Spatial expression of aquaporin 5 in mammalian cornea and lens, and regulation of Its localization by phosphokinase A

Animals

Wild type (WT; C57BL/6) and AQP5 knockout (AQP5-KO [33]) mice, and New Zealand white rabbits were used. Eyes were enucleated and transferred to mammalian physiological saline. Corneas and lenses were dissected out under a microscope. All experimental and animal care procedures were performed according to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research, and were approved by the State University of New York at Stony Book Institutional Animal Care and Use Committee.

RT-PCR, cloning and sequencing

Total RNA was extracted separately from mouse lacrimal gland, whole cornea, and lens anterior epithelial cells and fiber cells, using RNA STAT-60 (TEL-TEST Inc., Friendswood, TX). Following the manufacturer's instructions [5], RT-PCR was carried out in a Gradient Thermal Cycler (Stratagene) using AQP5-, AQP0-, or AQP1-specific primers (Table 1) and the products were fractionated on 1.2% agarose gels, stained with ethidium bromide and viewed over UV. Images were captured and digitized using Kodak Image Station [34]. cDNAs amplified from epithelial and fiber RNAs were cloned between the EcoRI and BamHI restriction sites of pIRES2-EGFP vector (Clontech, Mountain View, CA). Transformation was carried out using MAX Efficiency® DH5α™ E. coli competent cells (Invitrogen, Life Technologies, Grand Island, NY). Positive clones were selected using kanamycin-Luria-Bertani (LB) agar plates. Sequences of extracted DNAs were verified by automated sequencing using fluorescent dye terminators (Stony Brook University, DNA Sequencing Facility, Stony Brook, NY).

Western blotting, dephosphorylation, and immunostaining

Total membrane proteins extracted from whole cornea and lens of rabbit and mouse, as well as from mouse lacrimal gland (positive control) and different regions of lens [4] were used for western blotting [34,35]. In short, cornea, lens epithelium, lens cortex and lens nucleus were dissected out under ice-cold physiological saline. Tissues were homogenized using 1× gel loading buffer (Invitrogen, Life Technologies) in 25 mM Tris-HCl (pH 7.2) with proteinase inhibitor cocktail (Sigma) and/or phosphatase inhibitor ( Roche Diagnostics Corporation, Indianapolis, IN). The homogenates were spun at 45,000× g for 30 min and the membrane pellets were washed two times with the homogenizing buffer. The membrane pellets were homogenized in 50–100 μl of protein extraction buffer (2% SDS) and sonicated for 1 min with short bursts. The homogenates were centrifuged at 12,000× g for 30 min, and the supernatants were mixed with NuPAGE LDS sample buffer (Invitrogen) and fractionated using 10% NuPAGE Novex Bis-Tris gel with MOPS-SDS as the running buffer (Invitrogen). Proteins from the denaturing gels were transferred to PVDF nylon membranes and exposed to COOH-terminal-specific anti-AQP5, anti-AQP1 or anti-AQP0 antibodies (Chemicon, Billerica, MA or Santa Cruz Biotechnology, Inc., Santa Cruz, CA) as appropriate. Alkaline phosphatase conjugated secondary antibodies were used and detection of antibody binding was carried out using alkaline phosphatase detection kit (Vector Laboratories, Burlingame, CA).

In dephosphorylation studies, calf intestinal alkaline phosphatase (25 units) was added to ~25 μg of purified membrane protein samples and incubated for 60 min at 37 °C. Reactions were stopped with 50 mM EDTA and analyzed by western blotting as described above [34].

To investigate the spatial expression of different AQPs, eyes of WT and AQP5-KO mice were fixed in 4% paraformaldehyde, cryosectioned, and immunostained with anti-AQP5 (Chemicon, Billerica, MA or Epitomics, Inc., Burlingame, CA), anti-AQP1 or anti-AQP0 antibodies as appropriate, and imaged using a Zeiss confocal microscope as described [5]. Briefly, cornea and in vitro culture cells were fixed in 4% paraformaldehyde in phosphate-buffered saline (PBS) for 30 min; lenses were fixed for 12 h. Fixed corneas and lenses were cryoprotected, and sectioned at 10-16 μm thickness using a cryomicrotome (Leica) and stored at −20 °C. Cryosections, and fixed culture cells were permeabilized (0.2% [v/v] Triton X-100) for 20 min at room temperature, blocked with normal goat serum and treated with rabbit polyclonal antibody raised against AQP5, AQP0, or AQP1, at a dilution of 1:200 with 5% (w/v) bovine serum albumin (BSA) in PBS. After overnight incubation and three washes in PBS, the slides were exposed to fluorescein 5′-isothiocyanate (FITC) or Texas Red conjugated goat anti-rabbit IgG in PBS containing 5% BSA. The slides with the treated tissue sections or cells were washed again in PBS, mounted in anti-fade Vectamount (Vector Laboratories, Burlingame, CA) containing nuclear stain DAPI and viewed. Optimized Z-sectional digital images were acquired and deconvolved using Zeiss AxioVision software Version 5. Representative images are given.

Effects of cAMP and H-89 on AQP5 expression, trafficking, and localization

Mouse AQP5 expression construct (AQP5-pIRES2-EGFP) was transfected into Madin Darby Canine Kidney (MDCK) cells. MDCK cells were used for in vitro studies on AQP5 expression, trafficking, and localization because these cells do not express AQP5 and are epithelial cells, like the corneal and lens epithelial cells with apical and basolateral sides. In short, MDCK cells (American Type Culture Collection, Manassas, VA) were grown in Minimum Essential Medium (Invitrogen, Life Technologies, Grand Island, NY) that was supplemented with 5% heat-inactivated fetal calf serum (Hyclone Laboratories, Inc., UT) and 100 U/ml penicillin and 0.1 mg/ml streptomycin. Cell cultures were maintained at 37 °C, under 5% CO₂ in a humidified atmosphere. Transfections were carried out using Effectene reagent (Qiagen), following the manufacturer's recommendations. Stable cell lines were selected using Geneticin® (G418; A.G. Scientific, Inc., CA) Sulfate (400 mg/ml media).

Stable cells expressing AQP5 were exposed to PKA agonist mp-cAMP (membrane permeable cyclic AMP, 100 mM; Sigma-Aldrich, St. Louis, MO) or PKA antagonist H-89 (20 mM; Sigma-Aldrich) for 30 min. In another experiment, cells were incubated first with the H-89 for 30 min, and then stimulated with mp-cAMP for the same duration. Cells were washed in PBS, fixed in 4% paraformaldehyde and immunostained using anti-AQP5 antibody as described above [34].

For ex vivo investigation, WT mouse corneas were incubated in Minimum Essential Medium with 2% fetal bovine serum containing the PKA agonist or antagonist for 30 min or 6 h. Corneas were washed with PBS, fixed, cryosectioned, and immunostained using anti-AQP5 antibody as described above [5].

Expression of AQP5 transcripts in the mouse cornea and lens

Total RNA samples from the WT and AQP5-KO mice were subjected to RT–PCR to determine the expression of AQP5, AQP1, or AQP0 using gene-specific primers (Table 1), at the message level. Figure 2A shows the amplification of an approximately ~0.8 kb DNA segment from the WT lacrimal gland (positive control), cornea, lens epithelium, and lens cortical fiber cell samples (lanes 1, 3, 6, and 9, respectively). As expected, comparable samples of the AQP5-KO mouse (lanes 2, 5, 7, and 11, respectively) did not show any amplification. AQP5 genomic DNA sequence has three introns and four exons. The total RNAs prepared were not contaminated with the genomic DNA, since the RT–PCR-amplified products showed a single band of ~0.8 kb (Figure 2A, lanes 1, 3, 6, and 9) comparable to the PCR product of cloned AQP5 cDNA (798 bp; not shown). In addition, there was no amplification from the RNase I treated WT RNAs (Figure 2A, lanes 4, 8, and 10). To confirm that the transcripts were those of AQP5, we cloned and sequenced the RT–PCR products of mouse lens epithelial cells, and fiber cells. FASTA analyses showed 100% homology to the mouse AQP5 sequence in the GenBank (BC150769) verifying the authenticity of the identified AQP5 transcripts.

Figure 2B shows the expression of AQP1 transcripts in the epithelial cells of WT and AQP5-KO mice (lanes 1 and 3) and their absence in the RNase-treated samples (lanes 2 and 4). Similarly, AQP0 transcripts were present in the lens fiber cells of WT and AQP5-KO mice (Figure 2C, lanes 1 and 3), and were not seen when treated with RNase (lanes 2 and 4). In summary, RT–PCR data for lens revealed the presence of AQP5 and AQP1 transcripts in the epithelial cells and AQP5 and AQP0 transcripts in the fiber cells.

Expression of AQP5 protein in the cornea and lens

The presence of AQP5 expression at the message level in mouse cornea and lens prompted us to find out whether there is expression at the protein level. Information about the cellular and subcellular localization of AQP5 protein is critical for understanding the role of AQP5 in these tissues. Therefore, we performed western blotting and immunostaining.

Total membrane proteins from rabbit cornea and lens were immunoblotted with anti-AQP5 antibody (Figure 3A, lanes 1 and 2) which bound to two bands of ~28 (arrow) and ~34 kDa (arrowhead). Similar binding occurred with samples of mouse lacrimal gland (positive control; Figure 3B, lane 1), cornea (lane 2), lens (lane 3) and distinct regions of the lens (lanes 4–6). Total membrane proteins from AQP5-KO mouse lens immunoblotted with anti-AQP0 (Figure 3C, lane 1) and anti-AQP5 (Figure 3C, lane 2) antibodies separately, showed the expression of AQP0 and the absence of AQP5, respectively, substantiating the reaction specificity. In Figure 3A, the ~28 kDa band (arrow) corresponds to the unmodified AQP5 while the ~34 kDa band (arrowhead) could be due to post-translational modification such as phosphorylation. AQP5 has two consensus PKA/PKG-activated phosphorylation motifs (Figure 1), one at residues 153–157 (RRTSP) in loop D, and the second at residues 256–259 (RKKT) at the COOH-terminal domain. To test whether the 34 kDa band could be a phosphorylation product, we treated mouse corneal membrane proteins (Figure 3D, lane 1) with calf intestinal alkaline phosphatase. The ~34 kDa band disappeared (Figure 3D, lane 2) presumably, due to dephosphorylation.

To date, AQP1 in the epithelial cells and AQP0 in the fiber cells, have been functionally characterized in vitro using Xenopus oocytes and ex vivo using lens epithelial cells and fiber cell vesicles [4,5,9,34,36-38]. Published data on the expression of AQP5, in the lens are conflicting. Trace amounts of AQP5 transcript had been documented for rat [1] and dog [39] lens tissues. However, immunoblotting of rabbit [40] and dog [39] lens fiber cell membrane proteins did not detect AQP5. Mass spectrometry studies showed the presence of AQP5 among bovine [23] and murine [24] total fiber cell membrane proteins. We performed immunocytochemical studies to distinctly define the presence and localization of AQP5 within the specific regions of mouse cornea and lens.

Most mammalian corneas have five layers (Figure 4A) from anterior to posterior [41,42]: corneal epithelium, bowman's layer, stroma, descemet's membrane, and endothelium. AQP5 expression and localization have been reported only for the epithelial cell layer and not for other corneal cell layers in the WT mouse [43], rat [22] and dog [39]. Figure 4B shows that in the mouse cornea, AQP5 is expressed in the outermost multilayered epithelial cells, single layered basal columnar epithelium (abundant expression) and in the suprabasal cells. Low level expression is seen in the multilayered flat polygonal cells. Figure 4C corroborates the specificity of antibody binding by the lack of immunoreactivity in the AQP5 knockout mouse cornea. In the stroma, AQP5 is expressed at low levels in keratocytes close to Bowman’s membrane in the central cornea (Figure 4B), and at high levels in keratocytes at the limbal area (Figure 4D and E) which is the junction of the cornea and the conjunctiva. There is no expression of AQP5 in the other parts of the cornea including the endothelial cells. Immunostaining of cryosections of AQP5 knockout limbal area (Figure 4F) showed no antibody binding, validating the specificity of immunoreactions in the WT cornea.

AQP5 in the corneal epithelial cells and keratocytes could be involved in preventing corneal dehydration, promoting wound healing and maintaining corneal transparency. AQP5-KO mice showed significant reduction in corneal epithelial cell membrane water permeability (~2 fold) and increase in corneal thickness [29]. When corneal epithelial cells were subjected to hypotonic conditions, the recovery rate after swelling was significantly reduced in AQP5-KO compared to the WT mice [29]. The corneal epithelial layer serves as a barrier against pathogenic invasion and helps to keep the cornea moist. In the epithelial cells, AQP5 may be critical for corneal fluid dynamics; malfunction can cause corneal edema and dry eye disease [43,44]. Corneal edema can lead to loss of transparency. Kenney et al. [45] analyzed AQPs in normal, diseased and cataract-post-surgical corneas and observed AQP expression abnormalities under several disease conditions.

In the WT lens epithelial cells, anti-AQP1 antibody bound more intensely to the basolateral side than to the apical side (Figure 5A) while anti-AQP5 antibody bound uniformly throughout with punctate membrane localization (Figure 5B). In AQP5-KO mouse, there was no antibody binding to the epithelial cells (Figure 5C), substantiating the specificity of anti-AQP5 antibody binding in the WT (Figure 5B). Figures 5D and E exhibit the expression of AQP5 in the outer cortex and inner cortex fiber cells, respectively. In Figure 5D, antibody binding is more intense on the narrow sides (yellow arrows) of the fiber cells than on the broader sides (red arrows). In addition, it is apparent that the intensity of antibody binding gradually decreases as the fiber cells run deep toward the nucleus of the lens. This difference in antibody binding could be due to epitope masking or loss of epitopes due to fiber cell aging when the cells lose their COOH-terminal ends; the antibodies tested were raised against COOH-terminal epitopes. We have previously reported a similar decrease in antibody affinity in the fiber cells, with regard to anti-AQP0 antibody binding [37]. AQP5 is expressed at low levels in both lens epithelial (Figure 5B) and fiber cells (Figure 5D,E) compared to AQP1 (Figure 5A) and AQP0 (Figure 5F) in the respective cells types. Epithelial and fiber cells in the equatorial region also bound the anti-AQP5 antibody (Figure 5G); the window in Figure 5G is enlarged (Figure 5H) to show the immunoreactivity to the fiber cells. As expected, the equatorial (Figure 5I) and anterior (Figure 5J) regions of the AQP5-KO mouse lens sections showed no antibody binding.

It is estimated that AQP0 constitutes ~44.8% of the total lens fiber cell membrane protein in contrast to AQP5 that constitutes only ~0.36% [24]. In mammalian lens, AQP0 contributes ~80% of the water permeability [4,36,46] in the lens fiber cells even though it is >40-fold less efficient than AQP1 [47]. Mouse lens fiber cell lipid contributes only ~1.5% of the membrane water permeability due to the high levels of cholesterol and sphingomyelin [4,46,48]. We speculated [4] that ~18.5% of the fiber cell water permeability may be contributed by the co-transporters. The present study suggests that the 18.5% water permeability could come from AQP5 itself, or in combination with the membrane co-transporters. Lately, passage of water through fiber cell gap junctions has been postulated [49]. Even though, the expression level is ~100 fold less than AQP0 [24], the single channel water permeability of AQP5 is ~20 fold higher [50]. However, in the AQP0 knockout mice, the compensatory water permeability wielded by AQP5 appears insufficient since the animals developed cataract. In addition, the cell-to-cell adhesion capability could be unique to AQP0; crystallographic structural data [51] do not support the possibility of such a function for AQP5. In the WT, AQP5 could be complementing the water permeability needs of the fiber cells along with AQP0. Any additional function for AQP5 other than water permeability remains unclear. Lack of AQP1 in the lens epithelial cells of human [52] or mouse [53] did not cause lens abnormalities, opacity or cataract. However, AQP1 knockout mouse lenses subjected to stressful conditions, developed lens opacity/cataract with in vitro and in vivo models [53]. Under normal conditions, AQP5 present in the epithelial cells may fulfill the water permeability function in the AQP1 knockout lens, but apparently not under stressful conditions. Despite the nonoccurrence of phenotypes (based on the gross lens morphology) due to the absence of AQP5 in the knockout mouse, it cannot be ruled out that the function/s of AQP5 could be irrelevant, until more details emerge from a wide range of studies including the structural and functional aspects of this protein.

In the lens, while AQP1 is present in the epithelial cells and AQP0 is present in the fiber cells, AQP5 has a global expression. It is reasonable to speculate that the highly efficient AQP5 water channel may aid in maintaining lens transparency and homeostasis. Using AQP0, AQP1 and AQP5-KO mouse models, currently we are investigating the role of AQP5 in maintaining lens transparency and homeostasis.

Regulation of AQP5 localization

Several investigators have demonstrated that AQP5 expression, intracellular trafficking and localization in the epithelial cells are regulated by cAMP through the PKA pathway [54-56]. In our in vitro studies, plasma membrane localization of AQP5 decreased significantly when MDCK cells expressing mouse AQP5 were exposed to mp-cAMP (100 μM), a PKA agonist, for 30 min (Figure 6A). On the contrary, when the cells were exposed to a PKA antagonist, H-89 (20 μM) for the same duration, there was increase in AQP5 plasma membrane localization and abundance (Figure 6B). Cells that were incubated first with the PKA inhibitor H-89 for 30 min and then stimulated with mp-cAMP for 30 min, did not show significant AQP5 internalization (Figure 6C) as opposed to cells exposed only to mp-cAMP (Figure 6A). Similarly, in ex-vivo experiments, plasma membrane localization of AQP5 decreased significantly when WT corneas were exposed to mp-cAMP (100 μM) for 30 min or 6 h (Figure 6D,E). When corneas were exposed to the PKA inhibitor, H-89 (20 μM) for 30 min or 6 h there was increase in AQP5 plasma membrane localization and abundance (Figure 6F,G) depending upon the extent of exposure.

AQP5 has several potential phosphorylation sites (Figure 1). Two consensus PKA/PKG-activated phosphorylation motifs are present at residues RRTSP at 153–157 and RKKT at 256–259, in loop D and at the COOH-terminal domain, respectively. Sidhaye et al. [55] had shown cAMP-induced biphasic effect on AQP5 expression and membrane localization when cells expressing AQP5 were exposed to cAMP for short and long-terms. AQP5 protein internalization, lysosomal degradation and a subsequent decrease in AQP5 protein in the plasma membrane occurred in mouse lung epithelial cells exposed to cAMP for a short-term, but long-term exposure resulted in increased abundance and an increase in trafficking of AQP5 to the plasma membrane. Expression and localization of AQP5 in mouse lung epithelial cell line (MLE-12) were regulated at transcriptional and posttranscriptional levels, respectively [54]. Recently, Hasegawa et al. [56] reported that AQP5 is phosphorylated at Thr259 by PKA through cAMP. Based on the current investigation and previous studies, we hypothesize that dynamic regulation of AQP5 expression and localization through PKA pathway could be a potential tool for short and long-term regulation of membrane water permeability in the cornea and lens. During aging, when most of the lens proteins including aquaporins undergo modifications leading to alterations and/or loss of function, increasing the availability of AQP5 water channels in the plasma membrane by dephosphorylating the AQP5 present in the sub-cellular pool using small molecular pharmacological compound/drugs could augment membrane water permeability to maintain lens microcirculation and homeostasis. Further investigation of this regulatory pathway may provide a strong base for developing novel therapeutic drug targets to treat corneal diseases and lens cataracts.

In conclusion, we have defined the spatial distribution pattern of AQP5 in the cornea and lens. In the cornea, AQP5 is expressed in the stromal keratocytes in low levels compared to the epithelial layer. In the lens, AQP5 is expressed in the epithelial and fiber cells. AQP5 may play a significant role in maintaining transparency and homeostasis in the avascular cornea and lens. A thorough understanding of the functional complementation or uniqueness of the corneal and lens AQPs will facilitate development of therapeutic intervention strategies to avert pathological corneal swelling and opacity, as well as age-onset lens opacity and cataract.