INTRODUCTION

The urine concentration system in the kidney is regulated mainly by vasopressin, an antidiuretic hormone that increases the osmotic water permeability of the collecting duct cells, resulting in bulk reabsorption of free water (1-3). The cellular actions of vasopressin in the collecting duct cell have been partially resolved since the identification of the vasopressin water channel aquaporin-2 (AQP2)¹ (4, 5). Vasopressin has been shown to induce regulatory redistribution of AQP2 from endosomal compartments to the apical membrane (6-8). These observations have directly proved the shuttle hypothesis that exo- and endocytosis of water channel-containing vesicles account for the regulatory effects of vasopressin on water permeability of the collecting duct cells (9, 10).

Despite elucidation of the regulatory translocation of AQP2, the potential interactions between vasopressin-induced intracellular cAMP accumulation and AQP2 trafficking remain unknown. It has been shown that cAMP-dependent phosphorylation of AQP2 expressed in the Xenopus oocyte slightly increased osmotic water permeability without changing AQP2 surface expression (11), but the increase was far too small to account for the dramatic increase in water permeability of collecting duct cells. Many channels have been shown to be functionally regulated by their phosphorylation (12), but the direct regulation of AQP2 channel functions through phosphorylation is minimal at best and is still controversial (13). On the other hand, the physiological significance of regulatory effects on AQP2 trafficking has been emphasized. Regulatory delivery and sorting of membrane proteins have been observed in many cell types (14, 15). Many vesicle-associated proteins that regulate vesicle trafficking have been identified, and some of them have been shown to exist in association with AQP2-containing vesicles (16).

Recently, extrinsic expression of AQP2 in kidney-derived cell lines has been reported, and cAMP-dependent regulatory exocytosis has been shown (17). c-Myc-tagged AQP2 was expressed in LLC-PK1 cells, a cell type that is known to express the vasopressin type 2 receptor and increase intracellular cAMP concentration upon vasopressin stimulation, and vasopressin-induced translocation of the channel was observed (18). Although AQP2 was transferred to the basolateral membrane rather than to the apical membrane where AQP2 is bound in intact kidney cells, the study showed that LLC-PK1 cells are suitable for studying the regulatory translocation of AQP2. Furthermore, the study postulated that AQP2 itself is one of the key components of the vesicle trafficking system. This notion raises the possibility that in addition to its regulation through vesicle-associated proteins, the vesicle trafficking system may also be regulated through modifications of AQP2. In this study, by expressing wild-type and mutant AQP2 in LLC-PK1 cells, we examined the roles of phosphorylation of serine 256 (a residue phosphorylated by cAMP-dependent protein kinase) in the regulatory exocytosis of AQP2. Demonstrations of the regulatory effects of channel phosphorylation on channel translocation would provide new insights into the cell biology of membrane trafficking.

EXPERIMENTAL PROCEDURES

Serine 256 of rat AQP2 was replaced with alanine by polymerase chain reaction-based site-directed mutagenesis to produce the S256A AQP2 mutant. The nucleotide sequences of both strands of the mutant were verified by a fluorescence sequencer (model 373A, Applied Biosystems). Polymerase chain reaction fragment codings for open reading frames of wild-type (WT) and S256A AQP2 (4) were subcloned into the HindIII and XbaI site of the mammalian cell expression vector pcDNA3 (Invitrogen, San Diego, CA). Transfection was performed by electroporation using a Gene Pulser (Bio-Rad). Subconfluent cells were detached from dishes by trypsin treatment and were suspended. ~1 × 10⁷ cells suspended in 600 µl of PBS were electroporated at 350 V with 960 microfarads, using 20 µg of the appropriate plasmid DNA. In most experiments, cells ~24 h after transfection were used. In each transfection, 2 µg of pcDNA3.1/lacZ plasmid (Invitrogen) was co-transfected to verify transfection efficiency. Examinations using the -Gal staining kit (Invitrogen) revealed that >70% of the cells were expressing the lacZ gene.

Osmotic water permeability of LLC-PK1 cells was measured by a light-scattering method using a SX.18MV stopped-flow apparatus (Applied Photophysics Ltd., Leatherhead, United Kingdom) equipped with a circulating water bath. Transfected cells grown on plastic dishes were detached by incubating with 0.25% trypsin for 5 min at 37 °C. Cells were collected, and trypsin was neutralized by washing twice and incubating with DMEM containing 10% fetal calf serum for a few hours at 37 °C in a polystyrene tube to prevent attachment of cells. The cells were then suspended with DMEM at a concentration of ~5 × 10⁶ cells/ml. In some experiments, cells were incubated in DMEM containing 500 µM 8-(4-chlorophenylthio)-cAMP (cpt-cAMP) for 30 min at 37 °C before the assay. The cell suspension was abruptly mixed with PBS containing 600 mM mannitol, and cells were imposed with 300 mM inwardly directed osmotic gradient. A decrease in cell volume due to osmotic water efflux driven by the osmotic gradient was monitored as the time-dependent increase in 90° scattered light intensity monitored at 466 nm. A light intensity trace was obtained in each sample by averaging three to seven measurements. Data were fit to single exponential curves, and osmotic water permeability (P_f) was determined as described (19) by iteratively solving the following equation with Mathematica software (Wolfram Research, Champaign, IL),

(Eq. 1)

where V(t) is cell volume at time t, SAV is initial cell surface area to volume ratio (6 × 10⁵ cm¹, calculated from the average cell diameter of 10 µm measured by phase-contrast microscopy), vw is the molar volume of water (vw = 18 cm³/mol), Osm_in is osmolarity inside the cell, and Osm_out is osmolarity outside the cell.

SDS-polyacrylamide gel electrophoresis (PAGE) was performed as described (20). Cells grown on a 30-mm plastic dish were treated with 150 µl of 1 × reporter lysis buffer (Promega, Madison, WI) for 15 min. Cell lysates were centrifuged for 1 min at 15,000 × g at 4 °C to remove cell debris. Ten µl of the supernatant was denatured in SDS sample buffer (1.5% SDS, 30 mM Tris-HCl, pH 6.8, 2.5% -mercaptoethanol, and 5% (v/v) glycerol) at 80 °C for 10 min, resolved in 10-20% gradient SDS-PAGE for 1 h with a 40-mA current, and electrotransferred to an ECL nitrocellulose membrane (Amersham Corp.) using a Fastblot semi-dry blotting apparatus (Biometra, Goettingen, Germany). After the membrane was blocked with Superblock (Promega) for 1 h at 23 °C and washed once with TBS-T (20 mM Tris, 150 mM NaCl, 0.05% Tween 20, pH 7.4), it was incubated with affinity-purified antibody against a synthetic peptide corresponding to 15 carboxyl-terminal amino acid residues of rat AQP2 (4) diluted at ~1 µg/ml in TBS-T, washed three times with TBS-T, incubated with a 1:200 dilution of biotin-labeled anti-rabbit IgG antibody (Vector, Burlingame, CA), washed, and incubated again with a 1:50 dilution of ABC mixture (Vector). The blot was visualized by ECL-enhanced chemiluminescence using an ECL mini-camera (Amersham Corp.).

24 h after transfection, LLC-PK1 cells were washed with PBS and detached from the dishes by incubating with 0.25% trypsin, 50 mM EDTA. The cells were collected and washed twice with DMEM, incubated with DMEM for 1 h at 37 °C, and washed once with ice-cold PBS. ~1 × 10⁵ cells were suspended with 1 ml of ice-cold biotinylation buffer containing 0.5 mg/ml ImmunoPure Sulfo-NHS-LC-Biotin (Pierce) and incubated for 45 min with gentle rotation at 4 °C (21). The cells were then rinsed and suspended with 1 ml of lysis buffer (Promega), and the cell lysates were precleaned by incubation with 50 µl of protein A-agarose (Oncogene Research Products, Cambridge, MA) for 1 h at 4 °C using an end-over-end rotator. Precleaned lysates were incubated with 1:1,000 diluted affinity-purified anti-rat AQP2 antibody at 4 °C for 1 h with end-over-end rotation. 50 µl of protein A-agarose was then added and incubated 24 h at 4 °C with end-over-end rotation. The samples were washed 5 times with lysis buffer, suspended with 20 µl of SDS sample buffer, denatured at 80 °C for 10 min, loaded onto 10-20% gradient SDS-polyacrylamide gel, electrophoresed, and electrotransferred as described above. Blots were blocked as described above, reacted with streptavidin-horseradish peroxidase (KPL, Gaithersburg, MD), and visualized using the same method as that used for immunoblotting. In some experiments half of the immunoprecipitated sample was electrophoresed, blotted, and examined by ECL-enhanced chemiluminescence to ascertain that the amount of immunoprecipitated AQP2 from each sample was similar. Linearity of the biotin quantitation in the range of the amount used in our experiments was confirmed by applying a serial dilution of the samples to SDS-PAGE and then blotting and visualizing as described above.

About 10⁶ transfected cells were metabolically labeled by incubation with 1 mCi of [³²P]orthophosphate for 4 h in 1 ml of phosphate-free Dulbecco's modified Eagle's medium at 37 °C. 500 µM cpt-cAMP was then added to the medium and incubated for 30 min. The cells were pelleted, washed twice with ice-cold PBS, and suspended with 1 ml of lysis buffer. After 15 min of gentle mixing, AQP2 was immunoprecipitated as described for surface biotinylation, resolved with 10-20% gradient SDS-polyacrylamide gel electrophoresis, dried, and visualized by autoradiography.

RESULTS

WT AQP2 and a mutant AQP2 in which serine 256 was replaced with alanine (S256A) were expressed in LLC-PK1 cells by electroporation, and transient expression of AQP2 was examined by Western blotting (Fig. 1). Immunoreactive AQP2 was not detected in either native LLC-PK1 cells (data not shown) or LLC-PK1 cells transfected with mock vector, indicating that LLC-PK1 cells do not express an AQP2-like protein. A predominant band of ~29 kDa was detected in lysates from cells transfected with WT AQP2. The band was not detected with preimmune serum (data not shown). Interestingly, dispersed bands corresponding to glycosylated forms of AQP2 that were detected in membrane fractions from rat kidney medulla and oocytes expressing rat AQP2 (20) were not observed. A similar result was obtained by expressing c-Myc-tagged AQP2 in LLC-PK1 cells (18), indicating that AQP2 is not properly glycosylated in LLC-PK1 cells. However, in view of the previous observations that first, glycosylation of AQP2 is not necessary for proper folding and assembly as well as plasma membrane expression of AQP2 and second, that the non-glycosylated form of AQP2 possesses a function identical to that of the wild-type AQP2 water channel (22), it follows that a lack of the glycosylated form of AQP2 expressed in LLC-PK1 cells would have little effect in studying AQP2 function and trafficking. When S256A AQP2 was expressed in LLC-PK1 cells, the identical staining of a band of ~29 kDa shows similar expression efficiency, provided that the immunoreactivity of the WT and S256A AQP2 is the same.

To examine the functional expression of WT and S256A AQP2 in LLC-PK1 cells, the osmotic water permeability of the plasma membrane of LLC-PK1 cells transfected with constructs was examined by a stopped-flow light-scattering method (Fig. 2). Light-scattering traces in Fig. 2A show the time course of volume increase in response to inward osmotic gradient. Interestingly, without cAMP stimulation LLC-PK1 cells transfected with wild-type AQP2 showed faster volume shrinkage than did mock-transfected cells. Calculated P_f was about three times larger than the control, and the increase of P_f was ~80% inhibited by preincubation of cells with 50 µM HgCl₂ (data not shown), which is compatible with the characteristics of aquaporin water channels (23). In contrast to the previous observation (18), which showed that P_f of LLC-PK1 cells transfected with AQP2 was similar to that of the control without forskolin or vasopressin stimulation, our results indicated that AQP2 was constitutively expressed in LLC-PK1 cell plasma membrane without cAMP stimulation. This constitutive expression may be partially due to the larger amount of expressed AQP2 or to the lack of cell polarity in our experimental condition. Subsequently, we examined the effects of cAMP on the P_f. Suspended LLC-PK1 cells expressing WT AQP2 were incubated with 500 µM cpt-cAMP for 30 min at 37 °C, and the P_f was measured. A significant increase in the P_f was observed after this treatment, showing a regulatory enhancement of P_f in LLC-PK1 cells expressing WT AQP2. In contrast, when LLC-PK1 cells were transfected with S256A AQP2, the P_f was comparable with that of LLC-PK1 cells transfected with WT AQP2 without cAMP stimulation, but cAMP treatment failed to increase the P_f. These results suggested that serine 256 is required for a cAMP-dependent regulatory increase in the P_f of LLC-PK1 cells expressing AQP2.

To quantitatively determine the surface expression of WT and S256A AQP2, a cell-surface labeling method using biotin was employed. LLC-PK1 cells transfected with WT or S256A AQP2 were labeled with water-soluble Sulfo-NHS-LC-Biotin, and biotin-labeled AQP2 was immunoprecipitated and visualized. AQP2 expressed in the surface plasma membrane was quantified as biotin-labeled AQP2 (21). As shown in Fig. 3A, biotin-labeled AQP2 was detected without cAMP stimulation in LLC-PK1 cells transfected with WT or S256A AQP2, indicating constitutive surface expression of AQP2 and the S256A mutant. After cAMP treatment, surface expression of WT AQP2 increased by ~80% (Fig. 3B). This result, together with the observation of a cAMP-dependent increase in the P_f of LLC-PK1 cells expressing WT AQP2, proved the existence of cAMP-dependent regulatory trafficking of AQP2 in LLC-PK1 cells as well as the constitutive pathway. However, when serine 256 was replaced with alanine, surface expression of the mutant did not increase after cAMP stimulation, suggesting the significant role of serine 256 in cAMP-dependent regulatory exocytosis of AQP2.

Having shown the considerable importance of serine 256 of AQP2, we studied cAMP-dependent phosphorylation of WT and S256A AQP2 in LLC-PK1 cells to examine the role of channel phosphorylation in the trafficking process. As shown in Fig. 4, in vivo phosphorylation of AQP2 was detected without cAMP stimulation, and cAMP treatment increased the phosphorylation by about 39 ± 8% (mean ± S.D. from three independent experiments). On the other hand, in vivo phosphorylation of S256A AQP2 was very limited even after cAMP stimulation, showing that the major phosphorylation site of AQP2 in LLC-PK1 cells is serine 256 and that cAMP stimulation increased phosphorylated AQP2. cAMP-independent basal phosphorylation of WT AQP2 could be either due to basal activity of intracellular protein kinase A in LLC-PK1 cells or phosphorylation of AQP2 by other kinases. The slight phosphorylation observed in the S256A mutant may indicate weak phosphorylation of amino acid residues other than serine 256 of AQP2. Although phosphopeptide mapping was not done in this study, when our results are taken together with those of the previous phosphopeptide analyses of AQP2 in oocytes (11) and in rat kidney medulla (24), it can be concluded that the major phosphorylation site of AQP2 in LLC-PK1 cells is serine 256. 

DISCUSSION

Expression of AQP2 deficient of serine 256 (a consensus phosphorylation site for protein kinase A) in LLC-PK1 cells revealed that serine 256 is required for cAMP-dependent regulatory exocytosis of AQP2. Assessments by functional assay and surface biotin labeling confirmed translocation of WT AQP2 to the plasma membrane upon cAMP stimulation. In contrast, cAMP-dependent regulatory exocytosis of an AQP2 mutant with alanine in place of serine 256 (S256A) was not significant. In addition, in vivo phosphorylation of WT AQP2 expressed in LLC-PK1 cells was shown to occur and was enhanced by cAMP stimulation, while phosphorylation of S256A AQP2 was minimal. Taken together, our results indicated that phosphorylation of AQP2 at serine 256 may be required for regulatory exocytosis of AQP2.

Although it has been well known that functions of many channel proteins are modulated through protein phosphorylation (12), our results provide the interesting notion that the translocation of the channel protein may be regulated by phosphorylation of the channel itself. Our finding suggests that the major passenger of the vesicle trafficking system may be a site of the regulation. This postulation is compatible with the previous prediction that AQP2 itself is a crucial component of a regulated vesicle-recycling system (18). There have been numerous reports showing cAMP-dependent regulatory trafficking of secretory vesicles and membrane proteins (25, 26); however, most of them have suggested that vesicle-associated proteins are sites of regulation (14, 27). It has been shown that trafficking of the cystic fibrosis transmembrane regulator is regulated by cAMP (15, 28, 29), but regulatory effects of cystic fibrosis transmembrane regulator phosphorylation on exo/endocytosis of the cystic fibrosis transmembrane regulator have not been known.

According to the shuttle hypothesis (10), phosphorylation of AQP2 at serine 256 may be required for vesicle trafficking processes including transferring, docking, and fusing of AQP2-containing endosomes toward the plasma membrane (14). From accumulated observations, it is conceived that the AQP2 recycling system is composed of cytoskeletal components (30-33), vesicle fusion proteins (34, 35), and the AQP2 water channel. Vasopressin-responsive modification of actin (32, 36) and phosphorylation of several of the vesicle-associated proteins (37) have been shown. In our experiments, the increment in ³²P labeling of WT AQP2 was not proportional to the increment in channel surface expression, and apparent phosphorylation of WT AQP2 without cAMP stimulation was observed (Fig. 4). These observations may indicate that phosphorylation of AQP2 is not sufficient for exocytosis to occur. Thus, rather than being the initial regulatory effect on the translocation machinery, phosphorylation of AQP2 may be one of a series of regulatory processes. At the least, our results confirm that the phosphorylation of AQP2 plays a significant role in a series of the translocation signaling processes.

When transfected cells were examined immunohistochemically, results are qualitatively consistent with the conclusions from the data on water permeability and surface biotinylation, and after cAMP stimulation, a slight increment in basolateral staining was observed (data not shown) as in the previous study (18). In addition, constitutive surface expression of AQP2 and smaller cAMP-dependent changes were noticed, indicating that the dynamics of AQP2 trafficking in LLC-PK1 cells may be slightly different from those in natural collecting duct cells. However, in preliminary experiments with a cell line derived from OMCD cells (38), results were similar to those observed in LLC-PK1 cells, suggesting that although there may be some limitations, observations in LLC-PK1 cells are relevant to the collecting duct cells.

We found that the appearance of AQP2 in plasma membrane and an increment in P_f are well correlated in WT and mutant channels, suggesting that normal and mutant channels have the same permeability and that phosphorylation of serine 256 of AQP2 did not significantly change single channel water permeability in contrast to the previous study in oocytes (11). Thus, it can be concluded that the number of AQP2 channels, rather than single channel water permeability, is critical for vasopressin regulation of water permeability in collecting duct cells.