Claudin-18: a dominant tight junction protein in Barrett's esophagus and likely contributor to its acid resistance

Biljana Jovov, Christina M. Van Itallie, Nicholas J. Shaheen, Johnny L. Carson, Todd M. Gambling, James M. Anderson, Roy C. Orlando


Barrett's esophagus (BE) is a specialized columnar epithelium (SCE) that develops as replacement for damaged squamous epithelium (SqE) in subjects with reflux disease, and as such it is apparently more acid resistant than SqE. How SCE resists acid injury is poorly understood; one means may involve altered tight junctions (TJs) since the TJ in SqE is an early target of attack and damage by acid in reflux disease. To assess this possibility, quantitative RT-PCR for 21 claudins was performed on endoscopic biopsies on SCE of BE and from healthy SqE from subjects without esophageal disease. In SCE, Cldn-18 was the most highly expressed at the mRNA level and this finding is paralleled by marked elevation in protein expression on immunoblots. In contrast in SqE, Cldn-18 was minimally expressed at the mRNA level and undetectable at the protein level. Immunofluorescence studies showed membrane localization of Cldn-18 and colocalization with the tight junction protein, zonula occludens-1. When Cldn-18 was overexpressed in MDCK II cells and mounted as monolayers in Ussing chambers, it raised electrical resistance and, as shown by lower dilution potentials to a NaCl gradient and lower diffusion potentials to acidic gradients, selectively reduced paracellular permeability to both Na+ and H+ compared with parental MDCK cells. We conclude that Cldn-18 is the dominant claudin in the TJ of SCE and propose that the change from a Cldn-18-deficient TJ in SqE to a Cldn-18-rich TJ in SCE contributes to the greater acid resistance of BE.

  • claudin profiling
  • quantitative RT-PCR
  • paracellular permeability
  • stratified squamous epithelium

barrett's esophagus (BE) is defined as the presence of a metaplastic specialized columnar epithelium (SCE) within the tubular esophagus. It is readily identified on endoscopy by its reddish appearance macroscopically compared with the paler esophageal stratified squamous epithelium (SqE). The diagnosis of BE is confirmed on esophageal biopsy by the presence of goblet cells that stain positive with Alcian blue at pH 2.5 for acidic mucins (29). Biochemically BE is characterized by abundant expression of the intestinal protein villin and mucin 2 (20, 21, 31). The metaplastic SCE of BE arises in the setting of gastroesophageal reflux disease (GERD) likely as replacement for acid-damaged SqE, and, as such, SCE is more acid resistant than SqE and represents a form of “adaptive protection” against a hostile luminal environment (28). Clinical observations that support this concept include the fact that BE may be found in asymptomatic subjects without treatment for GERD and when stressed by esophageal acid perfusion as part of the Bernstein test experience either no symptoms or symptoms far less severe than those with GERD without BE (12, 17). In addition, BE is clinically stable for long periods of time, and this is the case irrespective of type or effectiveness of antireflux therapy (7, 32).

What contributes to the relative acid resistance of SCE in BE compared with native SqE remains unknown. However, candidate functions include the ability of SCE to secrete from its surface cells both mucins that form a viscoelastic surface layer (8) and bicarbonate that forms a more effective lumen-to-surface buffer zone for neutralization of hydrogen ions (H+) (1, 38). Another possible candidate may be the structure and function of the tight junction (TJ) in SCE since evidence suggests that in GERD the TJ of SqE is an early target for attack and damage by luminal acid. The result of this attack on SqE is an increase in paracellular permeability and the development of a lesion known as “dilated intercellular spaces.” Indeed dilated intercellular spaces in SqE are now recognized as an early and important feature of both the erosive and nonerosive forms of GERD (3, 4, 36, 40, 46). Since the TJs in SqE appear vulnerable to acid damage and the SCE of BE relatively more resistant to such injury, we hypothesized that the TJ of SCE may account, at least in part, for the greater acid resistance of BE.

The permeability of ions and uncharged molecules across the TJ is highly dependent on the nature of its bridging proteins, and these proteins are predominantly members of the multigene family known as the claudins. Consequently, we initially approached the hypothesis posed that the TJ of SCE is more acid resistant than that of SqE by performing a claudin gene expression profile on SCE in BE and comparing it to that of healthy SqE from subjects without esophageal disease. Not surprisingly, given the differences in phenotype, the profile for SCE differed dramatically from that of SqE. What was surprising, however, was that the claudin profile of SCE was quantitatively dominated by a qualitatively unique claudin: claudin-18 (Cldn-18). Therefore, we performed immunoblots and immunofluorescence microscopy (IF) to identify and characterize the distribution of Cldn-18 protein in SCE and overexpressed Cldn-18 in MDCK II cells to assess in Ussing chambers its effect on TJ function.



Two to three esophageal biopsies were obtained using large cup biopsy forceps from 20 adult subjects, ages 18–75 yr old, with either nondysplastic BE or history of BE with transient low-grade dysplasia. These subjects were undergoing upper endoscopic surveillance for cancer. Similarly healthy human SqE was obtained using large cup biopsy forceps from 12 subjects being endoscoped for clinical reasons. These subjects had no history of reflux symptoms or esophageal disease and a grossly normal esophagus on endoscopy. Patients gave written, informed consent prior to the procedure, and the study was approved by the Human Research Ethics Committee of University of North Carolina at Chapel Hill.


Quantitative RT-PCR (qRT-PCR) was performed on endoscopic biopsies using previously described methods (15). In brief biopsies underwent total RNA isolation using RNeasy kits (Qiagen, Valencia, CA) per the manufacturer's recommended protocols. RNA was treated with TURBO DNase (TURBO DNA-free kit, Ambion, Austin, TX) to remove contamination by genomic DNA. cDNA was synthesized from 2.5 μg of treated RNA for each tissue sample by using Superscript III reverse transcriptase (Invitrogen, Carlsbad, CA) with an equal amount of RNA included in a no-RT control for each separate RNA sample. Real-time PCR primers used in this study were validated primer sets (QuantiTect Primers Assays, Qiagen). Primer-specific details such as assay location, transcript detected, ensemble transcript identification, and amplicon length can be found at Real-time PCR was performed with 1:25 dilutions of the cDNA (in triplicate) with and no-RT control for each sample and as well as no template reaction controls. Reactions consisted of SYBR Green JumpStartTaq ReadyMix for quantitative PCR (Sigma-Aldrich, St. Louis, MO), premade primers, and 5 μl of sample (cDNA or control). Amplification was performed in a Rotor-Gene 3000 (Corbett Research, Mortlake, Australia) thermal cycler at 95°C for 3 min followed by 37 cycles of 94°C for 15 s, 54°C for 20 s, and 72°C for 25 s. Following amplification, a melting curve analysis was performed by heating the reactions from 50 to 99°C in 0.2°C intervals while monitoring fluorescence. The cycle at which each sample crossed a fluorescence threshold, Ct, was determined and triplicate values for each cDNA were averaged.

Eukaryotic translation elongation factor 1 alpha 1 (Eef1a1) served as a control gene for normalization between samples. It was included in each cycling run and demonstrated consistency among runs (14). Zonula occludens (ZO)-1 showed a constant relationship to Eef1a1 in all samples and was also considered a nonchanging control. To simplify reporting, gene expression was normalized to ZO-1 expression by calculating a ΔCt = (Ct of ZO-1 − Ct of gene). Relative expression values were calculated as 2Math, setting the expression value of ZO-1 to 1.0. Experimental error was estimated for each gene in each tissue by comparing the coefficient of variation (CV) of the average Ct value for four samples, error = [(2%CV)/100] × [relative expression value]. If a sample's signal did not rise above threshold within 37 cycles, it was considered not detectable. Amplification efficiency for each individual reaction was monitored by the Rotor-Gene software (v.5) comparative quantification function. Ct values were not adjusted for differences in amplification efficiencies, because efficiencies were consistently close to 1.9 for all reactions (2.00 is the value of a theoretically 100% efficient reaction, i.e., doubling each cycle). Determination of expression for villin and mucin 2 served as biochemical markers to confirm the presence of SCE in biopsies from subjects with BE (20, 21, 31).

IF and immunoblotting.

Methods for indirect IF and digital image processing have been described before (19, 30). Immunolabeled fluorescent MDCK cells were imaged with a Zeiss 510 Meta confocal microscope with a ×63 1.4-numerical aperture oil immersion objective with scanning in the x-y or x-z planes using a pinhole size of 1.0 Airy unit. Filters were set up for simultaneous scanning of Cy 2 (emission 505–530 nm) and Cy 3 (LP 585 nm). Lack of bleed-through was confirmed by transiently cutting off the 543-nm excitation and noting a lack of signal in the Texas red channel.

Biopsies for IF and immunoblots were flash frozen in liquid N2 and stored at −80°C. Tissue lysates were prepared by homogenizing tissue with the TissueLyser bead mill (Qiagen) in 20 volumes of a 50 mM HEPES buffer (pH 7.4) with 1% Triton X-100, 0.05% SDS, 0.2% sodium deoxycholate, 150 mM NaCl, 1.5 mM MgCl2, 1 mM EGTA, and Complete EDTA-free Protease Inhibitor Cocktail (Roche Diagnostics, Indianapolis, IN). Cell debris was removed by a short centrifugation at 5,000 rpm. An aliquot of cleared lysate was kept for protein quantitation using the BCA Protein Assay Kit (Pierce Biotechnology, Rockford, IL), and the rest was diluted with SDS-Laemmli sample buffer. Methods for electrophoresis and immunoblotting were standard and have been described previously (18, 45). Antibodies against claudins 1, 4, and 18 and ZO-1 were purchased from Zymed Laboratories (San Francisco, CA) and antibodies against actin from Sigma (St. Louis, MO). Secondary antibodies for immunoblots were goat anti-rabbit IRDye 800 (Rockland, Gilbertsville, PA) and goat anti-mouse Alexa680 (Molecular Probes, Carlsbad, CA). Signals were detected by use of an Odyssey Infrared Imaging System (LI-COR, Lincoln, NE).

Plasmid construction.

Cldn-18 isoform A2.1 was cloned by RT-PCR from mouse duodenum RNA by using the following primers: 5′-GAATTCGCCGCCATGTCGGTGACCGCCTGC and 3′-GAATTCGCCTACACATAGTCATACTTGGTAGGATGAG, into pCR2.1 TOPO (Invitrogen); the cDNA was digested and cloned into the EcoR I site of pTRE. The Cldn-18 plasmid was cotransfected with pSVZeo into the tet-off MDCK II cell line (Clontech) and stable cell lines were selected with 1 mg/ml zeocin. Plasmid structure for Cldn-2 was previously described (5).

Electrophysiology using MDCK II cell line.

Electrophysiological characterization of Madin-Darby canine kidney (MDCK) II monolayers was carried out according to published methods (43). Stable MDCK II tet-off cell lines (Clontech Laboratories) transfected with Cldn-18 or Cldn-2 were grown on Snapwell filters (Costar; Corning Life Sciences, Acton, MA) for 4 days, induced without doxycycline or noninduced with doxycycline (50 ng/ml). Transmonolayer resistance was measured by using a modified Ussing chamber with a microcomputer-controlled voltage-current clamp (Harvard Apparatus, Holliston, MA) with buffer A (120 mM NaCl, 10 mM HEPES, pH 7.4, 5 mM KCl, 10 mM NaHCO3, 1.2 mM CaCl2, and 1 mM MgSO4) in the apical and basolateral chambers. Transmonolayer dilution potentials were measured upon dilution of the apical chamber (buffer B: 60 mM NaCl, 120 mM mannitol, 10 mM HEPES, pH 7.4, 5 mM KCl, 10 mM NaHCO3, 1.2 mM CaCl2, and 1 mM MgSO4) relative to the basolateral chamber (buffer A) (43). Dilution potentials were immediately stable and repeatedly measured (every 6 s) for at least 30 s after buffer A had been replaced with buffer B in the apical chamber. Voltage and current electrodes consisted of a Ag-AgCl wire in 3 M KCl saturated with AgCl housed in a glass barrel with a microporous ceramic tip (Harvard Apparatus). Liquid junction potentials were calculated by using the Henderson diffusion equation for univalent ions as previously described (43).

Apical acidification protocol.

Cldn-18-transfected and nontransfected MDCK II cells were grown as described above. Three replicate filters from each group were placed in modified Ussing chambers with buffer A in both apical and basolateral chambers. Baseline readings of the electrical potential difference (PD) and conductance were obtained, and then the apical solution was changed (with one wash) to buffer A (above) that was previously equilibrated to pH 3.5 with concentrated HCl. At 120 min, the apical solution was again changed (with one wash) to solution A equilibrated to pH 2.5 with concentrated HCl. PD and conductance readings were obtained immediately after changes into the lower pH solutions and at 15- to 30-min intervals thereafter. The values of the junction potential were obtained by measuring the voltage across blank filters in the pairs of experimental buffers; these values were subtracted from subsequent measurements made on cell monolayers to determine the PD.


Claudin gene expression profiles in SCE and SqE.

qRT-PCR was performed for 21 claudins on endoscopic biopsies containing SCE from four subjects with Barrett's esophagus and containing healthy SqE from four subjects without esophageal disease. The 21 claudins tested included Cldn-1 to Cldn-12, Cldn-14 to Cldn-20, Cldn-22 and Cldn-23 (note: validated primer sets were unavailable for Cldn-13 and Cldn-21); and all 21 were expressed to varying degrees in SCE (Fig. 1). As shown in Fig. 1A, compared with ZO-1 serving as reference standard of 1.0, Cldn-4 and Cldn-18 were highly expressed at >1.0; Cldn-1, Cldn-12, and Cldn-23 were moderately expressed at 0.5–1.0; and all other claudins were considered to have low expression at levels significantly below 0.5. Notably, Cldn-18 expression was dominant in SCE and at levels that dwarfed all others (Fig. 1A, inset), including that of Cldn-4, which, although highly expressed in SCE, was more than fourfold lower than Cldn-18.

Fig. 1.

A: Claudin gene expression profiles for human esophageal specialized columnar epithelium (SCE), i.e., Barrett's esophagus (BE), and for healthy human esophageal stratified squamous epithelium (SqE; Sq.). Claudin expression levels are referenced to expression levels for zonula occludens-1 (ZO-1), which is set at 1.0, and the y-axis is broken into 2 scales to accommodate the broad range. Note the large numbers of claudins expressed in both SCE and SqE and, as emphasized in the inset, the overriding dominance of expression for claudin-18 over that of all other claudins in both SCE and SqE. Error bars = [(2%CV)/100] × [relative expression]. CV, coefficient of variation; NT, not tested; ND, not detected. B: comparative gene expression for mucin 2, villin, and claudin (Cldn)-18 in human gastric cardia epithelium and Barrett's SCE. Villin and mucin 2 are shown to be undetectable in stomach and to be highly expressed in SCE. In contrast, Cldn-18 is highly expressed in both epithelial types, with stomach expressing double amount of Cldn-18 expressed in SCE.

Similar to SCE, a large number, 19 of the 21 claudins tested, were expressed in SqE. However, unlike SCE, none of the 19 claudins in SqE were considered highly expressed since expression values in SqE were <1.0 (Fig. 1A). Indeed, only Cldn-4, the most highly expressed in SqE, reached a level compatible with moderate expression whereas Cldn-1 and Cldn-23 in SqE had expression levels considered low. The remaining claudins in SqE had very low levels of expression or, as for Cldn-2 and Cldn-6, were essentially undetectable. Also, and in stark contrast to SCE, Cldn-18 expression in SqE was barely detectable at the mRNA level (Fig. 1A).

As noted in materials and methods, villin and mucin 2 are established markers of SCE and their presence by qRT-PCR used to ensure that the biopsies analyzed for claudins were derived from BE and not from proximal stomach (20, 21, 31). Gastric epithelium, like SCE, however, is reported to have high levels of Cldn-18 (27). For this reason, we obtained endoscopic biopsies from the gastric cardia of three subjects and compared the results of qRT-PCR for villin, mucin 2, and Cldn-18 to those of Barrett's SCE. The results, as shown in Fig. 1B, confirm the differential expression of mucin 2 and villin in these tissues, with SCE having abundant villin and mucin 2 and gastric cardia epithelium being devoid of these genes. Despite these differences, however, both SCE and gastric epithelia were found to exhibit very high levels of expression for Cldn-18 (Fig. 1B).

Claudin protein expression in SCE and SqE.

Given the dominance of Cldn-18 expression at the mRNA level in SCE and minimal expression in SqE, we compared the level of protein expression in SCE and SqE by immunoblot. In addition to Cldn-18 we also determine protein expression levels of two other claudins (Cldn-1 and Cldn-4) readily detected at the mRNA level in SCE and SqE (Fig. 2A). As shown in Fig. 2A, protein expression levels for these three claudins paralleled their level of gene expression, with Cldn-18 protein again dominant in SCE and undetectable in SqE. Notably, the smearing of the blot for Cldn-18 in Fig. 2A was a reflection of the high level of protein expression rather than lack of antibody specificity, a conclusion supported by the high level of protein expression for Cldn-18 in SCE on a second blot in which we added only 20% of the protein load used in Fig. 1A. As shown in Fig. 2B, the lower protein load resulted in a single clean bend of appropriate size for Cldn-18.

Fig. 2.

Immunoblots of Cldn-1, Cldn-4, and Cldn-18 of tissue lysate prepared from SCE of BE, or from healthy human SqE. A: 10 μg of protein are loaded in each lane. Note: protein levels correlate with relative expression of transcripts (see Fig. 1A). Cldn-1 has similar protein expression level in SqE and BE, Cldn-4 has a higher level of expression in BE than SqE, and Cldn-18 is highly and exclusively expressed in BE. Molecular weight standards are displayed on the left in kDa. Prestained SDS-PAGE standard (wide range; Bio-Rad) was used (red bands). This immunoblot is representative of 3 separate experiments performed using tissue from different patients. B: immunoblot of Cldn-18 from BE tissue lysate. By reducing the total protein load to only 2 μg per lane, the band for Cldn-18 is now shown to be single and clean.

Cldn-18 localization to the cell membrane and TJ in SCE.

On the basis of confirmation that the high level of gene expression for Cldn-18 was mirrored by high levels of protein expression, immunolocalization and colocalization studies were performed in SCE for Cldn-18 and ZO-1, a protein known to be localized to the TJ (Fig. 3). As shown in Fig. 3A, ZO-1 was localized predominantly to the apical cell membranes, including TJ regions, of SCE whereas Cldn-18 was present in both apical and basolateral cell membranes of SCE (Fig. 3B). Indeed, and supporting a role for Cldn-18 within the TJ was its colocalization with ZO-1 in the apical membranes (Fig. 3C).

Fig. 3.

Immunofluorescent localization of the tight junction protein, ZO-1, and Cldn-18 in esophageal SCE from BE. A: ZO-1 localized to the apical cell membranes containing the tight junction region in SCE. B: abundantly expressed Cldn-18 localized to both apical and basolateral cell membranes in SCE. C: dual immunofluorescence demonstrates that Cldn-18 colocalizes with ZO-1 within the apical cell membrane and tight junction regions of SCE. Signals are pseudocolored green for ZO-1 and red for Cldn-18. Cell nuclei are stained blue with DAPI. Colocalization of ZO-1 and Cldn-18 results in the yellow color. Bar, 20 μm.

Expression of Cldn-18 in MDCK II cells influences TJ function.

Given the quantitative dominance of Cldn-18 in SCE and its localization in the TJ, we sought its effects on TJ function. This was done by transfection and expression of Cldn-18 in MDCK II cells using a previously described method (44). Cldn-18 is undetectable in uninduced MDCK II cells (Fig. 4A). Following transfection and induction, Cldn-18 expression could be demonstrated in the cell membranes of MDCK II cells (Fig. 4A) and its presence within the TJs confirmed by showing its colocalization with ZO-1 using confocal microscopy (Fig. 4B).

Fig. 4.

A: immunofluorescent localization of ZO-1 and Cldn-18 in uninduced and induced Madin-Darby canine kidney (MDCK) II cells. Note that uninduced MDCK II cells lack immunostaining for Cldn-18, whereas those induced to express Cldn-18 showed staining within the plasma membrane and in intracellular vesicular compartments. ZO-1 localization was unchanged by induction of Cldn-18 expression. Bar, 20 μm. B: z-plane section of immunofluorescent colocalization of ZO-1 and Cldn-18 in induced MDCK II cells. Top: dual staining for ZO-1 (red) and Cldn-18 (green) and their colocalization (yellow). Middle: ZO-1. Bottom: Cldn-18.

Electrophysiologically, induction of Cldn-18 increased transmonolayer electrical resistance fourfold compared with uninduced monolayers and decreased dilution potentials from +8 mV in uninduced monolayers to essentially zero for Cldn-18 expressing monolayers (Fig. 5, A and B). Furthermore, by calculation from the foregoing data, the observed decrease in dilution potentials by Cldn-18 expression was shown to reflect a selective decrease in paracellular permeability to cations (Na+), with no significant effect on anion (Cl) permeability (Fig. 5C). Note: It was shown previously that treatment of MDCK cells with doxycycline alone had no effect on paracellular permeability (6).

Fig. 5.

Electrophysiological effects of Cldn-18 expression in monolayers of MDCK II cells. A: resting transmonolayer epithelial resistance (TER) increases 4-fold, i.e., from 38.5 ± 2.3 Ω·cm2 in uninduced cells to 185 ± 23 Ω·cm2 following induction of Cldn-18. B: dilution potentials (in mV) following apical NaCl dilution from 120 to 60 mM declines from +8 mV for uninduced cells to effectively zero in the Cldn-18 induced matched clones. Note the positive dilution potentials for uninduced MDCK II cells reveal their cation selectivity and that this is effectively eliminated following expression of Cldn-18 creating a nonselective monolayer. C: calculation of ion permeabilities for Na+ and Cl reveals that the change in dilution potentials following Cldn-18 induction is due to a selective decrease in Na+ permeability (solid bars, compared with uninduced, open bars) with no change in Cl permeability. Average of 2 clones and duplicate measurements on each; means ± SE. D: following apical acidification (addition of HCl) to pH 3.5 a diffusion potential (in mV) develops in uninduced (▪) cells for 2 h raising the transmonolayer potential difference (PD) from −0.5 to −3.0 mV whereas the PD for Cldn-18 induced cells remains essentially unchanged (▴). Further apical acidification (addition of HCl) to pH 2.5 results in further increases of PD in uninduced cells to a maximum of −4.4 mV, whereas the PD for Cldn-18 induced cells increases to a maximum of −2.2 mV. At pH 2.5, the differences in diffusion potentials for uninduced and induced cells persists for almost 2 h. Values are means ± SE, n = 3, *P < 0.05.

In vivo, however, the cation of concern is the H+ ion since it is luminal exposure to refluxed gastric acid that poses the greatest risk of damage to esophageal epithelium irrespective of type. For this reason, we also assessed whether Cldn-18 expression reduced paracellular permeability to H+ as well as Na+. This was done by monitoring the diffusion potentials produced when MDCK II monolayers were apically exposed to solutions acidified with HCl to pHs 3.5 and 2.5. Following apical acidification to pH 3.5, a diffusion potential developed in uninduced monolayers but not in monolayers expressing Cldn-18, whereas apical acidification to pH 2.5 resulted in even higher diffusion potentials in the uninduced monolayers (2–3-fold greater) than those observed in Cldn-18 expressing monolayers. Given these observations, we next determined if the effect of Cldn-18 on the H+ diffusion potential was relatively specific by assessing whether it could be reproduced by overexpression of another claudin in MDCK II cells. Induction of Cldn-2, however, resulted in a H+ diffusion potential at pH 3.5 that was both indistinguishable from uninduced MDCK II cells (−1.9 ± 0.4 vs. −1.6 ± 0.3 mV at 2 min after acidification; n = 3) and similar in value to that observed in Fig. 5D for uninduced MDCK II cells.


TJs are the most apical of the cell-cell contacts in epithelia. They are known to function in cell-cell adhesion, membrane polarity, and regulation of ion and uncharged (aqueous) molecule passage through the paracellular pathway (11). On freeze-fracture replicas, TJs appear as a complex network of strands that bridge the space between the membranes of adjacent cells, with the strands themselves composed of proteins (34). Recent studies indicate that the majority of bridging strands are formed by members of a multigene family known as claudins (10, 24, 25, 41, 42). Claudins are small proteins with molecular weights ranging from 20 to 27 kDa (42). They have four transmembrane-spanning domains, two extracellular loops, and cytoplasmic -NH2 and -COOH termini (41). Currently at least 24 claudins have been identified in mammals. Within the epithelial TJ, homophilic and heterophilic interactions occur among the claudins as well as with other TJ proteins such as occludin and junctional adhesion molecule (2, 41). Furthermore, through PDZ (postsynaptic protein-95/discs large/zonula occludens-1) domains on the cytoplasmic COOH-terminal end, the claudins bind to members of the membrane-associated guanylate kinase family, ZO-1, ZO-2, and ZO-3; and through interactions with this latter group, they are connected to and communicate with the actin cytoskeleton (9, 16).

Quantitative and qualitative differences among the claudins account for much of the diversity in transepithelial resistance (TER) and (perm)selectivity of the TJ (44); and such differences also can account for varying degrees of vulnerability of the TJ to disease. For instance, the pathological effects of Clostridium perfringens, a common cause of foodborne gastroenteritis, is due to an enterotoxin with the capacity to bind to Cldn-3 and Cldn-4. By binding these claudins, other parts of the enterotoxin are then positioned to interact with and create small pores within the cell membrane. It is the pores that account for loss of cell osmoregulation and ultimately for cell death (22, 35). In similar fashion, then, differences among claudins within the TJ may theoretically make an epithelium either more or less vulnerable to damage from luminal acid.

In the present investigation, we compared the claudin gene profiles for two esophageal epithelia: one, SCE in BE, is considered significantly more acid resistant than the other, native SqE. Moreover, it has been shown that these two esophageal epithelial types have very different and distinctive functional characteristics with respect to ion transport and TER (39). And so, not surprisingly, given such differences in phenotype and function, the profiles were very different. For instance, SCE had two claudins, Cldn-4, and Cldn-18, that were expressed to levels that exceeded that of the reference standard, ZO-1, and three claudins, Cldn-1, Cldn-12, and Cldn-23, that came close to the level of expression of the reference standard. In contrast, and despite expressing 19 different claudins, none of the claudins expressed in SqE exceeded that of the reference standard, ZO-1, and only one, Cldn-4, was expressed at a level that was close to that of the reference. Even more dramatic, however, was the unquestioned quantitative dominance in SCE of a qualitatively unique claudin, i.e., Cldn-18. Cldn-18 not only exceeded the reference standard in SCE but exceeded the only other highly expressed claudin in SCE, i.e., Cldn-4, and this by greater than fourfold. In contrast, Cldn-18 was only barely detectable at the gene level in SqE and not at all at the protein level. Moreover, the literature indicates that Cldn-18 is abundantly expressed in mouse and human gastric epithelium (27), expressed to a limited extent in duodenum, and almost nonexistent throughout the rest of the small and large bowel (15).

Cldn-18 to our knowledge has not been described in BE, although other investigators have found in BE varying levels of gene and/or protein expression for Cldns 1–5 and Cldn 7 (13, 23, 26). Indeed, mouse Cldn-18 has only recently been isolated and characterized by Niimi et al. (27) and shown to have two isoforms, one lung specific and the other stomach specific; and each isoform has two transcripts, one for the full-length claudin and the other for a COOH-terminal truncated form of the claudin. COOH-terminally truncated isoforms were not found for human Cldn-18 (27, 33). Niimi et al., when searching for genes involved with pulmonary morphogenesis, noted that the lung-specific, but not stomach-specific, isoform of Cldn-18 was a downstream target gene for T/EBP/NKX2.1, a homeodomain transcription factor (27). To date, however, there is no known transcription factor for stomach (or SCE)-specific Cldn-18.

In the present investigation dominance of Cldn-18 expression at the mRNA level was also mirrored by dominance of protein expression in SCE. Further IF demonstrated that Cldn-18 was distributed within both apical and basolateral cell membranes within SCE. Moreover, and supporting a role in the TJ, Cldn-18 colocalized with the TJ protein, ZO-1. A similarly diffuse cell membrane pattern for Cldn-18 distribution within gastric epithelium was also observed by Niimi et al. (27). Additionally, they documented by immunogold staining that Cldn-18 was localized to the TJ strands on freeze fracture and suggested that excess Cldn-18 within the membrane served as a reservoir for its shuttling to and from the TJ.

Given the quantitative dominance of Cldn-18 within the TJ of SCE, it is likely to play a prominent role in TJ function; one critical function of the claudins within TJs is their ability to regulate the ion species traversing the paracellular pathway (37). Indeed Colegio et al. (6) have shown using site-directed mutagenesis that a change of charge from positive to negative on one amino acid within the extracellular domain of claudin-4 could increase paracellular permeability to Na+, and changing the charge on three amino acids from negative to positive in claudin-15 could reverse the ion preference of the paracellular pathway from Na+ to Cl. In the present study, Cldn-18 was expressed in MDCK II cells and shown by IF to be localized to the cell membrane and the TJ. Notably, and consistent with the ability of claudins to alter both the electrical resistance and ion conductance of the TJ (5), Cldn-18 expressing monolayers in Ussing chambers exhibited higher TER and lower NaCl dilution potentials than uninduced monolayers of MDCK II cells. Furthermore, since these dilution potentials were similar irrespective of sidedness of the gradient, the increase in TER and reduction in conductance with Cldn-18 expression reflected changes in paracellular, as opposed to transcellular, permeability. Moreover, and based on the direction of the conductance change, the reduction in paracellular permeability reflected a selective decline in cation conductance. It is also of interest that Cldn-4 which, like Cldn-18, is highly expressed in SCE has been shown to increase TER and lower paracellular permeability for cations (specifically Na+) in MDCK II cells (44). This suggests the emergence of a possible theme in which the claudin expression profile of SCE reflects a need for enhanced protection against the potentially noxious effects of luminal cations; see below.

Furthermore, and of potential clinical importance, the ability of Cldn-18 expressing monolayers to impede paracellular permeability of cations, extended to H+, these being the smallest of cations and the ones that pose the greatest risk of damage to esophageal epithelia in vivo. In the present experiments, resistance to the paracellular movement of H+ was demonstrated in Cldn-18 expressing MDCK II monolayers by the absence of a H+-induced diffusion potential upon apical acidification to pH 3.5 and lower diffusion potentials at apical pH 2.5 than observed in uninduced MDCK II monolayers. Moreover, this resistance to H+ diffusion by Cldn-18 expression was relatively specific, since it was not observed by expression of Cldn-2 and was long lasting rather than transient. The lack of a diffusion potential was sustained for 2 h at pH 3.5 and reduced significantly below that of uninduced monolayers for ∼2 h at pH 2.5. These observations, coupled with the fact that the TJ appears to be a weak link in the defense of SqE against damage by luminal acid (29), support the hypothesis that the presence of Cldn-18 within the TJ of SCE contributes to the acid resistance of BE. Consistent with this conclusion is the observation that the only other digestive epithelia, gastric and duodenal, where Cldn-18 is noted to be prominent are those that are exposed to and so require a strong defense against luminal acid (27).

In summary, Cldn-18 is the dominant claudin of SCE in BE. It is distributed widely within the cell membrane and colocalizes with the TJ protein, ZO-1. Among its properties Cldn-18 selectively decreases the permeation of cations through the paracellular pathway and this decrease extends to H+. We propose that the change from a Cldn-18-deficient TJ in SqE to a Cldn-18-rich TJ in SCE contributes to the greater acid resistance of BE.


This research was supported by National Institute of Diabetes and Digestive and Kidney Diseases grants DK-036013, DK-063669, and DK-45134.


We are thankful to Dr. McNaughton Kirk for the superb cut of fresh frozen tissue used for our immunolocalization studies.


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


View Abstract