Potential conflict of interest: Nothing to report.
Estradiol 17β-D-glucuronide (E217G) is an endogenous, cholestatic metabolite that induces endocytic internalization of the canalicular transporters relevant to bile secretion: bile salt export pump (Bsep) and multidrug resistance–associated protein 2 (Mrp2). We assessed whether phosphoinositide 3-kinase (PI3K) is involved in E217G-induced cholestasis. E217G activated PI3K according to an assessment of the phosphorylation of the final PI3K effector, protein kinase B (Akt). When the PI3K inhibitor wortmannin (WM) was preadministered to isolated rat hepatocyte couplets (IRHCs), it partially prevented the reduction induced by E217G in the proportion of IRHCs secreting fluorescent Bsep and Mrp2 substrates (cholyl lysyl fluorescein and glutathione methylfluorescein, respectively). 2-Morpholin-4-yl-8-phenylchromen-4-one, another PI3K inhibitor, and an Akt inhibitor (Calbiochem 124005) showed similar protective effects. IRHC immunostaining and confocal microscopy analysis revealed that endocytic internalization of Bsep and Mrp2 induced by E217G was extensively prevented by WM; this effect was fully blocked by the microtubule-disrupting agent colchicine. The protection of WM was additive to that afforded by the classical protein kinase C (cPKC) inhibitor 5,6,7,13-tetrahydro-13-methyl-5-oxo-12H-indolo[2,3-a]pyrrolo[3,4-c]carbazole-12-propanenitrile (Gö6976); this suggested differential and complementary involvement of the PI3K and cPKC signaling pathways in E217G-induced cholestasis. In isolated perfused rat liver, an intraportal injection of E217G triggered endocytosis of Bsep and Mrp2, and this was accompanied by a sustained decrease in the bile flow and the biliary excretion of the Bsep and Mrp2 substrates [3H]taurocholate and glutathione until the end of the perfusion period. Unlike Gö6976, WM did not prevent the initial decay, but it greatly accelerated the recovery to normality of these parameters and the reinsertion of Bsep and Mrp2 into the canalicular membrane in a microtubule-dependent manner. Conclusion: The PI3K/Akt signaling pathway is involved in the biliary secretory failure induced by E217G through sustained internalization of canalicular transporters endocytosed via cPKC. (HEPATOLOGY 2010)
Bile formation is a highly regulated process. It depends on the coordinated action of a number of transporters in the sinusoidal and canalicular domains of hepatocytes. The transfer of solutes across the canalicular membrane is the rate-limiting step in bile formation; therefore, functional alterations in canalicular transporters are more likely to impair bile flow generation. Bile salt export pump (Bsep; adenosine triphosphate–binding cassette b11) and multidrug resistance–associated protein 2 (Mrp2; adenosine triphosphate–binding cassette c2) are the more relevant canalicular transporters for bile formation. Bsep mediates the efflux of amidated bile salts, the main driving force for the so-called bile salt–dependent fraction of the bile flow.1 Mrp2 mediates the transport of glutathione (GSH) and GSH and glucuronide conjugates into bile, and these contribute to the bile salt–independent fraction of the bile flow.2
Estradiol 17β-D-glucuronide (E217G) is an endogenous metabolite of estradiol belonging to the family of glucuronide conjugates of the estrogen D-ring. These glucuronide conjugates induce acute and reversible cholestasis in vivo by impairing both fractions of the bile flow3; because the levels of this cholestatic metabolite build up during pregnancy, it has been suggested to be relevant to the pathogenesis of intrahepatic cholestasis in pregnant, susceptible women.3
The mechanism by which E217G induces cholestasis is multifactorial. E217G induces endocytic internalization of both Bsep4 and Mrp25, 6; this feature is common to many other cholestatic conditions in both experimental animals and humans.7 Transinhibition of Bsep-mediated transport of bile salts by E217G has also been proposed.8 Finally, E217G increases paracellular permeability and thus leads to dissipation of the plasma-to-bile osmotic gradient.9, 10
E217G modulates a number of signaling events in hepatocytes that may contribute to its cholestatic action. Recently, we have shown that exogenously administered E217G activates classical (Ca2+-dependent) protein kinase C (cPKC) isoforms and that this event is partially involved in the cholestatic effect of E217G.11 Because the contribution of cPKC is partial, we have speculated that other signaling pathways may be involved. A likely candidate is the phosphoinositide 3-kinase (PI3K)–dependent pathway. PI3K is involved in taurolithocholate-induced cholestasis12; this effect has been suggested to be mediated by the ability of PI3K lipid products to activate the novel protein kinase C (PKC) isoform PKCε.12 This finding has prompted us to ascertain here the role of PI3K in E217G-induced cholestasis. Our results have confirmed independent involvement of PI3K in E217G-induced cholestasis that complements the involvement of cPKC, and they point to protein kinase B (Akt) as a likely downstream effector.
Akt, protein kinase B; BSA, bovine serum albumin; Bsep, bile salt export pump; CLF, cholyl lysyl fluorescein; CMFDA, 5-chloromethylfluorescein diacetate; cPKC, classical protein kinase C; cVA, canalicular vacuolar accumulation; DMSO, dimethyl sulfoxide; E217G, estradiol 17β-D-glucuronide; Gö6976, 5,6,7,13-tetrahydro-13-methyl-5-oxo-12H-indolo[2,3-a]pyrrolo[3,4-c]carbazole-12-propanenitrile; GSH, glutathione; GS-MF, glutathione methylfluorescein; IgG, immunoglobulin G; IPRL, isolated perfused rat liver; IRHC, isolated rat hepatocyte couplet; LY294002, 2-morpholin-4-yl-8-phenylchromen-4-one; Mrp2, multidrug resistance–associated protein 2; pAkt, phosphorylated protein kinase B; PI3K, phosphoinositide 3-kinase; PKC, protein kinase C; PMSF, phenylmethylsulfonyl fluoride; SEM, standard error of the media; WM, wortmannin; ZO-1, zonula occludens 1.
Materials and Methods
Cholyl lysyl fluorescein (CLF) was a generous gift from Dr. Charles O. Mills (Birmingham, United Kingdom). E217G, collagenase type A (from Clostridium histolyticum), bovine serum albumin (BSA), trypan blue, L-15 culture medium, dimethyl sulfoxide (DMSO), Triton X-100, ethylene glycol tetraacetic acid, sodium dodecyl sulfate, tetramethylethylenediamine, dithiothreitol, ammonium persulfate, leupeptin, 2-morpholin-4-yl-8-phenylchromen-4-one (LY294002), urethane, phenylmethylsulfonyl fluoride (PMSF), reduced nicotinamide adenine dinucleotide phosphate, and colchicine were acquired from Sigma Chemical Co. (St. Louis, MO). 5-Chloromethylfluorescein diacetate (CMFDA) and Alexa Fluor 568 phalloidin were obtained from Molecular Probes (Eugene, OR). Wortmannin (WM) was acquired from Fluka. 5,6,7,13-Tetrahydro-13-methyl-5-oxo-12H-indolo[2,3-a]pyrrolo[3,4-c]carbazole-12-propanenitrile (Gö6976) and the Akt inhibitor (Calbiochem 124005) were obtained from Calbiochem (San Diego, CA). A cellular lysis buffer, rabbit anti-Akt, and rabbit anti-rat phosphorylated protein kinase B (pAkt; AktSer-473) were obtained from Cell Signaling Technology (Beverly, MA). Rabbit anti-rat Bsep was acquired from Kamiya Biomedical Co. (Seattle, WA). Mouse anti-human MRP2 (M2III-6) was obtained from Alexis Biochemicals (San Diego, CA). Donkey anti-rabbit immunoglobulin G (IgG; 31458), goat anti-mouse IgG (31430), a chemiluminescence reagent, and Hyperfilm ECL were obtained from Thermo Fisher Scientific, Inc. (Waltham, MA). [3H]Taurocholate was acquired from PerkinElmer Life and Analytical Sciences (Boston, MA). All the other reagents were analytical-grade.
Isolation and Culture of Isolated Rat Hepatocyte Couplets (IRHCs)
To obtain a preparation enriched in IRHCs, livers from adult, male Wistar rats weighing 300 to 350 g and bred in our animal house as described13 were perfused under urethane anesthesia (1 g/kg intraperitoneally) according to the two-step collagenase perfusion procedure and were further enriched by centrifugal elutriation.14 The viability, assessed with the trypan blue exclusion test, was greater than 90%. After isolation, IRHCs were plated onto 24-well plastic plates at a density of 0.5 × 105 U/mL in L-15 culture medium, and they were cultured for 5 hours to allow the restoration of couplet polarity.
IRHCs were exposed to the vehicle (DMSO; control group) or E217G (12.5-800 μM) for 20 minutes. To evaluate the role of PI3K in the effect of E217G, IRHCs were preincubated with the PI3K inhibitor WM (100 nM) or LY294002 (50 μM) for 15 minutes, and this was followed by the addition of E217G for another 20-minute period. Akt participation was confirmed by the administration of the Akt inhibitor (Calbiochem 124005; 20 μM) with a similar protocol. Studies of PI3K and cPKC coinhibition were also carried out by the coadministration of the cPKC inhibitor Gö6976 (1 μM) together with WM (100 nM) for 15 minutes before exposure to E217G (400 μM) for another 20-minute period. To ascertain the role of microtubules in the protective effect of WM, IRHCs were pretreated with the microtubule-disrupting agent colchicine (1 μM) for 30 minutes and then exposed to E217G (400 μM) for another 20-minute period with or without a 15-minute WM pretreatment.
Function and Localization of Bsep and Mrp2 in IRHCs
The functional changes in Bsep and Mrp2 exposed to E217G (12.5-800 μM for 20 minutes) were evaluated through the assessment of the canalicular vacuolar accumulation (cVA) of CLF, a fluorescent Bsep substrate, or glutathione methylfluorescein (GS-MF), a fluorescent Mrp2 substrate derived from CMFDA (CMFDA is taken up by passive diffusion and converted into GS-MF by intracellular esterases and glutathione S-transferases).15 cVA was assessed by the determination of the proportion of IRHCs (>100 per preparation) capable of accumulating in their canalicular vacuoles the fluorescent compounds, as described elsewhere.11
For the assessment of the intracellular localization of Bsep and Mrp2, cells fixed with 4% paraformaldehyde in phosphate-buffered saline were incubated with the specific antibodies to Bsep or Mrp2 (1:200) for 2 hours, and this was followed by incubation with cyanine 2–conjugated donkey anti-IgG (1:100) for 40 minutes. The densitometric analysis of images taken with a confocal microscope (Zeiss Pascal LSM 5, Carl Zeiss, Walldorf, Germany) was performed along a line perpendicular to the canalicular vacuole with ImageJ 1.34m, as described elsewhere.11 We identified the canalicular space on Bsep/Mrp2-labeled IRHCs by superposing each fluorescent image with its respective differential interface contrast (DIC) image.4
Western Blot Analysis of Akt Phosphorylation
The activation of PI3K was confirmed by an evaluation of the phosphorylation status of Akt, a final PI3K effector, via western blotting of the phosphorylated and nonphosphorylated forms of the protein in hepatocyte primary cultures. Briefly, isolated hepatocytes were obtained by collagenase perfusion, as previously described,16 and cultured in 3-cm Petri dishes at a density of 2 × 106 cells/mL. After a 5-hour culture period, cells were exposed to E217G (400 μM) for 10 to 60 minutes, then washed with cold phosphate-buffered saline, and finally resuspended in a cellular lysis buffer containing protease inhibitors (25 μg/mL leupeptin and 0.1 mM PMSF). Aliquots containing an equivalent total protein content, as determined by the Lowry procedure with BSA as the standard,17 were subjected to sodium dodecyl sulfate/12% polyacrylamide gel electrophoresis. Separated proteins were electrotransferred to Immobilon-P membranes and probed with an anti-pAkt antibody (1:2000) overnight. The membranes were then stripped and reprobed with an anti–total Akt antibody (1:1000). After the use of a donkey anti-rabbit IgG secondary antibody (1:5000), a chemiluminescence reagent, and Hyperfilm ECL, pAkt and total Akt bands were quantified by densitometry with ImageJ 1.34m.
Studies in Isolated Perfused Rat Livers (IPRLs)
Livers from bile duct–cannulated rats (Intramedic PE-10 tubing, Clay Adams) were perfused in situ, as described elsewhere.11 [3H]Taurocholate (2 μCi/L, 0.7 μmol/L) was added to the perfusion medium for bile salt secretion studies. After a 20-minute equilibration period, the selective PI3K inhibitor WM (200 nM final concentration) or its solvent (DMSO; 370 μL/L) was added to the reservoir. Fifteen minutes later, a 5-minute basal bile sample was collected, and this was followed by the administration of E217G (2 μmol/liver; a single intraportal injection over a 1-minute period) or its solvent [DMSO/10% BSA in saline (4:96)]. Bile was then collected at 5-minute intervals for another 30-minute period. In some experiments, a single intravenous dose of colchicine [1.25 mM in DMSO and saline (1:4), 1 μmol/kg] was administered to rats via the femoral vein 60 minutes before the liver perfusion procedure was started. Experiments were considered valid only if the initial bile flow was greater than 30 μL/minute/kg of body weight. The viability of the liver was monitored by the release of lactate dehydrogenase into the perfusate outflow; experiments exhibiting activities over 20 U/L were discarded.
The transport activity of Mrp2 and Bsep in IPRL was evaluated by the measurement of GSH and [3H]taurocholate biliary excretion, respectively. The total GSH content in bile was measured with the recycling method of Tietze.18 [3H]Taurocholate was determined in bile with a liquid scintillation counter (Packard Instruments, Meriden, CT).
At the end of the perfusion period, a liver lobe was excised, frozen immediately in isopentane precooled in liquid nitrogen, and stored at −80°C for further immunofluorescence and confocal microscopy analysis of Mrp2 and Bsep intracellular localization. Zonula occludens 1 (ZO-1) or F-actin staining was carried out to demarcate the limits of the canaliculi, as previously described,4, 11 in samples stained for Mrp2 or Bsep, respectively. Liver sections were obtained with a Zeiss Microm HM500 microtome cryostat, air-dried, and fixed with acetone at −20°C for Mrp2 and ZO-1 colocalization studies or with 3% paraformaldehyde in phosphate-buffered saline for Bsep and F-actin colocalization studies. After fixation, liver slices were incubated overnight with the specific antibodies to Bsep, Mrp2, and ZO-1, and this was followed by 1 hour of incubation with the appropriate cyanine 2–conjugated or cyanine 3–conjugated donkey anti-IgG or with Alexa Fluor 568 phalloidin for F-actin staining. All images were taken with a Zeiss Pascal LSM 5 confocal microscope. To ensure comparable staining and image capture performance for the different groups belonging to the same experimental protocol, liver slices were prepared on the same day, mounted on the same glass slide, and subjected to the staining procedure and confocal microscopy analysis simultaneously. Image analysis of the degree of Bsep and Mrp2 endocytic internalization was performed on confocal images with ImageJ 1.34m (National Institutes of Health), as described elsewhere.4
The results are expressed as mean ± standard error of the media (SEM). Statistical analysis was performed with one-way analysis of variance followed by the Newman-Keuls test. The variances of the densitometric profiles of Bsep and Mrp2 localization were compared with the Mann-Whitney U test. P values < 0.05 were considered statistically significant.
E217G Activates PI3K
Western blots of pAkt, an indicator of PI3K activation, showed that E217G increased the amount of pAkt in a time-dependent manner (Fig. 1). This increase became apparent at 40 minutes and reached a peak at 40 minutes. The total Akt content remained unchanged. Pretreatment with the PI3K inhibitor WM (100 nM) completely prevented the increase in pAkt.
PI3K Inhibition Prevents the Impairment of Canalicular Efflux and the Relocalization of Bsep and Mrp2 Induced by E217G in IRHCs
In E217G dose-response studies, the PI3K inhibitor WM (100 nM) significantly prevented E217G-induced decreases in cVA of CLF and GS-MF at virtually all E217G doses tested (Fig. 2A and B, respectively). The participation of the PI3K-dependent pathway in this effect was confirmed further by the use of LY294002, another inhibitor of PI3K not structurally related to WM. This inhibitor, tested selectively for the E217G dose of 400 μM, also prevented a decrease in cVA of CLF and GS-MF (Fig. 3A). Neither WM nor LY294002 modified these parameters when they were added alone.
The effect of E217G on Bsep and Mrp2 function was accompanied by a significant redistribution of these transporters from the canalicular membrane into intracellular vesicles (Fig. 4, top panels). The pretreatment of IRHCs with WM markedly prevented this relocalization. This was confirmed by densitometric analysis, which demonstrated an E217G-induced redistribution of both Bsep and Mrp2 over a greater distance from the canalicular vacuoles (Fig. 4, lower panels). The pretreatment of IRHC with WM markedly prevented this relocalization; the densitometric curves were similar to those of the group pretreated with WM alone, which exhibited a slight redistribution for Mrp2.
Akt Inhibition Prevents the E217G-Induced Secretory Failure of Bsep and Mrp2 in IRHCs
To ascertain whether the downstream effector of PI3K, Akt, is involved in the cholestatic effects of E217G, we evaluated the capacity of the Akt inhibitor (Calbiochem 124005; 20 μM) to prevent the secretory inhibition caused by E217G (400 μM) in IRHCs. Pretreatment with the Akt inhibitor did not modify cVA of CLF and GS-MF when it was added alone, but it partially prevented the decreases in these parameters induced by E217G (Fig. 3A). The magnitude of the prevention was of the order of that reached with the inhibitors of PI3K (WM and LY29400), and this suggested a critical, if not exclusive, role of Akt in the PI3K-dependent cholestatic pathway.
The cPKC and PI3K Signaling Pathways Are Involved in the E217G-Induced Canalicular Secretory Failure in a Complementary Manner
We have demonstrated that the cholestatic effect of E217G is partially prevented by the selective inhibition of cPKC with Gö6976,11 and this suggests a partial role for this kinase. Because PI3K involvement is also partial (as discussed previously), we assessed here whether PI3K and cPKC act in series or in parallel; if the latter were the case, additivity of the protective effects of Gö6976 and WM would be expected when they were administered together. Our data suggest complementary participation of both signaling pathways because the preventive effects of Gö6976 (1 μM) and WM (100 nM) on the decreases in cVA of CLF and GS-MF induced by E217G were additive in nature (Fig. 3B). However, the additivity of effects could be assumed only if the maximal protective effects of the individual inhibitors had been reached, and this actually was the case. Indeed, the protective effects of Gö6976 and WM were the same even when the concentrations of these inhibitors were increased up to five times in comparison with the concentration used in Fig. 3 (data not shown).
Microtubules Are Selectively Involved in WM Prevention of E217G-Induced Changes in the Function and Localization of Canalicular Carriers
E217G-induced cholestasis is reversible in nature,19 so the degree of internalization observed in IRHCs should reflect a balance between endocytic internalization and exocytic reinsertion of the canalicular transporters. Because cPKC activation favors endocytic internalization,11 we hypothesized that PI3K may inhibit exocytic reinsertion and thus favor the intracellular retention of the canalicular carriers previously endocytosed via cPKC. The E217G-induced endocytic internalization of Mrp2 (but not its reinsertion) was shown to be independent of microtubules.6 To distinguish the two processes, we took advantage of this differential dependency and determined whether the microtubule-disrupting agent colchicine selectively blocked the protective effect of the PI3K inhibitor WM without affecting the protective effect of the cPKC inhibitor Gö6976.
The results obtained from both functional and localization studies confirmed this assumption. As shown in Fig. 5, colchicine did not modify cVA of CLF and GS-MF when it was added alone, but it completely blocked the ability of WM to prevent the E217G-induced decrease in these parameters. In contrast, colchicine did not modify the capacity of Gö6976 to prevent these secretory alterations. Consistently, colchicine completely blocked WM but not Gö6976 from preventing the E217G-induced endocytic internalization of Bsep (Fig. 6). A similar pattern was observed for Mrp2 (data not shown).
cPKC Is Involved in the Initial Decay of Bile Secretory Function Induced by E217G, Whereas PI3K Blocks Its Recovery in the IPRL Model
The acute, initial reduction in bile flow due to transporter endocytosis after E217G administration and the subsequent recovery due to reinsertion of these transporters occur differentially in time. They can therefore be readily dissected with the IPRL model, which allows dynamic monitoring of changes in bile secretory function.
The bolus administration of E217G induced a 61% decrease in bile flow within 10 minutes, and the bile flow did not recover throughout the perfusion period (Fig. 7, upper panel). This was accompanied by a decrease in the biliary excretion of the Mrp2 and Bsep substrates GSH (−62%) and [3H]taurocholate (−79%), respectively (Fig. 7, middle and lower panels). Although the cPKC inhibitor Gö6976 prevented these initial drops, the PI3K inhibitor WM had little if any protective effect. In contrast, WM greatly accelerated the recovery of both the bile flow and the biliary excretion of [3H]taurocholate and GSH, which returned to normal within 20 to 25 minutes of E217G administration. This supports our contention that PI3K contributes to E217G-induced cholestasis by retaining the canalicular carriers once they are retrieved from their membrane domain via cPKC. More direct evidence supporting our hypothesis was obtained from experiments showing that disruption of microtubule integrity by colchicine abolished the ability of WM to accelerate both the recovery of bile secretory function (Fig. 7) and the recovery of the canalicular localization of Bsep and Mrp2 at the end of the perfusion period (Fig. 8). This suggests that PI3K hinders the otherwise spontaneous microtubule-dependent retargeting of the endocytosed transporters.
Our group11, 20, 21 and other groups12, 22, 23 have provided compelling evidence that cholestatic phenomena can involve the activation of intracellular signaling cascades. We have recently demonstrated that cPKC accounts in part for the acute cholestasis caused by E217G11 and by the pro-oxidizing agent tert-butyl hydroperoxide.20 The activation of cPKC induced by these cholestatic agents correlates well with their ability to induce endocytic internalization of canalicular transporters critical for bile secretion, such as Mrp2 and Bsep.11, 20 These findings agree with a report indicating that the selective activation of cPKC by thymeleatoxin induces both cholestasis and endocytic internalization of Bsep.23
The actions of cPKC in the acute cholestasis caused by E217G do not account, however, for all of the phenomena observed, and other signaling pathways may be involved. A likely candidate is the PI3K-dependent transduction pathway. Indeed, its participation in the cholestasis induced by the cholestatic bile salt taurolithocholate has been demonstrated,12 and marked physiopathological similarities, including endocytic internalization of Bsep13 and Mrp2,24 exist between the cholestasis caused by taurolithocholate and that caused by E217G. Furthermore, structural similarities are apparent among bile salts and estrogens, and this may justify a similar ability to evoke analogous signaling pathways. Supporting this hypothesis, the present study shows for the first time that E217G activates PI3K in the liver and that this event is involved in the cholestatic effects of this estrogen.
First, we demonstrated that E217G is actually capable of phosphorylating (and activating) Akt, the main final effector of the phosphorylation cascade initiated by PI3K (see Fig. 1). Then, we showed in the IRHC model that PI3K selective inhibition has a beneficial effect on the alteration induced by E217G in both the localization status of Mrp2 and Bsep (see Fig. 4) and the canalicular secretion of their fluorescent substrates (GS-MF and CLF, respectively; see Figs. 2 and 3A). The participation of PI3K in the E217G cholestatic effect was further confirmed in IPRLs. In this model, the PI3K inhibitor WM did not prevent the rapid decrease induced by E217G in the bile flow and in the biliary excretion of both [3H]taurocholate and GSH, but it significantly accelerated the recovery of these parameters to normal (see Fig. 7) and the reinsertion of both Bsep and Mrp2 into the canalicular membrane (see Fig. 8). These WM beneficial effects were abolished by the disruption of microtubules with colchicine, and this suggests that PI3K halts the spontaneous retargeting of the endocytosed transporters, an event that we have shown to be microtubule-dependent.6 These results are consistent with studies demonstrating that the apical exocytosis of fluid-phase markers, a microtubule-dependent vesicular route likely involved in the trafficking of transporters to the apical pole, is halted by activation of PI3K in the IPRL.12
These deleterious effects of PI3K seem to be complementary to those of cPKC. Indeed, as we have shown recently,11 the cPKC inhibitor Gö6976 efficiently prevents the initial impairment of the bile secretory function and the relocalization of Bsep/Mrp2 induced by E217G. This suggests that cPKC is responsible for the initial transporter internalization,11 an effect that, unlike transporter reinsertion, has been shown to be independent of microtubule integrity.6 The differential, cooperative effects of both cPKC and PI3K activation were also observed here in IRHCs. On the base of the dynamic and reversible nature of E217G-induced cholestasis,19 we speculate that, in the IRHC model, the degree of internalization of canalicular transporters reflects a balance between endocytic mechanisms leading to internalization and exocytic mechanisms leading to reinsertion. If so, Gö6976 should prevent the first process, whereas WM should prevent the latter one. In agreement with this, the PI3K inhibitor WM and the cPKC inhibitor Gö6976 had an additive effect when they were combined in IRHCs (see Fig. 3B), and this indicated complementary action mechanisms. Supporting this possibility further, the anticholestatic effects of WM in IRHCs (but not those of cPKC) required microtubules because pretreatment with colchicine prevented the ability of WM (but not that of Gö6976) to counteract the cholestatic effects of E217G (see Figs. 5 and 6).
In the present study, we have also investigated the role of Akt as the downstream mediator of the procholestatic effects of PI3K. Akt, a serine/threonine kinase, is the most important target and effector of PI3K,25 and its phosphorylation is used as a sensitive indicator of PI3K activation.26 Phosphorylated lipid products of PI3K trigger translocation of both Akt and phosphoinositide-dependent kinase 1 to the membrane and thus activate the latter kinase to phosphorylate (and activate) Akt.25 Unlike cPKC, which was activated acutely (from 5 minutes after E217G administration) and very transiently (up to 15 minutes),11 Akt was activated from 10 minutes onward and was far longer lasting (see Fig. 1). This differential temporal pattern of activation is consistent with our proposed sequential mechanisms of action of both kinases in E217G-induced cholestasis, with cPKC accounting for the acute cholestatic phase and PI3K interfering with the late recovery phase. The role of Akt was confirmed by the pretreatment of IRHCs with a selective Akt inhibitor. A preventive effect against E217G-induced inhibition of canalicular secretory function, which was similar to the effect obtained with pan-specific inhibitors of PI3K, was observed (see Fig. 3A); this suggests crucial, if not exclusive, participation of Akt in the PI3K-dependent component of this cholestasis. In line with this, the anticholestatic agent tauroursodeoxycholate, which efficiently counteracts E217G-induced cholestasis,27, 28 is capable of inhibiting Akt activation induced by PI3K.12 On the other hand, the alternative involvement of PKCε, another signaling kinase activated by PI3K that has been suggested to mediate the taurolithocholate cholestatic effect via PI3K,22 is unlikely because E217G does not activate PKCε.11
Apart from the procholestatic effects shown here, PI3K has been demonstrated to mediate the opposite effect on bile flow (i.e., choleresis) in other experimental contexts. Indeed, PI3K is involved in the choleresis induced by both the bile salt taurocholate29 and cyclic adenosine monophosphate30; this phenomenon is attributed to the stimulation of exocytic insertion of Bsep and Mrp2 into the canalicular membrane. The reason that PI3K activation is involved in both cholestatic and choleretic phenomena cannot be ascertained from our present data. Nevertheless, several possibilities may explain these contrasting actions. First, PI3K refers to a multiplicity of isoforms, and many classes and subclasses of PI3K have been described in the liver. All of them are inhibited by the pan-specific inhibitors that were used to study the role of PI3K in these cholestatic and choleretic phenomena (e.g., WM and LY294002).31 Therefore, it is impossible with this approach to discriminate between choleretic and cholestatic PI3K isoforms if they exist. This possibility is indeed likely because cholestatic bile salts have been shown to activate a group of PI3K isoforms different from those activated by choleretic bile salts.32 Alternatively, cholestatic or choleretic agents may activate different signaling pathways apart from PI3K that, by operating in concert with PI3K, result in opposite final effects. These alternative routes may differentially inhibit signaling molecules acting downstream of PI3K so that a cholestatic route (e.g., that mediated by Akt) or a choleretic route (e.g., that mediated by mitogen-activated protein kinases) is selectively activated according to the nature of the modulating agent. It is suggestive that the choleretic (and anticholestatic) compound tauroursodeoxycholate prevents Akt phosphorylation,12 whereas it simultaneously activates routes that mediate its choleretic effect, such as the mitogen-activated protein kinases of p38 and extracellular signal-regulated kinase 1/2 types.33, 34 The opposite also happens: the cholestatic bile salt taurolithocholate inhibits the activation by tauroursodeoxycholate of certain choleretic signaling pathways downstream of PI3K (extracellular signal-regulated kinase 1/2)34 and simultaneously activates potentially cholestatic ones (Akt and PKCε).12
In conclusion, our results indicate that PI3K contributes to the cholestatic effects of E217G by activating Akt. The PI3K/Akt signaling pathway contributes to cholestasis by maintaining internalized canalicular transporters that are key for generating bile flow, such as Bsep and Mrp2, which have been previously endocytosed through an alternative mechanism likely involving cPKC. The present results will help us to consolidate the novel concept that acute cholestatic phenomena can be caused by an imbalance between signaling pathways rather than the direct action of a cholestatic compound on hepatocellular structures relevant to bile formation.