Aquaporin-1 Channel Function Is Positively Regulated by Protein Kinase C*

Aquaporin-1 (AQP1) channels contribute to osmotically induced water transport in several organs including the kidney and serosal membranes such as the peritoneum and the pleura. In addition, AQP1 channels have been shown to conduct cationic currents upon stimulation by cyclic nucleotides. To date, the short term regulation of AQP1 function by other major intracellular signaling pathways has not been studied. In the present study, we therefore investigated the regulation of AQP1 by protein kinase C. AQP1 wild type channels were expressed in Xenopus oocytes. Water permeability was assessed by hypotonic challenges. Activation of protein kinase C (PKC) by 1-oleoyl-2-acetyl-sn-glycerol (OAG) induced a marked increase of AQP1-dependent water permeability. This regulation was abolished in mutated AQP1 channels lacking both consensus PKC phosphorylation sites Thr157 and Thr239 (termed AQP1 ΔPKC). AQP1 cationic currents measured with double-electrode voltage clamp were markedly increased after pharmacological activation of PKC by either OAG or phorbol 12-myristate 13-acetate. Deletion of either Thr157 or Thr239 caused a marked attenuation of PKC-dependent current increases, and deletion of both phosphorylation sites in AQP1 ΔPKC channels abolished the effect. In vitro phosphorylation studies with synthesized peptides corresponding to amino acids 154–168 and 236–250 revealed that both Thr157 and Thr239 are phosphorylated by PKC. Upon stimulation by cyclic nucleotides, AQP1 wild type currents exhibited a strong activation. This regulation was not affected after deletion of PKC phosphorylation sites in AQP1 ΔPKC channels. In conclusion, this is the first study to show that PKC positively regulates both water permeability and ionic conductance of AQP1 channels. This new pathway of AQP1 regulation is independent of the previously described cyclic nucleotide pathway and may contribute to the PKC stimulation of AQP1-modulated processes such as endothelial permeability, angiogenesis, and urine concentration.

The aquaporins (AQPs) 3 are a family of integral membrane proteins that are widely expressed in bacteria, plants, and animals (1). Their main physiological function is rapid transmembrane transport of water driven by osmotic gradients. In addition to water transport, some aquaporins also conduct ion currents. In humans, 11 aquaporins have been found so far (AQP0 -AQP10). Of these, AQP1 is the predominant and least specialized subtype. It plays a major role in constitutive water transport through the membranes of several cell types including endothelial cells, red blood cells, and renal proximal tubule cells (1). It has been shown that AQP1 may also function as a cyclic nucleotide-gated cation channel that is activated mainly by cGMP and indirectly also by cAMP (2,3). Recently, ion currents of native aquaporins were confirmed in choroid plexus epithelium and shown to modulate fluid transport of those cells (4).
Because of its significance for determining endothelial water permeability, AQP1 has been found to play a major physiological role in the peritoneal membrane (5). Subsequently, it has been shown to be the molecular correlate of the "ultrasmall pore" responsible for transcellular water permeability during peritoneal dialysis (6). Furthermore, recent reports have demonstrated a previously unexpected role of AQP1 in cell migration (7)(8)(9). Endothelial cells lacking AQP1 have impaired cell motility because of reduced formation of lamellipodia, resulting in impaired angiogenesis (7). Interestingly, AQP1 was also found to be essential for normal migration of renal proximal tubule cells and restitution of renal injury (8,9).
It is well documented that aquaporin channels may be subject to intense short term regulation by cellular signal cascades, with the most recognized example being the cAMP-dependent protein kinase-dependent regulation of AQP-2 in the kidney collecting duct (1,10). Interestingly, however, little is known to date about the short term regulation of AQP1, because this channel was originally considered to be constitutively open (1). The role of signaling pathways apart from cyclic nucleotides in the regulation of AQP1 has not been investigated to date.
The protein kinase C (PKC) system is a key component of intracellular signaling, and PKC activation is a central pathway downstream of G q/11 coupled receptors such as adrenergic ␣ 1 receptors and muscarinergic M 1 receptors (11,12). PKC has been shown to be involved in the regulation of structurally diverse membrane proteins, particularly ion channels (13,14).
Here, we show that PKC positively regulates AQP1 channels expressed in Xenopus oocytes. PKC activation induces an increase of both water permeability and ion currents mediated by AQP1. On the molecular level, the effect depends on both consensus PKC phosphorylation sites Thr 157 and Thr 239 with approximately half of the overall effect being attributable to each site.

EXPERIMENTAL PROCEDURES
Solutions and Drug Administration-Two microelectrode voltage clamp measurements of Xenopus oocytes were performed in isotonic NaCl saline containing (in mM) 100 NaCl, 2 KCl, 4.3 MgCl, and 5 HEPES, pH 7.3. Ca 2ϩ was omitted from the bath solution to minimize Ca 2ϩ -dependent Cl Ϫ currents in the oocytes as described by Barish (25). Current and voltage electrodes were filled with 3 M KCl solution. OAG (1-oleoyl-2acetyl-sn-glycerol), staurosporine, and chelerythrine (all from Calbiochem) were dissolved in Me 2 SO to stock solutions of 10 mM and stored at Ϫ20°C. Aliquots of the stock solutions were diluted to the desired concentration with the bath solution on the day of experiments. The maximum concentration of Me 2 SO in the bath had no effects on the measured currents. All of the measurements were performed at room temperature (20°C).
Electrophysiology and Data Analysis-The two microlelectrode voltage clamp configuration was used to record currents from Xenopus laevis oocytes as published previously (13). Pclamp software (Axon instruments) was used for generation of the voltage pulse protocols and for data acquisition. The statistical data are presented as the means Ϯ S.E. Statistical significance was evaluated using Student's t test for pairwise comparison and analysis of variance for comparison of several groups of data. The differences were considered to be significant if the p value was Ͻ0.05 and highly significant if the p value was Ͻ0.01. In the figures, significance levels are indicated by asterisks (*,p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001). All of the results were obtained in oocytes from at least three different batches.
Analysis of Osmotic Swelling-Water permeability assays at room temperature (20°C) were initiated with the transfer of AQP1-expressing or control oocytes at time zero from 200 mosm ND96 saline into 100 mM mosm ND96 diluted with distilled water as described by Anthony et al. (2). After the transfer of the cell, the images were recorded at intervals of 30 s for 5 min. Volume changes were analyzed with ImageJ on the basis of digital images captured every 30 s with a digital Nikon Coolpix 5400 camera on a Nikon SMZ 1500 stereomicroscope. The data were analyzed as proportional change in volume and normalized to the initial volume at time zero. All of the results were obtained in oocytes from at least three different batches.
Site-directed Mutagenesis-Sequence analysis revealed two potential PKC phosphorylation residues at positions Thr 157 and Thr 239 . To eliminate PKC-mediated phosphorylation at these positions, the two threonine residues were replaced with alanine. This resulted in the mutated channels T157A and T239A. Through repetitive mutagenesis, both point mutations were introduced into a single clone termed AQP1-⌬PKC. Point mutations were generated with the QuikChange protocol (Stratagene, La Jolla, CA). We used the primers CGTGCTG-GCTACTG CCGACCGGAGGCG (forward) and CGCCTC-CGGTCGGCAGTAGCCAGCACG (reverse) to introduce mutation T157A, and for mutation T239A, we used the primers CAGCAGTGACCTCGCAGACCGCGTGAAG (forward) and CTTCACGCGGTCTGC GAGGTCACTGCTG (reverse). All of the cDNAs used in this study were verified by complete sequencing (Sequence Laboratories Göttingen GmBH).
Expression of AQP1 Channels in Xenopus Oocytes-The human AQP1 wild type clone was a kind gift from Peter Agre (Baltimore, MD). cRNA was prepared from the corresponding cDNA (AQP1 WT, AQP1-T157A, AQP1-T239A, and AQP1-⌬PKC) with T3 RNA polymerase after linearization with SmaI using the mMESSAGE mMACHINE in vitro transcription kit (Ambion, Austin, TX). AQP1 cRNA of wild type and mutant channels (always at the same concentration of 20 ng/l) was injected into stage V and VI defolliculated oocytes using a Nanoject automatic injector (Drummond, Broomall, PA). The volume of injected cRNA solution was 46 nl/oocyte. The measurements were performed 2-3 days after expression.
Peptide Synthesis-Peptides corresponding to amino acid sequences 154 -168 (P154 -168, LATTDRRRRDLGGSG, with a single conservative replacement of alanine by glycine at residue 168 to improve solubility) and to 236 -250 (P236 -250, SDLTDRVKVWTSGEV, with a single conservative replacement of glutamine by glutamate at residue 249 to improve solubility) were obtained from Sigma-Genosys. Homologous peptides containing threonine-to-alanine mutations at residues 157 and 239, respectively, were obtained from the same source and referred to as P154 -168(T157A) and P236 -250(T239A), respectively.
In Vitro Phosphorylation Assays-10 l (5 g) of peptide and 25 ng of active PKC (Upstate Cell Signaling Solutions, Lake Placid, NY) were incubated at 30°C for 15 min in the presence of ADBII, which was composed of 20 mM MOPS, pH 7.2, 25 mM ␤-glycerol phosphate, 1 mM sodium orthovanadate, 1 mM dithiothreitol, 1 mM CaCl 2 ), 10 l of lipid activator (1 mg/ml phosphatidylserine, 0.1 mg/ml diacylglycerol, 0.3% Triton X-100, 1 mM dithiothreitol), and 20 Ci of [␥-32 P]ATP (GE Healthcare, Munich, Germany) including 15 mM MgCl 2 and 200 M unlabeled ATP in ADBII. The reactions were stopped by adding 4ϫ SDS sample buffer, and the mixture was boiled for 5 min. The peptides were resolved by 20% SDS-polyacrylamide gel electrophoresis. The gels were dried and subjected to autoradiography using Hyperfilm MS film (GE Healthcare, Munich, Germany). The bands were quantified densitometrically with ImageJ on the basis of digitalized images.
Oocyte Membrane Isolation and Western Blotting Analysis-Noninjected and AQP1-cRNA-injected (200 ng/l) Xenopus oocytes were used to isolate crude membrane fractions. 23 oocytes were incubated in 1000 l of ice-cold hypotonic phosphate buffer (7.5 mM Na 2 HPO4, 1 mM EDTA, pH 7.5, plus a protease inhibitor mixture tablet (Roche Applied Science)) for 5 min. All of the steps were done at 4°C. The samples were vortexed and pipetted repeatedly to lyse the oocytes. The yolk and cellular debris were pelleted at 500 ϫ g for 5 min. The membranes were then pelleted at 48,000 ϫ g for 30 min. The membrane pellets were resuspended in 30 l of 1ϫ SDS protein loading buffer, incubated at 37°C for 10 min, and electrophoresed on a 12% polyacrylamide SDS gel. Afterward they were electrophoretically transferred onto nitrocellulose filters (Whatman, Dasel, Germany). The blots were blocked for 1 h with TBS-T (10 mM Tris, pH 7.4, 138 mM NaCl, 0.05% Tween 20) containing 3% nonfat dry milk. The blots were then incubated with AQP1 primary antibody (1:1000; Santa Cruz Biotechnology, Heidelberg, Germany) in 3% milk at 4°C overnight. The blots were washed three times for 15 min with TBS-T and incubated with horseradish peroxidase-conjugated anti-mouse secondary antibody (1:2000) (Cell Signaling Technology, Frankfurt, Germany) for 1 h at room temperature and were washed again three times. After washing, immune complexes were visualized using enhanced chemiluminescence (ECL; GE Healthcare, Munich, Germany).

Protein Kinase C Activation Increases Water Permeability of
Aquaporin-1 Channels-Human AQP1 channels were expressed heterologously in Xenopus oocytes to allow measurement of water permeability and ion conductance. First, water permeability of the membrane of those cells was assessed by hypotonic swelling experiments according to Anthony et al. (2). The duration of these experiments had to be limited to 5 min (as in other studies (2)) to avoid cell destruction as a consequence of the volume increases. To activate PKC in a group of experiments, specific PKC activator OAG (a synthetic and more stable analogue of the physiological activator diacylglycerol) was added to the hypo-osmotic solution at a concentration of 10 M. No preincubation of cells in OAG was performed. Summary data of the time course of volume change in those cells is plotted in Fig. 1A. Within 5 min, cell volume increased to 131.1 Ϯ 2.6% (n ϭ 8; p Ͻ 0.01 in comparison with control experiments without OAG). Without addition of OAG to the hypoosmotic solution, cells expressing AQP1 exhibited a smaller volume increase to 115.3 Ϯ 1.9% (n ϭ 6; Fig.  1A; significantly different from the effect of OAG with p Ͻ 0.01). Cells that did not express AQP1 did not show any relevant volume change (100.3 Ϯ 0.3%; n ϭ 7; Fig. 1A; p Ͻ 0.001 in comparison with cells expressing AQP1).

Regulation of Aquaporin-1 Water Permeability by PKC Is Abolished in AQP1-⌬PKC Channels Lacking Phosphorylation Sites Thr 157 and
Thr 239 -Sequence analysis of AQP1 channel amino acid sequence revealed two consensus sites for PKC phosphorylation at Thr 157 and Thr 239 , respectively. To elucidate the functional role of those sites, we modified channel subunits by sitedirected mutagenesis. The respective threonin residue was replaced by alanine to abolish phosphorylation at the respective site. Channels lacking both PKC phosphorylation sites (i.e. including mutations T157A and T239A) were termed AQP1-⌬PKC and measured under conditions analogous to those described for wild type channels.
The results are summarized in Fig. 1B. After 5 min of exposure to the hypo-osmotic solution, the volume of oocytes expressing AQP1-⌬PKC channels increased to 109.8 Ϯ 1.1% of the respective initial values (Fig. 1B, left column, n ϭ 5). Then experiments were repeated, and OAG (10 M) was added to the hypo-osmotic solution to activate PKC. In contrast to the effect observed in AQP1 wt channels, volume increase was not enhanced by PKC activation in AQP1-⌬PKC channels. After the observation period of 5 min, the volume of those cells increased to 110.4 Ϯ 0.6% (Fig. 1B, right column, n ϭ 5), i.e. values without significant difference to those observed without application of OAG (p Ͼ Ͼ 0.05). Therefore, we concluded that inactivation of both PKC phosphorylation sites abolishes the positive regulation of AQP1 water permeability by protein kinase C.
Protein Kinase C Activation Increases Aquaporin-1 Ion Currents-It has been demonstrated previously that AQP1 channels also conduct ions and that this conductance is activated by an increase in intracellular cyclic nucleotide levels, particularly cGMP and potentially also cAMP (2,3). Hence, we examined whether the PKC system also regulates AQP1 ion currents.
A standardized voltage protocol was used to elicit AQP1 currents in Xenopus oocytes heterologously expressing these channels. From a holding potential of Ϫ30 mV that is close to the Oocytes that did not express AQP1 completely lacked cell swelling (white squares; n ϭ 7). In contrast, cells expressing AQP1 exhibited a time-dependent increase in cell volume as a consequence of water permeability (circles; n ϭ 6). Activation of PKC by OAG in AQP1-expressing cells induced a markedly stronger cell swelling, indicating a marked increase in water permeability (squares; n ϭ 8; p Ͻ 0.001). OAG was added to the hypo-osmotic medium; no preincubation was performed. Channels lacking both PKC phosphorylation sites (i.e. including mutations T157A and T239A) were termed AQP1-⌬PKC and measured under analogous conditions. Summary data of those experiments are shown in B. Under control conditions (i.e. without application of OAG), AQP1-⌬PKC channels induced an increase of relative cell volume in the hypo-osmotic medium by 9.8%. The addition of OAG to the medium did not induce any significant effect in those channels, resulting in a respective volume increase by 10.4% (n ϭ 5 each, p Ͼ Ͼ 0.05).
reversal potential of AQP1 currents in low K ϩ solution, voltage steps to potentials from Ϫ110 mV to ϩ60 mV (400 ms each) were applied. After recording a measurement under control conditions, phorbol ester PMA (100 nM) was perfused into the bath for 30 min to activate PKC. In contrast to the water permeability measurements, the electrophysiological experiments did not affect the integrity of the cells and could therefore be extended to an observation period of 30 min (instead of 5 min) that allowed a longer observation of the time courses. Measurements were recorded at intervals of 5 min until the end of the observation period. Outward current amplitudes during the step to ϩ60 mV were determined and compared with quantify effects. Typical recordings under control conditions and after 30 min of exposure to PMA are shown in Fig. 2. Under control conditions, the cells merely exhibited small currents ( Fig. 2A) with a reversal potential of approximately Ϫ30 mV and lack of inward or outward rectification (Fig. 2C). Application of PMA induced a marked increase of currents (Fig. 2B) without affecting reversal potential and rectification (Fig. 2C). During observation of the PMA effect, the currents increased exponentially and reached a plateau after ϳ20 min (Fig. 2D). Overall, a relative increase of current amplitudes by 102.5 Ϯ 4.1% was observed ( Fig. 2E; n ϭ 10; p Ͻ 0.001 in comparison with control experiments). Control experiments were performed with oocytes that did not express AQP1. Those cells exhibited a merely small current increase (Fig. 2, D and E; n ϭ 7 Again, only a small current increase was observed in those cells (Fig. 2, D and E; n ϭ 9, 17.1 Ϯ 2.2%).
PMA is a well established and highly potent activator of PKC, but it is less specific than OAG. There- In cells expressing AQP1, application of PMA induced a strong current increase that reached a plateau after ϳ20 min (black squares; n ϭ 10). In contrast, no current increase was observed if no PMA was applied (black triangles; n ϭ 9) or if the cells did not express AQP1 (black circles; n ϭ 7). As shown in E, after the observation period of 30 min, PMA induced a relative current increase by 103% that was significantly different from the observations in the two control groups (p Ͻ 0.001 each). Protocol: holding potential Ϫ30 mV; test pulses from Ϫ120 mV to ϩ60 mV in 10-mV increments (400 ms). FIGURE 3. Protein kinase C activation by OAG induces an increase of ionic conductance of aquaporin-1 channels that is suppressed by chelerythrine. Because PMA is not entirely specific for PKC, confirmatory experiments were performed with the use of the less potent, but more specific PKC activator OAG. Characteristic recordings of AQP1 currents before and after exposure to OAG (10 M) for 30 min are shown in A and B. Corresponding I-V curves are plotted in C. The time course of effects is shown in D. OAG induced a marked current increase comparable with PMA. Co-application of PKC inhibitor chelerythrine (10 M) suppressed this effect. Equally, no effect was observed if no OAG was added to the solution or if the cells did not express AQP1. As shown in E, the OAG-induced effect after 30 min was significantly different from the observations in the control groups (p Ͻ 0.001 each). Protocol: holding potential Ϫ30 mV; test pulses from Ϫ120 mV to ϩ60 mV in 10-mV increments (400 ms). fore, we repeated the experiments using OAG instead of PMA to further confirm the role of PKC in the observed effect. Because OAG is less potent than PMA, a higher concentration of OAG had to be applied (10 M) as already used in previous studies from our laboratory (13). The experiments were performed as described for PMA. The results are shown in Fig. 3. OAG also induced a marked current increase without affecting biophysical current characteristics (Fig. 3, A-C). Time course of effect was comparable with that seen with PMA (Fig. 3D). Overall, currents increased by 117.9 Ϯ 3.2% ( Fig.  3E; n ϭ 12; p Ͻ 0.001 in comparison with control experiments). Co-application of PKC inhibitor chelerythrine (10 M) suppressed the effect almost completely, resulting in a relative current increase by merely 35.6 Ϯ 3.2% (Fig. 3, D and E; n ϭ 9). Control experiments either without exposure to OAG or without expression of AQP1 yielded current increases of 17.1 Ϯ 2.2% (n ϭ 9) and 21.9 Ϯ 3.6% (n ϭ 7), respectively (Fig. 3, D and  E). The small current increases in all of the control experiments described in this study were stable and similar effects of statis-tical significance when compared with the respective initial values (p Ͻ 0.01 each). Their relative amplitude was within the range of nonspecific current increases that are commonly observed in the Xenopus oocyte expression system (26 -28). Sets of corresponding control experiments are shown in each figure together with the observed effects (Figs. 2, D and E; 3, D and E; 4, D and E; and 5, D and E).
Phosphorylation Sites Thr 157 and Thr 239 Are Essential for Protein Kinase C Regulation of Aquaporin-1 Currents-Given that PKC regulation of AQP1 water permeability was found to depend on the availability of phosphorylation sites Thr 157 and Thr 239 , we investigated the role of these residues for the PKC-induced current increase. Mutant channels AQP1-T157A, AQP1-T239A, and AQP1-⌬PKC (exhibiting both mutations in combination) were measured separately under conditions analogous to those described for wild type channels. OAG (10 M) was used to activate PKC.
In Fig. 4 (A-C), typical recordings of AQP1-⌬PKC currents before and after exposure to OAG are displayed. Biophysical current properties were indistinguishable from the wild type. However, in contrast to wild type channels, AQP1-⌬PKC channels did not exhibit the strong activation induced by OAG. Although there was a residual current increase by 41.2 Ϯ 3.0% (n ϭ 7; Fig. 4, D and E), the effect was dramatically weaker than in wild type channels (p Ͻ 0.001). The channels in which only one of the two PKC phosphorylation sites had been inactivated (AQP1-T157A and AQP1-T239A) showed an attenuated activation by OAG (Fig. 4, D and E; p Ͻ 0.001 in comparison with wild type). Relative current increases were 78.7 Ϯ 3.6% in AQP1-T157A ( Fig. 4E; n ϭ 9) and 78.6 Ϯ 3.0% in AQP1-239A ( Fig. 4E; n ϭ 7), respectively. Thus, relative effects in those channels were comparable and ranged halfway between wild type and AQP1-⌬PKC (Fig. 4E), indicating that half of the PKCdependent effect may be attributed to each of the two sites. Control experiments with AQP1-T157A, AQP1-T239A, and AQP1-⌬PKC channels without application of OAG showed a small current increase comparable with that of AQP1 wild type channels (Fig. 4, D and E).
Western blot analysis was performed with oocytes expressing AQP1 wild type and mutant channels. In membranes isolated from the same batch of oocytes, typical bands were found Functionally, however, AQP1 mutants exhibited a markedly attenuated response to OAG (A-C). Time course of measurements is plotted in D. For comparison, the effect observed in wild type (wt) channels is included. Summary data after 30 min are shown in E. Inactivation of either of the two sites reduced PKC effects to approximately half of the values observed in wild type channels with 78.7% in T157A and 78.6% in T239A (n ϭ 9 each; p Ͻ 0.001 each). In AQP1-⌬PKC channels lacking both sites, the effect was almost absent, with values approaching those of the control experiments (41%; n ϭ 7; p Ͻ 0.05 compared with T157A and T239A). Control experiments performed with all clones yielded results comparable with the control experiments with wild type channels and with those with oocytes that did not express AQP1. In F, results of the in vitro phosphorylation assays of peptides homologous to the phosphorylation sites and the respective mutants are displayed. The autoradiograph is shown in the inset (results were reproduced in two experiments). The relative intensity of the corresponding radiographic band is shown as column graph. Wild type sequence peptides P154 -168 (containing Thr 157 ) and P236 -250 (containing Thr 239 ) were subject to intense phosphorylation in vitro. By contrast, the corresponding mutant sequence peptides termed P154 -168(T157A) and P236 -250(T239A) were not phosphorylated (p Ͼ 0.001). Protocol: holding potential Ϫ30 mV; test pulses from Ϫ120 mV to ϩ60 mV in 10-mV increments (400 ms). JULY 20, 2007 • VOLUME 282 • NUMBER 29 at the expected sizes with an identical pattern in wild type and mutant channels (Fig. 4D). Noninjected oocytes were used as controls (Fig. 4D, inset, left lane).

Positive Regulation of AQP1 Channels by PKC
To examine phosphorylation of the two consensus sites that were found to be essential for the regulation of AQP1 by PKC in the functional measurements, we carried out in vitro phosphorylation. Two peptides were synthesized corresponding to the wild type amino acid sequences surrounding those sites: P154 -168 and P236 -250. In vitro, those peptides were phosphorylated by PKC (Fig. 4F). Additionally, two analogous peptides were synthesized that copied the corresponding region of the mutant channels AQP1-T157A and AQP1-T239A. The resulting peptides (referred to as P154 -168(T157A) and P236 -250(T239A)) were tested for phosphorylation under identical conditions. In contrast to the wild type sequences, PKC phosphorylation was absent in the mutant peptides (Fig. 4F).
Cyclic Nucleotide-induced Activation of AQP1 Currents Is Independent of the Protein Kinase C Pathway-It has been shown previously that both AQP1 ion conductance and water permeability may be positively regulated by cyclic nucleotides (2,3). Having demonstrated that protein kinase C also regulates AQP1 positively, we were interested in examining whether there is an interaction between those two signaling pathways.
Therefore, AQP1-⌬PKC channels lacking functional PKC phosphorylation sites were expressed in Xenopus oocytes, and adenylate cyclase activator forskolin (100 M) and phosphodiesterase inhibitor IBMX (100 M) were co-applied to increase intracellular cyclic nucleotide levels. Electrophysiological experiments were performed with the same design as used for examining the PKC dependent regulation, i.e. with the same voltage protocol and the same observation time (30 min). The results are shown in Fig. 5. In AQP1 wild type channels, application of forskolin and IBMX induced a strong increase of currents by 96.8 Ϯ 8.1% (n ϭ 6; p Ͻ 0.001 compared with controls) after 30 min (Fig. 5, D and E). AQP1-⌬PKC channels exhibited a virtually identical response with a current increase by 92.1 Ϯ 5.8% (n ϭ 7; p Ͻ 0.001 compared with controls; Fig.  5, D and E). Co-application of guanylate cyclase inhibitor ODQ (50 M) completely suppressed the effect, resulting in a small current increase by 13.9 Ϯ 1.4% (n ϭ 6; Fig.  5, D and E) that argues for a predominant role of cGMP in line with results from other groups (2,4). Control experiments were performed with cells that did not express AQP1 channels or with superfusion with the bath solution without forskolin and IBMX, respectively. In those experiments only small current increases within the limits of the nonspecific current run-up were observed (Fig. 5, D and E). Thus, we were able to reproduce the previously described positive regulation of AQP1 conductance by cyclic nucleotides in wild type channels with a predominant role of cGMP. Interestingly, mutated AQP1 channels lacking functional PKC phosphorylation sites exhibited a virtually identical response to this regulation, indicating that both pathways probably act independently on AQP1 channels.

DISCUSSION
To the best of our knowledge, this is the first study to show that protein kinase C positively regulates aquaporin-1 channels. Furthermore, we provide data demonstrating that this new pathway of AQP1 regulation is independent of the previously described cyclic nucleotide pathway.
Molecular Basis of AQP1 Regulation by Protein Kinase C-A parallel effect of protein kinase C on ion currents and water permeability mediated by aquaporin-1 channels was observed in this study. However, technically it is not feasible to measure both features of AQP1 function simultaneously. Furthermore, measurements of water permeability are limited by the mechanical robustness of the cells that do not tolerate excessive volume increases. Because the measurement of ion currents poses considerably less methodological limitations, the respective voltage clamp experiments were used for additional con- FIGURE 5. Cyclic nucleotide induced activation of AQP1 conductance is independent of the protein kinase C pathway. It has been demonstrated previously that ionic conductance of AQP1 channels is positively regulated by cyclic nucleotides, mainly cGMP and potentially cAMP (2,3). To test whether this pathway is still active in AQP1-⌬PKC channels lacking functional PKC phosphorylation sites, intracellular levels of cyclic nucleotides were increased by co-application of forskolin (100 M) and IBMX (100 M) to the cells. As shown in A-C, this induced a marked current increase after the observation period of 30 min. In line with previous reports, AQP1 wt currents also increased markedly upon exposure to forskolin and IBMX. The time course of effects is displayed in D. The summary data are shown in E, forskolin and IBMX induced similar strong current increases in AQP1 wt channels (97%, n ϭ 6) and in AQP1-⌬PKC channels (92%, n ϭ 7), respectively (p Ͼ Ͼ 0.05). This effect could be suppressed by co-application of guanylate cyclase inhibitor ODQ (50 M) (14%, n ϭ 6, p Ͻ 0.001). In control experiments performed without application of forskolin and IBMX, currents increased to 17% (n ϭ 9) and 15% (n ϭ 8), respectively (p Ͻ 0.001). Additional control experiments performed with cells that did not express AQP1 showed no effect, either (21%; n ϭ 6). Thus, effects in AQP1 wt and in AQP1-⌬PKC channels were comparable, indicating that cyclic nucleotide-dependent activation of AQP1 is independent of the protein kinase C pathway. Protocol: holding potential Ϫ30 mV; test pulses from Ϫ120 mV to ϩ60 mV in 10-mV increments (400 ms). firmatory experiments and for a more subtle analysis of the underlying molecular mechanisms.
Several lines of evidence are provided that protein kinase C regulates aquaporin-1 channel function. First, PKC activators OAG and PMA induced an increase of AQP1 function with a time course that is typical for kinase-dependent regulation in this expression system (13,14,26). This effect does not involve endogenous water channels or ion channels of Xenopus oocytes, because it was not observed in noninjected oocytes. Second, the effect could be suppressed by co-application of specific PKC inhibitor chelerythrine. Third, inactivation of the PKC consensus sites of AQP1 almost completely abolished the observed regulation. Interestingly, inactivation of either one of the two consensus sites attenuated the regulation to a relative current increase that was approximately in the middle in between the relative effects in wild type channels and in AQP1 ⌬PKC channels. Hence, the regulatory effects of the two phosphorylation sites appear to be independent of each other with each site conferring approximately half of the total effect. Finally, in vitro phosphorylation assays using synthesized peptides demonstrated that peptides corresponding to the wild type domains surrounding Thr 157 and Thr 239 are phosphorylated by PKC, whereas homologous peptides corresponding to the T157A and T239A mutants are not. Together with the functional data, the phosphorylation data provide complementary evidence arguing for a major role for both PKC phosphorylation sites in the mediation of the observed effects.
AQP1 Activation via the PKC Pathway and via the Cyclic Nucleotide Pathway-To date, little attention has been paid to the short term regulation of AQP1. So far, only cyclic nucleotide (CN) signaling pathways have been conclusively shown to modulate AQP1 function; cGMP and also indirectly cAMP induce an increase of water permeability and give rise to cationic currents (2,3). This effect could be reproduced in AQP1 wild type channels in this study. However, the molecular mechanism underlying CN regulation of AQP1 has not been fully elucidated yet because AQP1 does not exhibit a complete CNbinding domain and no cAMP-dependent protein kinase consensus sites (4,29).
In this study, we observed a similar pattern with protein kinase C, i.e. an increase of both water permeability and ionic currents. We were therefore interested in obtaining information about potential links between these two pathways. Hence, AQP1 ⌬PKC channels with almost completely abolished PKCdependent regulation were exposed to forskolin and IBMX that induce an increase of intracellular CN levels. Those channels exhibited a marked current increase that was nearly identical to that of AQP1 wild type channels. We therefore conclude that significant cross-talk between those two pathways is unlikely and that both pathways use different molecular mechanisms to regulate AQP1 function.
Potential Physiological Implications-Traditionally, AQP1 has been classified as constitutively available pore maintaining constant water permeability (1). Its regulation has mainly been investigated on the transcriptional level with experimental evidence for an up-regulation by corticosteroids and by hypertonic stimuli via respective promoters (5). In this study, we have shown a pronounced regulation of AQP1 function by the PKC system. PKC signaling has been implicated in various physiological processes that recently have also been linked to AQP1 such as endothelial permeability, angiogenesis, and urine concentration (17)(18)(19)(20)(21)(22)(23)(24). Notably, PKC promotes angiogenesis, and increased AQP1 function has been shown to be a determinant of endothelial cell migration and subsequent angiogenesis (7-9, 20 -22). It is well documented that PKC signaling increases endothelial permeability and modulates urine concentration, both of which may involve an increase of AQP1-mediated water permeation (17-20, 23, 24). Hence, it is plausible to hypothesize that an increase of AQP1 function may also contribute to the stimulation of those processes by PKC.
AQP1 is one of several aquaporins that have been shown to conduct ion currents in addition to water (30,31). However, the physiological role of those currents has remained a matter of debate because of the small ratio of ionic conductance to water permeation (32,33). It has been suggested that aquaporin-mediated ion flow may be involved in the maintenance of ionic and osmotic balance of AQP1-expressing tissues (31), possibly comparable with the function ascribed to water-specific AQP4 channels in neuroglia (34). Recently, a related function of AQP1 water and ion transport has been shown in choroid plexus epithelium, thereby proving for the first time a physiological function of AQP1-mediated currents in native tissue (4). Notably, both the CN pathway (2, 3) and the PKC pathway described in this study induce a parallel increase of both ionic conductance and water permeability.
Conclusions-In summary, this study provides experimental evidence that protein kinase C positively regulates aquaporin-1 channels with a predominant role of both PKC phosphorylation sites at Thr 157 and Thr 239 . This new pathway of AQP1 regulation is independent of the previously described cyclic nucleotide pathway. Activation of AQP1 may contribute to the PKCdependent stimulation of physiological processes that are modulated by AQP1 such as endothelial permeability, angiogenesis, and urine concentration.