Phosphorylation of aquaporin-2 does not alter the membrane water permeability of rat papillary water channel-containing vesicles.

Antidiuretic hormone modulates the water permeability (P) of epithelial cells in the rat kidney by vesicle-mediated insertion and removal of the aquaporin-2 (AQP-2) water channel. AQP-2 possesses a single consensus cAMP-dependent protein kinase A (PKA) phosphorylation site (Ser-256) hypothesized to regulate channel P(Kuwahara, M., Fushimi, K., Terada, Y., Bai, L., Sasaki, S., and Marumo, F.(1995) J. Biol. Chem. 270, 10384-10387). To test whether PKA phosphorylation of AQP-2 alters channel P, we compared the P values of purified AQP-2 endosomes after incubation with either PKA or alkaline phosphatase. Studies using [-P]ATP reveal that AQP-2 endosomes contain endogenous PKA and phosphatase activities that add and remove P label from AQP-2. However, the P (0.16 ± 0.06 cm/s) of endosomes containing phosphorylated AQP-2 (0.7 ± 0.3 mol of PO/mol of protein) is not significantly different from the same AQP-2 endosomes where 95 ± 8% of the phosphate has been removed (P 0.14 ± 0.06 cm/s). These data do not support a role for PKA phosphorylation in alteration of AQP-2's P. Instead, AQP-2 phosphorylation by PKA may modulate AQP-2's distribution between plasma membrane and intracellular vesicle compartments.

ADH 1 stimulation of rat kidney IMCD causes a rapid increase in its apical membrane osmotic water permeability (P f ) (1)(2)(3). This large ADH-elicited increase in P f occurs by the fusion of cytoplasmic vesicles containing water channels with the apical membrane (1,4,5). Withdrawal of ADH stimulation induces retrieval of apical membrane water channels by endocytosis and returns apical membrane P f to its low base-line value (6 -8). In the rat kidney IMCD, this insertion and re-moval process is mediated by an increase in intracellular cAMP and activation of cAMP-dependent PKA (9,10).
Data from many laboratories have now established the central role of aquaporin water channel proteins as water-selective pores in the plasma membranes of various renal epithelial cells (reviewed in Refs. 2, 3, and 11). Reconstitution studies of AQP-1 in liposomes suggest that AQP water channel proteins exist as homotetramers in plasma membranes where each AQP polypeptide forms a narrow water-selective channel (12). A total of four AQPs expressed in the mammalian kidney have been cloned and characterized. All four AQPs possess six transmembrane domains composed of highly conserved sequences, while the structures of their carboxyl-terminal regions are divergent (2,3). Antisera specific for the respective carboxylterminal sequences of various AQPs have been utilized to localize these proteins to either the apical or basolateral plasma membranes and/or vesicles within individual cell types within specific nephron segments (13)(14)(15)(16).
A large body of data demonstrates that AQP-2 is the ADHelicited water channel. AQP-2 is expressed exclusively by ADHresponsive cells in rat collecting duct (14,17,18), where it is prominently located in the apical membrane as well as a population of subapical vesicles (13,19). Recent ultrastructural studies have demonstrated that ADH stimulation and withdrawal redistributes AQP-2 from cytoplasmic vesicles to the apical membrane followed its retrieval into endosomes (19 -21). Purified endosomes originating from the apical membrane of rat IMCD are highly enriched for AQP-2 (22). Finally, structural alterations of the human AQP-2 gene are associated with the disease nephrogenic diabetes insipidus, where affected individuals lack the ability to produce hypertonic urine despite high serum ADH levels (23).
The cDNA sequence of AQP-2 predicts a 271-amino acid protein with one N-linked glycosylation site (17,18). This corresponds to a 29-kDa protein and its 35-40-kDa glycosylated form as identified by multiple anti-AQP-2 antisera (13,18). In addition, the cDNA sequence corresponding to the carboxylterminal domain of AQP-2 reveals several putative phosphorylation sites including a cAMP-dependent protein kinase A (PKA) site (Ser-256), a site for protein kinase C (Ser-226), and two potential sites of casein kinase phosphorylation (Ser-229 and Thr-244) (17). The presence of these potential phosphorylation sites has raised the possibility that the P f of individual AQP-2s may be altered by phosphorylation in a manner similar to that described for other ion channels (24 -27). This includes alteration of channel function by PKA phosphorylation (28) of the major intrinsic protein of the lens (29) that possesses a 59% sequence homology with AQP-2 (30).
Expression studies of AQP-2 in Xenopus oocytes demonstrate that preincubation with cAMP or its analogs increases the P f of oocytes injected with either total medullary RNA (31) or cRNA from wild type AQP-2 (17,32). In recent studies, Kuwahara and co-workers (32) have utilized site-directed mutagenesis techniques in an attempt to identify a potential role for cAMPmediated PKA phosphorylation of Ser-256 in modulation of AQP-2 function. Their data demonstrate that alteration of AQP-2 Ser-256 results in a loss of the 2-fold increase in P f induced by preincubation of oocytes expressing AQP-2 with cAMP. To distinguish between the cAMP-mediated increases in either 1) the P f of individual AQP-2 proteins resident in the oocyte plasma membrane or 2) alterations in the number of AQP-2 proteins present in the oocyte membrane by insertion and retrieval of AQP-2-containing membrane, these studies quantified binding of an anti-AQP-2 antiserum to the external surface of the oocyte membrane. Since no differences in anti-AQP-2 binding to oocytes were observed after cAMP stimulation, these authors suggested that cAMP-mediated phosphorylation of Ser-256 alters the P f of individual AQP-2 proteins.
To further test the conclusions from these data, we have utilized a homogeneous population of purified endosomes derived from the apical membrane of rat IMCD that are highly enriched for AQP-2 (22). We tested whether AQP-2 is phosphorylated or dephosphorylated by endogenous membrane-bound enzymes present in these purified endosomes and if phosphorylation of AQP-2 alters endosomal membrane P f . Our data show that purified IMCD endosomes possess endogenous PKA and phosphatase activities that phosphorylate and dephosphorylate AQP-2. However, paired measurements of membrane P f show no significant differences between endosomes containing phosphorylated AQP-2 after incubation with exogenous PKA as compared to the same endosomes where AQP-2 is dephosphorylated by alkaline phosphatase treatment. These data do not support a role for cAMP-mediated PKA phosphorylation in regulating the permeability of the AQP-2 water channel.
Isolation of AQP-2 Endosomes-Endosomes were prepared from the inner medulla and papilla of rats as described previously (22). In experiments using stopped-flow fluorimetery, endosomes were loaded with F-dextran (10 kDa) after intravenous injection of rats with 250 mg/animal. Purified AQP-2 endosomes were stored at Ϫ80°C prior to experiments with [␥-32 P]ATP.
Phosphorylation of AQP-2 Endosomes with [␥-32 P]ATP-Endosomes were incubated routinely under conditions described in Refs. 33 and 34 with modifications in the presence or absence of 0.02 g of the catalytic subunit of cAMP-dependent PKA (Sigma), where the final reaction mixture contained 20 units/ml PKA, 0.1 mM [␥-32 P]ATP (10 Ci/mmol, pH 7.5, DuPont NEN), 2 mM Mg 2ϩ , 5% (v/v) glycerol, and 0.05% (v/v) 2-mercaptoethanol for various intervals at 37°C in a total reaction volume of 33 l. The final concentration of endosome protein in these incubations was 1 mg/ml. In some reactions, either purified PKA regulatory subunit (34) (0.3 g/reaction tube; final concentration of 100 inhibitor units/ml) or a 20-mer inhibitory peptide of PKA (IP 20 ) (35) (0.02 g/reaction tube; final concentration of 78 inhibitory units/ml) were added. Phosphorylation reactions were terminated routinely by addition of EDTA to a final concentration of 2 mM. 32 P-Labeled endosomes were also incubated with highly purified alkaline phosphatase (Vector Laboratories, Burlingame, CA) for 30 min to remove 32 P label from endosomal proteins including AQP-2. After incubation with alkaline phosphatase, endosomes were pelleted by centrifugation at 14,000 ϫ g for 10 min and analyzed by autoradiography. Both 32 P-labeled endosome proteins and AQP-2 purified by immunoprecipitation (see below) were fractionated by SDS-PAGE as described previously (22) and gels were fixed, stained with 0.25% Coomassie Blue R-250, destained with 50% methanol/5% acetic acid, and dried under vacuum.
Autoradiography was performed on the dried gels using Kodak X-Omat AR film and Lightning Plus intensifying screens at Ϫ80°C. The intensity of individual bands appearing on the x-ray film was then quantified by laser scanning densitometer (model SL-504-XL; Biomed Instruments, Fullerton, CA). All values of autoradiogram densitometry are expressed as mean Ϯ standard deviation (S.D.).
Immunoprecipitation of AQP-2 Protein-After 32 P labeling, aliquots of endosomal proteins were solubilized by addition of Triton X-100 and leupeptin to a final concentrations of 1% (w/v) and 5 g/ml, respectively, and incubated at 4°C for 1 h. After centrifugation at 14,000 ϫ g for 30 min, the supernatant was mixed with rabbit anti-aquaporin-2 antiserum coupled to Sepharose 4B (Pharmacia LKB, Uppsala, Sweden) according to manufacturer's instructions. After incubation at 4°C overnight with continuous mixing, each tube containing 50 l of anti-AQP-2 Sepharose was centrifuged at 14,000 ϫ g for 3 min and the supernatant removed. The Sepharose pellet was then washed three times by resuspension and centrifugation with 1 ml of 50 mM NaPO 4 , 100 mM NaCl, pH 7.4. Bound AQP-2 protein was then removed from the Sepharose by solubilization with an equal volume of Laemmli buffer and analyzed by SDS-PAGE and autoradiography as described above.
Determination of the Phosphate Content of Purified 32 P-Labeled AQP-2 Protein-After phosphorylation of endosomes for 3 min using exogenous PKA as described above, purified 32 P-labeled AQP-2 protein was isolated by immunoprecipitation and SDS-PAGE. Incubation of immunoprecipitates in SDS solubilizing buffer without 2-mercaptoethanol released denatured 32 P-labeled AQP-2 without quantitative denaturation of light and heavy IgG chains of anti-AQP-2 antibody. Thus, subsequent SDS-PAGE purification of the AQP-2 35-45-kDa band resulted in AQP-2 protein localized to the 35-40-kDa region of the 35-45-kDa protein band that is completely free of contaminating IgG as determined by a combination of protein staining and ECL blotting using affinity-purified anti-rabbit antiserum as described above (data not shown).
Purified 32 P-labeled AQP-2 was then transferred to a polyvinylidene difluoride membrane (Immobilon, DuPont NEN) in 10 mM CAPS, pH 11.0, and its location determined by autoradiography and staining with 0.1% Amido Black. Equal portions of the same 32 P-labeled AQP-2 band was used to determine 1) its molar phosphate content by scintillation counting using appropriate calculations and determination of both 32 P isotope specific activity and scintillation counting efficiency, as well as 2) molar protein content by quantitative amino acid hydrolysis using published data of the molar ratios of individual amino acids of AQP-2 (17).
Determination of the Osmotic Water Permeability (P f ) of AQP-2 Endosomes-Paired measurements of endosomal water flux (J v ) were performed to compare the P f values of endosomes containing AQP-2 in either a maximally phosphorylated or dephosphorylated state. AQP-2 endosomes were first phosphorylated for 3 min as described above, and the J v of one half of the single preparation determined immediately. AQP-2 in the remaining half of the endosomes was then dephosphorylated using alkaline phosphatase as described above and the J v of the aliquot determined. J v was measured as described previously using a SF.17MV stopped-flow fluorimeter (Applied Photophysics, Leatherhead, United Kingdom) with a measured dead time of 0.7 ms configured in the fluorescence mode (22,36). Endosomal shrinkage due to water efflux was monitored as a function of the self quenching of F-dextran entrapped within the endosomal lumen. Stopped-flow fluorimetry data were collected, averaged, and fitted to single exponential curves (Applied Photophysics) and P f determined as described previously (12,22,36).
To insure that phosphorylation did not alter the mean diameters of AQP-2 endosomes, paired samples were fixed and sectioned and mean diameters of endosomes visualized under identical magnifications were determined as described previously (22).

RESULTS
In previous work, we have employed a protocol consisting of five separate differential centrifugation steps combined with Percoll gradient sedimentation to purify a homogeneous population of apically derived endosomes from homogenates of rat kidney inner medulla and papilla (22). These endosomes average 144 Ϯ 5 nm in diameter, contain functional water channels, and are highly enriched for AQP-2 protein as determined by immunoblotting using a specific rabbit anti-AQP-2 antiserum (14,22). In experiments described below, these AQP-2 endosomes were utilized to determine the functional consequences of AQP-2 phosphorylation by PKA.
As shown in Fig. 1, incubation of AQP-2 endosomes with Mg 2ϩ [␥-32 P]ATP alone (lane 1) results in the appearance of 7 major 32 P-labeled SDS-PAGE protein bands of approximately 100 -150, 94, 52, 40 -43, 30 -34, and 22-24 kDa as well as a band of very large molecular mass greater than 200 kDa. Their respective locations in autoradiograms of total endosomal proteins shown in Fig. 1 (lanes 1 and 3) are indicated by small arrows or brackets. However, Triton X-100 solubilization and immunoprecipitation of the endosomal proteins shown in lane 1 with anti-AQP-2 Sepharose shows AQP-2 protein is not phosphorylated (Fig. 1, lane 2). In contrast, addition of purified catalytic PKA subunit to these endosomes results in an overall increase in 32 P labeling of endosomal protein bands (Fig. 1, lane  3) and prominent 32 P labeling of AQP-2 protein in anti-AQP-2 immunoprecipitates (Fig. 1, lane 4). This includes protein bands of 28 and 35-45 kDa corresponding to the nonglycosylated and glycosylated forms of AQP-2 as described previously (18). The locations of AQP-2 bands present in autoradiograms of whole endosomal proteins are indicated by stars of Fig. 1  (lane 3). Under these conditions, phosphorylation of AQP-2 could be attributed directly to PKA since 32 P labeling of AQP-2 was not observed after addition of either an excess of purified PKA regulatory subunit (lanes 5 and 6) or IP 20 , a 20-mer peptide that is a highly specific competitive inhibitor of PKA catalytic activity (Fig. 1, lanes 7 and 8) (35). As expected, addition of excess regulatory subunit (molecular mass 55 kDa) to endosomes results in the appearance of a prominent 55-kDa 32

P-labeled band shown in lane 5 that is not present in lane 7.
A star in lane 3 denotes the location of a fainter 55-kDa band that appears upon addition of both [␥-32 P]ATP and PKA catalytic subunit that likely represents phosphorylation of endogenous PKA regulatory subunit by exogenous PKA.
The ratio of 32 P label between the 35-45-and 28-kDa AQP-2 bands after 5 min of 32 P incorporation by PKA and purification by immunoprecipitation was 3.75 Ϯ 0.57 (n ϭ 12) as determined by densitometry of autoradiograms of anti-AQP-2 immunoprecipitates. This value is comparable to the ratio of these AQP-2 protein bands as detectable by immunoblotting (14,18,22). These data demonstrate that under conditions described in Fig. 1, both the 28-and 35-45-kDa forms of AQP-2 present in these purified endosomes are specific substrates for PKA phosphorylation.
Previous work in brain (reviewed in Refs. 37 and 38), erythrocytes (39), epithelial (40), and neuroendocrine cells (41) has demonstrated that hormone activation causes cAMP accumulation in distinct cellular compartments where PKA subunits are differentially localized and often bound to membranes through their associations with a family of PKA-anchoring proteins. In the kidney medulla and papilla, a significant portion of total cAMP-dependent PKA activity is present in the particulate fractions of homogenates (42). To determine if bound endogenous cAMP-dependent PKA activity is present in endosomes where it could phosphorylate AQP-2, purified endosomes were incubated in the presence of 100 M cAMP and [␥-32 P]ATP only (Fig. 2, lanes 1 and 4). The patterns of phosphorylation obtained in SDS-PAGE analysis of both whole endosomal proteins (lane 1) and anti-AQP-2 immunoprecipitates (lane 4) were then compared to their respective counterparts after incubation of AQP-2 endosomes with either [␥-32 P]ATP alone (Fig. 2, lanes 2 and 5) or after addition of a combination of [␥-32 P]ATP and exogenous PKA catalytic subunit (Fig. 2,  lanes 3 and 6). Addition of 100 M cAMP consistently resulted in 32 P labeling of AQP-2 (lane 4) that was similar to that present after addition of both [␥-32 P]ATP and exogenous PKA (lane 6). 32 P labeling of AQP-2 was absent after incubation of   4 -6). The autoradiogram shown (exposed for 72 h) is from a single experiment representative of a total of four. The relative migration of proteins of known molecular mass are indicated as described in Fig. 1. endosomes with [␥-32 P]ATP only (lane 5). These data demonstrate the presence of endogenous PKA activity that is activated by 100 M cAMP and phosphorylates AQP-2 protein in these endosomes.
To obtain conditions where maximal phosphorylation of AQP-2 was achieved after addition of exogenous PKA catalytic subunit, the time course of 32 P labeling of AQP-2 was determined as shown in Fig. 3A. Maximal incorporation of 32 P into AQP-2 occurred within 3 min (Fig. 3A, lanes 2 and 6) and was followed by progressive loss of label over an interval of 20 min (Fig. 3A, lanes 3, 4, 7, and 8). Quantitation of autoradiograms of immunoprecipitates using anti-AQP-2 antiserum (lanes 1-4) showed loss of 47 Ϯ 23% (n ϭ 4) of AQP-2 32 P label after 10 min as compared to that present after 3 min of phosphorylation by PKA. After 20 min, only 25 Ϯ 13% (n ϭ 4) of 32 P label remained in AQP-2. There were no significant differences observed in ratios of 32  The progressive reduction of 32 P-labeled AQP-2 present in immunoprecipitates shown in Fig. 3A could result from either dephosphorylation by endogenous phosphatase activity or pro-teolysis of AQP-2 during the interval that endosome proteins are phosphorylated. To distinguish between these possibilities, identical aliquots of endosomes were either preincubated under phosphorylation conditions for 20 min at 37°C (Fig. 3B, lane 1) or held on ice (Fig. 3B, lane 2). After subsequent phosphorylation of endosomes for 5 min, the content of 32 P-labeled immunoprecipitable AQP-2 in each aliquot was compared by autoradiography. The 32 P-labeled AQP-2 content of endosomes preincubated under phosphorylation conditions (panel B, lane 1) was not significantly different (0.98 Ϯ 0.05; n ϭ 3) as compared to control samples (panel B, lane 2). These data suggest that the progressive loss of 32 P label from AQP-2 in purified endosomes results from endogenous phosphatase activity. Under these conditions, neither the 35-45-kDa nor the 28-kDa AQP-2 band appears to be preferential substrates for dephosphorylation.
As shown in Fig. 4, incubation of endosomes with 150 g/ml alkaline phosphatase for 20 min at 37°C resulted in a loss of greater than 95 Ϯ 8% (n ϭ 7) of 32 P label from purified immunoprecipitates of AQP-2. Prior to quantitation of the 32 P-labeled phosphate content of AQP-2, the 32 P label resulting from AQP-2 in endosomes subjected to preincubation with alkaline phosphatase was compared to that displayed by control endosomes. This experiment was performed to determine if AQP-2 present in endosomes already possesses a significant content of endogenous nonradioactive phosphate prior to its phosphorylation in vitro by exogenous PKA and [␥-32 P]ATP. Preincubation with alkaline phosphatase would remove any endogenous phosphate and thus be expected to increase incorporation of 32 P label into AQP-2 in subsequent exogenous PKA phosphorylation. However, as shown in Fig. 5, paired experiments revealed no significant difference (0.91 Ϯ 0.1; p Ͼ 0.05; n ϭ 4) in the 32 P label of AQP-2 derived from either control (lane 2) or alkaline phosphatase-treated (lane 1) endosomes. These data suggest that AQP-2 present in purified endosomes does not possess a significant content of phosphate prior to its phosphorylation with exogenous PKA.
To quantify the 32 P phosphate content of AQP-2, 32 P-labeled AQP-2 protein was purified by immunoprecipitation and SDS-PAGE. As described under "Experimental Procedures," the 35-  and 7), and 20 min (lanes 4 and 8) with [␥-32 P]ATP and PKA. The 32 P-labeled protein content of samples were then analyzed by SDS-PAGE and autoradiography as described in Fig. 1, either after addition of SDS solubilization buffer to endosomes (lanes 5-8) or following immunoprecipitation using anti-AQP-2 Sepharose (lanes 1-4). The autoradiogram (48-h exposure) shown is from a single experiment representative of a total of four. The relative migration of proteins of known molecular mass is displayed in a fashion identical to that described in Fig. 1. Panel B, to determine if the reduction of 32 P-labeled AQP-2 in the immunoprecipitates shown in panel A was the result of proteolysis, a single aliquot of endosomes was divided equally and one subaliquot (lane 1) exposed to phosphorylation conditions (see lanes 1 and 2 in Fig.  1) for 20 min at 37°C, while the other (lane 2) was incubated on ice. Subsequently, both aliquots received additions of PKA, were incubated for 5 min at 37°C and solubilized and their 32 P-labeled AQP-2 content determined by immunoprecipitation as shown in panel A, which displays the 28-and 35-45-kDa 32 P-labeled AQP-2 bands. The resulting autoradiogram from this single aliquot of endosomes was exposed for 48 h and is representative of a total of two experiments.
FIG. 4. Incubation of AQP-2 endosomes with alkaline phosphatase dephosphorylates 32 P-labeled AQP-2. AQP-2 in purified endosomes was 32 P-labeled by 5 min of incubation with PKA and [␥-32 P]ATP under conditions described in Fig. 1 and the reaction terminated by addition of EDTA to final concentration of 2 mM. One half of these endosomes were then incubated with 150 g/ml alkaline phosphatase for 20 min at 37°C, and both subaliquots were then analyzed after immunoprecipitation with anti-AQP-2 Sepharose 4B as described in Figs. 1 and 3, where lane 1 is AQP-2 protein derived from alkaline phosphatase-treated endosomes and lane 2 from control endosomes. The autoradiogram (exposed 48 h) displayed is from a single experiment and is representative of a total of seven. The relative migration of proteins of known molecular mass are displayed in manner identical to that described in Fig. 1. 40-kDa region of individual 35-45-kDa AQP-2 protein bands was used for analyses since prior studies had demonstrated that it was free of contamination by IgG protein present in AQP-2 immunoprecipitates (data not shown). Quantitation of 32 P content by scintillation counting and protein by amino acid analysis showed AQP-2 protein possessed 0.7 Ϯ 0.3 (n ϭ 3) mol of 32 P phosphate/mol of AQP-2 protein.
Having established conditions where AQP-2 in endosomes contains an average of 0.7 mol of phosphate/mol of protein (Fig.  5) or greater than 95 Ϯ 8% of the 32 P phosphate is removed by alkaline phosphatase (Fig. 4), the membrane water flux (J v ) of these respective endosomes was compared in a series of paired experiments. AQP-2 endosomes containing entrapped F-dextran were prepared from rats receiving intravenous F-dextran injections and the J v of aliquots of individual preparation of endosomes determined using stopped-flow fluorimetery (22,36). There was no significant difference between the magnitude of entrapped fluorescence as well as the J v of endosomes phosphorylated by PKA and ATP (Fig. 6, upper panel) as compared to those displayed when these endosomes were subjected to dephosphorylation by alkaline phosphatase treatment (Fig. 6, lower panel). Ultrastructural analyses also revealed no significant difference (p Ͼ 0.5) in the mean diameter of phosphorylated AQP-2 endosomes (144 Ϯ 15 nm; n ϭ 50) as compared to dephosphorylated endosomes (146 Ϯ 9 nm; n ϭ 50). Hence, the P f of endosomes containing phosphorylated AQP-2 (0.16 Ϯ 0.06 cm/s; n ϭ 3) was not significantly different from the P f of endosomes where AQP-2 had been dephosphorylated by alkaline phosphatase (0.14 Ϯ 0.06 cm/s; n ϭ 3). These data demonstrate that alterations in the phosphorylation state of AQP-2 do not result in significant changes in the P f of these endosomes. Kinetic studies of 32 P incorporation into AQP-2 reveals its rapid phosphorylation by PKA followed by a net loss of 32 P label (Fig. 3, panel A). Present data suggest that the loss of 32 P label from AQP-2 results from endogenous phosphatase activity rather than proteolysis. First, preincubation of endosomes under these conditions prior to AQP-2 phosphorylation results in no detectable loss of immunoprecipitable 32 P-labeled AQP-2 as compared to control (Fig. 3, panel B). Had significant proteolysis of AQP-2 occurred during the preincubation interval, we would have anticipated a reduction in the 32 P-content of AQP-2 as compared to control. Second, our anti-AQP-2 antiserum specifically recognizes the amino acid sequence (residues 258 -271) located immediately adjacent to the consensus PKA phosphorylation site at Ser-256 of AQP-2 (14). If limited proteolysis of AQP-2 had occurred, then additional 32 P-labeled AQP-2 bands would likely be present in autoradiograms of AQP-2 immunoprecipitates. However, we have observed no such additional 32 P bands in our AQP-2 immunoprecipitates. The presence of bound PKA and phosphatase activities in purified AQP-2 endosomes is intriguing because recent reports have demonstrated a specific role for both membrane-bound PKA (43) and phosphatase enzymes (44) in modulation of func-FIG. 6. Comparison of the membrane water flux (J v ) of endosomes containing AQP-2 present in phosphorylated and dephosphorylated states. AQP-2 endosomes containing entrapped F-dextran were isolated from rats as described in Ref. 22. A single aliquot of endosomes was first phosphorylated for 3 min using ATP and PKA under conditions described in Fig. 4 (lanes 1 and 2). While the membrane J v of one half of these endosomes (upper tracing) was being quantified using stopped-flow fluorimetry as described under "Experimental Procedures" and in Ref. 22, the remainder were incubated with alkaline phosphatase in a fashion identical to that described in Fig. 4 (lane 1). After dephosphorylation of AQP-2, the J v of these endosomes was determined (lower tracing) under identical conditions. These data display individual paired recordings from a single experiment performed a total of three times.
FIG. 5. Preincubation of AQP-2 endosomes with alkaline phosphatase does not alter its content of 32 P label after subsequent phosphorylation with exogenous PKA. Identical aliquots of AQP-2 endosomes were preincubated in the absence (Cont) or presence (Alk. P.) of alkaline phosphatase as described in Fig. 4. AQP-2 endosomes were then repeatedly centrifuged in ice-cold Buffer B to remove soluble alkaline phosphatase enzyme. Both endosome aliquots were then phosphorylated and analyzed as described for lane 4 of Fig. 1. The autoradiogram exposed for 36 h displays a representative experiment performed a total of four times. Arrows located on the right denote the 28and 35-45-kDa 32 P-labeled AQP-2 bands. tional aspects of various ion channels (45). Alternatively, it is possible that the presence of both PKA and phosphatase activities in purified AQP-2 endosomes results from the nonspecific binding of these enzymes to vesicles. A detailed characterization of both enzymatic activities that is beyond the scope of this report will be necessary to distinguish between these possibilities.
These data shown in Figs. 4 and 5 demonstrate that covalent phosphate can be removed from AQP-2 protein by alkaline phosphatase treatment and that AQP-2 present in purified endosomes is present in a relatively dephosphorylated state. In contrast, quantitation of the AQP-2 32 P content after its incubation with exogenous PKA shows it is efficiently phosphorylated in vitro. Although our analysis of AQP-2 phosphate content was limited to the 35-40-kDa region of the 35-45-kDa band, densitometry data showing that both the 28-and 35-45-kDa AQP-2 bands possess 32 P label proportional to their protein content suggest that both glycosylated and nonglycosylated forms of AQP-2 possess similar 32 P contents.
Paired experiments shown in Fig. 6 reveal no significant differences between the P f values exhibited by endosomes possessing AQP-2 phosphorylated by PKA and ATP or dephosphorylated by alkaline phosphatase. Having quantitated the phosphate content of AQP-2 using 32 P label, these experiments were designed to detect alterations in the P f exhibited by endosomes possessing AQP-2 in phosphorylated versus dephosphorylated states. Under our conditions, where a 30% difference in vesicle P f could be measured and endosome size did not change, no significant differences in either the magnitude of the entrapped vesicular fluorescence or its time course were noted. Thus, it is difficult to attribute the nearly 5-fold increase in apical membrane P f observed after incubation of rat inner medullary collecting duct with cAMP (46) to changes in P f resulting from AQP-2 phosphorylation. These data do not support the speculation that phosphorylation of AQP-2 by PKA alters the P f of AQP-2 proteins (18,32). However, our data cannot exclude entirely this possibility since more complex speculations such as interactions between AQP-2 protein and other proteins present in purified AQP-2 endosomes that may modulate AQP-2 protein function were not tested. Further study is necessary to reconcile the apparent differences between the role of phosphorylation of AQP-2 in modulation of plasma membrane P f in Xenopus oocytes (18,32) as compared to AQP-2 endosomes purified from rat kidney as reported here.