Molecular Vision 2020; 26:797-817
Received 31 March 2020 | Accepted 28 December 2020 | Published 30 December 2020
1Department of Ophthalmology and Eye Hospital, University of Leipzig, Leipzig, Germany; 2Institute of Anatomy, Medical Faculty, University of Leipzig, Germany
Correspondence to: Margrit Hollborn, Department of Ophthalmology and Eye Hospital, University of Leipzig, Faculty of Medicine, Liebigstrasse 10-14, D-04103 Leipzig, Germany; Phone: +49 (0) 341 97 21 561; FAX: +49(0) 341 97 21 659; email: firstname.lastname@example.org
Purpose: The expression of aquaporin-8 (AQP8), which plays a crucial role in the maintenance of the cellular fluid and electrolyte balance, was shown to be increased in RPE cells under hyperosmotic conditions. The aim of the present study was to investigate the mechanisms of hyperosmotic AQP8 gene expression and the localization of AQP8 in cultured human RPE cells.
Methods: Hyperosmolarity was produced with the addition of 100 mM NaCl or 200 mM sucrose. Hypoxia was induced by cell culture in a 0.2% O2 atmosphere or the addition of the hypoxia mimetic CoCl2. Oxidative stress was induced by the addition of H2O2. Gene expression was determined with real-time RT–PCR analysis. AQP8 protein localization and secretion of VEGF were evaluated with immunocytochemistry, western blotting, and enzyme-linked immunosorbent assay (ELISA).
Results: Immunocytochemical and western blot data suggest that the AQP8 protein is mainly located in the mitochondria. Extracellular hyperosmolarity, hypoxia, and oxidative stress induced increases in AQP8 gene expression. Hyperosmotic AQP8 gene expression was reduced by inhibitors of the p38 MAPK and PI3K signal transduction pathways, and by JAK2 and PLA2 inhibitors, and was in part mediated by the transcriptional activity of CREB. Hyperosmotic AQP8 gene expression was also reduced by autocrine/paracrine interleukin-1 signaling, the sulfonylureas glibenclamide and glipizide, which are known inhibitors of KATP channel activation, and a pannexin-blocking peptide. The KATP channel opener pinacidil increased the expression of AQP8 under control conditions. The cells contained Kir6.1 and SUR2B gene transcripts and displayed Kir6.1 immunoreactivity. siRNA-mediated knockdown of AQP8 caused increases in hypoxic VEGF gene expression and secretion and decreased cell viability under control, hyperosmotic, and hypoxic conditions.
Conclusions: The data indicate that hyperosmotic expression of AQP8 in RPE cells is dependent on the activation of KATP channels. The data suggest that AQP8 activity decreases the hypoxic VEGF expression and improves the viability of RPE cells which may have impact for ischemic retinal diseases like diabetic retinopathy and age-related macular degeneration.
Development of retinal edema is an important complication of various vision-threatening diseases, including exudative (neovascular) age-related macular degeneration (AMD) and diabetic retinopathy [1,2]. Edema is characterized by water accumulation in retinal tissue. In exudative AMD, fluid accumulation occurs in the subretinal space resulting in functional impairment of photoreceptors and serous retinal detachment. Water accumulation within retinal tissue results from an imbalance between the water influx from the blood into the retina and water clearance from retinal tissue into the blood . Normally, fluid absorption from retinal tissue is mainly mediated by the coupled transport of osmolytes (in particular, of potassium and chloride ions) and water through glial and RPE cells [3-6]. The transcellular water transport is facilitated by aquaporin (AQP) water channels. Thirteen members of the AQP protein family (AQP0‒12) were identified in mammals which mediate bidirectional movement of water across membranes in response to osmotic gradients and differences in hydrostatic pressure. Various AQP subtypes also mediate the transmembrane transport of small noncharged solutes, such as glycerol, lactate, urea, ammonia, and H2O2 . Facilitated water transport is important for the permission of rapid ion currents and the resolution of osmotic gradients within tissues and across membranes; the latter is important for the integrity and volumes of cells and mitochondria. Human RPE cells were reported to express gene transcripts of various AQP subtypes, including AQP1, AQP3, AQP5, and AQP8 [5,8-10].
Osmotic gradients between the blood and retinal tissue, and between intra- and extracellular compartments, contribute to the development of retinal edema . Hyperglycemia, which increases extracellular osmolarity , is the primary risk factor, and systemic hypertension is the main secondary risk factor of diabetic retinopathy [13,14]. In addition, the increased glucose flux through the polyol pathway produces intracellular sorbitol accumulation and increased intracellular osmotic pressure . Hypertension is also a risk factor of AMD [16,17]. The main condition that causes acute hypertension is increased extracellular osmolarity following intake of dietary salt (NaCl) . In experimental diabetic retinopathy, the expression of retinal AQPs is altered [19,20]; high salt intake aggravates the diabetic alterations of retinal AQP expression independently from changes in blood pressure . It was shown that extracellular hyperosmolarity induces the expression of AQP3 (Gene ID: 360; OMIM: 600170), AQP5 (Gene ID: 362; OMIM: 600442), and AQP8 (Gene ID: 343; OMIM: 603750) genes in human RPE cells [8,10]. Expression of the AQP5 gene in RPE cells was found to be regulated by extracellular osmolarity, with up- and downregulation in response to hyper- and hypo-osmotic conditions, respectively . However, the mechanisms of hyperosmotic AQP8 gene expression in RPE cells was not investigated until today. In various cell types, AQP8 is localized to the plasma membrane, intracellular vesicles, or inner mitochondrial membrane [22‒24]. Upon stimulation, AQP8 localized to secretory vesicles is inserted into the plasma membrane to increase the osmotic water and H2O2 membrane permeability . H2O2 plays a key role in the regulation of tyrosine phosphatase and kinase signaling induced, for example, by activation of growth factor receptors, like vascular endothelial growth factor (VEGF) receptors [26,27]. AQP8 localized to the inner mitochondrial membrane facilitates the efflux of metabolic water, which is a byproduct of adenosine 5′-triphosphate (ATP) synthesis, thus preventing mitochondrial swelling [23,24]. AQP8 in mitochondria also facilitates the transmembrane diffusion of solutes like H2O2 [28,29] and ammonia/ammonium, and thus, contributes to maintenance of the acid-base equilibrium, regulation of the cellular and mitochondrial oxidative stress levels, and detoxification of ammonia via mitochondrial urea synthesis [30-32]. However, the subcellular localization of AQP8 in RPE cells is unknown. The present study was performed to investigate in cultured human RPE cells the subcellular localization of AQP8 and the cellular mechanisms of hyperosmotic AQP8 expression, and to investigate whether knockdown of AQP8 has effects on three important physiologic properties of RPE cells implicated in retinal diseases in situ: cell proliferation and viability, and expression and secretion of VEGF. Because we found that hypoxia induces delayed expression of the AQP8 gene, we investigated the effects of AQP8 knockdown under hyperosmotic and hypoxic conditions. We also found that inhibition of the activation of ATP-sensitive potassium (KATP) channels reduces hyperosmotic AQP8 gene expression; therefore, we investigated the expression of Kir6.1 (Gene ID: 3764; OMIM: 600935) and Kir6.2 (Gene ID: 3767; OMIM: 600937) in the cells.
The use of human material was approved by the Ethics Committee of the University of Leipzig (#745, 07/25/2011) and was performed according to the Declaration of Helsinki. Postmortem eyes from human cornea donors without reported eye disease were obtained within 48 h of death; written informed consent for the use of retinal tissue in basic science was obtained from the relatives of each donor.
Human recombinant interleukin-1β (IL-1β), IL-1 receptor antagonist, platelet-derived growth factor-BB (PDGF), transforming growth factor-β1 (TGF-β1), and VEGF-A165 were purchased from R&D Systems (Abingdon, UK). The following compounds were obtained from Calbiochem (Bad Soden, Germany): AG490, 8-(3-chlorostyryl) caffeine (CSC), 8-cyclopentyl-1,3-dipropylxanthine (DPCPX), Gö6976, H-89, and U73122. Dithiothreitol was from Carl Roth (Karlsruhe, Germany), and PD173074 was kindly provided by Pfizer (Karlsruhe, Germany). 666–15, A-438079, AR-C 118925XX, caffeic acid phenethyl ester (CAPE), cyclosporin A, GSK650394, the hypoxia-inducible transcription factor (HIF)-1 inhibitor, LY294002, human recombinant matrix metalloproteinase-2 (purified active form), MRS2179, NS-398, the pannexin-blocking peptide 10panx, the scrambled control peptide 10panxScr, PD98059, PP2, SB203580, SP600125, SR11302, and SU1498 were purchased from Tocris (Ellisville, MO). Ac-YVAD-CMK, AG1478, and Stattic were from Enzo Life Science (Lausen, Switzerland). MCC950 was from InvivoGen (San Diego, CA). 4’,6-Diamidin-2-phenylindol (DAPI) was from Invitrogen (Paisley, UK). Human-specific small interfering RNA (siRNA) against nuclear factor of activated T cell 5 (NFAT5) and AQP8, respectively, and nontargeted control siRNA were obtained from Santa Cruz Biotechnology (Heidelberg, Germany). N-acetyl-L-cysteine, acetylsalicylic acid, AG1296, arachidonic acid, bis-(o-aminophenoxy)ethane-N,N,N',N'-tetra-acetic acid acetoxymethyl ester (BAPTA/AM), 4-bromophenacyl bromide, carbenoxolone, glibenclamide, glipizide, ibuprofen, indomethacin, PD150606, 1,10-phenanthroline, pinacidil, probenecide, prostaglandin E2, ruthenium red, SB431542, suramin, triamcinolone acetonide, VU0285655–1, VU0359595, and all other substances used were from Sigma-Aldrich (Taufkirchen, Germany), unless stated otherwise.
Preparation and culture of the RPE cells were described previously . Cell lines of passages 3–5 were used; each line was used in three to ten different experiments. Cells that achieved a confluency of about 90% were cultured in serum-free medium for 16 h. During this period, the cultures reached 100% confluency. Thereafter, test substances were added to the serum-free medium. Extracellular hyperosmolarity was produced with the addition of NaCl or sucrose to the medium. Extracellular hypo-osmolarity (60% of control osmolarity) was produced with the addition of distilled water. Hypoxia was induced with the addition of the hypoxia mimetic CoCl2 (150 µM)  or with cell culture in a 0.2% O2 atmosphere. Inhibitory agents were preincubated for 30 min.
Total RNA was extracted with the InviTrap Spin Universal RNA Mini Kit (Stratec Molecular, Berlin, Germany). The A260/A280 ratio of the optical density of the RNA samples (measured with NanoDrop1000; peQLab, Erlangen, Germany) was between 1.95 and 2.05, indicating adequate RNA quality. After treatment with DNase I (Roche, Mannheim, Germany), cDNA was synthesized from 0.5 µg RNA with a reverse transcription kit (ThermoFisher Scientific, Waltham, MA).
RT–PCR was performed with the Taq PCR Master Mix kit (Qiagen, Hilden, Germany) and the primer pairs described in Table 1. One microliter of the first-strand mixture and 0.25 µM of each gene-specific sense and antisense primers were used for amplification in a final volume of 20 µl. Amplification was performed for 40 cycles with the PTC-200 Thermal Cycler (MJ Research, Watertown, MA). Each cycle consisted of 30 s at 94 °C, 60 s at 58 °C, and 1 min at 72 °C.
Semiquantitative real-time RT–PCR analysis was performed with the CFX Connect Real-Time PCR Detection System (BioRad, Munich, Germany) and the primer pairs described in Table 1. The amplification reaction mixture (15 μl) consisted of 7.5 μl of 2×iQ SYBR Green Supermix (BioRad), the specific primer set (0.2 µM each), and 2 μl (2.5 ng for the AQP8 primer pair) and 1 µl (1.25 ng for the other primer pairs) of cDNA. The following conditions were used: one cycle of denaturation at 95 °C for 3 min, 45 cycles of denaturation at 95 °C for 30 s, annealing at 60 °C (AQP8) and at 58 °C for 20 s, extension at 72 °C for 45 s, and a melting curve at 55 °C with the temperature gradually increased for 0.5 °C up to 95 °C. The correct lengths of the PCR products were proved with agarose gel electrophoresis. mRNA expression levels were normalized to the level of β-actin (ACTB, Gene ID: 60; OMIM: 102630) mRNA. Changes in mRNA expression were calculated according to the 2-ΔΔCT method .
Cells were seeded at 7 × 104 cells per well in 12-well culture plates and were allowed to grow to a confluency of 60‒80%. Thereafter, the cells were transfected with NFAT5 siRNA and nontargeted siRNA (5 nM each), respectively, using HiPerfect reagent (Qiagen) in F-10 medium containing 10% fetal bovine serum (FBS; Invitrogen) according to the manufacturer's instructions. After 48 h (a time point at which the cultures were confluent), NaCl (100 mM) was added for a further 6 h. Total RNA was extracted, and the AQP8 mRNA level was determined with real-time RT–PCR. To knock down AQP8, the cells were cultured for 24 h (the time point of the strongest knockdown) in F-10 medium containing 10% FBS, transfection reagent, and AQP8 siRNA and nontargeted siRNA (10 nM each). Then, the cells were cultured in serum-free medium in the absence (6 and 24 h) and presence of high (+ 100 mM) NaCl (6 h) and CoCl2 (150 µM; 24 h), respectively, or in a 0.2% O2 atmosphere (24 h).
The cells were seeded at 5×105 cells per well in six-well plates and were cultured in FBS (10%)-containing F-10 medium. When a confluency of 80‒90% was achieved, the cells were growth arrested for 12 h in serum-free medium. After the medium was removed, the cells were washed twice with prechilled PBS (1X; 155 mM NaCl, 1.54 mM KH2PO4, 2.71 mM Na2HPO4-7H2O pH 7.2; Invitrogen) and scraped into 180 µl of lysis buffer (50 mM Tris-HCl, pH 8.0, 5 mM EDTA, 150 mM NaCl, 0.5% NP-40, 1% protease inhibitor cocktail, 1% phosphatase inhibitor cocktail). The cell lysates were centrifuged at 20,124 ×g for 10 min. Equal amounts of cytosolic protein (50 µg) were separated with 12.5% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE). Immunoreactive bands were probed with primary and secondary antibodies, and visualized using 5-bromo-4-chloro-3-indolyl phosphate/nitro blue tetrazolium. The antibodies used are described in Table 2.
The cultures were fixed with 1% paraformaldehyde for 20 min on ice. After several washing steps in prechilled PBS (pH 7.4; Invitrogen), the cultures were incubated in PBS containing 0.3% Triton X-100 for 15 min at room temperature. Blocking of nonspecific antibody binding was performed with PBS containing 10% normal goat serum and 0.3% Triton X-100 for 2 h at room temperature. Primary antibodies diluted in blocking solution were incubated at 4 °C overnight. After several washing steps in PBS plus 0.3% Triton X-100, the secondary antibody was applied for 2 h at room temperature. After additional washing steps, DAPI (1:10,000) was added for 15 min at room temperature. The coverslips were mounted with Fluorescence Mounting Medium (DakoCytomation, Glostrup, Denmark). Images were taken with an epifluorescence microscope (Olympus BX40, Olympus, Essex, UK), a charge-coupled device (CCD) camera (Olympus XM10), and cellSens software (Olympus). The antibodies used are described in Table 2.
A human donor eye was fixed in 4% paraformaldehyde at pH 7.4 for 4 days. The penetration of the fixative was improved with a small incision in the sclera of the eye. To prepare the paraffin sections, the eye was passed through an ascending series of alcohol, incubated in organic solutions, and embedded in Histowax (Histolab Products, Gothenburg, Sweden) according to a common protocol . Ten-micrometer-thick sections were cut with a rotation microtome (HM 355S; ThermoFisher Scientific).
After deparaffinization, the sections were pretreated with 0.01 M citrate buffer, pH 6.0, in a microwave oven for 2 × 20 min at 96 °C to achieve antigen retrieval. Nonspecific binding sites were blocked with PBS containing 5% normal goat serum plus 0.3% Triton X-100 for 2 h at room temperature. Primary antibodies diluted in blocking solution were incubated at 4 °C overnight. After washing in PBS plus 0.3% Triton X-100, secondary antibodies were applied for 2 h at room temperature. Following several washing steps with PBS, the cell nuclei were labeled with DAPI for 15 min at room temperature. The sections were mounted with Fluorescence Mounting Medium (DakoCytomation). The antibodies used are described in Table 2.
After transfection of AQP8 siRNA and nontargeted siRNA (10 nM each), respectively, the cells were cultured for 24 h in serum-free medium or in the cell proliferation assay, a medium that contained 0.5% serum, in the absence and presence of high (+ 100 mM) NaCl and CoCl2 (150 µM), respectively, or in a 0.2% O2 atmosphere. The level of VEGF-A165 in the cultured media (100 µl) was determined with enzyme-linked immunosorbent assay (ELISA; R&D Systems). Bromodeoxyuridine (BrdU) incorporation as a marker of the cell proliferation rate was determined with the Cell Proliferation ELISA BrdU Kit (Roche). BrdU (10 μM) was added to the culture medium 5 h before fixation. Cell viability was determined with a 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay (Serva, Heidelberg, Germany). Ten microliters of MTT solution (5 mg/ml) were added to each well after a 20-h period of stimulation. After 4 h, the culture supernatants were removed, and dimethyl sulfoxide (DMSO; 100 µl) was added. Absorbance was recorded at 570 nm with the Spectra Max 50 ELISA reader (Molecular Devices, Sunnyvale, CA).
At least three independent experiments with cell lines from different donors were performed for each test. Data are shown as means ± standard error of the mean (SEM). Statistical analysis was performed with Prism (GraphPad Software, San Diego, CA). Significant differences were evaluated with one-way analysis of variance (ANOVA), the Bonferroni multiple comparison post hoc test, and the Mann–Whitney U test. A p value of less than 0.05 was considered statistically significant.
As previously described [8-10], the cultured human RPE cells contained AQP8 gene transcripts (Figure 1A). Real-time RT–PCR analysis was performed to investigate whether alterations in the extracellular osmolarity induce changes in AQP8 gene expression in RPE cells. As shown in Figure 1B, expression of the AQP8 gene was statistically significantly (p<0.05) increased when the cells were cultured for 2 and 6 h in media that were made hyperosmotic by the addition of either 100 mM NaCl or 200 mM sucrose (both induce equal increases in extracellular osmolarity). Coaddition of NaCl and sucrose did not cause an additive increase in AQP8 gene expression after 2 h of stimulation (Figure 1B). There was a statistically significant difference (p<0.05) between the AQP8 expression levels induced by high NaCl and sucrose after 6 h of stimulation (Figure 1B). Although the hyperosmotic expression of AQP8 gene expression attenuated to the control level after 24 h of treatment with high NaCl or sucrose, coaddition of NaCl and sucrose induced a statistically significant (p<0.05) increase in AQP8 gene expression also after 24 h of treatment (Figure 1B). NaCl-induced expression of the AQP8 gene was dose-dependent; statistically significant (p<0.05) increases in the level of the AQP8 transcripts were found when more than 10 mM NaCl was added to the medium (Figure 1C). A hypo-osmotic medium (60% of control osmolarity) induced biphasic regulation of AQP8 gene expression, with downregulation and upregulation after 6 and 24 h of stimulation, respectively (Figure 1D).
Chemical hypoxia induced with the addition of CoCl2 (150 µM) to the culture medium  caused upregulation of AQP8 gene expression after 24 h of stimulation, while cell culture in a 0.2% O2 atmosphere caused time-dependent expression of the AQP8 gene (Figure 1E). Oxidative stress induced with the addition of H2O2 (20 µM) induced a transient increase in the level of the AQP8 transcripts after 6 h of stimulation (Figure 1E). In agreement with a previous study , cell culture in a high (25 mM) glucose medium was associated with a decrease in AQP8 gene expression after 2 h of stimulation (Figure 1F). Addition of the anti-inflammatory glucocorticoid triamcinolone acetonide to the culture medium induced a statistically significant (p<0.05) increase in the cellular level of the AQP8 transcripts after 6 and 24 h of stimulation, whereas addition of VEGF induced downregulation of AQP8 gene expression after 24 h of stimulation (Figure 1F). The following agents had no statistically significant (p>0.05) effects on AQP8 gene expression within 24 h of stimulation in iso-osmotic medium (data not shown): PDGF (10 ng/ml), TGF-β1 (10 ng/ml), IL-1β (10 ng/ml), prostaglandin E2 (10 ng/ml), and matrix metalloproteinase-2 (10 ng/ml). In agreement with a previous study , arachidonic acid (5 µM) and fetal calf serum (10%) also had no effects on AQP8 gene expression (data not shown). The data indicate that expression of the AQP8 gene in RPE cells is induced by extracellular hyperosmolarity, hypoxia, and oxidative stress.
It was shown in various cell types that the non-glycosylated type of AQP8 (with a molecular weight of about 27 kDa) is mainly expressed in the inner mitochondrial membrane, while glycosylated types of AQP8 (with molecular weights between 30 and 38 kDa) are predominantly localized to the plasma membrane [23,24,37]. To determine whether RPE cells contain non-glycosylated and/or glycosylated types of AQP8, we performed western blotting. As shown in Figure 2A, the AQP8 protein was found in different bands, e.g., around 27 and 37 kDa, suggesting that cultured RPE cells contain non-glycosylated and glycosylated types of AQP8. The different densities of the bands may suggest a higher expression level of AQP8 in mitochondria than in the plasma membrane.
We immunostained RPE cell cultures and an RPE monolayer that was freshly isolated from a donor eye with antibodies against AQP8, the mitochondrial marker cytochrome c, and the plasma membrane protein zonula occludens-1 (ZO-1; also known as tight junction protein-1). In the RPE cell cultures, AQP8 immunoreactivity partially colocalized with the cytochrome c immunoreactivity (Figure 2B). The RPE cell cultures were also labeled with the anti-ZO-1 antibody; in various areas, ZO-1 immunoreactivity was arranged in lines at the cell borders (Figure 2C). Cultures that were colabeled with the anti-AQP8 and anti-ZO-1 antibodies showed no colocalization of the two immunoreactivities (Figure 2D). Similarly, there was no colocalization of the two immunoreactivities in the RPE monolayer (Figure 2E). ZO-1 immunoreactivity was arranged in lines around the cells, while AQP8 immunoreactivity was found in spherical structures beside the nuclei within the cells, likely representing mitochondria and/or intracellular vesicles. However, there was colocalization of AQP8 and cytochrome c immunoreactivities in the freshly isolated RPE monolayer (Figure 2F). The data suggest that AQP8 in RPE cells is mainly located in the mitochondria.
It was shown that high extracellular NaCl induces activation of different signal transduction molecules in RPE cells, e.g., p38 mitogen-activated protein kinase (p38 MAPK) and extracellular signal-regulated kinases 1 and 2 (ERK1/2) . To investigate the intracellular signaling involved in mediating NaCl-induced expression of the AQP8 gene in RPE cells, we tested pharmacological blockers of various intracellular signal transduction molecules in cultures that were stimulated for 6 h with high (+ 100 mM) NaCl. As shown in Figure 3, the NaCl-induced expression of the AQP8 gene was statistically significantly (p<0.05) decreased in the presence of the following inhibitors: the inhibitor of p38 MAPK activation, SB203580, the inhibitor of phosphatidylinositol-3 kinase (PI3K)-related kinases, LY294002, the Janus kinase 2 (JAK2) inhibitor AG490, and the phospholipase A2 (PLA2) inhibitor 4-bromophenacyl bromide. The inhibitor of ERK1/2 activation, PD98059, the c-Jun NH2-terminal kinase (JNK) inhibitor SP600125, and the inhibitor of the serum and glucocorticoid-regulated kinase, GSK650394, had no effects on the NaCl-induced expression of the AQP8 gene (Figure 3). The following agents also had no statistically significant (p>0.05) effects on AQP8 gene expression within 6 h of stimulation with high NaCl (data not shown): the protein kinase A inhibitor H-89 (1 µM), the inhibitor of Src tyrosine kinases, PP2 (100 nM), the cyclooxygenase inhibitors indomethacin (10 µM), acetylsalicylic acid (2 mM), and ibuprofen (400 µM), the cyclooxygenase-2 inhibitor NS-398 (50 µM), triamcinolone acetonide (50 µM), the caspase-1 inhibitor Ac-YVAD-CMK (500 nM), the phospholipase D1 inhibitor VU0359595 (150 nM), and the phospholipase D2 inhibitor VU0285655–1 (500 nM). The data indicate that activation of p38 MAPK, PI3K, and JAK2 signal transduction pathways, as well as of PLA2, is involved in mediating the full expression of the AQP8 gene in RPE cells under high NaCl conditions.
It is unlikely that oxidative stress, e.g., induced by mitochondrial dysfunction or impaired AQP8-mediated H2O2 transport or both, plays a role in mediating NaCl-induced expression of the AQP8 gene in RPE cells, because the cell-permeable reducing agent dithiothreitol (300 µM), the inhibitor of reactive oxygen species, N-acetyl-L-cysteine (1 mM), and the inhibitor of the mitochondrial permeability transition pore, cyclosporin A (1 µM ), had no effects on AQP8 gene expression within 6 h of stimulation with high NaCl (data not shown). It is also unlikely that intracellular calcium signaling plays a significant role because the cell-permeable calcium chelator BAPTA/AM (5 µM), the phospholipase Cγ (PLCγ) inhibitor U73122 (Figure 3), the inhibitor of calcium-binding proteins, ruthenium red (30 µM), the inhibitor of protein kinase C (PKC) α/β, Gö6976 (Figure 3), and the calpain inhibitor, PD150606 (100 µM), had no effects on AQP8 gene expression within 6 h of stimulation with high NaCl (data not shown).
Extracellular hyperosmolarity was shown to induce the expression and activation of various transcription factors in RPE cells, including HIF-1α, nuclear factor (NF)-κB, activator protein-1 (AP-1), and NFAT5 [10,39,40]. Pharmacological blockers were used to investigate which transcription factors mediate NaCl-induced expression of the AQP8 gene in RPE cells. As shown in Figure 4A, AQP8 gene expression under unstimulated control conditions was statistically significantly (p<0.05) increased in the presence of the AP-1 inhibitor SR11302, suggesting that AP-1 activity exerts an inhibitory effect on constitutive expression of the AQP8 gene in RPE cells. NaCl-induced AQP8 gene expression was not altered in the presence of inhibitors of HIF-1 , signal transducer and activator of transcription 3 (STAT3) , NF-κB , and AP-1 (Figure 4A). However, the cAMP response element-binding protein (CREB) inhibitor 666–15 induced a statistically significant (p<0.05) decrease in NaCl-induced AQP8 gene expression in the RPE cells (Figure 4A).
In various cell systems, cellular survival under hyperosmotic conditions depends on the transcriptional activity of NFAT5 . It was shown that high extracellular NaCl induces expression of the NFAT5 (Gene ID: 10725; OMIM: 604708) gene and NFAT5 protein, and DNA binding of NFAT5, in RPE cells . To investigate whether NFAT5 activity is involved in mediating NaCl-induced expression of the AQP8 gene in RPE cells, NFAT5 was knocked down by transfection of the cells with NFAT5 siRNA. In accordance with a previous study , transfection of NFAT5 siRNA reduced the level of NFAT5 transcripts by 50‒60% in cells cultured under control and high-NaCl conditions; transfection with nontargeted siRNA had no effect (Figure 4B). However, transfection with NFAT5 siRNA or scrambled siRNA did not alter the cellular level of AQP8 mRNA under both conditions (Figure 4C). The data suggest that NaCl-induced expression of the AQP8 gene in RPE cells is (in part) mediated by the activity of CREB. However, it cannot be ruled out that the activities of additional transcription factors not investigated in the present study may contribute to NaCl-induced activation of the AQP8 gene.
Extracellular hyperosmolarity was shown to induce secretion of various growth and inflammatory factors from RPE cells, e.g., VEGF, basic fibroblast growth factor (bFGF), TGF-β1, and IL-1β [10,45]. To investigate whether autocrine/paracrine receptor signaling is required for NaCl-induced AQP8 gene expression in RPE cells, we tested pharmacological blockers of different receptor subtypes in cultures that were stimulated for 6 h with high (+ 100 mM) NaCl. As shown in Figure 5A, the following receptor blockers had no effects on control and NaCl-induced AQP8 gene expression: the inhibitor of the FGF receptor kinase, PD173074, the blocker of the VEGF receptor-2, SU1498, the inhibitor of the PDGF receptor tyrosine kinase, AG1296, the blocker of the epidermal growth factor (EGF) receptor tyrosine kinase, AG1478, the inhibitor of TGF-β1 superfamily activin receptor-like kinase receptors, SB431542, and the inhibitor of various G protein-coupled receptor subtypes, suramin. In addition, the broad-spectrum matrix metalloproteinase inhibitor 1,10-phenanthroline had no effect (Figure 5A); this suggests that shedding of growth factors from the extracellular matrix is not involved in mediating expression of the AQP8 gene. NaCl-induced expression of the AQP8 gene in RPE cells was also not dependent on the autocrine/paracrine signaling of purinergic receptors, because suramin also inhibits several subtypes of adenosine and purinergic metabotropic (P2Y) receptors , and because the following antagonists of purinergic receptors had no statistically significant (p>0.05) effects on AQP8 gene expression within 6 h of stimulation with high NaCl (data not shown): the P2Y1 receptor antagonist MRS2179 (30 µM), the P2Y2 receptor antagonist AR-C 118925XX (10 µM), the antagonist of ionotropic P2X7 receptors, A-438079 (50 nM), the adenosine A1 receptor antagonist DPCPX (50 nM), and the adenosine A2A receptor antagonist CSC (200 nM). However, a recombinant IL-1 receptor antagonist increased statistically significantly (p<0.05) NaCl-induced AQP8 gene expression (Figure 5A), suggesting that autocrine/paracrine IL-1 receptor signaling suppressed NaCl-induced expression of the AQP8 gene in RPE cells. To confirm the assumption that IL-1β-mediated signaling has an inhibitory effect on AQP8 gene expression, we tested exogenous IL-1β. As shown in Figure 5B, administration of IL-1β induced a statistically significant (p<0.05) decrease in NaCl-induced, but not of control, expression of the AQP8 gene. The effect of IL-1 β was apparently not dependent on inflammasome activation because the inhibitor of the nucleotide-binding oligomerization domain receptors-like receptor protein 3 (NLRP3) inflammasome, MCC950 , had no effect on NaCl-induced AQP8 gene expression, while the inhibitor increased the expression under control conditions (Figure 5A).
In addition to AQP-mediated water flux, channel-mediated ion currents contribute to the compensation of osmotic gradients across the plasma membrane under aniso-osmotic conditions . One subtype of potassium channels, KATP channels, which are heteromultimers of pore-forming Kir6.x subunits and the regulatory subunit sulfonylurea receptor (SUR) , was found to be expressed in RPE cells [50,51]. We tested two sulfonylureas, glibenclamide and glipizide, which induce, after binding to SUR, the closure of KATP channels and of ATP-sensitive nonselective cation channels NCCa-ATP [52,53]. As shown in Figure 6A, glibenclamide and glipizide caused statistically significant (p<0.05) decreases of NaCl-induced AQP8 gene expression in RPE cells. However, the competitive blocker of organic anion transporters, probenecid , had no effect (Figure 6A). Glibenclamide was also shown to inhibit activation of the NLRP3 inflammasome . However, because the NLRP3 inflammasome inhibitor MCC950 had no effect on NaCl-induced AQP8 gene expression (Figure 5A), it is unlikely that the effects of sulfonylureas on AQP8 gene expression are mediated by inhibition of the NLRP3 inflammasome. To further investigate whether the activity of KATP channels influences expression of the AQP8 gene, we tested the channel opener pinacidil, which is an activator of SUR2A and SUR2B, but not SUR1 . As shown in Figure 6B, pinacidil induced an increase in the expression level of the AQP8 gene after 24 h of stimulation. This suggests that activation of KATP channels stimulates AQP8 gene expression.
In addition to sulfonylureas, carbenoxolone decreased NaCl-induced AQP8 gene expression in RPE cells statistically significantly (p<0.05; Figure 6A). Carbenoxolone is an inhibitor of the steroid 11β-hydroxylase and of pannexin hemichannels [57,58]. There are two steroid 11β-hydroxylase isozymes encoded by the CYP11B1 (Gene ID: 1584; OMIM: 610613) and CYP11B2 (Gene ID: 1585; OMIM: 124080) genes; however, transcripts of both genes were not detected in the RNA extracted from the cells used (data not shown). Therefore, the effect of carbenoxolone was likely mediated by blocking pannexins. To prove this assumption, we tested the pannexin-blocking peptide 10panx. As shown in Figure 6A, the peptide decreased statistically significantly (p<0.05) NaCl-induced AQP8 gene expression in RPE cells, whereas a scrambled control peptide had no effect (Figure 6A).
The data suggest that activation of KATP channels is involved in mediating hyperosmotic expression of the AQP8 gene in RPE cells. Coexpression of the KATP channel subunits Kir6.2 and SUR1 was described in the rat RPE . We investigated whether human RPE cells express KATP channel subunits. As shown in Figure 7A,B, acutely isolated and cultured human RPE cells contained Kir6.1 but not Kir6.2 transcripts. The expression level of the Kir6.1 gene was not different between acutely isolated and cultured cells (Figure 7B). Cultured cells also contained SUR2B (Gene ID: 10060; OMIM: 601439), but not SUR1 (Gene ID: 6833; OMIM: 600509) and SUR2A (Gene ID: 10060; OMIM: 601439), transcripts (Figure 7A). The expression levels of the Kir6.1 and SUR2B genes were not altered under hyperosmotic compared to control conditions (Figure 7C). Cultured RPE cells were labeled with an antibody against Kir6.1 (Figure 7D).
To investigate whether AQP8 modulates the proliferation and viability of RPE cells, we knocked down AQP8 with siRNA. As shown in Figure 8A, transfection with AQP8 siRNA reduced the level of AQP8 transcripts by 60‒75% in cells cultured under control conditions; transfection with scrambled nontargeted siRNA had no effect. Transfection with AQP8 siRNA also reduced statistically significantly (p<0.05) the level of AQP8 mRNA in cells cultured under high NaCl conditions and in the presence of CoCl2, respectively, when compared to the respective controls and the effects of nontargeted siRNA (Figure 8B).
It was described that high extracellular osmolarity induces cell cycle arrest in various cell systems, including RPE cells [59,60]. High extracellular NaCl and the addition of the hypoxia mimetic CoCl2 to the culture medium induced decreases in the RPE cell proliferation rate (data not shown), as previously described . Transfection of AQP8 siRNA had no effects on the proliferation rate under control, high (+ 100 mM) NaCl, and CoCl2 (150 µM)-stimulated conditions (data not shown), suggesting that the activity of AQP8 does not influence the proliferation of RPE cells.
It was shown in various cell types that AQP8-mediated water and H2O2 transport contributes to the regulation of cell survival under stress conditions . In the cultured RPE cells, high extracellular NaCl caused a slight increase in cell viability, while cell culture in a CoCl2-containing medium was associated with a decrease in viability (Figure 8C). Cells transfected with AQP8 siRNA displayed statistically significantly (p<0.05) lower viability than non-transfected cells and cells transfected with nontargeted siRNA; this was found under all conditions tested (Figure 8C).
VEGF is a factor secreted by RPE cells implicated in the pathogenesis of different blinding retinal diseases, such as retinal edema and neovascular AMD . To investigate whether AQP8 modulates the expression and secretion of VEGF from RPE cells, we knocked down AQP8 with siRNA. Transfection of RPE cells with AQP8 siRNA had no statistically significant (p>0.05) effects on the cellular level of the VEGFA mRNA under unstimulated control and high NaCl conditions (Figure 9A). However, in CoCl2-induced chemical hypoxia, cells that were transfected with AQP8 siRNA displayed a statistically significantly (p<0.05) higher expression level of VEGFA mRNA compared to non-transfected cells and cells that had been transfected with nontargeted siRNA (Figure 9A). NaCl-induced extracellular hyperosmolarity and hypoxia caused the secretion of VEGF from the RPE cells, as indicated by the elevated levels of the VEGF-A protein in the cultured media compared to the non-stimulated control (Figure 9B). Transfection of AQP8 siRNA did not statistically significantly (p>0.05) alter the VEGF-A content of the cultured media under unstimulated control conditions and in the presence of high NaCl (Figure 9B). However, when the cells were cultured in the presence of CoCl2 or in a 0.2% O2 atmosphere, the media of the cells transfected with AQP8 siRNA contained statistically significantly higher (p<0.05) levels of VEGF-A protein compared to the media of non-transfected cells and of cells that had been transfected with nontargeted siRNA (Figure 9B). The data show that the knockdown of AQP8 results in stimulation of hypoxic expression and secretion of VEGF from RPE cells.
One major function of the RPE is the transport of excess subretinal water and potassium into the choroidal blood . In many cell types, bidirectional transport of water across membranes is facilitated by AQP water channels . In addition to transcripts of other AQP subtypes, human RPE cells contain AQP8 gene transcripts [8,10]. In the present study, we investigated the subcellular expression of AQP8 and the cellular mechanisms of hyperosmotic AQP8 gene expression in cultured human RPE cells. We found that expression of the AQP8 gene in RPE cells is dependent on extracellular osmolarity; it is transiently increased under hyperosmotic conditions (Figure 1B) and shows biphasic regulation under hypo-osmotic conditions (Figure 1D). AQP8 gene expression is also increased under hypoxic and oxidative stress conditions (Figure 1E), and in the presence of the anti-inflammatory glucocorticoid triamcinolone acetonide (Figure 1F). Western blotting and immunocytochemical data suggest that the AQP8 protein is mainly located in mitochondria (Figure 2A,B,F). However, because we found different bands of AQP8 immunoreactivity in western blots of RPE cell lysates (Figure 2A), it cannot be ruled out that a minor portion of AQP8 is also present in the plasma membrane.
We found that coaddition of NaCl and sucrose did not cause an additive increase in AQP8 gene expression after 2 h of stimulation (Figure 1B). The absence of an additive effect might be explained by the assumption that extracellular hyperosmolarity, but not the alteration of the transmembrane NaCl gradient, triggered expression of the AQP8 gene at this time point. The statistically significant difference (p<0.05) between the AQP8 expression levels induced by high NaCl and sucrose after 6 h of stimulation (Figure 1B) might be explained by the assumption that, at this time point, extracellular hyperosmolarity and the alteration in the transmembrane NaCl gradient induced AQP8 gene expression.
We found that activation of the p38 MAPK and PI3K signal transduction pathways contributes to the full NaCl-induced expression of the AQP8 gene (Figure 3). Interestingly, it was described that AQP8 is capable of increasing H2O2-induced phosphorylation of p38 MAPK and PI3K . We also found that inhibition of PLA2 decreases osmotic expression of the AQP8 gene in RPE cells (Figure 3). Osmotic challenge is a known activator of PLA2 . Activation of PLA2 is a critical factor that mediates AQP3  and AQP8 gene expression in RPE cells (Figure 3). However, AQP3 gene expression was shown to be induced by arachidonic acid and prostaglandin E2 , while AQP8 gene expression is not (data not shown). We also found that autocrine/paracrine IL-1 receptor signaling induced by high NaCl depresses hyperosmotic expression of the AQP8 gene in RPE cells (Figure 5A, B). Most of the blocking agents used in the present study did not inhibit completely hyperosmotic induction of the AQP8 gene. This may suggest that multiple signal transduction pathways contribute to osmotic induction of the AQP8 gene.
The high NaCl-induced expression of the AQP8 gene in RPE cells was decreased by the sulfonylureas glibenclamide and glipizide (Figure 6A). This suggests that activation of KATP and NCCa-ATP channels plays a role in mediating the effect of high NaCl on expression of the AQP8 gene. The KATP channel opener pinacidil induced a time-dependent increase in the expression level of the AQP8 gene (Figure 6B). This could be explained by the fact that increased transmembrane ion currents must be, for osmotic reasons, accompanied by facilitated water transport. We found expression of Kir6.1 and SUR2B in human RPE cells (Figure 7A,B,D). However, it cannot be ruled out that additional subtypes of potassium channels that are associated with SUR2B may contribute to the effects of the sulfonylureas on NaCl-induced expression of the AQP8 gene. The incomplete action of the KATP channel blockers (Figure 6A) might be explained by the assumption that currents through other subtypes of potassium channels compensate, in part, the blocked currents through KATP channels, or that AQP8-mediated water transport also supports the potassium currents through other subtypes of potassium channels, or both.
KATP and NCCa-ATP channels are activated by depletion of intracellular ATP . Closure of the KATP channels results in a reduction in the potassium efflux from the cells, membrane depolarization, and activation of voltage-gated calcium channels . In reactive astrocytes, opening of NCCa-ATP channels causes complete plasma membrane depolarization and cell blebbing; both are characteristics of cytotoxic edema . Cell depolarization and blebbing induced by depletion of intracellular ATP are prevented by glibenclamide, consistent with a role for NCCa-ATP channels in mediating cytotoxic edema . KATP channels are known to play an important role in the ischemic preconditioning of the retina . Generally, mitochondria and their KATP channels are crucially involved in mediating ischemic preconditioning in different tissues [67,68]. However, it is unlikely that mitochondrial dysfunction is involved in mediating NaCl-induced expression of the AQP8 gene in RPE cells because the inhibitor of mitochondrial permeability transition, cyclosporin A, had no effect (data not shown). In addition, AQP8-mediated H2O2 transport seems to be less relevant for high NaCl-induced induction of the AQP8 gene in RPE cells, because reducing agents and inhibitors of reactive oxygen species had no effects (data not shown). It is suggested that NaCl-induced induction of the AQP8 gene in RPE cells facilitates transmembrane water transport which supports ion currents, e.g., potassium currents through KATP channels, that mediate the rapid compensation of the ionic and osmotic gradients across the plasma membrane under high NaCl conditions.
It was described that high extracellular NaCl induces pannexin-mediated release of ATP from RPE cells . We found that inhibition of pannexins decreases NaCl-induced expression of the AQP8 gene (Figure 6A), suggesting that activation of pannexins is required for full AQP8 gene expression in RPE cells. However, AQP8 gene expression was not altered in the presence of inhibitors of various subtypes of purinergic receptors (data not shown). In cultured astrocytes, activation of pannexin-1 was shown to be involved in mediating the expression of AQP4 and the release of IL-1β under conditions of experimental ischemia . The functional role of pannexin activation in mediating high NaCl-induced AQP8 gene expression in RPE cells remains to be investigated.
In different cell systems, AQP8 activity either amplifies cell proliferation, e.g., via facilitating H2O2-mediated tyrosine kinase signaling upon activation of growth factor receptors , or inhibits cell proliferation, via inactivation of PI3K-mediated signaling . In glioma cells, knockdown of AQP8 induces cell cycle arrest, via downregulation of cell cycle proteins . We found that transfection of AQP8 siRNA had no effects on the proliferation rate of RPE cells (data not shown).
Different effects of AQP8 knockdown on cell viability were described in various cell systems. In hepatoma cells, knockdown of AQP8 causes mitochondrial depolarization and necrotic cell death which was attributed to the reduced H2O2 extrusion from the mitochondria resulting in an increased mitochondrial reactive oxygen species level . In 3T3-L1 preadipocytes, which express AQP8 predominantly in mitochondria, knockdown of AQP8 results in mitochondrial swelling, but has no effect on cell viability . Because cellular shrinkage mediated by ion and water efflux is a crucial step in apoptotic cell death , knockdown of AQPs may increase resistance to apoptosis . We found that knockdown of AQP8 reduces RPE cell viability under control, hyperosmotic, and hypoxic conditions (Figure 8C). This result suggests that water or H2O2 transport or both through AQP8 is required for the full viability of RPE cells under various conditions. It was shown that mitochondrial oxidative stress in the RPE leads to localized retinal degeneration , and that high salt loading in rats causes intracellular edema and mitochondrial swelling in retinal ganglion and glial cells . It could be that knockdown of AQP8 impairs water and H2O2 transport through mitochondrial membranes which results in a decrease in the viability of RPE cells. The relatively small effects of AQP8 knockdown on cell viability (10‒20% decrease) are statistically significant and might be explained by the fact that the siRNA used caused a decrease in the AQP8 transcript level by only 60‒75% (Figure 8A, B) and with the assumption that a part of the lost AQP8-mediated transmembrane water transport may be compensated by other subtypes of aquaporins. The decrease in ATP production induced by AQP8 knockdown may activate KATP channels. However, we found that the activity of KATP channels is implicated in mediating AQP8 gene expression under hyperosmotic, but not under control, conditions (Figure 6A). Further research is necessary to determine the mechanisms of the involvement of AQP8 in the maintenance of RPE cell viability. Triamcinolone acetonide, an anti-inflammatory steroid clinically used for rapid resolution of retinal edema , induced expression of the AQP8 gene under control conditions (Figure 1F), but not under hyperosmotic conditions (data not shown). An improvement in AQP8-mediated transmembrane transport of water and H2O2 by triamcinolone acetonide may contribute to the edema-resolving effect of this steroid, and may protect mitochondrial function and integrity in RPE cells.
VEGF is the most relevant hypoxia-induced angiogenic factor which promotes retinal edema and choroidal neovascularization . VEGF expression and secretion are also induced by other pathogenic conditions, e.g., extracellular hyperosmolarity and H2O2-induced oxidative stress [10,77]. We found that knockdown of AQP8 stimulated expression and secretion of VEGF under hypoxic conditions and had no effects under control and high NaCl conditions (Figure 9A,B). The effect of AQP8 knockdown specifically under hypoxic conditions might be explained by the fact that hypoxia affects the mitochondrial oxidative stress level. It was shown that administration of a mitochondria-targeted antioxidant decreases the retinal VEGF mRNA and protein levels in OXYS rats, an animal model of AMD . Mitochondrial complex III senses hypoxic conditions and generates reactive oxygen species which stabilize HIF-1α protein that induces transcription of the VEGFA (Gene ID: 7422; OMIM: 192240) gene . Inhibition of AQP8-mediated H2O2 transport by gene knockdown may exacerbate these conditions, resulting in increased expression of the VEGFA gene. VEGF is known to promote mitochondrial function and biogenesis [80,81]. It could be that mitochondrial dysfunction induced by coadministration of CoCl2 and AQP8 siRNA induces expression and release of VEGF to stimulate mitochondrial function and biogenesis in an autocrine/paracrine manner.
We found statistically significant effects of high extracellular NaCl on expression of the AQP8 gene when more than 10 mM NaCl were added to the culture medium (Figure 1C). It is generally accepted that the highest pathological blood osmolality in human subjects is around 360 mOsm/kg which can be achieved with the addition of 40 mM NaCl to the culture medium [82,83]. However, less well appreciated is that the local extracellular NaCl concentration in the interstitium may be considerably higher (160‒250 mM) than the plasma concentration of NaCl (about 140 mM) [84,85]. Therefore, the present results may have relevance for in vivo conditions.
RPE cells express different AQP subtypes, including AQP1, AQP3, AQP5, and AQP8 [5,8-10]. The AQP subtypes may have different functional roles in mediating water transport through RPE cells. It is conceivable that AQP1, which is not regulated by extracellular osmolarity , facilitates constitutive water transport through RPE cells. AQP3 and AQP5 may facilitate osmotic water transport under partly different conditions and with partly different functions; AQP3 is upregulated under inflammatory conditions  and may mediate glycerol transport. AQP8 may facilitate mitochondrial water and H2O2 transport. Further research is required to determine the functional roles of the different AQP subtypes in RPE cells.
The authors thank Ute Weinbrecht for excellent technical assistance.