β-Subunit of the Ostα-Ostβ Organic Solute Transporter Is Required Not Only for Heterodimerization and Trafficking but Also for Function*

Background: Deletion of Ostα-Ostβ, an intestinal transporter that participates in the enterohepatic circulation of bile acids, lowers cholesterol and triglyceride levels. Results: Residues of Ostβ required for interaction with Ostα and transport activity were identified. Conclusion: Distinct regions of Ostβ are essential for Ostα-Ostβ trafficking and transport activity. Significance: Findings suggest strategies for inhibiting the holotransporter and reducing lipid levels. The organic solute transporter, Ost/Slc51, is composed of two distinct proteins that must heterodimerize to generate transport activity, but the role of the individual subunits in mediating transport activity is unknown. The present study identified regions in Ostβ required for heterodimerization with Ostα, trafficking of the Ostα-Ostβ complex to the plasma membrane, and bile acid transport activity in HEK293 cells. Bimolecular fluorescence complementation analysis revealed that a 25-amino acid peptide containing the Ostβ transmembrane (TM) domain heterodimerized with Ostα, although the resulting complex failed to reach the plasma membrane and generate cellular [3H]taurocholate transport activity. Deletion of the single TM domain of Ostβ abolished interaction with Ostα, demonstrating that the TM segment is necessary and sufficient for formation of a heteromeric complex with Ostα. Mutation of the highly conserved tryptophan-asparagine sequence within the TM domain of Ostβ to alanines did not prevent cell surface trafficking, but abolished transport activity. Removal of the N-terminal 27 amino acids of Ostβ resulted in a transporter complex that reached the plasma membrane and exhibited transport activity at 30 °C. Complete deletion of the C terminus of Ostβ abolished [3H]taurocholate transport activity, but reinsertion of two native arginines immediately C-terminal to the TM domain rescued this defect. These positively charged residues establish the correct Nexo/Ccyt topology of the peptide, in accordance with the positive inside rule. Together, the results demonstrate that Ostβ is required for both proper trafficking of Ostα and formation of the functional transport unit, and identify specific residues of Ostβ critical for these processes.

The organic solute transporter, Ost/Slc51, is composed of two distinct proteins that must heterodimerize to generate transport activity, but the role of the individual subunits in mediating transport activity is unknown. The present study identified regions in Ost␤ required for heterodimerization with Ost␣, trafficking of the Ost␣-Ost␤ complex to the plasma membrane, and bile acid transport activity in HEK293 cells. Bimolecular fluorescence complementation analysis revealed that a 25amino acid peptide containing the Ost␤ transmembrane (TM) domain heterodimerized with Ost␣, although the resulting complex failed to reach the plasma membrane and generate cellular [ 3 H]taurocholate transport activity. Deletion of the single TM domain of Ost␤ abolished interaction with Ost␣, demonstrating that the TM segment is necessary and sufficient for formation of a heteromeric complex with Ost␣. Mutation of the highly conserved tryptophan-asparagine sequence within the TM domain of Ost␤ to alanines did not prevent cell surface trafficking, but abolished transport activity. Removal of the N-terminal 27 amino acids of Ost␤ resulted in a transporter complex that reached the plasma membrane and exhibited transport activity at 30°C. Complete deletion of the C terminus of Ost␤ abolished [ 3 H]taurocholate transport activity, but reinsertion of two native arginines immediately C-terminal to the TM domain rescued this defect. These positively charged residues establish the correct N exo /C cyt topology of the peptide, in accordance with the positive inside rule. Together, the results demonstrate that Ost␤ is required for both proper trafficking of Ost␣ and forma-tion of the functional transport unit, and identify specific residues of Ost␤ critical for these processes.
The heteromeric organic solute transporter (Ost) 2 ␣-␤ (Ost␣-Ost␤/Slc51) is the key basolateral plasma membrane bile acid and steroid conjugate carrier in many human tissues, including the small intestine, liver, and other steroidogenic organs (1)(2)(3)(4). Recently, Ost␣ Ϫ/Ϫ mice have been generated (5)(6)(7), and these animals exhibit a major defect in intestinal bile acid absorption, confirming that Ost␣-Ost␤ is the main intestinal bile acid efflux transporter in the enterohepatic circulation. Bile acids, which are major products of cholesterol catabolism, are required for hepatic bile secretion and for the emulsification and intestinal absorption of fat and fat-soluble vitamins (8). To better understand the biological roles of the two subunits and to gather biomolecular information potentially useful in developing strategies for modulating transport activity, studies were undertaken to investigate the role of mouse Ost␤ in the holotransporter.
The overall structure of Ost␣-Ost␤ resembles that of certain G protein-coupled receptors, including the G protein-coupled receptor-receptor activity-modifying protein (GPCR-RAMP) complexes. Ost␤, a 128-amino acid protein in humans and mice, is predicted to contain a single transmembrane (TM) domain, and to be oriented in the plasma membrane with its N terminus in the extracellular space and its C terminus in the cytosol (N exo /C cyt with no signal peptide, classified as a type Ia integral membrane protein) (5). Although its complete biological function(s) in the heteromeric complex has not yet been established, Ost␤ appears to serve as a chaperone facilitating the cell surface delivery of the 7-TM domain Ost␣ subunit, presumed to be the main catalytic subunit of the transporter (5,9,10).
Co-expression of Ost␣ and Ost␤ is required to elicit transport activity (4,11,12), and Li et al. (5) have demonstrated that Ost␣ and Ost␤ are present in cells as heterodimers and/or heteromultimers. Heterodimerization of Ost␣ and Ost␤ increases the stability of the individual proteins and is required for delivery of the heteromeric complex to the plasma membrane (5). Formation of the transporter complex is also coupled to posttranslational modifications of Ost␣; i.e. stable interaction between the subunits is required for progression through the biosynthetic-secretory pathway and to generate a glycosylated form of Ost␣ (5,9,10). Thus, the glycosylation status of Ost␣ can be utilized as an index of interaction between the two subunits.
Interestingly, although skate Ost␤ and human OST␤ display roughly 25% amino acid identity, previous studies have shown that human OST␣ can generate comparable taurocholate transport activity when complemented with either human OST␤ or skate Ost␤, suggesting that only a few conserved amino acids or the secondary structure of the OST␤/Ost␤ proteins is important for interaction (4). The goal of this work was to identify regions of Ost␤ that are critical for dimerization with Ost␣, trafficking, and transport activity. The three main amino acid segments of Ost␤ were analyzed: (i) C-terminal residues 54 -107; (ii) N-terminal residues 1-27; and (iii) the TM domain region. In addition, point mutants of evolutionarily conserved residues in Ost␤ were constructed and characterized.
Transient Expression in HEK293 Cells and Bile Acid Transport Activity Assay-HEK293 cells in 6-well plates were transfected with 900 ng of Ost␣ and 100 ng of each Ost␤ species DNA using LipoD293 (SignaGen). Twelve h later, the medium was replaced, and cells were incubated at 30°C or 37°C for an additional 36 h. At 48 h after transfection, medium was replaced with Hanks' balanced salt solution (HBSS) (Invitrogen) containing 25 M [ 3 H]taurocholic acid, and cells were incubated at 37°C for 30 min. After incubation, cells were washed twice with ice-cold HBSS containing 1 mM unlabeled taurocholate and 0.2% (w/v) BSA and then once with ice-cold HBSS alone. Monolayers were lysed overnight in 1 N NaOH, and an aliquot of the transport media and the lysate was analyzed by liquid scintillation to assess transport activity. Lysate protein concentrations were determined with the DC protein assay (Bio-Rad), and nonspecific cell-associated radioactivity was accounted for by subtracting radioactivity detected at time zero.
Construction and Visualization of Fluorescent Fusion Proteins-For bimolecular fluorescence complementation (BiFC) analysis (13)(14)(15), residues 1-155 of YFP (YN) were fused to the C terminus of Ost␣, and residues 156 -238 of YFP (YC) were fused to the C terminus of each Ost␤ construct. Ost␣-YN in pBiFCN and Ost␤-YC in pBiFCC were described previously (5). Truncations of Ost␤ were PCR-amplified and ligated into pBiFCC. Constructs in pBiFCN/pBiFCC were then subcloned into pcDNA3.3. BiFC-tagged Ost␤ point mutants were produced via site-directed mutagenesis. Cerulean (from Dr. David Piston, Vanderbilt University) was PCR-amplified and ligated 3Ј to Ost␣. The two halves of YFP in pBiFCN and pBiFCC were combined 3Ј to Ost␤, generating Ost␤-YFP. Two rounds of mutagenesis were conducted to change the YFP coding sequence to Topaz (16). The Topaz coding sequence was then PCR-amplified and ligated 3Ј to Ost␤ mutants. Stop codons in the linkers were changed to alanine codons via site-directed mutagenesis.
N-terminal Epitope Tagging of Constructs and Cell Surface ELISA-The N termini of Ost␣ and Ost␤ were tagged with V5 and triple HA (3ϫHA) epitopes, respectively, via site-directed mutagenesis. HEK293 cells were transfected with 900 ng of V5-Ost␣ and 100 ng of each 3ϫHA-Ost␤ species DNA in 24-well plates. At 48 h after transfection, cell surface ELISA was conducted as described previously (17). Briefly, plates were incubated with monoclonal anti-V5 (Invitrogen) and anti-HA (Covance) at 1:5,000 followed by anti-mouse IgG-HRP (Bio-Rad) at 1:5,000. Antibody binding was detected with 3,3Ј5,5Јtetramethylbenzidine, and the reaction was terminated with 10% sulfuric acid. Absorbance was read at 450 nm, and values from duplicate points were averaged. Absorbance in mocktransfected cells was subtracted, and the resulting values were normalized to those obtained in V5-Ost␣-3ϫHA-Ost␤-transfected cells, which were present on each plate.
Whole Cell Lysate Preparation and Immunoblotting-At 48 h after transfection, HEK293 cells were lysed with PBS containing 1% (v/v) protease inhibitor mixture (Sigma), 5 mM EDTA, 2 mM PMSF, and 0.2% Triton X-100. Lysates were centrifuged at 20,000 ϫ g for 15 min at 4°C, and the supernatant was taken as the whole cell lysate and immediately frozen at Ϫ80°C. Lysate protein concentrations were determined with the DC protein assay.
For immunoblotting of V5-Ost␣, 15 g of each sample was subjected to Laemmli-SDS-PAGE on 10% Tris-HCl ready gels (Bio-Rad) followed by wet transfer onto PVDF using Dunn carbonate buffer (10 mM NaHCO 3 , 3 mM Na 2 CO 3 , pH 9.9, 20% methanol) for 90 min at 100 volts. Blots were blocked in milk at room temperature for 3 h and then incubated at 4°C overnight with anti-V5 at 1:10,000. Anti-mouse IgG-HRP (KPL) at 1:5,000 was then applied for 1 h at room temperature. For immunoblotting of 3ϫHA-Ost␤ mutants, 15 g of each sample was subjected to Tricine-SDS-PAGE on 10 -20% or 16.5% Tris-Tricine precast gels (Bio-Rad) followed by wet transfer using Towbin buffer (Bio-Rad). Blots were blocked overnight at 4°C and then incubated with anti-HA HRP conjugate (Roche Applied Science) at 25 milliunits/ml for 1 h at room temperature. Antibody binding was detected with LumiGLO Peroxidase Chemiluminescent Substrate (KPL). For loading controls, blots were probed with anti-␣-tubulin or anti-␤-actin (Sigma).
N exo /C cyt Topology Assessment-Glycosylation tags (N*N*) were added to the N termini of constructs via mutagenic PCR. The glycosylation tag is a segment of the Saccharomyces cerevisiae ␣-factor receptor Ste2p with the amino acid sequence STI-NYTSIYGNGSTITSSS, which contains two Asn residues that are known to be N-glycosylated when luminal in the ER (18). HEK293 cells were transiently transfected with 900 ng of V5-Ost␣ and 100 ng of each N*N*-3ϫHA-Ost␤ species DNA, and whole cell lysates were collected 48 h later. Protein samples were processed with the Glycoprofile II, Enzymatic In-Solution N-Deglycosylation kit (Sigma) and subjected to SDS-PAGE for band shift analysis.
Statistical Analyses-All bar graphs show the mean Ϯ S.E. from three or four independent experiments, each performed in duplicate or triplicate. Data were evaluated with Prism 4 using one-way ANOVA followed by Bonferroni's multiple comparison test, or the two-tailed unpaired Student's t test. Differences were considered statistically significant at p Ͻ 0.05. Sequences were aligned using the MUSCLE program (multiple sequence comparison by log-expectation).

Domains of Ost␤ Required for Function-To identify regions
of Ost␤ that are important for function, five truncation mutants were constructed (Fig. 1A). These Ost␤ truncations were transiently transfected into HEK293 cells with Ost␣, and [ 3 H]taurocholate transport activity was measured (Fig. 1B). To correlate functional and visualization results with the BiFC-tagged proteins, cells were incubated at either 30 or 37°C before transport activity was measured (see below).
As expected, cells co-expressing wild-type Ost␣ and Ost␤ demonstrated robust taurocholate uptake (Fig. 1B). Deletion of the 21 C-terminal residues of Ost␤, generating Ost␤-1-107, did not affect transport activity; however, complete truncation of the C terminus, yielding Ost␤-1-53, markedly decreased taurocholate uptake (Fig. 1B). Deletion of the N terminus of Ost␤, producing Ost␤-28 -128, resulted in a heteromeric transporter complex that elicited transport activity, but only if cells were incubated at 30°C. Transport activity of the wildtype heteromer was also slightly higher if cells were incubated at 30°C (Fig. 1B). Culturing at low temperatures is known to increase protein production (19 -23) and rescue function of some mutant transporters (24,25). No transport activity was detected when Ost␣ was expressed with mutants of Ost␤ that contained only the TM domain region (Ost␤-29 -53) or lacked the entire TM domain (Ost␤-⌬34 -53) (Fig. 1B).
To determine whether the loss of transport activity observed with these Ost␤ mutants was due to failure of the resulting transporter complex to traffic to the plasma membrane, Ost␣ and the truncated Ost␤s were tagged at the N termini with V5 and triple HA (3ϫHA) epitopes, respectively, and the amounts of these proteins at the cell surface were determined (Fig. 1C). The tagged proteins gave taurocholate transport activity comparable with, although somewhat lower than, their untagged counterparts (compare Fig. 1B and supplemental Fig. S1), as observed previously (5). As expected, neither 3ϫHA-Ost␤ nor V5-Ost␣ was found at the cell surface when expressed alone. When expressed with V5-Ost␣, the Ost␤ constructs that displayed transport activity (3ϫHA-Ost␤, 3ϫHA-Ost␤-1-107, and 3ϫHA-Ost␤-28 -128) were readily detected at the plasma membrane along with the V5-Ost␣ subunit, although 3ϫHA-Ost␤-28 -128 was present at half the level of the full-length protein even though V5-Ost␣ was detected in normal amounts (Fig. 1C). One possible explanation for this result is that the 3ϫHA tag on 3ϫHA-Ost␤-28 -128 is closer to the TM domain of Ost␤ and may be partially occluded in the resulting V5-Ost␣-3ϫHA-Ost␤-28 -128 heteromer. The Ost␤ constructs that did not give transport activity (3ϫHA-Ost␤-1-53, 3ϫHA-Ost␤-29 -53, and 3ϫHA-Ost␤-⌬34 -53) were not localized at the plasma membrane, nor was co-expressed V5-Ost␣ (Fig. 1C), explaining the absence of functional activity.
BiFC Analysis of Ost␣-Ost␤ Interaction-To observe the interaction between Ost␣ and Ost␤ and its subcellular localization simultaneously in live cells, BiFC analysis was conducted using confocal microscopy. The N-terminal half of YFP (YN) was fused to the C terminus of Ost␣ and the C-terminal half of YFP (YC) to the C terminus of each Ost␤ species. If Ost␣ and Ost␤ interact directly, the two halves of the fluorophore form YFP. The BiFC constructs exhibited taurocholate transport activity roughly comparable with that of the untagged constructs (compare Figs. 1B and 2A). One interesting exception was Ost␤-1-53-YC, which showed significant transport activity upon co-expression with Ost␣-YN ( Fig. 2A), whereas their untagged counterparts did not (Fig. 1B).
Ost␤-YC, Ost␤-1-107-YC, and Ost␤-28 -128-YC showed clear YFP fluorescence at the plasma membrane, indicating that they heterodimerized with Ost␣-YN and trafficked to the cell surface (Fig. 2B). These findings are consistent with surface expression measurements described above. In contrast with the behavior of the epitope-tagged proteins, Ost␤-1-53-YC and Ost␣-YN heterodimerized at the plasma membrane (Fig. 2B), consistent with the transport activity generated by this complex (Fig. 2A). Ost␤-29 -53-YC (TM domain region of Ost␤ only) and Ost␣-YN formed a BiFC complex that was retained in the ER (Fig. 2B), in accord with the lack of detectable complex on the cell surface (Fig. 1C). The Ost␤ complex lacking the TM region (Ost␤-⌬34 -53-YC) did not interact with Ost␣-YN (Fig.  2B), in agreement with surface expression and immunoblotting data (Fig. 1, B and D). These findings suggest that the TM domain of Ost␤ is sufficient for interaction with Ost␣ but not for trafficking.
To ensure that these findings did not result from the affinity of the YFP fragments for one another, the fluorophores Cerulean and Topaz, which exhibit no affinity for each other, were fused to the C termini of Ost␣ and the Ost␤ truncations, respectively. The Topaz fusions of Ost␤, Ost␤-1-107, Ost␤-1-53, and Ost␤-28 -128 all co-localized with Ost␣-Cerulean at the plasma membrane (supplemental Fig. S2), in agreement with the BiFC results (Fig. 2B). Ost␤-29 -53-Topaz (TM domain region only) and Ost␤-⌬34 -53-Topaz (no TM domain) were detected in cells (supplemental Fig. S2), whereas the 3ϫHA-tagged versions were not (Fig. 1D). This is not surprising because GFP and its variants tend to increase the stability of proteins (26 -29). Ost␣-Cerulean and Ost␤-29 -53-Topaz co-localized in the ER (supplemental Fig. S2), in agreement with BiFC results. Ost␤-⌬34 -53-Topaz, which lacks a TM region, exhibited a diffuse intracellular localization, whereas Ost␣-Cerulean remained in the ER (supplemental Fig. S2).
Ost␤ Participates in Transport Mechanism-Additional studies examined the evolutionarily conserved amino acids in Ost␤ to determine their role in transport activity. Sequence alignment of Ost␤s from seven different species (1) revealed that the TM domain and several amino acids in close proximity to this membrane helix are most conserved (Fig. 3A). Four residues are completely conserved: a Glu-Asp sequence (Glu 29 -Asp 30 ) near the N terminus of the TM helix, a Trp (Trp 34 ) at the predicted start of the TM domain, and an Arg (Arg 61 ) located eight amino acids from the C terminus of the TM domain. Also of note is the highly conserved Asn  JUNE 15, 2012 • VOLUME 287 • NUMBER 25 (Asn 35 ), which is present in all species except the zebrafish (Fig. 3A). Site-directed mutagenesis was conducted to produce Ost␤-E29A D30A, Ost␤-W34A/N35A, Ost␤-W34A, Ost␤-N35A, and Ost␤-R61G, and each construct was expressed with Ost␣. All constructs exhibited transport activity except for the Ost␤ mutant in which the highly conserved Trp-Asn sequence at the beginning of the TM segment was mutated to Ala-Ala (Fig. 3B). When this Trp-Asn sequence was mutated to Phe-Gln (Ost␤-W34F/N35Q), function was intact (Fig. 3B). Despite their inability to form a functional transporter, V5-Ost␣ and 3ϫHA-Ost␤-W34A/ N35A were detected at the plasma membrane at ϳ50 and 75% of wild-type levels, respectively (Fig. 3C). Immunoblotting showed that the expression of the Ost␤ point mutants was equivalent although the presence of fully glycosylated V5-Ost␣ was decreased when it was paired with 3ϫHA-Ost␤-W34A/N35A, suggesting that this mutant may have difficulty interacting with Ost␣ (Fig. 3D).

Functional and Dimerization-related Regions of Ost␤
To examine interactions of these Ost␤ mutants with Ost␣ directly, BiFC-tagged versions were analyzed by fluorescence microscopy (Fig. 4), as were Cerulean-and Topaz-tagged constructs (supplemental Fig. S3). Both BiFC and co-localization approaches showed that all of the Ost␤ point mutants localized at the plasma membrane and interacted with Ost␣, including the functionally inactive Ost␤-W34A/N35A.

Importance of C Terminus of Ost␤ for Cell Surface
Localization-As shown above, untagged Ost␤-1-53, which lacks the entire region C-terminal to the TM domain, failed to traffic to the cell surface or generate transport activity ( Fig. 1  and supplemental Fig. S1). Conversely, Ost␤-1-53 tagged at the C terminus with YC or Topaz did traffic to the plasma membrane ( Fig. 2B and supplemental Fig. S3) and exhibit taurocholate transport activity ( Fig. 2A and data not shown), suggesting that the fused fluorophores can replace the natural amino acids 54 -107. To characterize the function(s) of the C terminus more precisely, successive truncations were made, generating Ost␤-1-103, 1-93, 1-83, 1-73, and 1-63 (Fig. 5A). When these constructs were co-expressed with Ost␣, all generated transport activity (Fig. 5B). Although Ost␣-Ost␤-1-53 was inactive, addition of the two natural Arg residues at positions 54 and 55 restored transport activity; however, substitution with two Ala residues, giving Ost␤-1-55 R54A/R55A, resulted in a loss of transport activity (Fig. 5B).
Cell surface ELISA was utilized to detect the presence of 3ϫHA-tagged versions of these constructs at the plasma membrane. Interestingly, as the C terminus of Ost␤ was progressively shortened, surface levels of both the mutant 3ϫHA-Ost␤ and V5-Ost␣ declined (Fig. 5C). Although almost no 3ϫHA-Ost␤-1-53 was detected at the plasma membrane, adding back the two Arg residues (3ϫHA-Ost␤-1-55) restored surface expression and transport activity, whereas adding two Ala residues did not (Fig. 5C). In cells expressing Ost␤-1-55, [ 3 H]taurocholate uptake was equivalent to that obtained with wild-type Ost␤ (Fig. 5B), even though surface expression of both V5-Ost␣ and 3ϫHA-Ost␤-1-55 was lower. This result is consistent with the correlation between the levels of wild-type V5-Ost␣ and 3ϫHA-Ost␤ on the plasma membrane and transport activity. Transport activity reached a maximum before surface expression of V5-Ost␣ or 3ϫHA-Ost␤ when cells were transfected with a constant amount of cDNA encoding V5-Ost␣ and increasing amounts of cDNA encoding 3ϫHA-Ost␤ (supplemental Fig. S4).
Immunoblotting revealed that the 3ϫHA-Ost␤ C-terminal mutant proteins were present at roughly comparable levels; however, the expression of V5-Ost␣ and its fully glycosylated form decreased as the C terminus was shortened or replaced with two Ala residues (3ϫHA-Ost␤-1-55 R54A/ R55A) (Fig. 5D). These results indicate that the two residues just C-terminal to the TM region of Ost␤-1-55, Arg 54 and Arg 55 , were sufficient for proper membrane localization and activity.
Positively Charged Residues in C Terminus of Ost␤ Establish Its N exo /C cyt Topology-Positively charged residues flanking the TM domain of integral membrane proteins are major determinants of topology (30 -33), with the positively charged side typically oriented toward the cytoplasm (positive inside rule). To examine whether Arg 54 and Arg 55 establish a N exo /C cyt orientation of Ost␤-1-55, a tag containing a pair of N-glycosylation sites (denoted N*N*) was fused to the N termini of 3ϫHAtagged versions of Ost␤, Ost␤-1-55, Ost␤-1-55 R54A/R55A, and Ost␤-1-53 (Fig. 6A). The glycosylation tags on these constructs can only be modified if the protein is inserted in the membrane of the ER during translation with a N exo /C cyt orien- tation. Constructs were co-expressed with V5-Ost␣, and the extent of glycosylation was determined by analyzing samples with and without enzymatic deglycosylation by PNGase F. As expected based on previous Ost␤ topology studies (5), the N-glycosylation tag was highly modified on N*N*-3ϫHA-Ost␤, which exhibited a sizeable molecular weight shift upon PNGase F treatment (Fig. 6B). N*N*-3ϫHA-Ost␤-1-55 was also glycosylated, but neither N*N*-3ϫHA-Ost␤-1-55 R54A/R55A nor N*N*-3ϫHA-Ost␤-1-53 appeared to be (Fig. 6B). The intensity of the glycosylated V5-Ost␣ band was lower when V5-Ost␣ was co-expressed with N*N*-3ϫHA-Ost␤-1-55 than with wild-type N*N*-3ϫHA-Ost␤ and negligible when it was co-expressed with the corresponding versions of Ost␤-1-55 R54A/R55A and Ost␤-1-53, although all of the Ost␤ constructs expressed well. These findings indicate that Arg 54 and Arg 55 are important for Ost␤ membrane orientation.

DISCUSSION
Ost␣-Ost␤ is a unique bile acid and steroid conjugate transporter composed of two distinct subunits that interact to generate the functional transporter. The current observations indicate that Ost␤ not only modulates Ost␣ glycosylation, membrane trafficking, and turnover rate (5,9) but also participates in the transport mechanism. The overall structure of the Ost transporter, in which a larger multi-TM domain subunit is accompanied by a single TM accessory protein, occurs frequently in membrane proteins. Examples include certain G protein-coupled receptors, like Frizzled-LRP5/6, CRLR-RAMP1, and MC2R-MRAP (34,35); the nicotinic acetylcholine receptor, ␣7-nAchR-RIC-3 (36, 37); the K ϩ channel, K v 1.4-K v ␤3 (38,39); the Na ϩ /K ϩ -ATPase, ATP1A1-ATP1B1 (40,41); and system L of the heterodimeric amino acid transporters, LAT1/2-4F2hc (42,43). The single TM accessory proteins fulfill several functions, including (i) assisting with the folding of

Functional and Dimerization-related Regions of Ost␤
the larger subunit(s), (ii) escorting the larger subunit(s) to the plasma membrane, (iii) forming part of the active complex and assisting in its retention at the membrane, (iv) forming part of the ligand/substrate binding pocket, and (v) participating in desensitization and internalization.
The studies described here were designed to dissect the functions of different regions of Ost␤ using a mutational strategy. The ability of Ost␤ mutants to interact with Ost␣ was evaluated by BiFC, analysis of post-translational modification of Ost␣, and determination of the stability of the individual subunits. The effect of mutations in Ost␤ on trafficking of the Ost␣-Ost␤ complex to the plasma membrane was monitored via localization of fluorescently labeled Ost␣ and Ost␤ using confocal microscopy and quantification of Ost␣ and Ost␤ on the cell surface using ELISA. Finally, the impact of Ost␤ mutations on the transport function of Ost␣-Ost␤ complexes was assessed by [ 3 H]taurocholate uptake. This multifaceted approach allowed specific roles to be assigned to distinct regions of the essential Ost␤ subunit of the Ost heteromer.

Role of N Terminus and TM Domain of Ost␤-Mouse
Ost␤ has 34 amino acids in the extracellular N-terminal domain, including a highly conserved Asp-His-Ser sequence immediately following the initiating Met and a conserved region preceding the transmembrane helix. Ost␤-28 -128, which lacks nearly all of the N terminus, was competent to deliver the complex to the plasma membrane and to generate transport activity, but only if cells were cultured at 30°C.
Co-immunoprecipitation and BiFC experiments have established that Ost␣ and Ost␤ form a tight complex when they are initially synthesized in the ER and remain associated in the functional transporter at the cell surface. Seward et al. (4) proposed that only a few conserved amino acids are required for interaction between Ost␣ and Ost␤. Skate Ost␤ has just 25% amino acid identity with the human ortholog, yet these two proteins generate similar transport activity when co-expressed with human OST␣. Although the overall Ost␤ amino acid identity is low among species (1), the Ost␤ TM region from evolutionarily divergent species exhibits Ͼ40% amino acid identity, suggesting that the TM domain is a key element for heterodimerization and transport activity.
The present findings support this hypothesis by demonstrating that the TM domain of Ost␤ is required for formation of a heteromer with Ost␣. When the TM domain was deleted, yielding Ost␤-⌬34 -53, no transport activity was detected upon coexpression with Ost␣, and no interaction between the subunits was found by BiFC. Ost␤-⌬34 -53-Topaz was distributed evenly throughout the cytoplasm. Ost␤-⌬34 -53 could in theory have interacted with Ost␣ in BiFC experiments even though the two proteins were synthesized in different compartments of the cell because the YFP fragment was fused onto the cytoplasmic side of Ost␣. The failure of Ost␣ and Ost␤-⌬34 -53 to interact may explain why both proteins were largely undetectable on immunoblots; native Ost␣ and Ost␤ are both unstable when expressed alone. Stronger evidence supporting the importance of the TM domain is provided by Ost␤-W34A/ N35A, a TM domain mutant that did not generate any transport activity although it did localize at the cell surface and interact with Ost␣ in BiFC experiments. The functional deficit of Ost␤-W34A/N35A and Ost␣ cannot be explained by the 25-50% decrease in the amount of the subunits at the plasma membrane, because cell surface levels of Ost␣-Ost␤-1-55 were much lower, and yet this complex exhibited normal transport activity. Taken together, the data provide compelling evidence that the TM domain of Ost␤, in particular the Trp-Asn sequence at the extracellular-membrane interface, is directly involved in the transport mechanism. When present in TM helices, Trp and Asn are often found near the membrane-water interface where they are more energetically favorable and confer stability to a TM helix (44,45). Asn can form hydrogen bonds with the peptide backbone (i.e. N-capping) to stabilize an ␣-helix (46) and drive TM-TM interactions via hydrogen bonding (47)(48)(49)(50)(51), which may occur between Ost␣ and Ost␤.
Because the Ost␣-Ost␤-29 -53 complex gave a BiFC signal, the TM domain of Ost␤ was sufficient for interaction with Ost␣, but it was not sufficient for trafficking to the cell surface. In fact, no single site within Ost␤ was found to be solely responsible for membrane trafficking of the heteromeric complex. Therefore, it is likely that other sites of interaction between the subunits exist and facilitate membrane trafficking of the transporter. In support of this, Sun et al. (10) demonstrated that the 50 N-terminal residues of human OST␣ are required for interaction with human OST␤ and in turn trafficking of the transporter complex to the plasma membrane. The stoichiometry of the Ost transporter is unknown. Ost␣ forms homodimers (5), and it is possible that interaction between two Ost␣ subunits is dependent on Ost␤ and essential for trafficking.

Role of Positively Charged Residues in C Terminus of Ost␤-
Ost␤ is a type Ia integral membrane protein with a N exo /C cyt topology (5). A stretch of 7-25 amino acids, which are uncharged and largely hydrophobic, constitutes a signal sequence (often a TM domain) that targets an integral membrane protein to the ER for co-translational insertion (52). Several factors influence the orientation of a signal sequence in the ER membrane and in turn establish the topology of the integral membrane protein. Most important is the positive inside rule, which states that the positively charged residues flanking the signal sequence determine its membrane orientation such that the more positive flank faces the cytoplasm. When the 15 N-terminal and C-terminal residues flanking a signal sequence are analyzed, the cytoplasmic end usually contains ϳ2-3-fold higher frequency of ArgϩLys residues than the end facing the ER lumen (30 -33). Ost␤ appears to abide by this rule, because the 15 residues C-terminal to the TM domain contain six positively charged amino acids whereas the extracellular 15 N-terminal residues have only one (5).
This positive charge distribution explains the cell surface localization of Ost␤ mutants with C-terminal truncations. As the C terminus was shortened, the proteins had less positive charge on the C-terminal side of the TM; there were 6, 4, 2 and 0 LysϩArg residues in wild type, 1-63, 1-55, and 1-53 or 1-55 R54A/R55A Ost␤, respectively. Based on the absence of modification on their N-glycosylation tags, Ost␤-1-53 and Ost␤-1-55 R54A/R55A were inserted in the ER membrane upside down (N cyt /C exo ) and did not support trafficking or function of the transporter unit. This probably occurred because the incorrectly oriented Ost␤s did not interact with Ost␣, which was therefore degraded. When Ost␤-1-53 was fused to YC or Topaz, the truncated Ost␤ regained the ability to interact with Ost␣, traffic to the plasma membrane and generate transport activity. The gain of functional activity is most likely explained by the increase in positive charge on the C-terminal side introduced by the tags (2 ArgϩLys) and the tendency of folded domains such as YC/Topaz to localize to the cytoplasmic side.
The novel findings presented here are summarized in Fig.  7, which illustrates the regions of Ost␤ and their proposed function(s). Mutations in several highly conserved amino acids did not, by themselves, disrupt Ost␤ activity. The N terminus of Ost␤ may be necessary for correct folding and/or assembly of the transporter, but if this requirement is bypassed by low temperature incubation, a transporter missing all but 5 amino acids N-terminal to the TM domain yields robust transport activity. Residues on the C-terminal side of the TM domain (yellow) are necessary for correct membrane orientation of Ost␤, which is essential for Ost␣-Ost␤ interaction. If this requirement is bypassed, however, then a transporter lacking the entire C-terminal domain can generate functional activity. Thus, all of the results obtained here point to the highly conserved TM domain region of Ost␤ as the major site of interaction with Ost␣. The TM helix also appears to be part of the functional component of the holotransporter, and the evolutionarily conserved Trp-Asn (W34/N35) sequence at the extracellular N terminus of the helix is absolutely required for transport activity. It is likely that future studies on the biochemistry of the transporter complex will identify additional contributions of the N-and C-terminal domains of Ost␤ and more specific roles of the TM helix. Because of the potential benefits of drugs targeting the Ost transporter in vivo, detailed understanding of the role of the subunits should be of great value. The TM region appears to be sufficient and necessary for interaction with Ost␣. The Trp-Asn residues shown in green are essential for transport activity, and the positively charged residues shown in yellow are important for establishing the N exo /C cyt topology of Ost␤. Asterisks denote fully conserved residues.