Aquaporin-1 Channels in Human Retinal Pigment Epithelium: Role in Transepithelial Water Movement

Abstract

purpose. Aquaporin (AQP) is a hexahelical integral membrane protein that functions as a constitutive channel for water and regulated channel for cations in fluid transporting tissues, including many in the eye. Although AQP1 has been cloned from a cDNA library prepared from cultures of retinal pigment epithelial (RPE) cells isolated from human fetal tissue, three separate studies failed with various immunochemical techniques to detect AQP1 protein in adult human or rat RPE preparations. The purpose of this study was to examine specifically the expression and distribution of AQP1 in adult human RPE in situ by using alternative methodologies and model systems and to determine the contribution of AQP1 to water movement across cultured RPE cells isolated from human cadaveric and fetal eyes.

methods. AQP1 in human RPE in situ was determined after biotinylation of proteins on cell surfaces and streptavidin chromatography, followed by immunoblot analyses. AQP1 distribution in a polarized in vitro RPE model was determined with indirect immunofluorescence confocal microscopy. The role of channel-mediated transport of water across RPE cell monolayers on filters was assessed by osmotic challenge assay. Expression levels of AQP1 were controlled with an adenovirus expression system and monitored by immunoblot analyses.

results. AQP1 protein was detected in human RPE in situ and in cultures of human adult and fetal RPE cells. In functional assays, AQP1 facilitated water movement across RPE monolayers in an expression-dependent manner in two complementary model systems.

conclusion. The expression of AQP1 by RPE in vivo probably contributes to the efficient transepithelial water transport across RPE, maintains retinal attachment, and prevents subretinal edema.

Tight junctions between the endothelial cells of retinal capillaries and cells of the retinal pigment epithelium (RPE) form the blood–retinal barrier (BRB). Similar to the blood–brain barrier, the BRB prevents leakage of fluid, protein, and potentially harmful agents from interacting with delicate neuronal tissues. While nutrients and metabolites are actively transported into the eye to support neuronal tissues, both passive and active forces move water in the opposite direction, out of the eye, and facilitate retinal attachment. For example, intraocular pressure continually forces water into the retina while choroidal osmotic pressure draws water across the RPE and toward the choroid. ¹ These two passive forces help maintain retinal attachment and remove water from the subretinal space. In conjunction with these passive forces, active transport of solute across the RPE contributes to the movement of water out of the eye. Thus, the net balance of solute flux across the RPE is in the apical (retinal) to basolateral (choroidal) direction and is a function of polarized expression and activity of an array of channels and transporters. ^{2 3 4 5}  

The net apical to basolateral movement of solute and water appears counterintuitive when considering the unusual localization of the Na/K adenosine triphosphatase (ATPase) to the apical domain of RPE. ^{2 6} The apical distribution of Na/K ATPase resembles the choroid plexus, but is unlike most other simple epithelia. To compensate for excess sodium on the apical surface, key transporters and channels are located basolaterally, moving chloride and bicarbonate into the choroid. ² Therefore, efficient movement of solute and water define healthy RPE and, because water movement across RPE is limited transcellularly by tight circumferential complexes of intercellular junction proteins, we hypothesized that the channel protein aquaporin (AQP)-1 facilitates the movement of water across RPE cells. 

Aquaporin-1 is an integral membrane protein at the plasma membrane having six transmembrane domains that function as a constitutive channel for water and regulated channel for cations. ^{7 8 9 10 11} The expression of AQP1 is generally limited to fluid secreting and absorbing tissues in the human body that demonstrate an enhanced permeability to water in comparison with other tissues. ¹² For example, AQP1 protein is present in cells of the adult human eye that require enhanced permeability to water, including corneal endothelium, ciliary epithelium, lens epithelium and cells of the conventional outflow pathway. ^{13 14 15} In previous studies, AQP1 protein was not detected in adult rat or human RPE preparations, possibly because of technical difficulties associated with the presence of melanin and lipofuscin in RPE cells. ^{12 14 15} AQP1 messenger RNA encoding AQP1 was detected, however, in cultured fetal RPE cells. ¹⁶ Because of the complex expression pattern for AQP1 during development of many tissues including in the eye, ¹⁷ it is not clear whether the AQP1 protein expression or cellular localization is maintained after birth in RPE. Thus, the purpose of the present study was to determine whether AQP1 protein is expressed by fetal and adult human RPE and whether AQP1 facilitates movement of water across RPE cells in culture. 

Methods

Human Adult RPE Cell Cultures

Human donor eyes were obtained from the North Carolina Eye Bank. RPE cells were isolated and cultured as described previously. ¹⁸ Briefly, the anterior segment was removed from whole eyes by dissection. The neurosensory retina was gently peeled away from the underlying RPE. The RPE was rinsed with versine (EDTA-saline) and the eyecups were filled with 0.3% trypsin in versine to remove the RPE from Bruch’s membrane. Enzymatically harvested RPE were collected in 10 mL of DMEM containing 10% FBS to inactivate the trypsin, and the cells were pelleted by centrifugation (800g for 10 minutes). Cells were resuspended in fresh medium, plated into standard cell culture flasks, and maintained at 37°C and 5% CO₂ in humidified air. Successful cultures were passaged once weekly, and the medium was changed three times per week. The cell strains used in the present study were isolated from cadaveric eyes aged 62, 50, 63, and 13 years at the time of death. 

Human Fetal RPE Cell Cultures

The culture method for human RPE cells obtained from aborted fetuses was as reported previously. ^{19 20 21} The tenets of the Declaration of Helsinki were observed, and the patients (or their guardians) gave consent for donation of the tissue. Institutional Human Experimentation Committee approval was obtained for the use of human eyes. 

For the current set of experiments, aliquots of cryopreserved cells were thawed and cultured in normal Ca²⁺ medium with 1% heat-inactivated calf serum (JRH Bioscience, Lenexa, KS) in culture wells (Millicell-PCF; Millipore, Bedford, MA) coated with mouse laminin (Collaborative Research, Bedford, MA). 

Adenovirus Vectors

The adenovirus (AV) backbone for the β-galactosidase (βgal) and AQP1 sense and antisense AV constructs was a replication-deficient first-generation AV with deletions of the E1 and E3 genes. ²² This empty AV contains the cytomegalovirus (CMV) promoter and bovine growth hormone polyadenylation (bHG) site separated by a polylinker that was used to clone AQP1 DNA, as described previously. ²³ Briefly, a recombinant AQP1 virus was constructed with a plasmid containing the coding sequence for AQP1, pCHIPev. ²⁴ pCHIPev was digested, and the AQP1 insert was subcloned into the shuttle vector pSKAC in either the sense (creating pSKAC/AQP1-S) or antisense (creating pSKAC/AQP1-AS) orientation. pSKAC contains map units 0.0 to 1.3 of the AV, which includes the left terminal repeat of AV, a CMV promoter, an alfalfa mosaic virus (AMV) translation enhancer, and a polylinker region. DNA fragments containing AQP1 DNA were obtained from pSKAC after restriction and ligated into the AV, as described previously. ²³ Human embryonic kidney cells (293 cells) were transfected with ligation mixture, and individual viruses were isolated from cell lysates by two consecutive rounds of plaque purification, using an agar overlay, as described previously. ²³ Individual viruses were amplified in 293 cells and purified over a cesium step gradient. Individual AV DNA titers were determined by three different methods: (1) plaque titration on 293 cells; (2) immunofluorescence microscopy of AV protein expression (anti-penton group antigen, clone 143; Biodesign, Kennebunk, ME) in 293 cells infected with serial dilutions of AV and (3) absorbance at 260 nm (plaque forming units [pfu]/mL = A₂₆₀ × dilution × 10¹⁰). The βgal AV was a gift from Karsten Peppel (Duke University). 

Cell Surface Biotinylation

Whole eye globes were obtained from the North Carolina Eye Bank less than 24 hours after death. Globes were bisected at the equator and the vitreous was decanted from the posterior pole. The neurosensory retina was removed gently to maintain the integrity of the RPE monolayer. RPE in the eyecup was gently washed with phosphate-buffered saline (PBS) to remove any remaining retinal cells or debris. Eyecups were rinsed twice with ice-cold biotinylation buffer (100 mM NaCl, 50 mM NaHCO₃, and 2 mM CaCl₂ [pH 8.0]) and incubated with freshly prepared NHS LC-biotin ([succinimidyl 6-(biotinamido)hexanoate] 1 mg/mL; Pierce, Rockford, IL) in biotinylation buffer at 4°C. In some experiments, 5 mM Na₂ EDTA was included in place of CaCl₂ in biotinylation buffer, as a control. After 1 hour, eyecups were rinsed with PBS at 22°C and incubated in quenching buffer (100 mM glycine, 25 mM Tris/HCl [pH 7.4]) for 15 minutes. Eyecups were again rinsed with PBS and RPE cells were gently scraped into ice cold hypotonic lysis buffer (5 mM N-ethyl maliemide, 10 mM EDTA, 10 μg/mL leupeptin, 1 mM benzamidine, and 1 mM phenylmethylsulfonyl fluoride). Cells scraped into the hypotonic lysis buffer were homogenized by 25 strokes in a homogenizer (Dounce; Bellco Glass Co., Vineland, NJ) then the nuclei were pelleted by centrifugation (2000g for 30 seconds). RPE homogenates were centrifuged at 20,000g, and pellets were resuspended in ice-cold PBS containing 1% NP-40 and homogenized (10 strokes) on ice. Homogenates were centrifuged at 14,000g, and the supernatant was incubated with streptavidin-agarose beads (Pierce) at 4°C with rotation. After 14 hours, beads were washed with 30 bed volumes of ice-cold PBS containing 0.1% NP-40. Proteins specifically bound to the beads were extracted by boiling in buffer containing 2% SDS. 

Confocal Microscopy

RPE cells were grown on filter supports for 2 months before fixation and immunocytochemistry. The filters with their attached cells were excised from their wells, washed three times with PBS (120 mM NaCl, 2.7 mM KCl, and 10 mM NaPO₄, [pH 7.4]; Sigma, St. Louis, MO), and fixed for 30 minutes with PBS-buffered 4% formaldehyde (Ted Pella Inc., Irvine, CA). The cells were washed twice with PBS and permeabilized with 0.1% Triton X-100/PBS (Roche Molecular Biochemicals, Mannheim, Germany) for 2 minutes and were then blocked for 1 hour with 1% BSA-PBS (Intergen, Purchase, NY) and 45 μL/mL goat serum (Sigma) before a 2-hour incubation with gentle shaking at 37°C with affinity-purified rabbit polyclonal antibodies against AQP1 (1:100 dilution of 400 μg/mL) in 0.1% blocking solution and PBS. ¹⁴ After the cells were washed three times for 10 minutes each on a shaker, they were incubated for 1 hour with FITC-conjugated goat anti-rabbit IgG (Molecular Probes, Eugene, OR) at a concentration of 1:200 in 0.1% blocking solution and PBS and washed thoroughly, as described earlier. All incubations and washes were conducted at room temperature, unless otherwise indicated. The treated cells on their filters were placed on glass slides, covered with mounting medium (5% n-propyl gallate in 100% glycerol) and coverslipped for viewing. 

The wholemount monolayers were examined with a laser confocal microscope (LSM-210; Carl Zeiss, Inc., Thornwood, NY). A 488-nm beam was used to excite the FITC-label. Confocal optical sections of 1 μm were taken in the z-axis. Cross-sections were obtained from the z-axis sections using the φ-z mode. 

Immunoblot Analyses

Sodium dodecyl sulfate (SDS)–solubilized whole cell lysates, cell fractions, or proteins isolated by cell surface biotinylation containing 5% β-mercaptoethanol were electrophoresed into 12% polyacrylamide gels containing 0.1% SDS. Fractionated proteins were blotted onto nitrocellulose with a commercial system (Transblot; Bio-Rad, Hercules, CA), according to the manufacturer’s instructions. The blots were preincubated for 30 minutes at 22°C in Tris-buffered saline containing 5% nonfat powdered milk and 0.2% Tween-20 (TBS-T) and were then probed with affinity purified anti-AQP1 IgG (1:7500), ¹⁴ pan anti-Na/K ATPase (1:10,000 dilution; gift from Ronald Lynch, The University of Arizona, Tucson, AZ) or anti-β-actin IgG (AC-15, 1:10,000) for 2 hours at 22°C. The blots were washed (three times for 15 minutes each) in TBS-T and were incubated for 2 hours with horseradish peroxidase–conjugated secondary antibodies (goat anti-rabbit or goat anti-mouse, 1:10,000; Pierce). Alternatively, blots were incubated with horseradish-peroxidase conjugated streptavidin (1:10,000). The blots were washed (three times for 15 minutes each) in TBS-T and specific labeling was visualized by enhanced chemiluminescence (ECL; Pierce) and exposure to ECL autoradiographic form (Hyperfilm; Amersham, Arlington Heights, IL). Immunoblots were digitized using the gel documentation system (UVP, Inc., Upland, CA), and densitometry was performed on computer (LabWorks software; UVP Inc., Upland, CA). 

Osmotic Challenge Assays

Water permeability was measured as the net fluid movement driven by an osmotic gradient across intact monolayers of RPE or AV-infected RPE monolayers, by methods similar to those reported previously. ^{23 25 26} Cells were seeded onto filters (Transwell; Corning Costar, Corning, NY, or Millicell PCF; Millipore, Bedford, MA; 1 cm², 0.4 μm pore size) at a density of 1.5 × 10⁵ cells per well. Cells were maintained in humidified air containing 5% CO₂ for up to 17 weeks to allow for cell–cell junctions to mature. Transepithelial electrical resistance of differentiated fetal RPE monolayers ranged from 554 to 636 Ω cm² before cells were infected at the apical surface ²⁷ with AV containing AQP1 cDNA in the sense or antisense orientation, βgal cDNA, or an empty AV at a multiplicity of infection (MOI) of between 0.1 and 10. Five days after infection, the medium was removed completely from both upper (apical) and lower (basolateral) chambers. One milliliter of fresh prewarmed medium was added to the lower chamber, and exactly 175 μL (100 μL for Millicell wells) of medium was added to the upper chamber at time 0. Medium added to basolateral chamber consisted of DMEM containing high glucose (isosmotic, ∼300 mOsM), whereas medium added to the apical chamber consisted of DMEM containing high glucose and supplemented with 60 mM sodium chloride or 100 mM sucrose (hyperosmotic, ∼450 mOsM). After incubation at 37°C, the entire medium from each of the upper chambers was removed carefully and the volume was measured using an analytical balance. 

Statistical Analyses

Differences between experimental categories were considered significant at a level of P < 0.05 using a Student’s t-test assuming unequal variance. 

Results

To avoid technical problems associated with imaging RPE in fresh-frozen and paraffin-fixed eyes, we used two complementary techniques to examine AQP1 expression and distribution in fetal and adult RPE. To examine AQP1 in fetal RPE, we used indirect immunofluorescence in conjunction with confocal microscopy of differentiated fetal RPE cells grown on permeable supports. Because there is currently no differentiated culture model of adult RPE, we used streptavidin chromatography to capture biotinylated cell surface proteins in situ. The presence of AQP1 among biotinylated proteins was visualized in immunoblot and streptavidin-horseradish peroxidase analyses. 

Localization of AQP1 in Fetal RPE

Figure 1 shows sequential confocal images of fetal RPE cells on a permeable support that have been labeled with affinity-purified anti-AQP1 IgG. Figure 1A shows an image taken near the apical surface, and Figures 2B 2C 2D are selected sequential 1-μm optical sections taken while moving toward the basal surface. Reconstruction of confocal images enabled the visualization of cell monolayers in cross-section. Figure 1E shows a representative cross-sectional view of AQP1 labeling in a fetal RPE monolayer. AQP1 labeling is clear in the apical cellular domain and possible in the lateral domain. Specificity of labeling to AQP1 was confirmed in parallel preparations of cells grown on permeable supports by Western blot (Fig. 1F) . 

Localization of AQP1 in Adult RPE

AQP1 was also detected in adult human RPE in situ. Figure 2 , lane 1, shows the presence of AQP1 (top) compared with Na/K ATPase (middle) in whole-cell lysates of RPE prepared from a human cadaveric eye. This type and level of expression was detected in samples from all 12 human eyes examined. The typical expression pattern for AQP1, namely both a 28-kDa unglycosylated and 40- to 50-kDa glycosylated form, ⁸ was observed in all samples. 

Cell surface proteins of human RPE were biotinylated in situ. During the preparation of eyecups, extreme care was taken to assure that the integrity of RPE monolayers was not compromised. In addition, calcium was included in the biotinylation reaction in an attempt to maintain intercellular junctional complexes. Figure 2 , lane 2, shows the amount of AQP1 (top), α₁-subunit of Na/K ATPase (middle) and total biotinylated proteins (bottom) remaining in lysate after exposure to streptavidin agarose. No proteins were observed after streptavidin agarose was washed extensively (Fig. 2 , lane 3). The biotinylated proteins at the surface of human RPE bound to streptavidin beads are shown in Figure 2 , lane 4. Because biotinylation and capture efficiencies varied between experiments, only qualitative estimates of total protein that was biotinylated at the apical plasma membrane for Na/K ATPase and AQP1 were obtained from three independent experiments. These results, taken together, indicate that a minority of AQP1 total protein (<25%) was biotinylated, whereas a majority (>60%) of Na/K ATPase was biotinylated. 

AQP1-Mediated Permeability of RPE

To study functionally the role of AQP1 in the movement of water across RPE monolayers we used two complementary models: RPE cells were isolated from human cadaveric or fetal eyes and were cultured and seeded onto porous filters. Expression of AQP1 was modulated in stable and confluent cell monolayers by infection with AV that contains AQP1 in the sense or antisense direction. RPE cell monolayers were assayed for water permeability in the presence of an osmotic gradient. 

The expression level of AQP1 in human RPE cells was compared with expression in human RPE in situ in comparison with β-actin. In Figure 3A the expression level of AQP1 in cultured RPE cells (2 weeks at confluence on permeable support) in comparison with β-actin was significantly lower than RPE in situ. 

Similar results were obtained from all four cell strains used in the present study (data not shown). We used an AV expression system to control AQP1 expression in adult cultured cells. With this system, we were able to incrementally control levels of AQP1 protein for functional assays and thus were able to attribute specifically the changes in permeability to expression of AQP1. Shown in Figure 3B is the expression of AQP1 in RPE cells that have been infected with AV containing the coding sequence for β-galactosidase as a control (CON) or AV containing the coding sequence for AQP1 at increasing multiplicity of infections (MOI = 0.1–10.0). RPE cell monolayers were maintained for 5 days to allow for AV-mediated protein expression. Figure 3C demonstrates the effect of AQP1 expression on permeability of RPE monolayers to water when faced with an osmotic challenge. As AQP1 protein increased, the volume of water moving across intact monolayers per unit of time increased in a significant and MOI-dependent manner. 

A second model system was used to extend findings with human adult RPE. The second approach used differentiated fetal RPE cells. Cells were plated onto filters and maintained in culture until differentiation marker proteins were expressed and transepithelial electrical resistance across monolayers was above 500 Ω cm². ²¹ The AV system was again used to control expression of AQP1. Because differentiated fetal RPE cells express robust levels of native AQP1 (Fig. 4A , lane 0), an AV containing DNA encoding AQP1 in the antisense orientation was used in experiments to knock down endogenous AQP1 to determine its specific contribution to water movement in this model system. Cells were infected and maintained for 5 days to facilitate knockdown. Figure 4A demonstrates AQP1 protein expression in fetal RPE cells that were not infected (0) or were infected with a control AV (containing no insert, E) or an AV containing antisense DNA (AS) coding for AQP1. Compared with expression of β-actin (β-ACT), the AV containing antisense AQP1 (AS) reduced endogenous AQP1 expression by approximately 80% 5 days after infection compared with cells infected with control AV (E). When fetal RPE monolayers were tested in an osmotic challenge assay, movement of water decreased significantly in AQP1 antisense-infected cells compared with cells infected with control AV (E) or uninfected cells (0; Fig. 4B ). 

Discussion

The present study shows for the first time the presence of AQP1 protein on cell surfaces of cultures of fetal RPE and in adult human RPE in situ. We used an AV expression system to control AQP1 protein levels and measured the contribution of channel-mediated transport compared with simple diffusion of water across intact monolayers of cultured adult and fetal RPE cells. AQP1 mediated water flux across RPE in an expression-dependent manner. 

We observed AQP1 among biotinylated proteins on cell surfaces of intact RPE in situ and in a differentiated fetal RPE model. Information from confocal imaging of the fetal RPE model was qualitative, showing fluorescence labeling of AQP1 on both apical and lateral cell surfaces. Results using the biotinylation assay illustrate that unlike the Na/K ATPase, most AQP1 in the adult was not biotinylated. This suggests that biotinylation of AQP1 is not efficient, recovery of AQP1 protein is not efficient, or AQP1 is expressed primarily on basolateral surfaces of RPE. Further study is needed to explore the later possibility. Currently, there are no published reports that AQP1 has a basolateral polarity. Two other aquaporins homologues, aquaporin-3 and -4, have a basolateral polarity. These aquaporin homologues have an expression pattern limited to the kidney. ^{28 29 30}  

Clearly the experimental approaches used in the present study improved on limitations found by others; however, it was not possible to distinguish between the nonbiotinylated AQP1 that resides on the basolateral plasma membrane and potential nonbiotinylated AQP1 located on intracellular vesicles. The remaining AQP1 protein was either on the plasma membrane and not biotinylated or located on intracellular vesicles. We reasoned that AQP1 localized predominantly in the plasma membranes, because this is the case with AQP1 in all other tissue but one. ³¹ In cholangiocytes the majority of AQP1 is located on intracellular vesicles. ³² When stimulated by secretin, intracellular cAMP levels rise and AQP1 containing vesicles shuttle to and fuse with the plasma membrane. This paradigm holds true for an AQP1 homologue, aquaporin-2, in the collecting ducts of kidney in response to vasopressin. ^{33 34} In both cases the insertion of aquaporin protein into the plasma membrane after stimulation dramatically increased transcellular permeability to water. By comparison, elevations in cAMP in RPE result in changes in transporter activity that decreases solute gradients across RPE, thereby decreasing transcellular fluid movement. ³⁵ Thus, we expect that as in all tissues but one (biliary duct), AQP1 in RPE is located primarily in the plasma membrane and not in intracellular vesicles. 

In epithelial cells, the Na/K ATPase localizes in a polarized fashion, generally to the basolateral domain. In some specialized epithelia, such as the choroid plexus and RPE, the Na/K ATPase exhibits a reversed polarity with apical Na/K ATPase. ^{6 36} In the case of the choroid plexus, AQP1 channels colocalize with Na/K ATPase to the apical domain, apparently facilitating the secretion of cerebral spinal fluid. ¹² In all other epithelium having a basolateral location of Na/K ATPase, AQP1 localizes to both apical and basolateral cell domains in a nonpolarized fashion. ¹² Data concerning AQP1 distribution in the present study were not conclusive. However, we surmise that the functional relationship between Na/K ATPase and AQP1 is different in RPE than in choroid plexus because of the unique role Na/K ATPase plays in RPE—namely, to supply a portion of sodium to outer segments of retinal photoreceptors and to provide a sodium gradient for chloride entry into the cell from the apical domain. 

To form the blood–retinal barrier, RPE forms intricate intracellular junctional complexes that have a transepithelial electrical resistance of approximately 2000 Ω cm². ³⁷ A consequence of this barrier function is that most of the water crossing the RPE must pass through its two plasma membrane domains, each containing differential amounts of AQP1. In the present study, we sought to determine the functional contribution of AQP1 to water movement across RPE monolayers. Using an AV expression system to control AQP1 protein levels in cultures of RPE, we found that AQP1 was requisite for enhanced water movement across RPE. At physiological expression levels of AQP1 in either model system, we observed water flux across RPE (apical to basal) at a rate of 2 to 18 μL/cm² per hour. These measurements are consistent with water movement across a healthy RPE that is apical to basal at a rate of 3 to 4 μL/cm² per hour. ³⁸ Therefore, it appears that the model systems used in the present study closely approximated the environment in vivo across the RPE. 

Efficient water movement across RPE in the apical to basolateral direction in vivo enables proper retinal attachment and function. There are several potentially blinding eye diseases in which fluid accumulates within the retina or in the subretinal space. For example, in eyes with macular edema, fluid accumulates within extracellular spaces and retinal cells. In eyes with serous retinal detachment, rhegmatogenous retinal detachment (retinal detachment associated with a retinal hole or tear), and central serous chorioretinopathy, fluid accumulates in the subretinal space. Agents that enhance vectorial movement of fluid across RPE, clearing the subretinal edema, would be very useful as therapy for these conditions. The adenoviral expression system used in the present study effectively enhanced vectorial fluid transport across RPE in vitro, demonstrating the promise of a gene therapy approach. Logistically, therapeutic agents, such as viral vectors, may be delivered to the RPE by intravitreal or subretinal injection. ³⁹ We hypothesize that overexpression of AQP1, delivered by gene transfer techniques, may be a useful treatment strategy for diseases in which intraretinal or subretinal fluid accumulates. Further work is needed in in vivo models to test this idea. 

 Supported in part by National Eye Institute Grants EY00444 (DB), EY00331 (DB), and EY09106 (GJJ); American Health Assistance Foundation (WDS); Research to Prevent Blindness Foundation (WDS, BSM) and the Dolly Green Endowed Chair (DB).

 Submitted for publication January 2, 2003; revised January 22, 2003; accepted January 30, 2003.

 Disclosure: W.D. Stamer, None; D. Bok, None; J. Hu, None; G.J. Jaffe, None; B.S. McKay, 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.

 

Localization of AQP-1 in cultured fetal human RPE by immunofluorescence confocal microscopy. Fetal human retinal pigment epithelial cells were seeded onto porous filters and cultured. (A–D) Selected sequential optical sections (1 μm) of RPE on filters in the apical (A) to basal (D) direction. (E) A computer-generated view of AQP1 labeling in cross section (z-axis) through the RPE monolayer. (F) Western blot analysis of AQP1 expression in RPE on filters. Shown are cells on one representative filter of three that were analyzed by confocal microscopy.

Figure 2.

 

Localization of AQP-1 in adult RPE in situ. Proteins at cell surfaces of intact RPE monolayers in situ were biotinylated, isolated, fractionated, and probed. Lane 1: the amount of AQP1 (top), the Na/K ATPase α₁ subunit (middle), and total biotinylated proteins (TOT BIOT, bottom) present in RPE whole-cell lysates (25 μg) after biotinylation and before exposure to streptavidin agarose. Lane 2: amount of protein remaining in lysate after exposure to streptavidin agarose (pass-through). Lane 3: amount of protein present in third wash buffer of streptavidin agarose. Lane 4: amount of proteins that specifically bound to streptavidin agarose. Lanes 1 to 3: volume equivalents; lane 4: a 3× concentrated sample compared with the others. Right: molecular mass markers in kilodaltons. Shown is one representative experiment of three total.

Figure 3.

 

Expression and functional analysis of AQP1 in adult human RPE monolayers. (A) Expression of native AQP1 in human RPE cell in situ (T) versus cultured cells (C, fourth passage) relative to expression of β-actin (β-ACT, bottom). (B) Adenovirus-mediated expression of recombinant AQP1 in cultured human RPE compared with β-actin (5 days after infection). Shown are cells infected with control AV containing insert coding for β-galactosidase (CON) or increasing doses of AV containing insert coding for AQP1 (MOI refers to the number of infective viral particles per cell). (C) Net amount of water movement across infected monolayers in response to osmotic gradient (∼150 mOsM). Data are expressed as rate of water movement, J_v. Significant differences between AQP1-expressing monolayers versus control: *P < 0.05, **P < 0.01. Shown are representative experiments of the five performed.

Figure 4.

 

Expression and functional analyses of differentiated fetal RPE monolayers. (A) Expression of endogenous AQP1 in fetal RPE monolayers not infected (0) or after infection with control AV (E) or AV-containing antisense AQP1 DNA (AS). (B) Amount of water movement across infected monolayers in response to osmotic gradient (∼150 mOsM). Data are expressed as rate of water movement, J_v. Significant differences between AQP1-expressing monolayers versus control (**P < 0.01). Shown is one representative experiment of two.

The authors thank Janice Burke for her careful review of the manuscript.