Since the first membrane water channel, AQP1, was cloned and functionally characterized (1,2), a growing family of water channel proteins has been identified (3). They have been studied extensively to obtain detailed knowledge of water movement across cellular plasma membranes in physiologic and pathologic conditions (4,5,6). Aquaporin-2 (AQP2) has been characterized as the major vasopressin-regulated water channel and is predominantly localized in the apical plasma membrane and subapical vesicles of principal cells in the kidney collecting duct (7,8). It is well established that AQP2 plays a critical role in urinary concentration (9). Water conservation in kidney is maintained through the regulated insertion of AQP2 in the apical plasma membrane of collecting duct principal cells and through the regulation of AQP2 expression. These two fundamentally different regulatory mechanisms of collecting duct water permeability act in a concerted manner. Wade et al. originally proposed the “shuttle hypothesis” (10) that has lately gained support by demonstration of vasopressin-regulated translocation of AQP2 from intracellular vesicles to the apical plasma membrane (11,12,13,14). Conversely, vasopressin withdrawal causes AQP2 retrieval from the apical plasma membrane into vesicles (12) for potential reuse by recycling. In addition, collecting duct water reabsorption is regulated by long-term mechanisms, which were functionally described by Lankford et al. (15). Subsequently, it was shown that high levels of circulating vasopressin (in response to thirsting or exogenously administrated arginine vasopressin [AVP]) cause a marked increase in the abundance of AQP2, including an increase in apical plasma membrane levels, which adaptively increase the osmotic water permeability of the collecting duct (7,16).
Defective long-term or short-term regulation of AQP2 plays a key role in several water balance disorders. Evidence from animal models of acquired nephrogenic diabetes insipidus (NDI) have shown reduced AQP2 expression in response to long-term lithium treatment (17), release of ureteral obstruction (18,19), hypokalemia (20), and hypercalcemia (21,22). Hence, downregulation of AQP2 appears to be a general mechanism in the development of multiple forms of acquired NDI. Moreover, evidence from animal models of liver cirrhosis (23,24) or congestive heart failure (25,26) suggest that upregulation and increased targeting of AQP2 also play a role in states of water retention.
Because AQP2 has such an important role in water balance and water balance disorders, determination of alterations in AQP2 levels in the kidney would be a valuable marker in the study of these disorders. Recent findings of urinary excretion of AQP2 (27,28,29) indicate that AQP2 levels in the kidney may be assessed by measurement of urinary AQP2 levels. However, the cellular mechanism involved in AQP2 excretion is unknown. Furthermore, it is not clear whether urinary excretion of AQP2 correlates with total AQP2 expression in the kidney or whether it correlates with AQP2 levels in the apical plasma membrane of collecting duct principal cells (i.e., with vasopressin activity). The present study addresses some of these issues.
To characterize these issues, rats were subjected to treatments that have known consequences on AQP2 expression levels and to changes in levels of AQP2 in the apical plasma membrane. These protocols include untreated rats (7), rats thirsted for 48 h (7,16), rats thirsted for 48 h and subsequently water loaded (11), and finally Brattleboro rats lacking vasopressin secretion (8,13,30) or rats treated with lithium for 1 mo (17) known to have extreme polyuria and virtually no AQP2 in the apical plasma membrane.
The study showed that: (1) Urinary excretion of AQP2 takes place via a selective apical pathway. (2) AQP2 excretion decreased with decreased kidney AQP2 expression and increased with increased kidney expression. However, in conditions with known severe changes in vasopressin action, urine excretion of AQP2 did not parallel kidney expression. (3) In contrast, changes in urinary AQP2 excretion closely paralleled conditions with distinct and known changes in vasopressin action, which dramatically changes levels of AQP2 in the apical plasma membrane. Thus, urinary AQP2 levels may be used under standardized conditions as a marker for the renal ability of water reabsorption in physiologic and pathologic conditions.
Materials and Methods
Munich-Wistar rats (Møllegard Breeding Center, Eiby, Denmark), initially weighing 200 to 300 g, were used. All rats had access to a standard rodent diet (Altromin, Lage, Germany) and water in normal rat cages before treatment.
Protocol 1: Normal Rats. To estimate daily AQP2 excretion during normal steady-state water balance, rats (n = 6) were kept in metabolic cages with free access to water and food for 24 h to collect urine (8 a.m to 8 a.m.). The rats were then anesthetized using halothane, and both kidneys were removed quickly and frozen in liquid nitrogen.
Protocol 2: Acute Desmopressin Acetate Treatment. To determine the relationship between the urinary excretion of AQP2 and the changes of AQP2 levels in the apical plasma membrane caused by the acute effect of desmopressin acetate (dDAVP), 20 μg of dDAVP in 300 μl of saline was injected subcutaneously into rats (n = 6). For controls, 300 μl of saline was injected subcutaneously (n = 6). The rats were kept in metabolic cages overnight with free access to water and food. Urine output was collected 8 h after treatment (12 a.m. to 8 a.m.). The rats were then anesthetized using halothane, and both kidneys were removed quickly and frozen in liquid nitrogen.
Protocol 3: Thirsting. To determine AQP2 excretion during 48-h thirsting (a condition known to increase endogenous vasopressin secretion), rats (n = 6) were kept in metabolic cages with free access to food but were deprived of water. Additionally, six control rats were kept in metabolic cages but had free access to water and food. Urine output was collected daily. Rats were sacrificed after thirsting for 48 h, and kidneys were taken out and frozen in liquid nitrogen. Urine collection during the last 24 h of thirsting (8 a.m. to 8 a.m.) was used for immunoblotting (and immunocytochemistry) as described below.
Protocol 4: Thirsting Followed by Water Loading. To determine the alterations in urinary excretion of AQP2 during a period in which water was resupplied to rats after 48 h of water deprivation, rats (n = 6) were maintained in metabolic cages and thirsted for 48 h. Urine collected during the last 24 h was used for immunoblotting and immunochemistry. Subsequently, rats were resupplied with water, and urine was collected during the following 24 h to examine the urinary AQP2 levels and compare this with the levels in the preceding 24 h (during water deprivation; 8 a.m. to 8 a.m.).
Protocol 5: Brattleboro Rats. Brattleboro rats (n = 6), which lack vasopressin secretion, were maintained in metabolic cages for urine collection over 24 h (8 a.m. to 8 a.m.). Kidneys were extracted for determination of AQP2 levels.
Protocol 6: Rats with Lithium-Induced NDI. Rats (n = 6) were treated with lithium for 4 wk, which induced an extreme polyuria (17). Urine was collected over 8 h (8 a.m. to 4 p.m.), during which rats were housed in metabolic cages. Kidneys were extracted to determine AQP2 levels.
From each of the protocols described above, additional animals were subjected to perfusion fixation of kidneys by retrograde perfusion through the abdominal aorta.
Preparation of Gel Sample from Urine
The urine osmolality was determined with a vapor-pressure osmometer (Osmomat 030, Gonotec, Berlin, Germany). Urine was collected in plastic tubes containing 100 μl (1 μg/μl) of leupeptin and 100 μl (5 μg/μl) of Pefabloc® (Boehringer Mannheim, Mannheim, Germany). Then, 0.25% bovine serum albumin was added, and the samples were subsequently centrifuged at 4,000 × g for 15 min. To the supernatant, Triton X-100 or Tween 20 was added to make up a 1% solution. Subsequently, the 4,000 × g supernatant was centrifuged at 200,000 × g for 1 h, and the resulting pellet was carefully rinsed twice and solubilized in Laemmli sample buffer. The urine sediment obtained from 4,000 × g centrifugation (pellet) was washed in phosphate-buffered saline (PBS) and centrifuged at 4,000 × g for 15 min, then vortexed in PBS containing 1% Triton X-100, and centrifuged at 200,000 × g for 1 h. The resultant pellet was finally dissolved into Laemmli sample buffer.
Preparation of Tissue Samples for Immunoblotting
After thawing on ice, rat kidneys were finely minced and homogenized in dissecting buffer (0.3 M sucrose, 25 mM imidazole, 1 mM ethylenediaminetetra-acetic acid, 8.5 μM leupeptin, 1 mM phenylmethylsulfonylfluoride, pH 7.2) using an Ultra-Turrax T8 homogenizer (IKA Labortechnik, Germany). The homogenates were centrifuged at 4,000 × g for 15 min at 4°C. The supernatant was centrifuged at 200,000 × g for 1 h, and the resultant pellet was resuspended in dissecting buffer. Gel samples (in Laemmli sample buffer containing 2% sodium dodecyl sulfate [SDS]) were made from this membrane preparation.
Electrophoresis and Immunoblotting of Tissue and Urine Samples
Samples were loaded onto 12% SDS-polyacrylamide gel electrophoresis (PAGE) gels and run on a BioRad minigel system. After transfer by electroelution to nitrocellulose membranes, blots were blocked for 1 h with 5% skim milk in PBS-T (80 mM Na2HPO4, 20 mM NaH2PO4, 100 mM NaCl, 0.1% Tween 20, pH 7.5), washed 3 times in PBS-T over 25 min, and subsequently incubated overnight with primary antibody: (1) LL127, an antibody raised in rabbit against peptide to the COOH-terminal 22 amino acids of rat AQP2. Either immune serum (diluted 1:2000) or a affinity-purified antibody (LL127AP, diluted to 40 ng IgG/μl) in PBS-T plus 1% bovine serum albumin; (2) LL178AP, an affinity-purified antibody raised in rabbit against a peptide corresponding to the COOH-terminal 26 amino acids of AQP3 (affinity-purified) (31). After washing as described above, the blots were incubated at room temperature for 1 h with horseradish peroxidase-conjugated secondary antibody (P448; Dako, Glostrup, Denmark; diluted 1:3000). The blots were finally washed with PBS-T, and antibody binding was visualized using the enhanced chemiluminescence system (Amersham International, Buckinghamshire, United Kingdom). Controls were made with exchange of primary antibody to antibody preabsorbed with immunizing peptide, or with nonimmune IgG (diluted 1:1000).
Enhanced chemiluminescence films were scanned using an AGFA ARCUS II (AGFA-Gevaert, Leverkusen, Germany) and Corel Photopaint Software (Corel, Toronto, Ontario, Canada). Both the 29-kD and the 35- to 50-kD bands (corresponding to nonglycosylated and glycosylated AQP2) were scanned. High molecular weight AQP2 bands in urine samples were also scanned. The labeling density was quantified using specially written software (available upon request). Bands from gels made with serial dilutions of protein from inner medulla, processed as above, were found to be linear over a wide range (17). For quantification of AQP2 expression, samples were chosen that gave bands within the linear range. Urinary AQP2 levelsa were estimated using the formula:
where P is urinary AQP2 level (% of total AQP2 in the two kidneys), Vu1 is total volume of urine collection, Vu2 is urine volume used for gel sample preparation, Vu3 is total volume of gel sample prepared from urine, Vu4 is volume of gel sample subjected to SDS-PAGE, Du is labeling density of urine sample on immunoblot, Vk1, total volume of one kidney after homogenization, Vk2, volume of kidney sample used for preparation of gel sample, Vk3 is total volume of gel sample prepared from kidney, Vk4 is volume of gel sample of kidney subjected to SDS-PAGE, and Dk is labeling density of kidney sample on immunoblot (×2 to correct for total expression in two kidneys).
Values are presented as means ± SEM. Comparisons between groups were made by unpaired t test. P values < 0.05 were considered significant.
Determination of AQP2 in Different Fractions of Rat Urine
To determine the levels of AQP2 in the fractions obtained during the different steps of ultracentrifugation, urine from three rats was subjected to the following procedure: (1) 4,000 × g spin for 15 min. A gel sample was prepared from the 4,000 × g pellet. (2) The 4,000 × g supernatant was subjected to centrifugation at 200,000 × g for 60 min, and a gel sample was prepared from the pellet. (3) The supernatant was concentrated using Centriprep (Centriprep-10; Amicon, Beverly, MA), and a gel sample was prepared from this. The gel samples were subjected to immunoblotting and densitometry.
Immunoelectron microscopy was carried out as described previously (32). Rat kidneys from Munich-Wistar rats subjected to protocols 1 to 6 were perfusion-fixed by retrograde perfusion through the abdominal aorta with 8% paraformaldehyde in 0.1 M sodium cacodylate buffer, pH 7.2. Tissue blocks from the inner medulla were post-fixed in the same fixative for 2 h, infiltrated for 30 min with 2.3 M sucrose containing 2% paraformaldehyde, mounted on holders, and quickly frozen in liquid nitrogen. The frozen samples were freeze-substituted in a Reichert Auto Freeze-Substitution Unit (Reichert, Vienna, Austria) as described before (33). Briefly, the samples were sequentially equilibrated over 3 d in methanol containing 0.5% uranyl acetate at temperatures gradually increasing from -80°C to -70°C, and then rinsed in pure methanol for 24 h while increasing the temperature from -70°C to -45°C. At -45°C, the samples were infiltrated with Lowicryl HM20 and methanol 1:1, 2:1, and, finally, pure Lowicryl HM20 before ultraviolet polymerization for 2 d at -45°C and 2 d at 0°C. Immunolabeling was performed for electron microscopy on ultrathin Lowicryl HM20 sections (60 to 80 nm), which were incubated overnight at 4°C with affinity-purified anti-AQP2 diluted in PBS with 0.1% BSA or 0.1% skim milk. The labeling was visualized with goat-anti-rabbit IgG conjugated to 10 nm colloidal gold particles (GAR.EM10; BioCell Research Laboratories, Cardiff, United Kingdom) diluted 1:50 in PBS with 0.1% BSA in PBS. The sections were stained with uranyl acetate and lead citrate before examination in Philips CM100 or Philips 208 electron microscopes.
Immunoelectron microscopy of urine samples was performed by applying methods as described previously (32,34). Munich-Wistar rats (n = 6) were accommodated in metabolic cages for 24 h, and urine was collected on ice in PLP fixative (0.01 M NaIO4, 0.075 M L-lysin, 0.0375 M Na2HPO4, pH 6.2) containing 2% paraformaldehyde and 0.1% glutaraldehyde. The urine samples were centrifuged at 4,000 × g for 15 min, the pellets were omitted, and the supernatants were subsequently centrifuged at 200,000 × g for 1 h. The pellet obtained after the 200,000 × g centrifugation was rinsed twice and then dissolved into PBS. A total of 3 to 5 μl of the dissolved urine pellet was applied to freshly negative-glowed carbon film-supported 300-mesh nickel grids (Graticules, Ltd.). Excess fluid was removed and the grids were floated on drops of PBS and rinsed twice in PBS. Subsequently, the grids were preincubated for 5 min with PBS containing 1% BSA, and then rinsed three times in PBS. After incubation with primary antibody (affinity-purified anti-AQP2 diluted in PBS with 0.1% BSA or 0.1% skim milk in PBS) for 30 min at room temperature, all grids were washed twice, and then incubated with goat-anti-rabbit IgG conjugated to 10-nm colloidal gold particles (BioCell Research Laboratories, diluted 1:50). The grids were finally washed three times with PBS, then three times with 10 mM imidazole, pH 7.4, and negatively stained with 1% uranyl acetate as described (34).
Urinary AQP2 Predominantly Accumulates in 200,000 × g Pellet Containing Vesicles and Membrane Fragments
To determine AQP2 levels in rat urine, urine samples collected over 24 h were centrifuged at 4,000 × g followed by 200,000 × g centrifugation of the supernatant for 1 h. AQP2 was abundant in the 200,000 × g pellet as demonstrated by immunoblotting (Figure 1A). As shown in Figure 1A, immunoblots of kidney membranes showed two distinct bands migrating at approximately 29 kD and 35 to 50 kD. This corresponds to nonglycosylated and glycosylated AQP2, respectively. In the urine samples the same bands were observed and, in addition, bands of higher molecular weight were also seen. As shown in Figure 1B, the anti-AQP2 labeling was ablated when blots were probed with antibody preabsorbed with the immunizing peptide, demonstrating the specificity of the labeling. Thus, the high molecular weight bands likely represent AQP2 aggregates and less likely oligomerized AQP2. To determine the relative amount of AQP2 in the 4,000 × g pellet and in the 200,000 × g pellet and supernatant, respectively, urine samples from three normal rats were analyzed using fractionation by ultracentrifugation, immunobloting, and densitometry. The results showed that 96.7 ± 9.1% (n = 3) of total urinary AQP2 was present in the 200,000 × g pellet and 2.6 ± 1.2% (n = 3) was present in the 4,000 × g pellet, whereas only 0.7 ± 0.3% (n = 3) of total urinary AQP2 was present in the 200,000 × g supernatant. Thus, we concluded that urinary excretion of AQP2 in rats can be determined by analysis of the 200,000 × g pellet. This does not cause a significant underestimation of the excreted amount of AQP2, since the total AQP2 levels in the 200,000 × g supernatant and the 4,000 × g pellet are very low.
To determine the levels of AQP2 excreted into the urine, adult rats (n = 6) were maintained in metabolic cages for 24 h to collect urine. AQP2 levels in both rat kidney and urine were quantified by immunoblotting and densitometry (semiquantification). As shown in Figure 1C, densitometry of urinary AQP2 of immunoblots indicated that daily urinary excretion of AQP2 in rats was 3.9 ± 0.9% (n = 6) of total kidney AQP2 levels.
AQP2 Is Predominantly Excreted via an Apical Pathway in the Collecting Duct
Immunoblotting of urine samples prepared by sequential 4,000 × g spin and subsequent 200,000 × g spin of the supernatants revealed that by far most AQP2 was associated with the 200,000 × g pellet. Only marginal excretion was associated with the 4,000 × g pellet containing whole cells and larger debris. As shown in Figure 2B, quantitative densitometry demonstrated that AQP2 in 4,000 × g sediment of urine represented only 0.09 ± 0.01% of AQP2 in total kidney, whereas urinary AQP2 levels in the consecutive 200,000 × g pellet of the same urine sample were 3.0 ± 0.7% of AQP2 in total kidney (compared with the same kidney samples). Only 2% of the total urinary AQP2 was present in the 4,000 × g sediment of urine. This indicates that the main excretion of AQP2 is not likely to be due to whole cell shedding since whole cells or lumps of cells would be present in the 4,000 × g pellet.
To further evaluate how AQP2 is excreted in urine, i.e., whether urinary AQP2 originates from whole cell shedding or whether it is excreted by a more specific process, we compared the expression and excretion levels of both AQP2 and AQP3 in the same samples. AQP3 is localized in the basolateral plasma membranes of the same cells and therefore represents an ideal basolateral marker. In previous membrane fractionation studies, it has been shown that AQP3 is abundant in 4,000 × g supernatant and hence in the 200,000 × g fraction (16,31). As shown in Figure 3, two prominent AQP3 bands migrating at approximately 27 kD and 33 to 40 kD were seen in kidney samples (200,000 × g pellet) after immunoblotting, consistent with previous observations (31). However, even at high exposures no AQP3 labeling was detected in the urine samples (Figure 3, top panel). In contrast, AQP2 was found abundantly both in the same kidney and urine samples (Figure 3, bottom panel). Thus, there was a disproportional excretion of AQP2 and AQP3 in urine. This strongly suggests that AQP2 excretion is not dependent on whole cell shedding, but strongly supports the view that a selective apical pathway is involved.
The absence of AQP3 in the urine samples was not due to a lower stability of AQP3 in urine since kidney membranes stored for 18 h in 30 × volumes of rat urine had remained intact antigenicity using the same anti-AQP3 antibody (recognizing a COOH-terminal epitope) with no signs of degradation (not shown).
dDAVP Treatment Markedly Increases Urinary Excretion of AQP2
Recent studies revealed that vasopressin causes translocation of AQP2 from intracellular vesicles to the apical plasma membrane (11,12,13,14), and in humans dDAVP treatment has been shown to increase urinary excretion of AQP2 (28,29). Thus, dDAVP treatment was used to evaluate whether there is a relationship between AQP2 levels in urine and AQP2 levels in apical plasma membrane. For this purpose, several protocols were applied that have all previously been characterized with respect to major changes in the levels of AQP2 in the apical plasma membrane (dDAVP treatment, thirsting, water loading, and different diabetes insipidus models). For these models, we have confirmed the changes in AQP2 labeling of the apical plasma membrane by applying immunoelectron microscopy of fixed kidneys.
Rats were treated with dDAVP subcutaneously and followed for 8 h. As shown in Figure 4, A and C, urinary AQP2 levels were 3.0 ± 0.7% of the AQP2 levels in total kidney of dDAVP-treated rats (Table 1). In comparison, urinary AQP2 levels were only 0.9 ± 0.3% of AQP2 levels in total kidney of control rats (note that this only represents collection during 8 h). The marked increase in urinary AQP2 levels in response to dDAVP treatment indicates that the dDAVP-induced increase in AQP in the apical plasma membrane levels is associated with increased urinary excretion of AQP2. (Multiple studies have previously shown that vasopressin or dDAVP induces a marked increase in AQP2 in the apical plasma membrane [12,13,14].) This is consistent with the view that AQP2 is excreted by an apical pathway (Figures 2 and 3).
Thirsting Increases both AQP2 Expression in Kidney and AQP2 Excretion into Urine
Thirsting is known to result in increases in AQP2 expression both in subapical vesicles and in the apical plasma membrane of collecting duct principal cells (7,16). To clarify whether urinary excretion of AQP2 parallels changes in AQP2 expression in kidney, rats were thirsted in metabolic cages for 48 h. Urine volume, urine osmolality, urinary AQP2 level, and renal AQP2 levels were determined as described in Table 2.
As shown in Figure 5A, thirsting for 48 h resulted in a marked increase in AQP2 levels and densitometry revealed a threefold increase (Figure 5B). This was associated with a decreased urine output in thirsted rats and a significant increase in urine osmolality (Table 2). Under these conditions, urine was collected during the second 24-h period and used for determination of urinary AQP2 excretion levels. Semiquantitative densitometry of immunoblots revealed that AQP2 levels in rat urine after thirsting was substantially increased to 15.1 ± 3.2% of levels in kidneys (Figure 6). AQP2 excreted in control rats averaged 3.2 ± 0.6% of levels in kidneys (Figure 6, B and C). Thus, urinary excretion of AQP2 was markedly increased in thirsted rats compared with control rats. Although urine AQP2 excretion indeed increases in parallel with increased kidney expression, the urinary excretion by far exceeds the increase in kidney levels. Since thirsting is known to increase endogenous vasopressin levels and apical plasma membrane AQP2 levels (7), this further supports a correlation between vasopressin, apical plasma membrane levels, and AQP2 urine excretion rats.
An approximately fourfold increase of AQP3 was also seen in rat kidneys (Figure 7, A and B) in response to thirsting for 48 h. However, urinary AQP3 levels were still undetectable in immunoblots (Figure 7C). This further supports the view that AQP2 was selectively excreted into rat urine via an apical pathway.
Water Administration to Previously Thirsted Rats Decreases Urinary AQP2 Excretion
To examine further whether urinary excretion of AQP2 parallels changes in levels of AQP2 in the apical plasma membrane, additional protocols were applied that are known to be associated with changes in plasma membrane levels of AQP2. In the first model, thirsted rats (48 h) were allowed free access to water for 24 h (11). This treatment has been shown to reduce AQP2 levels in the apical plasma membrane (11).
Urine output increased in response to water administration to previously thirsted rats (Table 3). As shown in Figure 8A, urinary AQP2 levels were markedly reduced after water administration as compared with the proceeding period during thirsting. Densitometry revealed an approximately threefold reduction in AQP2 urinary excretion (Figure 8B). This is consistent with the view that urinary excretion of AQP2 decreased in parallel with a reduction of AQP2 levels in the apical plasma membrane of collecting duct principal cells (11).
Very Low Urinary AQP2 Levels in Rat with No/Low Vasopressin Action
The above protocols strongly indicated that urine AQP2 levels are closely associated with vasopressin action to produce changes in levels of AQP2 in the apical plasma membrane rather than correlating closely with overall kidney AQP2 levels. To explore this further, we used two additional protocols that are associated with extreme urine production due to low or no vasopressin. In collecting duct, principal cells from Brattleboro rats (which lack vasopressin secretion and have extreme central diabetes insipidus ) have virtually no AQP2 in the apical plasma membrane (8,13,30), although they have quite high levels of AQP2 kidney expression corresponding to 30 to 50% of the levels in parent strain (Long-Evans rats) or Wistar rats (8). Even at high exposures, no detectable levels of AQP2 were observed in urine samples (Figure 9A). Moreover, in lithium-induced NDI, in which levels of AQP2 are also extremely low in the apical plasma membrane, AQP2 was not detected in urine (Figure 9B), as expected. Thus, in a steady-state condition with low vasopressin action but maintained relatively high levels of AQP2 kidney expression (30 to 50% of normal levels), urinary excretion of AQP2 was extremely low and thus much lower than what would have been predicted based on kidney levels. This strongly supports the view that urine AQP2 excretion closely parallels vasopressin action and hence levels in the apical plasma membrane.
AQP2 Is Associated with Vesicles in Kidney and in Rat Urine
As shown previously, AQP2 is predominantly localized in the apical plasma membrane and intracellular vesicles of collecting duct principal cells (7,36). In response to thirsting for 48 h, AQP2 is very abundant both in the apical plasma membrane, in small subapical vesicles, and in multivesicular bodies in collecting duct principal cells. As documented in previous studies (5,8,13,17,30), immunoelectron microscopy of AQP2 in lithium-treated rats or Brattleboro rats revealed extremely low levels of AQP2 in the apical plasma membrane (not shown), whereas rats subjected to acute dDAVP or vasopressin treatment had increased levels of AQP2 in the apical plasma membrane (not shown). Water administration for 24 h of rats that had been thirsted in the preceding 48 h also revealed reduced levels of AQP2 in the apical plasma membrane compared with rats thirsted for 48 h (not shown), as described previously (11).
To analyze the structures to which AQP2 was associated, urine samples of rat urine were subjected to immunolabeling and negative uranyl acetate staining. As shown in Figure 10, AQP2 was associated with vesicles with intact membrane structure. The labeled vesicles were of different sizes (Figure 10, A and B) and, as shown in Figure 10, C and D, AQP2 was also associated with larger membrane fragments and in structures resembling multivesicular bodies. Figure 10E shows that, in addition to AQP2, AQP1 was also found in membrane vesicles in rat urine samples. Thus, in urine AQP2 is associated with small vesicles, larger membrane fractions that may represent plasma membrane fragments, and in larger structures resembling intact or partly destroyed multivesicular bodies and/or vesicle aggregates.
Proteins in the urine of healthy individuals are: (1) plasma proteins that have been filtered at the glomerulus and subsequently escaped tubular reabsorption (37) and (2) kidney tubule-derived proteins that are excreted into urine. The latter are cytosolic proteins and membrane proteins from distinct segments of the renal tubule (38,39). Efforts have been made to isolate distinct proteins that are excreted in the urine and which could be used as valuable markers of kidney function or in identifying/diagnosing distinct kidney diseases (38,39). Previous clinical studies (27,28,29) have shown that AQP2 is excreted in urine and the excretion increases substantially in response to acute vasopressin treatment. In the same studies, it has been discussed whether urinary excretion of AQP2 would be a relevant/useful clinical or physiologic marker in water balance and water balance disorders. To fully answer that question, several issues must be addressed: (1) What are the cellular mechanisms underlying the excretion? (2) Does the excretion parallel total kidney expression? (3) Does the excretion parallel changes in the levels of AQP2 in the apical plasma membrane? These issues were addressed in the present study using several previously well-characterized rat models with major acute or chronic changes in water balance. The results demonstrated that the excretion of AQP2 takes place via an apical pathway different from whole cell shedding. The daily excretion corresponds to approximately 3 to 4% of total kidney levels, consistent with previous reports (40). Vasopressin treatment or thirsting (which induced endogenous vasopressin secretion) markedly increases the excretion, whereas water loading suppresses it. In conditions with severe nephrogenic or central diabetes insipidus in which AQP2 levels in the apical plasma membrane previously have been shown to be extremely low (confirmed in this study), AQP2 was undetectable in the urine. In these conditions with severe changes in water balance, AQP2 excretion closely follows conditions known to change AQP2 in the apical plasma membrane. In contrast, the excretion does not correlate with total kidney AQP2 levels under these conditions. Thus, the results strongly support the view that urinary excretion of AQP2 closely parallels changes in vasopressin action.
Urinary Excretion of AQP2 Occurs via a Selective Apical Pathway
A major question addressed in this study concerned the cellular mechanisms involved in AQP2 excretion into the urine. Two major pathways could be hypothesized: (1) excretion as a result of whole cell shedding into the collecting duct lumen and further excreted into the urine, or (2) a selective pathway including apical loss or apical release of AQP2 into the collecting duct lumen. To address this issue, urine was fractionated using sequential ultracentrifugation. After a 4,000 × g spin for 15 min, a procedure known to pellet whole cells and larger organelles such as mitochondria and nuclei, only very modest levels of AQP2 were found in the pellet (Figure 2). This argues against association of AQP2 with whole cells or fragments of cells and supports the view that a more selective mechanism is involved. However, at this stage it could not be completely excluded that whole principal cells were excreted but were then later dissolved or disintegrated within the bladder, although this remained very unlikely since only a very small fraction of urinary AQP2 was associated with the 4,000 × g pellet, which would include whole and partially disintegrated cells. To address this further, we tested the excretion of AQP2 being predominantly expressed in the apical plasma membrane and in subapical vesicles (7) and AQP3, which is located in the basolateral plasma membrane of the same collecting duct principal cells (31,41). Thus, AQP3 therefore represents an ideal basolateral marker and if whole cell shedding was underlying the excretion of AQP2, AQP3 also should be found in the urine in proportional amounts. The results revealed that AQP2, but not AQP3, was excreted into urine at a significant level. It was observed that even at long exposures in which strong levels of AQP3 (and AQP2) were found in kidney membranes, there was a disproportional excretion of AQP2 and AQP3 (the latter being undetectable). This was not due to a lower stability of AQP3, since storage of kidney membranes from inner medulla (with high expression of AQP3) in urine overnight maintained significant immunoreactivity. This is consistent with previous evidence that AQP2 is very stable even at storage in urine for several days (40). The disproportional excretion of AQP2 and AQP3 was further confirmed by long-term thirsting of rats, which resulted in a substantial increase in both AQP3 and AQP2 levels in the kidney as well as a solid increase in urinary excretion of AQP2, but not of AQP3 (Figures 6 and 7). These results support the view that urinary excretion of AQP2 takes place via a selective, apical pathway and not by whole cell shedding. Further evidence that urinary excretion of AQP2 is mediated by an apical pathway was provided by experiments using rats treated acutely with dDAVP or thirsted for 48 h (both conditions are known to increase apical plasma membrane levels of AQP2 due to translocation of AQP2 in intracellular vesicles to apical plasma membrane [7,8,12,14]) or rats that were water-loaded subsequent to prolonged thirsting. Urinary AQP2 levels were markedly increased in response to acute dDAVP administration, showing an association with the increased AQP2 levels in the apical plasma membrane. After long-term thirsting, AQP2 expression is increased in both intracellular vesicles and the apical plasma membrane of principal cells in collecting duct (7,17), and this resulted in increased excretion of AQP2 in the urine. In contrast, water loading (known to suppress endogenous release of vasopressin) caused a reduction in AQP2 levels in the plasma membrane by translocation into an intracellular reservoir (17). This procedure was associated with a decrease in urinary excretion of AQP2 and further supports the view than an apical pathway is involved. Therefore, all of these results strongly support the view that urinary excretion of AQP2 is mediated by an apical pathway that is tightly regulated by vasopressin.
It remains to be established how this apical excretion takes place. It may occur as a specific externalization of AQP2 (e.g., in an intracellular reservoir such as small multivesicular bodies or vesicles) in response to vasopressin. Alternatively, the excretion may be unspecific shedding of the apical plasma membrane. To explore these possibilities, we performed an analysis of the structures that AQP2 is associated with in the urine by immunogold labeling and negative staining of urine samples (described in Materials and Methods). AQP2 was found associated with small vesicle-like structures, with larger structures resembling multivesicular bodies, and also with larger membrane structures that may represent larger plasma membrane fragments.
The first hypothesis that vesicles and/or multivesicular bodies are specifically excreted in response to vasopressin is consistent with the observation that AQP2 is found to be associated with small vesicles and larger structures resembling multivesicular bodies in the urine. It is currently unknown what biologic role multivesicular bodies play in the collecting duct principal cell, although they have been shown to participate in endocytosis or exocytosis (42) and contain AQP2 as demonstrated previously (7). The turnover of multivesicular bodies is not known and their fate has not been described. The presence of multivesicular bodies in the urine (containing AQP2) indicates that they indeed may be shed.
Second, the excretion of AQP2 may relate to unspecific shedding of the apical plasma membrane. Since AQP2 density is increased in response to vasopressin treatment or in response to treatments that increase the release of endogenous vasopressin (like thirsting), an increased excretion would of course take place in case a constant fraction (i.e., constant rate) of apical plasma membrane is being lost by unspecific shedding. Third, the rate of unspecific shedding of the plasma membrane may increase in response to vasopressin. Additional studies are warranted to explore these possibilities.
Urinary AQP2 Levels versus Total Kidney AQP2 or Apical Plasma Membrane Levels
Approximately 3% of total kidney AQP2 is excreted daily in the urine in normal rats with steady-state water balance (Figure 1). This is consistent with previous reports by Rai et al. (40) in a study using an RIA. Apart from this study and the study by Rai et al., the other studies on urinary AQP2 excretion have been clinical (27,28,29).
As demonstrated, dDAVP treatment for 8 h (Figure 4) or thirsting for 48 h (Figure 6; which induces endogenous vasopressin secretion) markedly increases the excretion of AQP2 compared with time-matched control, whereas water loading for 24 h (Figure 8) suppresses it. This is consistent with studies in humans demonstrating that urinary excretion of AQP2 increased from 0.8 ± 0.3 to 11.2 ± 1.6 pmol per milligram of creatinine after dDAVP infusion (29). Elliot et al. (28) demonstrated that water loading induced a 76% decrease in urinary excretion of AQP2 and 3% sodium chloride produced a 760% increase in urinary excretion of AQP2. Thus, this is consistent with a close correlation between vasopressin action and excretion of AQP2 in the urine (both in rat and human). To further explore this finding, we used two well-described models with severe nephrogenic (lithium-induced NDI ) or central diabetes insipidus (Brattleboro rats [8,30]) in which AQP2 levels in the apical plasma membrane previously have been shown to be extremely low (as also confirmed in this study). In both of these models, AQP2 was undetectable in the urine. Thus, in conditions with lack of vasopressin action, a severe reduction in AQP2 excretion is noted.
With respect to a correlation between AQP2 kidney levels and AQP2 excretion in urine, several observations in the present study argue strongly against such a general correlation, although it cannot be ruled out that there may be a correlation under strict controlled or manipulated conditions. The urinary AQP2 excretion during 24 h increased to 15 ± 3% of AQP2 levels in the kidney after thirsting for 48 h (Figure 6), as compared to 3.2 ± 0.6% of AQP2 in the kidney of control animals. Thus, there is an increased excretion associated with the increased expression, but the excretion increases to a much higher degree consistent with an important role of vasopressin. Moreover, in the Brattleboro rats AQP2 expression is relatively high and constitutes approximately 30 to 50% of kidney levels in normal Wistar rats or Long-Evans rats (8) (D. Promeneur, B. M. Christsensen, J. Frøkiær, T-H. Kwon, and S. Nielsen, unpublished observations). In contrast, there is an almost complete lack of AQP2 in the apical plasma membrane (8,13,30). Because AQP2 was undetectable in the urine despite these quite high levels of AQP2 in the kidney cells, this speaks against a close correlation between kidney AQP2 levels and urine AQP2 excretion rates. Moreover, this finding also strongly favors an apical pathway (rather than whole cell shedding), which is vasopressin-regulated.
Thus, in these conditions with severe changes in water balance, AQP2 excretion closely follows conditions known to change AQP2 in the apical plasma membrane. In contrast, the excretion does not correlate with total kidney AQP2 levels under these conditions. Thus, the results strongly support the view that urinary excretion of AQP2 closely parallels changes in vasopressin action and may be used as a parameter for vasopressin action as suggested previously.
The authors thank Annette Blak Rasmussen, Gitte Christensen, Mette Vistisen, Zhila Nikrozi, and Annette Stockwell for expert technical assistance. Support for this study was provided by the Danish Research Academy, the Karen Elise Jensen Foundation, Novo Nordisk Foundation, Danish Medical Research Council, University of Aarhus Research Foundation, University of Aarhus.
American Society of Nephrology
aUrinary AQP2 excretion over a certain period (either 24 h or 8 h of urine collection) was compared with the kidney AQP2 protein levels at the end of the experimental period. Comparison of AQP2 excretion was done with time-matched controls; hence, values were not normalized to creatinine excretion to avoid inclusion of an unnecessary variable in the calculations.
1. Preston GM, Agre P: Isolation of the cDNA for erythrocyte integral membrane protein of 28 kilodaltons: Member of an ancient channel family. Proc Natl Acad Sci USA 88:11110–11114, 1991
2. Preston GM, Carroll TP, Guggino WB, Agre P: Appearance of water channels in Xenopus oocytes expressing red cell CHIP28 protein. Science 256:385–387, 1992
3. Agre P, Bonhivers M, Borgnia MJ: The aquaporins, blueprints for cellular plumbing systems. J Biol Chem273: 14659-14662,1998
4. Knepper MA: The aquaporin family of molecular water channels. Proc Natl Acad Sci USA 91:6255–6258, 1994
5. Nielsen S, Agre P: The aquaporin family of water channels in kidney. Kidney Int 48:1057–1068, 1995
6. Nielsen S, Marples D, Frøkiær J, Knepper M, Agre P: The aquaporin family of water channels in kidney: An update on physiology and pathophysiology of aquaporin-2. Kidney Int49: 1718-1723,1996
7. Nielsen S, DiGiovanni SR, Christensen EI, Knepper MA, Harris HW: Cellular and subcellular immunolocalization of vasopressin-regulated water channel in rat kidney. Proc Natl Acad Sci USA90: 11663-11667,1993
8. DiGiovanni SR, Nielsen S, Christensen EI, Knepper MA: Regulation of collecting duct water channel expression by vasopressin in Brattleboro rat. Proc Natl Acad Sci USA 91:8984–8988, 1994
9. Deen PM, Verdijk MA, Knoers NV, Wieringa B, Monnens LA, van Os CH, van Oost BA: Requirement of human renal water channel aquaporin-2 for vasopressin-dependent concentration of urine. Science264: 92-95,1994
10. Wade JB, Stetson DL, Lewis SA: ADH action: Evidence for a membrane shuttle mechanism. Ann NY Acad Sci372: 106-117,1981
11. Marples D, Knepper MA, Christensen EI, Nielsen S: Redistribution of aquaporin-2 water channels induced by vasopressin in rat kidney inner medullary collecting duct. Am J Physiol269: C655-C664,1995
12. Nielsen S, Chou CL, Marples D, Christensen EI, Kishore BK, Knepper MA: Vasopressin increases water permeability of kidney collecting duct by inducing translocation of aquaporin-CD water channels to plasma membrane. Proc Natl Acad Sci USA 92:1013–1017, 1995
13. Sabolic I, Katsura T, Verbavatz JM, Brown D: The AQP2 water channel: Effect of vasopressin treatment, microtubule disruption, and distribution in neonatal rats. J Membr Biol143: 165-175,1995
14. Yamamoto T, Sasaki S, Fushimi K, Ishibashi K, Yaoita E, Kawasaki K, Marumo F, Kihara I: Vasopressin increases AQP-CD water channel in apical membrane of collecting duct cells in Brattleboro rats. Am J Physiol 268:C1546 -C1551, 1995
15. Lankford SP, Chou CL, Terada Y, Wall SM, Wade JB, Knepper MA: Regulation of collecting duct water permeability independent of cAMP-mediated AVP response. Am J Physiol 261:F554 -F566, 1991
16. Terris J, Ecelbarger CA, Nielsen S, Knepper MA: Long-term regulation of four renal aquaporins in rats. Am J Physiol 271:F414 -F422, 1996
17. Marples D, Christensen S, Christensen EI, Ottosen PD, Nielsen S: Lithium-induced downregulation of aquaporin-2 water channel expression in rat kidney medulla. J Clin Invest95: 1838-1845,1995
18. Frøkiær J, Marples D, Knepper MA, Nielsen S: Bilateral ureteral obstruction downregulates expression of vasopressin-sensitive AQP-2 water channel in rat kidney. Am J Physiol270: F657-F668,1996
19. Frøkiær J, Christensen BM, Marples D, Djurhuus JC, Jensen UB, Knepper MA, Nielsen S: Downregulation of aquaporin-2 parallels changes in renal water excretion in unilateral ureteral obstruction. Am J Physiol 273:F213 -F223, 1997
20. Marples D, Frøkiær J, Dorup J, Knepper MA, Nielsen S: Hypokalemia-induced downregulation of aquaporin-2 water channel expression in rat kidney medulla and cortex. J Clin Invest97: 1960-1968,1996
21. Sands JM, Flores FX, Kato A, Baum MA, Brown EM, Ward DT, Hebert SC, Harris HW: Vasopressin-elicited water and urea permeabilities are altered in IMCD in hypercalcemic rats. Am J Physiol274: F978-F985,1998
22. Earm J-H, Christensen BM, Frøkiær J, Marples D, Han JS, Knepper MA, Nielsen S: Deceased aquaporin-2 water channel expression in kidney inner medulla of hypercalcemic rats. J Am Soc Nephrol 9:2181–2193, 1998
23. Fujita N, Ishikawa SE, Sasaki S, Fujisawa G, Fushimi K, Marumo F, Saito T: Role of water channel AQP-CD in water retention in SIADH and cirrhotic rats. Am J Physiol269: F926-F931,1995
24. Jonassen T, Nielsen S, Christensen S, Petersen JS: Decreased vasopressin-mediated renal water reabsorption in rats with compensated liver cirrhosis. Am J Physiol 275:F216 -F225, 1998.
25. Xu DL, Martin PY, Ohara M, St. John J, Pattison T, Meng X, Morris K, Kim JK, Schrier RW: Upregulation of aquaporin-2 water channel expression in chronic heart failure rat. J Clin Invest99: 1500-1505,1997
26. Nielsen S, Terris J, Andersen D, Ecelbarger C, Frøkiær J, Jonassen T, Marples D, Knepper MA, Petersen JS: Congestive heart failure in rats is associated with increased expression and targeting of aquaporin-2 water channel in collecting duct. Proc Natl Acad Sci USA 94:5450–5455, 1997
27. Deen PM, van Aubel RA, van Lieburg AF, van Os CH: Urinary content of aquaporin 1 and 2 in nephrogenic diabetes insipidus. J Am Soc Nephrol 7:836–841, 1996
28. Elliot S, Goldsmith P, Knepper M, Haughey M, Olson B: Urinary excretion of aquaporin-2 in humans: A potential marker of collecting duct responsiveness to vasopressin. J Am Soc Nephrol7: 403-409,1996
29. Kanno K, Sasaki S, Hirata Y, Ishikawa S, Fushimi K, Nakanishi S, Bichet DG, Marumo F: Urinary excretion of aquaporin-2 in patients with diabetes insipidus. N Engl J Med332: 1540-1545,1995
30. Grønbeck L, Marples D, Nielsen S, Christensen S: Mechanism of antidiuresis caused by bendroflumethiazide in conscious rats with diabetes insipidus. Br J Pharmacol 123:737–745, 1998
31. Ecelbarger CA, Terris J, Frindt G, Echevarria M, Marples D, Nielsen S, Knepper MA: Aquaporin-3 water channel localization and regulation in rat kidney. Am J Physiol 269:F663 -F672, 1995
32. Nielsen S, Marples D, Birn H, Mohtashami M, Dalby NO, Trimble M, Knepper M: Expression of VAMP-2-like protein in kidney collecting duct intracellular vesicles: Colocalization with aquaporin-2 water channels. J Clin Invest 96:1834–1844, 1995
33. Nielsen S, Pallone T, Smith BL, Christensen EI, Agre P, Maunsbach AB: Aquaporin-1 water channels in short and long loop descending thin limbs and in descending vasa recta in rat kidney. Am J Physiol 268:F1023 -F1037, 1995
34. Marples D, Schroer TA, Ahrens N, Taylor A, Knepper MA, Nielsen S: Dynein and dynactin colocalize with AQP2 water channels in intracellular vesicles from kidney collecting duct. Am J Physiol274: F384-F394,1998
35. Valtin H, Schroeder HA: Familial hypothalamic diabetes insipidus in rats (Brattleboro strain). Am J Physiol206: 425-430,1964
36. Fushimi K, Uchida S, Hara Y, Hirata Y, Marumo F, Sasaki S: Cloning and expression of apical membrane water channel of rat kidney collecting tubule. Nature 361:549–552, 1993
37. Galaske RG, Van LJ, Feld LG: Filtration and reabsorption of endogenous low-molecular-weight protein in the rat kidney. Kidney Int 16: 394-403,1979
38. Mantle D: Characterization of dipeptidyl and tripeptidyl aminopeptidases in human kidney soluble fraction. Clin Chim Acta 196:135–142, 1991
39. Scherberich JE, Wiemer J, Schoeppe W: Biochemical and immunological properties of urinary angiotensinase A and dipeptidylaminopeptidase IV: Their use as markers in patients with renal cell injury. Eur J Clin Chem Clin Biochem 30:663–668, 1992
40. Rai T, Sekine K, Kanno K, Hata K, Miura M, Mizushima A, Marumo F, Sasaki S: Urinary excretion of aquaporin-2 water channel protein in human and rat. J Am Soc Nephrol 8:1357–1362, 1997
41. Ishibashi K, Sasaki S, Fushimi K, Uchida S, Kuwahara M, Saito H, Furukawa T, Nakajima K, Yamaguchi Y, Gojobori T, Marumo F: Molecular cloning and expression of a member of the aquaporin family with permeability to glycerol and urea in addition to water expressed at the basolateral membrane of kidney collecting duct cells. Proc Natl Acad Sci USA 91:6269–6273, 1994
42. Nielsen S, Muller J, Knepper MA: Vasopressin- and cAMP-induced changes in ultrastructure of isolated perfused inner medullary collecting ducts. Am J Physiol 265:F225 -F238, 1993