Intestinal brush-border Na⁺/H⁺ exchanger-3 drives H⁺-coupled iron absorption in the mouse

Iron homeostasis is of significant public health interest because of the prevalence of iron deficiency and iron-overload disorders. Whole body iron homeostasis is achieved by regulating iron absorption. We have identified a role for the intestinal brush-border Na⁺/H⁺ exchanger-3 in generating the proton-motive force driving iron uptake via the proton-coupled divalent metal-ion transporter-1 and, thereby, energizing intestinal iron absorption.

Abstract

Divalent metal-ion transporter-1 (DMT1), the principal mechanism by which nonheme iron is taken up at the intestinal brush border, is energized by the H⁺-electrochemical potential gradient. The provenance of the H⁺ gradient in vivo is unknown, so we have explored a role for brush-border Na⁺/H⁺ exchanger (NHE) isoforms by examining iron homeostasis and intestinal iron handling in mice lacking NHE2 or NHE3. We observed modestly depleted liver iron stores in NHE2-null (NHE2^−/−) mice stressed on a low-iron diet but no change in hematological or blood iron variables or the expression of genes associated with iron metabolism compared with wild-type mice. Ablation of NHE3 strongly depleted liver iron stores, regardless of diet. We observed decreases in blood iron variables but no overt anemia in NHE3-null (NHE3^−/−) mice on a low-iron diet. Intestinal expression of DMT1, the apical surface ferrireductase cytochrome b reductase-1, and the basolateral iron exporter ferroportin was upregulated in NHE3^−/− mice, and expression of liver Hamp1 (hepcidin) was suppressed compared with wild-type mice. Absorption of ⁵⁹Fe from an oral dose was substantially impaired in NHE3^−/− compared with wild-type mice. Our data point to an important role for NHE3 in generating the H⁺ gradient that drives DMT1-mediated iron uptake at the intestinal brush border.

NEW AND NOTEWORTHY

Iron homeostasis is of significant public health interest because of the prevalence of iron deficiency and iron-overload disorders. Whole body iron homeostasis is achieved by regulating iron absorption. We have identified a role for the intestinal brush-border Na⁺/H⁺ exchanger-3 in generating the proton-motive force driving iron uptake via the proton-coupled divalent metal-ion transporter-1 and, thereby, energizing intestinal iron absorption.

iron deficiency, the most prevalent micronutrient deficiency worldwide, can result in iron-deficiency anemia and neurological and developmental disorders in children (4, 5, 32). Conversely, iron toxicity associated with iron-overload disorders such as hereditary hemochromatosis and thalassemia major can cause liver cirrhosis, cardiomyopathy, and endocrine disorders (17, 27). Since there exists no regulated mechanism for the excretion of iron, whole body iron homeostasis is achieved by regulating iron absorption (19).

Divalent metal-ion transporter-1 (DMT1) (see review in Ref. 54) is the principal mechanism by which nonheme iron is taken up at the intestinal brush border (53). This widely expressed iron transporter also plays a critical role in erythroid iron assimilation (24) via the transferrin cycle and functions similarly in many other cell types. We have shown that DMT1-mediated iron transport in vitro is functionally coupled with protons (25, 39); however, the provenance of the H⁺-electrochemical potential gradient driving DMT1-mediated iron uptake at the intestinal brush border in vivo is unknown.

The mucosal cell surface in the duodenum and proximal jejunum remains moderately acidic, despite the alkalization of the luminal contents. This “acidic microclimate” may reach pH 6.0 or lower in human proximal jejunum in vivo and in the small intestine of other mammals (35, 41, 42, 50, 55). We propose that the intestinal brush-border acidic microclimate, thought to rely on the activity of Na⁺/H⁺ exchanger (NHE) isoforms (55), plays a crucial role in the absorption of iron.

Gastrointestinal NHE isoforms play essential roles in pH balance and Na⁺ and fluid absorption (13, 61). Two isoforms are expressed at the brush border in the adult mammalian intestine: NHE2 is localized to the apical plasma membrane and is found in enterocytes at the absorptive surface and in the crypts (7, 13, 28, 40), and NHE3 is expressed predominantly in surface enterocytes, in which it is localized to the apical plasma membrane and apical recycling endosomes (6, 13, 28, 49, 61).

We have not included in our study the intestinal isoforms NHE8 and NHE1. NHE8 is expressed in the brush-border membrane in the suckling mammal (60) prior to the ontogenic induction of NHE2 and NHE3 around weaning (9); however, NHE8 is expected to play only a minor role in acidifying the surface microclimate in the adult. Whereas NHE1 is ubiquitously expressed (47), it is exclusively localized to the basolateral membrane (6, 59). Other NHE isoforms expressed in the intestine are confined to intracellular organelles (3, 12, 46, 47) and, therefore, were not considered in this study.

We tested the hypothesis that the intestinal brush-border NHE isoforms NHE2 and NHE3 promote iron absorption by examining iron homeostasis and intestinal iron handling in mouse models lacking NHE2 and NHE3. Mice in which the SLC9A2 gene coding NHE2 is globally inactivated [i.e., NHE2-null (NHE2^−/−) mice] exhibit no outwardly appearing disease phenotype (51) but display morphological changes to gastric mucosa (51), colonic crypts (23), and pituitary folliculostellate cell canaliculi (44). Whereas NHE2^−/− mice exhibit compromised recovery of intestinal barrier function after ischemic injury (45), they do not exhibit any morphological change in the small intestine. Moreover, ablation of NHE2 in NHE3-null (NHE3^−/−) mice does not exacerbate the phenotype of the NHE3^−/− mouse (33).

Knockout of the SLC9A3 gene coding NHE3 (i.e., NHE3^−/−) in mice produces defects in acid-base balance and Na⁺ and fluid absorption (associated with a mild secretory diarrhea) (52), revealing NHE3 to be the predominant NHE isoform at the intestinal brush border (21, 61).

EXPERIMENTAL PROCEDURES

Reagents and media.

Reagents were obtained from Sigma-Aldrich (St. Louis, MO) or Research Products International (Prospect, IL) unless otherwise indicated.

Expression of human DMT1 in Xenopus oocytes.

We performed laparotomy and ovariectomy on adult female Xenopus laevis frogs (Nasco, Fort Atkinson, WI) under 3-aminoethylbenzoate methanesulfonate anesthesia [0.1% (wt/vol) in 1:1 water-ice, by immersion] following a protocol approved by the University of Cincinnati Institutional Animal Care and Use Committee. Ovarian tissue was isolated and treated with collagenase A (Roche Diagnostics, Indianapolis, IN), and oocytes were isolated and stored at 17°C in modified Barth's medium, as described elsewhere (36). We expressed in Xenopus oocytes the 1A/IRE⁺ isoform of human DMT1, the product of the human SLC11A2 gene, as described elsewhere (30). Briefly, defolliculate stage V–VI oocytes were injected with ∼50 ng of DMT1 RNA and incubated for 6 days before being used in functional assays.

Radiotracer assays in oocytes expressing human DMT1.

Oocytes were incubated at room temperature (23°C) in transport medium (in mM: 100 NaCl, 1 KCl, 0.6 CaCl₂, 1 MgCl₂, and 1 l-ascorbic acid) and buffered using 0–5 mM 2-(N-morpholino)ethanesulfonic acid (MES) plus 0–5 mM N′,N′-diethylpiperazine (DEPP) (both buffers from GFS Chemicals, Columbus, OH) to obtain pH 5.4–7.5. We used ⁵⁵Fe (as FeCl₃ added to the ascorbate-containing medium) at final specific activity of 0.24 GBq/mg (Perkin-Elmer Life Science Products, Boston, MA). Radiotracer metal ion uptake was measured over 10 min, i.e., within the linear portion of the time course of ⁵⁵Fe²⁺ uptake (38). We terminated radiotracer uptake by rapidly washing the oocytes three times in radiotracer-free ice-cold pH 5.5 transport medium. Oocytes were then solubilized using 5% (wt/vol) sodium dodecyl sulfate, and radiotracer content was assayed by liquid scintillation counting using ScintiSafe 30% liquid scintillation cocktail (Fisher Scientific, Pittsburgh, PA). Transport rates (v) were fit by a four-parameter logistic function (Eq. 1) in which V_min and V_max are the minimum and maximum rates, K_0.5^H is the H⁺ concentration at which transport was half-maximal, and a is the Hill slope.

$$v=V_{\min }+\left\{\frac{\left(V_{\max }-V_{\min }\right)}{1+{\left(\frac{\left[{\text{H}}^{+}\right]}{K_{0.5}^{\text{H}}}\right)}^a}\right\}$$ (1)

Mouse models.

We studied iron metabolism in adult FVB/N mice lacking NHE2 or NHE3 following a protocol approved by the University of Cincinnati Institutional Animal Care and Use Committee. Generation of the NHE2^−/− and NHE3^−/− mouse models is described elsewhere (51, 52). At 21–28 days of age, DNA was obtained from tail clips treated with DirectPCR lysis reagent (Viagen Biotech, Los Angeles, CA) according to the manufacturer's instructions, and mice were genotyped. We analyzed PCR products using primers, as described elsewhere (8, 31). Mice were fed a standard (“normal”) diet containing 350 ppm Fe (catalog no. 7922 NIH-07, Harlan Laboratories, Indianapolis, IN), except where indicated as having been fed a low-iron (“low-Fe”) diet containing 2–6 ppm Fe (catalog no. TD.80396, Harlan-Teklad, Madison, WI). In the studies described in Figs. 3 and 5, we collected blood (via cardiac puncture) and tissues from mice subjected to isoflurane anesthesia (by inhalation, to effect) and then euthanized the animals by exsanguination. Secondary methods of euthanization comprised cutting of the diaphragm and removal of one or more vital organs. In all other experiments, mice were euthanized by CO₂ asphyxiation followed by removal of a vital organ.

Fig. 3.

Iron metabolism and homeostasis in male NHE2^−/− mice fed a low-Fe diet. Male NHE2^+/+ and NHE2^−/− mice were maintained for 6 wk on a low-Fe diet prior to analysis at ∼19 wk of age. A–H: hematological and blood iron variables in NHE2^+/+ and NHE2^−/− mice (n = 4–9): hematocrit, red blood cell count, mean corpuscular volume, hemoglobin (Hgb) concentration ([Hgb]), mean corpuscular hemoglobin, mean corpuscular hemoglobin concentration, serum iron (Fe), and transferrin (Tf) saturation. I–L: relative fold expression (geometric mean and SD, n = 5–7) of intestinal (I–K) and hepatic (L) iron-related genes: DMT1, cytochrome b reductase-1 (Cybrd1), ferroportin (Fpn), and hepcidin (Hamp1). Student's t-tests revealed no difference between NHE2^+/+ and NHE2^−/− mice in any of the variables (0.11 ≤ P ≤ 0.99).

Fig. 5.

Iron metabolism and homeostasis in male NHE3^−/− mice fed a low-Fe diet. Male NHE3^+/+ and NHE3^−/− mice were maintained for 6 wk on a low-Fe diet prior to analysis at ∼16 wk of age. A–G: hematological and blood iron variables (n = 7–14): hematocrit, red blood cell count, mean corpuscular volume, hemoglobin (Hgb) concentration ([Hgb]), mean corpuscular hemoglobin, mean corpuscular hemoglobin concentration, serum iron (Fe), and transferrin (Tf) saturation. I–L: relative fold expression (geometric mean and SD, n = 4–6) of intestinal and hepatic iron-related genes. Student's t-test revealed no difference between NHE3^+/+ and NHE3^−/− mice for hematocrit, mean corpuscular volume, hemoglobin concentration, or mean corpuscular hemoglobin (P ≥ 0.088, except ^bP = 0.039 ruled a false discovery rate). NHE3^−/− differed from NHE3^+/+ for the following variables: red blood cell count (^aP = 0.003), mean corpuscular hemoglobin concentration (^cP = 0.006), serum Fe (^dP = 0.001), Tf saturation (^eP ≤ 0.001), and all genes tested (P < 0.001).

Blood and tissue analyses.

Standard automated complete blood count (CBC) was performed by Antech (Chicago, IL). Serum iron (SI) and unsaturated iron-binding capacity (UIBC) were assayed using the SL and UIBC assay kits along with calibration standards (Sekisui Diagnostics, Charlottetown, PE, Canada) according to manufacturer's protocols, and transferrin saturation (Tf_sat, %) was computed according to Eq. 2

$${\text{Tf}}_{\text{sat}}=\left(\frac{\text{SI}}{\text{SI}+\text{UIBC}}\right)\times 100$$ (2)

Liver nonheme iron content was determined using a standard acid-digestion, chromogen-based colorimetric assay, as described elsewhere (58).

Quantitative PCR analyses.

Freshly isolated small intestine was rinsed and flushed with ice-cold saline. We collected an enterocyte-enriched preparation by lightly scraping the luminal surface of the proximal 1–5 cm of small intestine (i.e., duodenum) with the edge of a glass microscope slide. Enterocytes and liver tissue were collected into TRIzol reagent (Life Technologies, Carlsbad, CA), homogenized, and frozen at −80°C prior to their use in quantitative real-time PCR (qPCR) analyses. We performed reverse transcription by using 50 μg/ml oligo(dT)20 primer and reverse transcriptase (Qiagen, Valencia, CA) according to the manufacturer's instructions. For determination of sample cDNA concentrations, absorbance at 260 nm was measured using the NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA). We amplified 200 ng of sample cDNA with Fast SYBR Green Real-Time PCR master mix (Applied Biosystems, Life Technologies, Grand Island, NY) in a final volume of 20 μl by using the Step One PCR system (Applied Biosystems). Gene expression was determined using the cycle threshold (ΔΔC_T) method (34), with glyceraldehyde 3-phosphate dehydrogenase (GAPDH) mRNA in each tissue sampled as reference, and primers described in Table 1 to determine gene expression. GAPDH expression (C_T) did not differ between wild-type (NHE2^+/+) and NHE2^−/− enterocytes (rank-sum test, P = 0.43) or liver (P = 0.64) and did not differ between wild-type (NHE3^+/+) and NHE3^−/− enterocytes (P = 0.082) or liver (P = 0.22; data not shown). qPCR data for cytochrome b reductase-1 (Cybrd1), DMT1, ferroportin (Fpn), and hepcidin (Hamp1) were normalized by the mean −ΔΔC_T in NHE2^+/+ (see Fig. 3, I–L) and NHE3^+/+ (see Fig. 5, I–L) mice and expressed as 2^−ΔΔC_T, i.e., relative fold expression.

Table 1.

Primers used in real-time quantitative PCR assays of isolated mouse tissues

		Primer
Gene	NCBI Reference Sequence	Forward	Reverse	Reference No.
Cybrd1	AF_354666	5′-GCA GCG GGC TCG AGT TTA-3′	5′-TTC CAG GTC CAT GGC AGT CT-3′	43
DMT1	NM_008732	5′-CGC TCG GTA AGC ATC TCG AA-3′	5′-TGT TGC CAC CGC TGG TAT CT-3′	43
Fpn	NM_016917	5′-CAT TGC TGC TAG AAT CGG TCT T-3′	5′-GCA ACT GTG TCA CCG TCA AAT-3′	43
Hamp1	AF_503444	5′-AAG CAG GGC AGA CAT TGC GAT-3′	5′-CAG GAT GTG GCT CTA GGC TAT GT-3′	43
GAPDH	NM_008084	5′-CAT GGC CTT CCG TGT TCC TA-3′	5′-CCT GCT TCA CCA CCT TCT TGA-3′	18

Cybrd1, cytochrome b reductase-1; DMT1, divalent metal-ion transporter-1; Fpn, ferroportin; Hamp1, hepcidin; NCBI, National Center for Biotechnology Information.

Absorption of ⁵⁹Fe in NHE2^−/− and NHE3^−/− mice.

We measured radiotracer metal absorption in female control, NHE2^−/−, and NHE3^−/− mice at ∼5 mo of age. ⁵⁹Fe (0.1 μCi/g body wt added as FeSO₄ in 1 mM NaCl, 1 mM l-ascorbic acid, and 1 mM l-glutamine) was administered to conscious mice via oral intragastric gavage following an overnight fast.

Blood was collected into heparinized hematocrit tubes by tail-nick incision at 0.5, 1, 2, and 4 h. Blood samples were centrifuged for measurement of hematocrit and then expelled into 20-ml scintillation vials containing 1 ml of SOLVABLE solution (Perkin Elmer, Waltham, MA). Mice were euthanized at 4 h, and ∼100 mg of liver were collected into 1 ml of SOLVABLE. Freshly isolated small intestine was flushed with an ice-cold solution of (in mM) 130 NaCl, 5 KCl, 1 CaCl₂, 1 MgCl₂, 1 nitrilotriacetic acid, and 5 MES and buffered to pH 7.4 by using DEPP. Duodenal enterocytes were collected (as described above) into 1.5-ml microcentrifuge tubes containing 0.99 ml of SOLVABLE plus 10 μl of protease inhibitor cocktail I (EMD Biochemicals, San Diego, CA). Aliquots of this sample were used to count ⁵⁹Fe and to quantify protein concentration by using the Pierce bicinchoninic acid assay (Thermo Fisher Scientific). Tissue samples in SOLVABLE were further processed according to the manufacturer's protocol. We added ScintiSafe 30% cocktail and measured sample radioactivity by using liquid scintillation counting.

Statistical analyses.

Statistical analyses were performed using SigmaPlot version 13 (Systat Software), with critical significance level α = 0.05. Data are presented as arithmetic means and SD for n independent observations, with the following exceptions: 1) qPCR data (relative fold expression) are expressed as geometric mean and SD, and 2) enterocyte and liver ⁵⁹Fe content (see Fig. 6), for which data were not normally distributed, are presented as box plots (median, interquartile range, and whiskers representing 10th and 90th percentiles). Inferences were made using appropriate statistical tests specified in the figure legends. When between-group comparisons were made using Student's t-tests for multiple variables (see Figs. 3 and 5), we used a false discovery rate procedure (10) in which we declare that a significant effect exists only when the individual P value (P_i) is less than the individual critical significance level ${(d}_{i)}^{*}$ computed for each comparison tested.

Fig. 6.

Intestinal iron absorption in female wild-type (+/+), NHE2^−/−, and NHE3^−/− mice. Wild-type (n = 10), NHE2^−/− (n = 7), and NHE3^−/− (n = 7) mice, mean (SD) age 144 (140) days, were dosed with 0.1 μCi ⁵⁹Fe/g body wt via oral intragastric gavage. A: blood ⁵⁹Fe content as a function of time after intragastric dose. Data for +/+ mice are plotted as the black triangles and line, NHE2^−/− mice as the light gray inverted triangle and line, and NHE3^−/− as dark gray circles and line. Data were analyzed by 2-way ANOVA with repeated measures over time but, for clarity, are displayed as means ± SE. We observed a main effect of genotype (P = 0.030), but not time (P = 0.10), and no interaction (P = 0.55). Holm-Šidák pair-wise comparisons vs. wild-type revealed that NHE3^−/− (P = 0.019), but not NHE2^−/− (P = 0.48), mice differed from wild-type mice. B: enterocyte ⁵⁹Fe content at 4 h. One-way ANOVA on ranks (P = 0.004) and Dunn's pair-wise comparisons vs. wild-type: ^aP = 0.13, ^bP = 0.002 vs. wild-type. C: liver ⁵⁹Fe content at 4 h. One-way ANOVA on ranks (P = 0.024) and pair-wise analysis vs. wild-type: ^cP = 0.81, ^dP = 0.013 vs. wild-type.

RESULTS

DMT1-mediated iron transport in vitro is activated at low pH.

We measured DMT1 activity in vitro as a function of extracellular pH. Transport of 2 μM ⁵⁵Fe²⁺ in RNA-injected Xenopus oocytes expressing human DMT1 was stimulated at low pH (Fig. 1). We estimated K_0.5^H at 0.7 μM (equivalent to pH ∼6.1). As we and others observed previously (20, 39), residual DMT1-mediated transport activity persisted even at pH 7.5 (∼24% of maximum; Fig. 1).

Fig. 1.

Divalent metal-ion transporter-1 (DMT1)-mediated iron transport in RNA-injected Xenopus oocytes is activated at low extracellular pH (pH_o). Uptake of 2 μM ⁵⁵Fe²⁺ is shown as a function of pH_o in control oocytes (gray triangles) and oocytes expressing human DMT1 (black circles). Values are means ± SD (n = 10–16 oocytes per group). Two-way ANOVA revealed an interaction (P < 0.001); i.e., pH dependence of ⁵⁵Fe²⁺ uptake in oocytes expressing DMT1 differed from that in control oocytes. Data for oocytes expressing DMT1 were fit by Eq. 1, providing the relationship shown by the solid black line (gray shading indicates 95% confidence interval) and the following estimates: minimum transport rate (V_min) = 0.18 ± 0.03 pmol/min, maximum transport rate (V_max) = 0.74 ± 0.04 pmol/min, H⁺ concentration at which transport was half-maximal (K_0.5^H) = 0.72 ± 0.10 μM (i.e., pH ∼6.1), and Hill slope (a) = 2.7 ± 0.9 (r² = 0.98, P < 0.001, n = 8). Data for control oocytes could not be adequately fit by Eq. 1 (r² = 0.70, P = 0.15, n = 8) and are graphed as the gray line plot.

Iron metabolism and homeostasis in NHE2^−/− mice.

We measured liver nonheme iron content as an indicator of iron stores in male and female mice as a function of NHE2 genotype and as a function of dietary iron. Liver iron stores were lower in NHE2^−/− than wild-type (NHE2^+/+) and heterozygous (NHE2^+/−) mice, regardless of sex and whether the animals were stressed on a low-Fe diet for 6 wk (Fig. 2). Liver iron stores were lower in heterozygous than wild-type mice only when the animals were fed a low-Fe diet (Fig. 2).

Fig. 2.

Liver nonheme iron content in mice as a function of Na⁺/H⁺ exchanger-2 (NHE2) genotype. Male and female wild-type (NHE2^+/+), NHE2 heterozygous (NHE2^+/−), and NHE2-null (NHE2^−/−) mice were maintained for 6 wk postweaning on a normal or low-iron (low-Fe) diet. Values are means ± SD for 6–26 mice per group; mean (SD) age 101 (27) days. Three-way ANOVA revealed main effects of genotype, sex, and diet (all P < 0.001); 2-way interactions of sex × diet (P = 0.001) and genotype × diet (P < 0.001), but not genotype × sex (P = 0.38); and no 3-way interaction (P = 0.15). Pair-wise multiple comparisons using the Holm-Šidák test within diets indicated that NHE2^−/− differed from NHE2^+/+ and NHE2^+/−, regardless of diet (P < 0.001), and that heterozygotes differed from wild-type mice fed a low-Fe diet (P < 0.001), but not a normal diet (P = 0.30).

We examined the effect of loss of NHE2 on hematological and blood iron variables in mice. We limited our analysis to male NHE2^+/+ and NHE2^−/− mice fed a low-Fe diet, since the effect of NHE2 knockout on liver iron stores was independent of sex and diet (Fig. 2). Male NHE2^+/+ (n = 5) and NHE2^−/− (n = 9) mice were fed a low-Fe diet for 6 wk prior to analysis at ∼19 wk of age. Age did not differ between genotypes (P = 0.98), nor did body weight [25.9 (SD 3.1) and 27.1 (4.6) g for NHE2^+/+ and NHE2^−/−, respectively, P = 0.61]. Inspection of peripheral blood smears (not shown) did not reveal any difference in red cell morphology between male NHE2^+/+ and NHE2^−/− mice fed a low-Fe diet. Automated CBC revealed no change in any hematological variable, and we found no change in serum iron or transferrin saturation in NHE2^−/− mice compared with NHE2^+/+ mice (Fig. 3, A–H).

We examined the effect of ablation of NHE2 on the expression of genes involved in iron homeostasis and absorption. Loss of NHE2 had no effect on the intestinal expression of DMT1, the apical surface ferrireductase Cybrd1, or the basolateral iron exporter Fpn (Fig. 3, I–K). Although we observed no regulation at the mRNA level, it is possible that protein levels of DMT1, Cybrd1, or Fpn could be elevated in the intestine of NHE2^−/− mice; however, we found no change in hepatic expression of the iron-regulatory hormone hepcidin (Hamp1) (Fig. 3L), the primary regulator of ferroportin protein levels. In summary, our data indicate that ablation of NHE2 in the mouse results in reduced iron stores; however, NHE2^−/− mice maintain normal serum iron levels without the need for compensatory regulation of iron-related genes.

Iron metabolism and homeostasis in NHE3^−/− mice.

Liver nonheme iron stores were depleted in male and female NHE3^−/− mice compared with NHE3^+/+ or NHE3^+/− mice, regardless of sex or dietary iron (Fig. 4). Iron stores were lower in heterozygous than wild-type mice. Although our statistical analyses did not directly compare NHE3^−/− with NHE2^−/− mice, ablation of NHE3 appeared to have a more striking effect on liver iron stores (70–88%; Fig. 4) than did ablation of NHE2 (31–86%; Fig. 2).

Fig. 4.

Liver nonheme iron content in mice as a function of Na⁺/H⁺ exchanger-3 (NHE3) genotype. Male and female NHE3 wild-type (NHE3^+/+), heterozygous (NHE3^+/−), and null (NHE3^−/−) mice were maintained for 6 wk postweaning on a normal or low-Fe diet. Data are means and SD for 10–22 per group; mean (SD) age 113 (33) days. Three-way ANOVA revealed main effects of genotype, sex, and diet (all P < 0.001); 2-way interactions of genotype × sex (P = 0.001), but not genotype × diet (P = 0.53) or sex × diet (P = 0.060); and no 3-way interaction (P = 0.85). Pair-wise multiple comparisons using the Holm-Šidák test indicated that all genotypes differed from one another (P < 0.001), regardless of sex.

We examined the effect of loss of NHE3 on hematological and blood iron variables in mice. We limited our analysis to male NHE3^+/+ and NHE3^−/− mice fed a low-Fe diet, since the effect of ablation of NHE3 on liver iron stores was independent of diet and was qualitatively similar for male and female mice (Fig. 4). Male NHE3^+/+ (n = 14) and NHE3^−/− (n = 8) mice were fed a low-Fe diet for 6 wk prior to analysis at ∼16 wk of age. Age did not differ between genotypes (P = 0.48); however, NHE3^−/− mice were substantially underweight relative to their wild-type littermates [21.8 (SD 3.1) vs. 32.4 (4.5) g, P < 0.001]. In contrast, in the experiment described in Fig. 4, body weight of female NHE3^−/− mice did not differ from that of wild-type mice (data not shown; P = 0.37), despite a stronger effect of NHE3 ablation on liver nonheme iron in female than male mice (P = 0.0014), so it is unlikely that decreased food intake in NHE3^−/− mice explains their depleted iron stores.

Inspection of peripheral blood smears from male NHE3^−/− mice fed a low-Fe diet did not reveal any difference in red cell morphology compared with wild-type mice (not shown). We found no meaningful change in hematological variables of NHE3^−/− mice compared with wild-type mice (Fig. 5, A–F); however, we observed modest reductions in serum iron and transferrin saturation in NHE3^−/− mice (Fig. 5, G and H). Given that Na⁺ and fluid absorption are defective in the NHE3^−/− mouse (21, 52), it is conceivable that the characteristic dehydration in the NHE3^−/− mouse may have masked any effect of iron deficiency on hematocrit and hemoglobin concentration (we observed a trivial decrease in mean corpuscular hemoglobin concentration) and may underlie the slightly elevated red cell count.

Ablation of NHE3 resulted in upregulation of the intestinal expression of DMT1, Cybrd1, and Fpn (Fig. 5, I–K) and suppressed the hepatic expression of Hamp1 to <1% of that of wild-type mice (Fig. 5L). In summary, ablation of NHE3 in the mouse results in depleted iron stores, reduced serum iron variables, and appropriate compensatory upregulation of the iron-absorptive machinery.

Ablation of NHE3, but not NHE2, severely impairs intestinal iron absorption.

To directly test roles for NHE2 and NHE3 in driving intestinal iron absorption, we measured radiotracer iron absorption from an oral intragastric dose in wild-type, NHE2^−/−, and NHE3^−/− mice. We measured hematocrit over the 4-h time course to ensure against large changes in blood volume and to detect, if present, any indication of the onset of hypovolemic shock. We observed no change in hematocrit over time and no time-dependent differences between wild-type, NHE2^−/−, and NHE3^−/− mice (data not shown).

Appearance of ⁵⁹Fe in the blood following the oral dose did not differ between NHE2^−/− and wild-type mice. In contrast, ⁵⁹Fe appearance in the blood of NHE3^−/− mice was substantially blunted: only 21% of that of wild-type mice (from area under the curve; Fig. 6A). At 4 h, we obtained robust ⁵⁹Fe signals in enterocytes and liver (the latter tissue being expected to rapidly clear ⁵⁹Fe from the portal and peripheral circulation) from wild-type mice (Fig. 6, B and C). Whereas ⁵⁹Fe content of enterocytes and liver from NHE2^−/− mice did not differ from wild-type mice, enterocyte and liver ⁵⁹Fe content in NHE3^−/− mice was only a small fraction of that in wild-type mice. The lack of ⁵⁹Fe within enterocytes of NHE3^−/− mice demonstrates that the site of lesion in iron absorption in these mice is the intestinal brush border (we assume we had bypassed the need for luminal ferrireduction by providing Fe²⁺ in the presence of ascorbic acid). Our data show that iron uptake at the intestinal brush border largely depends on the activity of NHE3.

DISCUSSION

Our study demonstrates that intestinal absorption of nonheme iron relies in large part on the activity of NHE3, the primary determinant of the brush-border acidic microclimate in the mature mammalian intestine, and reveals NHE3 as the source of the H⁺-electrochemical potential gradient driving DMT1-mediated apical iron uptake.

We showed previously that DMT1-mediated iron transport in vitro is functionally coupled to the H⁺ flux and that the affinity of DMT1 for Fe²⁺ is increased at low pH (39). In the present study, we found that H⁺ activated human DMT1-mediated ⁵⁵Fe²⁺ transport in vitro (Fig. 1), with K_0.5^H ∼0.7 μM, i.e., pK ∼6.1, which coincides with the surface microclimate pH in the human proximal small intestine under physiological conditions (42). Our data illustrate that even modest changes in the microclimate pH (over the roughly “linear” region, pH 6.0–6.3) will be sufficient to modulate substantially DMT1-mediated apical iron uptake. Several factors, including nitric oxide and bacterial toxins (e.g., Escherichia coli heat-stable enterotoxin and cholera toxin), can rapidly alkalize the surface microclimate of mammalian intestine acutely (22, 41). Clostridium difficile toxin B decreases NHE3 mRNA and protein levels in human intestinal organoids (14) and triggers the internalization of NHE3 protein in mammalian epithelial cell lines (26). Surface microclimate pH of proximal small intestine was 6.1 in young rats but rose to 6.5 in senescent rats (29). Whether a similar change occurs in human subjects is not known; nevertheless, this observation raises the possibility that a diminished driving force for iron absorption could be one of several factors contributing to iron deficiency common in the elderly (15).

Several other mammalian intestinal transporters function, like DMT1, as H⁺-solute symporters (reviewed in Ref. 56). Among these are H⁺/amino acid transporter-1 (PAT1) (1), H⁺/folate transporter-1 (PCFT) (48), and H⁺/peptide transporter-1 (PEPT1) (16, 37). Although their dependence on NHE3 activity has not been demonstrated in vivo, PAT1 and PEPT1 transport activity in the Caco-2 intestinal cell line is inhibited by NHE3 inhibitors (2, 57).

Whereas gastric acid is thought to aid in iron absorption by promoting the solubility of Fe(III) complexes and the reduction of ferric ion by luminal ascorbic acid, we have found no clear requirement for gastric H⁺-K⁺-ATPase, the primary protein responsible for the gastric H⁺ secretion (M. A. Engevik, A. Shawki, S. R. Anthony, R. A. Baik, R. S. Kim, R. T. Worrell, G. E. Shull, and B. Mackenzie, unpublished observations). Gastric acid secretion has not been measured in NHE3^−/− mice; nevertheless, NHE3 is not expressed in mouse stomach, and NHE3^−/− mice exhibit no histological abnormalities in the stomach (G. E. Shull, unpublished observations).

DMT1 is a validated therapeutic target (54) in iron-overload disorders (11). Our observation that iron absorption was disrupted in NHE3^−/− mice raises the possibility that specific inhibition of NHE3 in the proximal small intestine may block iron absorption and could be used therapeutically to prevent iron overload in hereditary hemochromatosis or thalassemia major.

GRANTS

This study was supported primarily by National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) Grant R01 DK-080047 (to B. Mackenzie). R. A. Baik was supported by the University of Cincinnati Medical Student Summer Research Program, funded in part by NIDDK Short-Term Medical Student Training Grant DK-060444. Use of core services was supported in part by NIDDK Grant P30 DK-078392 (Digestive Health Center, Cincinnati Children's Hospital and University of Cincinnati). Additional support for this study was provided by NIDDK Grants R01 DK-050594 (to G. E. Shull) and R01 DK079979 (to R. T. Worrell) and by the University of Cincinnati (to B. Mackenzie).

DISCLAIMERS

The content of this work is solely the responsibility of the authors and does not necessarily represent the official views of the NIDDK or the National Institutes of Health.

DISCLOSURES

B. Mackenzie is a grant recipient of Vifor Pharma.

AUTHOR CONTRIBUTIONS

A.S., G.E.S., and B.M. developed the concept and designed the research; A.S., M.A.E., R.S.K., P.B.K., R.A.B., S.R.A., and B.M. performed the experiments; A.S., M.A.E., R.S.K., P.B.K., R.A.B., S.R.A., and B.M. analyzed the data; A.S., M.A.E., R.S.K., P.B.K., R.A.B., S.R.A., R.T.W., G.E.S., and B.M. interpreted the results of the experiments; A.S. and B.M. prepared the figures; A.S. and B.M. drafted the manuscript; A.S., M.A.E., R.S.K., P.B.K., R.A.B., S.R.A., R.T.W., G.E.S., and B.M. edited and revised the manuscript; A.S., M.A.E., R.S.K., P.B.K., R.A.B., S.R.A., R.T.W., G.E.S., and B.M. approved the final version of the manuscript.