NHE1, NHE2, and NHE3 contribute to regulation of intracellular pH in murine duodenal epithelial cells

J. Praetorius, D. Andreasen, B. L. Jensen, M. A. Ainsworth, U. G. Friis, T. Johansen


Na+/H+-exchangers (NHE) mediate acid extrusion from duodenal epithelial cells, but the isoforms involved have not previously been determined. Thus we investigated1) the contribution of Na+-dependent processes to acid extrusion, 2) sensitivity to Na+/H+ exchange inhibitors, and 3) molecular expression of NHE isoforms. By fluorescence spectroscopy the recovery of intracellular pH (pHi) was measured on suspensions of isolated acidified murine duodenal epithelial cells loaded with 2′,7′-bis(2-carboxyethyl)-5(6)-carboxyfluorescein. Expression of NHE isoforms was studied by RT-PCR and Western blot analysis. Reduction of extracellular Na+ concentration ([Na+]o) during pHirecovery decreased H+ efflux to minimally 12.5% of control with a relatively high apparent Michaelis constant for extracellular Na+. The Na+/H+exchange inhibitors ethylisopropylamiloride and amiloride inhibited H+ efflux maximally by 57 and 80%, respectively. NHE1, NHE2, and NHE3 were expressed at the mRNA level (RT-PCR) as well as at the protein level (Western blot analysis). On the basis of the effects of low [Na+]o and inhibitors we propose that acid extrusion in duodenal epithelial cells involves Na+/H+ exchange by isoforms NHE1, NHE2, and NHE3.

  • intracellular pH
  • 2′,7′-bis(2-carboxyethyl)-5(6)-carboxyfluorescein
  • ethylisopropylamiloride
  • amiloride
  • fluorescence spectroscopy
  • reverse transcriptase-polymerase chain reaction
  • Western blot
  • duodenum
  • sodium/hydrogen exchanger isoforms

four isoforms of the Na+/H+exchanger (NHE) have been localized in the gastrointestinal tract by immunologic investigation. NHE1 is an important isoform for regulation of intracellular pH (pHi) and cell volume in nearly all eukaryotic cells (27, 33). Data from rat intestine suggest that NHE1 is localized at the basolateral membrane of the epithelial cells (6). The function of NHE2 is less well defined, but there is evidence for a role in cell volume regulation in renal tubule cells (23). Expression of NHE2 has been demonstrated in duodenum (Ref. 7; for review, see Ref.18), and it is localized at the brush-border membrane in cells from the gastrointestinal tract (14). The apical isoform NHE3 has been associated with transepithelial transport of Na+ (6, 25). The apical NHE3 isoform was expressed in segments from jejunum to colon in rabbit (14), but an attempt to localize NHE3 to the duodenum was unsuccessful (25). Little is known about the physiological function of NHE4, which has been found in stomach and kidney cortex (33). Na+/H+ exchange has been demonstrated in isolated rabbit and human duodenal epithelial cells (2, 16). Most of the Na+/H+ exchange, measured as Na+-dependent H+ efflux, was inhibited by amiloride or ethylisopropylamiloride (EIPA). The physiological significance of duodenal NHEs has been regarded solely as defense against intracellular acidification. On the other hand, the duodenal mucosa contributes to intestinal Na+ absorption (21), a process that is mediated by the NHE3 isoform in ileum and jejunum. In the present investigation we have used a pharmacological and a molecular approach to reveal which isoforms are present in proximal duodenal epithelial cells from mice.



Amiloride hydrochloride, antibiotic-antimycotic solution (A9909, containing penicillin, streptomycin, and amphotericin), EDTA, HEPES,N-methyl-d-glucamine (NMDG), and RPMI 1640 medium (R-7388) were obtained from Sigma Chemical. EIPA and 2′,7′-bis(2-carboxyethyl)-5(6)-carboxyfluorescein acetoxymethyl ester (BCECF-AM) were purchased from Molecular Probes. Primers were synthesized by DNA-Technology. Taq polymerase, PCR buffer, collagenase A, and Complete-mini protease inhibitor were supplied by Boehringer-Mannheim. Deoxyribonucleotides were from Pharmacia. Oligo dT primer, Escherichia coli (DH5α), first-strand buffer, and reverse transcriptase were supplied by Gibco BRL. RNasin, EcoR I, BamH I, plasmid vector pSP73, and vector-specific primers SP6 and T7 were obtained from Promega. Affinity purified polyclonal antibodies against NHE1 (AB3081), NHE2 (AB3083), and NHE3 (AB3085) were obtained from Chemicon International. Horseradish peroxidase-conjugated goat anti-rabbit IgG and Renaissance Western Blot Chemiluminescence Reagent Plus were from NEN Life Science Products. Immobilon-P transfer membrane for semidry blotting was supplied by Millipore. All other chemicals were of analytical grade. The RPMI medium was adjusted to contain 1.0 mM Ca2+, 1 mg/ml albumin, 5.6 μM indomethacin, 5 IU/ml penicillin, 5 μg/ml streptomycin, 12.5 ng/ml amphotericin B, and 0.26 mg/ml phenylmethylsulfonyl fluoride. Drugs were dissolved in DMSO or ethanol, and final concentrations of DMSO or ethanol were <0.1% (vol/vol) except for the RPMI medium [0.4% (vol/vol)].

Cell preparation.

C57BL/6 mice had free access to water and food until 1 h before use. Animals were killed at 37.0 ± 1.4 days of age (mean ± SE, n= 38) by cervical dislocation. After laparotomy, the proximal duodenum was excised (starting 2 mm distal to pylorus) and transferred tosolution A (Table 1). The lumen was rinsed, and the luminal surface was exposed to solution A for 5 min to remove remaining mucus. Epithelial cells were isolated by a calcium chelation procedure modified from Isenberg et al. (16). The luminal surface was exposed to the EDTA-containing solution Bfor 20 min at 37°C. Epithelial cells were separated from the structural components of the duodenum by gentle manipulation and then washed by addition of RPMI and centrifugation (500 g) for 3 min at 4°C. The preparation was treated with collagenase A (0.5 mg/ml in RPMI) for 10 min at 2°C and washed twice by dilution and centrifugation as described above. Two hundred-microliter aliquots of RPMI cell suspension were kept at 2°C until use (within 30–120 min).

View this table:
Table 1.

Composition of solutions

Inspections of the duodenum by microscope at ×80 stereomagnification confirmed that epithelial cells had been removed and that subepithelial structures were preserved. Hematoxylin-eosin staining was performed on cells from all experiments showing single epithelial cells and groups of up to 50 cells. Fewer than 5% of cells differed in staining and size from the typical epithelial cell. No cellular or structural components from connective, muscle, or fatty tissue were seen. As an indirect measure of cell viability, we observed no leak of fluorescent probe during experiments. Furthermore, the cells were capable of regulating pHi after acidification.

Determination of pHi.

Measurement of pHi was performed by cuvette-based spectroscopy using the membrane-permeant form of the pH-sensitive fluorescent dye BCECF. After 15 min at 37°C the cells were loaded with BCECF-AM (5 μM) for an additional 15 min. Extracellular dye was removed by 25× dilution with solution 1, centrifugation for 2 min at room temperature (500 g), and resuspension of the cells in solution 1. Cuvettes containing 1.6 ml of cell suspension were placed in a dark chamber at 37°C and continuously stirred. Measurements were initiated after 2 min of equilibration. A rotating mirror shifted the excitation light between 440-nm and 490-nm bandpass filters (bandwidth 5 nm) with a sampling rate of 1 Hz. A photomultiplier tube (Photon Technology Instruments) collected the light emissions at 535 nm. From these data a ratio of the emitted light at excitation of 490 nm to that at 440 nm was calculated. No bleaching of the probe was observed even during prolonged measurements (up to 30 min of continuous excitation).

The fluorescence ratio (490/440) was calibrated to pHi by the high-K+/nigericin technique (24). Isolated cells were resuspended in 1.8 ml of solution 5, and the fluorescence ratio was measured after 2–5 min of equilibration. Extracellular pH (pHo) was adjusted stepwise from 7.4 to a maximum of 8.0 and then to a minimum of 6.0 by addition of small volumes (0.5–2 μl) of 2 N KOH or 1 N HCl (see Fig. 1 A). pHo was determined using a pH electrode (Radiometer) at each step of the fluorescence measurement. Assuming complete clamping of pHito pHo, the recorded fluorescence ratios (490/440) were plotted as a function of pHo and the points were fitted to a fourth-order polynomial using the software Igor Pro (WaveMetrics). The polynomial was used to calibrate the 490-to-440 ratio from the experiments on pHi.

The intrinsic buffering capacity was determined by a method modified from Boyarsky et al. (8) involving stepwise decreasing NH4+ concentrations. The cells were resuspended in 1.6 ml of solution 2, and, after 2 min, 1.6 ml of a 40 mM NH4+ solution (solution 4) was added. NH4+ concentration was then reduced from 20 to 10 mM by replacing one-half of the cell suspension with solution 2. Fluorescence was measured for 30 s, and the twofold dilution was repeated four times. Buffering capacity was calculated as values of change (Δ) in intracellular NH4+ concentration (Δ[ NH4+ ]i)/ΔpHi, assuming a permeability for NH3 >> NH4+ , steady-state distribution of NH3 across the plasma membrane, and Δ[ NH4+ ]i = change in intracellular H+ concentration (Δ[H+]i). Values of buffering capacity were grouped into pHi intervals of 0.15 in the pHi range of 6.7 to 7.6, and mean and SE values were computed. The data points were weighted by the SE for each value and fitted by linear regression using Igor Pro software (singular value decomposition algorithm).

The calibrated data from the experiments are presented as pHi vs. time traces (see Figs. 3 and 4, insets). For NH4+ prepulse experiments, the pHi recovery recordings were fitted to double-exponential functions. This function provided better fits of the data than single-exponential functions during pHi recovery. The slope of the pHi recovery curve (ΔpHi/dt) was plotted against pHi. Net H+ flux at a given pHi was calculated by multiplication of ΔpHi/dt by the buffering capacity for the corresponding pHi. In each single experiment, H+ fluxes were calculated in steps of 0.05 pH units throughout the pHi interval. The light emission was quenched in recordings on samples containing >0.5 mM amiloride. Therefore, these samples were calibrated in the presence of amiloride. Values of steady-state pHi were determined from the double-exponential function for each experiment. To compare decreases in steady-state pHi induced by low [Na+]o or inhibitors of Na+/H+ exchange, the mean change from control pHi was calculated as a Δ[H+]i value. This was necessary to correct for day-to-day variations of the control steady-state pHi. To illustrate these results in terms of pHi, the mean changes in [H+]i induced by low [Na+]o, EIPA, or amiloride were then converted to decreases of pHi from the mean control steady-state value (pH 7.31).

The measurements were validated by the observation of identical values of the buffering capacity and net H+ efflux at pHi 7.02 in mouse compared with rabbit duodenal cells (1). Furthermore, the initial pHi was close to the reported values for duodenal villus cells from rabbit (1) obtained with fluorescence microscopy and from rat (29) measured using a confocal technique. Data are presented as mean values ± SE. The Mann-Whitney rank-sum test was used for statistics.

Design of PCR primer pairs for NHE1, NHE2, NHE3, and NHE4.

Nucleotide sequences and Genbank accession numbers were retrieved using the NCBI resource Entrez (17): mouse NHE1, U51112 (13); rat and rabbit NHE2, L11004 and L13733, respectively (10, 26, 28); rat and rabbit NHE3, M85300 and M87007, respectively (20, 25); and rat NHE4, M85301(20). To identify highly conserved regions specific to NHE2 or NHE3, rabbit and rat sequences were aligned by use of the Genestream resource ALIGN (15). To facilitate cloning, the primers were constructed with adapters containing restriction sites at the 5′end forBamH I (antisense primers) and EcoR I (sense primers). Primer sequences were NHE1: sense 5′CTG TGG TCA TTA TGG CC′3, antisense 5′TGG GTT CAT AGG CCA GT′3, covered bases 1,938–2,458 of mouse NHE1 cDNA, 521 bases; NHE2: sense 5′CAG CAA GCT GTC AGT GA′3, antisense 5′TCG GGA GGT TGA AGT AG′3, covered bases 1,684–2,199 of rat NHE2 cDNA, 516 bases (adapter sequences are omitted). Two pairs of primers were used for NHE3: sense 5′TTC GAC CAC ATC CTC TC′3, antisense 5′TCC TTG TCC TGC TTC TC′3, covering bases 1,529–2,034 of rat NHE3 cDNA, 506 bases (pair a), and sense 5′TGG CCT TCA TTC GCT CC′3, antisense 5′TAC TCC TGC CGA GGC TTG′3, covering bases 1,746–1,974 of rat NHE3 cDNA, 229 bases (pair b). Primer sequences for NHE4 were sense 5′ACA TCT CTG CGA TCG AG′3 and antisense 5′ATC CCA GCC TTC AGA TG′3, covered bases 1,356–1,980 of rat NHE4 cDNA, 625 bases.

Isolation of mRNA and RT-PCR.

Isolated duodenal epithelial cells were lysed, and total RNA was extracted using the RNeasy Mini Kit (Qiagen). RNA (125 μg/ml) was denatured in the presence of oligo dT primer (62.5 μg/ml) and yeast tRNA (187.5 μg/ml) at 95°C for 3 min in a volume of 8 μl. Denatured RNA was reverse transcribed by incubation with (final concentrations) 0.5 mM (of each base) deoxyribonucleotides, 10 mM dithiothreitol, 1 IU/μl RNasin, 0.5 μg/μl bovine serum albumin, 10 U/μl reverse transcriptase, and buffer in a total volume of 20 μl for 60 min at 37°C. The reaction was terminated by heating at 95°C for 4 min.

The cDNA product was amplified by PCR. To 3 μl of cDNA produced by the RT reaction was added (final concentrations) 250 μM deoxyribonucleotides, 50 U/ml Taq polymerase, 0.5 μM each for primers, PCR buffer, and water to a final volume of 20 μl. After denaturation at 95°C for 3 min, 30 cycles of PCR amplification were performed: denaturation at 95°C for 30 s, annealing at 50°C for 30 s increasing with 2°C for each of the first five cycles to 60°C, and polymerization at 72°C for 30 s. The last cycle included 10 min at 72°C. A 15-μl sample of the PCR product and loading buffer was analyzed by electrophoresis on a 2% agarose gel. The gel was stained with ethidium bromide and photographed under ultraviolet illumination. Negative controls were yeast tRNA without duodenal RNA in the RT reaction and PCR and omission of reverse transcriptase in RNA containing samples during the RT reaction. Positive control for NHE3 and NHE4 was RNA obtained from kidneys.

Sequence analysis.

Bands of predicted molecular weight for each PCR product were excised from the gel and purified using the QIAquick Gel Extraction Kit (Qiagen). The eluted DNA fragments of NHE1, NHE2, NHE3 (primer pair a), and NHE4 were ligated into BamH I/EcoR I polylinker sites of the plasmid vector pSP73 for transformation of competent E. coli (DH5α) using standard procedures (22). The resulting plasmids were extracted from the relevant clones of E. coli by use of the Plasmid Maxi Kit (Qiagen). Both strands were sequenced using the vector-specific primers SP6 and T7 and analyzed on a 310 Genetic Analyzer (ABI PRISM, PE Applied Biosystems). Excised NHE3 products obtained with PCR primer pair b were sequenced directly after purification.

Western analysis.

Duodenal epithelial cells were lysed in a buffer containing 0.1% Triton X-100 and protease inhibitors (Complete mini, Boehringer-Mannheim), and a glycerol-bromphenol blue loading buffer was then added to obtain final concentrations of 1% SDS and 33 mM Tris. After boiling for 5 min samples were centrifuged at 13,000 rpm in a table top centrifuge. The protein-containing supernatant was loaded onto a 7% acrylamide gel and run at 200 V for 45 min. Gels were processed according to instructions for the Mini-Protean 3 Cell (Bio-Rad). Proteins were electrotransferred to a PVDF membrane at 0.8 mA/cm2 for 1 h, and the membrane was blocked by incubation in 5% nonfat dry milk in a Tris-buffered salt solution [TTBS, containing 20 mM Tris, 137 mM NaCl and 0.05% (wt/vol) Tween 20] overnight at 5°C. The membrane was incubated with primary antibody in 2% nonfat dry milk in TTBS for 2 h at room temperature in concentrations of 0.125–2.5 μg/ml as indicated in Fig. 6. The primary antibodies were affinity-purified anti-NHE1 antibody (Chemicon AB3081) raised in rabbit against a 22-amino acid domain of human NHE1 (30); affinity-purified anti-NHE2 antibody (Chemicon AB3083) raised in rabbit against a 20-amino acid domain of rat NHE2 (7); and affinity-purified anti-NHE3 antibody (Chemicon AB3085) raised in rabbit against a 22-amino acid domain of human NHE3 (31). The membrane was washed twice for 5 min and once for 15 min in TTBS before incubation with horseradish peroxidase-conjugated goat anti-rabbit IgG secondary antibody (0.2 μg/ml) for 1 h in 2% nonfat dry milk in TTBS. Excess antibody was then washed once for 15 min and four times for 5 min each in TTBS, and bound antibody was detected by chemiluminescence (Renaissance Western Blot Chemiluminescence Reagent Plus).



The high-K+/nigericin method was used for calibration of the fluorescence recordings to pHi (Fig.1). Figure 1 A shows the sequence of changes in fluorescence ratio (490/440) in response to changes in pHo. The final pHo was chosen near the initial value to ensure that 490-to-440 fluorescence ratios were not affected by the dilution of samples or by ratio drifting during the measurement. The data points were fitted to a fourth-order polynomial function: pHi(x) = 4.1702 + 5.1816 × x − 3.4078 × x 2 + 1.002 ×x 3 (Fig. 1 B). The changes in fluorescence ratio (490/440) in response to alterations in pHo were completed in a few seconds.

Fig. 1.

Calibration of intracellular pH (pHi). Duodenal epithelial cells were loaded with 2′,7′-bis(2-carboxyethyl)-5(6)-carboxyfluorescein-AM (BCECF-AM), and high-K+/nigericin technique was used to clamp pHi to extracellular pH (pHo) (solution 5). A: pHo was varied from 6.0 to 8.0 as indicated above trace. Ordinate scale: fluorescence excitation ratio of light emitted at 535 nm (excitation 490 nm/440 nm). B: data from 27 experiments performed on 11 mice were fitted by a 4th-order polynomial (curve).

Buffering capacity.

It was necessary to obtain values for the intrinsic buffering capacity for the conversion of pHi changes to H+ efflux. The relation between pHi and the intrinsic buffering capacity is shown in Fig. 2. Data were obtained as pHi changes in response to decreases of extracellular NH4+ (ΔH+/ΔpHi) (Fig. 2, inset). The calculations of the mean intrinsic buffering capacity (βint) were performed as described previously: βint = 186.1 − 23.5 × pHi for the pHi interval from 6.7 to 7.6. The correlation coefficient for the linear relation was 0.997. Except at 20 mM [ NH4+ ] the recordings were stable at the various levels of pHi. The decrease in pHi observed with 20 mM [ NH4+ ] is most likely the result of NH4+ influx (8).

Fig. 2.

Determination of buffering capacity of mouse duodenal epithelial cells. Relation between pHi and intrinsic buffering capacity was determined by stepwise removal of Formula . Cells were incubated in a sodium-free medium (solution 2, Table 1). After 30 s, solution 4 was added to obtain a solution with [ Formula ] of 20 mM (inset, recording of 1 experiment). Changes in pHi used for calculations were measured in stepwise decreasing concentrations of NH4Cl (from 20 to 1.25 mM). Mean values of buffering capacity were calculated in intervals of 0.15 pH units at pHi from 6.7 to 7.6. Mean values of the buffering capacity are shown. A total of 51 experiments were performed on 6 mice; bars show SE for mean pHi values and buffering capacity. Number of data points in each of the 6 groups were (in order of increasing pHi from 6.7 to 7.6) 13, 49, 53, 42, 37, and 9. Linear fitting was used to determine relation between pHi and buffering capacity.

H+ efflux from acid-loaded cells: Dependence on extracellular Na+.

NH3/ NH4+ induced a transient increase of pHi (Fig. 3,inset). After removal of NH3/ NH4+ , pHidecreased by ∼0.6 pH units. The acid extrusion rate was increased at low pHi and decreased at low [Na+]o in mouse duodenal epithelial cells (Fig. 3 A). The rate of pHi recovery was dependent on [Na+]o in the range investigated (in mM): 1, 10, 19, 37, 73, 109, and 145. pHirecovery was incomplete in experiments with low [Na+]o because of the limited observation period of 5 min. [Na+]oof 1 mM decreased the net H+ efflux at pHi 6.85 to 12.5% compared with the control H+ efflux at [Na+]o of 145 mM (decrease from 0.2251 ± 0.0249 to 0.0283 ± 0.0031 mM/s, n = 5). A strong H+ sensitivity of pHi recovery was illustrated by a 24-fold decrease of H+ efflux as pHi was raised from 6.75 to 7.25 in the presence of 145 mM Na+(0.3210 ± 0.0432 vs. 0.0134 ± 0.0056 mM/s). A similar dependence on pHi was observed at lower [Na+]o, and the level of H+ efflux for a given pHi was gradually decreased when the concentration of Na+ was reduced from 145 to 1 mM. The estimated concentration of Na+ was 1 mM when the sodium-free solution was applied to washed cells (∼150-fold dilution). The concentration-effect curve (Fig. 3 B) illustrates the [Na+]o dependence of the net H+ efflux in percentage of net H+ efflux with [Na+]o of 145 mM (control). For each value of [Na+]o, the day-matched efflux percentage was calculated. Values were obtained at five levels of pHi from 6.75 to 6.95. For each [Na+]o, mean values of the percentage of H+ efflux were calculated in the five experiments at the five different pHi levels. The curve obtained shows a biphasic shape with a steep slope at maximal [Na+]o. Although the curve does not allow the determination of an exact value, the overall apparent Michaelis constant (K m) for extracellular Na+ is 34 mM. From a Eadie-Hofstee plot (not shown) two apparent K m values could be estimated: a high value of 36 mM based on five data points and a low value of ∼2 mM based on only two data points.

Fig. 3.

Dependence of H+ efflux on extracellular Na+.A: duodenal epithelial cells were acid loaded by NH4Cl prepulse technique (30 mM for 3 min, solution 1). After washing procedure cells were resuspended in presence of various concentrations of extracellular Na+([Na+]o, in mM): 1, 10, 19, 37, 73, 109, and 145 by combinations of solutions 2 and 3. Data represent mean values + SE from 5 experiments. Inset, traces from experiments performed on 1 day. B: concentration effect curve for [Na+]o dependence of H+ efflux. Abscissa, [Na+]o (mM; log scale); ordinate, H+ efflux in % of control efflux at [Na+]o = 145 mM. Mean values ± SE were obtained at pHi 6.75–6.95 for all values of [Na+]o (SE of mean values from 5 different experiments).

H+ efflux from acid-loaded cells: Inhibition by EIPA and amiloride.

The effects of Na+/H+ exchange inhibitors were studied on acid loaded cells by the NH4Cl prepulse technique. Both amiloride and EIPA were used, the latter to exert a more potent inhibition of NHE isoforms (predominantly NHE1). EIPA gradually decreased the acid efflux in concentrations ranging from 0.1 to 40 μM (Fig. 4 A). The inhibition of net H+ efflux was observed throughout the pHi recovery. We used amiloride below millimolar concentrations to block NHE1 and NHE2 and very high concentrations to obtain inhibition of NHE3. Figure 4 B shows the effect of amiloride on net H+ efflux of acid-loaded duodenal epithelial cells. Four experiments were performed with amiloride at 0.1, 1.0, and 2.0 mM. Amiloride decreased the pHi recovery dose-dependently in this range and, in one experiment (same as Fig.4 B, inset) in the range of 0.001 to 2.0 mM. Control samples were incubated without amiloride. Figure 4 C shows the concentration-effect curve for the inhibition of H+ efflux by EIPA or amiloride in percentage of H+ efflux of control samples incubated without inhibitor. For each concentration of EIPA or amiloride, the day-matched inhibition percentage was calculated. Values were obtained at five levels of pHi from 6.75 to 6.95. For each concentration of inhibitor, mean values of the inhibition percentage were calculated in the five experiments at the five different pHi levels. The inhibition of H+efflux by EIPA resulted in an increased inhibition with EIPA concentrations up to 40 μM. Acid efflux was inhibited to a lower extent with 100 μM of EIPA than with 40 μM in all four experiments. This was likely caused by nonspecific effects of the high concentration of the drug. Maximal inhibition was 56.5 ± 1.2% of the net H+ efflux. IC50 was estimated as ∼1.48 × 10−7 M EIPA from an Eadie-Hofstee plot (not shown). Amiloride at 0.1 mM inhibited net H+ efflux by 53.7 ± 1.3%, whereas 1 or 2 mM amiloride resulted in inhibition of H+ efflux by 71.7 ± 1.0 or 79.9 ± 2.8%, respectively. This inhibition of H+ efflux was almost of the same magnitude as that obtained by low extracellular Na+, and a plateau of inhibition was not obtained with high concentrations of amiloride.

Fig. 4.

Inhibition of pHi recovery from an acid-load by ethylisopropylamiloride (EIPA) and amiloride. Mouse duodenal epithelial cells were acid loaded by a NH4Cl prepulse and then incubated with EIPA or amiloride (in solution 1). A: effect of EIPA on H+ efflux during recovery from low pHi. Control value and effects of 0.1, 1.0, and 40 μM EIPA are shown; effects of 0.5 and 100 μM EIPA are left out for clarity. Mean values + SE from 4 experiments are shown. Inset, traces from experiments performed on 1 day. B: effect of amiloride on H+ efflux during recovery from low pHi. Cells were incubated with 0.1, 1.0 ,or 2.0 mM amiloride or, as control, without amiloride. Mean ± SE values from 4 experiments are shown. Inset, traces from experiments performed on 1 day with 0.01, 0.05, 0.1, 0.5, 1.0, or 2.0 mM amiloride or without inhibitor. C: concentration-effect curve of inhibition of H+ efflux by EIPA and amiloride. Abscissa, concentration of EIPA ([EIPA]) or amiloride ([amiloride]) (M; log scale); ordinate, inhibition of H+ efflux in % of control efflux without inhibitor. For each [EIPA] (●) or [amiloride] (□), values of H+ efflux were determined as % of day-matched H+ efflux without inhibitor. A single measurement with 0.01 μM of EIPA was performed. ▵, values from single experiment shown in inset to B. Mean ± SE values were obtained at pHi 6.75–6.95 for all values of [EIPA] or [amiloride] (SE of mean values from 4 different experiments).

Effects of EIPA, amiloride and low [Na+]oon steady-state pHi.

An estimate of the control steady-state pHi of 7.31 ± 0.02 (n = 13) can be obtained from the double-exponential fits of the control samples. As shown in Fig. 5, both EIPA and amiloride decrease steady-state pHi after an acid load. This change of pHi is dependent on the concentration of inhibitor, and high concentrations of amiloride provide the largest deviation of pHi. EIPA at 5 μM decreased steady-state pHi by 0.036 ± 0.019 units, whereas amiloride at 0.1 mM decreased steady-state pHi by 0.072 ± 0.011 units (P < 0.05). Reduction of [Na+]o from 145 mM to 109, 73, 37, or 19 mM did not affect steady-state pHi significantly. At [Na+]o of 10 and 1 mM, steady-state pHi was decreased by 0.0639 ± 0.0080 and 0.1010 ± 0.0159 units, respectively.

Fig. 5.

Influence of EIPA, amiloride, and [Na+]o on steady-state pHi. Duodenal epithelial cells were acid loaded and incubated with various [Na+]o, [amiloride], or [EIPA] during recovery of pHi (as shown in Figs. 3 and 4). Deviations from control steady-state pHi with [Na+]o of 145 mM in absence of inhibitors were calculated for each experiment. Mean values + SE from 4 (EIPA, amiloride) or 5 ([Na+]o) experiments are shown.

Detection of NHE1, NHE2, and NHE3 by RT-PCR and Western blot analysis.

Expression of NHE isoforms was then investigated as exemplified in Fig.6 A. The figure shows the gel electrophoresis of the RT-PCR for NHE1, NHE2, NHE3, and NHE4 mRNA in mouse proximal duodenal epithelial cells. PCR yielded products of the expected molecular weight with all five primer pairs. Bands were detected for isoforms NHE1, NHE2, and NHE3 (using both primerpair a and pair b) in samples from duodenal epithelial cells, whereas a NHE4 band was only detected in samples with kidney extract cDNA. Nucleotide sequences were determined on excised DNA fragments. The NHE1 fragment of 521 bp shared 100% nucleotide sequence identity with the corresponding region of the published mouse NHE1 (13). The NHE2 fragment was 516 bp in length and shared 94% nucleotide sequence identity and 99% amino acid homology with rat NHE2 (10). The combined NHE3 fragment from amplification using primer pairs aand b provided a nucleotide sequence of 274 bp. The sequence shared 95% nucleotide sequence identity and 100% amino acid homology with rat NHE3 (20). The kidney NHE4 fragment of 585 bp shared 94% nucleotide sequence identity with rat NHE4 (20). The nucleotide sequence identity with rabbit NHE2 and NHE3 was lower than with rat sequences, amounting to 87% identity for NHE2 (26) and 82% for NHE3 (25). The sequences of the novel murine nucleotide fragments were given the Genbank accession numbers AF139193 (NHE2), AF139194 (NHE3), andAF139195 (NHE4).

Fig. 6.

Duodenal expression of Na+/H+ exchanger (NHE) isoforms. A: mRNA extracted from isolated duodenal epithelial cells was used as template for RT-PCR with 5 specific primer pairs for NHE isoforms 1–4. Markers for DNA molecular weight are shown inlanes 1 and 11. Lanes 2, 5, 8,12, and 13, reverse-transcribed mRNA from duodenal epithelial cells used as template for primers specific for NHE1, NHE2, NHE3 (primer pairs a and b), and NHE4, respectively.Lanes 4, 7, and 10, RT-PCR with similar template and primers as above without reverse transcriptase. Inlanes 3, 6, 9, 16, and 17template was omitted in RT-PCR with primers for NHE1, NHE2, NHE3 (primer pairs a and b), and NHE4, respectively.Lanes 14 and 15, reverse-transcribed mRNA from crude kidney extract used as template for NHE3 (primer pair b) and NHE4 primers. Typical results of RT-PCR on duodenal and kidney mRNA preparations are shown. B: Western blot analysis using antibodies against NHE1, NHE2, and NHE3. Protein fractions of isolated duodenal epithelial cells were separated by acrylamide gel electrophoresis and electrotransferred to PVDF membranes. Western blot analysis was performed with affinity-purified polyclonal antibodies against COOH-terminal domains of NHE1, NHE2, and NHE3. Lane 1, molecular mass standard. For detection of NHE proteins we used 2.5 μg/ml anti-NHE1 antibody, 0.125 μg/ml anti-NHE2 antibody, and 0.25 μg/ml anti-NHE3 antibody in lanes 2–4, respectively.

Western blot analysis was performed to demonstrate expression of the three NHE isoforms at the protein level, as shown in Fig. 6 B. The anti-NHE1 antibody recognized a 100-kDa protein from isolated duodenal epithelial cells as well as a 80-kDa protein with 2.5 μg/ml of the primary antibody. The anti-NHE2 and anti-NHE3 antibodies recognized proteins of ∼80–85 kDa in duodenal epithelial cells, and for NHE2 an additional band was detected at 65–70 kDa. Optimal concentrations were 0.125 and 0.625 μg/ml of the antibodies, respectively.


In this investigation we have demonstrated the presence of three isoforms of NHE in isolated epithelial cells from proximal mouse duodenum. Na+/H+ exchange was measured as Na+-dependent pHi recovery in acidified cells. Furthermore, the expression of the NHE isoforms was studied by RT-PCR and Western analysis.

The cellular extrusion of acid was characterized by a strong pHi dependency with maximal activity at low pHi. H+ efflux was dependent on [Na+]o in the higher range with a presumed inward gradient of Na+. However, even at [Na+]o in the range of 1 to 19 mM, which may be close to [Na+]i, H+ efflux was still dependent on extracellular Na+. The concentration-effect curve (Fig. 3 B) more clearly shows the [Na+]o dependency of H+ efflux, because the curve appears as the mean values of H+ efflux in percentage of control obtained at five different pHi levels. There was no saturation of the H+ efflux when [Na+]ovaried between 1 and 145 mM. Consequently, a direct estimation ofK m for extracellular Na+ was not possible, although estimations were made based on Eadie-Hofstee plots. Because the slope of the log-concentration-effect curve is steep at high concentrations of Na+, acid extrusion seems to be sensitive to changes in [Na+]o in the physiological range. It may be speculated that the biphasic shape of the curve and the determination of high and lowK m values could be explained by the action of isoforms with different sensitivity to extracellular Na+values. This view is supported by the observation that the apparentK m of rat NHE2 to extracellular Na+ was ∼5–10 times greater than the K m of NHE1 and NHE3 when expressed in Chinese hamster ovary cells (19, 32). However, comparison of the [Na+]osensitivity to previous determinations is difficult, because the NHEs very likely are regulated differently in various cell types. Intracellular messengers and regulating proteins control the function of NHE in cells with spontaneous expression of the exchangers, whereas elements of this regulation could be absent in transfected cells. The possible contribution of a high-K m NHE, like the NHE2, to acid extrusion might also be reflected in the steady-state pHi values obtained with low [Na+]o. Steady-state pHi was only slightly affected when [Na+]o was decreased from 145 to 19 mM, but [Na+]o values below that point led to larger decreases in steady-state pHi. A likely explanation is that only NHE2 is inhibited by lowering [Na+]o in the range of 109 to 19 mM. Na+/H+ exchange by NHE1 might then be sufficient to balance the metabolic acid production and provide a steady-state pHi near the control value. Cellular acid production likely decreases steady-state pHi when low [Na+]o inhibits NHE1 and NHE3 as well as NHE2.

The nature of the slow Na+-independent pHirecovery (Fig. 3 A) was not addressed in the present study. However, slow Na+/H+ exchange might explain the remaining pHi recovery even with low [Na+]o. Alternatively, leakage of either H+ or fluorescent probe from the cells might explain the alkalization. Such outward leak of H+ is unlikely, because the electrochemical driving force for protons is greatest in the inward direction, with a presumed membrane potential around −60 mV. Leakage of BCECF from the cells was ruled out because the fluorescence signal was not changed by renewal of the incubation medium after pHi recovery. Thus we conclude that Na+-dependent processes mediate most if not all acid extrusion in duodenal epithelial cells and that these processes reflect Na+/H+ exchange activity.

We next attempted to characterize the acid extrusion pharmacologically by the application of Na+/H+ exchange inhibitors. The inhibition of H+ efflux by EIPA in concentrations up to 40 μM allowed the determination of a single IC50, suggesting the involvement of only one NHE isoform sensitive to EIPA in the range tested. This isoform is likely NHE1, because it is inhibited by EIPA with 25 to 100 times lower IC50 than NHE2 and NHE3, respectively (18). Comparison of the IC50 value with earlier studies is difficult because the determinations were conducted with only trace amounts of [Na+]o and by measurements of22Na+ uptake (19). To our knowledge, IC50 values toward murine NHE1 and NHE2 have not been determined for EIPA. Another explanation of our finding could be that EIPA-sensitive H+ efflux as well as acid extrusion through NHE1 contained some degree of acid extrusion through NHE2. By applying a [Na+]o of 145 mM in our experiments, we expected higher IC50 values than reported with low [Na+]o, because amiloride and EIPA likely are competitive inhibitors of the exchangers. However, pharmacological discrimination between the isoforms should still be useful, because the relative sensitivities to EIPA of the isoforms were expected to be conserved at physiological Na+concentrations. The maximal inhibition revealed that a quantitatively large part of the H+ efflux was insensitive to EIPA. Amiloride was then used to study the EIPA-insensitive contribution to the acid extrusion. Both NHE1 and NHE2 are sensitive to inhibition by amiloride below millimolar concentrations, whereas NHE3 is relatively insensitive to amiloride. Amiloride inhibited H+efflux in a concentration-dependent manner in the range tested. At high concentrations acid extrusion was inhibited to a larger extent than the maximal inhibition with EIPA. This suggests that the H+efflux sensitive to micromolar amiloride is mediated by an EIPA-sensitive NHE1 and a relatively EIPA-resistant but amiloride-sensitive NHE2.

Another way of studying the contributions of NHE1 versus NHE2 is to determine the steady-state pHi after acid load during inhibition of Na+/H+ exchange. In the absence of CO2/ HCO3 , steady-state pHi would reflect a balance between cellular acid accumulation and acid extrusion. The steady-state pHi of duodenal cells is presumably obtained by the function of two or more NHEs. As noted above, amiloride inhibits NHE1 and NHE2 equally, but NHE2 is ∼25% less sensitive to EIPA than NHE1. If concentrations of the drugs are chosen to provide ∼50% inhibition of the net H+ efflux, one would expect amiloride to decrease steady-state pHi more than EIPA. This would be true if both isoforms participated in the maintenance of pHi near the observed steady-state level. Amiloride (0.1 mM) inhibited H+ efflux to almost the same extent as 5 μM of EIPA (53 and 51%, respectively). When the steady-state pHi values with these concentrations of inhibitors are compared, the deviation from control is significantly larger with amiloride than EIPA. Amiloride likely inhibits NHE1 and NHE2 to the same degree, and the inhibition of two isoforms provides a low steady-state pHi. This could be explained by an acid production exceeding the acid extrusion during inhibition of NHE1 and NHE2, leaving NHE3 to balance the acid production at a lower steady-state pHi. During inhibition with EIPA NHE1 is inhibited with only minor, if any, inhibition of NHE2. This would possibly allow NHE2 to maintain a steady-state pHi only slightly below control.

A very high concentration (2 mM) of amiloride inhibited the acid extrusion to almost the same extent as low [Na+]o. We did not observe a maximal inhibitory concentration for amiloride, and it is possible that further inhibition could be obtained with higher amiloride concentrations. However, we did not succeed in preparing stable solutions of amiloride >2 mM. The proportion of H+ efflux that was shown to be resistant to 2 mM amiloride amounted to ∼7.5% of the total H+ efflux. The fraction of the acid extrusion that was only sensitive to very high concentrations of amiloride could reflect the action of a third isoform with a low sensitivity to amiloride compared with NHE1 and NHE2. NHE3 is a likely candidate for mediating the relatively amiloride-resistant H+ efflux, because Na+/H+ exchange through NHE4 is slow (33). Theoretically, the amiloride-insensitive acid extrusion could be mediated by Na+-driven HCO3 transport into the cells. To reduce this problem, the present investigation was performed in the nominal absence of the CO2/ HCO3 buffer system. Even in the presence of bicarbonate, the duodenal Na+/H+-exchangers contributed ∼64% of the total “acid extrusion” compared with the bicarbonate-dependent processes (1). Furthermore, inhibitors of Na+- HCO3 cotransport (500 μM of DIDS or H2DIDS) had no effect on pHirestoration in duodenal epithelial cells in the absence of CO2/ HCO3 (unpublished data).

Our findings thus far point to a combined function of several isoforms of the NHE in acid-loaded duodenal cells. We then investigated the molecular expression of NHE isoforms in mouse duodenal epithelial cells. The demonstration of mRNA for the isoforms NHE1, NHE2, and NHE3 by RT-PCR and sequence analysis of the PCR products strongly support that these three are the NHE isoforms involved in duodenal pHi recovery. To validate this novel detection of NHE3 in duodenum we used two distinct primer pairs for PCR detection. Furthermore, the nucleotide sequences for the overlapping regions of the two NHE3 fractions were identical. However, a previous study by Northern blotting and RT-PCR reported that a message for NHE3 in rabbit duodenum was absent (25). This discrepancy might reflect interspecies variation or age-specific expression of isoforms as shown in rat jejunum for NHE2 and NHE3 (11). Western analysis revealed bands near the predicted molecular weights for all three isoforms using affinity-purified polyclonal antibodies directed against COOH-terminal peptides of the isoforms. The developers of the anti-NHE1 antibody, Yamashita and Kawakita (30), abolished the binding of the antibody to the NHE1 protein in Western blot analysis by incubation with the antigen, a COOH-terminal peptide. Likewise, the binding of anti-NHE2 and anti-NHE3 antibodies to the NHE isoforms were competed by the relevant COOH-terminal peptides by Bookstein et al. (7) and Yoshioka et al. (31), respectively. NHE1 has previously been identified as a 100- to 110-kDa protein in rodents (5, 12), whereas NHE2 and NHE3 were 80- to 90-kDa proteins (3, 4, 9, 14). Doublet bands have been reported for both NHE1 and NHE2 and may reflect nonspecific binding of the antibody or detection of immature, less glycosylated NHE proteins (5, 7, 9). Hence, the expression of all three isoforms was confirmed at the protein level in isolated duodenal epithelial cells. The presence of NHE3 in mouse duodenum provides a candidate for the mediation of duodenal Na+ absorption, although other functions of NHE3 and NHE2 might be considered. These could include pHiregulation in enterocytes exposed to acidic luminal contents under circumstances in which the mucous barrier is damaged. Alternatively, NHEs might be involved in volume regulation of duodenal epithelial cells, which could be very important for cell survival after challenge by changes in luminal osmolarity. Finally, the presence of a NHE with high K m for [Na+]o, as reported for NHE2, would allow duodenal cells to react to changes in the Na+concentration of the microenvironment in addition to changes in pHi. Future studies on the regulation and membrane distribution of duodenal NHEs are needed to elucidate the functional significance of the duodenal NHE isoforms.

In conclusion, we find that Na+/H+ exchange accounts for most acid extrusion in duodenal mucosal cells in the nominal absence of CO2/ HCO3 . The sensitivity toward inhibitors and reduced [Na+]o and the molecular biological findings suggest the involvement of the NHE1, NHE2, and NHE3 isoforms in pHi recovery from acid load in proximal duodenal epithelial cells.


The authors thank Annette K. Rasmussen for skilled technical assistance and Daniel L. Hogan for an inspiring introduction to the field of duodenal research.


  • Address for reprint requests and other correspondence: J. Praetorius, Physiology and Pharmacology, IMB, Univ. of Southern Denmark-Odense Univ., Winsloewparken 213, DK-5000 Odense C, Denmark (E-mail: jeppep{at}imbmed.sdu.dk).

  • This work was supported by the following foundations: Lily B. Lunds Fond, Overlægerådet Odense Universitetshospital, Direktør Ib Henriksens Fond, Nygård Fonden, and Danish Medical Research Council (9601778).

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.


View Abstract