Volume 141, Issue 5, Pages 1927-1937.e4 (November 2011) Complementary Functions of the Flippase ATP8B1 and the Floppase ABCB4 in Maintaining Canalicular Membrane Integrity  Annemiek Groen, Marta Rodriguez Romero, Cindy Kunne, Sarah J. Hoosdally, Peter H. Dixon, Carol Wooding, Catherine Williamson, Jurgen Seppen, Karin van den Oever, Kam S. Mok, Coen C. Paulusma, Kenneth J. Linton, Ronald P.J. Oude Elferink  Gastroenterology  Volume 141, Issue 5, Pages 1927-1937.e4 (November 2011) DOI: 10.1053/j.gastro.2011.07.042 Copyright © 2011 AGA Institute Terms and Conditions

Figure 1 Cellular expression of ABCB4 in the presence and absence of ATP8B1 and CDC50. (A) Western analysis of mutant and wild-type ABCB4 in HEK293T whole cell lysates. Lane 1, mock-transfected cells; lane 2, wild-type ABCB4 alone; lane 3, ABCB4 mutant K435M alone; lane 4, ABCB4 mutant E558Q alone. Upper panel was probed with the anti-ABCB family, monoclonal antibody, C219. Lower panel was probed with anti−β-tubulin as a loading control. (B) Western analysis of transporters in HEK293T whole cell lysates. Lane 1, vector control; lane 2, wild-type ABCB4 alone; lane 3, wild-type ABCB4 coexpressed with ATP8B1; lane 4, wild-type ABCB4 coexpressed with ATP8B1 and CDC50A. The top panel was probed with C219. The second panel was probed with the anti-his antibody to detect the histag on ATP8B1. The third panel was probed with anti-HA to detect the hemeagglutinin tag on CDC50A. The lower panel was probed with anti-Na+K+ adenosine triphosphatase (α-1 subunit) as a loading control. (C) ABCB4 localization by immunofluorescence. ABCB4 was detected using primary antibody (P3II-26) (middle column). Left column: Nuclear 4′,6-diamidino-2-phenylindole staining. The right-hand column shows the overlay. Counting from the top: row 1, vector-only control; row 2, ABCB4 only; row 3, ABC4 coexpressed with ATP8B1; row 4, ABCB4 coexpressed with ATP8B1 and CDC50A. (D) Cytotoxicity was estimated by measurement of LDH released from cells transfected as indicated. n = 3; *Significantly different (P < .01) from cells transfected with wild-type ABCB4. NS, no significant difference. (E) TC-stimulated phosphatidylcholine efflux by ABCB4 in the presence of the ATP8B1/CDC50A complex. Cells transiently coexpressing wild-type ABCB4 with ATP8B1 and CDC50A were fed 3H-choline for 48 h. The radioactivity effluxed to the medium (in the presence of the indicated concentrations of TC) is presented as the percentage of the total radioactivity in the dish after subtraction of the background level from similarly treated mock-transfected cells (n = 3; *P < .01, **P < .05). Gastroenterology 2011 141, 1927-1937.e4DOI: (10.1053/j.gastro.2011.07.042) Copyright © 2011 AGA Institute Terms and Conditions

Figure 2 Bile formation in mice of various genotypes. Mice were cannulated and bile was collected for 90 min. TC was intravenously infused at increasing rates that increased every 30 min with 400 nmol/min × 100 g (400–1200 nmol/min × 100 g). (A, B) Bile salt excretion. (C, D) Phospholipid excretion. (E, F) Cholesterol excretion. Left panels: wild-type mice and Atp8b1G308V/G308V mice as indicated. Right panels: Abcb4−/− mice and MF mice as indicated. Panel A also depicts the time frame of bile depletion (0–90 min) and the subsequent period of TC infusion 90–180 min). Data represent averages from 5–7 animals per group with standard deviations. Gastroenterology 2011 141, 1927-1937.e4DOI: (10.1053/j.gastro.2011.07.042) Copyright © 2011 AGA Institute Terms and Conditions

Figure 3 Biliary excretion of ectoenzymes and biliary phospholipid species in various genotypes. (A) Experimental conditions as in Figure 2. Top panels: Aminopeptidase N (APN). Lower panels: alkaline phosphatase (ALP). Left panels: Closed squares: wild-type mice; open squares: Atp8b1G308V/G308V mice. Right panels: closed circles: Abcb4−/− mice; open circles: MF mice. Data represent averages from 5–7 animals per group with standard deviations. The difference in both APN and ALP secretion between Atp8b1G308V/G308V and MF mice during the infusion phase was significant (P < .05). (B) Samples from the experiment of Figure 2 (taken at 160 min) were analyzed for phospholipid species by high-performance thin layer chromatography. Data were normalized for the amount of total bile salt in the same sample. Asterisks indicate a significant difference (P < .05) with the corresponding group lacking Atp8b1. Data represent averages from 5–7 animals per group with standard deviations. Note that the ordinate has a logarithmic scale. Gastroenterology 2011 141, 1927-1937.e4DOI: (10.1053/j.gastro.2011.07.042) Copyright © 2011 AGA Institute Terms and Conditions

Figure 4 Electron microscopic analysis of liver tissue from Atp8b1G308V/G308V, Abcb4−/−, and MF mice. (A) Atp8b1G308V/G308V liver specimen; the canalicular lumen contains membranous material as we have previously reported for this transgenic strain. (B) Abcb4−/− liver; the canalicular lumen does not contain membranous material. (C) MF liver; the canalicular lumen does not contain membranous material. In all cases, there is loss of microvilli as compare to control liver (not shown). Original magnification of all micrographs: 4200×. Bar (panel B) = 1 μm. Gastroenterology 2011 141, 1927-1937.e4DOI: (10.1053/j.gastro.2011.07.042) Copyright © 2011 AGA Institute Terms and Conditions

Figure 5 Hepatic damage and fibrosis in mice of various genotypes. Mice were fed a purified diet supplemented with 0.5% cholate for 7 days. (A) Serum alanine aminotransferase activity. Data represent mean values from 5 animals in each group with standard deviations. #Significant difference between the indicated samples (P < .05). *Significant difference compared to wild-type (P < .05). (B) Livers sections stained with picro-Sirius Red. (C) Stained sections were scored in a blinded fashion on a scale from 1 to 5 as described.14 Closed bars, control diet; open bars, cholate-supplemented diet. Data represent averages of 5 sections from 5 animals in each group with standard deviations. *Significant difference between the indicated samples (P < .05). Gastroenterology 2011 141, 1927-1937.e4DOI: (10.1053/j.gastro.2011.07.042) Copyright © 2011 AGA Institute Terms and Conditions

Figure 6 Model of the mechanism by which ATP8B1 and ABCB4 have a complementary function in the protection of the canalicular membrane against bile salts. (A) The function of ATP8B1 is to flip PS from the outer to the inner leaflet of the plasma membrane. This reduces the concentration of PS in the outer leaflet of the membrane and thereby reduces the total amount of glycerophospholipids in this leaflet. The glycerophospholipids (PC, phosphatidylethanolamine, and PS) stimulate a liquid-disordered phase in the membrane, which makes the membrane sensitive to detergents like bile salts. Hence, ATP8B1 reduces the sensitivity of the membrane toward bile salts. (B) ABCB4 flops PC from the inner to the outer leaflet of the canalicular membrane making PC available for extraction by bile salts. A fraction of the flopped PC may end up in the outer leaflet of the membrane, thereby increasing the liquid-disordered state and increasing the sensitivity to bile salts. Simultaneously, ATP8B1 flips PS and thus reduces the liquid-disordered state. Hence, particularly in the presence of ABCB4 function, ATP8B1 is important to keep the liquid-disordered state to a minimum. Gastroenterology 2011 141, 1927-1937.e4DOI: (10.1053/j.gastro.2011.07.042) Copyright © 2011 AGA Institute Terms and Conditions

Supplementary Figure 1 Transfection efficiency. Cells were transfected with 7.5 mg DNA, of which 2.5 mg was pTracer (Invitrogen) encoding green fluorescent protein (GFP). Forty-eight hours post-transfection the cells were harvested and analyzed by flow cytometry (FACSCAN; Becton Dickinson) for GFP fluorescence (green trace). The GFP-expressing population was compared with mock-transfected cells (orange trace) after gating for normal size and shape. Under these conditions, 77% of the cells expressed GFP (Gate M1). As the cells transfected with plasmids encoding ABCB4, ATP8B1, and CDC50A exhibit no increase in cytotoxicity over mock-transfected cells, we conclude that triple transfection is very efficient under these conditions and that very few, if any, cells express the ABCB4 PC floppase in the absence of the ATP8B1/CDC50A PS flippase. Gastroenterology 2011 141, 1927-1937.e4DOI: (10.1053/j.gastro.2011.07.042) Copyright © 2011 AGA Institute Terms and Conditions

Supplementary Figure 2 The 2 forms of ABCB4 produced differ in their glycosylation status. Two forms of ABCB4 protein with different apparent molecular weights were observed by Western analysis after transient expression in HEK293T cells. Treatment of whole cell lysates with deglycosidase PNGase F (as directed by the manufacturer, New England Biolabs) resulted in a single protein species that comigrated with the original smaller form (compare the untreated sample of ABCB4 K435M in lane 1, with PNGase F-digested sample in lane 2). PNGase F cleaves all N-linked glycans from the protein chain. No effect was observed with EndoH, which cleaves only immature core glycans (not shown). The smaller form, therefore, likely comprises a full-length protein chain that is either unglycosylated or core glycosylated (but if the latter, the 2 core glycans that could be present are insufficient to alter the mobility of the 140-kDa protein under the conditions tested). The larger and smaller forms of the protein with different glycosylation status most likely represent mature ABCB4 in the plasma membrane and immature ABCB4 in the endoplasmic reticulum, respectively. The glycans of membrane proteins are matured in the distal Golgi apparatus before trafficking to the plasma membrane. The ratio between the mature and immature forms therefore reflects subcellular localization and indeed all of the Western data is consistent with the confocal data (ie, for the loss-of-function mutants, or as increasing amounts of the ATP8B1/CDC50 flippase is coexpressed with the wild-type protein, the ratio shifts in favor of the larger mature form). Gastroenterology 2011 141, 1927-1937.e4DOI: (10.1053/j.gastro.2011.07.042) Copyright © 2011 AGA Institute Terms and Conditions

Supplementary Figure 3 Confocal microscopy showing that ABCB4 function is deleterious to cell survival. Cells were grown on glass cover slips treated with poly-L-lysine (Sigma), transfected with the relevant plasmids, fixed, and permeabilized, and stained with 4′,6-diamidino-2-phenylindole (blue) and anti-ABCB4 primary antibody (P3II-26) detected with a green fluorescent secondary antibody (Alexa488-conjugated anti-mouse IgG). Cells were viewed at 63× magnification under oil immersion by confocal microscopy (Zeiss, Leica). (A) Untransfected cells; (B) ABCB4; (C) ABCB4 K435M. Only a few cells were found with staining for wild-type ABCB4, which was consistent with a plasma membrane localization (B). More commonly, a punctate signal was evident (red arrows). We also noted that cells transfected with the pcDNA construct encoding the wild-type ABCB4 showed evidence of nuclear breakdown (white arrows). In contrast, the K435M mutant (C) showed much more staining (according to the Western data, it is more abundantly expressed), and the staining is at the periphery of the cells consistent with a plasma membrane localization. Gastroenterology 2011 141, 1927-1937.e4DOI: (10.1053/j.gastro.2011.07.042) Copyright © 2011 AGA Institute Terms and Conditions

Supplementary Figure 4 Immunofluorescence of ATP8B1 overexpressed in HEK293T cells. The figure shows the localization of ATP8B1-eGFP in green and the endoplasmic reticulum marker (ER) calnexin in red. The right panels give the merged picture of both markers. The localization of ATP8B1 is very similar to that previously reported for overexpression in CHO cells.7 In the absence of CDC50A overexpression, much of the ATP8B1 is in the ER colocalized with calnexin, with only a little at the plasma membrane. Coexpression of ATP8B1 and CDC50A redistributes most of the ATP8B1 out of the calnexin compartment to, or near, the plasma membrane. Gastroenterology 2011 141, 1927-1937.e4DOI: (10.1053/j.gastro.2011.07.042) Copyright © 2011 AGA Institute Terms and Conditions

Supplementary Figure 5 Triple transfection with ATP8B1 and CDC50A produces similar levels of wild-type ABCB4 in HEK293T cells as compared to human liver. In vivo, ABCB4 is restricted to the canalicular membrane of the liver. The canalicular membrane represents a small fraction (ie, about 10% of hepatocyte plasma membranes, so the expression level of ABCB4 in 2.5 μg liver homogenate [lane 1] was compared with 25 μg cell lysate from HEK293T cells transiently transfected with nonfunctional ABCB4 K435M [lane 2] or wild-type ABCB4 plus ATP8B1 and CDC50A [lane 3]). The blot was probed with C219 primary antibody. We judge that the in vitro expression level achieved for ABCB4 is comparable to the in vivo level (there is approximately 2–3-fold more ABCB4 in lane 1, but the canalicular membrane probably contributes significantly less than 10% of liver tissue). Gastroenterology 2011 141, 1927-1937.e4DOI: (10.1053/j.gastro.2011.07.042) Copyright © 2011 AGA Institute Terms and Conditions

Supplementary Figure 6 TC-stimulated phosphatidylcholine efflux by ABCB4 in the presence of the ATP8B1/CDC50A complex. Mock-transfected cells and cells transiently coexpressing wtABCB4, ATP8B1, and CDC50A, were fed 3H-choline for 48 h. The radioactivity effluxed to the medium (in the presence of the indicated concentrations of taurocholate) is presented as the percentage of the total radioactivity in the dish. White bars, mock transfected cells; black bars, cell transfected with ABCB4, ATP8B1, and CDC50A. Statistical analysis by unpaired and 2-tailed Student t test (n = 3; *P < .05; **P < .01). Gastroenterology 2011 141, 1927-1937.e4DOI: (10.1053/j.gastro.2011.07.042) Copyright © 2011 AGA Institute Terms and Conditions

Supplementary Figure 7 Bile composition in mice of various genotypes after feeding a cholate-supplemented diet. Mice were fed a purified diet supplemented with 0.5% cholate for 7 days. Subsequently, bile was collected for 15 min and analyzed for its composition. (A) Aminopeptidase N (APN) activity in bile. (B) Alkaline phosphatase (ALP) activity in bile. Data represent mean values from 5 animals in each group with standard deviations. #Significant difference between the indicated samples (P < .05). *Significant difference compared to wild-type (P < .05). Gastroenterology 2011 141, 1927-1937.e4DOI: (10.1053/j.gastro.2011.07.042) Copyright © 2011 AGA Institute Terms and Conditions

Supplementary Figure 8 The role of genetic background in biliary cholesterol excretion. The Atp8b1G308V/G308V mice were originally bred against 129/SvImJ background. Because the Abcb4−/− mice were bred against the FVB/N background, we have backcrossed Atp8b1G308V/G308V mice against the FVB/N background. We have previously reported on Atp8b1G308V/G308V mice on a C57Bl/6 background.13 Analysis of biliary parameters in wild-type mice of different genetic background has learned that the rate of cholesterol excretion depends on the genetic background. This figure gives the relation between biliary cholesterol and bile salt output of wild-type FVB/N (gray squares), wild-type 129Sv/ImJ (white squares), and wild-type C57Bl/6 mice (black squares) upon TC infusion. This figure shows that bile salt–driven cholesterol secretion is 4-fold higher in FVB/N than in C57Bl/6 mice. Gastroenterology 2011 141, 1927-1937.e4DOI: (10.1053/j.gastro.2011.07.042) Copyright © 2011 AGA Institute Terms and Conditions