Expression of Aquaporin-1 in Human Trabecular Meshwork Cells: Role in Resting Cell Volume

Abstract

purpose. Drainage of aqueous humor from the human eye appears dependent on intracellular volume of trabecular meshwork (TM) cells, the predominant cell type of the human outflow pathway. Thus, the modulation of water and solute flux across the plasma membrane of TM cells is predicted to be an important factor in regulating outflow facility. Aquaporin (AQP)-1 is a hexahelical integral membrane protein that functions as a regulated channel for water and cations in fluid-secreting and -absorbing tissues. AQP1 is present in many tissues of the human eye, including the TM; however, its role in outflow facility is unknown. The purpose of the present study was twofold: to evaluate the prospect of manipulating AQP1 protein levels in TM cells using sense and antisense mRNA and to investigate the functional role of AQP1 in TM cells.

methods. An adenovirus (AV) expression system was used to alter AQP1 protein levels. AQP1 protein expression was monitored using immunoblot analysis, and resting cell volume was measured by forward light scatter, electronic cell sizing, and [¹⁴C]-sucrose/urea equilibration. Permeability of TM monolayers to[ ¹⁴C]-sucrose was also assessed as an indirect evaluation of cell volume.

results. AV-mediated gene transfer of AQP1 cDNA to TM cells resulted in a titer-dependent increase in recombinant AQP1, whereas transfer of antisense cDNA decreased native AQP1 protein by 71.7% ± 5.5% (P < 0.01) after 5 days. A novel finding of this study is that mean resting volumes of AQP1(s) AV-infected TM cells in suspension were 8.7% ± 3.0% greater (P < 0.05) than control cells. Conversely, AQP1 antisense (as) AV-infected cells had resting volumes 7.8% ± 2.9% less than control cells (P < 0.05). Similar effects of AQP1 expression on resting cell volume were observed in TM monolayers. Consistent with this finding, paracellular permeability of AQP1(s) AV-infected TM monolayers to [¹⁴C]-sucrose decreased by 8.0% ± 1.4% (P < 0.001).

conclusions. In addition to influencing the osmotic permeability of TM plasma membranes, the level of AQP1 protein expression influences resting intracellular volume and thus paracellular permeability of TM cell monolayers in vitro. These data suggest that AQP1 expression may affect outflow facility in vivo.

Aquaporin (AQP)-1 is a hexahelical integral membrane protein that functions as a regulated channel for water and cations. ^{1 2 3 4} In general, the distribution of AQP1 is limited to fluid-secreting and -absorbing tissues in the human body that demonstrate an enhanced permeability to water relative to other tissues. ⁵ In some tissues such as the proximal tubules of the kidney, AQP1 functions in the bulk reabsorption of water. ^{6 7} In other tissues such as the biliary ducts, AQP1 participates in the concentration of bile in a manner regulated by the hormone secretin. ⁸ In fibroblastic cells, such as keratocytes, the function of AQP1 is unknown. ⁹  

In the human eye, AQP1 is present in many tissues that require the efficient movement of water, including the corneal endothelium, the lens epithelium, the iris epithelium, the nonpigmented ciliary epithelium, and the cells of the conventional outflow pathway. ^{9 10} The conventional outflow pathway contains two cell types, trabecular meshwork (TM) and Schlemm’s canal (SC) endothelial cells, both of which express AQP1. ^{9 10 11} TM cells cover collagen lamellae that form the maze of passages through which water must flow in the outflow pathway. As aqueous humor drains from the eye, it first passes through the tortuous TM until finally reaching and crossing a monolayer of endothelial cells that line SC. Because aqueous humor appears to move primarily by bulk flow around TM cells and not through them (i.e., paracellular rather than transcellular), ¹² previous studies indicate that intracellular volume of TM cells appears to be a determinant of outflow facility. ^{13 14 15} For example, agents that decrease TM-cell volume increase outflow facility, and agents that increase TM-cell volume decrease outflow facility. These findings suggest that intracellular volume of TM cells in vivo may dictate outflow resistance by affecting the dimensions or direction of the human outflow pathway; particularly in the juxtacanalicular region, where spaces between TM cells approach 1 μm. Whether AQP1 plays a physiological role in outflow across TM cells is unknown. 

Defining the functional contribution of AQP1 to the permeability of TM tissue and to outflow facility may be important for understanding the pathologic course of glaucoma, the second leading cause of blindness in the United States. ¹⁶ Glaucoma is an optic neuropathy that is generally characterized by elevated intraocular pressure resulting from impaired outflow facility. Unknown cellular defect(s) in the TM decrease outflow facility, resulting in an elevation in intraocular pressure that compresses nerve axons in the retina and results in blindness. ^{17 18}  

The present study was conducted as an initial investigation of the role of AQP1 in TM cells. Using adenovirus (AV)-mediated gene transfer of AQP1 cDNAs, we were able to control the expression of AQP1 in TM cells, thus demonstrating the utility of AQP1 AV as an important tool for studying AQP1 function in these cells. In the present study, we show that altering the expression of AQP1 influenced resting volumes of TM cells and consequently their paracellular permeability. In addition, these studies revealed that knockdown of native AQP1 in TM cells by AQP1 antisense AV decreased native AQP1 protein and correspondingly decreased TM cell volume. Taken together, these data suggest that AQP1 protein levels influence the set point for resting volume in TM cells. 

Methods

Human Trabecular Meshwork Cells

TM cells were isolated by using a blunt dissection technique in conjunction with extracellular matrix digestion and were cultured as previously described. ¹⁹ The cell strains used in this study were isolated from nonglaucomatous donor eye tissue from four different individuals (TM22, TM23, TM26, and TM29) of ages 55, 72, 15, and 0 years, respectively. 

AV Construction

The AV backbone for the AQP1 sense (s) and antisense (as) AV constructs was a replication-deficient first-generation AV with deletions of the E1 and E3 genes. ²⁰ This empty AV contains the cytomegalovirus (CMV) promoter and bovine growth hormone polyadenylation (bGH) site separated by a polylinker containing a unique XbaI restriction site. A large-scale preparation of empty AV DNA (described later) was digested with proteinase K (Sigma, St. Louis, MO) in the presence of 0.5% SDS at 55°C for 2 hours, followed by phenol-chloroform (1:1, vol/vol) extraction and ethanol precipitation. Viral DNA was then restricted with XbaI overnight, and the large fragment containing the bGH and AV (map units 9.3-100) was gel purified from a 0.6% agarose gel. This DNA fragment served as the right end of both AQP1(s) AV constructs. 

The left ends of AQP1 recombinant viruses were constructed uniquely (see Fig. 1 ). For AQP1(s) AV, a plasmid containing the coding sequence for AQP1, pCHIPev, ²¹ was digested with KpnI and BamHI (giving a 1162-bp fragment) and subcloned into the shuttle vector pSKAC, creating pSKAC/AQP1. pSKAC contains the left end cassette (map units 0.0-1.3) of the AV that is flanked by two unique restriction sites, PmeI and XbaI. The cassette includes the left terminal repeat of AV, a CMV promoter, an AMV translation enhancer, and a polylinker region. 

An 807-bp DNA fragment encoding AQP1(as) was generated by PCR, with pCHIPev used as a template and primers that delineated the full coding sequence of AQP1: 5′-CGCGAATTCCTATTTGGGCTTCATCTC-3′ and 5′-CGCGGATCCATGGCCAGCGAGTTCAAG-3′. Primers were engineered to contain EcoRI and BamHI restriction sites (in italics), respectively. The DNA fragment was gel purified, restricted with BamHI and EcoRI, and subcloned into pSKAC, creating pSKAC/AQP1as. Left end DNA fragments containing AQP1 sense or antisense DNA were liberated from pSKAC after restriction with PmeI and XbaI. 

Approximately 100 ng of appropriate gel-purified left-end fragment of AQP1(s) AV or AQP1(as) AV was ligated to approximately 1 μg of the right-end fragment overnight at 16°C, using T4 ligase (Gibco BRL, Gaithersburg, MD). The ligation mixture was transfected into single 60-mm dishes of strain 293 human embryonic kidney cells (293 cells) using lipofectamine reagent (Gibco). Transfected cells were allowed to lyse without an agar overlay. Cell lysates were subjected to three rounds of freezing-thawing using a dry ice-ethanol bath. Individual viruses were isolated from cell lysates by two consecutive rounds of plaque purification, using an agar overlay. 

After isolation, individual viruses were amplified in EBNA-293 cells (293 cells constitutively expressing EBNA-1 protein from Epstein-Barr virus; Invitrogen, San Diego, CA). EBNA 293 cells on forty 150-mm plates (at ∼50% confluence) were infected with appropriate AV at a multiplicity of infection (MOI) of 3. When the majority of cells were floating (36–48 hours after infection), the cells were harvested by gentle scraping and collected by a 5-minute centrifugation at 1000g. The cell pellet was resuspended in 20 mM Tris-HCl and 2 mM EDTA (pH 7.4) and the cells were homogenized with 20 strokes (Dounce homogenizer; Kontes Glass, Vineland, NJ). RNase A was added to 100 μg/ml, and the homogenate was incubated at 22°C for 5 minutes. Nuclei were removed by centrifugation at 2500g for 10 minutes. CsCl was added to the supernatant to 0.3 g/ml and the supernatant layered atop a CsCl step gradient (1.3 and 1.4 g/ml in virus storage buffer [VSB]: 137 mM NaCl, 20 mM Tris-HCl, 5 mM KCl, and 1 mM MgCl₂ [pH 7.4]) and centrifuged for 2 hours at 30,000 rpm (TH64 rotor; Sorvall, Newtown, CT). The virus band that formed at the 1.3- to 1.4-g/ml interface was removed with a 16-gauge needle and layered atop a 2-ml bed of Sepharose (Amersham Pharmacia Biotech, Piscataway, NJ) and centrifuged for 2 minutes at 1000 rpm in a tabletop centrifuge (Beckman Coulter, Fullerton, CA). This step was repeated once more, and the virus concentration was adjusted to 1 × 10¹¹ plaque forming units (PFU)/ml in VSB. Sucrose was added to 10% (final concentration) and the virus preparation was stored in aliquots at− 80°C. Each aliquot was used a maximum of two times and discarded. 

Individual AV DNA titers were determined by three methods: (1) plaque titration on 293 cells, (2) immunofluorescence microscopy of AV protein expression (anti-penton group antigen, clone 143; Biodesign, Saco, ME) in 293 cells infected with serial dilutions of AV, and (3) absorbance (A) at 260 nm (plaque forming units per milliliter = A₂₆₀ × dilution × 10¹⁰). Infection of cells with AV is expressed as MOI, indicating number of infective virus particles per cell. A range of titers was used to functionally test novel AQP1 AV constructs and was dependent on the specific requirements of a particular assay. 

Immunoblot Analysis

SDS-solubilized whole-cell lysates 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 using a commercial system, according to the manufacturer’s instructions (Transblot; Biorad, Hercules, CA). 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 (1:5000) or anti-β-actin IgG (AC-15, 1:2000) 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, 1:5000; Pierce, Rockford, IL). The blots were washed (three times for 15 minutes each) in TBS-T, and specific labeling was visualized after enhanced chemiluminescence (Pierce) and exposure to film (ECL-Hyperfilm; Amersham, Arlington Heights, IL). Immunoblots were digitized using a gel documentation system, and densitometry was performed on computer (LabWorks software; UVP, Upland, CA). 

Cell Surface Biotinylation

Monolayers of TM cells in 10-cm culture plates were infected with AQP1(s) AV (MOI, 1.0). Three days after infection, cells were rinsed with cation-free phosphate-buffered saline (CF/PBS; three times, 10 ml) and incubated with versene for 5 minutes at 4°C. Cells were rinsed twice with ice-cold biotinylation buffer (100 mM NaCl, 50 mM NaHCO₃ [pH 8.0]) and incubated with NHS LC-biotin (300 mg/ml, Pierce) in biotinylation buffer at 4°C with rocking. After 1 hour, cells were rinsed with PBS (three times, 10 ml) at 22°C and incubated in quenching buffer (100 mM glycine, 25 mM Tris/HCl [pH 7.4]) for 5 minutes. Cells were again rinsed with PBS (two times, 10 ml) and solubilized in 2 ml of 0.1 N NaOH. Cell lysates were diluted 1:10 in PBS and centrifuged at 3000g, and supernatants were incubated with streptavidin-agarose beads (Pierce) at 22°C with rotation. After 1 hour, beads were washed with 30 bed volumes of PBS. Proteins specifically bound to beads were extracted by boiling in buffer containing 3% SDS. 

Total RNA Isolation

Ten milliliters guanidine isothiocyanate buffer (4 M guanidine isothiocyanate, 25 mM sodium acetate, and 650 mM β-mercaptoethanol) was added to a 15-cm culture dish confluent with 293 cells at 4°C for 30 seconds. The solution was drawn up and down into a syringe with a 23-gauge needle three times, applied to 12.5 ml of 5.7 M cesium chloride solution in polyallomer tubes (Beckman) and centrifuged for 24 hours at 25,000 rpm at 20°C using an ultracentrifuge (model L565; with an SW28 rotor; Beckman). The liquid was removed by aspiration from the centrifuge tube, leaving a clear pellet. The RNA pellet was dissolved in diethyl pyrocarbonate–treated water and ethanol precipitated as described previously. ¹¹ The integrity of the RNA was verified after separation by electrophoresis into 0.8% agarose gel and visualization using ethidium bromide. 

Reverse Transcription–Polymerase Chain Reaction

With AMV reverse transcriptase (Boehringer-Mannheim, Indianapolis, IN), DNA copies of total RNA or AQP1 antisense RNA from 293 cells were made using oligo dT primers or a primer specific for AQP1(as) RNA transcripts (5′-CGCGAATTCCTATTTGGGCTTCATCTC-3′), respectively, as described previously. ¹¹ The presence of AQP1(as) RNA in infected 293 cells was determined using a primer set specific for AQP1 DNA. The primer set corresponded to nucleotides 316-336 (5′-ATCGGATCCACCGCCATCCTC-3′) and 646-671 (5′-ACAGAATTCTATCCCCCGATGAATGG-3′) of human AQP1. Amplification of RT cDNA by Taq polymerase was performed as previously described, ¹¹ using 25 PCR cycles (94°C for 30 seconds, 50°C for 30 seconds, and 72°C for 1.5 minutes). PCR products were analyzed on 1.4% agarose gels. 

Osmotic Challenge Assay

Water permeability was measured as the net fluid movement driven by an osmotic gradient across AV-infected Madin-Darby canine kidney (MDCK) cells. Cells were seeded onto filters (Transwell; Costar, Cambridge, 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 2 weeks to allow for cell–cell junctions to mature. Cells were infected at the apical surface with AV containing AQP1 in the sense or antisense orientation or with empty AV at an MOI of 10. Five days after infection, medium was removed from both upper and lower chambers. Exactly 1 ml fresh, prewarmed isotonic medium (300 mOsM) was added to the lower chamber, and exactly 175 μg hypertonic medium (400 mOsM) was added to the upper chamber. After 1 hour of incubation at 37°C, medium from the upper chamber was removed and volume measured using an analytical balance. 

Light-Scattering Measurements

Similar to methods described previously, changes in intensity of forward light scattering from cell monolayers were used as an indicator to assess changes in cell volume in real time. ^{22 23 24} In pilot experiments using both hypertonic and hypotonic challenges, optimal conditions for resolving volume changes for confluent TM cell monolayers were determined empirically. TM cells were seeded onto glass coverslips (13 × 37 × 1 mm) and grown to approximately 80% confluence. Cells were infected with AV (MOI, 0.1–1.0) and grown for 3 days. Confluence of cell monolayers was verified by phase-contrast microscopy before assay (IM-35 microscope; Carl Zeiss, Thornwood, NY). Coverslips containing cells were placed across the diagonal in a 3-ml polystyrene cuvette fitted with inlet and outlet lines. The cuvette was perfused at 3 ml/min with Hanks’ balanced salt solution supplemented with 25 mM HEPES (H-HBSS, pH 7.4), and maintained at 37°C. Once a stable baseline was obtained (typically, 10 minutes), medium osmolarity was changed by stopping perfusion and rapidly exchanging the contents of the cuvette with 9 ml prewarmed hypertonic medium (450 mOsM), using a 20-ml syringe. Control exchanges of prewarmed isotonic medium were performed in all cases before hypertonic challenge. Light (512 nm) was directed toward cell monolayers at a 45° angle of incidence, and light scattered forward from the monolayer (45°) was monitored in real time at a wavelength of 520 nm (RatioMaster Spectrofluorometer; PTI, South Brunswick, NJ). Light beam intersected 1.11 mm² of the cell monolayer (approximately 300 cells). Comparisons of monolayer permeabilities were obtained on computer (SigmaPlot software; SPSS, Chicago, IL), by fitting quadratic polynomials to scatter versus time curves (dV/dT) for initial response periods (10 seconds) for each condition. 

Intracellular Volume Measurements

TM-cell volume was measured by two methods. The volume of individual TM cells in suspension was determined by electronic cell sizing using a cell counter (Beckman Coulter, Fullerton, CA), as described previously. ²⁵ For TM cells attached to tissue culture plates, cell volume was determined by radioisotopic evaluation of intracellular water space, by using[ ¹⁴C]-urea and[ ¹⁴C]-sucrose as markers of total and extracellular water space, respectively. ²⁵  

To determine cell volume by electronic cell-sizing methods, mean cell volumes of TM cells in suspension were analyzed by cell counter (model ZM, with a C-1000 Channelizer; Beckman Coulter). The orifice diameter used for these experiments was 140 μm. Cells on T75 culture flasks were infected with appropriate AV (MOI, 0.1) when 80% to 90% confluent. After 4 days, cells were briefly trypsinized and rinsed free of trypsin with PBS that contained a trypsin inhibitor (Clonetics Corp., San Diego, CA). Cells were resuspended at a concentration of approximately 2 × 10⁵ cells/ml in HEPES-buffered minimal essential medium (MEM). Cells were diluted to a final concentration of approximately 50,000 cells/ml, and mean cell volumes of cells in suspension were calculated, by using a minimum of 5000 cells per measurement × three independent measurements for a single data point. Absolute cell volumes (picoliters per cell) were calculated from distribution curves of cell diameter, using a standard curve generated by polystyrene latex beads of known diameter (9.87 and 14.51 μm). Cell suspensions were maintained at 37°C throughout the assay period. 

For the radioisotopic method, cell monolayers were cultured in 12-well plates until 80% to 90% confluent and were infected with the appropriate AV (MOI, 0.1). After 4 days, cells were equilibrated for 10 minutes with HEPES-buffered MEM at 37°C in a water bath. The medium was changed to an assay medium that contained either[ ¹⁴C]-sucrose or[ ¹⁴C]-urea (final concentration, 1.0 μCi/ml) in HEPES-buffered MEM. After a 20-minute incubation, the medium was aspirated and each well was rapidly rinsed with ice-cold PBS (four times, 2 ml). Wells were air dried and extracted with 0.1 N NaOH for quantitation of radioactivity and protein. Specific activities (counts per minute per milliliter) of [¹⁴C]-sucrose or[ ¹⁴C]-urea in the assay medium were determined and used to calculate water space associated with trapped radioactivity (expressed as microliters per milligram protein). Intracellular volume was calculated as the difference between water space determined for[ ¹⁴C]-urea (a measure of intracellular plus trapped extracellular volume) and water space determined for[ ¹⁴C]-sucrose (a measure of trapped extracellular volume). 

Permeability Measurements

Similar to that described previously, ³⁵ flux of[ ¹⁴C]-sucrose across confluent TM cell monolayers was used to assess paracellular permeability. Cells were seeded onto filters (Transwell; Costar) at a density of 1 × 10⁵ cells per well, were maintained for 1 day in culture, and then were infected with AV at an MOI of 0.1. Five days after infection, cells were assayed. Medium was aspirated from upper and lower chambers, and exactly 1.0 ml of fresh isotonic medium was added to the lower chamber. To the upper chamber, exactly 200 μl of fresh isotonic medium containing [¹⁴C]-sucrose (1 μCi/ml) was added. Cells were maintained at 37°C with rotary shaking during the incubation period. After 10 minutes 100 μl medium from the lower chamber was removed, added to 5 ml of scintillation cocktail, and counted. In control experiments, hypotonic medium (200 mOsM) was used in place of isotonic medium in both chambers. 

Statistical Analyses

Experimental results were analyzed by a two-tailed Student’s t-test assuming unequal variance. The criterion for significance in all cases was P < 0.05. 

Results

The Results section is divided into two parts. The first part (Figs. 1 2 3 4 and 5) describes the characterization of two novel AQP1 AV constructions, AQP1(s) and AQP1(as). A full range of infective virus particles per cell (MOI, 0.1–100) was tested using in several assays in various cell models, including human TM cells. Using the characterized viruses to control AQP1 expression, the second part (Figs. 6 7 8 and 9) describes the specific effect of AQP1 expression on resting volume and paracellular permeability of human TM cells. 

AV-Mediated AQP1 Expression

AQP1 AVs were constructed by the directional ligation of a first-generation AV backbone (E1/E3 deletion) to DNA fragments that contain the terminal repeat of AV, a CMV promoter plus DNA that codes for AQP1 (Fig. 1) . Expression of recombinant AQP1 protein and AQP1(as) RNA (described later) was demonstrated after infection of TM cells and 293 cells. Figure 2 shows that increasing titers of AQP1(s) AV (MOI, 0.1–100) resulted in an incremental increase in expression of AQP1 protein in TM cells. Both the 28-kDa nonglycosylated and 40- to 50-kDa glycosylated forms of AQP1 were detected. After longer exposure of immunoblots, native AQP1 (MOI, 0) and recombinant AQP1 (MOI, 0.1) were observed (Fig. 2B) . A time course in days for the expression of AV-mediated AQP1 protein expression (MOI, 1.0) is shown in Figure 2C . A plateau of AQP1 expression was attained by 5 days for all virus titers tested (MOI, 0.1; not shown). For each AV, MOI (number of infective virus particles per cell) was determined empirically using three different assays (see Methods). 

Recombinant AQP1 as a Functional Water Channel

TM cell monolayers were infected with AQP1(s) AV (MOI, 1.0) and tested for the functional presence of recombinant AQP1 protein at their plasma membrane. To determine whether recombinant AQP1 was at the plasma membrane of TM cells, cell surface proteins were biotinylated, solubilized, and purified using streptavidin-agarose. We found recombinant AQP1 amid biotinylated cell surface proteins (Fig. 3 , lane 4). 

To determine whether recombinant AQP1 functions as a water channel in TM cells, cells were infected with empty or AQP1(s) AV, exposed to hypertonic medium (450 mOsM) and cell volume then assessed by means of a forward light-scatter assay. We predicted that an increase in AQP1 expression should increase the permeability of the TM-cell plasma membrane to water and thus alter the rate of initial cell volume changes as cells were exposed to anisosmotic media. By this method, volume changes corresponded to changes in forward light scatter from TM cell monolayers over time. As predicted, TM cells infected with AQP1(s) AV shrank more rapidly in response to a hypertonic challenge (100 mOsM gradient; Fig. 4A ) than cells infected with empty virus or uninfected cells (not shown). Cell volume decreased 1.8- or 4.0-fold (MOI, 0.1 or 0.5, respectively) more rapidly than control. Comparisons were obtained by fitting quadratic polynomials to scatter versus time curves (dV/dT) of initial response (first 10 seconds) to hypertonic challenge. 

In addition, we tested the function of recombinant AQP1 in a model system that measures channel-mediated transcellular movement of fluid across MDCK-cell monolayers. ²⁷ MDCK-cell monolayers infected with AQP1(s) AV were five times more permeable to fluid flux driven by a 100-mOsM osmotic gradient than cells infected with empty AV (P < 0.001; Fig. 4B ). MDCK cells were used in these experiments, because they form tight cell–cell junctions that limit paracellular fluid flux. 

Expression of AQP1(as) AV RNA

The ability of AQP1(as) AV to direct the expression of AQP1(as) RNA was assessed by RT-PCR in HEK293 cells, which do not express native AQP1. DNA copies of HEK293 mRNAs were made using reverse transcriptase and oligo dT primers or primers specific for AQP1(as) mRNA. Analysis of cDNAs verified the presence of AQP1(as) mRNA in AQP1(as) AV–infected cells but not in cells infected with empty virus. A primer set spanning two of AQP1’s three introns amplified DNA of appropriate size (355 bp). PCR products were obtained using either oligo dT or antisense-specific, oligonucleotide-primed cDNAs as templates (Fig. 5A) . The ability of AV-driven AQP1(as) RNA to hybridize to native AQP1 mRNA and inhibit expression of AQP1 protein was investigated in TM cells, which are known to express native AQP1. TM cells were infected with AQP1(as) AV (MOI, 10) and grown for 5 days. The presence of native AQP1 in empty AV and AQP1(as) AV–infected cells was determined by immunoblot analysis. Steady state levels of native AQP1 protein in cells infected with AQP1(as) AV were decreased by 71.7% ± 5.5% (mean ± SEM) of control (Fig. 4B) . Whereas native AQP1 expression varied between TM-cell strain and passage number, the knockdown level was very consistent. Comparisons were based on changes in the nonglycosylated 28-kDa AQP1 species and normalized to β-actin protein expression. A representative blot is shown (Fig. 5B , bottom). Similar levels of knockdown were obtained using lower concentrations of virus (Fig. 6) . 

AQP1 Protein Expression and Resting Cell Volume

While measuring effects of AQP1 expression on intracellular volume responses to anisotonic challenges, we observed a novel role for AQP1. Cellular resting volume of TM cells both in suspension and in monolayer correlated with AQP1 expression. For these experiments, resting volume of TM cells was determined using two different methods. Intracellular volume of individual cells in suspension was resolved by electronic cell sizing using a cell counter (Coulter), and intracellular volume of cell monolayers was determined by a urea-versus-sucrose equilibration assay. Similar to findings by others, ²⁷ pilot experiments in our laboratory showed that specific effects of AQP1 AVs on resting-cell volume was resolved optimally using low MOIs (< 1.0), primarily because of nonspecific effects of AV on cell volume that were noted at high MOIs (>1.0). Therefore, cells were infected with AQP1(s), AQP1(as), or empty AV at an MOI of 0.1 and grown for 4 days. Shown in Figure 6 is the protein expression pattern and specifically AQP1 expression levels in TM cells infected with AV at and MOI of 0.1. This figure illustrates the typical expression observed for AQP1 in AQP1(s) and AQP1(as)-infected TM cells. Native AQP1 expression levels in empty-AV–infected TM cells, however, varied and was dependent on the primary TM cell line used and its number of cell doublings at the time of use (see Fig. 5B ). 

As shown in Figure 7 , cells infected with AQP1(s) AV had significantly greater mean intracellular volume than cells infected with either empty AV (8.7%, P < 0.05) or AQP1(as) AV (16.5%, P < 0.01). Cells infected with AQP1(as) AV had significantly less mean intracellular volume than cells infected with empty AV (7.8%, P < 0.05). Shown are combined data from four independent experiments using three different TM cell strains. A minimum of 15,000 cells was analyzed for each data point in each experiment. 

The effects of AQP1 protein expression were also tested in monolayers of TM cells. Monolayers of TM cells were grown until 80% to 90% confluent and were infected at an MOI of 0.1. After 4 days, cells were assayed for intracellular water volume using a marker for intracellular plus extracellular water, [¹⁴C]-urea, and a marker for trapped extracellular water,[ ¹⁴C]-sucrose. Results shown in Figure 8 were similar to those obtained with individual cells. Thus, cells infected with AQP1(s) AV had a significantly greater mean intracellular volume than did cells infected with empty AV (19.9%, P < 0.05) or AQP1(as) AV (28.9%, P < 0.01). Conversely, cells infected with AQP1(as) AV had a smaller, but not significant, mean intracellular volume than empty-AV–infected cells (9.0%, P = 0.18). Retrospectively, we determined that low native AQP1 expression in three of six experiments affected our ability to reach significance. 

Effect of AQP1 Protein Expression on Paracellular Permeability

Previous studies have demonstrated that resting volume of TM cells affects paracellular permeability. ³⁵ To functionally determine the effect of AQP1’s expression on the paracellular permeability of TM cells, we measured the flux of a paracellular marker, [¹⁴C]-sucrose, across AV-infected TM cell monolayers grown on permeable filters. Figure 9 shows that TM cells infected with AQP1(s) AV were significantly less permeable to [¹⁴C]-sucrose than empty-AV–infected (8.0% ± 1.4%, P < 0.001) or antisense-infected (5.8% ± 1.6%, P < 0.01) cells. Conversely, the permeability of cell monolayers infected with AQP1(as) was not significantly different from cell monolayers infected with empty AV. Although AQP1(as) successfully knocked down low endogenous AQP1 (see Fig. 6 for example), this assay was not able to resolve differences between empty and AQP1(as) AV–infected cells. In control experiments, cell monolayers infected with AQP1(as) or empty AV were exposed to hypotonic medium (100 mOsM) and assayed for paracellular permeability. Similar to AQP1(s) AV–infected cells, cells assayed in hypotonic medium were significantly less permeable to[ ¹⁴C]-sucrose (due to cell swelling) than cells in isotonic medium (not shown). 

Discussion

We constructed and used AVs that contain AQP1 in the sense or antisense orientation as a mechanism to control AQP1 expression in TM cells. We showed that AQP1 influenced osmotic permeability of TM monolayers as a control and for the first time demonstrated, by three different methods, the association of the resting volume of TM cells with the expression of AQP1. Thus, intracellular volume of TM cells infected with AQP1(s) AV was greater, whereas those infected with AQP1(as) AV was less than cells infected with empty AV. Similar results were obtained, regardless of whether TM cells were examined as single cells or as part of a cell monolayer. Because a change in resting intracellular volume cannot occur without a change in the amount of osmotically active solute in the cells (with osmotically obliged water after solute), our findings suggest that AQP1 expression must in some manner alter intracellular solute. This hypothesis is supported by an observation in proximal tubule epithelial cells that hypertonic medium results in long-term increases in both AQP1 expression and cell volume. ⁷  

In general, resting cell volume is a function of the coordinated activities and expression of a variety of channels and transporters that control the influx and efflux of ions and water at the plasma membrane. The mechanisms whereby cells sense their volume and maintain intracellular volume at a given set point are not fully understood. However, a number of ion transporters and channels are known to participate in the regulation of intracellular volume of a variety of cell and tissue types. ^{29 30} These include ion flux pathways that tend to increase cell volume, such as the Na-K-2Cl cotransporter, the Na/H exchanger coupled to Cl/HCO₃ exchange, and the Na/K pump. Volume regulatory pathways also include those ion flux pathways that act to decrease intracellular volume, such as K-Cl cotransport, K channels, and Cl channels. With regard to Cl channels, it is noteworthy that the cystic fibrosis transmembrane conductance regulator (CFTR), a chloride channel involved in cell volume regulation, affects ion conductance of at least three different ion channels (Na, K, and Cl) and a water channel, AQP3. ^{26 31 32 33 34} In airway epithelial cells, water transport through the AQP3 water channel is functionally coupled to CFTR. ³³ Thus, AQP1 may influence the activity of an ion channel and/or transporter and affect resting cell volume in TM cells. 

Except for the recent characterization of Na-K-Cl cotransporter in TM cells, the specific participants that determine TM resting volume are not known. ³⁵ Using bumetanide and ethacrynic acid as inhibitors, the Na-K-Cl cotransporter was shown to contribute significantly to the regulation of TM-cell volume. The functional relevance to outflow cell volume of Na-K-Cl activity in the outflow pathway was detected in one study, but not in another. ^{13 36} Although Gabelt et al. ³⁶ failed to observe effects of bumetanide on outflow facility, Al-Aswad et al. ¹³ found that treatment of calf and human anterior chambers in organ culture with bumetanide (10 μM) decreased TM-cell volume by 10% and increased outflow facility by an average of 23%. Thus, small changes in TM-cell volume produced significant changes in outflow facility. This relationship between outflow cell volume and outflow facility is consistent with studies from two other laboratories using outflow perfusion models. ^{14 15} Accordingly, conditions that increase TM-cell volume decrease outflow facility, whereas those that decrease cell volume increase outflow facility. In the present study, AQP1 expression increased cell volume 8.7% to 19.9% (depending on the assay) resulting in decreased paracellular permeability of cell monolayers to sucrose by 8.0%. We propose that a coordinated change in the intracellular volume of TM cells influences outflow facility by modifying extracellular pathways for aqueous flow in the juxtacanalicular region of the outflow pathway. Such a mechanism may account for the nonuniform flow of aqueous through the TM that has been predicted by previous models and may contribute to increased outflow resistance predicted by Johnson’s funneling hypothesis. ^{37 38}  

A second possibility is that functional effects of outflow cell volume changes occur only at the level of the inner wall of Schlemm’s canal (IWSC), cells that also express AQP1. ^{9 10} Thus, the intracellular volume of IWSC cells may dictate a preferential paracellular or transcellular route for aqueous across the inner wall. Although IWSC cells are held together by tight junctions, a paracellular pathway between IWSC cells was revealed in experiments in which enucleated eyes were perfused with cationic ferritin. ⁴⁰ Thus, shrinkage of IWSC cells may weaken the cell–cell contacts and provide a paracellular route for aqueous humor. Evidence for this idea was first presented in work using vascular endothelial cells. Both in vivo and in vitro studies show that the intracellular volume of vascular endothelial cells modulates water and solute flux across a monolayer of endothelial cells. ^{41 42 43} The cells of the IWSC are functionally similar to vascular endothelium, in that they both present highly regulated barriers to water and solute flux. An alternative possibility must be considered. The enhanced permeability of the IWSC to water may simply be a function of the transcellular movement of water through AQP1 in the plasma membrane of SC cells. Such a mechanism is certainly operational in cell barriers of other tissues that express AQP1. ^{44 45} Thus, the functional role for AQP1 in SC may oppose that in TM, but must be determined experimentally. 

Using AQP1(s) AV and AQP1(as) AV to control AQP1 expression, the functional contribution of AQP1 to aqueous outflow facility can now be tested in systems that model the human outflow pathway. For example, infection of TM and SC cells in human anterior chambers in organ culture or in live primates will enable the examination of AQP1’s specific contribution to outflow facility. Further, the specific participation of AQP1 to outflow facility at the level of IWSC can be assessed using cultured SC cell monolayers on permeable supports. ²⁸ Such experiments are compelling, because information about channel-mediated (AQP1) versus non–channel–mediated (endocytic) fluid movement across SC can be measured directly. 

In addition to providing a valuable tool to understand aqueous outflow function, AQP1 AV may have therapeutic utility. The viability of adenoviral transfer of AQP1 to tissues with a secretory or absorptive deficit has been recently demonstrated. ⁴⁶ After radiation treatment, secretory function was restored to cells of the rat salivary glands using adenoviral delivery of recombinant AQP1 in vivo. These studies were extended in vitro to show that delivery of low virus titers of AQP1(s) AV (similar to those used in the present study) to salivary cell monolayers increased their permeability to water 10-fold. ²⁷ Implications of these results for patients with glaucoma who have depressed aqueous outflow function are encouraging but will depend ultimately on the net effect of AQP1 expression in TM and SC cells on outflow facility. 

 Supported in part by Grants F32EY0685, RO1EY01894, and P30EY05722 from the National Eye Institute; and by the American Health Assistance Foundation and Research to Prevent Blindness.

 Submitted for publication November 20, 2000; revised February 21, 2001; accepted March 19, 2001.

 Commercial relationships policy: N.

 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.

 

Diagram of recombinant AQP1 AV vectors. A first-generation AV vector backbone (E1/E3 deletions) was ligated with two AQP1 cDNAs oriented in either the sense or antisense direction (arrows). cDNAs were positioned behind the CMV promoter and in front of the bovine growth hormone polyA tail (bGHpA). Empty AV vector (not shown) contained no cDNA insert. TR, terminal repeat of AV; E2, E3, and E4, respective AV genes.

Figure 2.

 

Expression of recombinant AV AQP1 protein in human TM cells. Human TM cells were infected at an MOI of 1.0 (C) or as indicated in (A) or (B). Protein immunoblots were performed on whole-cell lysates (25 μg) of cells infected with AQP1(s) AV. (A) Expression of recombinant AQP1 protein in TM cells 4 days after infection (30-second exposure). (B) Expression of endogenous and recombinant AQP1 4 days after infection after extended exposure to film (30 minutes). (C) Time course (in days) for the expression of recombinant AQP1 (60-second exposure). Molecular weight markers are indicated to the left in kilodaltons.

Figure 3.

 

AV-directed recombinant AQP1 localized to plasma membrane of TM cells. Proteins at TM cell surfaces were biotinylated and isolated. Lane 1: the amount of AQP1 present in TM whole-cell lysates (25 μg) infected with AQP1(s) AV (MOI, 1.0, after biotinylation, before streptavidin-agarose); lane 2: the amount of AQP1 remaining in lysate after exposure to streptavidin-agarose; lane 3: the amount of AQP1 present in the fourth wash buffer of streptavidin-agarose; and lane 4: the amount of recombinant AQP1 that specifically bound to streptavidin-agarose. In control experiments using uninfected TM cells, visualization of endogenous AQP1 requires at least 10 minutes’ exposure time (rather than 30 seconds as shown in Figure 2 ). Molecular weight markers are indicated to the right in kilodaltons.

Figure 4.

 

AV-directed recombinant AQP1 functioned as a water channel. (A) Demonstrates the initial response of TM-cell monolayers infected with AV as indicated to a hypertonic challenge (100-mOsM gradient) over time. Shrinkage rate of cell monolayers was monitored as a function of change in forward light scatter by monolayer over time using a fluorometer (512/520 nm: incident versus scattered light). Comparisons of monolayer responses were obtained by fitting quadratic polynomials to scatter versus time curves (dV/dT) of initial response (first 10 seconds) to hypertonic challenge. Data are from one representative experiment of four total experiments. (B) The net movement of water across AV-infected MDCK monolayers (MOI, 10) in response to a hypertonic challenge (100-mOsM gradient). Data are from one representative experiment of six total experiments. Significant differences were determined by a Student’s t-test.

Figure 5.

 

AV-directed AQP1(as) RNA was expressed and knocked down native AQP1 protein. (A) Presence of AQP1(as) AV RNA in HEK293 cells 1 day after infection (MOI, 10.0). RNA was isolated from HEK293 cells infected with empty AV (lanes 1 and 2) and AQP1(as) AV (lanes 3 and 4). DNA copies were made of RNAs using reverse transcriptase and oligo dT primers (lanes 1 and 3) or a primer specific for AQP1(as) AV RNA (lanes 2 and 4). Shown are digitized negatives of ethidium-stained agarose gels showing PCR products for which primers designed to amplify a 355-bp fragment of AQP1(as) AV were used, with AQP1(as) AV cDNAs as templates. Lane 5: plasmid DNA encoding AQP1(as) [pSK/AQP1(as) AV] was used as a template (positive control). No template was used in lane 6 (negative control). Primer set spans two of AQP1’s three introns. No PCR product was observed when RNA was treated with RNase-A before cDNA reaction (not shown). (B) Effect of AQP1(as) AV on native AQP1 protein expression in TM cells 5 days after infection (MOI, 10.0). AQP1 protein expression was visualized by Western blot analysis of whole-cell lysates (50 μg) from AV-infected TM cells. Blots were digitized, and AQP1 protein expression was normalized to β-actin expression and quantified by densitometry. A representative blot probed with both anti-AQP1 IgG and anti-β-actin IgG is shown.

Figure 6.

 

SDS-PAGE of whole-cell lysates from AV-infected TM cells. TM cells were infected with either empty (lane E), AQP1(as) (lane A), or AQP1(s) (lane S) AV at an MOI of 0.1 and maintained in culture for 5 days before analyses. Left: Coomassie stain of TM cell proteins (whole-cell lysate, 50 μg) separated by SDS-PAGE. Right: duplicate of left, except proteins were transferred electrophoretically to nitrocellulose and analyzed for AQP1 expression by Western blot analysis. Blot was exposed to film for 30 minutes. Molecular weight markers are indicated on the left in kilodaltons.

Figure 7.

 

AQP1 protein expression affected mean resting volume of individual TM cells. TM cells were infected with empty AV (control), AQP1(s) AV or AQP1(as) AV at an MOI of 0.1. After 4 days, mean cell volumes were analyzed by electronic cell sizing on a cell counter, using at least 15,000 cells per data point and an orifice diameter of 140 μm. Absolute volumes (picoliters/cell) were calculated from distribution curves of cell diameter, using a standard curve generated by polystyrene latex beads of known diameter. Data are expressed as mean intracellular volume (±SEM) relative to the intracellular volume of empty-AV–infected cells. Shown are combined data from four independent experiments using three different TM cell strains (HTM22, 23, and 29). Significant differences: *sense or antisense versus empty and **sense versus antisense, by Students t-test.

Figure 8.

 

AQP1 protein expression affected resting volume of TM-cell monolayers. TM-cell monolayers were infected with empty AV (control), AQP1(s) AV or AQP(as) AV (MOI, 0.1), grown for 4 days, and assayed in an isotonic HEPES-buffered medium containing[ ¹⁴C]-urea or[ ¹⁴C]-sucrose. Intracellular volume was calculated as the difference between water space determined for[ ¹⁴C]-urea (a marker for intracellular plus trapped extracellular space) and [¹⁴C]-sucrose (a marker for trapped extracellular space). Data are expressed as mean intracellular volume (±SEM) relative to the intracellular volume of empty-AV–infected cells. Shown are combined data from six independent experiments, using two different TM cell strains (HTM26 and 29). Significant differences: *sense versus empty and **sense versus antisense, by Students t-test.

Figure 9.

 

Expression of AQP1 affected paracellular permeability of TM monolayers to [¹⁴C]-sucrose. Confluent TM-cell monolayers were infected with AV and were maintained in culture for 5 days before assay. Paracellular permeability of infected cell monolayers was determined by the flux of [¹⁴C]-sucrose from apical chamber to the basolateral chamber in 10 minutes. Data are expressed as mean counts per minute ± SEM of [¹⁴C]-sucrose that crossed TM-cell monolayers infected with AQP1(s) or AQP1(as), compared with the mean CPM of empty-AV–infected cells. Shown are combined data from three independent experiments. Significant differences: **sense versus empty and *sense versus antisense, by Students t-test.

The authors thank Neil Freedman, Brian McKay, Terete Borras, and Richard Whorton for their contributions to this project.