The KCNE2 Potassium Channel Ancillary Subunit Is Essential for Gastric Acid Secretion*

Genes in the KCNE family encode single transmembrane domain ancillary subunits that co-assemble with voltage-gated potassium (Kv) channel α subunits to alter their function. KCNE2 (also known as MiRP1) is expressed in the heart, is associated with human cardiac arrhythmia, and modulates cardiac Kv α subunits hERG and KCNQ1 in vitro. KCNE2 and KCNQ1 are also expressed in parietal cells, leading to speculation they form a native channel complex there. Here, we disrupted the murine kcne2 gene and found that kcne2 (-/-) mice have a severe gastric phenotype with profoundly reduced parietal cell proton secretion, abnormal parietal cell morphology, achlorhydria, hypergastrinemia, and striking gastric glandular hyperplasia arising from an increase in the number of non-acid secretory cells. KCNQ1 exhibited abnormal distribution in gastric glands from kcne2 (-/-) mice, with increased expression in non-acid secretory cells. Parietal cells from kcne2 (+/-) mice exhibited normal architecture but reduced proton secretion, and kcne2 (+/-) mice were hypochlorhydric, indicating a gene-dose effect and a primary defect in gastric acid secretion. These data demonstrate that KCNE2 is essential for gastric acid secretion, the first genetic evidence that a member of the KCNE gene family is required for normal gastrointestinal function.

Voltage-gated potassium (Kv) 2 channels repolarize excitable cells by opening in response to membrane depolarization to permit K ϩ ion efflux. In addition to the 40 known genes that encode the pore-forming (␣) subunits of Kv channels (1), a range of Kv channel ancillary subunits form heteromeric complexes with Kv ␣ subunits to alter their functional properties, thus increasing native Kv current diversity. One family of ancillary subunits, the MinK-related peptides (MiRPs, encoded by KCNE genes), contributes five known members to the human genome. MiRPs are single transmembrane domain subunits that co-assemble with Kv ␣ subunits, altering their gating, conductance, regulation, and pharmacology (2).
The MiRP1 protein, encoded by the KCNE2 gene, is now more commonly referred to as KCNE2, and this nomenclature is used here to avoid confusion. KCNE2 regulates hERG potassium channels, and KCNE2-hERG complexes are thought, at least in part, to generate the cardiac I Kr current, the major repolarizing force in human ventricles (3). Mutations in KCNE2 are associated with a form of inherited long QT syndrome, LQT6 (3)(4)(5). Further, relatively common polymorphisms in KCNE2 are associated with acquired (drug-induced) long QT syndrome, and some KCNE2 variants increase susceptibility to drug block of the I Kr channel complex (3,6).
Aside from interacting with hERG, KCNE2 has been found to modulate other Kv ␣ subunits in heterologous co-expression studies, including KCNQ1 (also known as Kv7.1) (7), Kv3.1, Kv3.2 (8), and Kv4.2 (9). Effects of KCNE2 on KCNQ1 are particularly dramatic: KCNE2 converts KCNQ1 to a voltage-independent "leak" channel that retains K ϩ selectivity but is constitutively active regardless of membrane potential (7). In addition to its role in human heart, KCNQ1 is required for normal gastric acid secretion (10) and is proposed to provide a K ϩ ion efflux in parietal cells located in the gastric glands of the stomach to balance K ϩ ion influx through the gastric H ϩ /K ϩ -ATPase (11,12). KCNE2, like KCNQ1, is expressed in parietal cells, leading to the suggestion that KCNE2 might co-assemble with KCNQ1 in parietal cells (10,13). Aside from heart and stomach, KCNE2 has also been detected in kidney, lung, bladder, brain, spinal cord, and skeletal muscle (2).
Thus, KCNE2 has the potential to play diverse roles in different tissues with a range of ␣ subunit partners. Here, we disrupted the murine kcne2 gene to identify its roles in mammalian physiology. The results demonstrate for the first time that KCNE2 is required for gastric acid secretion and also the first genetic evidence of a role for any of the KCNE proteins in the gastrointestinal tract. These findings have implications for the physiology, pharmacology, and pharmacogenetics of the mammalian gastrointestinal tract and potentially for treatment of pathophysiologic conditions such as gastro-esophageal reflux disease and peptic ulcers.
Immunohistochemistry-Immunohistochemical detection was performed with a Discovery XT System (Ventana Medical Systems) using slides prepared from stomach sections as for histology. For single labeling with diaminobenzidine detection a 30-min blocking step was employed, using 10% normal goat serum (Vector Laboratories; catalog number S-1000) and 2% bovine serum albumin (KCNQ1) or MOM Blocking reagent (Vector Laboratories; catalog number MKB-2213) (H ϩ /K ϩ -ATPase ␤-subunit) followed by avidin and biotin incubation for 4 -8 min each. Incubation with the primary antibodies (KCNQ1, Chemicon catalog number AB5932; 1 mg/ml; H ϩ /K ϩ -ATPase ␤-subunit, Affinity Bioreagents catalog number MA3-923; 0.5 mg/ml) was carried out for 3 h at room temperature followed by 60 min of incubation with: for KCNQ1, biotinylated anti-rabbit antibody at 1:200 dilution (Vectastain ABC kit (rabbit IgG) catalog number PK-6101) or for H ϩ /K ϩ -ATPase ␤-subunit, biotinylated anti-mouse IgG (HϩL) secondary antibody at 1:250 dilution (Vector MOM basic kit catalog number BMK-2202). Diaminobenzidine detection kit containing Blocker D, Copper D, Inhibitor D, streptavidinhorseradish peroxidase D and diaminobenzidine D (Ventana Medical Systems) was used according to the manufacturer's instructions. Slides were viewed with a Nikon Eclipse E600 microscope and photographed using a RT Color Camera and SPOT software (Diagnostic Instruments, Inc.).
Whole Stomach and Parietal Cell pH Measurements-Studies were conducted on gastric tissue from 3-month-old kcne2 (ϩ/ϩ), (ϩ/Ϫ), and (Ϫ/Ϫ) mice using methods described previously (18,19). Briefly, for whole stomach measurements, stomachs were ligated ex vivo at the esophageal and duodenal junctures and excised, infused with 0.3-0.5 ml of non-buffered, isotonic saline, and then incubated at 37°C for 1 h in oxygenated HEPES-buffered Ringer's solution (pH 7.4) with or without 100 M histamine (Sigma-Aldrich) before measuring pH of the aspirated stomach contents. For parietal cell measurements, individual gastric glands were hand dissected and loaded with the pH-sensitive dye 2Ј,7-bis(2-carbocyethyl)-5-(and 6) carboxyfluoresceinacetomethylester (BCECF) (Invitrogen). Proton extrusion by individual parietal cells was monitored by observing recovery of intracellular pH (pH i ) in the presence of 100 M histamine or 100 M carbachol (Sigma-Aldrich) after acidification using the NH 4 Cl prepulse technique (20).
Serum Gastrin Measurements-Mice were euthanized by carbon dioxide asphyxiation. Whole blood was collected postmortem by cardiac puncture following euthanasia. Following coagulation of the remaining whole blood at room temperature for 30 min, the clotted blood was centrifuged at 3000 rpm for 5-10 min at 4°C. Aliquots of serum were submitted to a commercial reference laboratory (Clinical Pathology Laboratory at Animal Medical Center, New York, NY) for determination of biochemical parameters. Serum gastrin levels were determined by AniLytics, Inc. (Gaithersburg, MD) using a double antibody radioimmunoassay (Diagnostic Products Corp., Los Angeles, CA).
Statistical Analyses-Numerical data were analyzed with EXCEL software (Microsoft) using one-way analysis of variance with statistical significance set at p Ͻ 0.05.

kcne2 (ϩ/Ϫ) and (Ϫ/Ϫ) Mice Have Impaired Gastric Acid
Secretion-Given the markedly abnormal appearance of the stomach in kcne2 (Ϫ/Ϫ) mice, we measured the pH in ex vivo stomachs to ascertain whether kcne2 is required for normal gastric acid secretion. The stomachs from kcne2 (ϩ/ϩ) mice had a mean basal pH of 3.9 Ϯ 0.1, which was significantly reduced to pH 2.8 Ϯ 0.2 when stimulated with 100 M histamine ( p Ͻ 0.01). In contrast, the mean luminal pH in stomachs from kcne2 (Ϫ/Ϫ) mice was strikingly more alkaline under basal conditions (pH 6.5 Ϯ 0.2) and was not significantly altered by histamine (pH 6.6 Ϯ 0.2). Importantly, the mean luminal pH in kcne2 (ϩ/Ϫ) mice was intermediate between the two (basal, pH 4.6 Ϯ 0.2) and was significantly shifted by histamine stimulation (pH 3.8 Ϯ 0.2, p Ͻ 0.05) ( Fig. 3; n ϭ 5-8 stomachs/group). These data demonstrate that disruption of the kcne2 gene impairs gastric acid secretion and that the effect of kcne2 on gastric acid secretion is gene-dose dependent.
kcne2 (Ϫ/Ϫ) Mice Have a Primary Defect in Gastric Acid Secretion-Gastrin, produced by G-cells in the antrum of the stomach, is a peptide hormone that stimulates parietal cell acid secretion. Impaired gastrin production can result in reduced gastric acid secretion (21). To determine whether the hypochlorhydria in kcne2 (ϩ/Ϫ) and achlorhydria in kcne2 (Ϫ/Ϫ) mice were due to a defect in gastrin production or release, we compared serum gastrin levels in kcne2 (ϩ/ϩ), (ϩ/Ϫ), and (Ϫ/Ϫ) mice. Gastrin levels were markedly elevated in kcne2 FIGURE 1. Generation and genotyping of kcne2 (؉/؊) and (؊/؊) mice. A, targeting vector designed to disrupt exon 1 of the murine kcne2 locus with a neomycin resistance cassette. LoxP sites for Cre-recombinase removal surround the neomycin resistance cassette. Regions between dashed lines are recombined during homologous recombination. B, genotyping of progeny from kcne2 (ϩ/Ϫ) ϫ (ϩ/Ϫ) matings by Southern blotting. Tail genomic DNA was digested with BglII and hybridized to a radiolabeled probe complementary to sequence within a 323-nucleotide region upstream of the kcne2 coding region. This 5Ј-probe binds to an 8.1-kb restriction fragment from the wild-type locus and a 4.2-kb restriction fragment from the targeted locus. C, left gel, product generated by PCR using kcne2-specific primers from a cDNA library generated from reverse transcription PCR of stomach mRNA from kcne2 (ϩ/ϩ) or (Ϫ/Ϫ) mice as indicated above the lanes. Numbers to the left indicate migration distance of DNA ladder markers on agarose gel. Right gel, negative controls using mRNA as in the left gel but omitting reverse transcriptase in the reverse transcription PCR step to demonstrate lack of genomic DNA contamination and positive control using KCNE2 cDNA (right lane). D, Western blots of stomach membrane preparations from kcne2 (ϩ/ϩ) and (Ϫ/Ϫ) mice using anti-KCNE2 polyclonal antibody (right lanes). Antibody specificity was confirmed using lysates from non-transfected Chinese hamster ovary cells and Chinese hamster ovary cells transfected with human KCNE2 cDNA (left lanes). Migration distance of molecular mass marker is shown on the left (in kDa).
(Ϫ/Ϫ) mice: 421 and 213 pg/ml in the two knock-out mice evaluated compared with 93 and 33 pg/ml for two wild-type littermates and 45 and 28 pg/ml for two kcne2 (ϩ/Ϫ) mice in this study. Previous reports indicated that normal murine gastrin levels are in the range of 32-67 pg/ml (12,22). These data demonstrated that achlorhydria in kcne2 (Ϫ/Ϫ) mice and hypochlorhydria in kcne2 (ϩ/Ϫ) mice were not due to a secondary defect in gastrin production or release. The results suggested that kcne2 disruption causes a primary defect in parietal cell acid secretion.
kcne2 (Ϫ/Ϫ) Mice Have Gastric Glandular Hyperplasia, Confined to Non-acid Secretory Cells-Histological examination demonstrated that kcne2 (Ϫ/Ϫ) mice exhibit significant diffuse gastric glandular hyperplasia characterized by lengthening of both the fundus and neck portions of the stomach. These hyperplastic gastric glands were lined primarily with tall, columnar cells exhibiting unusually abundant cytoplasm that was pale blue when stained with hematoxylin and eosin (Fig.  4A) and magenta when stained with periodic acid-Schiff (Fig.  4B), indicating these columnar cells were mucous neck and/or chief cells. In contrast, the gastric glands of kcne2 (ϩ/ϩ) and (ϩ/Ϫ) mice displayed no hyperplasia and were of normal size (Fig. 4A). In addition, there were mild to moderate numbers of neutrophils scattered within the lamina propria basally, the submucosa, and the muscularis in all kcne2 (Ϫ/Ϫ) stomachs studied (n ϭ 4) but in none of the stomachs from wild-type or heterozygous mice (Fig. 4A).
KCNQ1 Expression Is Up-regulated in kcne2 (Ϫ/Ϫ) Mice-KCNQ1 was previously suggested as a putative partner for KCNE2 in parietal cells (10,13). In gastric glands, KCNQ1 is expressed predominantly in parietal cells of wild-type mice but reportedly exhibits lower level expression in non-parietal cell types such as mucous-producing cells (10). Here, using antibodies raised against KCNQ1 ␣ subunit, we found KCNQ1 distribution in the stomach corpus of kcne2 (ϩ/ϩ) mice similar to that previously reported, with a region of high expression in parietal cells confined largely to a band in the upper and middle neck region of gastric glands and weaker expression in the fundus and surface cells of gastric glands. In contrast, in kcne2 (Ϫ/Ϫ) mice, KCNQ1 was highly expressed more uniformly throughout the gastric gland in parietal and non-parietal cells, from the fundus through to the surface cells (Fig. 5, C and D). Quantification of this redistribution by cell counting (n ϭ 25  glands/genotype) indicated that the stronger KCNQ1 labeling was detected in a mean of 11.4 Ϯ 0.4 cells per gastric gland in kcne2 (ϩ/ϩ) mice compared with a mean of 21.6 Ϯ 0.7 cells per gastric gland in kcne2 (Ϫ/Ϫ) mice, a 2-fold increase in the latter. This altered KCNQ1 distribution was suggestive of compensatory remodeling in response to the absence of kcne2. kcne2 (Ϫ/Ϫ) Mice Exhibit Abnormal Parietal Cell Morphology-Electron micrographs of the corpus region of the stomach showed highly abnormal parietal cell morphology in the absence of kcne2; in particular, the intracellular canaliculi of kcne2 (Ϫ/Ϫ) parietal cells were much larger than those in wildtype parietal cells, leading to abnormal distribution of mito-chondria in kcne2 (Ϫ/Ϫ) parietal cells, in which they were confined to the cell periphery. Parietal cells from kcne2 (ϩ/Ϫ) mice, however, appeared structurally normal (Fig.  6A). Chief cells in kcne2 (Ϫ/Ϫ) mice displayed no obvious differences from those of their wild-type littermates (Fig. 6B). These data demonstrate that KCNE2 is required for normal parietal cell morphology. It is noteworthy that although kcne2 (ϩ/Ϫ) mice have impaired gastric acid secretion, their parietal cell morphology and glandular architecture are normal. This strongly suggests that KCNE2 has an important role in gastric acid secretion.
KCNE2 Is Required for Parietal Cell Acid Secretion-Stimulation of parietal cell histamine H 2 receptors or muscarinic M 3 receptors triggers parietal cell proton release. To determine whether KCNE2 is required for agonist-triggered parietal cell proton release, we measured intracellular pH recovery in individual parietal cells in freshly isolated gastric glands after activation of H 2 receptors with histamine (100 M) or of muscarinic M 3 receptors with carbachol (100 M). The resting pH in parietal cells was similar in all three genotypes: 7.25 Ϯ 0.03 in kcne2 (ϩ/ϩ), 7.28 Ϯ 0.04 in kcne2 (ϩ/Ϫ), and 7.26 Ϯ 0.03 in kcne2 (Ϫ/Ϫ) (n ϭ 70 -150 cells from 3-5 mice/group). After an NH 4 Cl prepulse to acid load the cells and stimulation with either carbachol or histamine (100 M), kcne2 (Ϫ/Ϫ) parietal cells had greatly suppressed rates of proton secretion: a mean ⌬pH/minute of ϳ0.01 units compared with ϳ0.1 units in kcne2 (ϩ/ϩ) cells ( p Ͻ 0.01). kcne2 (ϩ/Ϫ) cells had an intermediate mean proton secretion rate of 0.05 pH units ( p Ͻ 0.05) (Fig. 7, A and B; n ϭ 70 -150 cells from 3-5 mice/group). These results demonstrate that KCNE2 is required for normal proton extrusion by parietal cells in vitro.

DISCUSSION
Gastric acidification is required for normal digestion and nutrient absorption in humans as well as for the sterilization of fluids and the prevention of bacterial overgrowth (24). The molecular regulation of parietal cell acid secretion is incompletely understood. The gastric H ϩ /K ϩ -ATPase is primarily responsible for acidification of the stomach lumen (11), and the 1:1 exchange of H ϩ ions for K ϩ ions by the H ϩ /K ϩ -ATPase requires a continuous supply of K ϩ from the luminal side (25). To avoid rapid depletion of K ϩ in the stomach lumen along with a reduction of acid secretion, a K ϩ recycling path at the apical surface is required.
Previous reports have provided genetic evidence that KCNQ1 is required for this K ϩ recycling mechanism (12,26), although a range of candidate potassium channels have been proposed and some or all of them may be involved in this process (27). Inward rectifier potassium channels Kir2.1, Kir4.1, Kir4.2, and Kir7.1 are expressed in gastric mucosa, with Kir2.1 reportedly showing the highest expression (28,29). Kir2.1 colocal-izes with H ϩ /K ϩ -ATPase in parietal cells and shows increased current when stimulated with lowered pH and protein kinase A (29). Kir4.1, but not Kir4.2 or Kir7.1, is also expressed in parietal cells, where it reportedly localizes to the apical membrane (28). Furthermore, Ba 2ϩ ion, a nonspecific blocker of inward rectifier channels, suppresses proton secretion from parietal cells (28). Although genetic disruption of either KCNQ1 or KCNE2 leads to achlorhydria due to a profound loss of parietal cell proton secretion, suggestive of a dominant role for KCNE2-KCNQ1 complexes in gastric acidification, it is possible that other potassium-selective channel types also contribute. Kir2.1 knock-out mice are not reported to exhibit a gastric defect, but they exhibit cleft palate and prolonged cardiac action potentials and die shortly after birth (30).
Recent studies incorporating heterologous characterization, native immunocolocalization, and pharmacological experiments and microarray analysis have suggested the possibility that KCNQ1 co-assembles with KCNE2 in parietal cells to form a complex important for gastric acid secretion (10,13,(31)(32)(33). However, in the absence of native KCNE2-KCNQ1 co-immunoprecipitation or genetic evidence for the role of KCNE2 in parietal cells, firm conclusions regarding the necessity of KCNE2 in these complexes have been lacking. The present data provide, for the first time, genetic and native functional evidence that KCNE2 is essential for gastric acid secretion.
Three conclusions can be drawn from our genetic data. First, the finding that kcne2 disruption drastically reduces carbachol-or histamine-stimulated gastric acidification suggests that homomeric KCNQ1 channels are not sufficient for gastric acid secretion. Second, kcne2 (ϩ/Ϫ) mice exhibited hypochlorhydria but unlike kcne2 (Ϫ/Ϫ) mice did not show either abnormal parietal cell morphology or significant gastric hyperplasia. This suggests that KCNE2 contributes directly to the K ϩ recycling required for proton secretion, as parietal cells from kcne2 (ϩ/Ϫ) mice have normal architecture but reduced ability to secrete acid. Previous work provided genetic evidence that KCNQ1 is required for normal gastric acid secretion and that KCNE2 is expressed in parietal cells (12,13,(31)(32)(33). Together, these results suggest that KCNE2-KCNQ1 channel complexes form a potassium leak channel that constitutes an important K ϩ ion recycling pathway required for parietal cell acid secretion. Third, other kcne family proteins are expressed in the gastrointestinal tract, including KCNE1 (MinK) and KCNE3 (MiRP2) (31), which also converts KCNQ1 into a voltage-independent leak channel (34). In fact, KCNQ1 is a particularly promiscuous partner for the KCNE gene family, with all five known KCNE peptides found to interact with KCNQ1 with various functional outcomes (2). Our data suggest that there is no redundancy of function in this instance: that nei-ther KCNE3 nor any other ancillary subunits can adequately compensate for loss of KCNE2 in parietal cells.
Striking morphological changes in gastric glands were apparent in kcne2 (Ϫ/Ϫ) mice. Glandular hyperplasia was observed, confined to non-parietal cells and in particular mucous-producing cells as evidenced by periodic acid-Schiff staining and lack of H ϩ /K ϩ -ATPase ␤-subunit staining. Although parietal cells were not reduced in number, those in kcne2 (Ϫ/Ϫ) cells showed abnormal morphology with highly vacuolated secretory canaliculi. This type of hyperplasia was previously observed in KCNQ1-deficient mice (12) and in H ϩ /K ϩ -ATPase ␤-subunit-deficient mice (35). In the latter, this hyperplasia was found to result from secondary hypergastrinemia because H ϩ /K ϩ -ATPase ␤-subunit/gastrin double knock-out mice did not display gastric hyperplasia (35). Thus, non-acid secretory cell hyperplasia in kcne2 (Ϫ/Ϫ) mice may also result from secondary hypergastrinemia. Unexpectedly, kcne2 (Ϫ/Ϫ) knock out also resulted in a form of compensatory remodeling in which KCNQ1, the probable parietal cell partner of KCNE2, was expressed at atypically high levels in non-parietal cells and across the length of gastric glands. This remodeling was not, however, sufficient to restore gastric acidification to kcne2 (Ϫ/Ϫ) mice, underlining the necessity for KCNE2 in proton secretion, most  likely via its conversion of KCNQ1 to a constitutively active channel. In some diseased heart models, KCNE2 itself is up-regulated, causing altered ionic currents without changes in the expression levels of its putative ␣ subunit partners (36). Our finding that loss of KCNE2 can alter KCNQ1 expression and distribution presents a novel form of remodeling, and further studies will be undertaken to elucidate the underlying molecular mechanisms and their possible roles in normal gastric physiology.
The role of KCNE2 in gastric acid secretion raises the possibility that genetic variants in the human KCNE2 gene might associate with hypochlorhydria. Importantly, our data demonstrate that KCNE2 has a gene-dose-dependent effect on parietal cell acid secretion. This indicates that single KCNE2 gene mutations could affect gastric acid secretion in heterozygous human subjects, potentially resulting in complications such as reduced absorption of ferric iron and other minerals (24) and vitamin B12 (37) and a range of orally ingested therapeutic drugs such as dipyridamole (38) and atazanavir (39).
Because the pharmacology of KCNE2-KCNQ1 complexes is markedly different from that of homomeric KCNQ1 channels and from that of KCNE1-KCNQ1 or KCNE3-KCNQ1 channels (13), the identification of KCNE2 as an ancillary subunit required for a parietal cell K ϩ ion-recycling channel will potentially allow specific pharmacological targeting of the KCNE2-KCNQ1 complex for hypersecretory disease states, with avoidance of effects on other KCNQ1-bearing channels such as cardiac I Ks , which is generated by KCNE1-KCNQ1 complexes (40,41) and has different pharmacology than KCNE2-KCNQ1 channels.
In summary, data derived from this first characterization of kcne2 knock-out mice support an important role for KCNE2 in mammalian gastric acid secretion. Further studies of kcne2 (Ϫ/Ϫ) mice are now intended to explore the potential roles of KCNE2 in other tissues in which it is expressed, including the heart and the brain.