The Interplay Between E-Cadherin, Connexin 43, and Zona Occludens 1 in Retinal Pigment Epithelial Cells

Abstract

Purpose: Cell-cell contact in retinal pigment epithelium (RPE) involves adherent junctions, gap junctions, and tight junctions, which are primarily composed by E-cadherin, zona occludens 1 (ZO-1), and connexin 43, respectively. Here, we aimed to explore the relationship and interplay between these junction-associated proteins.

Methods: E-cadherin, connexin 43, and ZO-1 expression in human primary RPE in the early phase after TGF-β1 stimulation was detected. The knockdown of E-cadherin, ZO-1, and connexin 43 was performed to characterize the regulatory network involving these three proteins. Dye transfer and FITC-dextran permeability assays were conducted to observe the epithelial functional alterations. Transmission electron microscopy (TEM) was used to observe the ultrastructure of the cell-cell junctions in mouse RPE. The immunofluorescence staining and coimmunoprecipitation were performed to observe the colocalization and the physical association of E-cadherin, ZO-1, and connexin 43.

Results: Among these three components, E-cadherin appeared to be the first protein that was downregulated after TGF-β1 treatment. The ultrastructures of adherent junctions, gap junctions, and tight junctions could be observed in mouse RPE by TEM. E-cadherin, ZO-1, and connexin 43 were colocalized and physically bound to each other. The knockdown of one of these three proteins led to downregulation of the other two proteins and compromised epithelial function.

Conclusions: E-cadherin, ZO-1, and connexin 43 were physically associated with each other and were mutually regulated. To enhance the understanding of cell-cell contacts, a holistic view is needed. Our results provide new insights in RPE disorders such as proliferative vitreoretinopathy.

Cell-cell contact and the cell-basement membrane in retinal pigment epithelium (RPE) is essential for the integrity of the outer blood-retinal barrier and the homeostasis of its metabolism. The loss of cell-cell contact in RPE leads to the loss of pigment,¹ retinal edema,² and age-related macular degeneration.³ During rhegmatogenous retinal detachment, RPE can be activated by cytokines that ultimately contribute to proliferative vitreoretinopathy (PVR), a severe blinding disease characterized by the formation of contractile epiretinal or subretinal membrane. Compared with RPE at the edges of culture sheets, RPE with complete cell-cell contact cannot be activated by TGF-β2 or undergo epithelial-mesenchymal transition (EMT), migration, and proliferation.¹ 

Cell-cell contacts in RPE consist of gap-junctions,⁴ adherens junctions (AJs),⁵ and tight junctions.¹ Gap junctions are composed of two hemichannels that allow the rapid transport of small ions and compounds. Of all currently identified connexin genes, connexin 43 is the most extensively expressed and has been investigated in the most detail.^6–8 Connexin 43 observed in RPE^4,9 is capable of mediating intercellular dye transfer⁴ and protecting against oxidative stress-induced cell death.⁹ Zona occludens 1 (ZO-1) was one of the first recognized and best-studied tight junction proteins.^10,11 The expression of ZO-1 in RPE is essential for maintaining the integrity of the outer blood-retinal barrier. As a defining feature of all epithelial sheets,¹² E-cadherin-based AJs are also present in RPE.¹³ 

In PVR, a portion of the RPE becomes dislodged due to a retinal tear and floats in the vitreous or subretinal space, resulting in gaps in the RPE monolayer and incomplete cell-cell contact in the RPE at the edge of the monolayer. Retinal tears also lead to the exposure of RPE to vitreous, which is abundant in various cytokines and growth factors. One of the crucial growth factors is transforming growth factor-β (TGF-β), which has been widely used to simulate the in vitro PVR environment.^14,15 Due to the compromised cell-cell contact, RPE at the edge of the monolayer is stimulated by cytokines such as TGF-β and undergoes EMT, migration, and dislodges from Bruch's membrane. As a result, new cells are exposed to the edge of the monolayer, which can be activated. Therefore, the prevention and treatment of PVR should focus on the cells at the edge of the RPE monolayer, especially on maintaining the cell junctions in these cells. However, the temporal sequence of the loss of the three kinds of cell-cell contact-related proteins is currently unknown. Furthermore, the potential interactions between E-cadherin, connexin 43, and ZO-1 in RPE have not been fully examined. Therefore, this study aimed to examine the early alteration of cell-cell contact-related genes in the RPE in PVR and the interactions of these genes. 

Materials and Methods

Reagents and Antibodies

Mouse anti-human E-cadherin antibody (category number: 610182) was purchased from BD Biosciences (San Jose, CA, USA). Rabbit anti-human ZO-1 (category number: 61-7300), Alexa Fluor 594–conjugated anti-rabbit antibody, and FITC-conjugated anti-mouse antibody were obtained from Invitrogen (Carlsbad, CA, USA). Mouse anti-human fibronectin antibody was purchased from Sigma-Aldrich Corp. (St. Louis, MO, USA). Mouse anti-human connexin 43 antibody (category number: ab11370) and rabbit anti-human β-actin antibody were purchased from Abcam (Cambridge, MA, USA). TRIzol reagent and 4′,6-diamidino-2-phenylindole (DAPI) were purchased from Invitrogen. The SuperReal PreMix Plus (SYBR Green) reagent kit was obtained from Takara Clontech (Kyoto, Japan). Most of the other reagents, such as salt and buffer components, were of analytical grade and were obtained from Sigma-Aldrich Corp. 

Cell Culture

Human primary RPE (phRPE) cells were isolated from the eyes of donors (obtained from the Eye Bank of Shanghai Tenth People's Hospital) as described previously, and low passage cells (passage 2–4) were used in this study. The human RPE cell line ARPE-19 in our lab^5,16 was also used in this study. 

PhRPE or ARPE-19 cells were cultured in Dulbecco's modified Eagle's medium (DMEM)/F12 culture media (Gibco, Thermo Fisher Scientific, Waltham, MA, USA) with 10% fetal bovine serum (Gibco, Thermo Fisher Scientific) at 37°C in a humidified incubator with 5% CO₂. The culture media was refreshed every 2 to 3 days. For further experiments, the cells were trypsinized, seeded in 6- or 12-well plates, and cultured for 12 to 24 hours after the cells attached. Thereafter, the cells were starved for 16 hours and were subjected to small interfering RNA (siRNA) transfection or stimulated with 10 ng/mL TGF-β1 (Invitrogen; Thermo Fisher Scientific) for various time periods. SiRNA transfection was conducted using the Lipofectamine 3000 reagent (Invitrogen; Thermo Fisher Scientific) according to the manufacturer's instructions. The siRNAs were synthesized by GenePharma (Shanghai, China). To compensate for the nonspecific effects of the nonsequence-specific RNA oligo, a negative control siRNA (si-NC) was used in all siRNA-involved experiments. To compensate for the nonspecific effects of siRNAs designed for knocking down specific genes, all the siRNA sequences were subjected to BLAST analysis (basic local alignment search tool; National Center for Biotechnology Information, Rockville, MD, USA) to ensure that the sequences were uniquely targeted to the specific gene. Furthermore, in most of the experiments, two siRNAs (which target different sites in one gene) were utilized to minimize the influence of potential nonspecific effects on the results. The siRNA sequences are listed in Table 1. 

Table 1

 

Sequences of the siRNAs

 

Real-Time Quantitative PCR

Total RNA was extracted using a TRIzol reagent kit. The RNA was reverse-transcripted using the PrimeScript RT reagent Kit (Takara Clontech). Real-time PCR was performed in triplicate using a TB Green Premix Ex Taq II (Tli RNaseH Plus) kit (Takara Clontech) with a CFX Connect Real-Time System (Bio-Rad, Hercules, CA, USA). Each reaction contained 12.5 μL 2×TB Green Premix Ex Taq II (Tli RNaseH Plus), 1 μL 10 μM oligonucleotide primers (Table 1), 1 μL cDNA, and 9.5 μL ddH₂O (final volume of 25 μL). The thermal cycling reaction comprised an initial denaturation step at 95°C for 30 seconds and 40 cycles of 95°C for 5 seconds and 60°C for 30 seconds. RNA expression was normalized to the level of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) mRNA. The primers for each gene were synthesized by Generay Biotech (Shanghai, China), and the sequences are listed in Table 2. 

Table 2

 

Primer Sequences Used in This Study

 

Western Blot (WB) Analysis

After treatment, cells were lysed with radio-immunoprecipitation assay (RIPA) buffer (Beyotime, Shanghai, China) supplemented with phenylmethylsulfonyl fluoride and PhoSTOP EASY pack phosphatase inhibitor (Roche, Mannheim, Germany) on ice for 30 minutes. The lysates were clarified by centrifugation at 12,000 rpm for 5 minutes, and the supernatants were collected. The total protein concentration was quantified using a bicinchoninic acid assay kit (Pierce, Rockford, IL, USA) according to the manufacturer's instructions. Fifty micrograms of proteins were loaded and separated with 8% SDS-PAGE and transferred onto nitrocellulose membranes (Bio-Rad). To avoid nonspecific binding, the membranes were blocked using 5% skim milk in PBS for 60 minutes at room temperature. The membranes were then incubated with primary antibodies diluted in 2% skim milk in PBS with 0.1% Tween-20 (PBS-T) at 4°C overnight. After washing with PBS-T three times, the membranes were incubated with IRDye 680 LT goat anti-rabbit or IRDye 800CW goat anti-mouse secondary antibody (Li Cor Biosciences, Lincoln, NE, USA) at room temperature for 1 hour. After washing with PBS-T three times, the bound antibody was detected with an Odyssey infrared imaging system (Li Cor Biosciences). The protein expression was quantified with Quantity One software (V4.6.6). The relative expression was normalized according to the level of β-actin. 

Transmission Electron Microscopy (TEM)

Female C57BL/6 mice (6–7 weeks old, specific pathogen-free) were purchased from Shanghai SLAC Laboratory Animal Co. Ltd. (Shanghai, China) and were maintained in a specific pathogen-free facility. Water and standard laboratory chow were provided ad libitum. A 12-hour day and night cycle was utilized. All animal experiments were conducted in compliance with the Institutional Animal Experiment Committee Guidelines and the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

After adaptive feeding for 7 days, the mice were killed. The eyes were harvested. Each eyeball was pierced with a syringe through the cornea, placed in electron microscope fixing solution, and fixed in a refrigerator at 4°C for 48 hours. The anterior segment of the eye and the lens were discarded. The eyecup was cut into small pieces of approximately 1 mm³ in size and were placed in fresh fixing solution for another 30 minutes. The tissue was then rinsed with PBS for 30 minutes and fixed with 1% citric acid for 1 hour. After rinsing with PBS, the tissue was dehydrated with an ethanol gradient. Afterward, the tissue was embedded and cut into ultrathin sections. Imaging was performed using a HITACHI TEM system. 

Immunofluorescence Analysis

After adaptive feeding for 7 days, the mice were killed. The eyes were harvested, and the RPE-Bruch's membrane-choroid complex (RBCC) was separated. The whole-mount RBCC was fixed in 4% paraformaldehyde (PFA) at 4°C overnight. The RBCC was soaked in 1% potassium permanganate for 5 minutes, washed three times with PBS, and infused with 2% oxalic acid for 1 minute. Then it was washed with PBS three times, soaked in 64-fold diluted vitamin C for 4 minutes, and then washed again with PBS three times. RBCC depigmentation was performed by using this method. The RBCC was then blocked with 2% BSA for 1 hour at room temperature. Subsequently, the RBCC was incubated with anti-mouse E-cadherin, connexin 43, and ZO-1 antibodies at 4°C overnight. After rinsing three times, the RBCC was incubated with Alexa Fluor 647-conjugated or Alexa Fluor 594-conjugated secondary antibody for 1 hour at room temperature. Afterward, the RBCC was mounted on a microscope slide. After counterstaining with DAPI, the stained coverslips were mounted and visualized under a confocal microscope (Carl Zeiss, LSM710, Jena, Germany). 

Coimmunoprecipitation Assay

Plasmids for the overexpression of E-cadherin (flag-E-cadherin) and connexin 43 were constructed by GeneChem Corporation (Shanghai, China). ARPE-19 cells were seeded in 10-cm dishes. After attachment, E-cadherin or connexin 43-overexpressing plasmid was transfected into cells with Lipofectamine 3000 reagent (Invitrogen; Thermo Fisher Scientific) according to the manufacturer's instructions. Forty-eight hours later, the cells were harvested and subjected to a co-immunoprecipitation (co-IP) assay with a Pierce Classic Magnetic IP/Co-IP Kit (Pierce; Thermo Fisher Scientific) according to the manufacturer's instructions. In brief, the cells were lysed with lysis/wash buffer and sonicated. The cell lysates were then incubated with 10 μL anti-E-cadherin antibody or anti-connexin 43 antibody overnight at 4°C. Nonspecific antibody was also used as a negative control. Afterward, protein A/G magnetic beads were added into the mixture and incubated at room temperature for 1 hour. The beads were then washed thrice and eluted with elution buffer. Immunoblotting was then performed as described in the Western Blot Analysis section. 

Dye Transfer Assay

The human RPE cell line ARPE-19 from the Tongji Eye institute^17–19 was used for the Lucifer yellow dye transfer assay. ARPE-19 cells were seeded in six-well plates and cultured in a humidified incubator with 5% CO₂ at 37°C. After confluence was reached, the cells were washed twice with Hank's buffer. In addition, 1 mM Lucifer yellow (Sigma-Aldrich Corp.) was added to the cells. Immediately, the confluent monolayer was scraped with a steel surgical scalpel. The cells were then incubated at 37°C for 5 minutes. Thereafter, cells were washed three times with Hanks buffer and fixed in 4% PFA. The dye transfer distance was visualized by a fluorescence microscope (DMI3000B, Leica) at an excitation wavelength of 488 nm. The number of cells with Lucifer yellow staining was counted in each group. Each experiment was repeated six times. 

Permeability Assays

PhRPE monolayer culture was conducted as previously described with minor modifications.²⁰ In brief, cells were seeded at 2 × 10⁵ cells/cm² onto Transwell filters (0.4 μm pore size; Costar, Corning, NY, USA). After confluence was reached, the cells were transfected with si-NC, si- E-cadherin-1, si-connexin 43-1, and si-ZO-1-2 with Lipofectamine 3000 reagent as described above. Forty-eight hours later, 1 mg/mL FITC-dextran (4 kDa; Sigma-Aldrich Corp.) was added to the upper chamber. After incubation for 3 hours, 100 μL culture medium from the lower chamber was collected. The concentration of FITC-dextran in these samples was determined by a microplate reader (ELx800; Bio-Tek, Winooski, VT, USA). Each experiment was repeated eight times. 

Statistical Analysis

All experiments were performed at least three times. The mean and SEM was calculated for all parameters measured in this study. The data were analyzed statistically using 1-way ANOVA or 2-tailed Student's t-test. A value of P < 0.05 was accepted as statistically significant. 

Results

In phRPE Cells, TGF-β1 Decreased E-Cadherin, Connexin 43, and ZO-1 Expression, and E-Cadherin Was Affected the Earliest

PhRPE cells cultured in cell culture plates reached 60% to 70% confluence and were treated with 10 ng/mL TGF-β1 for different time intervals. E-cadherin, connexin 43, and ZO-1 expression gradually decreased after TGF-β1 treatment (Figs. 1A, 1B). Among these three proteins, E-cadherin showed the earliest downregulation and showed significant reduction 3 hours after TGF-β1 treatment. Connexin 43 and ZO-1 expression were significantly decreased after 6 hours. The expression of the mesenchymal marker fibronectin was increased significantly at 12 hours after TGF-β1 exposure (Figs. 1A, 1B). 

Figure 1

 

TGF-β1 decreased E-cadherin, connexin 43, and ZO-1 expression and increased fibronectin expression in a time-dependent manner. PhRPE cells were incubated with TGF-β1 (10 ng/mL) for different time intervals. (A) The expression levels of E-cadherin, connexin 43, ZO-1, and fibronectin were detected by WB. (B) The statistical results of the grayscale values obtained from Western blotting were calculated after normalization to the level of β-actin. The data are presented as the mean ± SEM. *P < 0.05; **P < 0.01.

Knockdown of E-Cadherin Led to the Reduced Expression of Connexin 43 and ZO-1

PhRPE cells were transfected with two E-cadherin-specific siRNAs (si-E-cadherin-1 and si-E-cadherin-2) or a si-NC after starvation for 16 hours. The expression of E-cadherin, connexin 43, and ZO-1 was then detected by RT-PCR and WB. Compared with the transfection of si-NC, transfection with two E-cadherin siRNAs decreased the expression of E-cadherin by more than 50% (Figs. 2A–C). Furthermore, the knockdown of E-cadherin led to the reduced expression of connexin 43 and ZO-1 at both the mRNA and protein levels (Figs. 2A–C). 

Figure 2

 

Knockdown of E-cadherin downregulated the expression of connexin 43 and ZO-1 in phRPE cells. PhRPE cells were transfected with E-cadherin siRNA (si-E-cadherin-1 or si-E-cadherin-2) or si-NC and were cultured for another 48 hours. (A) The expression levels of E-cadherin, connexin 43, and ZO-1 were detected by WB. (B) The statistical results for the grayscale values obtained from Western blotting were calculated after normalization to the level of β-actin. (C) E-cadherin, connexin 43, and ZO-1 mRNA expression were detected by RT-PCR. Relative expression was calculated after normalization to the level of GAPDH. The data are presented as the mean ± SEM. *P < 0.05; **P < 0.01.

Knockdown of Connexin 43 Led to the Reduced Expression of E-Cadherin and ZO-1

PhRPE cells were transfected with two connexin 43-specific siRNAs (si-connexin 43-1 and si-connexin 43-2) or si-NC after starvation for 16 hours. Compared with the transfection of si-NC, transfection with both the connexin 43 siRNAs decreased the expression of connexin 43 by more than 70% (Figs. 3A–C). Furthermore, the knockdown of connexin 43 led to the reduced expression of E-cadherin and ZO-1 at both the mRNA and protein levels (Figs. 3A–C). 

Figure 3

 

Knockdown of connexin 43 downregulated the expression of E-cadherin and ZO-1 in phRPE cells. PhRPE cells were transfected with connexin 43 siRNA (si-connexin 43-1 or si-connexin 43-2) or si-NC and were cultured for another 48 hours. (A) The expression levels of connexin 43, E-cadherin, and ZO-1 were detected by WB. (B) The statistical results for the grayscale values obtained from Western blotting were calculated after normalization to the level of β-actin. (C) Connexin 43, E-cadherin, and ZO-1 mRNA expression were detected by RT-PCR. The relative expression was calculated after normalization to the level of GAPDH. The data are presented as the mean ± SEM. *P < 0.05; **P < 0.01.

Knockdown of ZO-1 Led to the Reduced Expression of E-Cadherin and Connexin 43

PhRPE cells were transfected with si-ZO-1-1, si-ZO-1-2, or si-NC after starvation for 16 hours. Compared with the transfection of si-NC, transfection with the ZO-1-specific siRNAs decreased the expression of ZO-1 by more than 70% (Figs. 4A–C). Furthermore, the knockdown of ZO-1 led to the reduced expression of E-cadherin and connexin 43 at both the mRNA and protein levels (Figs. 4A–C). 

Figure 4

 

Knockdown of ZO-1 downregulated the expression of E-cadherin and connexin 43 in phRPE cells. PhRPE cells were transfected with ZO-1 siRNA (si-ZO-1-1 or si-ZO-1-2) or si-NC and were cultured for another 48 hours. (A) The expression levels of ZO-1, E-cadherin, and connexin 43 were detected by WB. (B) The statistical results for the grayscale values obtained from Western blotting were calculated after normalization to the level of β-actin. (C) ZO-1, E-cadherin, and connexin 43 mRNA expression were detected by RT-PCR. The relative expression was calculated after normalization to the level of GAPDH. The data are presented as the mean ± SEM. *P < 0.05; **P < 0.01.

Ultrastructures of the Cell Junctions in Mouse RPE Monolayer

Eyecups from C57BL/6 mice were harvested and subjected to TEM observation. The ultrastructure of Bruch's membrane, the RPE, and the outer segment of the photoreceptor are clearly shown (Fig. 5A). Notably, in the apical lateral intercellular membrane, the ultrastructure of the zonula occludens, zonula adherens, macula adherens, and gap junctions can be observed and are relatively close to each other (Fig. 5B). 

Figure 5

 

Ultrastructures of the cell junctions in mouse RPE monolayers. Eyecups from C57BL/6 mice were harvested and subjected to TEM observation. (A) Sectional view of the ultrastructure in RPE. (B) Higher magnification view of the region contained in the dashed black box in A. In the apical lateral intercellular membrane, the zonula occludens (short white arrow, tight junction), zonula adherens (long white arrow, adherent junction), macula adherens (short black arrow, desmosome, adherent junction), and gap junction (long black arrow) are shown. BM, Bruch's membrane; OS of PR, outer segment of the photoreceptor.

E-cadherin, Connexin 43, and ZO-1 Were Colocalized and Physically Interacted With Each Other

C57BL/6 mice were killed, and the RBCCs were collected and subjected to IF staining. It was found that E-cadherin, connexin 43, and ZO-1 were colocalized with each other in cell-cell contacts (Figs. 6A–C). To further confirm their physical interaction, we first overexpressed E-cadherin and connexin 43 in ARPE-19 cells. The cell lysates were immunoprecipitated with anti-E-cadherin and anti-connexin 43 antibodies, respectively. The precipitated proteins were then subjected to immunoblot experiments with anti-E-cadherin, anti-connexin 43, and ZO-1 antibodies. It was found that E-cadherin coimmunoprecipitated with connexin 43 and ZO-1. Similarly, connexin 43 coimmunoprecipitated with E-cadherin and ZO-1, suggesting that a physical complex was formed by E-cadherin, connexin 43, and ZO-1 (Figs. 6D, 6E). 

Figure 6

 

E-cadherin, connexin 43, and ZO-1 colocalized and physically interacted with each other. (A–C) RBCCs from C57BL/6 mice were harvested and subjected to IF staining. Double staining of E-cadherin and connexin 43 (A), E-cadherin and ZO-1 (B), and connexin 43 and ZO-1 (C) are shown. Scale Bar: 20 μm. (D, E) E-cadherin (D) or connexin 43 (E) overexpressing plasmids were transfected into ARPE-19 cells. Immunoprecipitation of total protein in ARPE-19 cells was performed with anti-E-cadherin (D) or anti-connexin 43 (E) antibodies. The eluates were then subjected to WB analysis to detect the expression of E-cadherin, connexin 43, and ZO-1.

Knockdown of E-Cadherin, Connexin 43, or ZO-1 in ARPE-19 Cells Inhibits Lucifer Yellow Transfer via Gap Junctions

A lucifer yellow dye transfer assay was conducted to evaluate the functioning of gap junction, which are mostly characterized by connexin 43 expression. E-cadherin-, connexin 43-, or ZO-1-specific siRNA was transfected into ARPE-19 cells 48 hours before the Lucifer yellow transfer assay. We chose the siRNA with the highest knockdown efficiency for each protein. It was found that the knockdown of connexin 43, E-cadherin, or ZO-1 resulted in reduced Lucifer yellow transfer (Figs. 7A, 7B). 

Figure 7

 

Downregulating E-cadherin, connexin 43, and ZO-1 inhibited Lucifer yellow transfer via gap junctions. ARPE-19 cells were transfected with si-E-cadherin-1-, si-connexin 43-1-, or ZO-1-specific siRNAs. Forty-eight hours later, the cells were confluent and subjected to a Lucifer yellow transfer assay. (A) Images of ARPE-19 cells scraped and treated with Lucifer yellow in different treatment groups were captured with a fluorescence microscope (DMI3000B, Leica). (B) The number of cells with Lucifer yellow staining was counted in each group and subjected to statistical analysis. The data are presented as the mean ± SEM. *P < 0.05; **P < 0.01.

Knockdown of E-Cadherin, Connexin 43, or ZO-1 in ARPE-19 Cells Increases Epithelial Permeability

PhRPE cells were cultured in a Transwell chamber and transfected with E-cadherin-, connexin 43-, or ZO-1-specific siRNA. Forty-eight hours later, FITC-dextran was added into the upper chamber (Fig. 8A). A significant increase in the FITC-dextran concentration was observed in the lower chamber after the knockdown of E-cadherin, connexin 43, or ZO-1, suggesting the increased permeability of the RPE monolayer (Fig. 8B). 

Figure 8

 

Downregulating E-cadherin, connexin 43, and ZO-1 increased epithelial permeability in phRPE cells. (A) PhRPE cells were cultured in a Transwell chamber and transfected with E-cadherin-, connexin 43-, or ZO-1-specific siRNA. Forty-eight hours later, FITC-dextran was added to the upper chamber. After incubation for 3 hours, 100 μL culture medium from the lower chamber was collected. (B) The concentration of FITC-dextran in these samples was determined by a microplate reader (ELx800; Bio-Tek). The data are presented as the mean ± SEM. *P < 0.05; **P < 0.01. N = 8.

Discussion

EMT in RPE cells is a triggering event for the development of PVR. Tamiya et al.¹ revealed that the loss of cell-cell contact is a prerequisite for the initiation of EMT. In rhegmatogenous retinal detachment, RPE cells are dislodged and float in the vitreous or subretinal space. As a result, the integrity of the RPE monolayer is compromised. Thereafter, dislodged cells and RPE cells at the edges of the monolayer gradually undergo EMT, whereas cells in the center of the monolayer, where the original cell-cell junctions are preserved, retain an epithelial phenotype.¹ Because of this, to prevent PVR, we should focus on the cells at the edge of the RPE monolayer and especially on the maintenance of the intercellular junctions. Thus, it is necessary to further examine the interactions of different kinds of intercellular junctions in the RPE. 

Like those in most epithelial cells, the primary cell-cell contacts of RPE cells consist of adherent junctions,⁵ gap junctions,⁴ and tight junctions.¹ The primary molecular component of each of these three types of junctions is E-cadherin, ZO-1, and connexin 43, respectively. EMT is a dynamic process, during which epithelial cells lose cell-cell contact first and then pass through an intermediate stage before adopting mesenchymal characteristics.²¹ Therefore, we first detected the expression of the three cell junction-related proteins, E-cadherin, connexin 43, and ZO-1, in subconfluent RPE cells treated with TGF-β1. Previous studies indicated that TGF-β1 and TGF-β2 are capable of modulating the PVR environment in vitro (including initiating EMT and migration in RPE) and have both been widely used in related studies.^{5,14,16,22–25} Our previous study¹⁶ confirmed that TGF-β1 and TGF-β2 elicited similar EMT-initiating effects, including the downregulation of the expression of E-cadherin and ZO-1. Considering the similar effects of TGF-β1 and TGF-β2 in RPE cells and the predominant use of TGF-β1 in our laboratory, only TGF-β1 was used in this study. It was found that E-cadherin was the first downregulated protein of the three and showed a significant reduction as early as 3 hours after TGF-β1 treatment. The mesenchymal marker fibronectin was increased significantly after 12 hours, suggesting that an intermediate stage exists in TGF-β1-treated RPE cells before they undergo complete EMT. 

Previous studies revealed the potential interplay between different kinds of cell-cell junctions. For example, ZO-1 was found to directly bind to connexin 43 and to affect the assembly of connexin 43 in Hela cells and rabbit lens epithelial cells.^26,27 In mouse retinal endothelial cells, the knockdown of connexin 43 led to a reduction in ZO-1 expression.⁸ Studies have also indicated that E-cadherin modulates the trafficking and assembly of connexin 43 in squamous carcinoma cells and murine skin papilloma cells.^28,29 In contrast, in glioma stem cells, connexin 43 directly interacts with E-cadherin, and the overexpression of connexin 43 leads to the increased expression of E-cadherin.³⁰ Here, in our study, we confirmed that in phRPE cells, the knockdown of either E-cadherin, connexin 43, or ZO-1 leads to the downregulation of the other two proteins, indicating the interaction of these three proteins. We speculate that a junction complex exists that consists of these three proteins. 

Because previous studies also indicated that physical interactions occur between these three proteins in other cell types, we assume that E-cadherin, connexin 43, and ZO-1 also, at least partially, directly bind to each other in RPE. We first observed the ultrastructure of the mouse RBCC, in which the original RPE monolayer and cell-cell contacts are completely retained, with TEM. It was found that tight junctions, adherent junctions, and gap junctions were primarily located in the apical lateral intercellular membrane and were close to each other. This result is consistent with the observation by Akanuma et al.⁴ that connexin 43 was sporadically expressed and colocalized with ZO-1 and N-cadherin in the apical lateral intercellular membrane in mouse and human RPE cells via IF staining. Here, by utilizing immunofluorescence staining in the mouse RBCC, we observed more abundant connexin 43 expression, which colocalized with both E-cadherin and ZO-1 in the intercellular membranes of mouse RPE cells. Further co-IP assays confirmed the physical interaction among these three proteins. 

Cell permeability reflects barrier function, which is primarily affected by tight junctions, in epithelial cells.^31,32 The knockdown of ZO-1 leads to increased cell permeability in MDCK II cells. Similarly, our study found that ZO-1 is essential for maintaining the barrier function in RPE. In addition, the knockdown of either E-cadherin or connexin 43 led to the increased permeability of FITC-dextran, probably as a result of downregulated ZO-1 expression caused by E-cadherin or connexin 43 inhibition. We further examined the function of gap junctions using a dye transfer assay. The knockdown of E-cadherin, connexin 43, or ZO-1 led to reduced dye transfer, indicating that connexin 43 function is compromised when E-cadherin or ZO-1 is inhibited. 

In conclusion, as the molecule basis of adherent junctions, gap junctions, and tight junctions, E-cadherin, connexin 43, and ZO-1 physically interact with each other and are mutually regulated. To fully understand cell-cell contacts, a holistic view is needed rather than a focus on single cell junction-related proteins. 

Acknowledgments

Supported by the National Natural Science Foundation of China (No. 81800836 and No. 81770939). 

Disclosure: H. Bao, None; S. Yang, None; H. Li, None; H. Yao, None; Y. Zhang, None; J. Zhang, None; G. Xu, None; H. Jin, None; F. Wang, None