ATP8B1 requires an accessory protein for endoplasmic reticulum exit and plasma membrane lipid flippase activity

Authors


  • Potential conflict of interest: Nothing to report.

Abstract

Mutations in ATP8B1 cause progressive familial intrahepatic cholestasis type 1 and benign recurrent intrahepatic cholestasis type 1. Previously, we have shown in mice that Atp8b1 deficiency leads to enhanced biliary excretion of phosphatidylserine, and we hypothesized that ATP8B1 is a flippase for phosphatidylserine. However, direct evidence for this function is still lacking. In Saccharomyces cerevisiae, members of the Cdc50p/Lem3p family are essential for proper function of the ATP8B1 homologs. We have studied the role of two human members of this family, CDC50A and CDC50B, in the routing and activity of ATP8B1. When only ATP8B1 was expressed in Chinese hamster ovary cells, the protein localized to the endoplasmic reticulum. Coexpression with CDC50 proteins resulted in relocalization of ATP8B1 from the endoplasmic reticulum to the plasma membrane. Only when ATP8B1 was coexpressed with CDC50 proteins was a 250%-500% increase in the translocation of fluorescently labeled phosphatidylserine observed. Importantly, natural phosphatidylserine exposure in the outer leaflet of the plasma membrane was reduced by 17%-25% in cells coexpressing ATP8B1 and CDC50 proteins in comparison with cells expressing ATP8B1 alone. The coexpression of ATP8B1 and CDC50A in WIF-B9 cells resulted in colocalization of both proteins in the canalicular membrane. Conclusion: Our data indicate that CDC50 proteins are pivotal factors in the trafficking of ATP8B1 to the plasma membrane and thus may be essential determinants of ATP8B1-related disease. In the plasma membrane, ATP8B1 functions as a flippase for phosphatidylserine. Finally, CDC50A may be the potential β-subunit or chaperone for ATP8B1 in hepatocytes. (HEPATOLOGY 2007.)

ATP8B1 is a member of the type 4 subfamily of P-type ATPases (P4 ATPases).1, 2 Members of this protein family are implicated in the inward translocation of phospholipids in biological membranes and are termed flippases.3–7 ATP8B1 is expressed in the apical membrane of many epithelial cells, including hepatocytes and enterocytes.8–10 The deficiency of ATP8B1 in man causes the severe liver disease progressive familial intrahepatic cholestasis type 1 (PFIC1), which is characterized by impaired bile salt excretion from the liver into bile.11–13 PFIC1 primarily manifests as a chronic intrahepatic cholestasis that progresses to severe, end-stage liver disease and often requires liver transplantation during the first or second decade of life. Mutations in ATP8B1 also cause benign recurrent intrahepatic cholestasis type 1 (BRIC1), a milder form of PFIC1 that is characterized by periodic bouts of cholestasis that leave no liver injury.13 Finally, it has been recently shown that mutations in ATP8B1 may also play a role in the etiology of intrahepatic cholestasis of pregnancy (ICP).14, 15

We have previously shown in mice that Atp8b1 is essential for maintaining the detergent-resistant properties of the canalicular membrane.16 When challenged with bile salts, Atp8b1-deficient mice displayed enhanced extraction of cholesterol and ectoenzymes from the canalicular membrane. In addition, Atp8b1-deficient mice had significant amounts of phosphatidylserine (PS) in their bile, whereas PS is essentially absent from wild-type bile. Previously, Ujhazy and colleagues9 provided results suggesting that Atp8b1 can translocate a fluorescently labeled PS analog [7-nitro-2-1,3-benzoxadiazol-4-yl–labeled phosphatidylserine (NBD-PS)] from the outer leaflet to the inner leaflet of the canalicular membrane. Still, the physiological substrates and molecular mechanisms of ATP8B1-mediated activity in particular and intramembrane lipid transport in general are poorly understood.

Recently, it has been demonstrated in the yeast Saccharomyces cerevisiae and in Leishmania parasites that P4 ATPases assemble with members of the Cdc50p/Lem3p family.17–20 In yeast, this evolutionary conserved family of transmembrane proteins includes three close Cdc50p homologs, which are glycosylated proteins of approximately 60 kDa with two putative transmembrane domains. Saito and colleagues17 were the first to show that in yeast interactions between P4 ATPases and Cdc50p are required for the release of P4 ATPases from the endoplasmic reticulum (ER) and targeting to their proper membrane domains. In humans, three proteins (termed CDC50A, CDC50B, and CDC50C) have been identified that are homologous to the yeast Cdc50p/Lem3p family; however, their function is unknown.1, 2, 21 Analogous to yeast, these human CDC50 proteins may be important accessory factors for human P4 ATPase targeting and function.

In the present study, we have isolated complementary DNAs (cDNAs) encoding human CDC50A and CDC50B. In UPS-1 cells, a Chinese hamster ovary (CHO) mutant cell line that is defective in the nonendocytic uptake of NBD-PS,22 we show that ATP8B1 is retained in the ER when expressed alone, whereas coexpression with CDC50A or CDC50B results in rerouting of ATP8B1 to the plasma membrane and lysosome. In UPS-1 cells, we show that both CDC50A and CDC50B stimulate ATP8B1-mediated internalization of NBD-PS and natural PS. Finally, we show that ATP8B1 and CDC50A colocalize in the canalicular membrane of polarized WIF-B9 cells.

Abbreviations

ABCC2, adenosine triphosphate–binding cassette C2; ATP8B1-eGFP, C-terminal enhanced green fluorescent protein–tagged ATP8B1; BRIC1, benign recurrent intrahepatic cholestasis type 1; cDNA, complementary DNA; CHO, Chinese hamster ovary; CHO-K1, Chinese hamster ovary cell line; DMEM, Dulbecco's modified Eagle's medium; eGFP, enhanced green fluorescent protein; ER, endoplasmic reticulum; FITC, fluorescein isothiocyanate; GFP, green fluorescent protein; HA, hemagglutinin antigen; HA-CDC50, N-terminal hemagglutinin antigen–tagged CDC50; HBSS/H, Hank's balanced salt solution without phenol red supplemented with 20 mM 4-(2-hydroxyethyl)-1-piperazine ethanesulfonic acid at pH 7.4; ICP, intrahepatic cholestasis of pregnancy; IgG, immunoglobulin G; LAMP-2, lysosome-associated membrane protein 2; Lgp-2, lysosomal membrane glycoprotein 2; Lgp-B, lysosomal membrane glycoprotein B; mRNA, messenger RNA; NBD-PS, 7-nitro-2-1,3-benzoxadiazol-4-yl–labeled phosphatidylserine; PCR, polymerase chain reaction; PFIC1, progressive familial intrahepatic cholestasis type 1; PS, phosphatidylserine; SDS-PAGE, sodium dodecyl sulfate–polyacrylamide gel electrophoresis; TGN, trans-Golgi network; UPS-1, CHO-K1 mutant cell line with a defect in the nonendocytic uptake of fluorescent phosphatidylserine analogs.

Materials and Methods

Cell Lines and Cell Culture.

UPS-1 and CHO-K1,22 AKN-1,23 and HeLa, HEK 293T, and WIF-B9 cells24 were cultured in Dulbecco's modified Eagle's medium (DMEM; Cambrex) supplemented with 10% fetal calf serum, 1% glutamine (vol/vol), and 1% (vol/vol) penicillin/streptomycin at 37°C in a 10% CO2 atmosphere. WIF-B9 cultures were supplemented with 0.04 μM aminopterin, 10 μM hypoxanthine, and 1.6 μM thymidine.

Isolation of cDNAs and Generation of Constructs.

Human ATP8B1 cDNA (NM_005603) was isolated from a human control liver. C-terminal enhanced green fluorescent protein–tagged ATP8B1 (ATP8B1-eGFP) cDNA was generated by the fusion of an enhanced green fluorescent protein (eGFP) cassette to the C-terminus of ATP8B1. N-terminal hemagglutinin antigen–tagged CDC50A (HA-CDC50A; NM_018247) and N-terminal hemagglutinin antigen–tagged CDC50B (HA-CDC50B; XM_090844) cDNAs were isolated from HeLa and AKN-1 cells, respectively. The primers used to amplify CDC50 cDNAs and to introduce the hemagglutinin antigen (HA) tag are available on request. cDNAs were cloned into a phosphoglycerate kinase promoter–containing lentiviral transfer vector, and a recombinant lentivirus was produced as described.25

Lentiviral Transduction of Cells and Preparation of Cell Lysates.

UPS-1 or WIF-B9 cells, grown to 50%-60% confluence, were incubated with virus-containing supernatants/DMEM (1:1) supplemented with 10 μg/mL diethylaminoethyl-dextran for 4 hours. Total cell lysates were prepared by the scraping of cells in a hypotonic lysis buffer [10 mM trishydroxymethylaminomethane/HCl (pH 7.4), 10 mM KCl, and 1.5 mM MgCl2] supplemented with protease inhibitors (2 mM phenylmethylsulfonylfluoride, 1 μg/mL aprotinin, 1 μg/mL leupeptin, 1 μg/mL pepstatin, and 0.1 mg/mL Pefabloc SC). Cells were homogenized by sonication. The protein content was measured with the bicinchoninic acid method.

Quantitative Polymerase Chain Reaction (PCR) Analyses.

Total RNA was extracted from mouse livers with the Trizol reagent (Invitrogen). cDNA was synthesized from total RNA with an oligo-dT12–18 primer and Superscript II reverse transcriptase (Invitrogen). Real-time PCR measurements were performed at 60°C in a Lightcycler apparatus (Roche) with Lightcycler Faststart DNA Master Plus SYBR Green I (Roche). The primer sequences are available on request. The expression levels were calculated with respect to the housekeeping gene 36b4 (acidic ribosomal phosphoprotein P0).

Coimmunoprecipitation and Western Blot Analyses.

Coimmunoprecipitation was performed as described previously with some modifications.17 UPS-1 cell pellets were incubated for 0.5 hours on ice in 1 mM NaHCO3 supplemented with protease inhibitors. Cells were homogenized with a Dounce homogenizer and centrifuged at 210g for 5 minutes at 4°C to remove intact cells. The supernatant was centrifuged in a Beckman SW40 rotor for 1 hour at 100,000g at 4°C. The pellet was incubated in an immunoprecipitation buffer [10 mM trishydroxymethylaminomethane-HCl (pH7.5), 150 mM NaCl, 2 mM ethylene diamine tetraacetic acid, and 1% 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate] supplemented with protease inhibitors for 2 hours at 4°C. Insoluble material was removed by centrifugation for 5 minutes at 19,900g at 4°C. The supernatant was incubated with monoclonal anti-HA agarose conjugate clone HA-7 (Sigma-Aldrich) overnight at 4°C. Anti-HA agarose beads were pelleted and washed with an immunoprecipitation buffer. Immunoprecipitated proteins were subsequently removed from the beads by incubation with a sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) sample buffer, fractionated by 8% SDS-PAGE, and transferred to Protran nitrocellulose transfer membranes (Schleicher & Schuell BioScience). The membranes were incubated in a block buffer (phosphate-buffered saline/5% milk powder/0.05% Tween 20) for 1 hour at room temperature and incubated with rabbit polyclonal anti-ATP8B1 (2K), mouse monoclonal anti-HA (clone 12CA5), rabbit polyclonal antibody to Atp1a1 (C356-M09),26 or mouse monoclonal antibody to green fluorescent protein (GFP; JL-8, Clontech Laboratories) in a block buffer for 2 hours at room temperature. Immune complexes were visualized with peroxidase-conjugated goat–anti-rabbit immunoglobulin G (IgG) or goat–anti-mouse IgG2b (anti-HA and anti-GFP). Immune complexes were visualized with Lumi-Light western blotting substrate (Roche). Chemiluminescence was detected with a Lumi-Imager F1 and LumiAnalyst 3.1 software (Roche).

Indirect Immunofluorescence.

UPS-1 and WIF-B9 cells were grown on glass coverslips and fixed in 2% paraformaldehyde in phosphate-buffered saline or in methanol/acetone (4:1) for 20 minutes at room temperature. Paraformaldehyde-fixed cells were incubated with anti-HA (clone 3F10; Roche), anti–early endosome antigen 1 (clone 14; BD Biosciences Pharmingen), or anti–adenosine triphosphate–binding cassette C2 (anti-ABCC2; M2III6). Methanol/acetone-fixed cells were washed twice with 70% ethanol and were incubated with the antibody to lysosomal membrane glycoprotein B (anti–Lgp-B) or the antibody to lysosome-associated membrane protein 2 (anti–LAMP-2; clones UH3 and H4B4, respectively; Developmental Studies Hybridoma Bank) or with anti-calnexin [C-terminal (575-593); Calbiochem]. The immunoreactivity was visualized with goat–anti-rat, affinity-purified, Texas Red–conjugated IgGs (Rockland) or goat–anti-mouse or goat–anti-rabbit alexa 594 (Molecular Probes). Sections were mounted in Vectashield/DAPI (Vector Laboratories) and were studied with a Leica TCS-SP2 confocal microscope.

Translocation Assay for NBD-PS.

The fluorescently-labeled, 16:0-06:0 NBD-PS [1-palmitoyl-2-[6-[(7-nitro-2-1,3-benzoxadiazol-4-yl)amino]hexanoyl]-sn-glycero-3-phospho-L-serine (ammonium salt)] was from Avanti Polar Lipids. NBD-PS stock in chloroform/methanol (3:2) was transferred to a glass tube and dried under a stream of nitrogen. The lipid film was dissolved in 40μL of 96% ethanol with glass beads. Hank's balanced salt solution without phenol red (Cambrex) supplemented with 20 mM 4-(2-hydroxyethyl)-1-piperazine ethanesulfonic acid at pH 7.4 (HBSS/H) was added during vortexing. The final concentration of NBD-PS was 20 μM. The translocation of NBD-PS was determined as described before with some modifications.22 Briefly, cells grown to confluence in 48-well plates (Corning) were washed, equilibrated in HBSS/H for 10 minutes at 15°C, and incubated with 20 μM NBD-PS for 20 minutes at 15°C. Subsequently, the cells were washed with HBSS/H (4°C). To remove NBD-PS from the outer leaflet of the membrane, the cells were incubated 3 times for 10 minutes in Hank's balanced salt solution supplemented with 2% (essentially fatty acid–free) bovine serum albumin (Sigma-Aldrich). The cells were washed with HBSS/H (4°C), and 200 μL of HBSS/H was added. Intracellular fluorescence was measured with a Novostar analyzer (BMG Labtech; excitation, 485 nm; emission, 540 nm). Intracellular fluorescence was completely derived from NBD-PS as determined after lipid extraction and high-performance liquid chromatography.

Annexin V Assay.

The cells were trypsinized, washed, and incubated for 2 hours at 37°C in DMEM/10% fetal bovine serum at a cell density of 4.106/mL. The cells were washed with an incubation medium [132 mM NaCl, 6 mM KCl, 1 mM MgSO4, 1.2 mM potassium phosphate buffer, p47.4 (Kpi), 20 mM 4-(2-hydroxyethyl)-1-piperazine ethanesulfonic acid (pH 7.4), 10 mM glucose, and 0.5% human serum albumin], resuspended in an incubation medium supplemented with 2.5 mM CaCl2, and incubated on ice for 0.5 hours with FITC-labeled annexin V. The cells were analyzed on a Becton Dickinson FACScan flow cytometer. A data analysis was performed with the WinMDI 2.8 software program (Scripps Research Institute).

Results

Localization of ATP8B1 and ATP8B1-eGFP in UPS-1 Cells.

In order to study the lipid flippase activity of ATP8B1, we stably expressed ATP8B1 in UPS-1 cells by lentiviral transduction. UPS-1 is a nonpolarized CHO-K1 mutant cell line with a defect in the nonendocytic uptake of the NBD-PS analog.22 Because PS is the putative substrate for ATP8B1 and UPS-1 cells have low background NBD-PS flippase activity, this cell line is very suitable for testing ATP8B1 function. ATP8B1-transduced UPS-1 cells displayed high ATP8B1 protein levels (Fig. 1A). No endogenous Atp8b1 protein or messenger RNA (mRNA) was detected in untransduced UPS-1 cells (or in parental CHO-K1 cells). In contrast to a previous report by Ujhazy and colleagues,9 we could not detect any significant ATP8B1-dependent internalization of NBD-PS in these cells (shown later in Fig. 6). To study subcellular localization of ATP8B1, we expressed ATP8B1-eGFP in UPS-1 cells. ATP8B1-eGFP predominantly localized to the ER, and this was confirmed by costaining for the ER-specific marker calnexin (Fig. 1B). At cell-cell contacts, very faint plasma membrane staining was observed in some cells. The same ATP8B1-eGFP localization and NBD-PS internalization data were obtained for the ATP8B1-eGFP transduced parental CHO-K1 cell line (not shown).

Figure 1.

ATP8B1 localizes to the ER in the nonpolarized CHO mutant cell line UPS-1. (A) ATP8B1 and ATP8B1-eGFP protein expression levels in UPS-1 cells. Cells were transduced with lentiviral stocks encoding ATP8B1 or ATP8B1-eGFP. Total cell lysates were prepared and subjected to SDS-PAGE and immunoblotting. Blots were incubated with anti-ATP8B1 (upper panel) and anti-GFP (lower panel). (B) ATP8B1-eGFP costains with the ER marker calnexin. Sometimes, faint ATP8B1-eGFP plasma membrane staining was observed at cell-cell contacts (white arrow). Cells were studied with a confocal microscope. The bar is equal to 11.9 μm.

Isolation of CDC50A and CDC50B cDNAs.

Analogous to the mechanism in the yeast S. cerevisiae, ATP8B1 may assemble with a member of the conserved Cdc50p/Lem3p family. To study this possibility, we isolated cDNAs encoding human CDC50A and CDC50B. Attempts to amplify CDC50C cDNA from a large panel of cell lines failed with several primer sets.21 CDC50A (361aa) and CDC50B (351aa) are ∼60-kDa proteins with two putative transmembrane domains, a large presumed exoplasmic loop, and a few potential N-glycosylation sites (Fig. 2). CDC50A and CDC50B share 47% amino acid sequence identity, whereas sequence identities with the yeast homologs range from 27% to 30% and from 22% to 29% for CDC50A and CDC50B, respectively. In addition, this conserved family of proteins is characterized by 2 highly conserved amino acid stretches, that is, the PCG-X-IANS-X-FND and F-XX-WMR-XX-A stretches, both in the presumed exoplasmic loop. The functional role of these conserved stretches is unknown.

Figure 2.

Alignment of members of the human and yeast S. cerevisiae Cdc50p/Lem3p protein family. The alignment was generated with ClustalW analyses (http://www.genome.ad.jp/SIT/SIT.html). The prediction of transmembrane domains (shaded) was determined with SOSUI software (http://www.genome.ad.jp/SIT/SIT.html). Potential N-glycosylation sites are shown for the human CDC50 proteins (underlined) and were predicted with the Center for Biological Sequence Analysis prediction server (http://www.cbs.dtu.dk/services). Conserved amino acids are indicated by asterisks. The GenBank accession numbers are Lem3p (NP_014076), Cdc50p (NP_010018), Crf1p (NP_014446), CDC50A (NP_060717), CDC50B (NP_001017970), and CDC50C (sequence from Katoh and Katoh21).

In Mouse Liver, Cdc50a mRNA Is Preferentially Expressed.

We tested if Cdc50 mRNAs were expressed in the liver of wild-type and Atp8b1-deficient mice (Fig. 3). Cdc50a was highly expressed in the liver of both genotypes. The Cdc50b and Cdc50c mRNA levels were 100 and 1000 times lower, respectively, than the level of Cdc50a. No differences in the liver Cdc50 expression levels were observed between wild-type and Atp8b1-deficient mice, and this suggests that Atp8b1 deficiency does not lead to compensatory expression of any of the Cdc50 mRNAs. These data suggest that Cdc50a may be a potential binding partner for Atp8b1 in mouse liver.

Figure 3.

In a mouse liver, Cdc50a is preferentially expressed. Real-time PCR analyses were performed on liver cDNA from wild-type mice (white bars, n = 4) and Atp8b1-deficient mice (black bars, n = 4). The expression levels were calculated with respect to the housekeeping gene 36b4.

Coexpression of ATP8B1 and CDC50 Proteins Retargets ATP8B1 to the Plasma Membrane.

To study a possible interaction between ATP8B1 and CDC50 proteins, we coexpressed HA-CDC50A or HA-CDC50B in UPS-1 cells stably expressing ATP8B1-eGFP (Fig. 4A). In these cells, we studied subcellular localization of ATP8B1-eGFP. Upon coexpression with HA-CDC50A, ATP8B1-eGFP localized to the plasma membrane (Fig. 4B, panels a-c). In these cells, discontinuous plasma membrane staining for HA-CDC50A was observed that costained with ATP8B1-eGFP. Coexpression with HA-CDC50B resulted in plasma membrane staining of ATP8B1-eGFP but also staining of intracellular vesicular structures (Fig. 4B, panels d-f). These vesicular structures represented lysosomes/late endosomes, as determined by costaining with the lysosome-specific marker Lgp-B (Fig. 4C). We could not detect any HA-CDC50B staining in the plasma membrane. These data show that ATP8B1-eGFP is released from the ER when coexpressed with HA-CDC50A and HA-CDC50B and is targeted to the plasma membrane and lysosome/late endosome.

Figure 4.

ATP8B1-eGFP localizes to the plasma membrane and to lysosomes when coexpressed with HA-CDC50A or HA-CDC50B. (A) ATP8B1 and CDC50 protein expression levels in UPS-1 cells. UPS-1 cells were (co)transduced with a recombinant lentivirus encoding ATP8B1, ATP8B1-eGFP, HA-CDC50A, and HA-CDC50B. The protein expression was analyzed by immunoblotting with anti-ATP8B1 (upper panel) and anti-HA (lower panel). (B) Immunofluorescent detection of ATP8B1-eGFP and HA-CDC50 proteins. (a-c) ATP8B1-eGFP localized to the plasma membrane when coexpressed with HA-CDC50A. The merge images show partial costaining of ATP8B1-eGFP with HA-CDC50A in intracellular compartments and faint costaining in the plasma membrane (b, white arrows). (d-f) ATP8B1-eGFP localized to the plasma membrane and intracellular vesicular structures when coexpressed with HA-CDC50B. No costaining between ATP8B1-eGFP and HA-CDC50B was observed. (C) Intracellular vesicular GFP staining in cells coexpressing ATP8B1-eGFP and HA-CDC50B costains with the lysosomal marker lysosomal membrane glycoprotein 2 (Lgp-2). Nuclear 4′,6-diamidino-2-phenylindole staining is in blue. The bar is equal to 11.9 μm.

ATP8B1 Physically Interacts with HA-CDC50A and HA-CDC50B.

To study a possible interaction between ATP8B1-eGFP and the CDC50 proteins, we performed coimmunoprecipitation experiments. Crude membrane preparations from the various UPS-1 cell lines were used to immunoprecipitate HA-CDC50A and HA-CDC50B with anti-HA agarose. The immunoprecipitates were analyzed for the presence of ATP8B1. We could not detect any ATP8B1 in immunoprecipitates of untransduced UPS-1 cells and cells expressing only ATP8B1, HA-CDC50A, or HA-CDC50B (Fig. 5A, upper panel, lanes 1-4). The coexpression of ATP8B1 with HA-CDC50A or HA-CDC50B resulted in the coimmunoprecipitation of ATP8B1 with the CDC50 proteins (Fig. 5A, upper panel, lanes 5 and 6). ATP8B1-eGFP was also coimmunoprecipitated with both HA-CDC50 proteins (not shown), and this indicated that the eGFP tag did not impair the interaction with either of the CDC50 proteins. No signal for the Na+,K+-ATPase was detected in the coimmunoprecipitates, and this indicated that the solubilization of the membrane preparations was efficient and that there was no aspecific binding of ATP8B1 to the beads. CDC50A was hardly detectable in the total lysates of cells expressing HA-CDC50A alone (Fig. 5B, lower panel, lane 2). However, the coexpression of ATP8B1 and either of the CDC50 proteins resulted in higher expression levels of both the CDC50 proteins and ATP8B1 (Fig. 5B). These data indicate that ATP8B1 physically interacts with both CDC50 proteins and suggest that the interaction of ATP8B1 with CDC50 proteins increases the stability of ATP8B1 and that of the CDC50 proteins.

Figure 5.

ATP8B1 physically interacts with CDC50A and CDC50B. (A) ATP8B1 is coimmunoprecipitated with HA-CDC50A and HA-CDC50B. HA-CDC50A and HA-CDC50B were immunoprecipitated with anti-HA agarose. The immunoprecipitates were analyzed by immunoblotting with anti-ATP8B1 (upper panel) or anti-HA (lower panel). ATP8B1 coimmunoprecipitates with both HA-CDC50A and HA-CDC50B (lanes 5 and 6). HA-CDC50A is not detectable, possibly because of the breakdown of this protein when it is expressed alone (lane 2, lower panel). (B) ATP8B1 and HA-CDC50 protein expression in total cell lysates. Equal amounts of protein of solubilized membrane extracts were analyzed by immunoblotting with anti-ATP8B1 (upper panel) and anti-HA (lower panel). The ATP8B1 expression levels are higher when ATP8B1 is coexpressed with HA-CDC50A or HA-CDC50B (upper panel; compare lanes 4-6). The HA-CDC50A and HA-CDC50B expression levels are higher upon coexpression with ATP8B1 (lower panel; compare lanes 2 and 5 and lanes 3 and 6). The results are shown for a representative experiment.

Coexpression of ATP8B1 and CDC50A/B Stimulates Plasma Membrane NBD-PS Internalization.

Next, we tested the panel of UPS-1 transductants for their ability to internalize NBD-PS (Fig. 6). All experiments were performed at 15°C to reduce the bulk endocytic uptake (but not flippase activity) of phospholipid analogs.27 CHO-K1 cells, included as a positive control, accumulated approximately 50% more fluorescence than UPS-1 cells. The expression of only HA-CDC50A or HA-CDC50B did not result in enhanced translocation of NBD-PS in comparison with UPS-1 (not shown). The expression of only ATP8B1 resulted in a very small (20%) increase in intracellular fluorescence. However, the coexpression of ATP8B1 and HA-CDC50A resulted in a 250% increase in NBD-PS internalization in comparison with UPS-1 cells (Fig. 6A). Coexpression with HA-CDC50B showed an even stronger increase in NBD-PS internalization: 300% in comparison with UPS-1 cells (Fig. 6B). Also, ATP8B1-eGFP displayed high NBD-PS internalization activity when coexpressed with either CDC50 protein. The uptake of 7-nitro-2-1,3-benzoxadiazol-4-yl–labeled phosphatidylcholine (PC), phosphatidylethanolamine (PE) and sphingomyelin (SM) was not dependent on ATP8B1-CDC50 expression. These data show that (1) ATP8B1 stimulates the internalization of NBD-PS, provided that one of the CDC50 proteins is present, and (2) the C-terminal eGFP tag does not interfere with the activity of the protein.

Figure 6.

ATP8B1 stimulates NBD-PS internalization when coexpressed with either HA-CDC50A or HA-CDC50B. (A) NBD-PS internalization in UPS-1 cells (co)expressing ATP8B1(-eGFP) and HA-CDC50A. (B) NBD-PS internalization in UPS-1 cells (co)expressing ATP8B1(-eGFP) and HA-CDC50B. Cells were incubated with NBD-PS for 20 minutes at 15°C to allow the uptake of NBD-PS. Cells were back-extracted with 2% bovine serum albumin to remove NBD-PS that was associated with the exoplasmic leaflet of the plasma membrane, and intracellular fluorescence was quantified fluorimetrically. The data are expressed as the percentage of fluorescence in UPS-1 cells, which was set at 100%. CHO-K1, the parental cell line of UPS-1, was included as a positive control. The results are shown for a representative experiment and are expressed as means ± the standard deviation (n = 4). Statistical analyses were performed by a one-way analysis of variance with Bonferroni's correction for multiple testing. *P < 0.01.

Coexpression of ATP8B1 and CDC50A/B Stimulates Natural PS Internalization.

Next, we determined the ability of ATP8B1 to stimulate the internalization of endogenous PS by measuring annexin V binding to the cell surface (Fig. 7). In CHO-K1 cells, 22% of the cells stained positive for annexin V. In contrast, in UPS-1 cells expressing only ATP8B1, over 50% of the cells were annexin V–positive. However, the coexpression of ATP8B1 with HA-CDC50A or HA-CDC50B resulted in a significant reduction of the percentage of annexin V–positive cells (36% and 28%, respectively). These data demonstrate that (1) UPS-1 cells, apart from being impaired in the uptake of NBD-PS analogs, are also impaired in the internalization of natural PS and (2) ATP8B1, when coexpressed with CDC50A or CDC50B, stimulates the internalization of natural PS.

Figure 7.

ATP8B1 stimulates the internalization of endogenous PS when coexpressed with HA-CDC50A or HA-CDC50B. Cells were trypsinized and labeled on ice with an FITC-labeled annexin V probe. Cells were analyzed for surface-bound FITC–annexin V by fluorescence-activated cell sorting analysis. (A) A histogram of a representative experiment displaying the differences in the fluorescence intensity between CHO-K1 (dotted line), UPS-1 cells coexpressing ATP8B1 and HA-CDC50B (dashed line), and UPS-1 cells expressing ATP8B1 alone (gray area). The x axis and y axis represent the fluorescence intensity of the cells and the cell number, respectively. UPS-1 cells coexpressing ATP8B1 and HA-CDC50A showed a histogram similar to that depicted here for ATP8B1 and HA-CDC50B. (B) The percentage of FITC–annexin V–positive cells is expressed as the mean ± the standard deviation of 4 independent experiments. Statistical analyses were performed by a one-way analysis of variance with Bonferroni's correction for multiple testing. *P < 0.001 (compared with UPS-1 cells expressing ATP8B1 alone).

ATP8B1-eGFP and HA-CDC50A Colocalize in the Canalicular Membrane of WIF-B9 Cells.

ATP8B1 localizes to the canalicular membrane of hepatocytes. Because Cdc50a is preferentially expressed in mouse liver, we studied the localization of ATP8B1-eGFP and HA-CDC50A coexpressed in WIF-B9 cells, a hepatocyte model cell line.24 In polarized WIF-B9 cells, ATP8B1-eGFP localized to the canalicular membrane, as demonstrated by costaining with the canalicular marker ABCC2 (Fig. 8A-C). ATP8B1-eGFP also localized to subapical vesicles and weakly stained the basolateral membrane. Importantly, HA-CDC50A costained with ATP8B1-eGFP in the canalicular membrane (Fig. 8D-F). In addition, ATP8B1-eGFP and HA-CDC50A costained in subapical vesicles and in the basolateral membrane, the latter being much weaker than the canalicular membrane. These subapical vesicles represented lysosomes/late endosomes, as demonstrated by costaining with the lysosomal marker LAMP-2 (Fig. 8G-I). These data demonstrate that ATP8B1-eGFP and HA-CDC50A colocalize in the canalicular membrane and in subapical, lysosomal/late endosomal vesicles in WIF-B9 cells.

Figure 8.

ATP8B1-eGFP and HA-CDC50A colocalize to the canalicular membrane of WIF-B9 cells. WIF-B9 cells were cotransduced with a transgenic lentivirus encoding ATP8B1-eGFP and HA-CDC50A. (A-C) ATP8B1-eGFP costains with ABCC2 in the canalicular membrane of polarized WIF-B9 cells. (D-F) ATP8B1-eGFP and HA-CDC50A costain in the canalicular membrane and in subapical vesicles; also, weak costaining of the basolateral membrane can be observed. (G-I) The ATP8B1-eGFP–positive subapical vesicles costain with the lysosomal/late endosomal marker LAMP-2. No LAMP-2 staining of the canalicular membrane can be observed. Nuclear 4′,6-diamidino-2-phenylindole staining is in blue. The asterisks indicate canalicular vacuoles. The bar is equal to 11.9 μm.

Discussion

Genetic defects in ATP8B1 are the cause of PFIC1 and BRIC128 and are associated with ICP.14, 15 Previously, we have shown in mice that Atp8b1 deficiency leads to enhanced biliary recovery of PS and cholesterol.16 We hypothesized that ATP8B1 is a flippase for PS in the canalicular membrane, but direct evidence for this hypothesis remained elusive in our hands. Here we have shown, using the CHO-K1 mutant cell line UPS-1, that ATP8B1 mediates the translocation of PS from the exoplasmic leaflet to the cytoplasmic leaflet of the plasma membrane. Importantly, plasma membrane PS flippase activity was observed only when ATP8B1 was coexpressed with CDC50 proteins. We show that the coexpression of ATP8B1 with HA-CDC50 proteins allows the exit of ATP8B1 from the ER and localization to the plasma membrane and lysosome/late endosome. Furthermore, we show that ATP8B1-eGFP and HA-CDC50A colocalize to the canalicular membrane of WIF-B9 cells.

Saito and colleagues17 showed that P4 ATPases in the yeast S. cerevisiae assemble with members of the Cdc50p/Lem3p family. S. cerevisiae expresses five P4 ATPases that are implicated in the translocation of phospholipids and participate in distinct vesicular trafficking pathways.3, 29 With a broad panel of mutants, it has been demonstrated that P4 ATPase-Cdc50p/Lem3p assembly is required for exit from the ER and proper membrane trafficking of the P4 ATPase. Such a mechanism is highly reminiscent of that for non–phospholipid-translocating human P-type ATPases, including the Na+,K+-ATPase and H+,K+-ATPase.30 The assembly of the α-subunits and β-subunits of these pumps in the ER is essential for α-subunit maturation and subsequent exit of both subunits out of the ER. Saito and colleagues proposed that Cdc50p and Lem3p are possible β-subunits for yeast P4 ATPases. Although we have found that ATP8B1 assembles with CDC50 proteins, it is not clear from our data whether CDC50 proteins are β-subunits or chaperones for ATP8B1. In the case of a subunit, a stable, permanent interaction occurs, whereas a chaperone interacts only transiently during trafficking.

Our data indicate that the specificity between ATP8B1 and CDC50 interactions determines the trafficking pathways in which these proteins are involved. In S. cerevisiae, differential localization of the yeast P4 ATPases Drs2p, Dnf1-Dnf3, and Neo1p has been observed.29 Drs2p assembles with Cdc50p and cycles between the trans-Golgi network (TGN), late endosomes, and the plasma membrane.3, 17, 18 Dnf1p and Dnf2p interact with Lem3p17, 19 and cycle between the endosomal system and the plasma membrane.3, 4 Finally, Neo1p cycles between ER and TGN and TGN and the endosomal system,31, 32 whereas Dnf3p has overlapping localizations with Drs2p;3, 4 however, no assembly with Cdc50 proteins has been reported thus far.

In humans, three CDC50 homologs have been identified, whereas there are 14 P4 ATPases.1, 2 At present, it is not known which CDC50 proteins assemble with ATP8B1 in the hepatocyte, although our data suggest that CDC50A is the most likely candidate. First, we find that in mouse liver, Cdc50a mRNA is preferentially expressed. Second, we demonstrate the colocalization of ATP8B1-eGFP and HA-CDC50A in the canalicular membrane of WIF-B9 cells. Whether CDC50A is crucial for canalicular trafficking of ATP8B1 or whether additional factors are required remains to be determined. At present, it is also not known if CDC50 proteins are binding partners of any of the other 13 human P4 ATPases. It is of interest to note that CDC50C and ATP8B3, a close homolog of ATP8B1, are exclusively expressed in the testes.2, 33–35 Their functional interaction, however, remains to be determined.

Using the same cells and experimental setup used for this article, Ujhazy and colleagues9 previously showed a relation between ATP8B1 overexpression and NBD-PS internalization. The authors showed a 150% increase in NBD-PS internalization when ATP8B1 was overexpressed alone. Under this condition, we show that ATP8B1 sticks in the ER, and we find only a 20% increase in ATP8B1-dependent NBD-PS uptake. Apparently, a very small amount of ATP8B1 is able to exit from the ER to the membrane in the absence of CDC50 proteins. In contrast, the coexpression of ATP8B1 and CDC50A resulted in a relocalization of ATP8B1 to the plasma membrane, which coincided with a 250% increase in NBD-PS internalization. The coexpression of ATP8B1 and CDC50B led to a plasma membrane and lysosomal localization and a 300% increase in NBD-PS internalization. NBD-PS internalization activity was somewhat higher for ATP8B1-eGFP, and this observation may be explained by a somewhat lower turnover of this fusion protein. Similarly, the internalization of endogenous PS, as measured by annexin V binding, was enhanced in the cell lines coexpressing ATP8B1 and either of the CDC50 proteins. Despite the lysosomal localization of ATP8B1 in cells coexpressing CDC50B, both NBD-PS and natural PS internalization were enhanced. One explanation for this may be that the ATP8B1-CDC50B complex rapidly cycles between the plasma membrane and intracellular stores, including the lysosomes. Our data demonstrate that ATP8B1 is involved in the internalization of PS. In yeast, the flippase function of P4 ATPases is important for protein trafficking.7, 29 It remains to be established whether ATP8B1 also plays a role in vesicular trafficking events.

If an interaction between ATP8B1 and CDC50 proteins is crucial in hepatocytes, one can speculate that amino acid substitutions in ATP8B1 that interfere with ATP8B1-CDC50 assembly can cause PFIC1, BRIC1, or ICP. The complete impairment of the protein complex assembly may result in ER retention and subsequent degradation of ATP8B1, causing PFIC1. Partial impairment of the complex assembly may result in an instable protein complex that is prone to breakdown. Indeed, in CHO cells, we have observed that in the absence of its partner, both ATP8B1 and the CDC50 proteins are less stable. Such mutations in ATP8B1 may be a cause of BRIC1 and ICP. Alternatively, mutations in CDC50 proteins that interfere with ATP8B1 interaction may also cause PFIC1, BRIC1, and ICP. Furthermore, if CDC50 proteins catalytically contribute to ATP8B1-mediated PS flipping, mutations in catalytic CDC50 domains may result in PFIC1, BRIC1, and ICP.

In conclusion, we have shown that ATP8B1 is involved in the internalization of PS. Our data implicate a role for CDC50 proteins as possible β-subunits or chaperones for human P4 ATPases that are pivotal factors in the trafficking of P4 ATPases. More specifically, our data suggest that CDC50A is a potential binding partner for ATP8B1 in hepatocytes. Finally, if CDC50 proteins are in vivo accessory factors important for the trafficking and activity of P4 ATPases, they may be important determinants of yet unknown ATP8B1-associated and other P4 ATPase–associated diseases.

Acknowledgements

We thank Dr. R.E. Pagano, Dr. D. Cassio, and Dr. A.K. Nussler for the CHO-K1 and UPS-1 cell lines, WIF-B9 cell line, and AKN-1 cell line, respectively, and Dr. L.W. Klomp, Dr. C.J.M. de Vries, and Dr. J.B. Koenderink for the polyclonal antibody to Atp8b1, the monoclonal antibody to HA, and the polyclonal antibody to Atp1a1, respectively. The mouse monoclonal antibodies UH3 and H4B4, developed by Dr. B.L. Granger, Dr. S. Uthayakumar, J.T. August, and J.E.K. Hildreth, were obtained from the Developmental Studies Hybridoma Bank, which was developed under the auspices of the National Institute of Child Health and Human Development and is maintained by the Department of Biological Sciences, University of Iowa (Iowa City, IA).

Ancillary