Water Permeability of C-Terminally Truncated Aquaporin 0 (AQP0 1-243) Observed in the Aging Human Lens

Abstract

purpose. To first assess the distribution of posttranslationally truncated products of aquaporin 0 (AQP0) in dissected sections of a normal human lens and to determine the effect of backbone cleavage on the water permeability of AQP0.

methods. A 27-year-old lens was concentrically dissected into six sections. Membrane protein was isolated from each section and cleaved with cyanogen bromide, and the peptides were separated and analyzed by reverse-phase (RP)-HPLC-mass spectrometry (MS). The sites of posttranslational AQP0 C-terminal truncation were determined by mass spectrometry. Truncated forms of AQP0 were expressed in a Xenopus laevis oocyte system, and the effect of truncation on AQP0 water permeability was assessed in an oocyte osmotic swelling assay.

results. The extent of truncation at many sites within the C terminus increased with fiber cell age, and the effects of truncations after residues 234, 238, and 243 on AQP0 water permeability were examined. Truncation after residue 243 resulted in an approximate 15% decrease in permeability compared with the full-length protein, AQP0 1-263. However, rather than a direct effect on water transport, analysis of surface protein expression indicated that the decrease in permeability was a result of less efficient protein trafficking to the oocyte surface and that the permeabilities of full-length and 1-243 AQP0 were indistinguishable. Further, C-terminal truncation of AQP0 to 1-234 and 1-238, completely impaired trafficking into the plasma membrane, precluding the measurement of permeability.

conclusions. These data provide evidence that loss of 20 amino acids from the C terminus may not directly affect the ability of AQP0 to transport water.

Aquaporin 0 (AQP0), also known as the major intrinsic protein (MIP), is a water-permeable channel abundant in the ocular lens. The first sequenced member of the aquaporin family, ¹ AQP0 shares the highest sequence similarity (∼80%) with the vasopressin-regulated AQP2 of the kidney’s collecting duct. The aquaporins are ubiquitously distributed transmembrane water channels responsible for maintaining the water homeostasis of entire organisms as well as individual tissues. ² Impairment of aquaporin function results in a wide spectrum of diseases including nephrogenic diabetes insipidus, ³ deafness, ⁴ Sjögren’s syndrome, ⁵ and cataract. ⁶ Of the 11 known mammalian aquaporins, 6 have been identified in the eye and are thought to act in a concerted manner to provide osmotic balance, a critical necessity for maintaining lens transparency. ⁷ Within the lens, AQP1 is restricted to epithelial cells lining the anterior lens surface, whereas AQP0 is distributed throughout the fiber cells composing the bulk of the lens. Many functions have been attributed to AQP0; however, demonstration of the ability of AQP0 to confer membrane water permeability ^{8 9 10} raises the possibility that it may contribute to the movement of water and circulation of nutrients within the avascular lens. ¹¹ Although the precise role of AQP0 in vivo is not entirely known, the importance of AQP0 in maintaining lens transparency is evidenced by mutations in AQP0 in humans ⁶ and mice ^{12 13 14 15} that result in the formation of cataracts. In addition to the development of lens opacities, phenotypic characteristics of the AQP0 knockout mouse include loss of the sharpness of focus, as well as a decreased fiber cell membrane water permeability. ¹⁶  

Many proteins within the human lens, including AQP0, are retained throughout the lifetime of an individual and consequently accumulate age-related posttranslational modifications. This is a result of normal lens development and aging in which mature lens fiber cells, rather than being shed or turned over, are buried under newly differentiating cells at the lens surface. Once the older fiber cells are buried to a particular depth, 300 to 500 μm as measured in the chick lens, they abruptly lose their nuclei and organelles ¹⁷ and concomitantly lose their capacity for protein synthesis and protein turnover. The accrual of age-related posttranslational modifications of lens proteins can affect protein function in a number of ways. Age-related modifications are implicated in the loss of Na,K ATPase activity, ¹¹ loss of pH sensitivity of gap junctional connexin 50, ¹⁸ cytoskeletal remodeling through cleavage of α-spectrin, ¹⁹ and partial loss of the chaperone activity of αA-crystallin. ²⁰ AQP0 undergoes extensive backbone cleavage in normal human lenses of all ages ^{21 22 23} and an accelerated rate of truncation has been observed in cataractous lenses. ^{24 25 26} Structural characterization by mass spectrometry (MS) has permitted the identification of deamidation at residues Asn246 and -259, phosphorylation of Ser235, and backbone cleavage at many sites within the C terminus of human AQP0. ²⁷  

Sequence alignment of the aquaporins demonstrates that the C termini are the most diverse regions of these proteins, and as such these domains are thought to regulate aquaporin function in a tissue-specific manner. Molecular models of the prototypical aquaporin AQP1, resolved by cryoelectron microscopy, depict a right-handed bundle of six tilted α-helices with two short helices dipping into the membrane from either side, ²⁸ allowing the juxtaposition of two Asn-Pro-Ala repeats necessary for water permeability. ²⁹ Studies have shown that the aquaporins are assembled as tetramers ^{30 31 32} with each monomer serving as a water-permeable pore. ^{33 34 35} Although unresolved by electron microscopy, atomic force microscopy placed the C termini of each monomer intracellularly at the interface of two monomers and the central cavity of the tetrameric structure. ^{36 37} Site-directed mutagenesis and naturally occurring mutations of AQP0, AQP1, and AQP2 provide evidence of the involvement of the C termini in routing newly synthesized protein to the target membrane ^{38 39 40} and in regulating protein function. ^{37 41 42} Nemeth-Cahalan and Hall ⁴³ have shown that the water permeability of AQP0 is modulated by calcium and calmodulin, which, based on other studies, may be mediated through a direct interaction of calmodulin with the C terminus. ^{41 44}  

The most abundant membrane protein in lens fiber cells, AQP0 has been postulated to play a role in maintaining water homeostasis within the lens. Recent studies have shown that the movement of water within the lens is heterogeneous, and that fiber cell membrane permeability to water changes with the cell’s age and location in the lens. ^{45 46 47} Directed by ionic currents, the movement of water in the lens has been proposed to supply the nucleus with nutrients from the more metabolically active lens cortex. ⁴⁸ Therefore the distribution of functional aquaporins, ion channels, gap junctions, and transporters throughout the lens is crucial for understanding the circulation of nutrients within the lens. The purpose of the present study was to examine the spatial distribution of truncated forms of AQP0 in the lens and to determine whether age-related C-terminal truncation affects the ability of the protein to transport water. A normal 27-year-old human lens was concentrically dissected into regions of varying fiber cell age and the sites of backbone cleavage within the C terminus of AQP0 were determined by mass spectrometry. Based on the truncation of AQP0 observed in this and previous studies, ^{27 49} the permeabilities of full-length AQP0, residues 1-263, and truncated forms of AQP0, residues 1-234, 1-238, and 1-243, were measured in a Xenopus oocyte osmotic swelling assay. The effect of substitution of a known phosphorylation site, Ser235, on membrane water permeability was also evaluated. Because the level of aquaporin expression at the surface of the oocyte directly affects the water permeability imparted to the membrane, an extracellular binding assay was developed to ascertain the efficiency of wtAQP0 and mutant AQP0 protein expression at the oocyte surface. 

Methods

Dissection of Human Lens and Preparation of AQP0

Human eyes were obtained from the National Disease Research Interchange (Philadelphia, PA). Lenses were removed and the tenets of the Declaration of Helsinki for research involving human tissue were strictly followed. A 27-year-old lens was concentrically dissected into six layers as previously described, ⁵⁰ frozen, and shipped to our laboratory on dry ice. Each lens section was homogenized in 10 mM NaF, 5 mM EDTA, and 1 mM NaHCO₃ (pH 8.0) and the membrane protein pellet isolated by centrifugation at 88,000g for 20 minutes at 4°C. The pellet was washed sequentially with Tris buffer (1 mM CaCl₂, 1 mM EDTA, 5 mM Tris base [pH 9.1]), 4 and 7 M urea in Tris buffer, and finally deionized (d)H₂O. The membrane pellet was resuspended in 1:1 n-propanol-1.5 M Tris (pH 8.7; 140 μL), and cysteine residues were reduced with tributyl phosphine (7 μL; Sigma-Aldrich Co., St. Louis, MO) and alkylated with 4-vinyl pyridine (7 μL; Sigma-Aldrich). After a 1-hour incubation with intermittent sonication, the reaction mixture was centrifuged as just described and the membrane protein pellet washed sequentially with 1:1 n-propanol-Tris-HCl and water. The lipids were extracted by overnight incubation in 95% ethanol at −20°C followed by centrifugation in the same conditions. The membrane protein pellet was then washed with acetone and water. After solubilization in 75% trifluoroacetic acid (TFA), the membrane protein was cleaved with 5 M cyanogen bromide (5 μL) in acetonitrile (Sigma-Aldrich) overnight in the dark. The reaction was quenched by the addition of water and the peptide mixture dried completely under vacuum. 

Reversed Phase-HPLC and Matrix-Assisted Laser Desorption Ionization-Mass Spectrometry Analysis of the C Terminus of AQP0

After cyanogen bromide cleavage, peptides were resuspended in TFA (5 μL) and 5% (vol/vol/) 2:1 isopropanol-acetonitrile in water (100 μL). Peptides were separated on a 1 × 150-mm C4 column (Vydac, Hesperia, CA) using solvents A (0.05% TFA) and B (0.05%TFA in 2:1 isopropanol-acetonitrile), with a gradient of 5% to 97.5% solvent B in 60 minutes, at a flow rate of 400 μL/min with a 10:1 preinjection and 10:1 postcolumn split. ^{51 52} Fractions were collected every 2 minutes and then dried in a vacuum. Fractionated peptides were resuspended in 70% acetonitrile, 0.1% TFA (4 μL) and mixed (1:3, vol/vol/) with 50 mM α-cyano-4-hydroxy cinnamic acid matrix in 70% acetonitrile and 0.1% TFA. Peptide-matrix solutions (0.5 μL) were spotted and dried on a matrix-assisted laser desorption ionization (MALDI) sample plate and were mass analyzed with a time of flight (TOF) mass spectrometer (Voyager DE; Applied Biosystems, Foster City, CA) after desorption with a 337 nm nitrogen laser. An internal standard, insulin m/z 5734, was included for mass calibration. 

Cloning of Human AQP0

Total lens cDNA, kindly supplied by Kirsten Lampi, served as a template for the amplification of human AQP0 by polymerase chain reaction (PCR) using a standard protocol supplied with a kit (Taq PCR Core kit; Qiagen, Valencia, CA). The sense oligonucleotide primer incorporated a SacI restriction site, an alfalfa mosaic virus enhancer sequence, ⁵³ and the first 22 nucleotides of the coding region of human AQP0, as determined from the gene sequence ⁵⁴ : 5′ GAC TGC GAG CTC GGA TCC GTT TTT ATT TTT AAT TTT CTT TCA AAT ACC TCC ACC ATG TGG GAA CTG CGA TCA GCC TCC 3′. The antisense oligonucleotide primer was complementary to nucleotides 846-868 in the 3′ untranslated region of AQP0 and incorporated an ApaI restriction site: 5′AGT CGC GGG CCC TTC TTC ATC TAG GGG GCT GGC TAA A 3′. The PCR product was digested with SacI and ApaI and cloned into a vector (pBluescript SK+; Stratagene, La Jolla, CA) containing a 3′ polyA tail. ⁵³ The entire sequence and correct orientation of AQP0 in the vector were confirmed by automated DNA sequencing at the Medical University of South Carolina Biotechnology Facility. 

Construction of AQP0 Mutants

C-terminally truncated forms and the S235A mutant of AQP0 were constructed by PCR with complementary pairs of primers using site-directed mutagenesis (Quickchange kit; Stratagene). Truncated forms of AQP0 were synthesized by incorporating stop codons at positions 235, 239, or 244, with the following oligonucleotide primers and their complements: 5′-GAG AGA CTG TCT GTC CTC TAG GGC GCC AAA CCC GAT-3′, 5′-ATT TCT GAG AGA CTG TAG GTC CTC AAG GGT GCC AAA-3′, and 5′-GGT GCC AAA CCC GAC TAG TCC AAT GGA CAA CCA GAG G-3′. The resultant plasmids were termed AQP0 1-234, 1-238, and 1-243, respectively. The serine at position 235 was substituted for alanine (S235A) with the primer set 5′-ATT TCT GAG AGA CTG GCG GTC CTC AAG GGT GCC AAA-3′. Incorporation of the desired mutations was confirmed by automated DNA sequencing, starting at nucleotide 458. 

Preparation of AQP RNA

Wild-type and mutant plasmids were linearized with NotI endonuclease, and capped RNA was synthesized by in vitro transcription with T7 RNA polymerase (mMessage mMachine; Ambion, Austin, TX). Human AQP1 cDNA, cloned into a Xenopus oocyte expression vector, ⁵⁵ was obtained from American Type Culture Collection (Manassas, VA) and transcribed with T3 RNA polymerase. After precipitation, drying, and resuspension of the RNA in nuclease-free water, the concentration of RNA was determined by measuring the optical density at 260 nm. Before injection into Xenopus oocytes, the quality of the RNA was assessed by electrophoresis on a nondenaturing 1% agarose gel. Electrophoresis was performed at 60 V until the RNA had migrated 4 to 5 cm into the gel. Degraded RNA that appeared as a smear rather than a band was discarded. To check for degradation of RNA during electrophoresis, a molecular weight RNA ladder (New England Biolabs, Beverly, MA) was used as a positive control. The RNA was visualized by ethidium bromide staining. 

Preparation and Injection of Xenopus Oocytes

Oocytes, harvested from female Xenopus laevis (Xenopus Express, Homosassa, FL), were defolliculated by incubation in 1 mg/mL collagenase A (Sigma-Aldrich) for 2 hours in OR2 buffer (82.5 mM NaCl, 2.5 mM KCl, 20 mM MgCl₂, and 5 mM HEPES [pH 7.6], filter sterilized). ⁵⁶ Stage VI oocytes of similar size and markings were selected and stored at 18°C in ND96 buffer (96 mM NaCl, 2 mM KCl, 1 mM MgCl₂, 1.8 mM CaCl₂, 5 mM HEPES [pH 7.6], filter sterilized) supplemented with 40 μg/mL gentamicin (Sigma-Aldrich), 2.5 mM Na-pyruvate, and 3% horse serum. Twenty-four hours after they were harvested, oocytes were injected with variable amounts of RNA (0.1–50 ng) in 50 nL of water using a positive-displacement pipet (Nanoject; Drummond, Broomall, PA). Control oocytes were not injected. Oocytes were incubated in ND96 buffer plus supplements at 18°C for 48 hours before analysis. The media were replaced once daily, and unhealthy oocytes were discarded. 

Water Permeability Assay

Before the osmotic swelling assay, oocytes were equilibrated at room temperature in fresh ND96 buffer without supplements (190 mOsm). The swelling assay was performed at room temperature and involved the transfer of oocytes to a chamber containing 30% ND96 in water (70 mOsm). ⁸ Images were recorded every 5 seconds for 1 minute with a microscope equipped with a computer-interfaced camera (IM 35; Carl Zeiss Meditec, Thornwood, NY). The osmolarity of the solutions was determined with an osmometer (model 5500; Wescor, Logan, UT). Relative oocyte volumes were calculated from the oocyte cross-sectional area by the equation: (A/A ₀)^3/2 = V/V ₀ where A is the cross-sectional area at time 0 and V is the volume as a function of time. Because the volume of AQP0-injected oocytes would often decrease during the first 10 seconds of the assay, possibly because of an endogenous regulatory volume decrease mechanism, the initial rate of oocyte swelling was determined from the slope of relative volume change with time (d(V _t/V ₀)/dt) from 10 to 60 seconds. The permeability coefficient (Pf) was calculated from the initial oocyte volume (9 × 10⁻⁴ cm³), initial oocyte surface area (S = 0.045 cm²) assuming the oocyte is a smooth sphere, the molar ratio of water (V _w = 18 mL/mol), the osmotic gradient, and the initial rate of oocyte swelling, using the formula: Pf = [(1000 × V ₀)/(S × V _w × Δosm)]) × (d(V _t/V ₀)/dt). ⁵⁷ The mean permeabilities ± SE are shown. The significance of difference between mean permeabilities was determined by the two-tailed paired Student’s t-test. P < 0.05 was considered significant. The data presented are representative of multiple assays performed in batches of oocytes from multiple frogs. 

Production of Antibodies against the Amino Terminus of AQP0 and Full-length AQP0

Antibodies were generated to the N terminus of AQP0 by immunizing New Zealand White rabbits with the N-terminal peptide, residues 1-9, of AQP0 coupled to a carrier protein. The peptide, MWELRSASF, was synthesized by solid-phase peptide synthesis with a cysteine residue incorporated at the C terminus for coupling to keyhole limpet hemocyanin. Bovine AQP0 was purified by anion exchange chromatography ⁵⁸ and used to generate polyclonal rabbit anti-AQP0 antibodies. Both antibodies worked well for immunoblotting; however, only the polyclonal anti-AQP0 antibody was effective for immunocytochemistry (data not shown). 

Immunocytochemistry

Forty-eight hours after injection with 10 ng of wild-type or mutant AQP0 RNA, oocytes were fixed with calcium acetate buffered 4% para-formaldehyde for 30 minutes at room temperature. Oocytes were washed with PBS, placed into a cryomold containing optimal cutting temperature compound (OCT 4583; Sakura, Torrance, CA) and frozen and 10-μm sections obtained. Oocyte sections were blocked with 1% BSA and 5% nonfat dry milk in Tris buffer (20 mM Tris-HCl [pH 7.4], 137 mM NaCl) then incubated with anti-AQP0 antibody (1:100) in 0.1% BSA, 0.5% nonfat dry milk in Tris buffer overnight at 4°C. Secondary antibody incubation was performed using 1:1000 anti-rabbit IgG conjugated to tetrarhodamine isothiocyanate (TRITC; Sigma-Aldrich) under subdued lighting for 1 hour at room temperature. The sections were washed with Tris buffer and a final rinse of water and images recorded by fluorescence microscope (Axioplan2; Carl Zeiss Meditec). 

Immunoblot Analysis

After the swelling assay, oocytes expressing wild-type and mutant AQP0 were homogenized, and total membranes were isolated by centrifugation and washed with aqueous buffer. Samples were suspended in buffer containing 0.25 M Tris-HCl (pH 6.8, 5% wt/vol) SDS, 0.05% bromophenol blue, 10% (vol/vol) glycerol, and 2.5% β-mercaptoethanol and incubated at room temperature for 30 minutes. Oocyte membranes (5 μL) were loaded onto a 14% Tris-glycine gel, and electrophoresis was performed at 125 V for 2 hours. Protein was electroblotted onto 0.2 μm nitrocellulose membrane and incubated with anti-N-terminal AQP0 antibody (1:1000) overnight at 4°C. Protein was detected by enhanced chemiluminescence and quantified with a fluorescence imager (FluorS Imager; Bio-Rad, Hercules, CA). 

Protein Expression at the Oocyte Membrane Surface

The protocol for measuring the surface protein expression at the oocyte membrane was modified from Shih et al. ⁵⁹ Immediately after each swelling assay, seven to eight oocytes were fixed in 4% para-formaldehyde in ND96 buffer (pH 7.4) for 15 minutes at room temperature. Because oocytes expressing AQP0 burst after approximately 10 minutes in 30% ND96, oocytes were subject to hypotonic solution for no longer than 70 seconds before fixing. After they were fixed, oocytes were washed three times by gentle transfer to fresh 100% ND96 buffer supplemented with 5% horse serum in a 24-well plate, and those that were damaged were discarded. Primary antibody incubation was performed with 1:100 rabbit anti-AQP0 antibody in ND96 buffer overnight at 4°C followed by three washes. Oocytes were then incubated with 0.1 μCi of [¹²⁵I] goat anti-rabbit IgG (specific activity 1190 Ci/mmol; New England Nuclear, Boston, MA) in a total volume of 0.6 mL ND96 buffer overnight at 4°C and subsequently washed as just described. [¹²⁵I] counts per minute were measured in a gamma counter (Compu Gamma Cs 1282; LKB, Turku, Finland). The specific binding was obtained by subtracting nonspecific binding of the primary and secondary antibodies to uninjected control oocytes from the total binding of antibody to oocytes heterologously expressing AQP0. The mean specific binding ±SE is shown. The SE of specific binding includes the errors associated with mean binding to uninjected oocytes and the mean binding to injected oocytes. For each amount of RNA injected, three measurements were made, with seven to eight oocytes per measurement. 

Results

Truncation of the C Terminus of AQP0

The sites of posttranslational C-terminal truncation of AQP0 and the distribution of truncated products within a 27-year-old human lens were determined by mass spectrometry. A normal 27-year-old human lens was dissected into six concentric sections labeled outer cortex, cortex 1 to 3, nucleus, and inner nucleus. The outer cortical layer contained the newly differentiated, young fiber cells, whereas the inner nuclear layer was composed of the aged fiber cells of the fetal and embryonic nucleus. AQP0 from each lens section was cleaved with cyanogen bromide and the peptides fractionated by reverse phase (RP)-HPLC. The intact and truncated forms of the C-terminal peptide, residues 184-263, eluted together and fractions containing those peptides were analyzed by MALDI-TOF-MS (Fig. 1) . In the outer cortex, the predicted C-terminal cyanogen bromide fragment, residues 184-263, m/z 8636, was the predominant signal observed. In the older fiber cells, the signal intensity of the intact C-terminal peptide decreased relative to truncated forms of the protein represented by peptides, residue 184 through residues 228, 234, 238, 239, 241, 243, 245, 246, 247, 248, and 259. The signals in the mass spectra were assigned by comparing the observed masses with the predicted masses of the intact and truncated C-terminal peptides. The sequences of peptides 184-263 and 184-259 had been confirmed previously by tandem MS. ²⁷ Previous analysis of AQP0 from whole lens homogenates analyzed in a similar manner revealed these sites and additional less abundant truncation after residues 242, 244, 250, 251, 252, 253, 254, and 258. ²⁷ Although MALDI-MS analysis is not quantitative, these and the available data suggest that truncation occurs first after residues 243, 246, and 259 in the normal aging human lens. The pattern of age-related truncation at specific residues of AQP0 within a single lens is in agreement with that observed in whole lens homogenates of increasing age. ²⁷  

The Permeability of Human AQP0

Human AQP0 was cloned and the osmotic water permeability (Pf) determined using a Xenopus oocyte swelling assay. Because of the low permeability of AQP0 relative to other aquaporins, Xenopus oocytes were injected with a variable amount of human AQP0 RNA to determine the amount necessary to elicit a measurable response in the swelling assay. Figure 2 shows the membrane water permeability and total AQP0 protein expression of oocytes injected with 0.1, 0.5, 1, 5, or 10 ng of AQP0 RNA. At the 10 ng injection of RNA, the oocyte membrane permeability increased by approximately threefold compared with the uninjected control. This is consistent with the Pf of AQP0s in other species ^{8 9 10} and human AQP0, as measured previously by Francis et al. ⁶ As the amount of AQP0 RNA injected into the oocyte increased, the level of total protein expression and the membrane water permeability increased in a dose-dependent manner (Fig. 2) . However, at or above the level of 50 ng of AQP0 RNA injected, the oocytes’ membrane water permeability was unpredictable within and among different batches of oocytes (data not shown). Oocytes heterologously expressing AQP1, used as a positive control for the assay, demonstrated approximately a 10-fold increase in permeability compared with the uninjected control. 

Permeability of AQP0 Mutants

To assess the effects of C-terminal truncation on water permeability, stop codons were incorporated into the AQP0 cDNA to generate proteins truncated on the C-terminal side of residues 234, 238, and 243. The effect of truncation at residue 243 on AQP0 permeability was addressed because it is an abundant C-terminal truncation product in the human lens. ²⁷ Truncation after residue 234, observed in the human lens, would also remove a known site of phosphorylation at residue 235. ²⁷ The effect of truncation at residue 238 was tested because, in addition to being observed in the human lens, the extent of cleavage at this site was accelerated in the selenite-induced cataract rat model. ⁴⁹ After heterologous expression of the full-length and truncated forms of AQP0 in oocytes, the swelling assay indicated that C-terminal truncation resulted in a decrease in membrane permeability to water. Figure 3A shows the mean permeability of oocytes injected with 10 ng of the full-length or mutant AQP0 RNA. The permeability of oocytes injected with 1-234 or 1-238 RNA was similar to that of uninjected control oocytes, whereas the permeability of the AQP0 truncated after residue 243 was only approximately 15% lower than the full-length AQP0 permeability. 

The effect of removing a known site of phosphorylation on the water permeability of AQP0 was also examined. Studies have shown that approximately 20% of serine 235 in human AQP0 is phosphorylated. ²⁷ AQP0 with an alanine substituted at position 235 was expressed in oocytes and the water permeability of the membrane measured. The permeability of oocytes expressing S235A was not significantly different from oocytes expressing the wild-type protein (Fig. 3A) . 

Immunocytochemistry

To determine whether the decreased membrane permeability of C-terminally truncated AQP0 was a result of impaired protein trafficking to the oocyte membrane, immunocytochemistry was performed on oocyte slices. Figure 3B shows the immunofluorescence localization of wild-type and mutant AQP0 proteins expressed in Xenopus oocytes. Immunostaining with a polyclonal anti-AQP0 antibody indicated that the wild-type AQP0, S235A, and 1-243 proteins were expressed at the oocyte surface. However, AQP0 protein truncated after residue 234 or 238 was retained intracellularly and therefore did not contribute to the water permeability of the oocyte membrane. AQP0 1-234 was diffusely distributed throughout the oocyte cytosol, whereas AQP0 1-238 was concentrated just inside the oocyte membrane. Western blot analysis of protein isolated from total oocyte membranes probed with an anti-N-terminal AQP0 antibody confirmed that the truncated forms of AQP0, 1-234 and 1-238, were shifted in molecular weight (data not shown). 

Effect of Truncation on Protein Expression

Because the removal of 25 (AQP0 1-238) or 29 (AQP0 1-234) amino acid residues from the C terminus impaired routing of the protein to the oocyte plasma membrane, the effect of losing 20 residues on the expression of AQP0 1-243 protein was examined. After injecting oocytes with equal amounts of full-length or 1-243 AQP0 RNA (1 or 10 ng), the total level of protein expression in the oocyte membrane was compared by immunoblot analysis (Fig. 4) . Although the permeability of AQP0 1-243 was consistently approximately 15% lower than that of the full-length protein (Fig. 4A) , the amount of full-length and 1-243 protein expressed in the oocytes appeared nearly identical, indicating that truncation did not affect the level of protein synthesis. Because total oocyte membranes includes intracellular stores of overexpressed protein, this approach did not address the amount of protein at the oocyte surface contributing to the membrane water permeability. To determine whether the truncated protein exhibited a lower permeability than the full-length AQP0 or whether it was not as efficiently incorporated into the oocyte membrane, an extracellular binding assay was developed to measure the surface expression of AQP0 protein. 

Effect of Truncation on Expression of Protein at the Oocyte Surface

The heterologous expression of AQP0 protein at the oocyte plasma membrane was assessed by anti-AQP0 antibody binding to the surface of the oocyte. Although other methods for quantitating protein expression at the oocyte surface have been described, ^{60 61 62 63} modification of the approach of Shih et al. ⁵⁹ did not require incorporation of additional mutations, surface accessible lysine or cysteine residues, or microdissection of oocyte membranes. After the swelling assay, whole oocytes were fixed and incubated with a rabbit anti-AQP0 primary antibody and an anti-rabbit [¹²⁵I] labeled secondary antibody. The level of [¹²⁵I] detected was assumed to be directly related to the amount of protein expressed at the oocyte surface. As expected, as the level of AQP0 RNA injected increased, the permeability and the amount of protein expressed on the oocyte surface increased (Fig. 5A 5B) . Note that the anti-AQP0 1-9 antibody directed against the intracellular N terminus of AQP0 showed no reactivity against intact oocytes expressing wild-type AQP0. In addition, binding of the anti-AQP0 antibody to intact oocytes expressing AQP0 1-238 was similar to that of uninjected control oocytes, both under slightly different fixing conditions that those shown in Figure 5 . As seen previously, for equal amounts of RNA injected into the oocytes, the osmotic membrane permeability of 1-243 was less than that of the full-length protein. However, the binding assay showed that the specific binding of antibodies to 1-243-expressing oocytes was less than oocytes expressing the full-length protein. To examine the relationship between permeability and surface expression, the data are presented as specific permeability versus specific binding (Fig. 5C) . These data suggest that at equal levels of protein expression, the water permeability of the full-length and truncated protein is indistinguishable. These data are consistent with the idea that removal of the C-terminal 20 amino acids results in decreased efficiency of the routing and/or incorporation of the truncated protein into the oocyte membrane. This analysis, also applied to the S235A mutant, indicated that when expressed at equal levels at the oocyte surface the wtAQP0 and S235A protein had the same water permeability (data not shown). 

Discussion

Formation of cataracts, as a result of mutations in the AQP0 gene, provides evidence that AQP0 is critical to establishing the ordered architecture of the lens ³⁸ and preserving lens transparency. ⁶⁴ The phenotypic characteristics of these cataracts vary ⁶ and therefore have not illuminated the precise role of AQP0 in maintaining lens homeostasis or in the development of opacities. The cause of the cataract in many of these cases could be due to loss of AQP0 function, gain of mutant AQP0 function, differences in genetic background, or impaired trafficking of the AQP tetramer or other proteins through the co- and posttranslational machinery. Characteristics of the knockout mouse indicate that AQP0 imparts water permeability to the fiber cell membranes and that the AQP0 is necessary for lens focusing and transparency. ¹⁶ These observations are consistent with the suggestion that, in addition to water transport, AQP0 may perform other functions in the lens such as a role in maintaining the structural integrity of fiber cell architecture. 

Previous biochemical analyses of human AQP0 have shown that there is an increase in the level of C-terminally truncated forms of AQP0 in whole lens homogenates of increasing age. ^{21 22 23} For the first time, mapping the distribution of specific truncated products of AQP0 within a single human lens by MS revealed which sites are truncated most readily and that backbone cleavage begins in the outer cortical layers of the lens. The first sites of truncation observed in the young fiber cells of a 27-year-old lens were on the C-terminal side of residues 243, 246, and 259. Consistent with studies comparing AQP0 truncation in the cortex and nucleus, the extent of C-terminal truncation increased with fiber cell age and was extensive in the lens nucleus. The pattern of backbone cleavage observed as the fiber cells aged was identical with that observed in whole lens homogenates of increasing age, suggesting a consistent pattern of C-terminal truncation of human AQP0. ²⁷ This, in addition to the finding that truncation of AQP0 occurs as early as age 7, ²⁷ suggests that truncation is a normal age-related event. 

C-terminal truncation of AQP0 at residue 243 consistently resulted in a 15% decrease in membrane water permeability compared with the full-length protein. However, assessment of the expression of protein at the oocyte surface showed the permeabilities of the full-length protein and AQP0 1-243 to be indistinguishable. This suggests that within the human lens one of the major truncation products remains functionally viable during the aging process. Further truncation to residues 234 and 238 resulted in protein that was improperly routed to the oocyte plasma membrane which necessitates the use of an alternative method to examine the effects of truncation at these residues on AQP0 water permeability. Chandy et al. ⁸ obtained similar results after the expression of bovine AQP0 1-228 in oocytes. These data are consistent with previous findings that the C termini of the aquaporins are involved in routing newly synthesized protein to the target membrane. However within the lens, backbone cleavage of AQP0 is a postmembrane insertion event. Therefore, misrouting of truncated AQP0 to the fiber cell membrane would not be of consequence, unless the protein was regulated by a shuttling mechanism. Although AQP0 has been observed in intracellular vesicles in newly differentiating fiber cells ⁶⁵ and in hepatocytes, ⁶⁶ there is no evidence that lens fiber cell membrane water permeability is regulated by redistribution of AQP0 from the plasma membrane to intracellular vesicles. Endogenously expressed AQP2 and AQP8 undergo shuttling in response to phosphorylation. ^{42 67} However substitution of the major phosphorylation site Ser235 and of other potential phosphorylation sites ¹⁰ did not affect routing or membrane water permeability of AQP0, as measured in Xenopus oocytes. ⁶⁸  

Many lens proteins including AQP0, crystallins, connexins, and cytoskeletal proteins, are posttranslationally truncated with age. Whether these structural changes produce a physiologically relevant effect or are the result of degradation of vestigial protein is the subject of current investigation. The data presented herein are consistent with the idea that AQP0 1-243, one of the major truncation products formed during the normal course of aging, maintains the ability to transport water. The ability of the truncated protein to retain water channel activity is significant with respect to understanding its potential contribution to circulation throughout the lens. Although the truncated protein 1-243 retains water permeability, removal of the C-terminal residues may affect the regulation of channel activity, the localization of the protein in the fiber cell membrane, or the structural properties attributed to AQP0. ³⁶ For example, modulation of water permeability by calmodulin would be lost if the proposed calmodulin-binding domain, residues 225-241, ⁴⁴ in the C terminus of AQP0 were truncated. The involvement of AQP0 in the decreased water permeability of aged fiber cell membranes ⁶⁹ and the formation of a physiological barrier to the movement of water into the lens nucleus ^{45 46 70} of aged lenses remain to be elucidated. 

 Supported by National Eye Institute Grant EY13462.

 Submitted for publication December 20, 2002; revised April 10, 2003; accepted April 12, 2003.

 Disclosure: L.E. Ball, None; M. Little, None; M.W. Nowak, None; D.L. Garland, None; R.K. Crouch, None; K.L. Schey, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

MALDI-mass spectra of C-terminal peptides of AQP0 from a 27-year-old lens. The expected C-terminal peptide of AQP0, generated by cyanogen bromide cleavage, contains residues 184-263. Signals from peptides corresponding to in vivo truncation within the C terminus are labeled with the C-terminal residue, 184-x. An additional cyanogen bromide-cleaved peptide, residues 47-90, also eluted with the C-terminal peptides. The internal standard used for mass calibration, indicated by an asterisk, is not shown for the outer layers because it repressed the signals observed for low-intensity peaks.

Figure 2.

 

(A) Osmotic water permeability of Xenopus oocytes injected with 0.1, 0.5, 1, or 10 ng of AQP0 RNA (▪); 0.6 or 5 ng of AQP1 RNA (□). Control oocytes ( ) were not injected. The mean permeability of three swelling assays with five oocytes per assay ±SE is shown. (B) After the permeability assay, total membrane protein was prepared from five oocytes from each amount of RNA injected. Protein from one third of the membrane preparation was separated by SDS-PAGE. Immunoblot analysis was performed with an anti-AQP0 1-9 antibody and the protein detected by enhanced chemiluminescence.

Figure 3.

 

(A) Osmotic water permeability of Xenopus oocytes injected with 10 ng of AQP0 or mutant AQP0 RNA. The mean permeability of multiple assays with five to eight oocytes per assay ±SE is shown. *Permeabilities that are significantly different from the Pf of wtAQP0-injected oocytes, as determined by a paired t-test (P ≤ 0.05). †Permeabilities that are not significantly different from that of uninjected control oocytes. (B) Immunofluorescent staining of Xenopus oocytes injected with 10 ng of wild-type or mutant AQP0 RNA. Control oocytes were not injected. Forty-eight hours after RNA injection, oocytes were fixed and sections were mounted and stained with a polyclonal rabbit anti-AQP0 antibody and detected by an anti-rabbit IgG labeled with TRITC.

Figure 4.

 

(A) Membrane water permeability for oocytes injected with 0.1, 1, 5, 10, or 50 ng of wtAQP0 RNA (▪) or 1-243 AQP0 RNA (□). Oocytes injected with 0.1 ng of RNA were not different from control oocytes. The mean permeability of multiple assays (number in parentheses) from different preparations of RNA and different batches of oocytes ±SE is shown. *Permeabilities that are significantly different from the Pf of wtAQP0 injected oocytes, as determined by a paired t-test (P ≤ 0.05). (B) After the permeability assay, total oocyte membrane protein was prepared from 5 oocytes injected with 1 or 10 ng of wtAQP0 RNA or 1-243 AQP0 RNA and separated by SDS-PAGE. Immunoblot analysis was performed with an anti-AQP0 1-9 antibody and the protein detected by enhanced chemiluminescence.

Figure 5.

 

(A) Expression of wtAQP0 protein (▪) and 1-243 protein (□) at the oocyte surface as determined by [¹²⁵I] labeled antibody binding. Oocytes were incubated with a rabbit anti-AQP0 antibody followed by [¹²⁵I]-labeled anti-rabbit IgG. The specific binding was calculated by subtracting nonspecific binding of antibody to the uninjected oocyte from the total antibody binding to AQP0 expressing oocytes. The mean binding of three assays with eight oocytes per assay ±SE is shown (n = 3). (B) Water permeability of oocytes injected with various levels of wtAQP0 RNA (▪) and 1-243 RNA (□). The water permeability of uninjected oocytes was subtracted from that of experimental oocytes. The mean permeability of three assays with eight oocytes per assay ±SE is shown (n = 24). *Binding or permeability of 1-243 that is significantly different from that of the wtAQP0-injected oocytes as determined by a paired t-test (P ≤ 0.05). (C) Full-length and 1-243 AQP0 protein expression at the oocyte surface versus the membrane water permeability measured in oocytes.

The authors thank the members of the Medical University of South Carolina core laboratories including the DNA Sequencing Facility, Antibody Facility, Imaging Facility and Ophthalmology Department, Mass Spectrometry Facility, and the Peptide Synthesis Facility for their assistance.