Endothelin-1 Synthesis and Secretion in Human Retinal Pigment Epithelial Cells (ARPE-19): Differential Regulation by Cholinergics and TNF-α

Abstract

purpose. Endothelin (ET)-1 can produce nerve damage analogous to that in optic neuropathies such as glaucoma. The precise source of endothelin in the posterior segment of the eye remains unclear. The present study was conducted to determine whether the retinal pigment epithelium (RPE), which helps maintain the outer blood–retinal barrier, is a local source for ET-1 and whether the amount of ET-1 secreted by the RPE is be differentially regulated by cholinergics and the cytokine TNF-α.

methods. Human retinal pigment epithelial cells (ARPE-19) were cultured either to a mature state (mRPE) for 4 weeks, with well-defined tight junctions, or to a young state (yRPE) for 4 days, with incompletely formed tight junctions. ET-1–like immunoreactivity was determined by immunocytochemistry, and secreted ET-1 was quantified by radioimmunoassay in both cell types. Cells were stimulated with the cholinergic agonist carbachol or with the cytokine TNF-α for specific periods. The expression of muscarinic receptor subtypes M₁ and M₃ and the peripheral membrane protein ZO-1 were analyzed by immunoblot and immunocytochemistry, respectively. Expression of preproendothelin-1 (ppET-1) mRNA after application of different stimuli at specific time points was determined by real-time RT-PCR. Carbachol-mediated elevation in intracellular calcium ([Ca²⁺]_i) was also measured in the presence or absence of a selective muscarinic receptor antagonist.

results. Constitutive synthesis and secretion of ET-1 was greater in mRPE than in yRPE cells. TNF-α caused a significant increase in ppET-1 mRNA levels and ET-1 secretion in both phenotypes. The disruption and subsequent breakdown of the tight junction barrier was evident in either phenotype after treatments with TNF-α. There was a concentration-dependent increase in [Ca²⁺]_i in both y- and mRPE cells in response to CCh. CCh at 1 μM significantly increased ET-1 secretion, a response observed in yRPE but not in mRPE cells. This effect may be mediated primarily by the M₃ receptor subtype and is phospholipase C (PLC) dependent.

conclusions. Regulation of ET-1 release in ARPE-19 cells was differentially regulated by TNF-α and CCh and was dependent on the age of the culture. RPE may be a source for ET-1 in the retina, and its increased release may become more important during breakdown of the blood–retinal barrier, as seen after TNF-α treatment.

Endothelins (ET) are 21-amino-acid potent vasoactive peptides, processed and secreted locally in the eye, ^{1 2 3 4} with immunoreactive ET-1 and -3 expressed most notably in the iris, choroid, retina, optic nerve head, ciliary body, lens, and corneal endothelium. ^{1 3 4 5} ET-1 is also present in the aqueous and vitreous humors. ^{6 7} Within in the retina, the retinal pigment epithelium (RPE), photoreceptor inner segments, inner plexiform layer, retinal ganglion cells, and retinal pericytes have also been shown to be immunoreactive for ET-1 mRNA and mature ET-1 protein. ^{3 5 8 9 10} Our laboratory has demonstrated that nonpigmented ciliary epithelium can serve as a source of ET-1 in the anterior chamber and that cholinergics, TNF-α, and glucocorticoids may regulate its secretion. ^{4 11} The ciliary pigment and nonpigmented epithelium (NPE) together constitute the blood–aqueous barrier in the anterior segment, which is contiguous with the outer blood–retinal barrier formed by the RPE at the ora serrata. The ciliary epithelium, primarily the NPE and the RPE secrete a multitude of growth-factor–like substances and thus act as source cells for peptides within the immune-privileged environment of the eye. ^{12 13} Based on these reports and similarities between the NPE and the RPE in regulating fluid transport and acting as secretory cells within their respective local environments, we hypothesized that RPE can act as a source for ET-1. 

Our primary intent was to provide a quantitative description of both the mRNA and protein levels of ET-1 in a cell culture model of human RPE. In addition, we quantitatively analyzed both constitutive and regulated secretory pathways for ET-1 in these cells, which emphasizes the temporal aspects of ET-1 synthesis and secretion in the RPE. The rationale for considering the actions of cholinergics on RPE stems from observations that the uvea, including the choroid, is parasympathetically innervated by varicosities arising postganglionically from the pterygopalatine and ciliary ganglions. ¹⁴ The dense plexus of cholinergic innervation at the choriocapillaris just beneath the RPE is thought to act postsynaptically on choroidal smooth muscles as well as the RPE. ¹⁵ In addition, both immunofluorescence and binding studies have shown that mammalian and avian RPE express abundant muscarinic receptors throughout development and adult stages, similar to that in the brain. ^{15 16 17 18 19}  

The physiological and pathophysiological implications of endothelins at specific regions in the eye, including the RPE, are not well understood. ET-1 along with ET-3 and nitric oxide (NO) may help regulate optic nerve head, retinal, and choroidal blood flow. ^{20 21 22 23} Higher levels of ET-1 have been implicated in severe cardiovascular and developmental dysfunctions ²⁴ and more recently in the pathophysiology of glaucoma. ²⁵ Exogenous ET-1 administered at the retrobulbar region of the optic nerve results in a neuropathy in a manner similar to glaucoma, ²⁶ and intravitreally injected ET-1 can significantly alter the rate of membrane-bound organelles associated with fast axonal transport in the optic nerve. ²⁷ Presently, we have established that the retinal pigment epithelium (RPE) can act as a local source for ET-1. 

The RPE, like most epithelia, form apical tight junctions and polarize with distinct apical and basolateral domains. ²⁸ The development of the outer blood–retinal barrier is a gradual, multistep process that parallels changes in expression and recruitment of proteins involved in formation of the tight junction complex with concomitant decrease in paracellular permeability, akin to most epithelia, including the RPE. ^{28 29 30 31 32} Sorting and secretion of proteins in epithelial cells are critically dependent on cell polarity. ^{33 34} We have therefore used two phenotypes of RPE to delineate differences in ET-1 secretion that may be dependent on the polarity and maturity of the epithelium. We report that secretion of ET-1 may be differentially regulated (muscarinic or TNF-α mediated stimulation) in a cell culture system of polarized or mature RPE (mRPE) and nonpolarized or young RPE (yRPE) cells. 

Material and Methods

Antibodies

The rabbit anti-muscarinic receptor M₁ (epitope corresponding to part of the i3 loop-residues 231-350) and goat anti-M3 subtypes (epitope corresponding to the C-terminal region) were purchased from Santa Cruz Biotechnology, Santa Cruz, CA. Rat heart lysate and the blocking peptide purchased from Santa Cruz were used as controls for M₁ and M₃ receptor detection respectively. Mouse anti-ZO-1 was purchased from Zymed Laboratories (San Francisco, CA). Rabbit anti-endothelin-1 (anti-ET-1) was purchased from Bachem/Peninsula Laboratories (Belmont, CA). The same antibody was used in radioimmunoassay measurements (secreted ET-1) and immunofluorescence experiments (intracellular ET-1). Rabbit IgG and mouse IgG were purchased from Vector Laboratories (Burlingame, CA). Secondary antibodies including donkey anti-rabbit IgG, donkey anti-mouse IgG, and bovine anti-goat IgG conjugated to horseradish peroxidase (HRP) were purchased from Amersham Biosciences (Piscataway, NJ). Fluorescent probes, including goat anti-rabbit Alexa 488 and goat anti-mouse Alexa 594, were purchased from Molecular Probes (Eugene, OR). 

Cell Culture

ARPE-19 human retinal pigment epithelial cells, a spontaneously transformed cell line, was purchased from the American Type Culture Collection (ATCC, Manassas, VA). ARPE-19 cells (passages 20–23) were maintained at 37°C and 5% CO₂ in a 1:1 mixture of Dulbecco’s modified Eagle’s medium (DMEM) and Ham’s F-12 (Invitrogen, Carlsbad, CA) supplemented with 10% fetal bovine serum (Hyclone, Logan, UT), 2 mM l-glutamine, 23 mM NaHCO₃, and penicillin and streptomycin (Invitrogen). The spontaneously arising ARPE-19 cells, characterized by Dunn et al., ³⁵ display typical polarized epithelial morphology when grown according to conditions described by them. In addition, mature polarized ARPE-19 cells that are maintained in culture for 3 to 4 weeks have reduced paracellular permeability and higher transepithelial resistance (TER) compared with cultures that are grown for 1 week. ³⁵ We used culture conditions similar to those described by Dunn et al. ^{35 36} to obtain mRPE cells. Cellular morphology for both young (3–4 days in culture) and mature phenotypes was similar to those shown by Dunn et al. ³⁵ Cells were seeded at 1.4 × 10⁵ cells/well (six-well plate) or 4 × 10⁵ cells/100-mm dish and maintained in culture according to Dunn et al. ARPE-19 cells were grown either for 4 to 5 weeks (mRPE) with well-defined tight junctions or for 3 to 4 days with incompletely formed tight junctions (yRPE). 

Treatments

ARPE-19 cells were subjected to different treatments in serum-free DMEM-F12 for various periods, as specified in each experiment. The agonists used in this study were carbachol (CCh; Sigma-Aldrich, St. Louis, MO) and tumor necrosis factor (TNF)-α (PeproTech, Rocky Hill, NJ). The antagonists used were pirenzepine, 4-diphenylacetoxy-N-(2-chloroethyl)-piperidine hydrochloride (4-DAMP; both from Sigma-Aldrich, St. Louis, MO), and U73122 (Biomol, Plymouth Meeting, PA). In treatments that included an antagonist or inhibitor, cells were pretreated for 20 to 30 minutes before treatment with the agonist. Each treatment condition was performed at least six times in most experiments. 

ET-1 Extraction and Measurement by Radioimmunoassay

ARPE-19 cells were grown to either young (3-4 days in culture, yRPE) or mature states (4 weeks in culture, mRPE) in six-well culture plates (35 mm diameter/well, ∼1.4 × 10⁵ cells/ well) in 1:1 DMEM+Ham’s F12 culture medium containing 10% FBS. On the day of treatment, cells were rinsed three times with serum-free 1:1 DMEM+Ham’s F12 culture media (SF-DMEM/F12) and treated with 1 mL SF-DMEM/F12 containing either CCh (CCh: 1, 10, 100 μM) or TNF-α (10 nM), a concentration previously shown to stimulate ET-1 synthesis and secretion in human nonpigmented epithelial cells. ⁴ Treatment incubations were for 24 hours in most of the experiments or during a time course (1, 4, 8, 16, and 24 hours). The extraction protocol for ET-1 was performed as previously described by Prasanna et al. ⁴ Efficiency of ET-1 recovery was 75% ± 3% (n = 3). Measurement of immunoreactive ET-1 (ir-ET-1) was according to manufacturer’s instructions in a commercially available RIA kit for ET-1 (Peninsula Laboratories, Belmont, CA). ⁴  

Intracellular Ca²⁺ Measurement

Intracellular Ca²⁺ ([Ca²⁺]_i) in y- and mRPE cells was measured at 37°C by the ratiometric technique using fura-2AM (excitation at 340 nm and 380 nm, emission at 500 nm) according to Prasanna et al. ³⁷  

Total RNA Extraction, cDNA Synthesis, and Quantitative Reverse Transcriptase–Polymerase Chain Reaction

Total RNA was isolated from y- and mRPE cells, grown in 100-mm dishes to subconfluent or confluent states and treated as described earlier. Extraction reagent (TRIzol; Invitrogen) was used for total RNA isolation, as previously described. ³⁸ Five micrograms of total RNA was used to synthesize the corresponding cDNA, using avian myeloblastosis virus (AMV) reverse transcriptase (Promega, Madison, WI) and random primers (Promega) in a reaction volume of 50 μL at 42°C for 30 minutes. Quantitative PCR (QPCR) primers for human preproET-1 (ppET-1; Fisher Scientific-Genosys, Plano, TX) were designed from the respective cDNA sequence, using an automated sequencing program (GeneJockey II; Biosoft, Ferguson, MO). The primer sequences for human ppET-1 were designed so that they spanned different exons. β-Actin served as an internal control that accounted for variability in the initial concentration, quality of the total RNA, and conversion efficiency of the reverse transcription reaction. The primer sequences for ppET-1 and β-actin were as follows: ppET-1-forward/sense 5′-TATCAGCAGTTAGTGAGAGG-3′ and reverse/antisense 5′-CGAAGGTCTGTCACCAATGTGC-3′, with an expected amplicon/product size of 180 bp; β-actin- forward/sense 5′-TGTGATGGTGGGAATGGGTCAG-3′ and reverse/antisense 5′-TTTGATGTCACGCACGATTTCC-3′ with an expected amplicon/product size of 514 bp. 

QPCR was performed as described by Zhang et al. ¹¹ Product authenticity was confirmed by DNA sequencing followed by a BLAST homology search of the resulting sequences (data not shown). Quantitation of relative ppET-1 transcript levels in ARPE-19 was achieved using the comparative threshold cycle (C_T) method, as described by the manufacturer (User Bulletin #2: ABI Prism 7700 Sequence Detector; http://docs.appliedbiosystems.com/pebiodocs/04303859.pdf). 

QPCR data are presented as the mean percentage of the value of its corresponding untreated control in three separate experiments. 

Membrane Isolation and Immunoblot Analysis

Membrane preparation and immunoblot analysis were performed according to Zhang et al. ¹¹ Sixty micrograms per lane (for M₁ receptor detection), 100 μg/lane (for M₃ receptor detection), or 50 μg/lane (for ZO-1 detection) of the membrane protein was boiled with the SDS page sample buffer and loaded onto 7.5% SDS-PAGE gels and run at 80 V for 2 to 3 hours. Gel contents were electrophoretically transferred to 0.45-μm pore size polyvinylidene difluoride (PVDF) membrane (Millipore, Bedford, MA) overnight at 30 V and at 4°C. Blots were probed with the desired primary antibody (in 3% nonfat dry milk): rabbit anti-muscarinic M₁ receptor (final concentration, 0.7 μg/mL), goat anti-muscarinic M₃ receptor (2 μg/mL), or mouse anti-ZO-1 (1.5 μg/mL) for 1 hour. The rat heart lysate was used as a control for the muscarinic M₁ receptor (Santa Cruz Biotechnology) and the M₃ blocking peptide (control antigen; Santa Cruz Biotechnology) for the M₃ receptor detection according to the manufacturer’s instructions. After secondary antibody incubation, blots were rinsed and developed using enhanced chemiluminescence (ECL) reagents (Amersham Pharmacia Biotech). 

Immunofluorescence Microscopy

yRPE and mRPE cells were grown on 25-mm glass coverslips for the desired period and treated as indicated. Cells were fixed in 4% paraformaldehyde in PBS (15 mM KCl, 468 mM NaCl, 580 mM Na₂HPO₄.7H₂O and 27 mM KH₂PO₄) for 30 minutes at room temperature followed by permeabilization with 0.2% Triton X-100 for 15 minutes. Cells were rinsed in PBS and incubated twice in 50 mM glycine, 15 minutes per incubation. Each coverslip was carefully inverted (cell-side facing solution) onto 200 μL of blocking solution containing 3% BSA+3% normal goat serum in PBS for 30 minutes. The coverslips were then incubated in rabbit anti-ET-1 (10 μg/mL) for 4 hours at 4°C followed by incubation in a mixture containing rabbit anti-ET-1 (10 μg/mL) and mouse anti-ZO-1 (10 μg/mL), overnight at 4°C. The antibody used for intracellular ET-1 detection was the same antibody as that used in the ET-1 RIA measurements. Coverslips were rinsed and allowed to incubate in a mixture of secondary antibodies containing Alexa 594–conjugated donkey anti-mouse (5 μg/mL) and Alexa 488–conjugated donkey anti-rabbit (5 μg/mL) for 1 hour in the dark at room temperature. Nuclei were stained with DAPI (300 nM; Molecular Probes) for 10 minutes. Coverslips were mounted on glass slides in antifade medium (FluorSave; Calbiochem, La Jolla, CA) and allowed to dry for 20 minutes in the dark. Cells were viewed with a digital fluorescent microscope (Microphot FXA; Nikon, Tokyo, Japan) and images at the red, green, and blue wavelengths were acquired with a CCD-camera and digitally processed using image-analysis software (IPLab; Scanalytics, Billerica, MA). All images were deconvolved with the same software and transferred for further analysis (Photoshop, ver. 7.0; Adobe Systems, Mountain View, CA). 

Data Analysis

Quantitative data are represented as the mean ± SEM. Statistical comparisons were performed by t-test in most experiments, except for ET-1 RIA measurements, for which comparisons between control and multiple treatments were made with ANOVA and the Student-Newman-Keuls (SNK) test. In [Ca²⁺]_i measurements, comparisons between baseline, peak, and 1-minute postpeak values were made by one-way, repeated-measures ANOVA. Sample size and probabilities for each experiment are indicated in the figure legends. 

Results

In the present study we compared two different phenotypes of RPE in cell culture: a mature culture (4–5 weeks, mRPE), similar to an intact polarized and differentiated epithelium with few intercellular gaps, and a young culture (3–4 days, yRPE) that has an incomplete barrier formed with more intercellular gaps and possible lack of polarization. The two RPE cell phenotypes were tested for their ability to secrete ET-1 after application of a cytokine (TNF-α) or a cholinergic agonist (CCh). 

Secretion and Regulation of ET-1 in ARPE-19 Cells

yRPE and mRPE cells were incubated with either TNF-α (10 nM) or CCh (0.01–100 μM), a nonselective muscarinic receptor agonist for 24 hours (Fig. 1) . In yRPE cells (Fig. 1A) , TNF-α and CCh (1 μM) significantly enhanced ir-ET-1 secretion compared with the untreated control. The extent to which TNF-α potentiated ir-ET-1 secretion was higher than that produced by CCh 1 μM. It was only at the 1-μM concentration that CCh was consistently able to stimulate ET-1 release in yRPE. There are five muscarinic receptor subtypes (M_1–5), ^{26 39} of which M₁, M₃, and M₅ are directly coupled to the G_q-IP₃-Ca²⁺ signaling pathway, whereas the M₂ and M₄ subtypes are coupled to the G_i-cAMP cascade. Activation of the IP₃-Ca²⁺ cascade requires prior activation of the G-protein–coupled receptor-mediated transducer phospholipase Cβ (PLCβ). ^{40 41} To determine whether this was the mechanism responsible for CCh-mediated ir-ET-1 release, yRPE cells were preincubated with 2 μM U73122, a PLC inhibitor, for 20 to 30 minutes with subsequent stimulation with 1 μΜ CCh. U73122 completely inhibited ET-1 release suggesting that activation of PLC was a critical determinant in CCh-mediated ET-1 release (Fig. 1A) . Selective muscarinic receptor antagonists were then used for further delineation of the receptor subtype(s) that were involved in ET-1 release in yRPE cells. The compounds 4-DAMP, a selective M₁/M₃ receptor antagonist (pKi: 9.4 and 9.1 for M₁ and M₃, respectively), and pirenzepine (PZE), an M₁-selective antagonist (pKi: 6.9 for M₁), ^{42 43} were used in our study. Both 4-DAMP and PZE were effective in inhibiting CCh-induced ET-1 release with an apparent relative order of potency of 4-DAMP > PZE (Fig. 1A) . mRPE cells were allowed to grow for 4 weeks to form an intact epithelial barrier (see Fig. 5 ). The basal amounts (untreated controls) of both ppET-1 mRNA (see Fig. 8 ) and secreted ir-ET-1 measured in these cells (mRPE) was higher than yRPE cells after 24 hours (compare scales in Figs. 1A and 1B ). TNF-α continued to potentiate ET-1 secretion in mRPE cells to the same degree as that observed in yRPE (four to fivefold increase over control). A similar increase in released ET-1 was not observed in mRPE after CCh treatments for 24 hours (Fig. 1B) . 

M₁ and M₃ Muscarinic Receptors in ARPE-19

Differences in secretion of ir-ET-1 in m- and yRPE cells in response to CCh and TNF-α may be due to differential expression of M₁ and M₃ receptors or differences in the intracellular calcium ([Ca²⁺]_i) trends after muscarinic receptor activation. The primary reason M₁ and M₃ receptor subtypes were considered was that 4-DAMP and PZE were effective in inhibiting CCh-mediated ET-1 release in the yRPE (Fig. 1A) . In addition, CCh mediated phosphoinositide hydrolysis and the subsequent increase in [Ca²⁺]_i in human RPE cells may be predominantly M₃ receptor mediated. ¹⁶ Because 4-DAMP has a 6- to 13-fold lower affinity for the M₅ receptor compared with that for M₃ or M₁ receptors ⁴² and the concentrations we used in our studies and previous reports on muscarinic receptor expression in RPE, ^{15 44} it was evident that either the M₃ or M₁ or both subtypes were principal targets for CCh-induced ET-1 secretion in yRPE. 

Both M₁ and M₃ receptors are expressed in ARPE-19 cells (y- and mRPE; Fig. 2 ). The apparent molecular weights of the protein bands (M₁R: 60 kDa and M₃R: 70 kDa) were confirmed by using appropriate controls (rat heart lysate for M₁R and the blocking peptide for M₃R). Detection of the M₃ receptor subtype in y- and mRPE cells required higher amounts of total protein (100 μg/lane for M₃ detection as opposed to 60 μg/lane for M₁ detection). 

CCh-Mediated [Ca²⁺]_i Mobilization in y- and mRPE Cells

M₁ and M₃ muscarinic receptors belong to the class of G_q-coupled receptors that on activation mobilize [Ca²⁺]_i in an IP₃-dependent manner. ⁴⁵ Receptor mediated increase in [Ca²⁺]_i can activate several downstream effectors including those involved in regulated exocytosis in both excitable and nonexcitable cells. ⁴⁶ RPE cells express muscarinic receptors that mobilize [Ca²⁺]_i but not cAMP after acetylcholine or CCh. ¹⁵ Because the immunoblot analysis indicated there were differences in the expression of M₁ and M₃ receptors between y- and mRPE cells, it was possible that functional differences between the receptors existed as well. Representative [Ca²⁺]_i trends in y- and mRPE cells after CCh (1, 10, 100 μM) stimulation are shown in Figures 3A and 3B , respectively, and a summary of the results are shown in Tables 1 and 2 . There was a concentration-dependent increase in the mean [Ca²⁺]_i mobilized by CCh in both y- and mRPE with characteristic biphasic transients observed in both phenotypes (Fig. 3) . Because 1 μΜ CCh was effective in evoking significant release of ET-1 in yRPE cells (Fig. 1A) , this concentration was used for all future experiments in these cells. To address the receptor subtypes that were functionally coupled to the CCh response, different concentrations of 4-DAMP (M₁/M₃ selective inhibitor) and PZE (M₁ selective inhibitor) were used. U73122 (2 μΜ) blocked CCh-mediated [Ca²⁺]_i mobilization (Table 1) , consistent with results observed in which similar treatments effectively blocked CCh-mediated ET-1 secretion in yRPE cells (Fig. 1A) . PZE (100 nM) failed to inhibit CCh-mediated [Ca²⁺]_i elevation and required a concentration of 400 nM or more to do so (Table 1) . 4-DAMP unlike PZE, continued to inhibit CCh-mediated [Ca²⁺]_i elevation at concentrations similar to those used in Figure 1A . 

To determine whether the RIA results were due to functional uncoupling of muscarinic receptors (receptor desensitization and/or internalization) in their ability to mobilize [Ca²⁺]_i, cells were preincubated with CCh (1 and 100 μM) for 24 hours followed by CCh (1 and 100 μM) challenge (Table 1) . Cells pretreated with CCh 1 μM were able to retain approximately 50% of their response to acute CCh 1 μM as opposed to CCh 100 μM, where the response was reduced to 10% of the initial response without pretreatment. Similar results were observed in mRPE cells (Table 2) in their ability to mobilize [Ca²⁺]_i after CCh (1, 10, 100 μM) additions before or after pretreatments. A comprehensive study using muscarinic receptor antagonists were not performed in the mRPE cells, because CCh failed to evoke any significant increase in ET-1 release in these cells (Fig. 1B) . However, 4-DAMP (5 and 10 nM) completely inhibited CCh-mediated [Ca²⁺]_i elevation in these cells (data not shown), similar to that observed in yRPE cells, suggesting that both phenotypes had similar responses to CCh in their ability to mobilize [Ca²⁺]_i. 

Disruption of Tight Junctions by TNF-α and Its Influence on Intracellular and Secreted ET-1

Several recent studies have proposed that tight junction and sub–tight-junction domains form clusters of scaffolding proteins that could be important in regulating paracellular transport, cell motility, membrane integrity, and recruitment of exocytotic machinery. ⁴⁷ We hypothesized that the presence of a mature tight junction complex may regulate secretion of ET-1 in mRPE cells. mRPE cells expressed abundant amounts of ZO-1, a peripheral tight junction–associated protein in epithelial cells. Immunoblot and immunofluorescence analysis was used to determine the extent of ZO-1 expression in both phenotypes. There was a significant increase in ZO-1 expression in mRPE as opposed to yRPE cells (50 μg/lane, n = 3) with both isoforms of ZO-1 (α+ and α−; data not shown). Immunofluorescence studies demonstrated that the mRPE cells expressed greater amounts of ZO-1 with well-defined tight junctions compared with yRPE cells (Figs. 4 5) . 

TNF-α caused visible changes in morphology, an increase in intercellular gaps and disruption of tight junctions in both y- and mRPE cells (Figs. 4 5) . These changes were not observed after CCh treatment. Intracellular ET-1 was detected as punctate stains in both phenotypes. Although there were differences in the intensities of intracellular ET-1 content between cells on the same coverslip, basal intracellular ET-1 in yRPE was visibly higher than in mRPE cells (compare Figs. 4C and 5C ). Negative controls including nonimmune serum (Fig. 5) and no primary or secondary antibodies (data not shown) confirmed the authenticity of detection. 

Time-Dependent Changes in ZO-1 and ET-1 after TNF-α

To determine whether TNF-α–mediated changes in mRPE cells were time dependent, we measured the amount of secreted ir-ET-1 (Fig. 6) , immunofluorescent ZO-1, and intracellular ET-1 (Fig. 7) expression at 1, 4, 8, 16, and 24 hours after treatment with TNF-α. Constitutive ppET-1 (preproET-1) mRNA expression in mRPE cells was more than four times that in yRPE cells during 24 hours (Fig. 8) . This finding was in agreement with differences in constitutive secretion of mature ir-ET-1 over the same time period (Figs. 1A 1B) . ppET-1 mRNA expression was measured in response to TNF-α or CCh in y- and mRPE cells at the indicated time points (Fig. 9) . TNF-α significantly increased the amount of secreted ET-1 at the end of 8 hours (Fig. 6) and continued to do so until 16 and 24 hours, when the highest amounts of secreted ET-1 were measured. In addition, TNF-α caused visible alteration in cellular morphology and disruption of tight junctions, an effect that was first detected at 8 hours and persisted until 24 hours (Fig. 7) . ppET-1 mRNA levels in mRPE cells were significantly elevated (∼4-fold) at the end of the first hour after TNF-α stimulation and gradually decreased to approximately 1.5-fold of control at the end of 24 hours (Fig. 9B) . There was a gradual and significant increase in ET-1 release at 8, 16, and 24 hours after TNF-α administration (Fig. 6) . CCh failed to increase ppET-1 transcription in both phenotypes at the end of 24 hours. 

Discussion

Several physiological or pathophysiological stimuli can cause the release of ET-1 and are considered to be regulators of the secretion of ET-1. Cytokines, including TGF-β, ⁴⁸ IL-1, ⁴⁹ TNFα, ^{4 50} and interferon (IFN)-γ, ⁵¹ can induce both transcription and release of ET-1. ET-1 secretion typically results from activation of the constitutive or regulated pathways. ^{52 53}  

Unlike CCh, TNF-α can influence mRNA synthesis and secretion of ET-1, ⁵⁴ and several studies have reported the ability of TNF-α to cause cytoskeletal changes including breakdown of the tight junction barrier in epithelial cells and RPE. ^{55 56 57} TNF-α has been shown to decrease the turnover of occludin, one of the integral proteins of the tight junction complex. ⁵⁸ In yRPE cells, the tight junction complex appears to be premature and the epithelial phenotype to be nonpolarized and undifferentiated. ^{31 59} Failure to recruit desired proteins at the tight junction complex and execution of the barrier, fence, and signaling functions ⁴⁷ may alter both constitutive and regulated release (by CCh and TNF-α) of ET-1 in RPE cells. Constitutive synthesis and secretion of ET-1 was higher in mature RPE cells than in young RPE cells. However, basal intracellular ET-1 (endogenous ET-1) content appeared to be higher in y- than in mRPE cells. This suggests that the rate of constitutive secretion may be higher in m- than in yRPE cells. Constitutive and regulated secretion may be influenced by several factors including polarization by plasma membrane asymmetry and/or Golgi asymmetry, differential sorting of proteins including membrane receptors, and decreased paracellular permeability. The finding that yRPE (nonpolarized) cells had higher amounts of ET-1 secretion when stimulated with CCh than did mRPE (polarized) cells was consistent with this view. In contrast, TNF-α enhanced secretion of ET-1 by four- to fivefold at the end of 24 hours in both phenotypes. This was probably due to TNF-α’s ability not only to enhance ET-1 secretion but also to increase ppET-1 transcription significantly and disrupt tight junctions in RPE cells, all of which were time dependent in mRPE cells. CCh, on the contrary, although able to regulate ET-1 secretion in yRPE cells, failed to enhance ppET-1 transcription or cause significant alterations in cell shape and membrane integrity. Of interest, CCh mobilized [Ca²⁺]_i to a similar degree in both y- and mRPE cells, indicating that the cumulative muscarinic receptor–mediated [Ca²⁺]_i trends remained unaltered in either phenotype and that ET-1 release mechanisms may not necessarily be coupled to calcium mobilization alone. This was particularly evident in yRPE where significant ET-1 release only occurred at the 1-μM concentration of CCh. The inability of lower (<1 μM) and higher (10–100 μM) concentrations of CCh to elicit similar or greater release of ET-1 may have been due to the inability to mobilize the required [Ca²⁺]_i and to activate the ET-1 secretory pathway at lower concentrations and/or due to receptor desensitization and internalization at higher concentrations. 

Our results suggest that the CCh-mediated ET-1 release may predominantly involve the M₃ muscarinic receptor subtype. However, we cannot totally exclude the participation of M₁ receptors, because 4-DAMP has similar affinities for M₁ as for M₃ receptors. ^{42 43} In addition, PZE at 100 nM inhibited CCh-mediated ET-1 release in yRPE cells. Concentrations of 400 nM PZE or more were necessary to inhibit 1 μM CCh-mediated [Ca²⁺]_i increase in yRPE cells, suggesting that most of this increase was M₃ mediated and that CCh-mediated ET-1 release was both M₁ and M₃ dependent. The inability of CCh to increase ET-1 secretion in mRPE may have been due to limited paracellular permeability in mRPE cells in addition to recruitment of the tight junction complex that may affect its actions on M₁ receptors. 

In conclusion, these results favor a role for CCh in the regulated release of ET-1 in yRPE cells, whereas the actions of TNF-α reflect a generalized disturbance in cell morphology, disruption of tight junctions, and enhanced ppET-1 transcription, so that the increased release of ET-1 after TNF-α may occur through de novo synthesis and release of ET-1. Such actions would be reminiscent of an inflammatory response during breakdown of the blood–retinal barrier, as seen in proliferative vitreoretinopathy (PVR) and diabetic retinopathy. ¹² Our results suggest that the RPE may be the source for ET-1 at the posterior pole of the eye. The implications for physiological function of ET-1 under normal conditions in the RPE are presently unknown. In diseased conditions, a pathologic increase in ET-1 secretion by barrier-compromised RPE may be important in cell migration and proliferation, as seen in PVR. Locally secreted ET-1 could act to produce vasoconstriction and affect cellular responses that minimize damage to a compromised blood–retinal barrier. In addition, released ET-1 may mediate vascular homeostasis as a mechanism to balance vasodilator influences; however, with excessive secretion, ET-1 may promote prolonged vasoconstriction and induce ischemic episodes in the retina. We are presently working on models that will address the role of ET-1 at the outer blood–retinal barrier. 

 Supported by National Eye Institute Grant EY11979 (TY).

 Submitted for publication April 16, 2003; revised June 18, 2003; accepted July 18, 2003.

 Disclosure: S. Narayan, None; G. Prasanna, None; R.R. Krishnamoorthy, None; X. Zhang, None; T. Yorio, None

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

 

Figure 1.

 

ET-1 RIA in yRPE (ARPE-19 cells grown for 4 days) and mRPE (ARPE-19 cells grown for 4 weeks) cells. Cells were treated with various agonists and/or antagonists for 24 hours in serum-free DMEM/F-12 medium. Immunoreactive ET-1 (ir-ET-1) released in the media was extracted and measured by RIA. (A) In yRPE cells, TNF-α and CCh (1 μM) significantly increased ir-ET-1 secretion versus control. U73122, 4-DAMP, and PZE inhibited CCh-mediated ir-ET-1 secretion. (B) In mRPE cells, TNF-α significantly increased ir-ET-1 secretion versus the control. Note the difference in the x-axis scales. mRPE cells produced higher amounts of ir-ET-1 (compare controls in A and B). Data are expressed as the mean ± SEM. Statistical comparisons were performed using ANOVA and the SNK test. *Significance versus control (P < 0.05) of at least six treatments.

Figure 2.

 

Immunoblot analysis in yRPE and mRPE cells: (A) Total membrane fraction loaded per lane was 60 μg. Blots probed with the rabbit anti-M₁ receptor. Rat heart lysate (30 μg) was used as a control to verify the apparent size (∼60 kDa) and expression of the M₁ receptor. (B) Total membrane fraction per lane required to detect M₃ receptor expression in both phenotypes was 100 μg. The goat anti-M₃ receptor antibody was used for detection. The apparent size of the M₃ receptor was ∼70 kDa. Control antigen (bottom) preincubated with the M₃ antibody for 1 hour at room temperature was used as the negative control (n = 4 per condition).

Figure 3.

 

Intracellular [Ca²⁺]_i measurements in y- and mRPE cells. Representative [Ca²⁺]_i trends in response to CCh (1, 10, 100 μM) in (A) yRPE and (B) mRPE cells. In both phenotypes, there was a concentration-dependent increase in CCh-mediated mean [Ca²⁺]_i mobilization (see Tables 1 2 ) as well as an increase in biphasic transients.

Table 1.

 

Summary of CCh-Mediated [Ca²⁺]_i Mobilization in yRPE Cells Measured by Fura-2AM Imaging

Treatment	[Ca2+]i	Cells (n)
CCh dose response
Baseline	74 ± 5	98
CCh 1 μM	1504 ± 154*	98
Baseline	57 ± 5	55
CCh 10 μM	1629 ± 276*	55
Baseline	137 ± 21	26
CCh 100 μM	2851 ± 1392*	26
Antagonist studies with CCh 1 μM
Baseline	65 ± 5	34
CCh 1 μM	1414 ± 203*	34
U73122, 2 μM/Baseline	132 ± 12	26
U73122, 2 μM + CCh 1 μM	177 ± 14	26
PZE 100nM/Baseline	68 ± 7	16
PZE 100nM + CCh 1 μM	1584 ± 360*	16
PZE 400nM/Baseline	90 ± 8	17
PZE 400nM + CCh 1 μM	181 ± 30	17
4-DAMP 10nM/Baseline	130 ± 6	68
4-DAMP 10nM + CCh 1 μM	122 ± 5	68
Receptor internalization studies
No Pretreatment
Baseline	65 ± 5	34
CCh 1 μM	1414 ± 203*	34
Pretreatment with CCh 1 μM (24 h)
Baseline	77 ± 5	21
CCh 1 μM	736 ± 308*	21
No Pretreatment
Baseline	137 ± 21	26
CCh 100 μM	2851 ± 1392*	26
Pretreatment with CCh 100 μM (24 h)
Baseline	156 ± 9	35
CCh 100 μM	202 ± 16	35

Table 2.

 

Summary of CCh-Mediated [Ca²⁺]_i Mobilization in mRPE Cells Measured by Fura-2AM Imaging

Treatment	[Ca2+]i	Cells (n)
CCh dose response
Baseline	101 ± 8	35
CCh 1 μM	1260 ± 270*	35
Baseline	105 ± 11.1	57
CCh 10 μM	1352 ± 158*	57
Baseline	72 ± 6	54
CCh 100 μM	2349 ± 209*	54
Receptor internalization studies
No Pretreatment
Baseline	100 ± 9	33
CCh 1 μM	1236 ± 316*	33
Pretreatment with CCh 1 μM (24 h)
Baseline	117 ± 28	22
CCh 1 μM	411 ± 134*	22
No Pretreatment
Baseline	65 ± 9	16
CCh 100 μM	2500 ± 532*	16
Pretreatment with CCh 100 μM (24 h)
Baseline	87 ± 10	22
CCh 100 μM	82 ± 22	22

Figure 4.

 

Indirect immunofluorescence microscopy in yRPE cells. Cells were treated with TNF-α or CCh for 24 hours, and expression of ZO-1 and ET-1 were analyzed by light microscopy. (A–D) Control/untreated cells; (E–H) TNF-α (10 nM)–treated cells; and (I–L) CCh (1 μM)-treated cells. Fixed cells were probed with mouse anti ZO-1. (B–F) and rabbit anti ET-1. (C–K) followed by incubation with goat anti-mouse Alexa 594 antibody and goat anti-rabbit Alexa 488 antibody. Nuclei were stained with DAPI (blue fluorescence) Merged images are shown in (D), (H), and (L). (A), (E), and (I) represent the differential interference contrast (DIC) images. TNF-α but not CCh caused visible changes in morphology and disruption of tight junctions (ZO-1 staining). At least five different fields were viewed per coverslip, and three coverslips per condition (n = 4) were tested under similar conditions. Scale, 10 μm.

Figure 5.

 

Indirect immunofluorescence microscopy in mRPE cells. Cells were treated with TNF-α or CCh for 24 hours, and expression of ZO-1 and ET-1 was analyzed by light microscopy. (A–D) Control/untreated cells; (E–H) TNF-α (10 nM)–treated cells; and (I–L) CCh (1 μM)-treated cells. Fixed cells were probed with mouse anti ZO-1 (B–F) and rabbit anti ET-1 (C–K), followed by incubation with donkey anti-mouse Alexa 594 antibody (5 μg/mL, red fluorescence) and donkey anti-rabbit Alexa 488 antibody (5 μg/mL, green fluorescence). Nuclei were stained with DAPI (300 nM, blue fluorescence), and merged images are shown in (D), (H), and (L). TNF-α but not CCh caused visible changes in morphology and disruption of tight junctions (ZO-1 staining). At least five different fields were viewed per coverslip and a total of three coverslips per condition (n = 3) were tested under similar conditions. Negative controls (M–P) using nonimmune IgG at concentrations identical for anti-ZO-1 and anti-ET-1 showed little or no staining. Scale bar, 10 μm.

Figure 6.

 

Time-dependent increase in ir-ET-1 secretion in confluent ARPE-19 cells (4 weeks old) after TNF-α stimulation. Confluent ARPE-19 cells were treated with TNF-α (10 nM) for 1, 4, 8, 16, and 24 hours. The medium was collected and assayed for ir-ET-1 content, as previously described. TNF-α stimulated ir-ET-1 secretion in a time- dependent manner. A significant increase in ir-ET-1 was observed at the end of 8, 16, and 24 hours compared with control. Secretion of ir-ET-1 reached a plateau after 16 hours. Data are represented as the mean ± SEM. Statistical comparisons were performed by t-test. Significance versus controls at 8, 16, and 24 hours, respectively (*P < 0.001; n = 6).

Figure 7.

 

Immunofluorescence analysis in mRPE cells after TNF-α at the indicated time points. (A–F) Differential interference contrast (DIC) images; (G–L) merged fluorescence images of cells labeled with mouse anti-ZO-1 (red), rabbit anti-ET-1 (green), and DAPI (blue; similar to Fig. 5 ). TNF-α (10 nM) caused visible breakdown of cell–cell contact and tight junction disruption that was time dependent. The first detectable change in cell–cell contact was observed at 4 hours, and progressive damage was seen thereafter. Scale bar, 10 μm.

Figure 8.

 

Quantitative RT-PCR in y- and mRPE cells. Quantitative RT-PCR was performed with PCR core reagents (SYBR-green; Applied Biosystems, Foster City, CA). Quantitation of ppET-1 transcripts was performed by the comparative C_T method. Bar graph represents basal levels of ppET-1 mRNA expression in y- and mRPE cells. Data are represented as mean ± SEM. Statistical comparisons were performed by t-test. *Denotes significance versus control (P < 0.05; n = 4 per treatment).

Figure 9.

 

Quantitative RT-PCR in y- and mRPE cells. (A, B) ppET-1 mRNA levels after CCh (1, 10, 100 μM) and TNF-α (10 nM) in y- and mRPE cells, respectively. CCh (1, 10, 100 μM) did not result in significant elevation of ppET-1 mRNA in yRPE (A) or mRPE cells (B) after 24 hours. TNF-α (10 nM) significantly increased ppET-1 mRNA after 24 hours (A) and within 1, 4, and 8 hours versus control (B). Data are represented as the mean ± SEM. Statistical comparisons were performed by t-test. *Significance versus control (P < 0.05; n = 4 per treatment).

The authors thank Anne Marie Brun and Larry Oakford for help with and suggestions on the immunofluorescence analysis, Jerry Simeka and Xiangle Sun for help with the real time RT-PCR analysis, and Christina Hulet for technical assistance.