Regulation of Plasma Membrane Localization of the Na+-Taurocholate Cotransporting Polypeptide (Ntcp) by Hyperosmolarity and Tauroursodeoxycholate*

Background: Little is known about the effect of hyperosmotic hepatocyte shrinkage on Na+-taurocholate cotransporting polypeptide (Ntcp) regulation. Results: Hyperosmolarity induces a Fyn-dependent retrieval of Ntcp, which is reversed by tauroursodeoxycholate via a β1-integrin-mediated formation of cAMP. Conclusion: Fyn and β1-integrins are novel regulators of Ntcp localization. Significance: The study provides new insights into the regulation of bile salt transport. In perfused rat liver, hepatocyte shrinkage induces a Fyn-dependent retrieval of the bile salt export pump (Bsep) and multidrug resistance-associated protein 2 (Mrp2) from the canalicular membrane (Cantore, M., Reinehr, R., Sommerfeld, A., Becker, M., and Häussinger, D. (2011) J. Biol. Chem. 286, 45014–45029) leading to cholestasis. However little is known about the effects of hyperosmolarity on short term regulation of the Na+-taurocholate cotransporting polypeptide (Ntcp), the major bile salt uptake system at the sinusoidal membrane of hepatocytes. The aim of this study was to analyze hyperosmotic Ntcp regulation and the underlying signaling events. Hyperosmolarity induced a significant retrieval of Ntcp from the basolateral membrane, which was accompanied by an activating phosphorylation of the Src kinases Fyn and Yes but not of c-Src. Hyperosmotic internalization of Ntcp was sensitive to SU6656 and PP-2, suggesting that Fyn mediates Ntcp retrieval from the basolateral membrane. Hyperosmotic internalization of Ntcp was also found in livers from wild-type mice but not in p47phox knock-out mice. Tauroursodeoxycholate (TUDC) and cAMP reversed hyperosmolarity-induced Fyn activation and triggered re-insertion of the hyperosmotically retrieved Ntcp into the membrane. This was associated with dephosphorylation of the Ntcp on serine residues. Insertion of Ntcp by TUDC was sensitive to the integrin inhibitory hexapeptide GRGDSP and inhibition of protein kinase A. TUDC also reversed the hyperosmolarity-induced retrieval of bile salt export pump from the canalicular membrane. These findings suggest a coordinated and oxidative stress- and Fyn-dependent retrieval of sinusoidal and canalicular bile salt transport systems from the corresponding membranes. Ntcp insertion was also identified as a novel target of β1-integrin-dependent TUDC action, which is frequently used in the treatment of cholestatic liver disease.

Disturbances of transcellular transport of bile acids in hepatocytes can result in the accumulation of potentially toxic bile acids inside the hepatocyte and can lead to cell damage and liver dysfunction (2)(3)(4)(5)(6). It is therefore critical that hepatocellular uptake and secretion of bile acids are coordinately regulated. At the canalicular membrane, bile salt excretion is mediated by the bile salt export pump (Bsep) 2 (7) and to a lesser extent by the multidrug resistance-associated protein 2 (Mrp2) (8). The expression of Bsep and Mrp2 in the canalicular membrane of the hepatocyte is regulated by oxidative stress (9 -11), lipopolysaccharide (LPS) (12)(13)(14), drugs (15,16), and ambient osmolarity (1,10,(17)(18)(19). The cellular hydration state is a dynamic parameter that can change physiologically within minutes depending on the ambient osmolarity, nutrient supply, influence of hormones, and oxidative stress. Alterations in the cellular hydration represent an important mechanism for the metabolic control and a messenger linking cell function to hormonal and environmental alterations. Liver cell hydration also controls biliary excretion through activation of osmosensing and osmosignaling pathways (20 -22). Hyperosmotic liver cell shrinkage is cholestatic and induces a Fyn-dependent (1) retrieval of Bsep (10) and Mrp2 (17) from the canalicular membrane, whereas hypoosmotic cell swelling increases the transport capacity for bile acids in a microtubule-and mitogen-activated protein kinase (MAPK)-dependent fashion (19,23,24). Choleretic agents, such as cAMP and tauroursodeoxycholate (TUDC), also stimulate insertion of canalicular transporters from an intracellular compartment into the membrane via acti-vation of phosphatidylinositide 3-kinase (PI3K)/protein kinase C (PKC)-and MAPK-dependent pathways (19,(25)(26)(27)(28)(29).
The Na ϩ -taurocholate cotransporting polypeptide (Ntcp) (30) is the major uptake system for conjugated bile salts from the blood into liver parenchymal cells. Loss of Ntcp results in increased plasma levels of conjugated bile acids, as also confirmed by a recent report on human NTCP deficiency (31), which is clinically characterized by mild hypotonia, growth retardation, and delayed motor milestones. Functional studies show that the underlying mutation results in a markedly reduced uptake activity of taurocholate (TC), which is due to an absence of the mutant protein in the plasma membrane of the hepatocyte (31). Recent studies suggest that the cholestatic bile salt taurolithocholate inhibits hepatic TC uptake and decreases plasma membrane levels of Ntcp (32). Also, taurochenodeoxycholate (TCDC) inhibits TC uptake at the sinusoidal membrane in a chelerythrine-and cypermethrin-dependent way (33). Ntcp is a serine/threonine phosphorylated protein (34) and underlies complex regulation (35). cAMP can induce an increase in TC uptake by insertion of Ntcp into the plasma membrane (36) via increases in intracellular Ca 2ϩ and subsequent activation of protein phosphatase 2B (PP2B) via Ca 2ϩcalmodulin kinase (35). This is associated with a dephosphorylation of Ntcp at Ser-226 (37). Hepatocyte swelling stimulates rapid Ntcp insertion into the basolateral membrane and increases TC uptake in hepatocytes (38), but nothing is known about a regulation of Ntcp by hyperosmotic cell shrinkage and the hydrophilic nontoxic bile acid TUDC.
The hydrophilic bile acid ursodeoxycholic acid, which in vivo is rapidly converted to its taurine conjugate TUDC (39), is a mainstay in the treatment of cholestatic liver disease, such as primary biliary cirrhosis or intrahepatic cholestasis of pregnancy (40 -44). Intrahepatic cholestasis results in an accumulation of potentially toxic bile acids inside the hepatocyte that can cause hepatocyte death due to activation of CD95-dependent apoptosis (45,46). TUDC has potent anticholestatic and antiapoptotic properties. Experimental and clinical evidence suggests that at least three mechanisms are responsible for the hepatoprotective properties of TUDC in cholestatic disorders (47) as follows: 1) stimulation of hepatobiliary secretion; 2) cytoprotection of hepatocytes against bile acid-induced apoptosis and cytokine-induced injury; and 3) protection of cholangiocytes against cytotoxicity of hydrophobic bile acids. We could recently demonstrate that both the choleretic and antiapoptotic effects of TUDC are mediated by an activation of ␣ 5 ␤ 1 -integrins (48,49). TUDC-induced integrin activation leads to the formation of cAMP, which triggers inactivation of the CD95 and induction of a MAPK phosphatase (49). The stimulation of biliary excretion is also triggered via TUDC-induced integrin activation and subsequent activation of the MAPKs p38 MAPK and ERKs (19,50). This dual MAPK activation results in a TUDC-induced stimulation of biliary excretion by rapid targeting and insertion of intracellularly stored transport ATPases, such as Bsep or Mrp2 into the canalicular membrane (25,29). However, nothing is known about the effects of TUDC on the localization of the basolateral Ntcp, the major uptake system for conjugated bile acids into the hepatocyte. This study shows that TUDC triggers the insertion of the bile salt transporter Ntcp into the plasma membrane and that integrins are the sensor for this mechanism. Our study also shows that hyperosmotic hepatocyte shrinkage triggers an oxidative stress-, PKC-, and Fyn-dependent internalization of Ntcp, which is reversed by TUDC in a ␤ 1 -integrin-and cAMP-dependent manner. The findings are not only of interest for the understanding of hepatocellular bile acid handling and TUDC action in liver, but may also be relevant for hepatitis B virus entry into the hepatocyte, which was recently shown to be mediated by NTCP (51). Indeed, recent data suggest that interleukin-6 inhibits hepatitis B virus entry into the hepatocyte through down-regulation of NTCP (52).
Liver Perfusion-Livers from male Wistar rats (140 -160 g) were perfused in a nonrecirculating manner as described previously (54). As a perfusion medium the bicarbonate-buffered Krebs-Henseleit saline plus L-lactate (2.1 mmol/liter) and pyruvate (0.3 mmol/liter) gassed with 5% CO 2 and 95% O 2 at 37°C was used (305 mosmol/liter, normo-osmotic). Hyperosmotic (385 mosmol/liter) perfusion was performed by raising the NaCl concentration in the medium. Addition of inhibitors, Bt 2 cAMP and TUDC to the influent perfusate was made by dissolution into the Krebs-Henseleit buffer. Viability of the perfused livers was assessed by measuring lactate dehydrogenase leakage into the perfusate. The portal pressure, the effluent K ϩ concentration, and pH were continuously monitored. Livers from male control mice (wild type, 8 -10 weeks; on C57BL/6 background) or p47 phox knock-out mice (8 -10 weeks, C57BL/ 6J-Ncf1m1J/J, The Jackson Laboratory) were perfused in an adapted version.
Preparation and Culture of Primary Rat Hepatocytes-Hepatocytes were isolated from livers of male Wistar rats (160 -180 g) by a collagenase perfusion technique in an adapted version as described previously (55). Animals were fed ad libitum with a standard diet. Aliquots of rat hepatocytes were plated on collagen-coated culture plates and maintained in bicarbonate-buffered Krebs-Henseleit medium (115 mmol/liter NaCl, 25 mmol/ liter NaHCO 3 , 5.9 mmol/liter KCl, 1.18 mmol/liter MgCl 2 , 1.23 mmol/liter NaH 2 PO 4 , 1.2 mmol/liter Na 2 SO 4 , 1.25 mmol/liter CaCl 2 ), supplemented with 6 mmol/liter glucose in a humidified atmosphere of 5% CO 2 and 95% air at 37°C. After 2 h, the medium was removed, and the cells were washed twice. Subsequently, the culture was continued for 24 h in William's Medium E, supplemented with 2 mmol/liter glutamine, 100 nmol/liter insulin, 100 units/ml penicillin, 0.1 mg/ml streptomycin, 100 nmol/liter dexamethasone, and 5% FBS. After 24 h, experimental treatments were performed using William's Medium E that contained 2 mmol/liter glutamine and 100 nmol/liter dexamethasone. The viability of hepatocytes was more than 95% as assessed by trypan blue exclusion. To allow comparisons with previous studies, in these cell culture experiments hyperosmolarity of 405 mosmol/liter was used, whereas in rat liver perfusions we used 385 mosmol/liter.
Densitometric Fluorescence Intensity Analysis-Cryosections of perfused rat liver for the analysis at the basolateral membrane were stained for Ntcp and for Na ϩ /K ϩ -ATPase (plasma membrane marker protein). For negative controls, primary antibodies were omitted in each experiment. Na ϩ /K ϩ -ATPase profiles were selected according to apparent integrity and comparability. Acceptable Na ϩ /K ϩ -ATPase intensity profiles have a sufficiently high peak of fluorescence in the central part (corresponding to the basolateral membrane) and a low intracellular fluorescence. For analysis of the canalicular membrane, cryosections were stained for Bsep and the tight junction protein ZO-1, which forms the sealing border between canalicular and sinusoidal membrane. Apparent integrity and comparability of the canaliculi was assumed when the bordering tight junction lines (detected by the immunostained ZO-1) were intact, run in parallel, and showed a width of 1.12 to 1.82 m. Densitometric analysis was performed as described previously (10,17). For analysis of digitalized microscopic pictures of the membranes, the software Image-Pro Plus (Media Cybernetics, Rockville, MD) was used. The profile of the fluorescence intensity was measured over a thick line at a right angle to the mem- brane. The length of the line was always 8 m. The mean fluorescence intensity to each pixel over the line perpendicular to the length was calculated by Image-Pro Plus. The obtained data (pixel positions with the associated pixel intensities, red and green channel) were transferred to an Excel data sheet (Microsoft Excel for Windows, Redmond, WA). Each measurement was normalized to the sum of all intensities of the respective measurement. Values are given as means Ϯ S.E. Densitometric analysis of protein distribution in immunofluorescence images was performed using Wilcoxon rank sum test. p Ͻ 0.05 was considered statistically significant. Data from at least 10 different areas per tissue sample and from at least three independent liver preparations were processed.
Immunoblot Analysis-Liver samples were immediately lysed at 4°C by using a lysis buffer containing 20 mmol/liter Tris-HCl (pH 7.4), 140 mmol/liter NaCl, 10 mmol/liter NaF, 10 mmol/liter sodium pyrophosphate, 1% (v/v) Triton X-100, 1 mmol/liter EDTA, 1 mmol/liter EGTA, 1 mmol/liter sodium vanadate, 20 mmol/liter ␤-glycerophosphate, protease inhibitor, and phosphatase inhibitor mixture. The lysates were kept on ice for 10 min and then centrifuged at 8000 rpm for 8 min at 4°C, and aliquots of the supernatant were taken for protein determination using the protein assay (Bio-Rad, Munich, Germany). Equal amounts of protein were subjected to SDS-PAGE and transferred onto nitrocellulose membranes using a semidry transfer apparatus (GE Healthcare, Freiburg, Germany). Membranes were blocked for 60 min in 5% (w/v) bovine serum albumin containing 20 mmol/liter Tris (pH 7.5), 150 mmol/liter NaCl, and 0.1% Tween 20 (TBS-T) and exposed to primary antibodies overnight at 4°C. After washing with TBS-T and incubating at room temperature for 2 h with horseradish peroxidase-coupled anti-mouse or anti-rabbit IgG antibody, respectively (all diluted 1:10,000), the immunoblots were washed extensively, and bands were visualized using the Fluo-rChem E detection instrument from ProteinSimple (Santa Clara, CA). Semi-quantitative evaluation was carried out by densitometry using the Alpha View image acquisition and analysis software from ProteinSimple. Protein phosphorylation is given as the ratio of detected phosphoprotein/total protein.
Detection of ROS-Primary hepatocytes were seeded on collagen-coated plates (Falcon, Heidelberg, Germany) and cultured for 24 h. Cells were incubated with PBS containing 5 mol/liter CM-H 2 DCFDA for 30 min at 37°C in a humidified atmosphere of 5% CO 2 and 95% air. To detect ROS generation, CM-H 2 DCFDA-loaded cells were supplemented again with culture medium and then exposed to the respective stimuli for the indicated time period. Thereafter, cells were washed briefly using ice-cold PBS and lysed in 0.1% Triton X-100 (v/v) dissolved in aqua bidest. Lysates were centrifuged immediately (10,000 ϫ g, 4°C, 1 min), and fluorescence of the supernatant was measured at 515-565 nm using a luminescence spectrometer LS-5B (PerkinElmer Life Sciences, Rodgau, Germany) with an excitation wavelength of 488 nm. Fluorescence intensity of controls was set to 1, and fluorescence found under the respective treatment is given relative to it.
Statistical Analysis-Results from at least three independent experiments are expressed as mean values Ϯ S.E. n refers to the number of independent experiments. For each experimental treatment and time point analyzed, a separate control experiment was carried out. Differences between experimental groups were analyzed by Student's t test or one-way analysis of variance following FIGURE 2. Regulation of Ntcp in hyperosmotically perfused rat liver. A, rat livers were perfused with normo-osmotic (305 mosmol/liter) or hyperosmotic (385 mosmol/liter) Krebs-Henseleit buffer for up to 60 min and immunostained for Ntcp and Na ϩ /K ϩ -ATPase. Densitometric analysis of fluorescence profiles and intensity of Ntcp and Na ϩ /K ϩ -ATPase distribution (visualization by confocal LSM) is shown. Under control conditions (green (0 min) and black (60 min)), Ntcp-bound fluorescence was largely localized in the center of the basolateral membrane. Hyperosmotic perfusion (red (30 min) and blue (60 min)) resulted in a significant lateralization of the Ntcp-bound fluorescence, but no significant changes of Na ϩ /K ϩ -ATPase fluorescence profiles were observed. The fluorescence profiles depicted are statistically significantly (p Ͻ 0.05) different from each other with respect to variance and peak height. Means Ϯ S.E. of 30 measurements in each of at least three individual experiments for each condition are shown. B, rat livers were perfused with normo-osmotic (305 mosmol/liter) or hyperosmotic (385 mosmol/liter) Krebs-Henseleit medium for up to 60 min, immunostained for Ntcp and Na ϩ /K ϩ -ATPase, and visualized by SR-SIM. Under normo-osmotic conditions, Ntcp was largely localized in the membrane, and after hyperosmotic exposure Ntcp, but not the Na ϩ /K ϩ -ATPase, was internalized. Representative pictures of at least three independent experiments are depicted. The scale bars correspond to 10 m in the overviews (upper row) and 1 m in the four bottom rows with higher magnification of the plasma membrane. C, primary rat hepatocytes were cultured for 24 h and thereafter stimulated with normo-osmotic (305 mosmol/liter) or hyperosmotic (405 mosmol/liter) medium for up to 24 h. RNA was extracted, and Ntcp mRNA expression levels were analyzed by real time PCR. Ntcp mRNA expression levels are given relative to the unstimulated control (0 min). Data represent the mean Ϯ S.E. of four independent experiments *, p Ͻ 0.05 statistical significance compared with the unstimulated control. #, p Ͻ 0.05 statistical significance between normo-osmotic and hyperosmotic stimulations.

Regulation of Ntcp Trafficking in the Liver
Dunnett's multiple comparison post hoc test where appropriate (GraphPad Prism; GraphPad, La Jolla, CA; Microsoft Excel for Windows). p Ͻ 0.05 was considered statistically significant.

Results
Hyperosmolarity Triggers Ntcp Retrieval from the Basolateral Membrane in the Perfused Rat Liver-To study the subcellular localization of the bile salt transporter Ntcp in hepatocytes, immunofluorescence studies on cryosectioned tissues using confocal laser-scanning microscopy were performed (Fig. 1). For labeling of the plasma membrane, liver sections were stained simultaneously with specific antibodies against the plasma membrane marker protein Na ϩ /K ϩ -ATPase and against Ntcp, respectively. The profile of the fluorescence intensity of Ntcp and Na ϩ /K ϩ -ATPase was measured over a thick line (8 m) perpendicular to the membranes of adjacent  . Distribution of Ntcp in hyperosmotically perfused wild-type and p47 phox knock-out mouse liver. Mouse livers were perfused for 30 min with either normo-osmotic (305 mosmol/liter, black) or hyperosmotic Krebs-Henseleit buffer (385 mosmol/liter, green) and stained for Ntcp and Na ϩ /K ϩ -ATPase. Densitometric analysis of fluorescence profiles and intensity of Ntcp and Na ϩ /K ϩ -ATPase staining is shown. The means Ϯ S.E. of 30 measurements in each of three individual experiments for each condition are shown. As shown by colocalization with Na ϩ /K ϩ -ATPase, Ntcp is largely localized in the membrane under normo-osmotic conditions. After hyperosmotic perfusion, Ntcp is no longer colocalized (p Ͻ 0.05) with Na ϩ /K ϩ -ATPase and appears inside the cells in wild-type animals but remained unchanged in p47 phox knock-out mice. OCTOBER  hepatocytes. The mean fluorescence intensity of each pixel along this line was calculated by Image-Pro Plus, and the obtained and normalized data were plotted on a line graph (Fig. 1). Subcellular Ntcp distribution in normo-osmotically (305 mosmol/liter) and hyperosmotically (385 mosmol/liter) perfused livers was quantified by densitometric fluorescence intensity analysis as described above. Within 30 min of hyperosmotic perfusion, the fluorescence profile of Ntcp shifted to significantly lower peak values and became more broadened (p Ͻ 0.05, Fig. 2A), suggestive for an internalization of the Ntcp in response to hyperosmolarity. In contrast, no changes in Na ϩ / K ϩ -ATPase distribution were observed under these conditions ( Fig. 2A). To discriminate Ntcp distribution at a subcellular level, SR-SIM in Ntcp and Na ϩ /K ϩ -ATPase-costained cryosections was performed. In SR-SIM images, obtained from hyperosmotically (385 mosmol/liter) perfused rat livers, intracellular vesicular Ntcp fluorescence intensity was markedly increased within 30 min and maintained for at least 60 min (Fig. 2B). Localization of the basolateral membrane protein Na ϩ /K ϩ -ATPase remained unchanged as compared with normo-osmotic control conditions (Fig. 2B). These results suggest that hyperosmolarity induces within 30 min endocytosis of Ntcp in the intact rat liver, which lasts for at least 60 min. Hyperosmolarity-induced retrieval of Ntcp was not accompanied by changes in Ntcp mRNA expression during the first 60 min of hyperosmotic exposure (0.96 Ϯ 0.07-fold, not significant, n ϭ 4, Fig. 2C). However, a significant decrease of Ntcp mRNA levels was seen after treatment with hyperosmotic medium for 180 min (0.76 Ϯ 0.04-fold, p Ͻ 0.05, n ϭ 4) or more (Fig. 2C).

Regulation of Ntcp Trafficking in the Liver
Hyperosmolarity-induced Retrieval of Ntcp from the Basolateral Membrane Is Mediated by Activation of the Src Family Kinase Fyn-Hyperosmolarity was shown to trigger NADPH oxidase-dependent formation of ROS (6,57) and to activate the Src family kinases Yes and Fyn, although no c-Src activation became detectable within 180 min of hyperosmotic exposure (1). Hyperosmotic Fyn and Yes activations were sensitive to inhibition by NAC, apocynin, and SU6656, whereas PP-2 inhibited hyperosmotic Fyn but not Yes activation (1). Ntcp distribution in hyperosmotically (385 mosmol/liter) perfused livers in the absence and presence of inhibitors of the NADPH oxidase-driven ROS generation (apocynin and NAC) or of Src kinases (PP-2, SU6656) was quantified by densitometric fluorescence intensity analysis (Fig. 3). Fig. 3A depicts the perfusion plans chosen for these experiments. As shown in Fig. 3, B-D, NAC (10 mmol/liter) and apocynin (20 mol/liter) inhibited the hyperosmolarity (385 mosmol/liter, t ϭ 30 min)-induced Ntcp retrieval from the basolateral membrane, whereas Na ϩ / K ϩ -ATPase distribution remained unchanged. These findings point to a role of NADPH oxidase-driven ROS formation in triggering transporter retrieval from the plasma membrane. Also, the Src kinase inhibitors SU6656 (1 mol/liter) and PP-2 (250 nmol/liter) prevented hyperosmotic Ntcp retrieval from the basolateral membrane (Fig. 3D). Because PP-2 (58) inhibits Fyn but not Yes activation, and SU6656 (59) inhibits both hyperosmotic Yes and Fyn activation, these findings suggest that Fyn is involved in the hyperosmotic retrieval of Ntcp from the basolateral membrane. The role of NADPH oxidase in triggering hyperosmotic Ntcp retrieval from the plasma membrane was also studied in livers from p47 phox knock-out mice. As shown in Fig. 4, hyperosmolarity induced within 30 min the internalization of Ntcp in livers from wild-type mice, whereas no retrieval was detectable in livers from p47 phox knock-out mice.
TUDC Reverses Hyperosmolarity-induced Internalization of Ntcp via Integrin-dependent Protein Kinase A Activation-The choleretic bile salt TUDC induces a rapid insertion of intracellularly stored Mrp2 and Bsep into the canalicular membrane of the hepatocyte (19,25), which increases the bile salt excretory capacity in a microtubule-dependent way (60,61). As shown by SR-SIM, TUDC (20 mol/liter, 30 min) was able to reverse the hyperosmolarity-induced internalization of Ntcp (Fig. 5). Because ␤ 1 -integrins were identified as sensors for TUDC in rat liver (19,48), the integrin-inhibitory peptide GRGDSP, which binds to ␣ 5 ␤ 1 -integrin with a K D of 396 nM (62), was used to study the involvement of ␤ 1 -integrins in TUDC-induced Ntcp insertion into the plasma membrane. GRGDSP (10 mol/liter) inhibited the counteracting effect of TUDC on hyperosmotic Ntcp retrieval, although the inactive control peptide GRADSP (10 mol/liter) was ineffective (Fig. 5, A and C). As shown recently, TUDC activates via ␤ 1 -integrins the PKA (49). When H89 (2 mol/liter), a PKA inhibitor, was perfused 30 min before and during the experiment, the effect of TUDC on hyperosmolarity-induced Ntcp retrieval was largely abolished, whereas Na ϩ /K ϩ -ATPase immunostaining remained unchanged under all conditions (Fig. 5C). Similar findings were obtained with Bt 2 cAMP (50 mol/liter), which is known to increase the uptake capacity of TC by increasing plasma membrane content of Ntcp (36). This effect was also H89 (2 mol/liter)-sensitive (data not shown). All inhibitors, i.e. GRADSP, GRGDSP, and H89, had per se no effect on Ntcp localization (data not shown).

Hyperosmolarity-induced Fyn Activation Is Inhibited by TUDC via an Integrin-and Protein Kinase A-dependent
Mechanism-Rat liver perfusion with hyperosmotic medium (385 mosmol/liter) (Figs. 6 and 7) induced an activating phosphorylation of Fyn and Yes, which was inhibited by TUDC (20

JOURNAL OF BIOLOGICAL CHEMISTRY 24247
ity (30 min)-induced activation of Fyn and Yes (Fig. 7). When H89 (2 mol/liter) was added 30 min before and during the perfusion, the effect of Bt 2 cAMP on hyperosmolarity-induced Src kinase activation was blunted (Fig. 7).
TUDC Reverses the Hyperosmolarity-induced Internalization of the Bile Salt Transporter Bsep-In line with a TUDC-induced reversal of hyperosmotic Fyn activation, TUDC not only prevented the hyperosmolarity-induced Ntcp retrieval from the plasma membrane, but also the hyperosmolarity-induced Bsep retrieval from the canalicular membrane (Fig. 9), which was previously shown to be Fyn-dependent, too (1). Fig. 9A shows the experimental approach, which was also used in previous studies (1,9,10,17). Immunofluorescence stainings of the canalicular bile salt transporter Bsep and of ZO-1 (a protein of the tight junction complex) were investigated by confocal laser scanning microscopy and densitometric analysis. ZO-1 delineates the bile canaliculi and a linear staining pattern of ZO-1 was found under all conditions. During normo-osmotic perfusion (305 mosmol/liter), Bsep was located predominantly in the canalicular membrane, i.e. between the two rows of ZO-1 fluorescence (Fig. 9, A and B). Hyperosmotic perfusion (385 mosmol/liter) led to an increase of Bsep fluorescence intensity inside the hepatocytes at the expense of its canalicular localiza-tion, suggesting Bsep internalization (Fig. 9B). Subsequent addition of TUDC (20 mol/liter) to the same liver preparation led to an almost complete disappearance of intracellular Bsep, indicating re-insertion of Bsep into the canalicular membrane. Under all these conditions, ZO-1 fluorescence distribution remained unaffected (Fig. 9B).
Paralleling the inhibition of ROS formation, the hyperosmolarity-induced Fyn phosphorylation was sensitive to inhibition of PKC by the specific PKC pseudosubstrate and by the broad spectrum PKC inhibitors chelerythrine and Gö 6850 (Fig. 11A). Gö 6850 also prevented the hyperosmolarity-induced retrieval of Ntcp (Fig. 11B). Gö 6976, a selective inhibitor of Ca 2ϩ -dependent PKC isoforms, had no significant effect on hyperosmotic Fyn phosphorylation (Fig. 11A). These findings suggest that activation of PKC in response to hyperosmolarity may play a role in triggering hyperosmotic Fyn phosphorylation (Fig. 11A) and Ntcp retrieval (Fig. 11B).

Discussion
Hyperosmolarity was shown to stimulate the retrieval of Bsep and Mrp2 from the canalicular membrane (1,10,17), whereas the hydrophilic bile acid TUDC stimulates the insertion of these transporters into the canalicular membrane (25,29). However, the effects of hyperosmolarity and TUDC on the short term regulation of the basolateral transporter Ntcp remained unexplored. This study shows the following: (i) hyperosmolarity induces an oxidative stress-and Fyn-dependent retrieval of Ntcp from the membrane, and (ii) TUDC is able to stimulate re-insertion of Ntcp into the plasma membrane by a ␤ 1 -integrin-and PKA-dependent mechanism. To what extent TUDC-induced PI3K activation (64) contributes to this effect remains to be established. Fig. 12 schematically summarizes our current view on short term regulation of the basolateral bile salt transporter Ntcp.
Under normo-osmotic conditions, Ntcp was found mainly in the basolateral membrane and also in transporter-containing vesicles below the plasma membrane (Figs. 2 and 5). Hyperosmolarity increased the fraction of Ntcp, which is located inside the hepatocytes (Figs. 2 and 5). This shift in Ntcp localization occurred within 30 min, persisted for at least 60 min, and was reversible in response to TUDC (Fig. 5). Activation of the Src kinase Fyn was shown to occur in response to hyperosmotic cell shrinkage and to trigger the retrieval of Bsep and Mrp2 from the canalicular membrane (1). As shown in this study, hyperosmotic Fyn activation also triggers Ntcp retrieval from the basolateral membrane (Fig. 3). PP-2, which blocks the hyperosmolarity-induced phosphorylation of Fyn (1), abolished internalization of Ntcp (Fig. 3). Likewise, inhibitors of the upstream events leading to Fyn activation, such as the Nox inhibitor apocynin or the antioxidant NAC, prevented transporter internalization (Fig. 3). Furthermore, Ntcp retrieval was inhibited in p47 phox knock-out mice (Fig. 4), in which the hyperosmolarity-induced activation of Fyn is abolished (1). These data strongly suggest that Fyn mediates not only Bsep and Mrp2 retrieval from the canalicular membrane, but simultaneously it also triggers the retrieval of Ntcp from the basolat-eral membrane. This suggests a coordinated regulation of sinusoidal bile salt uptake and canalicular export systems, which may help to avoid accumulation of potentially toxic bile acids in the hepatocyte. This view is also corroborated by the finding that TUDC and cAMP could reverse hyperosmolarityinduced Fyn activation (Figs. 6 and 7), Ntcp retrieval from the plasma membrane (Fig. 5), and Bsep retrieval from the canalicular membrane (Fig. 9). Thus, TUDC exerts its choleretic effect by a coordinated insertion of both Ntcp and Bsep into plasma and canalicular membrane, respectively. A recent study showed that vesicle motility and ATP production is transiently reduced in a variety of cell types, including hepatocytes after treatment with hyperosmotic medium (65). Movement of dextran-loaded endosomes was tracked by live-cell imaging. Because dextran is distributed to early, late, and recycling endosomes, this approach, however, will not distinguish between differences in motility among these compartments. Whether the motility is restricted also in newly endocytosed vesicles therefore remains unclear (65).
Hyperosmotic perfusion of rat liver induced NADPH oxidase-dependent oxidative stress response as an upstream signal for Fyn activation. Cell hydration and oxidative stress are apparently closely interlinked. On the one hand cell shrinkage induces oxidative stress (66), and on the other hand oxidative stress leads to hepatocyte shrinkage (67). Inflammation is frequently accompanied by ROS formation and can induce cholestasis. It is likely that inflammation also triggers Ntcp retrieval from the plasma membrane. Indeed, proapoptotic bile salts such as taurolithocholic acid 3-sulfate or glycochenodeoxycholate, which induce a NOX2-dependent ROS formation and cell shrinkage (68), also trigger Fyn activation and Ntcp internalization.
Hyperosmolarity-induced retrieval of Ntcp was not accompanied by a down-regulation of Ntcp mRNA levels within 60 min, whereas LPS triggered both Ntcp retrieval and a significant decrease of Ntcp expression at the mRNA and protein level (69). This difference between hyperosmotic and LPS-induced Ntcp retrieval may be explained by cytokine effects.
PKCs belong to a family of serine/threonine kinases, which consists of at least 12 isoforms (70). These kinases are involved in the regulation of diverse cellular functions (70), including bile formation (35,71). It was suggested that choleretic and cholestatic effects may involve different PKC isoforms; however, the picture is not entirely consistent. The toxic bile acid TCDC inhibits bile salt transport across the sinusoidal mem-  brane by PKC-and PP2B-mediated retrieval of Ntcp from the plasma membrane (33). Phorbol ester-induced endocytosis of Ntcp was shown to be mediated in part by Ca 2ϩ -dependent PKCs (56). Furthermore, it was shown recently that PKC␣ can directly interact with Ntcp in stably Ntcp-transfected HepG2 cells (72). PKC (73,74) was suggested to mediate a choleretic effect by inserting Ntcp into the membrane, whereas PKC⑀ (75, 76) may trigger cholestasis by retrieving Mrp2 from the canalicular membrane. However, PKC was also shown to trigger p47 phox phosphorylation and to induce oxidative stress (57) as an upstream event for Bsep and Mrp2 retrieval from the canalicular membrane (1). This study shows that PKC is involved in the Fyn-regulated retrieval of the Ntcp from the membrane (Fig. 11), whereas PKC inhibitors were reported to prevent the cAMP-induced Ntcp translocation to the plasma membrane (73). These discrepant findings strongly suggest that regulation of Ntcp retrieval/insertion by PKC may be dependent on the signaling context and the underlying stimulus.
As part of its choleretic effect, TUDC stimulates insertion of the canalicular transporter Bsep and Mrp2 by activation of Erks and p38 MAPK (25,50) and counteracts internalization in taurolithocholate-induced cholestasis (29). In this study, the choleretic effect of TUDC may reside in the prevention of hyperosmotic Fyn activation and cAMP formation. Recently, ␤ 1 -integrins were identified as sensors for TUDC in the liver (19,48), and molecular dynamics simulation studies showed that TUDC directly interacts with ␣ 5 ␤ 1 -integrins resulting in an activation of ␤ 1 -integrin and the initiation of the downstream integrin signaling (19,48). This also explains the TUDCinduced activation of c-Src as observed in Fig. 6C. Translocation of Ntcp to the plasma membrane by TUDC was sensitive to GRGDSP and H89 (Fig. 5), which suggests that also Ntcp insertion by TUDC involves a ␤ 1 -integrin-mediated activation of PKA. Similarly to TUDC, cAMP significantly increased Ntcp translocation to the basolateral membrane in a PKA-dependent way (data not shown). The effect of cAMP is known to be mediated via two major pathways. Activation of PI3K, probably via PKA (77), leads to activation of Akt (78) and PKC isoforms (73,74,79), which in turn induce the translocation of Ntcp from an intracellular compartment to the membrane. Fyn was immunoprecipitated and detected for phosphorylation by Western blotting. Total Fyn served as a loading control. Hyperosmolarityinduced Fyn phosphorylation was sensitive to inhibition of PKC (PKC pseudosubstrate, chelerythrine, and Gö 6850). Fyn phosphorylation under control condition was set to 1. Data represent means Ϯ S.E. of three independent experiments. *, p Ͻ 0.05 statistically significant compared with the unstimulated control. #, p Ͻ 0.05 statistical significance between hyperosmolarity and hyperosmolarity plus PKC inhibitor. B, rat livers were perfused with either Gö 6850 (1 mol/liter)-containing normo-osmotic (305 mosmol/liter) buffer (black) or hyperosmotic (385 mosmol/liter) buffer in the absence (blue) or presence of the inhibitor (green) (see perfusion plan in Fig. 3A) and immunostained for Ntcp and Na ϩ /K ϩ -ATPase. Densitometric analysis of fluorescence profiles and intensity of Ntcp and Na ϩ /K ϩ -ATPase staining at t ϭ 30 min (see perfusion plan) is shown. The means Ϯ S.E. of 10 measurements in each of three individual experiments for each condition are shown. Gö 6850 significantly inhibited hyperosmolarity-induced retrieval of Ntcp from the membrane (p Ͻ 0.05), whereas the inhibitor by itself did not significantly affect Ntcp localization.
Ntcp is a serine/threonine-phosphorylated protein (34), and dephosphorylation at Ser-226 (37) is associated with translocation to the plasma membrane. cAMP activates the serine and threonine protein phosphatase PP2B by increasing intracellular Ca 2ϩ . Activated PP2B directly dephosphorylates Ntcp, allowing translocation to the basolateral membrane (80). It was shown recently that TUDC by itself can induce a ␤ 1 -integrindependent rapid generation of cAMP (49), which may explain Ntcp dephosphorylation in response to TUDC.
Taken together, this study for the first time shows that hyperosmolarity induces the retrieval of Ntcp from the plasma membrane and that this effect is mediated by the Src family kinase Fyn, which is activated by a PKC-dependent and NADPH oxidase-driven ROS formation. TUDC counteracts the effect of hyperosmolarity through activation of a ␤ 1 -integrin/PKA signaling pathway, which prevents hyperosmotic ROS formation and Fyn activation and is associated with dephosphorylation of the Ntcp on serine residues.