The human hepatocellular carcinoma cell line, HepG2, and the monkey kidney cell line, COS7, were acquired from ATCC (Manassas, VA). HepG2 cells were cultured in MEM with non-essential amino acids (ATCC) containing 10% FBS and 1% penicillin-streptomycin, at 37°C with 5% CO₂. COS7 cells were cultured in DMEM containing 10% FBS and 1% penicillin-streptomycin, at 37°C with 5% CO₂.

After HepG2 cells reached ~70% confluence, they were washed and cultured in fresh medium containing 10% charcoal-stripped serum in the presence or absence of 50 μM chenodeoxycholate (CDCA) (Sigma, St Louis, MO), or 2 μM 6-ethyl CDCA (Dr. Roberto Pellicciari, Universita Di Perugia, Italy). Twenty four to forty-eight hours after addition of CDCA, RNA and protein were isolated or cells were fixed for immunofluorescence as described below. To inhibit glycosylation, tunicamycin (Sigma) was added at concentrations indicated in the figure legends 6 hrs after the addition of CDCA treatment in HepG2 cells or 4 hrs after the initiation of transfection in COS7 cells.

Lysates from COS7 cells transfected for 48 hrs with OSTα-FLAG and OSTβ-MYC were digested with peptide:N-glycosidase F (PNGase F) and endoglycosidase H (EndoH) according to the manufacturer's instructions (New England Biolabs) and subjected to SDS-PAGE as described below.

HepG2 cell cDNA was used as a template. We generated a 1.03 kb cDNA fragment encoding the full-length of human OSTα and a 0.4 kb cDNA fragment encoding the full-length of human OSTβ by PCR. The primers for amplification of human OSTα and OSTβ were based on the published human sequences (GenBank accession number AK172837 and AY194242). The forward primer OSTα 5'-GCTTGGTACCATGGAGCCGGGCAGGACCCAGATAA-3' and the reverse primer OSTα 5'-CCGCTCGAGTTACTTGTCATCGTCGTCCTTGTAATCCCCGGCTTTGAGGTTCAAGTCCAGGTC-3' were used. The forward primer OSTβ 5'-GCTGGATCCACCATGGAGCACAGTGAGGGGGCTCC-3' and the reverse primer OSTβ 5'-GCACTCGAGGCTCTC AGTTTCTGGTACATCCGG-3' were used. The amplified cDNA fragment encoding the full-length of OSTα was then subcloned into the Kpn I and Xho I sites of the mammalian expression vector pcDNA3.1 (Invitrogen) and the cDNA fragment encoding the full-length of OSTβ was subcloned into the BamH I and Xho I sites of pcDNA3.1 Myc/His vector (Invitrogen). PcDNA3.1-OSTα-FLAG and pcDNA3.1-OSTβ-Myc/His were sequenced using Yale Keck DNA sequencing facility. The coding sequences were identical to the published sequences with the GenBank accession numbers for OSTα [AY194243] and for OSTβ [BC103842].

COS7 cells were transfected with FuGene 6 (Roche) using 1 μg OSTα-FLAG or OSTβ-Myc DNA/9 cm² surface area, according to manufacturer's instructions. pcDNA vector control (1 μg DNA) was used when only one subunit was transfected. Cells were harvested 24-48 hr after transfection, as described for Western blotting or immunofluorescence.

Cells were extracted with Trizol (Invitrogen, Carlsbad, CA) and RNA was isolated according to manufacturer's instructions. Quantitative RT-PCR was carried out as described previously 4 using Applied Biosystems 7500 DNA sequence detector system with TaqMan universal master mix (Applied Biosystems, Foster City, CA). Specific primer pairs for hOSTα and hOSTβ were the same as previously described 4.

Cells were washed with PBS and then extracted directly in RIPA buffer (25 mM Tris, pH 7.2, 150 mM NaCl, 10 mM EDTA, 1% Triton X-100, 1% deoxycholate, 0.1% SDS) or in 1% Triton X-100, 50 mM Tris HCl, pH 7.4,150 mM NaCl, 1 mM EDTA for immunoprecipitation. Lysates were centrifuged at 10,000 × g for 20 min and the supernatant was collected for analysis using SDS-PAGE. Immunoprecipatation was performed using anti-FLAG affinity gel (M2, Sigma, St Louis, MO) or anti-Myc polyclonal antibody (abcam, Cambridge, MA) and Protein A/G beads (Santa Cruz). Lysates were precleared and negative controls were performed with non-specific anti-mouse IgG. In the case of immunoprecipitation of endogenous protein from HepG2 cells, a Native IgG kit from Pierce was used with polyclonal antibodies raised against OSTα (hOSTα-327) and OSTβ (hOSTβ-1) provided by Ned Ballatori (Rochester, NY).

Pulse-chase experiments were carried out in COS7 cells 24 hr after transfection. After incubation for 15 min with Cys/Met minus media, cells were pulsed for 15 min with media containing 135 μCi ³⁵S Trans-label (MP Biomedicals, Solon, OH). Cells were either extracted immediately in 1% Triton X-100, 50 mM Tris Hcl, pH 7.4,150 mM NaCl, 1 mM EDTA or chased for 2 hr in complete media. Immunoprecipitation was carried out as described above.

Cells were fixed with cold methanol for 10 min or with 4% paraformaldehyde for 15 min. Quenching of non-specific fluorescence in formaldehyde fixed cells was done with 50 mM NH₄Cl for 20 min prior to blocking 20 minutes in blocking buffer (PBS, 1% BSA, 0.05% Triton X-100). In the case of OSTβ, non-permeablized conditions using no detergent was found to give better surface labeling. Primary antibody was diluted in blocking buffer and incubated on the cells for 2 hours at room temperature. After washing in PBS, secondary antibody (Alexa 594 or 488 anti-IgG (Invitrogen)) was incubated for 1 hour at room temperature. Fluorescence was visualized with a Zeiss LSM510 (Carl Zeiss Inc, Thornwood, NY) confocal microscope and images processed with Photoshop (Adobe, Mountainview, CA).

HepG2 cells were cultured in 35 mm dishes as described above. At ~70% confluency 50 μM CDCA or vehicle was added and the culture continued for 48 hrs. ³H-Taurocholate (1 μM) or ³H-estrone 3-sulfate (15 nM) were made up in transport buffer (116 mM NaCl, 5.3 mM KCl, 1.1 mM KH2PO4, 0.8 MgSO4.7H2O, 1.8 mM CaCl, 11 mM glucose, 10 mM HEPES) and warmed to 37°C. For each time point, triplicate dishes were washed 3 times with warm transport buffer alone and then incubated for the given time with 1 ml transport buffer containing ³H-substrate. Uptake of substrates was stopped by rapid addition and aspiration of 1 ml of cold transport buffer three times. Cells were lysed with 1 ml 0.5% Triton X100. Cell lysates (600 μl) were combined with OptiFluor scintillation fluid (5 ml) and counted in a PerkinElmer WinSpectral LSC (PerkinElmer, Waltham, MA). Protein content of the lysates was determined with the BCA reagent (Pierce Biotechnology, Rockford, IL) and used to normalize the counts.

Previous studies indicate that human hepatocytes express OSTα-OSTβ on their basolateral membranes 4. In untreated human hepatoma, HepG2, cells, there was little expression of OSTα-OSTβ mRNA or protein. However, treatment with the FXR agonist, CDCA, up-regulated mRNA levels of both subunits rapidly within one hour and peaked between 12 and 24 hours. mRNA levels increased ~12 fold for OSTα and ~20 fold for OSTβ (Figure 1A). Protein levels for both subunits were not detectable until at least 12 hrs after treatment with 50 μM CDCA or 2 μM 6-ethyl CDCA (Figure 1B). Both subunits were visualized on the plasma membrane at that time, with 70–90% of cells expressing the transporter after 48 hrs (Figure 1C). A time course using immunofluorescence detected no OSTα prior to 12 hrs, although OSTβ could be visualized in a perinuclear localization at earlier time points (data not shown). Finally, uptake of both ³H-taurocholate and ³H-estrone 3-sulfate was increased 3–4 fold in CDCA treated cells, indicating that OSTα-OSTβ was functional in these cells when expressed on the plasma membrane (Figure 1D). Uptake of taurocholate can also be mediated by other basolateral transporters such as the sodium taurocholate co-transporting protein (NTCP) and the family of organic anion transporter proteins (OATPs). Therefore, we also measured mRNA levels for OATP1A2, OATP1B1, OATP1B3, and NTCP by Q-PCR. While most were unchanged after CDCA treatment, mRNA for OATP1B3 was up-regulated (data not shown), consistent with previous reports 9. Thus, we cannot rule out that some of the uptake was due to OATP1B3.

Examination of the amino acid sequence of the alpha subunit has indicated that, unlike in the mouse, rat and skate sequences, the human OSTα subunit does not have the traditional N-glycosylation consensus sequence of Asn-X-Ser/Thr 1. Therefore, it was of interest to determine if the human OSTα protein was glycosylated. Tunicamycin is an antibiotic that has been shown to inhibit N-glycosylation of proteins by blocking the addition of N-acetylglucosamine to dolichol phosphate, the first step in the formation of core oligosaccharide 10. Treatment of HepG2 cells with tunicamycin (1 μg/ml), 5 hours after the addition of 50 μM CDCA and for a total of 15 hrs, reduced the molecular weight of OSTα from ~36 kD to ~30 kD (Figure 2A). Treatment with the glycosidases, Endo H and PNGase F, confirmed that these proteins were the mature and non-glycosylated forms of OSTα (data not shown). There was no shift in the molecular weight of the OSTβ subunit after tunicamycin treatment. This confirms previous data demonstrating that the only potential N-glycosylation consensus sequence in mouse Ostβ is in a presumably transmembrane domain and, thus, is not utilized 6. These data indicate that endogenously expressed human OSTα is a glycoprotein. However, despite the lack of N-glycosylation in these treated cells, immunofluorescence demonstrated that the OSTα subunit was still expressed on the plasma membrane at similar levels to untreated cells (Figure 2B). Furthermore, the tunicamycin treated HepG2 cells showed no difference from untreated cells in ³H-estrone 3-sulfate uptake, demonstrating that the non-glycosylated OSTα was still capable of associating with OSTβ to form a functional transporter (Figure 2C). In contrast, previous work has suggested that N-glycosylation of Oatps is essential to their functioning1112. The lack of difference in uptake after tunicamycin treatment suggests that little, if any, of the ³H-estrone 3-sulfate transport was through OATPs.

COS7 cells were transfected with OSTα-FLAG and OSTβ-Myc in order to evaluate the behavior of the individual subunits. When OSTα-FLAG alone was transfected, immunofluorescence revealed only an intracellular signal and never detected protein on the plasma membrane (Figure 3A). However, when both OSTα-FLAG and OSTβ-Myc were transfected together, both proteins were detected on the plasma membrane by immunofluorescence (Figure 3A). Transfection of OSTβ-Myc alone resulted in plasma membrane and intracellular localization (Figure 3A), demonstrating the ability of Ostβ to traffic to the plasma membrane independently of Ostα in an over-expressing system. This has also been reported in transfected MDCK and HEK293 cells 78.

Western blotting revealed that transfection of OSTα-FLAG alone resulted in two bands of approximately 31 and 35 kD (Figure 3B). Co-expression of OSTβ with OSTα resulted in an additional, higher molecular weight species of OSTα of ~40 kD (Figure 3B). These data are similar to that reported for mouse ileum and for HEK293 cells transfected with mouse Ostα and Ostβ 6, and suggest that this band represents a mature, complex glycosylated form of the OSTα subunit. Indeed, when N-glycosylation was inhibited with tunicamycin (0.5 μg/ml, 24 hrs) this 40 kD band disappeared (Figure 4A). In addition, the 35 kD band also disappeared, suggesting that this band represents the core glycosylated or precursor form found in the ER. The 31 kD band was not eliminated by tunicamycin treatment indicating that it is the non-glycosylated OSTα subunit. In contrast to HepG2 cells, tunicamycin treatment of COS7 cells appeared to reduce the expression of a higher molecular weight band of the OSTβ subunit (Figure 4A), although this was not a consistent finding. This subunit is not glycosylated, thus the change may reflect a non-specific effect of the tunicamycin treatment in this experiment.

The glycosylation status of OSTα was further clarified by treatment of cell lysates with the glycosidases, Endo H and PNGase F. These two enzymes can distinguish between N-glycans that only contain the core oligosaccharide that has been added in the ER (Endo H sensitive) and those that have trafficked through the Golgi and have had their carbohydrate chains modified (PNGase F sensitive, Endo H resistant) 13. Figure 4B shows that the 40 kD band was sensitive to PNGase F, but not Endo H, treatment, indicating that the mature alpha subunit has exited the Golgi. The 35 kD band was sensitive to both Endo H and PNGase F and, thus, represents a glycoprotein that has not trafficked through the Golgi. The 31 kD band remains after both glycosidase treatments, confirming that it represents the non-glycosylated OSTα subunit. These data demonstrate that human OSTβ is required for human OSTα to be processed from the high mannose type N-linked glycan in the ER to complex oligosaccharides in the cis/medial Golgi region. Immunofluorescence of tunicamycin treated COS7 cells transfected with OSTαFLAG and OSTβ-Myc showed that the lack of glycosylation did not prevent OSTα trafficking to the plasma membrane, confirming data seen in HepG2 cells (Figure 4C). Thus, the interaction of the beta subunit with the alpha subunit in the ER and the subsequent trafficking through the Golgi does not require that OSTα be glycosylated.

In order to gain further insight into the interaction of the two subunits, immunoprecipitation was conducted using antibodies to the tags associated with both the alpha and beta subunits. In COS7 cells, when OSTαFLAG was immunoprecipitated with anti-FLAG agarose beads, the precipitate also contained OSTβ (Figure 5A). Although all the OSTα was efficiently precipitated, only a fraction of the OSTβ was found in the precipitate. When OSTβ was immunoprecipitated with anti-Myc antibody, only the two lower molecular weight forms of OSTα were found in the precipitate (Figure 5A). Despite repeated attempts using more protein and over-exposure of the blots (data not shown) the mature, complex glycosylated form of OSTα was never seen. Metabolic labeling of transfected COS7 cells demonstrated that the 40 kD form of OSTα appeared only after co-transfection of the alpha and beta subunits and after a time lag of > 15 min (Figure 5B; earlier time points not shown). Furthermore, anti-FLAG precipitated all forms of OSTα as well as OSTβ, but anti-Myc co-precipitated only OSTβ and the immature forms of OSTα (31 and 35 kD bands), confirming the previous data. This suggested that the physical association of the two subunits may be necessary for the transporter to be processed and trafficked through the intracellular compartments, but possibly not necessary once OSTα reaches the plasma membrane.

To test whether this could be an artifact of transfection of exogenous DNA, we also conducted immunoprecipitation of endogenous protein in HepG2 cells. In this case it should be noted that after CDCA treatment we see expression only of the mature form of OSTα (Figure 2A), suggesting that the immature form(s) move very rapidly through the Golgi and are of too low abundance to detect. Immunoprecipitation using the polyclonal anti-OSTα and OSTβ antibodies does not demonstrate co-precipitation of the two subunits (Figure 6A). This is consistent with the lack of association between the mature, plasma membrane subunits.