Effect of AQP1 Expression Level on CO₂ Permeability in Bovine Corneal Endothelium

Abstract

purpose. Corneal endothelial fluid transport is dependent on HCO₃ ⁻ and CO₂ fluxes. CO₂ permeability (Pco ₂) measurements in an oocyte expression system and in reconstituted proteoliposomes have suggested that the water channel AQP1 can transport CO₂. An AQP1 knockout mouse model, however, showed no evidence for CO₂ transport through AQP1 in erythrocytes or lung. Because HCO₃ ⁻ and CO₂ fluxes are essential to endothelial function, the current study was conducted to determine whether AQP1 expression levels in confluent cultures of bovine corneal endothelial cells (BCECs) affects membrane Pco ₂.

methods. BCEC endogenous AQP1 expression was reduced by antisense oligonucleotide (AO) transfection or adenoviral antisense-AQP1 (AV) infection. AQP1 was overexpressed by adenoviral sense-AQP1 (SV) infection, which directs expression of recombinant AQP1.

results. Expression of AQP1 and osmotic water permeability (control P _f = 0.046 ± 0.005 cm/sec) were reduced 45% and 36.5%, respectively, by AO transfection and reduced 67% and 49%, respectively, by AV infection. SV infection induced a more than threefold overexpression of AQP1 but showed only a 37% increase in P _f. Adenoviral empty virus (EV) infection did not change AQP1 expression or P _f. Pco ₂ was determined by measuring the rate of intracellular pH decrease after exposure to CO₂/HCO₃ ⁻-rich solutions, as measured by the pH-sensitive fluorescent dye 2′,7′-bis-(2-carboxyethyl)-5-(and-6)-carboxyfluorescein (BCECF). Apparent Pco ₂ of BCEC (0.0036 ± 0.00023 cm/sec) was not different among control, oligonucleotide-transfected, and adenoviral-infected cells. P _f could also be reduced more than 50% by 3 to 5 minutes’ exposure of control cells to 0.5 mM p-chloromercuriphenylsulfonic acid (pCMBS), but this had no effect on rates of intracellular pH decrease.

conclusions. AQP1 does not contribute to Pco ₂ in corneal endothelial cells.

Expression of the water channel protein aquaporin-1 (AQP1) confers significant increases in membrane osmotic water permeability (P _f; see ¹ for review). AQP1 and other aquaporins are often highly expressed in epithelial cells involved in large amounts of osmotically coupled water transport (e.g., the kidney ² ) and active salt pumping (e.g., salivary gland ³ ) and AQP1 knockout mice ⁴ show significant reductions in P _f and tissue water fluxes. In contrast, absence of aquaporins in lung alveoli reduces P _f but has no effect on isotonic fluid transport, indicating that epithelia involved in small levels of fluid transport may not require the high P _f provided by aquaporins. ^{5 6} Furthermore, aquaporins can be highly expressed in cell types with no known water transport function. ⁷ Thus, there may be other undiscovered functions of membrane aquaporin expression. 

Small solutes are excluded from AQP1; however, recent reports indicate that AQP1 may be permeable to CO₂. CO₂ permeability (P co ₂) was increased approximately 40% in oocytes expressing AQP1. ^{8 9} Further, proteoliposomes reconstituted with AQP1 were reported to have four times higher P co _2. ¹⁰ These reports have suggested that increased P co ₂ may be particularly advantageous for those cells involved in rapid CO₂ exchange, such as erythrocytes, lung alveoli, renal proximal tubule, choroid plexus, and ciliary epithelium. However, a recent study using AQP1 knockout mice showed no significant change in erythrocyte or alveolar P co ₂ and did not demonstrate an effect of AQP1 on P co ₂ in proteoliposomes. ¹¹  

Using an antisense transfection approach to alter AQP1 expression, we examined whether AQP1 is permeable to CO₂ in confluent cultures of bovine corneal endothelial cells (BCECs). AQP1, but not AQP2 through 5, ¹² is highly expressed in the corneal endothelium ^{13 14} and confers a relatively high P _f. Ion-coupled fluid transport by the corneal endothelium maintains corneal hydration and transparency, but at 6 μl/cm² per hour, its secretion is at a relatively low level. ^{15 16} Endothelial fluid transport is dependent on the presence of HCO₃ ⁻ and is significantly slowed by carbonic anhydrase inhibitors. ^{15 16 17 18 19} In addition to strong expression of cytosolic carbonic anhydrase (CAII), ²⁰ corneal endothelium also expresses the membrane-bound carbonic anhydrase (CAIV) on its surface. ²¹ A recent model of HCO₃ ⁻ transport across the apical membrane of corneal endothelium includes CO₂ flux from intracellular to extracellular compartments with conversion of CO₂ to HCO₃ ⁻ catalyzed by CAIV. ²² Thus, if AQP1 were permeable to CO₂, it could have an important role in vectorial transport of HCO₃ ⁻ across the corneal endothelium. In this study, we show that reduction in endogenous AQP1 expression or overexpression of AQP1 significantly alters osmotic water permeability in cultured BCEC but has no effect on P co ₂. 

Materials and Methods

Cell Culture

Bovine corneal endothelial cells (BCECs) were cultured to confluence as previously described. ²³ Briefly, primary cultures from fresh cow eyes were established in T-25 flasks with 3 ml of Dulbecco’s modified eagle’s medium (DMEM) and 10% bovine calf serum gassed with 5% CO₂ and 95% air at 37°C and fed every 2 to 3 days. Primary cultures were subcultured to three T-25 flasks and grown to confluence in 5 to 7 days. The resultant second-passage cultures were then further subcultured onto coverslips and used for the experiments described herein. 

Antisense Oligonucleotides

Antisense oligodeoxynucleotides (AOs), 5′-CAGCCCTCCAGAAGAGCTTCTTCTT-3′, complementary to bases +16 to +40 of AQP1 mRNA were used to reduce expression in cultured BCECs. As a control, sense oligodeoxynucleotides (SOs; +14 to +38, 5′-TCAAGAAGAAGCTCTTCTGGAGGGC-3′) were also used. Computer searches of GenBank and EMBL (Vector NTI ver 5.2 software; InforMax; North Bethesda, MD) showed no significant homologies to other sequences. The first nine bases from the 5′ end and the penultimate seven bases from the 3′-end of both oligonucleotides were phosphorothioated (Gibco Life Technology, Grand Island, NY). In preparation for immunoblot analysis, cells were grown in six-well plates. For physiological measurements cells were grown on glass coverslips. When BCEC cultures were 60% to 80% confluent (1 day after seeding), 5μ g/ml oligonucleotides was applied in complete culture medium. Culture media were changed on days 3 and 5 and fresh oligonucleotides were added. On day 7, cells were collected for protein extraction. Cell-coated coverslips were used for measurement of osmotic water P _f and separate coverslips for measurement of P co ₂. 

AQP1 Adenoviruses

The adenovirus (AdV) backbone for the AQP1 sense and antisense AdV constructs was a replication-deficient first-generation AdV with deletions of the E1 and E3 genes. ²⁴ This empty AdV (EV) 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. ²⁵  

Two AQP1 recombinant viruses were constructed by a unique method. For AdV sense-AQP1 (SV), a plasmid containing the coding sequence for AQP1, pCHIPev, ²⁶ was digested and subcloned into the shuttle vector pSKAC, creating pSKAC/AQP1. pSKAC contains map units 0.0 to 1.3 of the AdV, which includes the left terminal repeat of AdV, a CMV promoter, an AMV translation enhancer, and a polylinker region. For AdV antisense-AQP1 (AV), an 807-bp DNA fragment encoding antisense AQP1 was generated by polymerase chain reaction using pCHIPev as a template and primers that delineated the full coding sequence of AQP1, as described previously. ²⁵ The DNA fragment was subcloned into pSKAC, creating pSKAC/AQP1as. DNA fragments containing AQP1 sense or antisense DNA were liberated from pSKAC after restriction and ligated into adenovirus, as described previously. ²⁵ Human embryonic kidney cells (strain 293) 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. ²⁵ Individual viruses were amplified in the 293 cells and purified over cesium step gradient. Individual AdV DNA titers were determined by three methods: 1) plaque titration on 293 cells, 2) immunofluorescence microscopy of AdV protein expression (anti-penton group antigen, clone 143; Biodesign, Kennebunk, ME) in 293 cells infected with serial dilutions of AdV; and 3) absorbance at 260 nm (plaque-forming units [pfu]/ml = A ₂₆₀ × dilution × 10¹⁰). 

Adenoviral Infection of BCECs

The three adenoviruses EV, AV, and SV were used to infect BCECs at a multiplicity of infection of 10. Cultured BCECs grown to 80% to 90% confluence on six-well tissue culture plates or coverslips were exposed to virus in complete culture media. On days 3 and 5, culture media (without virus) were replenished. On day 7, the cells from six-well plates were collected for extraction of protein to perform Western blot analysis. Coverslips were used for osmotic water P _f or P co ₂. 

Immunoblot Analysis

Cultured BCECs were dissolved directly in sample buffer, sonicated (model 250 sonicator; Branson, Danbury, CT) briefly on ice, and centrifuged at 6000g for 5 to 10 minutes. An aliquot of the supernatant was taken for protein assay using the Lowry method (Bio-Rad, Richmond, CA). β-Mercaptoethanol (5%) and bromphenol blue were added to the remainder of the supernatant. The samples were boiled for 5 minutes and applied to a 10% polyacrylamide gel with a 4.5% stacking gel. Proteins were transferred to a polyvinylidene difluoride (PVDF) membrane overnight at 4°C. Membranes were incubated in phosphate-buffered saline (PBS) containing 5% nonfat dry milk for 1 hour at room temperature. The blots were then incubated with rabbit polyclonal AQP1 antibody (1:200; Alpha Diagnostics; San Antonio, TX), in PBS containing 5% nonfat dry milk for 1 hour at room temperature with shaking. Next, the blots were washed five times for 5 minutes each with PBS/Tween-20, incubated with goat anti-rabbit secondary antibody coupled to horseradish peroxidase (Sigma, St. Louis, MO) for 1 hour at room temperature, washed with PBS/Tween-20 five times for 5 minutes each, and developed by enhanced chemiluminescence (ECL). Films were scanned to produce digital images, and the density of the bands was estimated using Un-Scan-It software (Silk Scientific, Orem, UT). 

Determination of Osmotic P _f

Relative changes in cell volume of confluent BCECs cultured on glass coverslips were determined by a fluorescence-quenching technique. ²⁷ Briefly, collisional quenching of a halide sensitive quinolinium fluorescent dye by intracellular quenchers is reduced during cell swelling and increased during cell shrinkage. Strict volume sensitivity is obtained by perfusion in the absence of halides. ²⁷ Cells were loaded with 6-methoxy-N-ethylquinolinium (MEQ) by exposure to 10 μM 6-methoxy-N-ethyl-1,2-dihydroquinoline (DiH-MEQ; Molecular Probes, Eugene, OR) ²⁸ for 10 minutes followed by 20 minutes’ washing at room temperature. Coverslips (25 mm in diameter) were clamped into a microscope perfusion chamber that was modified from that previously described. ²³ Briefly, the chamber formed a perfusion slot (3.5 × 16 × 1.5 mm deep) with the clamped coverslip at the bottom and the slot left open at the top. One end of the slot was fit with 23-gauge stainless steel tubing connected to PharMed tubing (Fisher Scientific; Fairlawn, NJ). Ringer solution was delivered to the chamber periodically for 7 seconds through glass syringes at 0.5 ml/sec and rapidly removed at the opposite end of the perfusion slot by suction. Cells were viewed with a ×40 (0.75 numeric aperture) objective (Zeiss, Thornwood, NY) on the stage of an inverted microscope (Diaphot; Nikon, Melville, NY). Fluorescence excitation (365 ± 10 nm) and data collection (50 sec⁻¹) were obtained using a DeltaRam PTI fluorescence system controlled by Felix software (Photon Technology International, Monmouth Junction, NJ). The composition of the halide-free Ringer solution used was (in millimolar): 150 Na⁺, 4 K⁺, 0.6 Mg²⁺, 1.4 Ca²⁺, 146 NO₃ ⁻, 1 HPO₄ ⁻, 10 HEPES⁻, 2 gluconate⁻, and 5 glucose, with pH adjusted to 7.50 at 22°C. Osmolality was adjusted to 300 mOsm/kg with sucrose. Hypotonic solutions were obtained by 33% to 50% dilution with nanopure water. 

After exposure to hypotonic solutions, fluorescence increased and reached a new steady state within 20 seconds. These data were smoothed (Savitzky–Golay), and the initial slope (dF/dt) determined using Felix analysis software. P _f was calculated by  

\[P_{f}{=}\ \frac{dF{/}{[}dt\ {\cdot}\ (F_{I}{-}F_{b}){]}}{(S{/}V)\ {\cdot}\ V_{w}\ {\cdot}\ dC}\]

where P _f is the osmotic water permeability coefficient (in centimeters per second), dF/dt is the initial slope of fluorescence change, F _I is the baseline fluorescence level, F _b is the background fluorescence determined at the end of each experiment by complete quenching of MEQ fluorescence with 150 mM thiocyanate, S/V is the surface-to-volume ratio assumed to be 2500 cm for 4-μm-thick confluent cells, V _w is the partial molar volume of water (18 cm³/mole), and ΔC is the difference in osmolality between solutions. 

Determination of Pco ₂

BCEC P co ₂ was determined by measuring the rate of intracellular pH (pH_i) change after exposure to CO₂. The same perfusion setup as was used for P _f was used for determination of P co ₂. Solutions were Na⁺-free to remove any effect of the Na⁺/HCO₃ ⁻ cotransporter or Na⁺/H⁺ exchanger on pH_i. ²³ HCO₃ ⁻-rich Ringer contained (in millimolar): 80 K⁺, 70 N-methyl-d-glucamine (NMDG⁺), 0.6 Mg²⁺, 1.4 Ca²⁺, 118 Cl⁻, 28.5 HCO₃ ⁻, 1 HPO₄ ⁻, 10 HEPES⁻, 2 gluconate⁻, and 5 glucose, equilibrated with 5% CO₂ and pH adjusted to 7.50 at 22°C. For HCO₃ ⁻-free solutions, KHCO₃ was replaced by K-gluconate, and the solution was equilibrated with air. Confluent BCECs cultured on glass coverslips were loaded with the pH-sensitive fluorescent dye 2′,7′-bis-(2-carboxyethyl)-5-(and-6)-carboxyfluorescein (BCECF), by incubation in an Na⁺-free, HCO₃ ⁻-free Ringer solution containing 5 μM of the acetoxymethyl ester of BCECF (BCECF-AM; Molecular Probes) at room temperature for 30 to 60 minutes. Dye-loaded cells were then kept in Ringer for at least 30 minutes before use. Fluorescence ratios (495 ± 10 and 440 ± 10 nm excitation) were obtained at 50 sec⁻¹ and were calibrated against pH_i by the high K⁺-nigericin technique. ²⁹ A calibration curve that follows a pH titration equation has been constructed for BCECs. ²³  

After exposure to the CO₂/HCO₃ ⁻ solution, pH_i rapidly decreased (half-life ∼0.5 seconds) and reached a new steady state in 4 to 5 seconds. These data were smoothed (Savitzky–Golay) and the initial slope (dpH_i/dt) determined using analysis software (Felix). The apparent H⁺ influx rate, JH⁺, due to dissociation of CO₂ and formation of H₂CO₃ is (dpH_i/dt) · β · V/S, where β is the intrinsic buffering capacity of BCECs (10 mM/pH ²³ ) and V/S is the volume-to-surface ratio. The fraction of CO₂ entering the cell that forms H₂CO₃ is K/(H + K), where H is the intracellular H⁺ concentration (steady state pH_i in the presence of CO₂) and K is the apparent dissociation constant (7.24 × 10⁻⁷). Therefore, CO₂ influx (JCO₂) is JH⁺ · [(H + K)/K], ⁹ and  

\[P\mbox{\textsc{co}}_{2}{=}\ \frac{J\mbox{\textsc{co}}_{2}}{{[}\mathrm{CO}_{2}{]}}\]

where [CO₂] is 1.36 mM (0.036 mM/mm Hg × 37.7 mm Hg). 

Results

AQP1 Expression

Antisense oligonucleotides were first used in an attempt to reduce AQP1 expression in cultured BCECs. Oligonucleotide concentrations higher than 10 μg/ml caused an increase in dead floating cells, reduced cell proliferation, and reduced cell density after 7 days of culture compared with control. Cultures exposed to 5 μg/ml of oligonucleotides were morphologically indistinguishable from control cultures. Furthermore, fluorescence levels after MEQ or BCECF loading were in the same range as control cultures, indicating cell viability and membrane integrity. Therefore, in all oligonucleotide experiments reported, 5 μg/ml was used. In pilot experiments (not shown), visual inspection of BCEC cultures after 7-day exposure to Texas red–labeled oligonucleotides indicated that approximately 99% of cells had been transfected. Figure 1A shows an immunoblot for AQP1 after sense and antisense oligonucleotide exposure. To determine the approximate percentage of decrease, the densities of the native band (28 kDa), and the glycosylated band (∼45 kDa) were summed. In the experiment shown in Figure 1A , antisense exposure reduced AQP1 expression to approximately 60% of control (by densitometry), whereas parallel incubation with sense oligonucleotides produced an approximately 11% reduction. In a separate experiment (not shown), immunoblots indicated approximately 50% antisense oligonucleotide decrease. 

In an attempt to obtain a greater reduction in AQP1 expression, adenoviral antisense infection was used. As in oligonucleotide-transfected cells, cell density after 7 days’ culture after viral infection and fluorophore loading were indistinguishable from control. Figure 1B shows quantitative immunoblot results indicating approximately 50% reduction in AQP1 expression 4 days after antisense virus exposure and a further reduction to approximately 40% of control after 7 days. Conversely, 7 days after exposure to AQP1-sense adenovirus, which directs expression of recombinant AQP1, AQP1 expression was increased approximately 3.5-fold over control. EV-exposed cells showed no significant change relative to control. 

BCEC Permeability Measurements

Transfected BCECs were first characterized to show that the altered expression in AQP1 was concomitant with P _f. Figure 2A shows representative data of relative volume changes (F − F _b/F _I− F _b) for each condition. In each case, 20 seconds of data are shown, starting at the time when MEQ fluorescence began to increase on exposure to hypotonic solution. A new steady state level of fluorescence was achieved in approximately 10 seconds for SV-infected cells; approximately 15 seconds for control, EV-infected, and SO-transfected cells; and between 25 and 35 seconds for AO-transfected and AV-infected cells. The averaged P _f for each condition is summarized in Figure 2B . AO transfection and AV infection showed 36.5% and 49% reduction in P _f, respectively, relative to control cells (P _f = 0.046 cm/sec). SO-transfected and EV-infected cells had P _f that was not significantly different from control. However, SV infected cells showed a small (37%) but significant increase in P _f relative to EV-infected cells. 

P co ₂was determined from the initial rate of pH_i decrease after exposure to a known concentration of CO₂, corrected for the proportion of CO₂ that dissociates to H₂CO₃ (see the Methods section). Figure 3A shows a typical experiment using control cells. Cells were bathed in bicarbonate-free Ringer, which was then rapidly exchanged with CO₂/HCO₃ ⁻-rich solution. After a new steady state pH_i is reached (5–7 seconds), bicarbonate-free Ringer is reintroduced. This sequence is repeated at least three times for each experiment, and the initial dpH_i/dt is calculated from the pH_i decrease trial and averaged for that experiment. Figure 3B shows three examples (control and EV- and AV-infected cells) of exponential decreases in pH_i after CO₂ exposure on an expanded time scale. Note that the steady state pH_i varied among the experiments. In normal Na⁺-containing Ringer, pH_i ranges from 7.0 to 7.5. In the high-K⁺, 0-Na⁺ Ringer used in these experiments, the beginning pH_i varied from 6.6 to 7.5. The initial dpH_i/dt values of these three examples were 0.375, 0.50, and 0.54 pH/sec for control and EV- and AV-infected cells, respectively. However, after correction for the final steady state pH_i, the P co ₂ for all three examples was approximately 3.35 × 10^{− 3} cm/sec. In four experiments using control cells, addition of 2 mg/ml carbonic anhydrase, which could enhance CO₂ flux by reducing the effects of unstirred layers, ¹¹ had no effect. Figure 3C shows the averaged P co ₂ for all experimental conditions and indicates that there was no effect of AQP1 expression on P co ₂. 

AQP1 water permeability can be inhibited by mercurials, which were also shown to impede P co ₂ in oocytes. ^{9 10} p-Chloromercuriphenylsulfonic acid (pCMBS) has been shown to reduce P _f in corneal endothelial cells, whereas HgCl₂ was not used because it quickly kills BCECs. ¹³ Figure 4A shows that MEQ fluorescence changed in response to a 33% hypotonic challenge. The cells were then bathed in isotonic Ringer with 0.5 mM pCMBS for 3 minutes and again exposed to hypotonic Ringer in the presence of pCMBS. The initial dF/dt was reduced 58% in pCMBS (mean reduction, 56% ± 11%, n = 4). pCMBS was then removed, and cells were bathed in Ringer containing 5 mM dithiothreitol (DTT) for 10 minutes, in an attempt to reduce the sulfhydryl groups of remaining pCMBS. ¹³ This caused an approximately 10% decrease in fluorescence that reversed after a 5-minute wash in isotonic Ringer. Repeat hypotonic challenge showed approximately 75% reversal of the pCMBS inhibition. Although pCMBS can inhibit water permeability, we could not demonstrate an effect on P co ₂. Figure 4B shows pH_i measurements in BCECs before and after pCMBS exposure. The initial dpH_i/dt for control and pCMBS-treated cells were 0.55 ± 0.03 and 0.53 ± 0.04 pH/sec, respectively, which was not significantly different by paired t-test (n = 5, P > 0.05). 

Discussion

In this study we attempted to show whether AQP1 expression in cultured BCECs could affect P co ₂. Corneal endothelial ion-coupled fluid transport is dependent on the presence of HCO₃ ⁻ and is significantly reduced by carbonic anhydrase inhibitors (CAIs). ^{15 17 18 19} Basolateral-to-apical HCO₃ ⁻ flux gives rise to a small (0.5-mV) apical-side negative transepithelial potential, which is also reduced by CAIs. ¹⁶ HCO₃ ⁻ influx at the basolateral membrane is through Na/HCO₃ ⁻ cotransport (pNBC). ³⁰ HCO₃ ⁻ efflux across the apical membrane, however, appears to have a significant component that is due to CO₂ flux. The apparent apical CO₂ flux (as measured by changes in pH_i) is facilitated by both cytosolic and membrane-bound carbonic anhydrase. ²² Thus, AQP1 permeability to CO₂ could play a significant role in CO₂/HCO₃ ⁻ transport in BCEC. Further support for this notion comes from the fact that BCECs strongly express AQP1, ^{12 14} yet the fluid transport rate across corneal endothelia is relatively low, indicating that functions other than contributions to rapid water flux may be associated with AQP1. However, despite the approximately 2.5-fold difference in P _f among AQP1-overexpressing, AQP1-basally expressing, AQP1-underexpressing, and pCMBS-treated BCECs, no differences in P co ₂ was elicited. 

AQP1 expression in BCECs was altered using antisense oligonucleotides or CMV-promoted expression of AQP1 sense and antisense mRNA. The oligonucleotide approach was reasonably effective, inhibiting approximately 45% of AQP1 expression and leading to a decrease in P _f of 37%. The adenoviral approach was more effective, inhibiting approximately 60% of AQP1 expression and reducing P _f by 49%. This is a reasonable relation between reduced AQP1 expression and reduced P _f, because P _f includes both AQP1-mediated water flux as well as non-AQP1 mediated flux (e.g., through the naked membrane). This has been demonstrated previously, for example, in heterozygous AQP1 null mice, in which red blood cell AQP1 protein expression was reduced by 20%, and P _f was reduced by approximately 10%. ¹¹ By contrast, SV infection increased AQP1 expression more than threefold; however P _f increased by only 37%. AQP1 expression in control BCECs is exclusively at the apical and basolateral plasma membranes. ¹⁴ Considering that AQP1 membrane protein density can be very high, ¹ it may be that in cells overexpressing AQP1, trafficking to the outer membranes is inefficient. Further, functionality of human AQP1 in bovine cells may be compromised. Regardless, the difference in functional range in AQP1 expression between SV- and AV-infected cells was approximately 2.5-fold. 

The calculated P _f of control cells was 0.046 cm/sec, giving a density of approximately 7500 AQP1 monomers/μm². This P _f is slightly higher than that reported for erythrocytes (0.018 cm/sec) ¹¹ and airspace-capillary barrier (0.022 cm/sec), ⁵ but almost four times lower than that reported in renal proximal tubules (0.15 cm/sec). ³¹ The previous estimate of P _f for cultured BCECs by Echevarria et al. ¹³ was approximately 0.0093 cm/sec. One significant difference between the studies is that Echevarria et al. used trypsinized BCECs that had settled onto coverslips for 4 hours. Thus, there was not a confluent monolayer. The calculation of P _f by Echevarria et al. ¹³ used the entire cell membrane, giving an S/V ratio that is approximately five times greater than that used in the present study. As they suggest, because water would enter the cells primarily through the apical membranes, they may have underestimated P _f. It is remarkable that the differences in P _f are completely explained by the choice of the S/V ratio. Considering that water flux may also occur across lateral membranes through the leaky intercellular junctions, then the current estimate of P _f could be considered the upper limit. Furthermore, our results show that accurate P _f estimates can be obtained with confluent cultures using the fluorescence quenching technique. 

P co ₂ in BCECs was measured using the rate of pH_i decrease after exposure to CO₂, as measured by the fluorescence ratio of BCECF. The experiments were performed in the absence of Na⁺, to block any pHi regulatory effects from the Na/HCO₃ ⁻ cotransporter or the Na⁺/H⁺ exchanger. Chloride was left in the solutions, because the Cl⁻/HCO₃ ⁻ exchanger is not expressed in cultured BCECs. ³² That pH_i regulation was fully blocked is apparent from Figure 3A , which shows no increase in pH_i over 20 seconds during exposure to CO₂/HCO₃ ⁻ solutions. The apparent Pco ₂ of cultured confluent BCEC was approximately 0.0036 cm/sec. This value is similar to that measured in Xenopus oocytes expressing AQP1 ⁹ and approximately 3× less than that measured in erythrocytes using a rapid stopped-flow procedure. ¹¹ In comparison with bilayer studies performed in the presence of carbonic anhydrase and high buffering capacity (P co ₂ = 0.35 cm/sec), ³³ the values reported for cells are low, which is probably due to unstirred layers, intracellular barriers to diffusion, and/or carbonic anhydrase kinetics. 

Given the density of AQP1 monomers estimated for BCECs from the measured P _f, it is possible to predict the cellular P co ₂ from estimates of single-channel P co ₂. The single-channel P co ₂ (2 × 10^{− 14} cm³/sec) ¹¹ for AQP1 estimated from oocyte data ⁹ would predict a control P co ₂ in BCECs that is approximately four times that measured. The single-channel AQP1 P co ₂ (5 × 10^{− 15} cm³/sec) ¹¹ estimated from the proteoliposome data of Prasad et al. ¹⁰ predicts a P co ₂ of 0.0038 cm/sec in BCECs, similar to that found. However, the approximately 50% decrease in P _f in antisense-treated BCECs should have reduced P co ₂ by 0.0019 cm/sec, a change that would have been easily detected by the current method. On the contrary, using the upper limit on single-channel AQP1 P co ₂ estimated for erythrocytes (3 × 10^{− 16}/cm³), ¹¹ a value that suggests little CO₂ transport by AQP1, a 50% decrease in AQP1 expression in BCECs would cause a 0.0001 cm/sec decrease in P co ₂, a change that falls within the SE of our data. This demonstrates that the current findings are consistent with those reported by Yang et al., ¹¹ indicating that AQP1 P co ₂ is insignificant. 

In AQP1-transfected oocyte studies, 15 minutes’ exposure to 1 mM pCMBS slowed CO₂ fluxes; however, reversal with a reducing agent was not shown. ⁹ In addition, a mercurial insensitive AQP1 mutant that retains water permeability showed increased P co ₂ and no effect of pCMBS. ⁹ This indicates that pCMBS was not toxic to the oocytes; however, with BCECs the pCMBS dosage and time used for oocytes would sometimes prove toxic, as evidenced by accelerated losses of BCECF fluorescence. In contrast, 3 to 5 minutes’ exposure to 0.5 mM pCMBS, which reduced P _f by more than 50% and did not reduce P co ₂, was well tolerated by BCECs. Prasad et al. ¹⁰ measured increased rates of pH change in proteoliposomes when reconstituted with AQP1, which was eliminated in the presence of 1 mM HgCl₂. Yang et al. ¹¹ were unable to measure an effect of AQP1 reconstitution on proteoliposome P co ₂, and their study suggests that HgCl₂ may inhibit carbonic anhydrase. 

In summary, antisense oligonucleotide transfection and adenoviral infection of sense and antisense AQP1 led to significant changes in cultured BCEC AQP1 expression and osmotic water permeability. Measurements of P co ₂ in untransfected cells versus transfected cells as well as pCMBS-treated cells, show that AQP1-dependent P _f in BCECs does not significantly affect P co ₂. 

 Supported by Grant EY08834 from the National Institutes of Health.

 Submitted for publication August 9, 2000; revised October 4, 2000; accepted October 16, 2000.

 Commercial relationships policy: N.

 

Figure 1.

 

Immunoblot analysis of BCECs probed with polyclonal AQP1 antibody. (A) Oligodeoxynucletide transfection. Lanes 1 and 2: untreated cells (control); lanes 3 and 4: cells exposed to sense and antisense oligonucleotides for 7 days (7d). (B) Effect of adenoviral transfection with AV, SV, and EV. All lanes (A, B) contained 60μ g of protein.

Figure 2.

 

P _f of BCECs from control and transfected cells measured by MEQ fluorescence quenching. (A) Relative volume (F − F _b/F _I− F _b). Data are arranged in descending order of rate of volume change (CON, control). Cells were bathed in 300 mOsm isotonic Ringer. Beginning data indicate exposure to Ringer diluted by 50%. (B) P _f averages ± SE; number of experiments is indicated above each bar; *AO significantly different from CON (P < 0.05); #SV and AV significantly different from EV.

Figure 3.

 

Pco ₂ of cultured BCECs. (A) Control BCECs were loaded with the pH-sensitive fluorescent dye BCECF initially bathed in a CO₂/HCO₃ ⁻-free Ringer. Down arrows: Exposure to CO₂/HCO₃ ⁻-rich Ringer; up arrows: return to CO₂/HCO₃ ⁻-free Ringer. (B) Expanded section of pH_i decrease after exposure to CO₂ for control (CON) and EV- and AV-transfected cells. Calculated P co ₂ (1.85 × 10^{− 3} cm/sec) was similar for all three. (C) Pco ₂ average ± SE; number of experiments is indicated above each bar.

Figure 4.

 

Effects of pCMBS on P _f and CO₂ P _f. (A) Cells were loaded with MEQ and the bathing solution was switched from isotonic (300 mOSM) to 33% diluted Ringer, indicated by the hatched boxes. After the control challenge, cells were exposed to 0.5 mM pCMBS for 3 minutes and challenged again with hypotonic Ringer. pCMBS was then removed and cells were bathed in 5 mM DTT for 10 minutes, followed by a 5-minute wash in isotonic Ringer and then challenged with hypotonic Ringer. (B) pH_i changes due to CO₂ exposure before and after exposure to pCMBS. Down arrows: Exposure to CO₂/HCO₃ ⁻-rich Ringer; up arrows: return to CO₂/HCO₃ ⁻-free Ringer. At the break in the data, cells were pre-exposed to 0.5 mM pCMBS for 3 minutes. Two subsequent CO₂ exposures in the continued presence of pCMBS were similar to control.