An inducible intestinal epithelial cell-specific NHE3 knockout mouse model mimicking congenital sodium diarrhea

Sodium (Na⁺) absorption is accomplished by several different processes. Postprandial, secondary active transport processes including Na⁺/glucose cotransport and Na⁺/amino acid cotransport both significantly contribute to intestinal Na⁺ absorption. In addition, Na⁺/H⁺ exchange, in particular via the Na⁺/H⁺ exchanger isoform 3 (NHE3), is a critical mechanism for transepithelial movement of Na⁺, HCO₃⁻ and water in the basal and postprandial state [1,2]. In the mouse, NHE3 is highly expressed in all small intestinal segments (jejunum > duodenum > ileum) and the colon [3,4]. Functionally, luminal NHE3 is coupled to Cl⁻/HCO₃⁻ exchange (DRA, down-regulated in adenoma) [5] or H⁺/amino acid cotransport (PAT1, putative anion transporter 1) [6] that mediates electroneutral NaCl absorption. Genetic global deletion of NHE3 in mice (NHE3^−/−) causes diverse problems associated with NHE3 function in the intestine and the kidney [7,8]. NHE3^−/− mice suffer from mild diarrhea, mild metabolic acidosis, dilation of the intestinal tract and a high mortality rate subsequent to weaning, with only ∼30% of mice surviving into adulthood [9]. When fed a low NaCl diet, NHE3^−/− mice experience severe dehydration resulting in hypovolemia, body weight loss and death [10]. When NHE3^−/− mice are fed a high–NaCl diet (5% NaCl), the small intestine becomes severely swollen, consequently resulting in death within 2 days [11]. This is attributed to the osmotic effect exerted by the high NaCl levels in the intestine. Taken together, a high mortality rate, deletion of NHE3 in both intestine and kidney, and prenatal NHE3 deletion leading to potentially confounding developmental problems are significant drawbacks of previously generated NHE3 knockout models.

In 1985, two groups independently described diarrheal disorders that were classified as congenital sodium diarrhea (CSD) [12,13] and ascribed it to defective handling of Na⁺ and H⁺. The molecular origin remained unknown until Janecke et al. [14] identified recessive missense, splicing and truncation mutations in NHE3 that resulted in reduced basal surface expression and/or loss of basal transport activity when the mutants were overexpressed in NHE-deficient fibroblasts. Overall, ∼50 cases of CSD have been identified [15]. These patients are characterized by intrauterine onset of diarrhea resulting in polyhydramnios, and, after birth, severe dehydration, weight loss, hyponatremia and metabolic acidosis [16]. The electrolyte imbalances in the most severe cases are associated with a high risk of mortality.

Our previous studies indicated that renal NHE3 plays a minor role in acid–base regulation under baseline conditions [3], emphasizing the importance of the intestine. Therefore, in the present study we wanted to categorically determine the role of NHE3 in the intestine to acid–base regulation and test the hypothesis that lack of intestinal NHE3 in adulthood mimics CSD. As conditional non–inducible intestinal epithelial cell–specific NHE3 knockout mice and NHE3^−/− mice have high mortality [17], we generated tamoxifen–inducible intestinal epithelial cell–specific NHE3 knockout mice (NHE3^IEC-KO) to test our hypothesis. Our results in NHE3^IEC-KO mice demonstrate that intestinal NHE3 significantly contributes to (i) acid–base homeostasis, (ii) Na⁺/K⁺ homeostasis and (iii) fluid absorption, defects in which cannot be fully compensated for by altered renal function. Cumulatively, the phenotype of NHE3^IEC-KO mice resembles the phenotype of patients with CSD.

All animal experiments were conducted in accordance with the Guide for Care and Use of Laboratory Animals (National Institutes of Health, Bethesda, MD) and were approved by the local Institutional Animal Care and Use Committee, USF. All animal experiments were performed at the USF. Mice were housed under a 12:12-h light:dark cycle in standard rodent cages with free access to standard rodent chow (TD.2018, Envigo, Madison, WI) and tap water. Floxed NHE3 (NHE3^loxlox) female mice [18] were crossed with hemizygous male villin–CreER^T2 mice (Stock No: 020282, The Jackson Laboratories, Bar Harbor, ME) that express the Cre recombinase transgene specifically in intestinal epithelial cells. From the filial (F) 1 progeny, male mice heterozygous for the floxed NHE3 allele (NHE3^{lox/wild-type}) and villin–creER^T2 were bred to NHE3^loxlox mice to generate final breeder pairs in the F2 progeny. To generate male experimental intestinal epithelial–specific NHE3 knockout (NHE3^IEC-KO) mice and control (NHE3^loxlox) mice, female NHE3^loxlox mice were bred to male NHE3^IEC-KO mice. Mice were genotyped by polymerase chain reaction from genomic DNA isolated from ear punch as published previously [3,19]. NHE3 deletion was induced by application of tamoxifen (67 mg/kg), initially dissolved in 5% (v/v) of ethanol followed by adding 95% (v/v) of corn oil. Tamoxifen was administered via oral gavage (volume 1% of body weight) for five consecutive days to control and NHE3^IEC-KO mice.

Body weight, fluid and food intake were determined daily [20] before induction, during induction and up to 27 days after induction. One week before induction, plus at the end of the experimental period, blood was collected under brief isoflurane anesthesia from the retro-orbital plexus for analysis. Spontaneous voided urine was collected by reflex urination holding mice over a clean Petri dish [21].

Blood chemistry was determined by an OPTI^® CCA-TS2 blood gas analyzer [3] using an E-Cl Type cassette (OPTIMedical, Roswell, GA). Urine osmolality was measured using an Osmomat 3000 (Gonotec GmbH, Berlin, Germany). Urinary pH was determined using a pH electrode (9810BN, Fisher Scientific, Pittsburgh, PA). Plasma aldosterone was determined using a solid phase enzyme-linked immunosorbent assay (IBL America, Minneapolis, MN). Blood glucose was determined by a glucose meter (Contour, Bayer, Parsippany, NJ) from blood collected via tail snip.

Mice were gavaged with fluorescein isothiocyanate (FITC) conjugated-dextran (FD–4, molecular weight of 3–5 kDa, Sigma-Aldrich, 22 mg/ml, 1% of body weight) 2 weeks after tamoxifen induction [22]. Blood from a tail snip was collected in 10 µl minicaps^® (Hirschmann Laborgeräte, Eberstadt, Germany) at 1, 2 and 4 h after gavage. After centrifugation, samples were diluted 1:10 in 0.5 mol/L HEPES (pH 7.4) and FD–4 concentrations determined using a fluorescent microplate reader (Cytation 3, Biotek, Winooski, VT). Fecal water content in freshly defecated stool was calculated as the difference in fresh weight versus dry weight. Fecal dry weight was measured after 24 h at 80°C when mass was stable [23].

Freshly isolated epithelial cells were prepared by calcium (Ca²⁺) chelation [23]. Briefly, the small intestine was removed and flushed with 10 ml ddH₂O, The flushing solution was collected, and the pH determined. The intestine was everted, one end was ligated, and the everted small intestine was filled with Ca²⁺–free PBS containing 5 mmol/L ethylenediaminetetraacetic acid (EDTA), pH 7.4. The everted pieces were put in 50 ml tubes containing 40 ml Ca²⁺–free PBS+EDTA and incubated in a water bath at 37°C for 20 min. Every 5 min, the tubes were vigorously shaken to release the epithelial cells. After removal of sacs devoid of epithelial cells, the solution was split in half and the tubes were centrifuged. One pellet was prepared for mRNA analysis the other pellet for Western blotting.

In another set of mice, the small intestine and colon were collected from each animal and measured in length. In addition, small intestinal weight was determined before flushing with ice-cold PBS. Afterward, tissue was fixed overnight in 4% paraformaldehyde, paraffin embedded, and sectioned at 4–6 μm. After deparaffinization and rehydration, sections were stained with hematoxylin and eosin (H&E) for morphometry. In H&E-stained jejunal sections, villus length and crypt depth were measured in 12–15 well-oriented crypt-villus units per animal using a 40–1000× infinity research laboratory compound microscope with built in camera display. Measurements were performed by the software integrated measurement tools in TouchScope (version 2.5.6; Amscope, Irvine, CA) [24]. For cleaved caspase-3 staining, small intestine and colon sections were incubated in 3% H₂O₂ for 10 min, heated in Antigen Decloaker (Biocare Medical, Concord, CA) for 45 min, blocked with 20% normal goat serum (Vector Laboratories, Burlingame, CA), and then incubated with rabbit polyclonal cleaved caspase-3 antibody (1:100; Cell Signaling Technology, Beverly, MA) for 1 h at room temperature. Sections were then incubated with goat anti-rabbit biotinylated secondary antibody (1:200; Vector Laboratories) for 30 min at room temperature. Vectastain Elite ABC reagent (Vector Laboratories) was then applied, followed by diaminobenzidine as substrate. Sections were counterstained with hematoxylin, dehydrated and mounted with coverslips. Apoptosis in the intestinal epithelium was quantified using the number of cleaved caspase-3-positive cells 100 well–oriented contiguous crypts (jejunum and colon) and 50 well–oriented villi (jejunum) per animal [24]. All counting was performed by an examiner (J.A.D.R) blinded to sample identity.

H&E-stained small intestine and colon sections were graded by using the following scheme: 0 = normal colonic mucosa with no active inflammation; 1 = mild active inflammation; 2 = moderate active inflammation; 3 = severe active inflammation. The highest score seen in sections was reported for each mouse. All scoring was performed by a pathologist (M.T.) blinded to sample identity.

Tissue preparation, sectioning and labeling were performed as previously described [21,23]. Goat anti-rabbit Texas Red conjugated secondary antibody (TI–1000, Vector Laboratories, Burlingame, CA) was used for visualization of NHE3 labeling (SPC–400, StressMarq Biosciences, Cadboro Bay, Victoria, British Columbia, Canada). Sections were mounted using hardset antifade mounting medium with DAPI (H–1500, Vector Laboratories). An Olympus (Center Valley, PA) FV1000 MPE Multiphoton Laser Scanning Microscope with a 40× UPLFL objective lens (numerical aperture: 0.75) was used for imaging of labeled sections.

Total RNA from freshly isolated intestinal epithelial cells and whole kidney was isolated from control and NHE3^IEC-KO mice and homogenized using a Tissue Tearor (Bartlesville, OK) and QIAshredder (Qiagen, Valencia, CA) according to the manufacturer’s instructions. The extracted RNA was further purified using RNeasy Plus Mini Kit (Qiagen). cDNA was made by reverse transcribing total RNA using a QuantiTect reverse transcription kit (Qiagen) according to the manufacturer’s instructions. Quantitative real-time PCR was performed using TaqMan™ Universal PCR Master Mix (Fisher Scientific) in a QuantStudio 3 (Applied Biosystems, Foster City, CA). Template concentration was 12.5 ng reverse-transcribed RNA/25–µl reaction (performed in triplicates) and was used in conjunction with primer pairs specific for NHE3 (Mm01352478_g1, efficiency 98%) and with GAPDH (Mm99999915_g1) as a reference gene. Data analysis used the ΔΔC_t method, i.e. cycle thresholds (C_t), were normalized to GAPDH expression and compared with control [25].

Renal tissue and acutely isolated intestinal epithelial cells were homogenized in buffer containing protease inhibitor (250 mmol/L sucrose, 10 mmol/L triethanolamine, Sigma-Aldrich, St. Louis, MO; Fisher Scientific, respectively) and phosphatase inhibitor (Fisher Scientific). The homogenate was centrifuged at 1000 g for 15 min, and the resultant supernatant further centrifuged at 17,000 g for 30 min. Pellets were resuspended and used for Western blotting. For ZO–1, the supernatant of the 1000 g spin was used for Western blotting. Equal lane loading (15 µg for kidney and 20 µg for intestine) was achieved using a Bio-Rad DC Protein assay (Bio-Rad Laboratories, Richmond, CA). Samples were resolved on NuPAGE 4–12% Bis-Tris gels in MOPS buffer or NuPAGE 3-8% Tris-Acetate gels (for occludin and ZO-1, Fisher Scientific). Gel proteins were transferred to PVDF membranes and immunoblotted with antibodies against NHE3 (dilution 1:500, rabbit, MilliporeSigma, Burlington, MA), occludin (dilution 1:500, mouse, Invitrogen, Carlsbad, CA), ZO–1 (dilution 1:250, rabbit, Invitrogen) and ß–actin (dilution 1:30,000, mouse, Sigma-Aldrich; dilution 1:5,000, rabbit, Proteintech, Rosemont, IL). Detection was performed with secondary antibodies against rabbit or mouse and detected with an Odyssey^® CLx (LI-COR Biosciences, Lincoln, NE). Densitometric analysis was performed by using Image Studio Lite (LI–COR Biosciences).

Data are expressed as mean ± S.E.M. Unpaired Student’s t test or Mann–Whitney U test was performed as appropriate to analyze for statistical differences between groups. One-way analysis of variance (ANOVA) or two–way mixed effects ANOVA followed by two–stage linear step–up procedure of Benjamini, Krieger and Yekutieli or Tukey multiple comparison tests as indicated were used to test for significant differences between genotype and experimental conditions. Significance was considered at P < 0.05.

NHE3^lEC-KO mice were born at predicted Mendelian frequencies, appeared grossly indistinguishable from control mice, and developed normally. At the end of the experimental period (27 days after tamoxifen administration) qualitative immunofluorescence identified NHE3 along the villi of the small intestine in control mice (Figure 1A). In contrast, this labeling was completely absent in NHE3^lEC-KO mice (Figure 1B). Staining intensity and NHE3 localization in the kidney were not different between control and NHE3^lEC-KO mice (Figure 1C,D). NHE3 messenger RNA (mRNA) and protein levels were virtually undetectable in acutely isolated intestinal epithelial cells (Figure 1E,F). Renal NHE3 protein levels were ∼2-fold higher in NHE3^lEC-KO compared with control mice after tamoxifen administration (Figure 1F). Of note, survival analysis identified that 3 weeks after tamoxifen administration NHE3^lEC-KO mice have a ∼25% mortality rate (Figure 1G).

Fecal samples from NHE3^IEC-KO mice showed visible signs of diarrhea (Figure 2A), with ∼1.45–fold higher fecal water content compared with control mice (Figure 2B). Around 35% of NHE3^IEC-KO mice had visible watery diarrhea. Morphologically, all NHE3^IEC-KO mice had dilated small intestines, cecum and colon (Figure 2C). This was associated with significantly higher intestinal permeability, demonstrated by ∼1.9-fold higher FD–4 plasma concentrations 4 h after administration (Figure 2D). Analysis of tight junction protein expression showed that occludin in the small intestine was decreased by ∼28% in NHE3^IEC-KO compared with control mice (Figure 3A). In contrast, ZO–1 expression was not significantly different between genotypes (Figure 3A). Histologically, no observable difference in inflammation was detected in the small intestine between genotypes. Examination of colon sections (Figure 3B and Supplementary Figure S1) showed that 44% of NHE3^IEC-KO mice had pathological findings: 11% showed patchy cryptitis with occasional crypt abscesses and 33% showed focal/patchy mild cryptitis. Of note, all the control mice and 56% of NHE3^IEC-KO mice had no pathological findings. Cleaved caspase-3 staining, a specific marker for apoptosis, was increased in crypts and villi (∼2.4– and ∼1.6-fold, respectively) of the small intestine in NHE3^lEC-KO compared with control mice (Figure 3C), indicating increased apoptosis. In colonic crypts, apoptosis was also increased ∼1.9–fold in NHE3^lEC-KO compared with control mice (Figure 3C).

Non-flushed small intestines of NHE3^IEC-KO mice were >2-fold heavier compared with control mice (Figure 4A). Similarly, the weight–to–length ratio of the small intestine was ∼2.5–fold higher (Figure 4B), a finding not related to the overall small intestinal length, which was not significantly different between genotypes (Figure 4C). Colons from NHE3^IEC-KO mice were ∼1.6 cm longer (∼25%) compared with control mice (Figure 4C) 27 days after tamoxifen administration. Villi in proximal small intestine from NHE3^IEC-KO mice were ∼100 µm longer compared with control mice (Figure 4D). In contrast, villus length in the distal small intestine, and crypt depth in the proximal and distal small intestine were not significantly different between genotypes (Figure 4D). Colonic flushes from control mice were ∼0.9 pH units lower compared with small intestine (Figure 4E). In contrast, the pH of small intestinal flushes from NHE3^IEC-KO mice were ∼0.8 pH units higher than controls and colonic flushes were ∼1.6 pH units higher (Figure 4E). In contrast with controls, no significant differences in flush pH between small intestine and colon were observed in NHE3^IEC-KO mice.

Control (n = 19) and NHE3^lEC-KO (n = 15) mice had no differences in body weight before tamoxifen (baseline) administration (31 ± 1 vs 29 ± 1 g, NS). In control mice, body weight did not change during or after tamoxifen administration (Figure 5A). In contrast, NHE3^lEC-KO had a significant loss of body weight during initial tamoxifen administration (maximum −6 ± 1% vs baseline), which was followed by a significant increase in body weight compared with baseline (Figure 5A). The immensely fluid-filled cecum and other parts of the intestine (Figure 2C) might have, in conjunction with the increased fluid intake (Figure 5C), contributed to the body weight gain. Food intake was not significantly different between genotypes at baseline (Figure 5B). In control mice, tamoxifen administration caused a transient decrease in food intake (maximum −25%, P < 0.05 vs baseline) followed by a return to baseline levels. NHE3^lEC-KO mice showed a significantly greater decrease in food intake (maximum −45%, P < 0.05 vs baseline) followed by a return to baseline levels (Figure 5B). Fluid intake was not significantly different between genotypes at baseline (Figure 5C) and remained similar in control mice during the remainder of the experimental period. In NHE3^lEC-KO mice, 3 days after the end of tamoxifen administration, fluid intake increased significantly compared with baseline and remained elevated for the remainder of the experimental period (Figure 5C). Despite increased fluid intake, relative to controls, NHE3^lEC-KO mice had a significantly higher hematocrit (50 ± 2 vs 42 ± 1%, P < 0.05) and blood urea nitrogen (BUN) was ∼2.8-fold higher (76 ± 16 vs 27 ± 1, P < 0.05), together suggesting dehydration and volume depletion.

Baseline blood pH, HCO₃⁻, base excess, plasma Na⁺, K⁺ and aldosterone as well as blood glucose were not significantly different between genotypes (Figure 6A–G). At the end of the experimental period, blood pH (Figure 7A), base excess (Figure 7C) and plasma K⁺ (Figure 7E) were not affected in control mice; however, significant increases in blood HCO₃⁻ (Figure 7B) and plasma Na⁺ (Figure 7D) were observed; aldosterone (Figure 7F) and blood glucose (Figure 7G) showed a significant decrease. After tamoxifen administration, blood pH (Δ –0.16 ± 0.04 pH units; Figure 7A), HCO₃⁻ (Δ –4 ± 1 mmol/L; Figure 7B) and base excess (Δ –7 ± 2 mmol/L; Figure 7C) significantly decreased in NHE3^lEC-KO mice. Consistent with impaired water and Na⁺ handling in NHE3^lEC-KO mice, plasma Na⁺ decreased (Δ –10 ± 3 mmol/L; Figure 7D). Potentially due to the body’s effort to defend against metabolic acidosis in combination with hypovolemia, plasma K⁺ significantly increased (Δ 1.9 ± 0.5 mmol/L; Figure 7E). Consistent with an attempt to compensate for hypovolemia and hyponatremia, there were significantly increased plasma aldosterone levels (Δ 3085 ± 551 pg/ml; Figure 7F) in NHE3^lEC-KO mice. The elevated plasma aldosterone levels possibly contributed to a ∼4.6–fold higher ratio of urinary K⁺/Na⁺ in NHE3^lEC-KO compared to control mice (11.9 ± 2.5 vs 2.6 ± 0.3, P < 0.05). Mice with knockout of NHE3 along the renal tubule and collecting duct have a significantly elevated urinary pH and lower urine osmolality [3]. In contrast, urinary pH (5.97 ± 0.1 vs 5.95 ± 0.1, NS) and urine osmolality (2133 ± 212 vs 2220 ± 206 mmol/kg, NS) were not significantly different between NHE3^lEC-KO mice and control mice.

CSD is a disorder with high mortality that is difficult to treat. Determining the role of NHE3 in CSD has been hampered by the lack of a suitable mouse model, with global deletion of NHE3 resulting in very high mortality rates [9,17]. Additionally, prenatal deletion of NHE3 affects intestinal development and epithelial barrier function [26], hampering its usefulness in studies during adulthood. To navigate this problem, we generated tamoxifen–inducible intestinal epithelial cell–specific NHE3 knockout mice (NHE3^IEC-KO) by interbreeding NHE3^loxlox mice with villin-CreER^T2 mice. The phenotype of NHE3^IEC-KO mice clearly demonstrates the important homeostatic role of intestinal NHE3. When knockout is induced in adulthood, NHE3^IEC-KO mice have features that are consistent with patients suffering from CSD: alkaline diarrhea, hyponatremia and metabolic acidosis, alongside activation of the renin–angiotensin–aldosterone system and volume contraction.

NHE3^IEC-KO mice had a complete absence of NHE3 mRNA and protein expression in the small intestine. Although villin is expressed in the mouse kidney [27], NHE3 mRNA expression in kidneys from NHE3^IEC-KO mice was not significantly different from controls. In contrast, renal NHE3 protein expression was significantly greater in NHE3^IEC-KO mice. It is possible that this is a compensatory response. For example metabolic acidosis (which NHE3^IEC-KO mice have, see below) increases renal NHE3 protein expression [28,29] without affecting NHE3 mRNA expression [28]. Why renal NHE3 is not affected remains elusive, but may relate to certain regions of the mouse villin gene being vulnerable to positional effects. The site of chromosomal integration determines transgene expression, which can be altered by neighboring regulatory regions, presumably acting on chromatin conformation. In an unrelated study, we used a non–inducible villin–Cre mouse model and we also did not see differences in renal gene expression between genotypes [30]. We previously attempted to generate constitutive non–inducible intestinal epithelial cell–specific NHE3 knockout mice. This model represented an extreme form of CSD, with only one mouse out of >50 offspring surviving ∼2 weeks and none survived into adulthood [17]. NHE3^−/− mice have a 30% survival rate at the age of 30 weeks [9]. NHE3^IEC-KO mice had a ∼27% mortality rate 3 weeks after tamoxifen administration; these mice developed progressive body weight loss and lethargy, requiring them to be euthanized.

NHE3^IEC-KO mice had distended abdomens, fluid-filled intestinal loops and alkaline diarrhea. This is consistent with NHE3^−/− mice, which have alkaline diarrhea, increased intestinal weight and increased fluid content [8,31], but importantly NHE3^IEC-KO mice mimic the clinical features of CSD patients [32,33]. The more alkaline pH of the small intestinal and colon contents in NHE3^IEC-KO mice is consistent with reduced H⁺ secretion and impaired buffering of secreted HCO₃⁻. NHE3^−/− mice are more susceptible to mucosal injury [34], with dilated apical junction complexes in the small intestine of NHE3^−/− mice [34]. Although this previous observation has, at least in part, been ascribed to developmental effects in NHE3^−/− mice, plasma levels of FD–4, a marker of intestinal permeability, is significantly increased 4 h after administration in adult NHE3^IEC-KO mice. Thus, our data are indicative of a continuous important role of NHE3 in the maintenance of epithelial integrity and cell–cell interactions. Our data point toward reduced occludin expression in NHE3^IEC-KO mice contributing to increased intestinal permeability. Reduced occludin expression is a major contributor to increased macromolecule flux across the tight junction barrier of the intestine [35]. In addition, changes in pH, as found in NHE3^IEC-KO mice, were found to affect inter–molecular forces and adhesion strength of tight junction proteins [36]. Additionally, NHE3 is important for intestinal microbial composition and NHE3^−/− mice develop spontaneous colitis possibly because of microbial dysbiosis [37–39]. A subset of patients with mutations in NHE3 [16] develop inflammatory bowel disease (IBD). Intestinal epithelial cell apoptosis and loss of barrier function are characteristics of IBD [40,41]; consistent with this, NHE3^IEC-KO mice show increased intestinal apoptosis. Along those lines, our data also indicate that a subset of NHE3^IEC-KO mice develop mild to moderate active colitis. Tenapanor, an orally non–absorbable drug that inhibits NHE3, was just recently approved by the FDA for the treatment of IBD with predominant constipation symptoms [42]. Although the long-term effects of NHE3 inhibition in humans for the course of IBD is unknown, Tenapanor does not appear to cause altered epithelial integrity. This suggests that the altered epithelial integrity in NHE3^IEC-KO and NHE3^−/− mice is not due to a lack of Na⁺/H⁺ exchange, but more a direct role of NHE3 as a scaffolding protein.

Classically, Na⁺/H⁺ exchange in the proximal tubule of the kidneys is described as the primary determinant of acid–base homeostasis [43]. However, deletion of NHE3 throughout the renal tubule [3,44], or NHE3 knockout in the S1/2 segment of the proximal tubule, does not cause an acid–base phenotype [45]. In contrast, NHE3^IEC-KO mice show a clear metabolic acidosis, lower HCO₃⁻ levels and lower base excess. Thus, our model unravels the importance of intestinal NHE3 in the regulation of acid–base homeostasis, with the kidney unable to compensate for the magnitude of acid–base disturbance caused by lack of NHE3 in the intestine. Intestinal NHE3 therefore has a physiological role in fending off metabolic acidosis, as emphasized by the significantly increased NHE3 activity and protein in response to NH₄Cl-induced metabolic acidosis [46].

Hyponatremic dehydration is a common finding in children with diarrhea [47], consistent with a role of intestinal NHE3 in transepithelial Na⁺ and water absorption [48]. Supporting this role, NHE3^IEC-KO mice show a significantly higher hematocrit and BUN levels, both indirect signs of hypovolemia (we cannot completely exclude a role of reduced glomerular filtration rate to elevate BUN). Subsequent hypovolemic–mediated arginine–vasopressin secretion, coupled to the significantly greater fluid intake to compensate for intestinal fluid loss, is the likely cause of hyponatremia in NHE3^IEC-KO mice. In addition, hypovolemia–mediated increases in ANGII and increased proximal tubule Na⁺ reabsorption would limit distal Na⁺ delivery, resulting in reduced Na⁺/K⁺ exchange. This can account for the hyperkalemia in the NHE3^IEC-KO mice, although it is conceivable that the drastic metabolic acidosis contributed to the hyperkalemia. The drastic increase in plasma aldosterone levels in NHE3^IEC-KO mice can be explained by the volume contraction (activation of renin–angiotensin system) and the hyperkalemia. Hyperkalemia and hyperaldosteronism are common features of patients with CSD [14,33,49].

Another point of discussion from our studies is related to changes in intestinal architecture. In terms of unchanged intestinal but increased colon length, our data from NHE3^IEC-KO mice are consistent with NHE3^−/− mice [31]. The latter is possibly a compensatory response to the chronic exposure to an increased volume/electrolyte amount. A similar situation is observed in a mouse model of short bowel syndrome, which develop significantly longer villi and deeper crypts compared with Sham mice [50]. Of note, longer villi were not associated with a better outcome after surgery and could be considered maladaptive. Patients with CSD show mild villus atrophy with hyperplastic crypts [49,51]. In contrast, even though pathophysiological changes were observed in NHE3^−/− mice, no histopathological changes in the intestine were described, including no changes in apoptosis [9,34]. Of note, genes involved in apoptosis were significantly up–regulated in NHE3^−/− mice [34]. The significantly longer villi in NHE3^IEC-KO mice compared with control mice is indicative of increased mucosal mass. This is supported by a ∼6–fold greater yield of intestinal epithelial cells per unit length of intestinal segment in NHE3^−/− mice compared with wild–type mice [34].

In summary, we have generated an inducible intestinal epithelial cell–specific NHE3 knockout mouse model. Studies in these NHE3^IEC-KO mice provide direct in vivo evidence that NHE3 has several important homeostatic functions, including but not limited to acid–base regulation, electrolyte balance and water homeostasis. This new mouse model mimics and explains the phenotype of individuals carrying mutations in SLC9A3. NHE3^IEC-KO mice are an essential new model to further examine the role of intestinal NHE3 in various physiological and pathophysiological mechanisms.

• The intestine has an important role in acid–base regulation, volume homeostasis and electrolyte transport, some of them directly related to the functional presence of NHE3 in the intestinal mucosa. In patients with mutations in NHE3 congenital sodium diarrhea develops.

• Mice with whole body and constitutive intestinal epithelial cell–specific NHE3 knockout show high mortality rates. We developed a new, tamoxifen–inducible, intestinal epithelial cell–specific NHE3 knockout mouse model that allows for studying the disease in adulthood. The model closely resembles the features seen in patients with congenital sodium diarrhea with a low mortality rate.

• Given that a subset of patients with congenital sodium diarrhea develop inflammatory bowel disease and the recent development of an intestinal–specific NHE3 inhibitor for the treatment of constipation in inflammatory bowel disease and hyperphosphatemia in chronic kidney disease, this new model will help to gain a better understanding of the pathophysiology and pharmacology of NHE3.

The authors declare that there are no competing interests associated with the manuscript.

This work was supported by the National Institute of Diabetes and Digestive and Kidney Diseases [grant number 1R01DK110621 (to T.R.)]; American Heart Association Transformational Research Award [grant number 19TPA34850116 (to T.R.)]; the LeDucq Foundation (to R.A.F.); the Danish Medical Research Council; the Novo Nordisk Foundation; American Heart Association predoctoral fellowship [grant number 18PRE33990236 (to J.X.)]; and L.T. by a postdoctoral fellowship [grant number 19POST34400026].

T.R. designed the work. J.X., L.T., M.T., A.V., R.A.F., J.A.D.R. and T.R. contributed to the acquisition, analysis or interpretation of data for the work. T.R. drafted the work; J.X., L.T., R.A.F. and J.A.D.R. revised it critically for important intellectual content. J.X., L.T., M.T., A.V., R.A.F., J.A.D.R. and T.R. approved the final version of the manuscript.

CSD

congenital sodium diarrhea

DRA

Cl⁻/HCO₃⁻ exchanger

EDTA

ethylenediaminetetraacetic acid

NHE3

Na⁺/H⁺ exchanger isoform 3

PBS

phosphate–buffered saline