The initial response of renal epithelial cells to the antidiuretic hormone arginine vasopressin (AVP) is an increase in cyclic AMP. By applying immunofluorescence, cell membrane capacitance and transepithelial water flux measurements we show that cAMP alone is sufficient to elicit the antidiuretic cellular response in primary cultured epithelial cells from renal inner medulla, namely the transport of aquaporin‐2 (AQP2)‐bearing vesicles to, and their subsequent fusion with, the plasma membrane (AQP2 shuttle). The AQP2 shuttle is evoked neither by AVP‐independent Ca2+ increases nor by AVP‐induced Ca2+ increases. However, clamping cytosolic Ca2+ concentrations below resting levels at 25 nM inhibited exocytosis. Exocytosis was confined to a slow monophasic response, and readily releasable vesicles were missing. Analysis of endocytic capacitance steps revealed that cAMP does not decelerate the retrieval of AQP2 from the plasma membrane. Our data suggest that cAMP initiates an early step, namely the transport of AQP2‐bearing vesicles towards the plasma membrane, and do not support a regulatory function for Ca2+ in the AQP2 shuttle.
In mammals, water homeostasis is regulated by epithelial cells of the renal collecting duct. Key components in the regulation of collecting‐duct osmotic water permeability (Pf) are the vasopressin V2 receptor (V2 R) and the water channel protein aquaporin‐2 (AQP2). This is evident from mutations in the human V2 R and AQP2 genes causing congenital nephrogenic diabetes insipidus (Rosenthal et al., 1992; Deen et al., 1994). In a first step, the antidiuretic hormone arginine vasopressin (AVP) binds to V2 R expressed in principal cells, which are the main epithelial cell type of the collecting duct in the inner medulla. Stimulation of V2 R leads to an increase in cyclic AMP levels. The subsequent phosphorylation of AQP2 by cyclic AMP‐dependent protein kinase (PKA) is followed by the exocytic insertion of AQP2 into the plasma membrane, resulting in an increase in Pf. This process might be accompanied by an inhibition of AQP2 endocytosis (Agre et al., 2000). However, the relative contributions of exocytosis and endocytosis to Pf changes and the molecular mechanisms controlling them are not known. Two recent reports suggest that both an increase in cAMP levels and in the cytosolic Ca2+ concentration ([Ca2+])i are required for the AQP2 shuttle and the accompanying increase in Pf (Chou et al., 2000; Yip, 2002). We found that cAMP is the sole trigger of the AQP2 shuttle and that cytosolic Ca2+ is not able to promote or evoke the AQP2 shuttle in primary cultured inner‐medulla collecting‐duct cells from rat kidney (IMCD cells); in this, IMCD cells differ from other systems that display slow Ca2+‐independent exocytosis (Hille et al., 1999).
Ca2+ transients and the AQP2 shuttle
The basal [Ca2+]i level in IMCD cells was 187 ± 15 nM (number of cells n = 45; mean ± s.e.m.). AVP‐stimulation increased [Ca2+]i by 108 ± 14 nM (n = 45) (Fig. 1A, left panel) in 80% of the cells. Of the responding cells, 64% exhibited a time‐invariant increase in [Ca2+]i, in 13% [Ca2+]i decayed to baseline within 60 s, and in 23% [Ca2+]i oscillations (frequencies 0.025–0.01 Hz) were observed. In all cells, [Ca2+]i was distributed homogenously. The AVP‐stimulated AQP2 shuttle is shown in Fig. 1A (right panel).
To elucidate the role of Ca2+ in this shuttle, [Ca2+]i was clamped to 50 nM (Fig. 1B, left panel), 25 nM (Fig. 1C, left panel) and resting levels (150 nM) (Fig. 1D, left panel). Within the resolution limit of the fluorescence microscope, no subdomains with higher [Ca2+]i were observed. Subsequent stimulation with AVP failed to induce the AQP2 shuttle only at a [Ca2+]i of 25 nM (Fig. 1C, right panel). In all other cases, the AQP2 redistribution was fully preserved (Fig. 1B, D, right panel, and E). [Ca2+]i ≫ 1 μM, achieved by treatment of cells with 10 μM ionomycin in the presence of 1.8 mM extracellular CaCl2, neither initiated the AQP2 shuttle in the absence of AVP nor promoted the AVP‐induced shuttle (data not shown).
Capacitance measurements to monitor exocytosis
Fusion of vesicles with the plasma membrane increases electrical cell membrane capacitance, Cm. Cm for unstimulated IMCD cells in the monolayer was 12.1 ± 0.6 pF (n = 48). Cm for isolated cells, obtained by pretreatment of the monolayer for 15–30 min with a Ca2+‐free and Mg2+‐free external solution containing 1 mM EGTA, was similar (11.8 ± 0.9 pF; n = 17). Thus, significant electrical coupling between cells in the monolayer did not occur.
Fast application of AVP increased Cm by 1.1 ± 0.015 pF, or 9.1 ± 1.4% (n = 10) (Fig. 2A and C). Fusion of vesicles larger than 280 nm in diameter yielded stepwise increases in Cm (more than 2.5 fF; Fig. 2A, box). The fusion of smaller vesicles led to smooth increases in Cm, accounting for 77% of the total increase. Treatment of IMCD cells with the V2 R antagonist SR121463A (supplied by Claudine Serradeil‐Le Gal (Serradeil‐Le Gal et al., 1996)) for 15 min before the application of AVP inhibited the response to AVP (Fig. 2A and C), indicating the involvement of V2 R. The effect of extracellularly applied AVP was mimicked by an intracellular application of cAMP through the patch pipette (Fig. 2A and C).
When the weakly Ca2+‐buffering internal pipette solution (IS1) was replaced by the strongly Ca2+‐buffering internal solution (IS2, containing 40 nM free Ca2+), the increase in Cm in response to cAMP was similar to that observed with IS1 (Fig. 2C). The data show that increases in [Ca2+]i are not required for triggering exocytosis in IMCD cells.
Besides aquaporins, which (at pH 7.4) are known to exclude ions (Pohl et al., 2001), cAMP upregulates epithelial Na+ channels. However, the lack of increase in plasma membrane conductance, Gm, (Fig. 2A, box) suggests that neither plentiful Na+‐channel insertion into the plasma membrane nor activation of already inserted channels occurred. The result is consistent with the finding that the abundance of epithelial Na+ channels in non‐overexpressing cells is too low to determine whether cAMP alters their cell surface expression (Snyder, 2002).
Microelectrode‐based Pf measurements
If the cAMP‐triggered exocytosis reflects the fusion of AQP2‐bearing vesicles with the plasma membrane, the time courses of increases in Cm and Pf should be identical. In agreement with this prediction, forskolin increased Pf from 13 ± 2 to 26 ± 3 μm s−1 with a rate similar to that of Cm (Fig. 2B). Steady‐state measurements (n = 20; Fig. 3A) revealed the same result. Bis‐(o‐aminophenoxy)ethane‐N,N,N′,N′‐tetraacetic acid acetoxymethyl ester (BAPTA‐AM) did not reduce the incremental Pf (Fig. 3A). Forskolin also doubled Pf in the presence of BAPTA‐AM, whereas the PKA inhibitor H89, used as a control agent, abolished the stimulating effect of forskolin (Fig. 3B). These results confirm that an increase in [Ca2+]i does not promote the AQP2 shuttle.
Triggering of exocytosis by flash photolysis of caged cAMP
IMCD cells required 24 ± 6 s (n = 10) or 55 ± 19 s (n = 10) to respond with an increase in Cm to AVP or cAMP, respectively (Figs 2A, C and 4D). To exclude diffusion restrictions, we employed a new type of caged cAMP, the [6,7‐bis(carboxymethoxy)coumarin‐4‐yl]methyl (BCMCM) ester of 8‐Br‐cAMP (BCMCM‐caged 8‐Br‐cAMP), which, after flash photolysis, liberates within a few nanoseconds the active compound 8‐Br‐cAMP and a fluorescent, inactive photoproduct, 6,7‐bis(carboxymethoxy)‐4‐(hydroxymethyl)coumarin (BCMCM‐OH) (Hagen et al., 2001). In the presence of strongly (IS2) and weakly (IS1) Ca2+‐buffering solutions in the patch pipette, increases in Cm followed the ultraviolet flash with a delay of 8 ± 2 s (n = 10) and 11 ± 4 s (n=4), respectively, whereas photoproduct fluorescence confirmed 8‐Br‐cAMP release within milliseconds (Fig. 4A, B and C). Although, significantly shortened (Fig. 4D), the persistence of the delay indicates that a pool of readily releasable vesicles is missing from IMCD cells.
AQP2 immunogold electron microscopy (EM)
In resting IMCD cells, AQP2 was found almost exclusively in vesicles (Fig. 5A, B and C). Most AQP2‐bearing vesicles (99%) were found more than 50 nm away from the plasma membrane (Fig. 5D); 15% were 280–850 nm in diameter (Fig. 5C). The size distribution was in reasonable agreement with that calculated from Cm steps (Fig. 5C). After stimulation with forskolin, the gold particles decorated the apical and basolateral membranes (Fig. 5E and F).
Histogram analyses of exocytic and endocytic steps
In resting cells, exocytosis was balanced by endocytosis (Fig. 6A). Whereas the number of endocytic events remained nearly unchanged after stimulation with cAMP or AVP, the number of exocytic events clearly increased (Fig. 6B and C). Thus, at least within 10 min after application of the stimulus, IMCD cells responded with a selective increase in exocytic events.
The present study shows that the AQP2 shuttle in renal epithelial (principal) cells is exclusively triggered by cAMP. It is neither evoked nor promoted by increasing [Ca2+]i. However, clamping [Ca2+]i at 25 nM abolished the AQP2 shuttle, indicating a requirement for Ca2+ at very low levels. Although there is no direct proof that only the fusion of AQP2‐bearing vesicles contributes to the observed increase in Cm, the similarity of time courses of increases in Pf and Cm together with immunofluorescence and EM data strongly suggest that increases in Cm reflect exocytosis of AQP2‐bearing vesicles. The delay and slow kinetics of both increases in Cm and Pf as well as the distance of AQP2‐bearing vesicles from the plasma membrane are consistent with the view that cAMP initiates the long‐range targeting of AQP2 to the plasma membrane. It remains possible that—besides the distance from the plasma membrane—the lack of the release readiness of vesicles (Kasai, 1999) contributes to the slow response. The release readiness and fusion of vesicles might not be regulated by cAMP. At present we cannot exclude a role of low levels of [Ca2+]i at these final steps of the AQP2 shuttle.
In contrast with our results, elimination of increases in [Ca2+]i in isolated collecting ducts from rat kidney was reported to inhibit the AVP‐induced AQP2 shuttle and the accompanying increase in Pf (Chou et al., 2000; Yip, 2002). In collecting ducts, a [Ca2+]i transient lasting 100 s seemed to be important for an increase from a resting value of 100 μm s−1 to 300 μm s−1 that was reached after 10–20 min. There is at present no explanation for the unexpectedly high resting Pf because (1) water moves primarily transcellularly (Kovbasnjuk et al., 1998) and (2) the number of water channels in the apical membrane of resting cells is rather low (Nielsen et al., 1995). Thus, in the resting collecting duct, water transport is limited by the Pf of the apical epithelial membrane, which, according to measurements on bilayers reflecting its composition, is 5 μm s−1 (Hill & Zeidel, 2000; Krylov et al., 2001). This value is in good agreement with our Pf of the resting IMCD monolayer and with the Pf of single IMCD cells obtained by laser‐scanning reflection microscopy (Maric et al., 2001).
The AVP‐induced increase in the Pf of principal cells is due to enhanced exocytosis. Within the first 10 min, AVP does not decelerate endocytosis and hence does not prolong the persistence of AQP2 in the plasma membrane, if one assumes that decreases on Cm reflect the endocytosis of AQP2‐bearing vesicles. Our data support the hypothesis that recruitment of AQP2 to the plasma membrane and its retrieval into a pool of intracellular vesicles are independently regulated events (Zelenina et al., 2000).
The number of cellular systems showing a slow exocytic response while lacking a fast response is very limited. As well as principal cells, they include insulin‐responding cells (adipocytes and skeletal muscle cells) and pancreatic exocrine acinar cells activated by cholecystokinin or acetylcholine. However, the signalling cascades (Yang et al., 2002) leading to slow exocytic responses in these cells differ fundamentally from the cascade inducing the AQP2 shuttle in principal cells. Given the differences of regulated exocytosis in principal cells and in other cells, namely the triggering signal molecule (solely cAMP, solely Ca2+ or a synergistic action of both) and the kinetics (exclusively slow, exclusively fast or biphasic kinetics), the AQP2 shuttle differs from known types of exocytosis. A major challenge for future research is the identification and characterization of the cellular compartment in which cAMP initiates AQP2 targeting to the plasma membrane.
IMCD primary cultures (Maric et al., 1998) were used from days 4 to 8 after seeding.
Ratiometric measurements of [Ca2+]i.
IMCD cells in the monolayer were incubated (45 min) in 5 μM fura‐2/AM and 0.02% Pluronic F 127 (both from Molecular Probes, Leiden, The Netherlands). Calculation of [Ca2+]i (Lorenz et al., 1998) was possible in 20% of the cells that showed sufficient fura‐2 uptake.
For the detection of AQP2, confocal laser‐scanning microscopy was performed with a specific anti‐rat AQP2 antibody and a Cy3‐conjugated anti‐rabbit secondary antibody (Maric et al., 1998).
Cell membrane capacitance measurements.
IMCD cells were transferred to the external solution (in mM: 140 NaCl, 5 KCl, 2 CaCl2, 1 MgCl2, 10 HEPES, 15 glucose; pH 7.2; 298 mOsm kg−1) 15 min before measurement. The standard pipette solution 1 (IS1) contained (in mM) 135 potassium glutamate, 22 NaCl, 1 MgCl2, 10 HEPES, 0.2 MgATP, 0.3 NaGTP, 0.2 EGTA (pH 7.2; 295 mOsm kg−1). Pipette solution 2 (IS2) was similar to IS1 but was adjusted to 40 nM free Ca2+ with 5 mM EGTA and 1 mM CaCl2.
Cm, series (Gs, 4–15 MΩ) and membrane (Gm) conductances were measured at 32–35 °C in the whole‐cell configuration with an EPC‐9 amplifier with the lock‐in extention to Pulse software (Heka Elektronik, Lambrecht, Germany). Sinusoid stimuli (794 Hz, 60 mV peak‐to‐peak; 20–40 ms duration) were superimposed over the d.c. holding potential of −50 mV. Currents were sampled at 8 kHz and filtered at 1.6 kHz. The generated Cm, Gs and Gm were averaged (time resolution 8–11 Hz).
Pf of a confluent epithelial monolayer.
Water, passing through the IMCD cell monolayer grown on a Transwell filter, diluted K+ ions in the basolateral compartment that contained the osmolyte (0.4 M sorbitol). Pf was derived from microelectrode‐aided measurements of (1) the steady‐state K+ concentration distribution adjacent to the basolateral membrane (Pohl et al., 1997) and (2) the time course of K+ concentration changes 20 μm away from the epithelium (see Supplementary Information).
Flash photolysis of BCMCM‐caged 8‐Br‐cAMP.
BCMCM‐caged 8‐Br‐cAMP was synthesized by analogy with BCMCM‐caged cAMP (Hagen et al., 2001). Flash photolysis liberated 8‐Br‐cAMP and the fluorescent compound BCMCM‐OH (see Supplementary Information).
IMCD cells grown on Transwell‐polycarbonate filters (Costar, Bodenheim, Germany) were fixed and permeabilized (Maric et al., 1998). For immunogold labelling, monolayers were incubated (16 h, 4 °C) with specific anti‐rat AQP2 antibody (diluted 1:600), followed by washing and incubation (5 h, 37 °C) with secondary gold‐labelled (1 nm) anti‐rabbit antibody (diluted 1:200; British BioCell International, Cardiff, UK). Cells were postfixed (1% glutaraldehyde, 10 min) and treated with OsO4 (2%, 1 h). After silver enhancement of gold particles (Hacker et al., 1988), the samples were embedded in Epon 812, sectioned (60 nm) and stained with uranyl acetate and lead citrate before examination with a Zeiss 902 A electron microscope at 80 kV (Lorenz et al., 1998).
Supplementary data are available at EMBO reports Online (http://www.emboreports.org).
We thank E. Klussmann and B. Wiesner for suggestions, and A. Geelhaar, B. Oczko and M. Ringling for technical assistance. Financial support was provided by the Deutsche Forschungsgemeinschaft (Ro 597/6‐3, Po 533/7‐1), the EU (QLRT‐2001‐02149) and the Fonds der Chemischen Industrie.
- Copyright © 2003 European Molecular Biology Organization