INTRODUCTION

Top Abstract Introduction Procedures Results Discussion References

Bile formation is an important function of vertebrate liver (1). It is mediated by hepatocytes that generate bile flow within bile canaliculi by continuous vectorial secretion of bile salts and other solutes across their canalicular (apical) membrane (2). Studies in isolated membrane vesicles of rat and human livers have shown that canalicular bile salt transport is an ATP-dependent process (3-7). However, the molecular identity of the primary active canalicular bile salt transporter or bile salt export pump (BSEP)^1,2 has not yet been elucidated (8, 9). Although the canalicular ecto-ATPase has been proposed as a possible candidate (9, 10), other investigations have provided evidence that BSEP of mammalian liver is an ABC (ATP binding cassette)-type of membrane transporter (11, 12). This assumption has recently been further supported by the cloning of an ATP-dependent bile salt transporter from Saccharomyces cerevisiae (13). This yeast bile salt transporter (BAT1) belongs to a subgroup of ABC-type proteins that includes also the canalicular multiorganic anion transporter or multidrug resistance protein MRP2 (human)/mrp2 (rat) (14-16). Although MRP2/mrp2 mediates canalicular excretion of a broad range of divalent amphipathic anionic conjugates (1, 14, 17), it does not transport primary bile salts such as taurocholate or glycocholate (1, 18). Therefore, we designed degenerate oligonucleotide primers spanning the Walker A and B motifs of ABC proteins and performed reverse transcription-polymerase chain reactin with total rat liver mRNA. One of the amplified fragments revealed an 88% identity with the published pig liver cDNA-fragment of the so called "sister of P-glycoprotein" (spgp), a novel putative canalicular ABC transporter of unknown function (19). Here we report the full-length isolation of spgp from rat liver, demonstrate its function as an ATP-dependent bile salt transporter in cRNA-injected Xenopus laevis oocytes and in transfected Sf9 cells, and document its subcellular localization at the canalicular microvilli and at subcanalicular smooth membrane vesicles of rat hepatocytes using immunofluorescence and immunogold electron microscopy.

	EXPERIMENTAL PROCEDURES

Top Abstract Introduction Procedures Results Discussion References

Materials-- [³H]Taurocholic acid (2.1-4.6 Ci/mmol), [³H]cholic acid (13.2 Ci/mmol), and [⁴C]glycocholic acid (44.6 mCi/mmol) were obtained from NEN Life Science Products. ³H-Labeled taurochenodeoxycholic and tauroursodeoxycholic acids of high specific activity (2-60 Ci/mmol) were prepared as described previously (20-22). All other chemicals and reagents were of analytical grade and were readily available from commercial sources.

cDNA Cloning of Rat Liver spgp-- A cDNA probe spanning the Walker A and B motifs of ABC proteins was constructed by the reverse transcription-polymerase chain reaction of rat liver poly(A)⁺ RNA using the following primers: forward primers, GGCGGATCCTCIGGIKSIGGIAARAGYAC and GGCGGATCCTCIGGIKSIGGIAARTCIIC; reverse primers, CGGGAATTCTCIARIGCRCTIACIGSYTCRTC, CGGGAATTCTCIARIGCRCTIGTIGSYTCRTC, CGGGAATTCTCIARIGCICAIACIGSYTCRTC, and CGGGAATTCTCIARIGCIGAIGTIGSYTCRTC. After cloning and sequencing a single polymerase chain reaction product of 430 bp was identified that revealed an 88% identity to the published portion of the pig liver spgp cDNA (19). It was used to sequentially isolate the full-length spgp cDNA from two cDNA libraries constructed from total rat liver mRNA with the Superscript Plasmid System (Life Technologies, Inc.) and from size-fractionated poly(A⁺) RNA (4.0-6.0 kb) using the ZAP Express^TM cDNA Synthesis and ZAP Express^TM cDNA Gigapack III Gold Cloning kits (Stratagene), respectively. The final single clone was sequenced by Microsynth Gmbh (Balgach, Switzerland).

Functional Expression of spgp in X. laevis Oocytes-- Oocytes were prepared, injected with cRNAs, cultured, and preloaded with 28.5 pmol of [³H]taurocholate (60 nl of 475 µM [³H]taurocholate (2.1 Ci/mmol)) as described elsewhere (23, 24). [³H]Taurocholate efflux from individual oocytes was determined at 25 °C as described previously (18, 24) using a Na⁺-free incubation medium of 100 mM choline chloride, 2 mM KCl, 1 mM CaCl₂, 1 mM MgCl₂, and 10 mM HEPES/Tris, pH 7.5.

Expression and Functional Characterization of spgp in Sf9 Insect Cells-- The Bac-to-Bac system (Life Technologies, Inc.) was used to generate the recombinant baculovirus. Sf9 cells were infected with virus and kept in culture (27 °C, air) for 72 h. Thereafter, Sf9 cells were scraped from culture dishes, centrifuged at 1400 × g_av, and homogenized with a glass-Teflon tissue homogenizer in 50 mM mannitol, 2 mM EGTA, 50 mM Tris/HCl, pH 7.0, 1 µg/ml leupeptin/antipain, and 0.5 mM phenylmethylsulfonyl fluoride (25). Undisrupted cells, nuclear debris, and large mitochondria were pelleted at 500 × g_av for 10 min. The supernatant was centrifuged for 60 min at 100,000 × g_av. The resulting pellet was resuspended in taurocholate uptake buffer consisting of 50 mM sucrose, 100 mM KNO₃, 12.5 mM Mg(NO₃)₂, and 10 mM HEPES/Tris, pH 7.4. The vesicles were stored frozen in liquid nitrogen. Protein was determined using the BCA assay with bovine serum albumin as a standard (26).

Western and Northern Blotting-- A polyclonal antibody was raised in rabbits against an oligopeptide containing the C-terminal 13 amino acids coupled via cysteine to keyhole limpet hemocyanin (Neosystem, Strasbourg, France) and used for Western blotting as described previously (27). A multiple tissue Northern blot (CLONTECH) (~2 µg of poly(A⁺) RNA per lane) was hybridized with the full-length spgp cDNA clone or with a -actin probe labeled with [-³²P]dCTP by random priming. Hybridization was performed at 65 °C overnight in Amersham hybridization buffer. The blot was washed at 68 °C in 0.2× SSC, 0.1% SDS.

Immunofluorescence and Immunogold Electron Microscopy-- Immunofluorescence studies (27) and immunogold electron microscopy studies on ultrathin frozen-thawed sections of perfusion-fixed (3% formaldehyde, 0.1% glutaraldehyde) liver were performed as described previously (28, 29).

	RESULTS

Top Abstract Introduction Procedures Results Discussion References

The isolated full-length spgp cDNA³ contained 5,036 base pairs and encoded a polypeptide of 1321 amino acids with 12 putative membrane-spanning domains, four predicted N-glycosylation sites in the first extracellular loop, and the typical structural features of ABC-transporting polypeptides (Fig. 1). Comparison of the full-length amino acid sequence with other members of the ABC-transporter superfamily revealed the following similarities/identities to spgp: mdr1a (mouse) 71%/50%, mdr1b (rat) 70%/49%, mdr2 (rat) 69%/48%, MRP1 (human) 51%/26%, mrp2 (rat) 49%/25%, BAT1 48%/23%, and cystic fibrosis conductance regulator (mouse) 48%/22%. These data confirm the close homology of spgp to the MDR/mdr gene family (19).

View larger version (42K): [in this window] [in a new window]   Fig. 1.   Amino acid sequence of rat liver spgp. Potential transmembrane segments predicted by using the algorithm of Eisenberg et al. (30) and hydropathy values of Kyte and Doolittle (31) are underlined twice (M1-M12). The Walker A and B motifs of the putative nucleotide binding folds are underlined once (WA and WB). The ABC transporter family signatures (32) are underlined by dotted lines (FS). Potential extracellular N-glycosylation sites are at amino acids 109, 116, 122, and 125.

To elucidate its function, spgp was expressed in X. laevis oocytes (18) and in Sf9 cells. As illustrated in Fig. 2, an approximately 1.5-fold increase in [³H]taurocholate efflux was observed in spgp cRNA-injected oocytes as compared with water- and mrp2 cRNA-injected oocytes (18). In addition, membrane vesicles isolated from spgp cDNA-infected Sf9 cells (see "Experimental Procedures") demonstrated marked ATP-dependent uptake of taurocholate, whereas control vesicles from wild type cells (or cells infected with wild type or with -galactosidase-containing baculovirus; data not shown) did not (Fig. 3A). Initial rates of this spgp-mediated, ATP-dependent [³H]taurocholate uptake exhibited saturability with increasing concentrations of taurocholate (K_m value ~5.3 µM) (Fig. 3B). Furthermore, only minimal or no stimulation of [³H]taurocholate uptake into spgp expressing Sf9 vesicles was observed with other nucleotides (UTP, CTP, and GTP), ATP degradation products (ADP, AMP, and adenosine), or nonhydrolyzable ATP analogs (adenosine-5'-[-thio]triphosphate and adenosine-5'-[,-imido]triphosphate) (data not shown). Also, increasing concentrations of vanadate between 5 µM and 100 µM inhibited ATP-dependent [³H]taurocholate uptake between 32% and 63%. And finally, comparison of initial ATP-dependent uptake rates (linear time-dependent uptake phase) of various bile salt derivatives in spgp-expressing Sf9 cell vesicles and in isolated canalicular liver plasma membrane vesicles (6) showed the highest ATP-dependent uptake rate for taurochenodeoxycholate followed by tauroursodeoxycholate and taurocholate in both vesicle preparations (Table I). In contrast, glycocholate and cholate were less efficiently transported in canalicular vesicles, and no ATP-dependent transport of these bile salt derivatives could be detected in spgp-expressing Sf9 cell vesicles. These latter results most probably resulted from the overall lower expression of ATP-dependent bile salt transport activities in spgp-expressing Sf9 cells compared with canalicular vesicles (Table I), thus preventing the detection of low residual spgp-mediated cholate and glycocholate transport.

View larger version (20K): [in this window] [in a new window]   Fig. 2.   The spgp-mediated taurocholate efflux from X. laevis oocytes. Oocytes were injected with spgp cRNA (5 ng) (), mrp2-cRNA (5 ng) () or water (). [3H]Taurocholate efflux was determined as described under "Experimental Procedures." Values represent the means ± S.D. of 8-10 determinations.

View larger version (13K): [in this window] [in a new window]   Fig. 3.   ATP-dependent taurocholate transport in membrane vesicles isolated from spgp-expressing Sf9 cells. A, effect of ATP on taurocholate uptake. Membrane vesicles isolated from wild type (control) and spgp-expressing Sf9 cells were incubated in 50 mM sucrose, 100 mM KNO3, 12.5 mM Mg(NO3)2, and 10 mM HEPES/Tris, pH 7.4, supplemented with 2 µM [3H]taurocholate. Vesicle uptake of [3H]taurocholate was determined in the presence and absence of ATP (5 mM) (6). , control vesicles without ATP; , control vesicles with ATP; , spgp-expressing vesicles without ATP, , spgp-expressing vesicles with ATP. Data represent the means ± S.D. of three determinations with the variability being smaller than the symbols. B, saturability of spgp-mediated ATP-dependent taurocholate uptake. Initial uptakes of increasing taurocholate concentrations were determined at 30 s in the presence and absence of ATP (5 mM). ATP-dependent uptake values were determined by subtraction of uptake in the absence from uptake in the presence of ATP. The values are the means ± SD of triplicate determinations in one out of two separate experiments.

To directly correlate transport activity with protein expression in Sf9 cells, a polyclonal antiserum against the C-terminal 13 amino acids of BSEP/spgp was raised in rabbits. The antiserum recognized a predominant protein band with an apparent molecular mass of ~140 kDa (140.5 ± 2.7; mean ± SD, n = 3) in vesicles isolated from BSEP/spgp cDNA-transfected Sf9 cells (Fig. 4A, lane 2), but not in vesicles isolated from wild type cells (Fig. 4A, lane 1) nor in vesicles isolated from Sf9 cells infected with wild type or with -galactosidase containing baculovirus (data not shown). Thus, the expression of BSEP/spgp correlates with ATP-dependent taurocholate transport. An immunopositive protein band was also observed in canalicular, but not in basolateral plasma membrane vesicles isolated from rat liver (Fig. 4A, lanes 3 and 4). This canalicular form of BSEP/spgp exhibited a slightly higher molecular mass of ~160 kDa (157.4 ± 3.1; mean ± SD, n = 3), which most probably reflects a different N-linked glycosylation pattern of hepatocytes as compared with Sf9 cells (25). Based on Northern blot analysis of various rat tissue mRNAs, BSEP/spgp is predominantly, if not exclusively, expressed in the liver (Fig. 4B). When tested for its presence in other animal species the BSEP/spgp cDNA yielded positive hybridization reactions also with mRNAs of mouse (~5.0 kb), chicken (~2.0 kb), and turtle (~4.0 kb) livers (data not shown), indicating the occurrence of BSEP/spgp-related proteins also in lower vertebrates.

View larger version (103K): [in this window] [in a new window]   Fig. 4.   Tissue distribution, cellular expression, and subcellular localization of BSEP/spgp. A, Western blot analysis of membrane vesicles isolated from Sf9 cells and rat liver. Lane 1, membrane vesicles (75 µg of protein) from wild type Sf9 cells. Lane 2, membrane vesicles (75 µg of protein) from Sf9 cells transfected with BSEP/spgp cDNA. Lane 3, canalicular plasma membrane vesicles (100 µg of protein) from rat liver (33). Lane 4, basolateral plasma membrane vesicles (100 µg of protein) from rat liver (33). Lane 5, molecular mass standards. B, Northern blot analysis of various rat tissue mRNAs. The different lanes represent mRNAs of 1, heart; 2, brain; 3, spleen; 4, lung; 5, liver; 6, skeletal muscle; 7, kidney; and 8, testis. A human -actin probe served as a control. The different -actin size transcripts are due to the detection of various isoforms of -actin mRNAs as specified by the supplier of the multiple tissue Northern blot. C-F, immunogold electron microscopic localization of BSEP/spgp in rat hepatocytes. C, gold particle labeling is restricted to the canalicular (apical) plasma membrane (Ca). Lateral plasma membranes (arrowheads) are unlabeled. D, labeling is enriched in the canalicular microvilli but absent from the smooth intermicrovillar plasma membrane regions (arrowheads). E and F, in addition to the apical plasma membrane, membrane invaginations (arrow in E), subplasmalemmal vesicles, and electron lucent vacuoles (V) exhibit labeling. Besides hepatocytes, no other cell type of rat liver was labeled. Scale bars, 0.5 µm (C), 0.6 µm (D), and 0.8 µm (F and G).

The cellular and subcellular distribution of BSEP/spgp in rat liver was further investigated by immunofluorescence and immunogold electron microscopy. At low magnification positive immunoreactivity was confined homogenously to bile canaliculi of periportal and perivenous hepatocytes (data not shown). No immunoreactivity was detected in bile ductular epithelial cells. Within bile canaliculi the immunoreactivity for BSEP/spgp was predominantly associated with the canalicular microvilli, while the smooth intermicrovillar plasma membrane regions were essentially free of gold particle labeling (Fig. 4, C and D). No labeling was found along the basolateral plasma membrane of hepatocytes. Intracellularly, gold particle labeling was associated with subplasmalemmal smooth membrane vesicles, lucent vacuoles probably representing endosomes (Fig. 4, E and F) and the Golgi apparatus.

	DISCUSSION

Top Abstract Introduction Procedures Results Discussion References

This study has identified spgp (19) as a canalicular ATP-dependent BSEP of mammalian liver. This conclusion is derived from the functional expression of the full-length BSEP/spgp cDNA in X. laevis oocytes (Fig. 2) and in Sf9 insect cells (Figs. 3 and 4A). Vesicles isolated from BSEP/spgp-overexpressing Sf9 cells exhibited similar high affinity ATP-dependent taurocholate uptake (K_m = ~5.3 µM) (Fig. 3B) as previously determined under identical conditions in canalicular rat liver plasma membrane vesicles (K_m = ~2.1 µM) (6). Since canalicular secretion represents the rate-limiting step in the overall transport of bile salts from sinusoidal blood plasma into bile (2, 8), its high affinity for taurocholate indicates that BSEP/spgp represents the major canalicular bile salt transporter of rat liver. Based on Northern blot analysis, BSEP/spgp is predominantly, if not exclusively expressed in the liver (Fig. 4B) (19). This organ-selective expression of BSEP/spgp is not astonishing considering its bile salt transport function. However, BSEP/spgp-related proteins appear also to occur in the livers of other vertebrates including chicken and turtle (this study) as well as the winter flounder (34), supporting the concept that the origin of the BSEP/spgp gene predates the divergence of fish and mammals (19) and that BSEP/spgp mediates the transport of C₂₇ bile acids (present in turtle) as well as C₂₄ bile acids (present in most mammals and chickens). Finally, Western blot analysis as well as immunofluorescence and immunogold labeling studies demonstrated the selective localization of BSEP/spgp at the canalicular microvilli (Fig. 4, A, C, and D) and at subcanalicular smooth membrane vesicles (Fig. 4, E and F) of rat hepatocytes. Since infusion of taurocholate into intact animals has been shown to increase the canalicular bile salt transport capacity (2, 35), the labeled subcanalicular endosomes might represent a regulatory carrier pool that could be rapidly inserted into the canalicular plasma membrane under high bile salt load conditions.

The full length of the rat liver BSEP/spgp represents a canalicular protein with a molecular mass of ~160 kDa (Fig. 4A). Its homology to the MDR/mdr gene family is in contrast to the closer relatedness of the recently cloned yeast BAT1 with the MRP/mrp gene family of ABC transporters (13). Since mammalian MRP/mrp analogues do not mediate transport of monovalent bile salt derivatives (1, 4, 17, 18), and since BAT1 has been shown to exhibit an approximately 12-fold lower affinity for taurocholate (K_m = ~63 µM) (13) as compared with BSEP/spgp (Fig. 3B), the participation of BAT1-related gene products in bile salt transport in mammalian liver seems unlikely. Furthermore, since fungi are thought to be devoid of bile salts, the true endogenous substrate(s) of BAT1 as well as the functional significance of bile salt transport in yeast remain unclear. Nevertheless, ATP-dependent taurocholate transport has also been demonstrated in plant vacuoles despite the absence of typical bile salts in plant tissue (36). Hence, amphipathic bile salt transport might be a nonspecific "bystander" function of several multispecific ABC-type transporters in a variety of biological tissues. In contrast, ongoing bile salt secretion serves an essential function in mammalian liver, where it mediates phospholipid and cholesterol excretion and water flow into bile canaliculi. Furthermore, in vertebrates bile salts are 100-1000-fold concentrated in bile as compared with plasma, stimulate absorption of lipids from the intestine and undergo efficient enterohepatic circulation (2, 8, 9). Therefore, it is conceivable that vertebrates have developed a separate high affinity bile salt export pump within the large superfamily of ABC-type proteins.

Several observations from this and other recent studies strongly indicate that BSEP/spgp represents the major, if not the only bile salt export pump in mammalian liver. First, BSEP/spgp-mediated bile salt transport exhibits similar kinetics (see above) and similar substrate preference (Table I) as ATP-dependent bile salt transport in canalicular liver plasma membrane vesicles (6). Second, preferential BSEP/spgp-mediated transport of the dihydroxy-conjugated bile salt taurochenodeoxycholate is consistent with the preferential secretion and transport of this bile salt in the isolated perfused liver and in canalicular plasma membrane vesicles (37). Third, in contrast to the ecto-ATPase canalicular expression of BSEP/spgp is decreased in various forms of cholestatic liver disease (38-40), and its level of expression correlates well with the activity of ATP-dependent canalicular bile salt transport (37, 38). Fourth, overexpression of BSEP/spgp is associated with increased bile salt secretion in a gallstone susceptible mouse strain (41). And fifth, the BSEP/spgp gene is localized on the same chromosome (chromosome 2) as the locus for progressive familial intrahepatic cholestasis PFIC2 (41-43), which is characterized by defective canalicular bile salt secretion in the presence of normal serum cholesterol and normal plasma -glutamyltranspeptidase (44). These patients secrete exceptionally low levels of conjugated primary bile salts into bile and have particularly low levels of biliary chenodeoxycholic acid (45, 46), which represents the preferred substrate of BSEP/spgp (Table I). In fact, and most importantly, in preliminary studies missense mutations of the human liver SPGP gene have been identified as the most probable cause of PFIC linked to the PFIC2 locus (42).⁴ These very recent developments corroborate the findings presented in this study and further support the conclusion that BSEP/spgp represents the major, if not the only canalicular bile salt transporter of mammalian liver.