Expression and Distribution of Claudin Subtypes in Human Corneal Endothelium

Abstract

Purpose.: To investigate the expression pattern of claudins in human corneal endothelium, and to evaluate the functional role of the claudin-10b subtype.

Methods.: Corneal endothelium with Descemet's membrane and the corneal epithelium were stripped from donor human corneal stroma. Reverse transcription–polymerase chain reaction (RT-PCR) was performed to evaluate the claudin subtypes expressed in corneal endothelium, stroma, and epithelium. Immunohistochemistry was performed to confirm the expression of claudin subtypes in corneal endothelium, and the expression pattern was compared to that of corneal epithelium. Finally, transendothelial resistance (TER), short-circuit current (SCC), and potential difference (PD) were measured in human corneal endothelial cell line B4G12 cells with or without claudin-10 small interfering RNA (siRNA) transfection by Ussing chamber system.

Results.: Transcripts for claudin-1, ‐2, ‐3, ‐4, ‐7, ‐10b, ‐11, ‐15, ‐22, ‐23, and ‐24 were identified in corneal endothelium sample by RT-PCR. Immunohistochemistry confirmed the expression of claudin-1, ‐2, ‐4, ‐7, ‐10, ‐11 ‐15, ‐22, and ‐23 in corneal endothelium. In corneal stroma, claudin-1, ‐2, ‐3, ‐4, ‐5, ‐6, ‐7, ‐8, ‐10b, ‐11, ‐12, ‐14, ‐15, ‐22, ‐23, and ‐24 were identified by RT-PCR. In corneal epithelium, claudin-1, ‐3, ‐4, ‐7, ‐11, ‐14, and ‐23 were identified by immunohistochemistry and RT-PCR. Downregulation of claudin-10b by siRNA resulted in the decrease of SCC and PD, but not TER, in B4G12 cells.

Conclusions.: The expression pattern of claudin-10b⁺/claudin-14⁻ was specific in corneal endothelium among the three corneal layers. Claudin-10b may play an important role in the tight junction of corneal endothelium.

Introduction

The human corneal endothelium is a single layer of hexagonal cells with Descemet's membrane covering the posterior surface of the cornea in a well-arranged mosaic pattern. The primary function of corneal endothelium is to maintain corneal transparency by regulating corneal hydration and nutrition through a “leaky” barrier and metabolic pump function. As the cornea is avascular, glucose and other solutes for the cornea must diffuse from the aqueous humor across the corneal endothelium. In the early 1970s, Maurice first presented this well-known “pump and leak” hypothesis. ^1–3 However, the molecular mechanisms underlying regulation of the balance between the pump and leak functions remain unknown. 

Tight junctions restrict the flow of ions and aqueous molecules between cells by forming a selective barrier to the paracellular pathway. Permeability of the tight junctional barrier is determined by claudins. ⁴ Since the novel discovery of claudins by Tsukita and his colleagues, claudin family members have been found to play critical roles in maintaining the integrity of epithelial and endothelial tight junctions. ^5–7 To date, 24 members of this gene family have been identified. The claudin family form complex patterns in various organs, with distinct expression patterns in a tissue-specific and cell type–specific manner. ^1,8–10 Moreover, claudins regulate the ion- and size-selective pores through cell barriers and therefore influence paracellular charge selectivity. ¹¹  

Since specific markers for corneal endothelial cells have not been identified, the combination of claudin subtype expression may be used as corneal endothelial cell marker(s). Previous studies have revealed the claudin subtypes of human corneal epithelium. ^12–15 However, the subtypes of claudin in corneal endothelium have not been reported. In this study, we determined the expression pattern of claudin subtypes, and we also analyzed the functional role of claudin-10b in human corneal endothelium. 

Materials and Methods

Human Corneal Tissue Preparation

Human corneas were obtained from Sight Life (Seattle, WA) and preserved for experiments in Optisol-GS (Bausch & Lomb, Rochester, NY) at 4°C after central corneal buttons were used for transplantation. The donor was a 65-year-old male, and the preservation time was less than 10 days. The corneal rim tissue was placed under a stereoscopic microscope (MZ16; Leica Microsystems, Wetzlar, Germany) with the endothelium side up, and the endothelium and Descemet's membrane were carefully dissected from the cornea along Schwalbe's line. The stripped corneal endothelium was fixed with 4% paraformaldehyde (PFA) and preserved at 4°C for immunohistochemical staining. Some tissue samples were used for RNA extraction. Next, conjunctiva and excess cornea were surgically removed from the rest of the corneoscleral rim. Corneal epithelium was totally isolated by treatment with Dispase II (1.2 U/mL in Dulbecco's modified Eagle medium [DMEM]/F-12; Roche, Basel, Switzerland) in a 35-mm dish (Iwaki, Tokyo, Japan) for 1 hour at 37°C, followed by dissociation using a Cell Scraper (Iwaki) for RNA extraction. The remaining corneal stroma was cut into small pieces and digested, and then placed in 0.1 mg/mL collagenase type IA (Sigma-Aldrich, Dorset, UK) in basal culture medium DMEM/F-12 (Life Technologies Corporation, Carlsbad, CA). The tissue was incubated overnight (37.0°C, 5% CO₂, 95% humidity) and filtered through a 40-μm mesh (Cell Strainer; BD Biosciences, San Jose, CA) to remove the remnants of any undigested tissue. The extracted stromal cells were used for RNA extraction. 

Our research protocols followed the tenets of the Declaration of Helsinki. The approval of the Keio University Ethics Committee was obtained for the use of human materials for this research. 

RNA Isolation and Reverse Transcription–Polymerase Chain Reaction (RT-PCR)

Total RNA was extracted from stripped human corneal epithelium and endothelium using a commercial RNA isolation kit (RNeasy kit; Qiagen, Valencia, CA), and cDNA was synthesized using another kit (RevaTra Ace; Toyobo, Osaka, Japan). The amount of cDNA was amplified by PCR (Veriti 96-Well Thermal Cycler; Applied Biosystems, Foster City, CA) for each primer pair shown in Table 1. Complementary DNA of human iPS cell line 201B7 induced by transducing Oct3/4, Sox2, Klf4, and c-Myc, ¹⁶ which was obtained from the RIKEN Cell Bank (Tsukuba, Japan), and purchased cDNA from human kidney tissue samples (Human Kidney QUICK-Clone cDNA; Clontech, Mountain View, CA) were used for positive control for the primers (Fig. 1D). Polymerase chain reaction products were examined by 2% agarose gel and ethidium bromide staining. Since claudin-13 and ‐21 are not expressed in human cells, we omitted claudin-13 and ‐21 from analysis. 

Figure 1

 

Messenger RNA expression of claudins in human corneal epithelium (A), stroma (B), and endothelium (C) by RT-PCR. (D) Positive control for claudin RT-PCR using human iPS cells and human kidney cells. Multiple bands observed in the claudin-1 lane are artifacts.

Table 1

 

The Sequences of Primers Used for RT-PCR

Claudin Subtype	Forward	Reverse	Size, bp
1	TCCACTGAACAAAACCTACGC	TGAAAAGCAACACCAAAACG	451
2	ACACACAGCACAGGCATCAC	TCTCCAATCTCAAATTTCATGC	319
3	CACGCGAGAAGAAGTACACG	TGAGGTTTCACAGTCCATGC	345
4	CACCCTCCTCTGGATATTGG	TCCCAGGCTCATTAGTGTCC	485
5	GTTTTACGACCCGTCTGTGC	AGTGGCAGGAGAAGGTCAGC	215
6	CACCTGTGTGGAGGAGAAGG	CAGGAGGCAGAAACAAAAGG	448
7	CCCTCCACCTTTTGTTTGCC	TACCCAGCCTTGCTCTCATTCC	821
8	GGCTAAAACTCATCCCAGAGG	AAGCCAAGAAGGACATCACG	306
9	TTCATCGGCAACAGCATCGT	GCCCAGCCCAGGTAGAGGGA	403
10a	GCGGCGCGACATGTCCAGG	CGAGCTCTTTTAGACATAAGC	700
10b	TCATCGCCTTCATGGTCTCC	TCCGTTGTGATTTTGTTTGC	411
11	TCGTGAGCTTTGGCTACTCC	TAATAGGGGCAACACCAAGG	418
12	AGTGTGGGCGAGTAAAATGC	GTCACTGCTCCCGTCATACC	401
14	TGTACCTGGGCTTCATCTCC	CCTCGCATTCACATTATTTCC	364
15	TGACAGAGCAGAAGCTCAGG	ACTGACGGATACAGCCAAGG	412
16	TTTGGATTTCTCACCCTGCTC	TGTGCGAGGGGCTGAGTATGA	398
17	TAGCTCCTTGTTGGCTCTCC	CCCTTGCTTCTTTCTGTTGC	381
18	TTCCATCCCAGTACCAAAGC	GTGGCCTTCAACTCAAATGG	488
19a	GGGAGGAATGACACAAATGG	CTGGGATTACAAGCGTGAGC	485
19b	CCAGGCATACACACAGATGC	AACTCAGACCCTCCCTGTCC	308
20	TTCATCCTGGCCTTATCTGG	GGCTCTTGGTTTCTCTGTCC	315
22	GCGGCATTATTTCCCTATCC	ACACAGGTTTGCCAGAGTCC	359
23	GGATCAGGGTCCTCTTAGGG	GTTCTCCTCTTGGCTTCTGG	458
24	CACACTGGAAGAACCTCAACC	ACTCCTGAACCGTCTTGTGG	350
GAPDH	GTCAAGGCTGAGAACGGGAA	GCTTCACCACCTTCTTGATG	613

 GAPDH, glyceraldehyde-3-phosphate dehydrogenase.

Immunohistochemistry

Human corneal endothelium and Descemet's membrane were carefully dissected from the cornea along Schwalbe's line by forceps, expanded on glass slides, and fixed at room temperature for 10 minutes in 4% PFA in PBS. After three PBS washes, the specimens were incubated for 15 minutes in Morphosave (Ventana Medical Systems, Tucson, AZ). After two PBS washes, the specimens were incubated for 30 minutes in blocking solution consisting of normal donkey serum (10%), Triton X-100 (0.1%) in PBS for 1 hour at room temperature to block nonspecific binding. This was followed by overnight incubation at 4°C with the appropriate dilution of each primary antibody. 

The primary antibodies used in this study were the following polyclonal antibodies: rabbit anti-claudin-1, rabbit anti-claudin-2, rabbit anti-claudin-3, rabbit anti-claudin-4, rabbit anti-claudin-7, and rabbit anti-claudin-23, purchased from Abcam (Cambridge, UK). Rabbit anti-claudin-11, goat anti-claudin-22, and goat anti-claudin-24 polyclonal antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Rabbit anti-claudin-10, rabbit anti-claudin-12, and rabbit anti-claudin-15 polyclonal antibodies were purchased from LifeSpan BioSciences (Seattle, WA). Mouse anti–zonula occludens-1 (ZO-1) monoclonal antibody was purchased from Life Technologies Corporation. 

All primary antibodies were diluted according to the manufacturers' recommendations. Immunoreactivity of primary antibodies was visualized with secondary antibodies conjugated with Cy3 (Jackson ImmunoResearch Laboratories, West Grove, PA) and incubated for 1 hour at a 1:500 dilution. After two PBS washes, specimens were mounted on dishes with an antifading mounting medium containing 4′,6-diamino-2-phenylindole (DAPI; Dojindo Laboratories, Kumamoto, Japan). Imaging of the stained samples was done using a microscope (Axio Imager; Carl Zeiss, Inc., Thornwood, NY) equipped with a digital camera (Axiocam; Carl Zeiss, Inc.). Since appropriate human tissue for positive control was unavailable, positive control for immunohistochemistry was performed with mouse kidney specimens (Supplementary Fig. S1). 

Cell Culture

Human corneal endothelial cell line B4G12 cells, which were immortalized by SV40 large T- and small T-antigens, were purchased from DSMZ (Braunschweig, Germany) and cultured in Human Endothelial-SFM (Life Technologies Corporation) supplemented with 1.0% fetal bovine serum and 10 ng/mL FGF-2 under a humidified atmosphere of 5% CO₂ at 37°C. 

Medium was changed every 2 to 3 days. B4G12 cells reached semiconfluence in a week, were replated on 35-mm dishes or Snapwell inserts (Corning, Lowell, MA), and were cultured for subsequent experiments. All dishes and flasks used for culture were polystyrene, noncoated vessels obtained from Asahi Techno Glass (Tokyo, Japan). For the Ussing chamber study, cells were dissociated into single cells, suspended at a cell density of 2 × 10⁵ cells/cm², plated, and cultured for an additional 9 days on 0.1% gelatin-coated polycarbonate Snapwell inserts with a membrane pore size of 0.4 μm. 

Inhibition of Claudin Expression by Small Interfering RNA

Small interfering RNA (siRNA) against claudin-10 (final concentration 5 nM; Santa Cruz Biotechnology) and a control siRNA (5 nM; Santa Cruz Biotechnology) were transfected into B4G12 cells using HiPerfect Transfection Reagent (Qiagen, Mississauga, ON, Canada) according to the manufacturer's instructions. Cells were seeded on 24-well or Snapwell inserts (Corning) for 7 days prior to the transfection. Cells on 24-well inserts were harvested 48 hours after transfection; total RNA was isolated using RNeasy kit (Qiagen), and cDNA was synthesized using RevaTra Ace (Toyobo). Cells on Snapwell inserts were used for subsequent barrier and pump function analysis by Ussing chamber system. Quantitative RT-PCR analysis was performed to check the expression level of claudin-10b by SYBR Green enzyme mixture (Thunderbird SYBR qPCR mix; Toyobo) and thermal cycler (Step One Plus; Applied Biosystems). Primer pairs are listed in Table 2. The expression levels of claudin-10b in cells transfected with nonspecific control siRNA were used as controls. 

Table 2

 

The Sequences of Primers Used for Quantitative RT-PCR

Gene	Forward	Reverse	Size, bp
Claudin-10b	ACGGCACGGTCATCACAACC	GACAGCAGCGATCATAAGTCC	154
ATP1A1	CCAAAGACAGGGTGCTATCG	TTGCTTGGACACATCTGAGC	117
GAPDH	AACGGATTTGGTCGTATTGG	CATGGGTGGAATCATATTGG	135

In Vitro Measurement of Short-Circuit Current, Potential Difference, and Resistance by the Ussing Chamber System

The short-circuit current (SCC), potential difference (PD), and transendothelial resistance (TER) of confluent monolayers of B4G12 cells were measured with the use of an Ussing chamber as described previously. ¹⁷ Before measurements, cell densities were manually counted by microscope (CKX41; Olympus, Tokyo, Japan). The cells cultured on Snapwell inserts (Corning) were held by a specific holder (P2302) with a measurement area of 1.12 cm². The entire setup was then placed in the Ussing chamber EM-CSYS-2 (Physiologic Instruments, San Diego, CA). The cell surface side was in contact with one chamber, and the Snapwell membrane side or collagen sheet side was in contact with another chamber. The chambers were carefully filled with 4 mL Krebs-Ringer bicarbonate (120.7 mM NaCl, 24 mM NaHCO₃, 4.6 mM KCl, 0.5 mM MgCl₂, 0.7 mM Na₂HPO₄, 1.5 mM NaH₂PO₄, and 10 mM glucose bubbled with a mixture of 5% CO₂, 7% O₂, and 88% N₂ to pH 7.4). The chambers were maintained at 37°C by an attached heater. The SCC was measured by narrow polyethylene tubes positioned close to either side of the Snapwell, filled with Krebs-Ringer bicarbonate and 4% agar gel, and connected to silver electrodes. These electrodes were connected to a computer through the Ussing system EVC4000 (World Precision Instruments, Sarasota, FL) and iWorx IX/408 data acquisition system (iWorx Systems, Dover, NH). Short-circuit current was recorded by Labscribe 2 Software for Research (iWorx Systems). After the Snapwell inserts with cells were placed in the chamber, SCC was continually measured for 10 minutes until it reached a steady state. Next, the cells on Snapwell membrane were loaded with 1-mV currents through the electrodes three times, and SCC change was recorded. The TER was calculated from average SCC change and loaded voltage according to Ohm's law. 

Next, 100 μL Krebs-Ringer solution in the cell surface-side chamber was drained, and the same amount of 25 mM ouabain (the specific Na,K-ATPase inhibitor; Sigma-Aldrich) solution was added to the chamber. This process was repeated 21 times, and the final ouabain concentration in the chamber was 10.3 mM. This concentration was confirmed to have no effect on B4G12 cell viability (Supplementary Fig. S2). The SCC was continually recorded during this procedure. Finally, recorded SCC during ouabain addition was translated to PD by multiplying TER. 

Statistical Analysis

Data are presented as mean ± SD and were compared by Student's t-test or multiple t-test with Bonferroni and Dunn correction with the use of Excel 2007 software (Microsoft, Redmond, WA). A P value of <0.05 was considered statistically significant. 

Results

Expression of Claudin Subtypes in Human Cornea by RT-PCR

Transcripts for claudin-1, ‐2, ‐3, ‐4, ‐6, ‐7, ‐8, ‐14, ‐15, ‐22, ‐23, and ‐24 were identified in corneal epithelium (Fig. 1A), and transcripts for claudin-1, ‐2, ‐3, ‐4, ‐5, ‐6, ‐7, ‐8, ‐10b, ‐11, ‐12, ‐14, ‐15, ‐22, ‐23, and ‐24 were identified in the corneal stroma by RT-PCR (Fig. 1B). Transcripts for claudin-1, ‐2, ‐3, ‐4, ‐7, ‐10b, ‐11, ‐15, ‐22, ‐23, and ‐24 were identified in the corneal endothelium by RT-PCR (Fig. 1C). Claudin-10b and ‐11 were not identified in human corneal epithelium but were present in human corneal stroma and endothelium. 

Expression of Claudin Subtypes in Human Cornea by Immunohistochemistry

We investigated immunohistochemistry of claudin subtypes confirmed by RT-PCR in the corneal epithelium (Fig. 2) and endothelium (Fig. 3). Claudin-14 expression in corneal stromal cells was confirmed in horizontal sections of the corneal stroma (Figs. 2B, 2C). Positive controls for each primary antibody are shown in Supplementary Figure S1. Among these, expression of claudin-1, ‐2, ‐4, ‐7, ‐10, ‐11 ‐15, ‐22, and ‐23 was confirmed in the corneal endothelium. Expression of claudin ‐3, ‐12, and ‐24 was not observed, suggesting that these claudin subtypes do not participate in tight junction formation. ZO-1 expression was also observed in order to visualize the junctional complex. Expression of claudin-1, ‐4, ‐7, and ‐10 was observed linearly along the cell junction. Expression of claudin-11 was not continuous and was present in a dot-like pattern along the cell junction. Expression of claudin-2, ‐15, ‐22, and ‐23 was observed in the cytoplasm. Although specific antibodies for alternatively splicing variants of claudin-10a and ‐10b were not commercially available, we confirmed by RT-PCR that claudin-10b, but not ‐10a, was expressed in corneal endothelium. Therefore, these results suggest that claudin-10b is a component of the endothelial tight junction. 

Figure 2

 

(A) Expression of claudin subtypes (red) in human corneal epithelium by immunohistochemistry. Nuclei were stained with DAPI (blue). Note that claudin-14 was also expressed in corneal stroma cells (white arrows). Claudin-14 expression in corneal stroma cells was also confirmed by horizontal section of corneal stroma layers in low (B) and high (C) magnification. Scale bars: 100 μm (A, B); 50 μm (C).

Figure 3

 

Expression of claudin subtypes and ZO-1 (red) in human corneal endothelium by immunohistochemistry. Nuclei were stained with DAPI (blue). Note that claudin-14 was not expressed in corneal endothelial cells. Scale bars: 100 μm.

Expression of claudin-1, ‐3, ‐4, ‐7, ‐11, ‐14, and ‐23 was confirmed in corneal epithelium, while expression of claudin-2, ‐5, ‐6, ‐8, ‐10, ‐12, ‐15, ‐22, and ‐24 was not observed. Taken together, the observations show that claudin-1, ‐4, ‐7, ‐11, and ‐23 were commonly expressed in both the corneal epithelium and endothelium, whereas the expression of claudin-10 (presumably -10b) was unique to the corneal endothelium. 

Functional Role of Claudin-10b in Human Corneal Endothelial Cell Line (B4G12 Cells)

To assess the physiological role of claudin-10b in barrier function and pump function of the corneal endothelial cell line, we used corneal endothelial cell line B4G12 cells. Transcripts for claudin-1, ‐2, ‐3, ‐4, ‐9, ‐10b, ‐11, ‐12, ‐15, ‐17, ‐20, ‐22, ‐23, and ‐24 were identified in B4G12 cells by RT-PCR (Fig. 4A). Expression of claudin-10 and ZO-1 along the cell–cell border was also confirmed by immunofluorescence (Figs. 4B, 4C). 

Figure 4

 

(A) Messenger RNA expression of claudins in B4G12 cells. (B) Expression of ZO-1 (red) in B4G12 cells. (C) Expression of claudin-10 (red) in B4G12 cells. (D) Expression of claudin-10 (red) in claudin-10 siRNA-treated B4G12 cells. ([B–D] Nuclei were stained with DAPI [blue]. Scale bars: 100 μm.) (E) Claudin-10b expression in control siRNA and claudin-10 siRNA-transfected B4G12 cells. (F) Expression of ATP1A1 between control siRNA and claudin-10 siRNA-transfected B4G12 cells. (G) Representative tracing of short-circuit current (μA/well) obtained with B4G12 cells in an Ussing chamber. After SCC had reached a steady state, the cells were charged at 1 mV three times, and the average SCC change was recorded (1). The TER was calculated from SCC change and voltage. Next, ouabain concentration was increased gradually in the chamber until the final concentration of ouabain reached 10 mM (2). (H) Comparison of TER between B4G12 cells without siRNA, B4G12 cells with control siRNA, and B4G12 cells with claudin-10 siRNA. (I) Comparison of SCC between B4G12 cells without siRNA, B4G12 cells with control siRNA, and B4G12 cells with claudin-10 siRNA. (J) Comparison of PD between B4G12 cells without siRNA, B4G12 cells with control siRNA, and B4G12 cells with claudin-10 siRNA. ([E, F] Data are expressed as mean ± SD of values from three replicates from four representative experiments [**P < 0.01, Student's t-test].) ([H–J] Data are expressed as mean ± SD of values from four replicated experiments [*P < 0.05, **P < 0.01, multiple t-test with Bonferroni correction following ANOVA].)

Using this cell line, we knocked down the expression of claudin-10 by siRNA. Figure 4D shows immunofluorescence of claudin-10 siRNA-treated B4G12 cells. Cell surface expression of claudin-10 was downregulated by claudin-10 siRNA compared to B4G12 cells without claudin-10 siRNA treatment (Fig. 4C). Since the only available antibody stained both isotypes of claudin-10, quantitative PCR was performed using primers specific to claudin-10b. Compared to that in control siRNA-transfected cells, the expression level of claudin-10b was significantly reduced by almost 40% with claudin-10 siRNA (Fig. 4E). In this siRNA concentration, the expression level of Na,K-ATPase alpha1 subunit (ATP1A1) was not affected (Fig. 4F). 

Figure 4G shows a representative tracing of SCC (μA/well) obtained with B4G12 cells in an Ussing chamber. The TER was calculated from SCC change by 1-mV voltage load (Fig. 4G, 1), and subsequently Na,K-ATPase-dependent SCC was calculated from SCC change by ouabain addition (Fig. 4G, 2). Na,K-ATPase-dependent SCC was also translated to PD by multiplying TER. The TER of B4G12 cells without siRNA, with control siRNA, and with claudin-10 siRNA is shown in Figure 4H, and Na,K-ATPase-dependent SCC and PD of these cells is shown in Figures 4I and 4J, respectively. Cell densities of these cells were 4080.0 ± 465.6, 3952.5 ± 264.3, and 4037.5 ± 473.3 cells/mm², respectively. There was no statistical difference in TER between these cells; however, knockdown of claudin-10b resulted in a significant decrease in SCC and PD compared to cells without siRNA or cells with control siRNA. 

Discussion

Corneal endothelium has two important functions; one is the pump function, and the other is the barrier function. In healthy human corneas, the tight junction barrier prevents the bulk of flow of fluid from aqueous humor to corneal stroma; on the other hand, it allows moderate diffusion of small nutrients, water, and other metabolites from aqueous humor to stroma through the intercellular spaces. ¹⁸ However, the molecular mechanism of the barrier function has not been fully investigated. 

Tight junctions form complex barriers with ion and size selectivity. Claudins have four transmembrane domains, two extracellular domains, and N- and C-terminal cytosolic domains. ¹⁹ The first extracellular domain is sufficient to determine both paracellular charge selectivity and transepithelial resistance. ²⁰ Increasing evidence indicates that the first extracellular domain lines the paracellular pathway and determines paracellular charge selectivity. ^11,20–23  

The tightness of barrier function varies significantly by cell type. These cell type–specific properties of tight junction strands are determined by the combination and expression ratios of claudin subtypes. ^8,24 For example, each nephron segment in the kidney expresses multiple claudin subtypes, which enables different functions on the reabsorption system of the individual segment. ²⁵ The characteristic patterns of claudin distribution in the kidney indicate that claudins in renal tubules form cation- or anion-selective pores that regulate paracellular transport. ^22,26  

In this study, we identified the claudin subtypes of the cornea. In the corneal endothelium, claudin ‐1, ‐2, ‐4, ‐7, ‐10 (presumably ‐10b), ‐11, ‐15, ‐22, and ‐23 were identified by both immunohistochemistry and RT-PCR. These claudin subtypes, except for claudin-2, ‐15, ‐22, and ‐23, were present along the cell junction. Claudin-2 and ‐15, which have been reported to form a cation-selective pore, ²⁷ were observed in the cytoplasm rather than the cell junction, so whether they participate in tight junction formation is unclear. On the other hand, the expression of claudin-1, ‐3, ‐4, ‐7, ‐11, ‐14, and ‐23 was confirmed in the corneal epithelium. Yoshida et al. ¹⁵ evaluated the expression pattern of claudin including ‐1 to ‐10 subtypes in human corneal epithelium, indicating claudin-1, ‐4, and ‐7 expression, and our data support this information. 

Among the claudin subtypes expressed in corneal endothelium, claudin-1 has been reported to be expressed ubiquitously, and other claudin subtypes have been reported to be restricted to specific tissues: claudin-2 in intestine, renal proximal tubule, and descending limb; claudin-4 in lung, intestine, and renal collecting duct; claudin-7 in intestine, renal distal convoluted tubule and collecting duct, and ovary; claudin-10b in uterus, renal medullary collecting duct, and ascending limb of Henle's loop; and claudin-11 in oligodendrocytes in the brain. Claudin-11 is also called oligodendrocyte-specific protein. The combined expression pattern of these claudin subtypes in corneal endothelium has not been reported in other tissues or organs in previously published studies. ^{8,13,15,22,24,28–30}  

Immunohistochemistry revealed that claudin-1, ‐4, ‐7, ‐11, and ‐23 were commonly expressed in both corneal endothelium and epithelium. Since corneal endothelium is a simple monolayer, it may have characteristics in common with corneal epithelium in tight junctional components. On the other hand, all claudin subtypes of corneal stroma except claudin-8 and ‐14 were also detected in corneal endothelium by RT-PCR, so corneal endothelium also has characteristics in common with corneal stroma. The reason may be that both corneal endothelium and stroma are of neural crest origin. ³¹ A claudin subtype unique to the corneal endothelium was not identified; however, the combination of positive claudin-10b expression and negative claudin-14 expression (claudin-10b⁺/claudin-14⁻) may be used to identify the corneal endothelium. 

Claudin-2, ‐10b, and ‐15 are known to construct cation-selective pores. ²³ These three claudin subtypes were not expressed in corneal epithelium. The participation of claudin-10b in tight junction formation of the corneal endothelium is reasonable, since Na⁺ ions, which are transported to lateral intercellular space by Na,K-ATPase of corneal endothelial cells, must pass through tight junctions into the anterior chamber in order for corneal endothelial pumps to function. To evaluate the importance of claudin-10b in corneal endothelial tight junction for Na⁺ ion transport, we analyzed changes of SCC, PD, and TER due to claudin 10 (presumably ‐10b) knockdown using the Ussing chamber system. Decrease of SCC and PD by claudin-10 knockdown, despite equal expression levels of Na,K-ATPase alpha1 subunit, suggested that Na⁺ ion transport through tight junctions might be prevented by a decrease in claudin-10b expression. On the other hand, TER was not changed by claudin-10 knockdown. A possible reason is that other claudin subtypes may contribute to the tight junction of corneal endothelium, and therefore knockdown of only one subtype may have little effect on TER. Further investigations are required to determine whether the changes in SCC and PD were due to cation selectivity of claudin-10b. 

We used corneal endothelial cell line B4G12 cells for the pump function and barrier function analysis. A cell line was used because culturing sufficient numbers of primary endothelial cells for this analysis was technically difficult. Interestingly, the claudin subtype expression pattern of B4G12 cells was slightly different from that of primary human corneal endothelium; B4G12 lacks claudin-7 expression and additionally expressed claudin-17. The difference may be due to individual variation in the expression of claudin subtypes. In order to pursue this possibility, we performed PCR analysis using a different donor cornea and found that claudin-6, ‐7, ‐9, ‐12, ‐17, and ‐20 were not always expressed (data not shown). 

The corneal endothelium also expressed claudin-4, which is known to have anion-selective pores ³² and cation barrier function. ^30,33 The combination of cation- and anion-selective pores may contribute to the balance of cation/anion transport of the corneal endothelium. Further investigations are required to investigate the role of other claudin subtypes in ion transport. 

In conclusion, corneal endothelium ubiquitously expressed claudin ‐1, ‐2, ‐4, ‐7, ‐10 (presumably ‐10b), ‐11, ‐15, ‐22, and ‐23, which include cation-selective and anion-selective subtypes along the cell junction. The expression pattern of claudin-10b⁺/claudin-14⁻ was specific in corneal endothelium among the three corneal layers, and claudin-10b may play an important role in the tight junction of corneal endothelium. 

Supplementary Materials

pdf S1 

Acknowledgments

Supported by a grant from the Ministry of Education, Culture, Sports, Science and Technology, Japan. The authors alone are responsible for the content and writing of the paper. 

Disclosure: E. Inagaki, None; S. Hatou, None; S. Yoshida, None; H. Miyashita, None; K. Tsubota, None; S. Shimmura, None