Loss of gastric acid secretion is pathologically known as achlorhydria. Acid-secreting parietal cells are characterized by abundant expression of ezrin (Vil2), one of ezrin/radixin/moesin proteins, which generally cross-link actin filaments with plasma membrane proteins. Here, we show the direct in vivo involvement of ezrin in gastric acid secretion. Ezrin knockout (Vil2−/−) mice did not survive >1.5 wk after birth, making difficult to examine gastric acid secretion. We then generated ezrin knockdown (Vil2kd/kd) mice by introducing a neomycin resistance cassette between exons 2 and 3. Vil2kd/kd mice born at the expected Mendelian ratio exhibited growth retardation and a high mortality. Approximately 7% of Vil2kd/kd mice survived to adulthood. Ezrin protein levels in Vil2kd/kd stomachs decreased to <5% of the wild-type levels without compensatory up-regulation of radixin or moesin. Adult Vil2kd/kd mice suffered from severe achlorhydria. Immunofluorescence and electron microscopy revealed that this achlorhydria was caused by defects in the formation/expansion of canalicular apical membranes in gastric parietal cells.
Gastric acid, i.e., hydrochloric acid (HCl), is secreted from highly differentiated gastric epithelial cells called “parietal” cells, which are characterized by numerous H+,K+-ATPase-rich tubulovesicles and mitochondria in the cytoplasm (Urushidani and Forte, 1997; Yao and Forte, 2003). This secretion is tightly regulated. In parietal cells, ligands such as acetylcholine, gastrin, and histamine bind to their receptors localizing at basolateral plasma membranes (Urushidani and Forte, 1997; Lindstrom et al., 2001; Yao and Forte, 2003). Protein kinase cascades are then activated, and H+,K+-ATPase-rich tubulovesicles are fused with apical membranes to generate and secrete HCl to the gastric lumen (Forte and Yao, 1996; Forte et al., 1977; Duman et al., 2002; Zhou et al., 2003). However, the molecular mechanisms underlying the regulation of this tubulovesicle fusion remain elusive. One of the characteristic features of parietal cells is the abundant expression of ezrin, which is concentrated at the subapical actin cortical layers (Hanzel et al., 1991; Urushidani and Forte, 1997; Yao and Forte, 2003). Thus, it has been speculated that ezrin functions as one of the key regulators for gastric acid secretion in parietal cells. Recently, the NH2-terminal phosphorylation of ezrin was suggested to play a critical role in gastric acid secretion (Zhou et al., 2003). Ezrin belongs to the ezrin/radixin/moesin (ERM) family (Bretscher, 1983; Sato et al., 1992). ERM proteins have been shown to play crucial roles in various events in cell survival and differentiation through organizing actin-based cytoskeletons underlying plasma membranes (Tsukita and Yonemura, 1999; Bretscher et al., 2002). However, the possible involvement of ERM proteins in gastric acid secretion in parietal cells has not been well evaluated. Therefore, we attempted to examine the function of ezrin in gastric acid secretion in parietal cells using gene targeting technology.
Results And Discussion
We first generated mice completely lacking the expression of ezrin (Vil2−/− mice). When Vil2+/− mice were interbred, Vil2−/− mice were born at the submendelian segregation ratio, but they did not grow, and showed no increase in body weight after birth. All Vil2−/− mice died within 1.5 wk after birth before weaning (unpublished data). These findings were consistent with the recent report by Saotome et al. (2004). Considering that gastric acid secretion develops after birth and matures after weaning in mice (Helander, 1970; Ackerman, 1982), we concluded that Vil2−/− mice were not suitable for studying the role of ezrin in HCl secretion from parietal cells. We then attempted to produce ezrin knockdown mice (Vil2kd/kd) in which the expression of ezrin was not completely abolished, but largely suppressed. For that purpose, we targeted embryonic stem cells using a replacement construct that was designed to insert a neomycin-resistance cassette between exons 2 and 3 of the ezrin gene, according to the previous method described previously (Nagai et al., 2000; Fig. 1 A). Correctly targeted cells were selected by Southern blotting with internal and external probes, and by blastocyst injections of two independent embryonic stem cell clones, two lines of mice heterozygous for a mutation in the ezrin locus (Vil2+/kd) were obtained. When Vil2+/kd mice were interbred, wild-type (Vil2+/+), heterozygous (Vil2+/kd), and homozygous mutant (Vil2kd/kd) mice were born at the expected Mendelian segregation ratio, as observed from Southern blots (Fig. 1 B).
Vil2kd/kd mice exhibited severe growth retardation and a high mortality up to the weaning period (∼3 wk after birth; Fig. 2 A). However, in contrast to Vil2−/− mice, ∼7% of Vil2kd/kd mice survived beyond the weaning period. These mice gradually increased their body weight, and reached ∼75% of the body weight of their wild-type littermates by 7 wk (Fig. 2 B). They survived at least up to 2 yr old, and showed no severe gross abnormalities in histological scan. For example, in Vil2kd/kd mice, intestinal villi as well as microvillar structures of intestinal cells appeared to be normal, though they were sometimes slightly deformed (Fig. 2 C). Furthermore, the serum profile in these mice suggested that the intestinal absorption of glucose, trigycerol, cholesterol, or amino acids was not affected (unpublished data).
We then examined the expression of ERM proteins in various organs of adult sex-matched Vil2+/+ and Vil2kd/kd littermate pairs aged ∼8 wk. When immunoblotting was performed with anti-ERM pAb, which equally recognized ERM proteins, ezrin was mostly undetectable without compensatory up-regulation of radixin or moesin in most Vil2kd/kd organs such as heart, liver, lung, and small intestine (Fig. 3 A). In the Vil2kd/kd stomach, a small amount of ezrin was reproducibly detected, possibly because exceptionally large amounts of ezrin are expressed in Vil2+/+ stomach (Fig. 3 A). Immunoblotting with antibodies specific for ezrin, radixin, or moesin confirmed that in the Vil2kd/kd stomach, ezrin was expressed in small amounts (∼5%) and that the expression of radixin or moesin was not up-regulated in a compensatory manner (Fig. 3 B). Quantitative RT-PCR analyses revealed that in the Vil2kd/kd stomach, ezrin mRNA levels were decreased to <5% of Vil2+/+ stomach levels (not depicted). The stomach of adult Vil2kd/kd mice appeared normal macroscopically. Histological examination of the Vil2kd/kd stomach revealed that parietal cells, as well as chief and mucous cells, were normally developed, which were indistinguishable from those in the Vil2+/+ stomach. However, there was a thickening of the gastric mucosa in Vil2kd/kd mice (Fig. 3 C). Ultra-thin section EM also easily identified these differentiated gastric epithelial cell types in the Vil2kd/kd stomach (not depicted). Immunofluorescence microscopy revealed that in Vil2kd/kd parietal cells, which were positive for H+,K+-ATPase, no compensatory up-regulation of radixin or moesin was detected (Fig. 3 D). These findings indicated that the stomach of adult Vil2kd/kd mice could provide a useful system to evaluate the possible involvement of ezrin in HCl secretion in parietal cells.
Because the gastric juice of Vil2+/+ and Vil2kd/kd stomachs showed a different pH by strips of pH indicator paper, acidic and neutral, respectively, suggesting that Vil2kd/kd mice suffered from achlorhydria (unpublished data), we prepared the isolated stomach samples from adult sex-matched Vil2+/+ and Vil2kd/kd littermate pairs aged 6–20 wk, and measured histamine-triggered acid secretion using the Ussing chamber system (Suzuki and Kaneko, 1987). As shown in Fig. 4 A, when the isolated stomach was stimulated by histamine under physiological conditions, acid secretion was measured as 1.76 ± 0.38 and 0.14 ± 0.21 μmol/h/cm2 (mean ± SD, n = 4) for the Vil2+/+ and Vil2kd/kd stomach, respectively, confirming achlorhydria in the Vil2kd/kd mice.
The question has naturally arisen how the suppression of ezrin expression abolishes HCl secretion in parietal cells. There were no differences detected between Vil2+/+ and Vil2kd/kd parietal cells in the expression level and subcellular distribution of gastric H+,K+-ATPase (Fig. 4, B and C). In the Vil2+/+ stomach, ezrin was highly concentrated at the well-developed actin cortical layers of parietal cells (Fig. 4 D). By contrast, in the Vil2kd/kd stomach, immunofluorescence signal for ezrin was very weak and it was difficult to determine its subcellular distribution in parietal cells. In these parietal cells, the subcortical actin layer became significantly thinner as compared with that in Vil2+/+ stomach. As reported previously (Mercier et al., 1989; Agnew et al., 1999; Zhou et al., 2003), when isolated Vil2+/+ stomach was incubated with histamine, the organization of actin-based cytoskeletons in parietal cells changed markedly (Fig. 4 E). In addition to the cortical layers, the deeper regions of parietal cells were also positive for rhodamine-phalloidine staining. This change was thought to be attributed to the expansion of canalicular apical membranes into the deeper regions through successive tubulovesicle fusion that was associated with the facilitation of actin polymerization beneath these newly formed apical membranes. Interestingly, in parietal cells in the Vil2kd/kd stomach, the rhodamine-phalloidine staining pattern, i.e., thin cortical actin layers, did not change significantly with histamine stimulation (Fig. 4 E). This suggested that ezrin knockdown affected the molecular mechanism behind the expansion of canalicular apical membranes in parietal cells.
Finally, we examined the Vil2+/+ and Vil2kd/kd parietal cells by ultra-thin section EM (Fig. 5). We fasted Vil2+/+ and Vil2kd/kd mice in order to suppress the secretion of gastric juice. In the Vil2+/+ stomach, parietal cells are characterized by a meandering invagination of the apical surface called the secretory canaliculus, which bears numerous microvilli with actin filament cores (Fig. 5 A, red). Tubulovesicles were densely distributed just beneath the apical canalicular membranes (Fig. 5 A, blue). They were similar in dimension and appearance to canalicular microvilli, though they were vacant vesicular structures and not cellular protrusions. Close inspection easily distinguished these two membranous structures (Fig. 5 A). Vil2+/+ parietal cells bore a fairly developed secretory canaliculus in addition to cytoplasmic tubulovesicles. In contrast, Vil2kd/kd parietal cells mostly lacked invaginated apical canaliculi, with the cytoplasm filled with numerous tubulovesicles (Fig. 5 B, −histamine). The stomach, which was dissected out from fasted Vil2+/+ and Vil2kd/kd mice, was stimulated with histamine (Fig. 5 B, +histamine). In histamine-stimulated Vil2+/+ parietal cells, the apical secretory canaliculus markedly expanded, and tubulovesicles were decreased in number, though the size and morphological aspects of the remaining tubulovesicles did not change. However, in Vil2kd/kd parietal cells, histamine stimulation did not induce the formation/expansion of apical secretory canaliculi, leaving the cytoplasm densely packed with tubulovesicles. This difference in the behavior of the apical canalicular membrane and tubulovesicles between Vil2+/+ and Vil2kd/kd parietal cells was confirmed semi-quantitatively (Fig. 5 C). Therefore, we concluded that ezrin is not required for formation of tubulovesicles per se, but plays a crucial role in formation/expansion of apical secretory canaliculi in parietal cells. This formation/expansion process may include many steps such as signaling, recruitment/docking/fusion of tubulovesicles to apical membranes, etc. It is still premature to further discuss how ezrin is involved in the secretion of gastric acid at the molecular level, but we believe that Vil2kd/kd mice, as well as radixin-deficient mice (Kikuchi et al., 2002), will offer novel systems to understand how ERM proteins are involved in the formation of epithelial apical membranes in general.
Materials And Methods
Antibodies and reagents
Rabbit anti-ERM pAb (TK89) recognized all ERM proteins almost equally (Kondo et al., 1997). Rat anti-ezrin, anti-radixin, and anti-moesin mAbs reacted specifically with the respective ERM proteins (Kondo et al., 1997). Anti–α-subunit of H+,K+-ATPase mAb (Calbiochem), anti–β-subunit of H+, K+-ATPase mAb (Affinity BioReagents), histamine, and omeprazole (Sigma-Aldrich) were purchased.
Generation of ezrin knockdown (Vil2kd/kd) mice
The ezrin genomic clone (13.7 kb) including exons 2–6 was isolated from a λ-phage 129/Sv mouse genomic library. The targeting vector (Fig. 1 A) was constructed by standard recombinant DNA techniques and the ezrin knockdown mice were obtained as described previously (Doi et al., 1999; Nagai et al., 2000; Kikuchi et al., 2002).
Various tissues were taken from Vil2kd/kd and Vil2+/+ mice and homogenized in SDS sample buffer (50 mM Tris-HCl, pH 6.8, 2% SDS, 20% glycerol, 2% 2-mercaptoethanol, 0.01% bromophenol blue). Total lysates (25 μg) were separated by SDS-PAGE (10% SDS-PAGE). Proteins were then electrophoretically transferred from gels to nitrocellulose membranes and then processed for immunoblotting. Representative data were obtained from the independent immunoblots for five separate combinations of Vil2kd/kd and Vil2+/+ mice.
Organs from Vil2kd/kd and Vil2+/+ mice were cut into small pieces and frozen in liquid nitrogen for immunofluorescence microscopy. Frozen sections of Vil2kd/kd and Vil2+/+ stomachs were cut on a cryostat, mounted on glass slides, and air dried. They were fixed with 100% methanol at −20°C for 10 min or 2% formaldehyde in PBS at RT for 15 min, and processed for immunofluorescence microscopy. Fluorescence images were obtained with a confocal microscope (model LSM 510 META; Carl Zeiss MicroImaging, Inc.), equipped with an Axioplan2 (Plan Apochromat 63×/1.40 NA oil immersion objective; Carl Zeiss MicroImaging, Inc.).
Vil2kd/kd and Vil2+/+ stomach samples were fixed in 2.5% glutaraldehyde and 2% formaldehyde in 0.1 M cacodylate buffer, pH 7.2, and processed for ultra-thin section EM as described previously (Kikuchi et al., 2002).
Histamine stimulation and measurement of acid secretion in the Ussing chamber
The gastric region of the stomach was cut off and the muscle layer was largely removed with fine scissors. The resultant tissue preparations were mounted vertically between chambers that provided an exposed area of 0.2 cm2. The volume of the bathing solution in each side was 15 ml. The temperature of the bathing solution was kept at 37°C. The serosal bathing solution (119 mM NaCl, 21 mM NaHCO3, 2.4 mM K2HPO4, 0.6 mM KH2PO4, 1.2 mM CaCl2, 1.2 mM MgCl2, and 10 mM glucose) was gassed with 95% O2/5% CO2, pH 7.3–7.4. The mucosal bathing solution (140 mM NaCl, 5.4 mM KCl, 1.2 mM CaCl2, 1.2 mM MgCl2, 0.5 mM Hepes, pH 7.4, and 10 mM glucose) was gassed with 100% O2. The H+ secretion rate was determined from the pH change in the mucosal bathing solution and the titration curve of the solution. The pH was monitored using a combined pH electrode (GS-5015C; DKK-TOA). Histamine (1 mM) was added to the serosal side and omeprazole (200 μM), a gastric H+,K+-ATPase inhibitor, was added to the mucosal side.
This work was supported in part by a grant-in-aid for Scientific Research (B) (to Sa. Tsukita) and a grant-in-aid for Cancer Research (to Sh. Tsukita) from the Ministry of Education, Science and Culture of Japan.
Abbreviations used in this paper: ERM, ezrin/radixin/moesin; HCl, hydrochloric acid.