A Yeast ATP-binding Cassette-type Protein Mediating ATP-dependent Bile Acid Transport*

ATP-dependent transport of bile acids is a key determinant of bile flow in mammalian liver and is associated with cholesterol excretion, gallstone formation, and numerous inherited and acquired hepatobiliary diseases. Secretory vesicles and a vacuole enriched fraction purified from Saccharomyces cerevisiaealso exhibit ATP-dependent bile acid transport. ATP-dependent transport of bile acids by the vacuolar fraction was independent of the vacuolar proton ATPase, responded to changes in the osmotically sensitive intravesicular space, and was saturable, exhibiting a K m of 63 μmfor taurocholate. The BAT1 (bileacid transporter) gene was isolated from yeast DNA by polymerase chain reaction amplification using degenerate oligonucleotides hybridizing to conserved regions of ABC-type proteins. ATP-dependent bile acid transport was abolished when theBAT1 coding region was deleted from the genome and restored upon reintroduction of the gene. The deduced amino acid sequence predicts that Bat1p is an ABC-type protein 1661 amino acids in length, similar to mammalian cMOAT/cMRP1 and MRP1 transporters, yeast Ycf1p, and two yeast proteins of unknown function. Information obtained from the yeast BAT1 gene may aid identification of the gene encoding the mammalian bile acid transporter.

Bile acids (BA) 1 are the main product of cholesterol catabolism in mammals. These amphipathic organic detergents are synthesized in the liver and excreted in bile into the intestine, where they are required for lipid emulsification prior to absorption. Over 90% of BA in the intestine undergoes an efficient and regulated process of enterohepatic circulation, in which BAs are taken up from the portal circulation and ultimately secreted into the canaliculus. Transport across the canalicular membrane is rate-limiting in transfer of BA from blood to bile (1)(2)(3) and is a major determinant of bile flow. Purified rat canalicular membrane vesicles (CMV) exhibit ATP-dependent BA transport (4 -6), and a similar activity has been characterized in the microvillous membrane of the human trophoblast (7).
ATP-dependent transport activity in the canalicular membrane is essential for translocation of many bile components from the hepatocyte into the canaliculus. Molecular characterization has indicated that ATP-binding cassette (ABC)-type proteins are responsible for these activities. ABC-type proteins are a large family of integral membrane proteins that effect ATP-dependent membrane translocation of an enormous variety of substrates and are present in organisms ranging from bacteria to man (12). Members of this superfamily typically contain one or two highly conserved nucleotide binding domains (NBD), which are paired with hydrophobic regions capable of spanning the membrane multiple times. ABC-type proteins that have been identified in the canalicular membrane include MDR1, which transports hydrophobic drugs into bile (13); MDR3, a phospholipid flippase (14 -16); and cMOAT/ cMRP1 (17)(18)(19), which is associated with transport of glutathione, glucuronide, and sulfate conjugates across the canalicular membrane. These transporters are not abundant in liver, and the genes encoding these proteins were initially identified in other tissues, or because of their similarity with other ABCtype proteins. MDR1 is highly expressed in multidrug-resistant cell lines, and MDR2 was first identified due to its high degree of homology with MDR1. Similarly, cMOAT/cMRP1 was isolated because it is similar in function and sequence to MRP1 (20), another ABC-type protein involved in multiple drug resistance. However, a mammalian homolog of the gene or genes that encode the ATP-dependent BA transporter has not been identified.
Heterologous expression of MDR1 (21), MDR2 (15), and MRP1 (22) in yeast has been useful in the study of mammalian ABC-type proteins. Secretory vesicles purified from transgenic sec1-1 mutant yeast provide an abundant source of right-sideout membrane vesicles that are ideal for biochemical characterization of transport activity. These vesicles also contain ATP-dependent transport activities for BA and glutathione conjugates, which are similar to activities present in the bile canaliculus (23). ATP-dependent transport of glutathione conjugates is mediated by Ycf1p (24), 2 an ABC-type protein that * This work was supported in part by National Institutes of Health Grants DK51005-01 (to D. F. O.) and DK35652 (to I. M. A.), and by the Center for Gastroenterology Research on Absorptive and Secretory Processes under Public Health Service Grant 1 P30 DK39428. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Preparation of Organelle and Membrane Fractions-Secretory vesicles were prepared from temperature-sensitive sec1-1 strains as described previously (23). Vacuoles were prepared as described (32) with the following modifications. Buffers A, B, and C were supplemented with the protease inhibitors as follows: pepstatin A (2 g/ml), leupeptin (2 g/ml), aprotinin (2 g/ml), and phenylmethylsulfonyl fluoride (1 mM). Glycerol was added to vacuoles in buffer C to a final concentration of 5% (v/v) before freezing at Ϫ80°C. Protein concentrations were determined by the method of Lowry et al. (33) on trichloroacetic acidprecipitated samples. Enzyme assays for marker enzymes were carried out as described (32). As has been reported (14,16), vacuoles prepared by the Ficoll flotation procedure are highly enriched in vacuolar markers ␣-mannosidase and alkaline phosphatase (20-and 15-fold increase in specific activity relative to total homogenate, respectively). There is little or no contamination (less than 5% of the specific activity observed with the total homogenate) with enzyme markers for mitochondria (succinate dehydrogenase), endoplasmic reticulum (cytochrome c-reductase), or cytoplasm (glucose-6-phosphate dehydrogenase). Total membranes were prepared from yeast grown in SG. Yeast cells were disrupted by bead beating in five volumes of 20 mM MES/Tris, pH 8.0, 50 mM KCl, 1 mM EDTA, 1 mM dithiothreitol, 10 mM phenylmethylsulfonyl fluoride, 2 M pepstatin A, and 2 M leupeptin. Cell debris was pelleted by centrifugation at 250 ϫ g for 10 min. Total membranes were pelleted by centrifugation at 100,000 ϫ g, resuspended in disruption buffer, and frozen at 80°C.
Transport Assays-ATP-dependent transport of [ 3 H]DNPSG and BA was measured by the standard vesicle filtration assay. Transport by secretory vesicles was determined as described previously (23). Vacuolar transport was measured as follows; radiolabeled substrate was added to 50 l of 2 ϫ reaction buffer (20 mM MES/Tris, pH 8.0, 10 mM ATP, 20 mM MgCl 2 , 10 mM creatine phosphate, 10 g/ml creatine kinase) and mixed with 50 l of vacuolar vesicles (20 -30 g of protein) prewarmed to 30°C. At the specified time points, 15-25-l aliquots of the reaction mix were diluted to 1 ml with ice-cold wash buffer (10 mM MES/Tris, pH 8.0, 5 mM MgCl 2 , 20 mM KCl) and filtered under mild vacuum through glass fiber filters (25-mm GF/C, Whatman). The filter disks were washed with 5 ml of wash buffer, dried, and the retained counts determined by liquid scintillation.
Gene Disruptions-Disruption constructs were generated by PCR using chimeric oligonucleotide primer pairs (30), which consisted of 12-15 nucleotides homologous to sequences flanking the URA3 gene insert in the Yep24 plasmid (34) on the 3Ј end linked with 40 -45 nucleotides homologous to yeast genomic sequences flanking the gene of interest. PCR amplification of a Yep24 template using these primers generated DNA fragments consisting of the URA3 gene flanked by short stretches of homology to the gene of interest. Transformation of the RSY12 strain with these PCR products resulted in replacement of chromosomal DNA between the regions of microhomology with the URA3 gene. DNA was prepared from URAϩ transformants and deletion of the chromosomal copy of the gene confirmed by Southern analysis. PCR conditions used were 3 times at 95°C, 45°C, and 72°C; and PCR-Degenerate oligonucleotide primers used to isolate novel ABCtype genes were designed by reverse translation of the consensus aa sequences TGAGKS (forward: GTCGAATTC(A/T)CIGGIGCIGGIAA(A/ G)TC) and DEATA (reverse: GTCGAATTCG(A/C)IG(A/C)IGTIGCT(T/ C)TC(A/G)TC). PCR parameters were 3 ϫ 95°C, 40°C, 72°C; and 30 ϫ 95°C, 55°C, 72°C (all steps for 1 min). PCR products of 400 -450 bp were gel-purified and ligated with the pCRII vector (Invitrogen). Plasmid inserts were classified into groups by restriction mapping and sequenced.
Plasmids-Three clones hybridizing to the gene "C" PCR fragment were isolated from a Saccharomyces cerevisiae genomic DNA library in YES-R bacteriophage (ATCC 7256) (37). Transfection of the Escherichia coli strain BNN132 with purified phage resulted in CRE-LOX excision of yeast shuttle vector pSE396 containing yeast DNA inserts 4 -6 kb in length. Inserts from two plasmids were sequenced and combined using a unique SphI site residing 2429 bp downstream of ATG to generate an 8-kb insert containing the 5-kb open reading frame flanked by 1.5 kb of upstream and downstream sequence, respectively. In pAA17 the NruI-SmaI fragment containing the URA3 gene in pSE936 was replaced with a HindIII fragment containing the LEU2 gene. pAA43 was generated by insertion of the 8-kb BAT1 genomic fragment in pAA17.
Antibody Production and Immunoblot Analysis-A 318-bp PCR fragment encoding aa 900 -1006 of Bat1p was subcloned into the pQE70 expression vector (QIAGEN), which ligates it in frame with an ATG codon on the 5Ј end and a six-histidine coding region on the 3Ј terminus. Induction of E. coli containing this plasmid with isopropyl-1-thio-␤-Dgalactopyranoside led to overexpression of a 12.5-kDa peptide, which was purified by affinity chromatography on a nickel-agarose column (QIAGEN) and used to immunize two rabbits. Antisera from both rabbits reacted strongly on immunoblots with the 12.5-kDa peptide expressed in E. coli and with a 125-190-kDa protein in extracts derived from RSY12, or yeast strains harboring the pAA43 plasmid containing the BAT1 gene. Little or no cross-reactivity could be detected with extracts prepared from E. coli lacking the expression plasmid, or DOY11bat1⌬2 yeast containing the pAA17 empty vector.
IgG was purified from rabbit antisera on a protein A-Sephadex (Sigma) column (38). Two milligrams of purified histidine-tagged Bat1p (see above) antigen were bound to Affi-Gel 10 (Bio-Rad) matrix according to the manufacturer's instructions and used for affinity purification of Bat1p specific antibodies (38).
For SDS-PAGE analysis an equal volume of sample loading buffer (5% SDS, 5% dithiothreitol, 50 mM Tris-HCl, pH 6.8, 10% glycerol) was added to total membrane and vacuolar fractions and heated at 75°C for 8 min. Equal amounts of protein were separated on 12% SDS-polyacrylamide gels overlaid with a 4% stacking gel. Polyacrylamide gels were prepared from National Diagnostics reagents according to the manufacturer's instructions. Immunoblots were done essentially as described (39) using affinity-purified IgG (40 ng/ml) and goat anti-rabbit IgG (Bio-Rad) conjugated with horseradish peroxidase as secondary antibody. Immunoblots were developed using enhanced chemiluminescence (DuPont). Membrane samples were delipidated by ethanol-acetone extraction or ether extraction of trichloroacetic acid-precipitated samples as described (40). Immunoprecipitation of Bat1p was done essentially as described (38). Vacuolar membranes (200 g of protein) were solubilized with RIPA buffer containing pepstatin A (2 g/ml), leupeptin (2 g/ml), aprotinin (2 g/ml), and phenylmethylsulfonyl fluoride (1 mM) and incubated overnight with 20 g of affinity-purified IgG. IgG-Bat1p complexes were bound to Protein A-Sephadex beads for 1 h (Sigma), washed three times with RIPA buffer, and dissolved in sample loading buffer by heating at 75°C before running of SDS-polyacrylamide gel electrophoresis. Proteins were electroblotted unto PVDF-Immobilon (Millipore) in 100 mM CAPS-piperidine, pH 9.5. The bound proteins were stained with Coomassie Blue, and the Bat1p-containing membrane fragment was submitted for aa sequence determination using an Applied Biosystems Inc. model 477 protein sequencer.

RESULTS
A BA Transporter in the Vacuolar Fraction-Yeast secretory vesicles exhibited ATP-dependent transport of various BAs (see Table I and Ref. 23). Conjugated BAs were transported more efficiently than unconjugated BAs. A subcellular fraction enriched in vacuolar enzyme markers also exhibited ATP-dependent transport of [ 3 H]taurocholate (Fig. 1A). The rate of BA transport in the vacuolar fraction was dependent on substrate concentration and was saturable. ATP-dependent transport as a function of taurocholate concentration can be fitted assuming Michaelis-Menten kinetics with a predicted K m of 63 M and V max of 14.9 nmol/mg/min (Fig. 1B). Because the specific activity and substrate affinity of ATP-dependent BA transport was higher in the vacuolar fraction than in purified secretory vesicles, and because preparation of the vacuolar fraction does not require temperature-sensitive strains, subsequent experiments were performed with the vacuolar fraction.
Transport of BAs into membrane-bound vesicles should be sensitive to changes in the intravesicular space. Increasing the osmolarity of the reaction mix inhibited ATP-dependent transport of taurocholate in vacuolar vesicles (Fig. 1C). When the rate of BA transport was plotted against the inverse of the sucrose concentration in the reaction mix, a straight line with an origin of approximately zero was obtained, indicating that taurocholate was translocated into the vesicle rather than merely binding to the membrane.
BA transport by the vacuolar fraction was virtually nonexistent at 4°C, in the absence of ATP, or when ATP was replaced by the non-hydrolyzable ATP analog, AMP-PNP. Vacuolar ATP-dependent BA transport was not inhibited by azide, an inhibitor of the mitochondrial proton ATPase, bafilomycin A 1 , an inhibitor of the vacuolar proton pump, or ionophores that dissipate membrane potential and ⌬pH (Table II). NH 4 Cl reduces uptake of taurocholate by 31%; however, the same degree of inhibition is observed when vacuoles lacking V-ATPase activity are used (see below), suggesting that the effect represents direct inhibition of the BA transporter rather than quenching of the ⌬pH. ATP-dependent BA transport in the vacuolar fraction did not depend on activity of the V-ATPase. SF838 -5Atfp1-⌬8 yeast lack the vma1 gene (31), which codes for the catalytic subunit of the V-ATPase, and consequently do not acidify the vacuolar space. Vacuoles purified from this strain exhibit no significant difference with regard to BA transport when compared with the vacuolar fraction prepared from an isogenic strain expressing VMA1 (Fig. 2). 100 M sodium orthovanadate inhibits ATP-dependent BA transport by 56%. The plasma membrane proton ATPase Pma1p is also inhibited by vanadate with a K i of 1 M (41). However, BA transport was not significantly affected by 10 M vanadate, which inhibits Pma1p activity by 80%. This is consistent with previous findings, which show that secretory vesicles lacking Pma1p are still capable of vigorous taurocholate transport (23). Thus, vanadate appears to inhibit the vacuolar BA transporter directly, rather than by abrogation of Pma1p activity. These data indicate that BA transport in the yeast vacuolar fraction is energized directly by ATP.
Identification of the Gene Encoding the BA Transporter-Transport mediated by a number of ABC-type proteins is energized directly by ATP and, in some cases, is inhibited by vanadate, suggesting that BA transport may be mediated by an ABC-type protein. The YCF1 gene encodes an ABC-type protein that is sorted primarily to the yeast vacuole and transports glutathione conjugates, which are also amphipathic organic anions. However, vacuoles from the ycf1 deletion mutant JWY53 (28), although deficient in ATP-dependent glutathione conjugate transport, were normal with regard to ATP-dependent BA transport, indicating that Ycf1p is not responsible for this activity. Likewise, yeast strains deficient in expression of the ABC-type proteins Ste6p or Pdr5p (42), or which overexpress Snq2p (43) exhibited no alteration in ATP-dependent BA transport (not shown).
To identify other yeast ABC proteins, a homology search of the Saccharomyces Genome Database was performed using a consensus NBD amino acid (aa) sequence. At that time, the search identified seven putative yeast ABC-type proteins of  (23), we hypothesized that the BA transporter was more closely related to Ycf1p than to other ABCtype proteins. Alignment of the Ycf1p aa sequence with Yhd5p, Yk83p, and mammalian MRP1, the proteins in the data base that most closely resembled Ycf1p, identified highly conserved aa sequence motifs in the carboxyl-terminal NBDs. Degenerate oligonucleotide primers, designed by reverse translation of these motifs, were used for PCR amplification of a genomic DNA template derived from DOY2J, a yeast double mutant deleted for the YCF1 and YHD5 genes. PCR products of the size expected for genes encoding ABC-type proteins (400 -450 bp) were cloned and sequenced. Three novel DNA fragments were identified that contained open reading frames exhibiting similarity with the NBD of Ycf1p (Fig. 3).
BA transport activity was measured in vacuoles prepared from mutant yeast deleted for the DNA encompassed in the novel PCR fragments. These deletions, although partial, were expected to affect activity of the cognate proteins, as deletion or FIG. 1. Characterization of ATP-dependent transport of taurocholate by yeast subcellular fractions. A, uptake of radiolabeled substrate by purified secretory vesicles (Ç, å) or a subcellular fraction highly enriched in vacuoles (E, q) was measured by the standard vesicle filtration assay in the presence (Ç, E) or absence (å, q) of ATP. Secretory vesicles were prepared from temperature-sensitive NY3 cells, which accumulate secretory vesicles in the cytoplasm when shifted to the 38°C. Vacuoles were purified from NY3 cells grown at 30°C. [ 3 H]Taurocholate was present at a final concentration of 5 M. B, substrate dose response of ATP-dependent BA transport in vacuoles. Data points represent ATP-dependent uptake of radiolabeled substrate during the first minute of the transport reaction using vacuolar vesicles prepared from NY3 yeast. The line was drawn by fitting the Michaelis-Menten equation to the data using the RS/I program (BBN Software products). C, sensitivity of taurocholate transport to reduction in the intravesicular space. Vacuole vesicles were incubated with increasing concentrations of sucrose for 1 min; transport was then initiated by addition of taurocholate to 5 M and Mg-ATP. The initial velocity of ATP-dependent transport was plotted against the inverse of the sucrose concentration in the reaction. Values in all cases represent the mean (n Ն 3) Ϯ standard error (except where smaller than symbol). F IG. 2. ATP-dependent transport of taurocholate does not depend on the activity of the V-ATPase. Vacuoles were prepared from SF838 -5Afp1-⌬8 strain that is deleted for the vma1 gene and the QL1 strain in which a copy of VMA1 has been integrated in the genome of SF838 -5Atfp1-⌬8. Transport of 5 M taurocholate was measured as in Fig. 1 for vacuoles from SF838 -5Atfp1-⌬8 (Ç, å) and QL1 (E, q) in the presence (Ç, E) and absence (å, q) of 5 mM ATP. Values represent the mean (n Ն 3) Ϯ standard error (except where smaller than symbol). alteration of the NBD abolishes transport activity of natural (46) and synthetic (47,48) ABC protein mutants. Vacuolar BA transport in the DOY8 and DOY9 knockout strains, which are partially deleted for genes A and B, respectively, proved indistinguishable from the progenitor RSY12 strain. On the other hand, ATP-dependent transport of [ 3 H]taurocholate was completely abrogated in vacuoles derived from the DOY10 mutant that had been disrupted for gene C.

TABLE II Effect of various inhibitors on ATP-dependent transport of taurocholate by the vacuolar fraction
Analysis of the BAT1 Sequence-DNA clones encompassing PCR fragment C and flanking sequences were isolated from a bacteriophage library of genomic S. cerevisiae DNA by hybridization to radioactively labeled probes. Analysis of the nucleotide sequence of the yeast DNA inserts identified a single long open reading frame encoding a putative 1661-aa protein with an estimated M r of 189,147 (Fig. 4). The nucleotide sequence was derived from DNA clones of genomic origin. Posttranscriptional processing could therefore produce an mRNA that differs significantly from the DNA template. However, PCR fragments generated from genomic DNA and DNasetreated RNA are indistinguishable in size, suggesting that the BAT1 primary transcript does not contain large introns (Fig. 5).
The Bat1p (bile acid transporter) exhibits considerable aa sequence identity with a large number of ABC-type transporters in the protein sequence data bases. BLAST and FASTA homology searches of protein sequence data bases indicated that the aa sequences most closely resembling Bat1p include two putative proteins of unknown function encoded by the S. cerevisiae genes YHD5 (57% amino acid sequence identity with Bat1p) and YK84 (46%), the rat cMOAT/cMRP1 (32%) canalicular multispecific organic anion transporter, the human MRP1 (30%) glutathione conjugate transporter, and the yeast Ycf1p (27%) glutathione conjugate transporter. A cluster analysis showing putative phylogenetic relationships of Bat1p and several other ABC-type proteins is presented in Fig. 6.
The Bat1p deduced aa sequence contains two NBDs typical of ABC type proteins. Both NBDs contain a reasonable match with the consensus Walker A and B motifs as well as the ABC signature sequence. Associated with each NBD is a hydrophobic domain capable of spanning a membrane multiple times (Fig. 5). It is difficult to ascertain the membrane topology of Bat1p from the primary sequence, as the predictions made by different algorithms do not match exactly. Thus TMAP (EMBL server) suggests an arrangement of 9 amino-proximal trans-membrane helices ϩ 5 membrane-spanning domains carboxylterminal to the first NBD, TMpred (49) proposes 11ϩ5, and PhdTopology (50) predicts 11ϩ4. It is unlikely that there are 5 transmembrane domains between the first and second NBD, as this would place the ATP binding domains on opposite sides of the membrane, suggesting that there are 4 or 6 transmembrane helices. Comparison of the hydropathy plots of closely related proteins reveals strong similarity in the arrangement and placement of the transmembrane domains with Yhd5p, Yk83p, MOAT, MRP1, and SUR (51). One notable difference is that Bat1p and the putative Yhd5p and Yk83p proteins appear to contain an additional amino-terminal transmembrane helix that is separated from the next transmembrane domain by a highly polar stretch of approximately 100 aa. This suggests the presence of a signal peptide. However, no good match to the von Heijne cleavage consensus sequence was detected in this region. Prosite analysis of the Bat1p sequence identified numerous potential glycosylation, myristylation, and phosphorylation sites, the relevance of which require experimental validation.
BAT1 Deletion Mutants Are Deficient in BA Transport Activity-ATP-dependent transport of BA was abolished in vacuoles prepared from DOY11 yeast (Fig. 7A) in which the BAT1 coding region has been deleted from the genome. It is conceivable that loss of Bat1p function affected ATP-dependent transport activities in the vacuolar fraction in a nonspecific manner; however, ATP-dependent transport of the glutathione conjugate DNPSG was unaffected in vacuoles derived from the DOY11 knockout strain (Fig. 7B), suggesting that this was not the case. Transformation of DOY11 with the centromeric plasmid pAA43, which carries an 8-kb genomic fragment containing the BAT1 coding region and 1.5 kb of upstream sequence, restored ATP-dependent BA transport to the level observed in the RSY12 control (Fig. 7A). Because overexpression of membrane proteins may result in mislocalization to the vacuole (52), a low copy number centromeric plasmid was used for the complementation experiments and samples from the progenitor RSY12 strain were included in all experiments.
A sec1-1 bat1⌬3 double mutant (DOY12) was generated by disruption of the BAT1 gene in the sec1-1 NY3 strain. Secretory vesicles prepared from DOY12 were essentially lacking ATP-dependent transport of taurocholate relative to the NY3, indicating that Bat1p is also responsible for the BA transport activity detected in secretory vesicles.
Polyclonal antibodies were raised against a histidine-tagged FIG. 3. Alignment of the Ycf1p NBD sequence with aa sequences deduced from DNA fragments generated by PCR amplification of yeast DNA. Fragment A is identical to EMBL accession number E238715 on chromosome XII, fragment B is identical to the YOR1 gene (65), and fragment C is identical to sequence accession number Z73153 on chromosome XII.
Bat1p partial peptide expressed in E. coli. The antibodies recognized a protein that was enriched 10 -15-fold in the vacuolar fraction relative to total yeast membranes, and was absent in the DOY10 and DOY11 bat1 deletion mutants (Fig. 8). It was difficult to estimate the M r of Bat1p by SDS-PAGE as the electrophoretic mobility of Bat1p varied relative to common protein M r standards depending on the acrylamide percentage of the gel, e.g. Bat1p has an apparent M r of 115,000 or 190,000 in immunoblots generated from protein extracts separated on 6% or 15% SDS-PAGE, respectively. Bat1p that had been delipidated by extraction with ether or ethanol-acetone behaved in an identical fashion, suggesting that it is not an effect of associated phospholipids. Migration of the major glycoprotein of human erythrocyte membranes is known to vary with the acrylamide concentration of the SDS-PAGE system (53). Other membrane proteins also migrate aberrantly on SDS-PAGE, including yeast uracil permease (54); mammalian sulfonylurea receptor SUR (51), which shows significant similarity with Bat1p at the level of aa sequence and transmembrane topology; and the bacterial ABC-type proteins, AbcA (55) and HlyB (56). At this time it is difficult to rule out post-translational modification or processing of Bat1p that would affect the size or mobility of the protein. The deduced aa sequence suggests the presence of a signal peptide that could be cleaved after membrane insertion. Bat1p was purified by immunoprecipitation and gel electrophoresis, and protein sequencing was attempted on the purified protein. However, the amino terminus was blocked. Further experiments will address this question. DISCUSSION The Bat1p ABC-type protein mediates ATP-dependent transport of bile acids and is overrepresented in a subcellular fraction enriched in vacuolar markers. Uptake by this fraction represents true translocation of the BA across the membrane, as transport is susceptible to reduction of the osmotically sensitive intravesicular space and permeabilization of the vesicles results in loss of the radiolabeled substrate. The yeast transporter is similar to the bile acid transport activity that resides in the canalicular membrane. Transport is completely dependent on the presence of Mg 2ϩ -ATP and is inhibited by nonhydrolyzable nucleotide analogs. Like canalicular transport, the yeast transporter is inhibited by the phosphate analog vanadate, which also inhibits the activity of a number of ABCtype proteins. Substrate competition studies with CMV indicate that the mammalian transporter has higher affinity for negatively charged, conjugated BA (57); likewise, yeast membrane fractions recognize glycine-and taurine-conjugated BA more efficiently than unconjugated BA. Transport kinetics are also similar, CMV manifest a K m for taurocholate of 2-47 M and V max of 0.5-4 nmol/min/mg (4 -6), whereas the vacuolar fraction of yeast has a K m of 63 M and V max of 15 nmol/min/mg. While it is possible that the higher V max observed in the yeast vacuolar fraction represents a real difference between the two proteins, it may also reflect a higher yield of vacuoles in the right configuration for transport. Approximately 80% of CMV have a right-side-out orientation and do not manifest ATP-de- pendent BA transport (58).
The deduced aa sequence indicates that Bat1p is similar to the subgroup of ABC-type proteins that includes mammalian CFTR, EHBR, cMOAT or cMRP1, and SUR, yeast YCF1 and YOR1, and Leishmania tarentolae ltpgpA (59). These proteins are highly diverse with regard to the functions they perform. For example, MRP1, cMOAT/cMRP1, and YCF1 have been shown to transport glutathione conjugates, whereas CFTR and EHBR are cAMP-gated chloride channels, and SUR1 and SUR2 (60) appear to confer ATP sensitivity upon K ATP inwardly rectifying channels. However, these transporters can be distinguished from other ABC-type proteins on the basis of aa sequence identity and common structural features, which are shared by Bat1p. The members of this subfamily are asymmet-rical and exhibit little sequence similarity between the two halves of the protein, unlike MDR1 and related polypeptides, which are believed to have originated from duplication and fusion of an ancestral gene encoding a single NBD and polytopic transmembrane region (61). Many members of this subgroup exhibit an amino-terminal transmembrane region that is significantly longer than the carboxyl-terminal counterpart and spans the membrane more times. Although membrane topology of polytopic membrane proteins is hard to predict in the absence of experimental data, three different algorithms predicted that Bat1p contains 8 -10 transmembrane helices in its amino terminus versus only 4 in the carboxyl region. This matches the predicted topologies for SUR, EHBR, cMOAT/ cMRP1, and MRP1, the mammalian proteins most closely resembling Bat1p, and is similar to the topologies predicted for the yeast putative proteins, Yk83p and Yhd5p.
ATP-dependent transport of BA has also been observed in the vacuoles of plants (62) and the fission yeast, Schizosaccharomyces pombe (63). Conservation of this activity over such a broad evolutionary range suggests that BA transport may have a basic metabolic relevance. Yeast vacuoles are digestive compartments implicated in autophagy and degradation of cellular organelles (64). It is conceivable that Bat1p concentrates fun- 5. Analysis of the BAT1 transcript. A, Northern blot of total RNA prepared from the RSY12 progenitor strain and the DOY10 strain (right lane), which is deleted for 440 bp encoding the NBD found in PCR fragment C. Fragment C DNA was radioactively labeled and used to probe RNA blotted unto GeneScreen Plus membrane (DuPont). Autoradiography detects a band corresponding to 6000 nucleotides in length in the RSY12 lane and no signal in the lane containing DOY10 RNA. B, PCR products generated from DNase-treated RNA (R) and genomic DNA (G) prepared from the RSY12 progenitor strain. PCR products spanning the BAT1 transcript (boxed region) suggest that no introns, or only very small ones, are excised from the primary transcript. The numbers above the photograph correspond to the PCR products drawn below the diagram of the BAT1 transcript. The arrows on either side of the lines represent the approximate position of the oligonucleotide primers used for PCR. The dotted line corresponding to fragment 5 indicates that the downstream primer resides outside the BAT1 transcript and thus is expected to generate a PCR product with genomic DNA but not with the RNA template. The numbers left of the photograph represent the size of the M r markers in kb (Life Technologies, Inc. 1-kb ladder).

FIG. 7. ATP-dependent transport by vacuolar subfractions derived from yeast that express or lack Bat1p. Time course of [ 3 H]taurocholate (A) or [ 3 H]DNPSG (B) transport in vacuolar fractions
was measured in the presence (Ⅺ, Ç, E) or absence (f, å, q) of 5 mM ATP. Vacuoles were prepared from the progenitor strain RSY12 carrying the pAA17 empty vector (Ⅺ, f), the bat1⌬ disruption strain DOY11 with pAA17 (Ç, å), and DOY11 harboring pAA43 (E, q), a pAA17 construct containing the Bat1p open reading frame and promoter region. Addition (after 7 min) of Triton X-100 to 0.02% resulted in release of most of the incorporated radioactivity from the vesicles, indicating that internalization of radiolabel into the vesicles had taken place rather than mere binding to membranes. Taurocholate and DNPSG were present at final concentrations of 5 M and 2 M, respectively. Values are mean of two experiments done in triplicate Ϯ standard error (except where smaller than symbol). gal equivalents of BA in the vacuole where these detergents aid in breakdown of organelle membranes and lipids. Alternatively, transport of soluble sterol derivatives across the vacuolar membrane may fulfill a function in sterol or steroid metabolism. BAT1 is closely related to two genes implicated in detoxification of xenobiotics: YCF1 in Cd tolerance and YOR1/ YRS1, which is involved in oligomycin and multidrug resistance. It is therefore possible that BAT1 also plays a role in toxin resistance. Identification of BAT1 and the availability of a knockout strain should aid in elucidating the function of BA transport in plants and fungi.
ATP-dependent transport of BA in the liver directly affects cholesterol secretion and water flow into the canaliculus, and impaired BA secretion occurs in numerous inherited and acquired hepatobiliary diseases. Despite the importance of ATPdependent BA transport, the protein responsible for this activity and the cognate gene have not been isolated, mainly due to low abundance in the liver. Hepatic BA transport activity resembles Bat1p-mediated BA transport with regard to energy requirements, substrate specificity, and response to inhibitors. Yeast Ycf1p, which exhibits similarities in substrate affinity, energy requirements, and inhibitor response with mammalian MRP1 and cMOAT (24), 2 also displays significant aa sequence identity with these proteins, suggesting that isolation of the yeast gene encoding an ATP-dependent BA transporter should facilitate identification of the mammalian counterpart. FIG. 8. Immunoblot analysis of proteins present in total membranes and in the vacuolar subfraction. Total membranes and vacuoles were prepared from the DOY10 and DOY11 bat1 ⌬ deletion mutants and the RSY12 progenitor strain. Yeast were carrying either the pAA17 centromeric plasmid containing no insert or pAA43, which contains an 8-kb genomic DNA fragment harboring the BAT1 gene. 20 g of protein were loaded in each lane and separated by electrophoresis on a 12% SDS-polyacrylamide gel. Proteins were transferred to nitrocellulose and incubated with affinity-purified IgG recognizing Bat1p. The numbers on the right represent the size in kDa of protein molecular size markers (Pharmacia HMW-SDS calibration Kit).