent isoform-specific drugs with therapeutic potential.

EXPERIMENTAL PROCEDURES
Materials-Carrier-free 22 NaCl (radioactivity, 5 mCi/ml) was obtained from PerkinElmer Life Sciences. Amiloride and ouabain were purchased from Sigma, and the amiloride derivative EIPA was obtained from Molecular Probes (Eugene, OR). HOE694 ((3-methylsulphonyl-4piperidinobenzoyl) guanidine methanesulfonate) was generously provided by Dr. Hans J. Lang (Hoechst AG). ␣-Minimal essential medium, fetal bovine serum, kanamycin sulfate, and trypsin-EDTA were purchased from Life Technologies. Polyclonal antibodies to NHE1 were generated by injecting rabbits with a fusion protein containing Escherichia coli ␤-galactosidase linked to the C-terminal 157 amino acids (658 -815) of the human exchanger and subsequently affinity-purified as previously described (34). The horseradish peroxidase-conjugated goat anti-rabbit secondary antibody was purchased from New England BioLabs (Beverly, MA). Cell culture dishes and flasks were purchased from Becton Dickinson and/or Corning (Montreal, Quebec, Canada). All other chemicals and reagents used in these experiments were purchased from British Drug House Inc. (St. Laurent, Quebec) or Fisher Scientific and were of the highest grade available.
Construction of Na ϩ /H ϩ Exchanger Mutants-The rat NHE1 and NHE3 cDNAs, engineered to contain a series of unique restriction endonuclease sites to create convenient DNA cassettes for mutagenesis, were inserted into a mammalian expression vector under the control of the enhancer/promoter region from the immediate early gene of human cytomegalovirus (modified cDNAs called NHE1Ј and NHE3Ј, respectively) as previously described (16). These nucleotide changes either preserved the native amino acid sequence or, in a few cases, resulted in substitution of an amino acid present in one isoform with the equivalent residue of the other isoform and had no obvious effect on their respective functional properties. Substitution of various amino acids was accomplished by site-directed mutagenesis using the unique site elimination method (35). The cDNAs were sequenced to confirm the presence of the mutations and to ensure that other random mutations were not introduced.
Stable Transfection and Expression of the Na ϩ /H ϩ Exchanger cDNAs-Chemically mutagenized Chinese hamster ovary (AP-1) cells devoid of endogenous NHE activity (36) were transfected with plasmids containing the various NHE constructs by the calcium phosphate-DNA coprecipitation technique of Chen and Okayama (37). Starting 48 h after transfection, the AP-1 cells were selected for survival in response to repeated (5-6 times over a 2-week period) acute NH 4 Cl-induced acid loads (i.e. H ϩ -killing technique) (10,38) to discriminate between Na ϩ / H ϩ -exchanger positive and negative transfectants. The positive clones for each transfectant were pooled and used for subsequent analyses. 22 Na ϩ Influx Measurements-The cells were grown to confluence in 24-well plates. NHE activity was determined by preloading the cells with H ϩ using the NH 4 Cl technique (39), then measuring the initial rates of 22 Na ϩ influx essentially as described (10). Briefly, the cell culture medium was aspirated and replaced by isotonic NH 4 Cl medium (50 mM NH 4 Cl, 70 mM choline chloride, 5 mM KCl, 1 mM MgCl 2 , 2 mM CaCl 2 , 5 mM glucose, 20 mM HEPES-Tris, pH 7.4). Cells were incubated in this medium for 30 min at 37°C in a nominally CO 2 -free atmosphere. After acid loading, the monolayers were rapidly washed twice with isotonic choline chloride solution (125 mM choline chloride, 1 mM MgCl 2 , 2 mM CaCl 2 , 5 mM glucose, 20 mM HEPES-Tris, pH 7.4). 22 Na ϩ influx assays were initiated by incubating the cells in isotonic choline chloride solution containing 1 mM ouabain and 1 Ci of 22 NaCl (carrier-free)/ml (ϳ120 nM NaCl). The assay medium was K ϩ -free and included ouabain to prevent the transport of 22 Na ϩ by the Na ϩ -K ϩ -Cl Ϫ cotransporter and the Na ϩ ,K ϩ -ATPase. Influx of 22 Na ϩ was terminated by rapidly washing the cell monolayers three times with four volumes of ice-cold isotonic saline solution (130 mM NaCl, 1 mM MgCl 2 , 2 mM CaCl 2 , 20 mM HEPES-NaOH, pH 7.4). The washed cell monolayers were solubilized in 0.25 ml of 0.5 N NaOH, and the wells were washed with 0.25 ml of 0.5 N HCl. Both the solubilized cell extract and wash solutions were added to vials, and radioactivity was assayed using a liquid scintillation counter. Under the conditions of H ϩ loading used in this study, uptake of 22 Na ϩ was linear with time for 8 -10 min (at low Na ϩ concentrations, 22°C). Therefore, 22 Na ϩ uptakes were measured after 5 min except when examining the kinetics of NHE activity as a function of the extracellular Na ϩ concentration. We found that 22 Na ϩ influx was linear with time for only 4 min when Na o ϩ was increased to 40 mM. Hence, an uptake time of 1 min was chosen for this set of experiments, i.e. when the extracellular Na ϩ concentration ranged from 1.25 to 40 mM. Measurements of 22 Na ϩ influx specific to the Na ϩ /H ϩ exchanger were determined as the difference between the initial rates of H ϩ -activated 22 Na ϩ influx in the absence and presence of 2 mM amiloride or 100 M EIPA (concentrations sufficient to inhibit NHE1 or NHE3 under these experimental conditions). Protein content was determined using the Bio-Rad DC protein assay procedure. To examine NHE activity as a function of the intracellular H ϩ concentration, the pH i was clamped at different concentrations over the range of 5.4 -7.4 by suspending the cells in media of varying K ϩ concentrations containing the K ϩ /H ϩ exchange ionophore nigericin as previously described (40).
Immunoblotting-Stably transfected cells were grown to confluence in 10-cm dishes and lysed with 1% Triton X-100. Total cellular protein extracts (30 g) were resolved by 6% SDS-polyacrylamide gel electrophoresis and transferred onto polyvinylidene difluoride membranes (Amersham Pharmacia Biotech). The blots were briefly rinsed with phosphate-buffered saline (PBS), blocked with 5% nonfat skim milk in PBST (phosphate-buffered saline with 0.1% Tween 20), and then incubated with a rabbit polyclonal anti-NHE1 antibody (dilution 1:5,000). After extensive washes with PBST, the blots were incubated with goat anti-rabbit IgG secondary antibody conjugated with horseradish peroxidase (dilution 1:2,000). Immunoreactive bands were visualized by enhanced chemiluminescence (Amersham Pharmacia Biotech) recorded on x-ray film. For quantitation of protein levels, the amount of protein was titrated in preliminary experiments, and the film exposure times were selected to ensure that the chemiluminescent signals were within the linear range of the x-ray film.

Determinants of Drug Sensitivity in M9 -Earlier studies
have identified sites within M4 (Phe 165 , Phe 166 , Leu 167 , and Gly 178 , numbered according to the rat NHE1 sequence) and a 66-amino acid segment encompassing M9 (residues 327-392) as important elements of NHE drug recognition. The location of these sites are depicted in a secondary structural model of rat NHE1 (Fig. 1), as defined by recent topological mapping studies (41). The region surrounding M9 is of particular interest because it appears to account for large differences (2-3 orders of magnitude) in the differential drug sensitivity of NHE1 (high affinity) and NHE3 (low affinity) (16) and is proposed to form part of the pore-lining region with M4 (41). These two isoforms share ϳ50% amino acid identity in this region, but the precise amino acids that confer this differential drug sensitivity are unknown. To identity these sites, 14 divergent amino acids distributed throughout this region of NHE1 were individually substituted with the corresponding residues present in NHE3 by site-directed mutagenesis ( Fig. 2A; mutated sites are boxed in gray and black shading). The modified constructs were then transfected in Chinese hamster ovary AP-1 cells that lack endogenous plasmalemmal NHE activity and stably selected for functional NHE1 activity. All the mutants yielded stable transfectants with varying levels of plasmalemmal NHE1 activity, as assessed by measuring rates of H ϩ -activated 22 Na ϩ influx (summarized in Table I).
To characterize their drug sensitivity, concentration-response measurements were performed with amiloride and the more selective NHE antagonists, EIPA and/or HOE694. As an initial screen, mutations that did not alter the apparent halfmaximal inhibition (K 0.5 ) for EIPA by more than 2-fold were considered wild type in behavior and not examined further. Of these mutations, only the G356A substitution significantly altered the drug sensitivity of NHE1Ј, effectively reducing its apparent half-maximal inhibition by amiloride, EIPA, and HOE694 by ϳ5-, 32-, and 342-fold, respectively ( Fig. 2, B-D and Table II). To further examine the molecular nature of this interaction, other non-polar (Leu) and polar (Ser, Asp, and Lys) amino acids were substituted at this site. Stable transfectants were generated for all four mutants, but only the G356S and G356D mutants yielded transporters with sufficient activity to reliably measure their drug inhibition profiles (Table I). Like the G356A substitution, both G356S and G356D significantly decreased EIPA sensitivity by comparable extents (72-and 33-fold, respectively) ( Fig. 3 and Table II). These data suggest that increases in side-chain length rather than polarity are detrimental to drug recognition. The length of the side-chain group, however, did not closely correlate with the extent of inhibition. It was also noted that the G356S mutant displayed a modest but reproducible biphasic response to EIPA, with transport activity stimulated (10 -15%) in the presence of low drug concentrations. To further establish the importance of this site, the reciprocal substitution was made in NHE3Ј. As shown in Fig. 4, replacement of alanine with glycine at the analogous position of NHE3Ј (i.e. A305G) caused a modest 4-fold gain in its sensitivity to inhibition by EIPA. Taken together, these data indicate that this site in M9 is an important determinant of drug recognition of the NHEs, with the greatest effects observed for the more NHE-selective drugs.
Although the above approach revealed a single amino acid that partly accounts for the differential drug sensitivity between NHE1 and NHE3, it is likely that amino acids common to both isoforms in this same region are also involved in drug recognition since even NHE3 can be inhibited by high millimolar drug concentrations. To define likely candidate sites, we compared the sequences of the plasmalemmal NHE isoforms (NHE1 to NHE5) and identified 25 highly conserved residues. Of these, 12 residues within M9 and the predicted extracellular loop of NHE1Ј were selected for mutagenesis ( Fig. 5A; mutated sites are boxed in gray and black shading; S342A, Y343N/D, Y346N/D, E350Q/D/N, S355A, Y370A, N374A, S376A, S379A, T381A, T382A, Y385A). All the mutants were capable of rescuing AP-1 cells from repeated lethal acid loads and, correspondingly, displayed measurable plasmalemmal NHE activity (Table I). However, only mutations at Glu 350 caused a significant reduction in drug sensitivity. Isosteric replacement of Glu 350 with Gln, which should minimally perturb the protein structure, dramatically reduced sensitivity to amiloride and EIPA by 20-and 127-fold, respectively (Fig. 5, B and C; values for half-maximal inhibition are summarized in Table II). Other conservative substitutions (E350D and E350N) that reduced the side-chain length also produced substantial losses in EIPA recognition (Fig. 5D). Like G356S, these mutants displayed The shaded boxes indicate sites in NHE1Ј that were mutated to the equivalent residues in NHE3Ј. The gray and black shading indicates those sites that, when mutated, had either no effect or had significantly altered drug sensitivity, respectively. B-D, AP-1 cells stably expressing either wild type NHE1Ј, wild type NHE3Ј, or mutant NHE1Ј/G356A were grown to confluence in 24-well plates. Before 22 Na ϩ influx measurements, the cells were loaded with H ϩ using the NH 4 Cl prepulse technique. Cells were washed with Na ϩ -free isotonic choline chloride solution and then incubated in assay medium containing carrier-free 22 NaCl (1 Ci/ml) and increasing concentrations of amiloride (B), EIPA (C), or HOE694 (D) (for details, see "Experimental Procedures"). Data were normalized as a percentage of the maximal rate of H i ϩ -activated 22 Na ϩ influx in the absence of inhibitor. Values represent the average of two or three experiments, each performed in quadruplicate. modest increases (10 -20%) in their rates of transport in the presence of low concentrations of EIPA. It was also noted that the activities of the E350Q/D/N mutants could not be completely blocked by amiloride or EIPA, achieving only 80 -85% inhibition at the highest concentrations tested, whereas wild type NHE1Ј and NHE3Ј were rendered inactive.
Critical Sites in M3-M4 Region-As mentioned earlier, mutations of residues within M4 of NHE1 (i.e. Phe 165 , Phe 166 , Leu 167 , and Gly 178 ) have been shown to influence drug sensitivity. Of these, only the residue at position 167 is different between NHE1 and NHE3. To determine whether additional residues in this region may contribute to isoform-specific drug recognition, we compared the amino acid sequences in the predicted extracellular loop between M3 and M4 of selected NHE isoforms that display broad differences in drug sensitivity. As shown in Fig. 6A, three residues in this region (Gly 152 , Pro 157 , Pro 158 , numbered according to the rat NHE1 sequence) are common among the more drug-sensitive isoforms (e.g. NHE1 and NHE2) but differ in the more drug-resistant isoforms (e.g. NHE3 and NHE5). Because glycine and proline residues are known to significantly influence the secondary conformations of proteins (i.e. disrupt formation of ␣-helices and, in the case of proline, also ␤-structures), they may contribute to the differential drug sensitivity of the NHE isoforms. To test this possibility, these sites were mutated in NHE1Ј to the corresponding residues present in NHE3Ј (i.e. single G152A and double P157S/P158F substitutions). For comparison, a previously characterized mutation within M4 (L167F) that is known to reduce drug sensitivity was also constructed (30). As shown in Fig. 6B and summarized in Table II, both the single G152A and dual P157S/P158F substitutions modestly reduced sensitivity to EIPA by 3-and 7-fold, respectively. Consistent with previous studies, the L167F mutation significantly reduced (ϳ30-fold) the effectiveness of EIPA to block transport. These data implicate the M3-M4 exomembranous loop as an additional component along with M4 in conferring drug recognition.
The above mutational analyses suggest that M4 and M9 may be in close spatial proximity to each other to form part of the drug binding pocket. If so, then one would predict that simultaneous mutation of key sites in both helices should produce additive effects. We tested this supposition by constructing and examining a NHE1Ј/L167F/G356A mutant. As shown in Fig. 7, this double mutant showed a 164-fold lower affinity for EIPA compared with wild type and closely mimicked the affinity of the drug for NHE3Ј. However, rather than being additive, the dual substitution caused a synergistic reduction in drug sensitivity. These data strongly support the notion that these two sites are critical determinants of the differential drug sensitivity between NHE1 and NHE3.
Kinetic Properties of Drug-resistant Mutant Na ϩ /H ϩ Exchangers-To evaluate whether the above mutations had additional functional consequences other than affecting drug sensitivity, we performed a detailed comparison of their intrinsic kinetic properties (cation affinities and catalytic turnover). Earlier kinetic measurements indicate that NHE antagonists can function as either simple competitors (26) or, under Cl Ϫfree conditions, mixed competitors (27,28) of external Na ϩ binding. Consistent with these analyses, mutation of certain sites that decreased drug sensitivity (Leu 167 , Gly 178 , and His 353 ) were found to have no demonstrable affect on Na o ϩ affinity (30,31,33), whereas mutation of a neighboring residue, F166S, influenced both Na ϩ and drug affinities (32). Hence, it was of interest to examine whether the single (E350Q; G356A) or double (P157S/P158F; L167F/G356A) substitutions generated in this study also influenced Na o ϩ affinity. For comparative purposes, the previously characterized L167F mutation was included in the analysis. NHE activity, defined as the initial rates of EIPA-inhibitable H i ϩ -activated 22 Na ϩ influx, was measured as a function of the Na o ϩ concentration in the wild type and mutant exchangers. As illustrated in Fig. 8, the rates of 22 Na ϩ influx gradually approached saturation, with increasing Na o ϩ concentrations for wild type and mutant exchangers, consistent with simple Michaelis-Menten kinetics. Analysis of the data using a hyperbolic fit function yielded apparent Na o ϩ affinity constants (K Na ) for the mutants that were not appreciably different from wild type (Table III) and corroborated earlier findings for L167F. Thus, the new drug-sensitive sites identified herein do not contribute to Na ϩ binding but, rather, are more likely involved in non-competitive drug interactions.
The transport activities of the M9 drug-resistant mutants (i.e. E350Q and G356A) were also measured as a function of pH i to assess whether H i ϩ affinity was affected. The residues in the exomembranous loop between transmembrane helices M3 and M4 were not examined because they are less likely to directly affect internal H ϩ sensitivity. As shown in Fig. 9, both mutations had no appreciable effect on the H i ϩ dependence of the exchanger over the range of pH i 5.4 -7.4. Similarly, other substitutions at these sites (i.e. E350D, E350N, G356S, and G356D) did not alter H ϩ affinity (data not shown). Last, we assessed whether the catalytic turnover of the mutated transporters was affected. To measure this parameter, it was necessary to estimate the quantity of mature, fully glycosylated NHE1 transporter at the cell surface (separate from the immature, core-glycosylated transporter that resides within intracellular compartments) of the different stable transfectants and express these values in relation to the cellular rates of H i ϩ -activated 22 Na ϩ influx under near maximal acid-load conditions. This was accomplished immunologically by Western blot analysis using an affinity-purified rabbit anti-NHE1 antibody directed to the cytosolic domain of the exchanger followed by densitometry of the chemiluminescent signals recorded on x-ray film. To obtain a reasonably accurate measure of cellular NHE1 levels, the amount of total cellular protein analyzed by Western blotting was titrated in preliminary experiments, and film exposure times were varied to ensure that the chemiluminescent signals were within the linear range of the x-ray film. Representative immunoblots are shown in Fig.  10A. The anti-NHE1 antibody recognized two major bands, a slower migrating, fully glycosylated form with an apparent molecular mass of ϳ100 kDa that has been demonstrated by biochemical and immunological means to reside at the plasmalemma (42,43) and an immature core-glycosylated form of the protein of ϳ75 kDa that resides intracellularly, presumably within the endoplasmic reticulum. A faint, nonspecific band of ϳ62 kDa was consistently observed in all cell extracts and served as a convenient indicator of equal protein loading on the gels. Among the various mutants, it was noted that the upper protein band for NHE1Ј/G356A migrated more diffusely, suggestive of incomplete glycosylation, albeit minor. A more drastic effect was observed for the NHE1Ј/G356L mutant that showed a marked reduction in the level of the fully glycosylated, plasmalemmal form and a corresponding increase in the core glycosylated form. It is likely that this substitution destabilized the protein, resulting in incomplete processing and retention in endomembrane compartments. Densitometric analysis of the fully glycosylated band and normalization of the data to that of the wild type exchanger is shown in Fig. 10B. In general, the majority of the mutants proteins were expressed at equivalent or higher levels relative to the wild type transporter in stably transfected cells but generally displayed equivalent or lower rates of H i ϩ -activated 22 Na ϩ influx on a per-cell basis (Fig. 10C). To estimate the relative activity or turnover of the transporters, the cellular rates of 22 Na ϩ influx were expressed as a function of their respective plasmalemmal protein levels (Fig. 10D). Of the various NHE1Ј constructs, only the L167F mutant had a level of activity that was comparable with wild type. All the others showed a marked reduction in activity, suggesting that although these sites do not affect Na o ϩ or H i ϩ affinity, they do influence the velocity of the transporter and, hence, may be important in conformational changes underlying cation translocation.   4. Effect of mutation of A 305 on the drug sensitivity of NHE3. AP-1 cells separately expressing wild type (NHE1Ј, NHE3Ј), and mutant (NHE3Ј/A305G) exchangers were grown to confluence in 24-well plates, and their inhibition was assessed by varying the concentrations of EIPA . Transport activity was measured as described in the legend of Fig. 2. Values represent the average of three experiments, each performed in quadruplicate.

DISCUSSION
The aim of this study was to further define the structural elements involved in drug recognition by mammalian Na ϩ /H ϩ exchangers. Prior analyses of chimeras of the drug-sensitive NHE1 and drug-resistant NHE3 isoforms revealed a 66-amino acid sequence containing transmembrane helix M9 as a major segment responsible for drug recognition (16). The present study further demarcates this region by showing that Gly 356 of NHE1, which is thought to reside within M9, is a crucial determinant of NHE isoform-specific drug sensitivity. Mutation of this amino acid to the corresponding residue present in NHE3 (i.e. Ala 305 ) substantially reduced drug recognition. Significantly, the effects were greatest for the more NHE-selective drugs, EIPA and HOE694, rather than amiloride, which is consistent with previous analyses of the NHE1/3 chimeras (16). The more substantial changes for EIPA compared with amiloride suggest that Gly 356 either directly or indirectly influences the interaction of the 5-amino-substituted moiety of EIPA with this region. This may equally explain the enhanced sensitivity to HOE694, which is structurally similar to EIPA and also has a large substituent group at an equivalent position of its benzoyl ring. Conversely, the reciprocal mutation in NHE3 (A305G) caused a modest increase in drug sensitivity. Although the magnitude of this increase for NHE3/A305G is less than the reduction of drug sensitivity for NHE1/G356A, the result is nevertheless significant since a gain-of-function mutation is more indicative of a site that directly contributes to drug recognition than a loss-of-function mutation, which could cause nonspecific alterations in protein structure. The relevance of this site in conferring isoform-specific drug sensitivity is further indicated by the absence of an effect of mutations at neighboring amino acids that differ between NHE1 and NHE3. It is also noteworthy that NHE5, which shows drug affinities that are intermediate to NHE1 and NHE3 (12), also has an alanine residue at the equivalent position (i.e. Ala 301 ). Taken together, these data support the critical importance of this site in drug recognition and possibly binding. In addition to Gly 356 , residues that are divergent between NHE1 and NHE3 in the exomembrane loop between M3 and M4 were also found to influence drug sensitivity. Substitutions of Gly 152 , Pro 157 , and Pro 158 in NHE1 with the corresponding residues present in NHE3, Ala, Ser, and Phe, respectively, moderately reduced drug sensitivity. We also corroborated earlier findings that replacement of Leu 167 with Phe (present at the equivalent position of NHE3) in M4 also significantly reduced drug sensitivity (30). Indeed, this site seems to be generally important for drug recognition by the NHEs, since mutagenesis of the equivalent residue in rabbit NHE2 (L143F) also reduced its sensitivity to amiloride compounds (44). To further confirm the involvement of transmembrane helices M4 and M9 in drug recognition, we combined the L167F and G356A mutations. Significantly, this dual substitution caused a synergistic reduction in drug sensitivity of NHE1 to levels closely approaching those observed for wild type NHE3. Thus, these two sites are sufficient to account for much of the difference in drug sensitivity between NHE1 and NHE3 and possibly other isoforms as well. Taken together, these data now extend the elements involved in isoform-specific drug recognition to include not only those sites within M4 but also those that reside on the exofacial surface between M3-M4 and within M9.
Aside from these sites, we also identified a highly conserved residue (Glu 350 ) in M9 that when mutated caused a profound reduction in drug sensitivity. This effect was particularly striking given that isosteric or conservative replacements (Gln, Asp, and Asn) were used that would be expected to minimally per-turb protein structure. Substitutions of other conserved residues flanking this site had no effect on drug sensitivity, confirming an important role for this particular site in drug recognition and one that may be shared among all mammalian NHEs, including plasmalemmal (NHE1-5) as well as the more distantly related organellar (NHE6 -7) isoforms.
To determine whether these sites were involved in other aspects of exchanger function, we performed a detailed kinetic analysis of the various mutants. Previous studies show that the amiloride-based compounds display either simple-competitive (26,45,46) or mixed-competitive (27,28) inhibition at the external Na ϩ transport site, suggesting that the external Na ϩ and amiloride binding sites may not be identical. Furthermore, certain guanidinium derivatives that block Na ϩ /H ϩ exchange by competing with Na o ϩ were not effective competitors of [ 3 H]ethylpropyl amiloride binding to the exchanger (47), again implicating the involvement of at least two discrete sites in drug binding. In this regard, kinetic analyses of point mutations of NHE1 at positions Leu 167 (30) and Gly 178 (31) showed no change in Na ϩ affinity, although a combined mutation did produce a modest 2-fold reduction in Na ϩ affinity (K m 14 to 28 mM) (31). More significantly, mutation of a neighboring residue, F166S (equivalent to F162S in human NHE1), was found to substantially decrease both Na o ϩ affinity (ϳ11-fold) and sensitivity to HOE642 (ϳ1550-fold), although affinities for other transportable cations such as Li ϩ and H ϩ or the inhibitor guanidinium were unaffected (32). In the present report, mutations that affected drug sensitivity (P157S/P158F; L167F; E350Q; G356A; and L167F/G356A) did not affect Na o ϩ affinity. In the cases of E350Q/D/N and G356A/S/D, intracellular H ϩ affinity was also unaffected. The absence of an effect on cation affinities for the Glu 350 mutants was initially somewhat surprising since it seemed like a good candidate site for cation binding given its negative charge, location within a transmem- FIG. 8. Transport activity of wild type (WT) and mutant NHE1s as a function of the extracellular Na ؉ concentration. AP-1 cells expressing wild type and mutant NHE1s were preloaded with H ϩ using the NH 4 Cl technique, and 22 Na ϩ influx was measured at increasing concentrations of extracellular Na ϩ . Low levels of background 22 Na ϩ influx that were not inhibitable by 100 M EIPA were subtracted from the total influx. The background influx as a function of the Na ϩ concentration varied between 0.5 and 15% of the total for wild type and mutant NHE1s, with the exception of E350Q, where background influx was ϳ30%. The higher background for E350Q most likely results from the inability of EIPA to fully inhibit this mutant (see Fig. 5). Na ϩ /H ϩ exchanger activity was expressed as EIPA-inhibitable 22 Na ϩ influx (nmol/min/mg of total cellular protein). Values represent the average of two experiments, each performed in quadruplicate. The apparent affinity constants (K Na ) for Na o ϩ are summarized in Table III.  9. Transport activity of wild type and mutant NHE1s as a function of the intracellular H ؉ concentration. AP-1 cells stably expressing wild type NHE1Ј (WT) or mutants of NHE1Ј, E350Q (A), and G356A (B), were grown to confluence in 24-well plates. Initial rates of amiloride-inhibitable H ϩ -activated 22 Na ϩ influx were measured at various intracellular H ϩ concentrations over the range of pH i 7.4 -5.4. The pH i was adjusted by the potassium nigericin method, as described in "Experimental Procedures." To facilitate comparison of the effects of mutating these sites, data were normalized to the maximal uptake rates. Values represent the average of two experiments, each performed in quadruplicate. brane helix, and its high conservation among all mammalian NHEs. However, more detailed phylogenetic sequence comparisons of the NHEs reveal that many lower organisms (i.e. prokaryotes and simple eukaryotes such as yeast) can tolerate other, mainly polar, amino acids at this position, including Ala, Cys, Ser, Thr, His, Gln, Asp, and Asn (analysis not shown). This suggests that a glutamate residue at this position is not essential for function.
However, despite the absence of an effect on cation affinities, the vast majority of mutations that decreased drug binding also dramatically reduced the intrinsic velocity or turnover of the transporter relative to wild type, with the exception of L167F. This suggests that these sites may be important for the ensuing conformational changes that occur after cation binding and during translocation. In a sense, mutations at these sites may partly mimic potential conformational constraints imposed by antagonist binding, thereby compromising optimal substrate translocation.
In summary, we have defined several distinct sites between helices M3-M4 (Gly 152 , Pro 157 , Pro 158 ) and within M9 (Gly 356 ) that partly account for the differential drug sensitivities between NHE1 and NHE3 and which may apply to other isoforms as well. Significantly, mutations of Gly 356 appear to have a greater effect on binding of the more NHE-selective antagonists such as EIPA when compared with amiloride. In addition, we have identified a highly conserved glutamate residue (Glu 350 ) in the ninth transmembrane helix that is also a critical determinant of drug recognition by mammalian NHEs. Mechanistically, these sites are not involved in the competitive interaction between Na o ϩ and drug binding but, rather, appear to contribute significantly to efficient cation translocation. Further molecular dissection of the drug binding region should facilitate the rational design of more potent and isoform-specific drugs that may be therapeutically beneficial in the treatment of certain human diseases. NHE1s. Stably transfected AP-1 cells were lysed, and the protein extracts were subjected to SDS-polyacrylamide gel electrophoresis followed by protein transfer to polyvinylidene difluoride membranes. The membranes were probed with a rabbit polyclonal affinity-purified anti-NHE1 antibody and detected using chemiluminescence as detailed in "Experimental Procedures." The plasmalemmal fully glycosylated (fg) and intracellular core-glycosylated (cg) forms of NHE1Ј are indicated to the left of the panel. Molecular mass (kDa) is indicated to the right. B, densitometric analysis of the fully glycosylated form of wild type and drug-resistant mutants of NHE1. Band densities were normalized to that of the wild type exchanger, which was present on each Western blot. Values represent the mean Ϯ S.D. of 2-5 independent measurements (number indicated in parentheses). C, maximal cellular rates of H i ϩ -activated 22 Na ϩ influx of wild type and drug-resistant mutants of NHE1 following acid loading by the NH 4 Cl prepulse technique (for details, see "Experimental Procedures"). The rate of transport was expressed as pmol/min/mg of total cellular protein. Values represent the mean Ϯ S.D. of 2-12 independent measurements (number indicated in parentheses). D, comparison of the relative activity of wild type and drug-resistant mutants of NHE1. Cellular rates of 22 Na ϩ influx (panel B) were normalized to levels of fully glycosylated NHE1 protein expression determined by densitometry (panel C). Values represent the mean Ϯ S.D.