The NTCP (Na+–taurocholate co-transporting protein)/SLC10A [solute carrier family 10 (Na+/bile acid co-transporter family)] 1 is tightly controlled to ensure hepatic bile salt uptake while preventing toxic bile salt accumulation. Many transport proteins require oligomerization for their activity and regulation. This is not yet established for bile salt transporters. The present study was conducted to elucidate the oligomeric state of NTCP. Chemical cross-linking revealed the presence of NTCP dimers in rat liver membranes and U2OS cells stably expressing NTCP. Co-immunoprecipitation of tagged NTCP proteins revealed a physical interaction between subunits. The C-terminus of NTCP was not required for subunit interaction, but was essential for exit from the ER (endoplasmic reticulum). NTCP without its C-terminus (NTCP Y307X) retained full-length wtNTCP (wild-type NTCP) in the ER in a dominant fashion, suggesting that dimerization occurs early in the secretory pathway. FRET (fluorescence resonance energy transfer) using fluorescently labelled subunits further demonstrated that dimerization persists at the plasma membrane. NTCP belongs to the SLC10A protein family which consists of seven members. NTCP co-localized in U2OS cells with SLC10A4 and SLC10A6, but not with SLC10A3, SLC10A5 or SLC10A7. SLC10A4 and SLC10A6 co-immunoprecipitated with NTCP, demonstrating that heteromeric complexes can be formed between SLC10A family members in vitro. Expression of SLC10A4 and NTCP Y307X resulted in a reduction of NTCP abundance at the plasma membrane and NTCP-mediated taurocholate uptake, whereas expression of SLC10A6 or NTCP E257N, an inactive mutant, did not affect NTCP function. In conclusion, NTCP adopts a dimeric structure in which individual subunits are functional. Bile salt uptake is influenced by heterodimerization when this impairs NTCP plasma membrane trafficking.
During a meal, bile acids are released from the gall bladder into the small intestine, aiding in the solubilization and absorption of fats and fat-soluble nutrients. At the distal end of the ileum ~95% of the bile acids are reabsorbed and returned to the liver via the portal venous circulation. The liver efficiently clears bile acids from the portal blood. In humans, the majority of bile acids are absorbed in a Na+-dependent manner [1–4] mediated by the NTCP (Na+–taurocholate co-transporting protein)/SLC10A [solute carrier family 10 (Na+/bile acid co-transporter family)] 1 which operates as a symporter, cotransporting two Na+ ions per bile acid molecule. Besides bile acids, NTCP is capable of transporting other substrates, e.g. estrone sulfate, thyroid hormones and drugs covalently bound to taurocholate . NTCP is a glycoprotein of 349 amino acids, with an extracellular N-terminus containing two N-linked glycosylation sites and an intracellular directed C-terminus flanking seven or nine transmembrane α-helices. The C-terminus is essential for targeting to the basolateral plasma membrane of hepatocytes as shown by rat NTCP lacking the C-terminus mainly residing in the ER (endoplasmic reticulum) . NTCP is the founding member of the bile acid transporter family which contains seven members: SLC10A1–7 . NTCP and ASBT (apical Na+-dependent bile acid transporter)/SLC10A2 are the best characterized family members, playing pivotal roles in hepatic and intestinal bile acid uptake respectively. Non-synonymous ethnicity-dependent genetic variations in human NTCP have been identified, resulting in a reduced or near complete loss of taurocholate uptake and plasma membrane expression [7,8]. NTCP protein expression is decreased in PFIC (progressive familial intrahepatic cholestasis)  and rat extrahepatic cholestasis , showing an important role for NTCP in bile salt transport. SLC10A6 [SOAT (Na+-dependent organic anion transporter)] is known to transport steroid sulfates. The physiological role of the other members remains largely elusive; SLC10A3, SLC10A4, SLC10A5 and SLC10A7 are orphan transporters [3,6,11–13]. A higher oligomeric organization of SLC10A family members is suggested. Older studies using photoaffinity labelling correlated a 93 kDa integral membrane protein with intestinal Na+-dependent bile acid uptake . This size is in line with an ASBT dimer. SLC10A5 and SLC10A7 proteins are often also detected in immunoblots at a position corresponding with twice the size of the monomer [6,11]. Furthermore, it was suggested that membrane transporters with less than 12 membrane-spanning helices require oligomerization to create a functional protein . The aim of the present study was to assess the oligomeric structure of human NTCP, to investigate possible heteromerization with other SLC10A proteins or NTCP variants, and the functional consequences thereof regarding bile salt transport in vitro. In the present study we demonstrate that NTCP adopts a dimeric structure.
Full-length human NTCP (GenBank® accession number NM-003049) was obtained by PCR (5′-GCGCGGATCCATGGAGGCCCACAACGCGTCT-3′ and 5′-GCGCCTCGAGCTAGGCTGTGCAAGG-GGAGCA-3′) using human liver cDNA as the template and was cloned into the pcDNA3 vector. Adaptor DNA encoding FLAG or HA (haemagglutinin) epitope tags was inserted between the KpnI and BamHI sites thereby generating FLAG–NTCP and HA–NTCP respectively. The ACP (acyl carrier protein) coding sequence was obtained by PCR (5′-GAAGCTGGGTACCAGCTG-3′ and 5′-GCGCGGATCCCGCCTGGTGGCCGTTGATGTA-3′) using a ACP–NKI (neurokinin receptor) construct (from Dr Nils Johnsson, Center for Molecular Biology of Inflammation, Münster, Germany) as a template and inserted at the 5′-end of the NTCP ORF (open reading frame) between the KpnI and BamHI sites. NTCP constructs with E257N (NTCP E257N) or Y307stop (NTCP Y307X) mutations were prepared by mutagenesis according to the manufacturer's protocol (Strategene). All constructs were verified by sequence analysis. Constructs encoding V5-tagged human SLC10A3–7 were provided by Dr Barbara Döring and Stephanie Schmidt (Institute of Pharmacology and Toxicology, Justus Liebig University of Giessen, Germany) and are related to GenBank® accession numbers NM_019848 (SLC10A3), NM_152679 (SLC10A4), AY825924 (SLC10A5), EF437223 (SLC10A6) and DQ122860 (SLC10A7).
Cell culture and transfection
U2OS cells were cultured in DMEM (Dulbecco's modified Eagle's medium) containing GlutaMax™-I, supplemented with 10% FBS (fetal bovine serum; Invitrogen) and 100 units/ml penicillin/streptomycin (Invitrogen). Cells were transfected with PEI (polyethylenimine; Polyscience). To establish U2OS cells stably expressing HA–NTCP, U2OS clones were selected by culturing in the presence of 800 μg/ml G418 (Gibco). Expression of HA–NTCP was determined by taurocholate uptake, immunoblot analysis and immunofluorescence microscopy. Over 98% of the cells of selected clones expressed HA–NTCP. U2OS cells stably expressing NTCP were used for cross-linking and co-immunoprecipitation experiments. For visualization of plasma membrane localization, U2OS cells were transfected with ACP–NKI.
[3H]Taurocholate uptake assay
U2OS cells transiently transfected with NTCP were seeded in 24-well plates, harvested 48 h later and washed with uptake buffer [5 mM KCl, 1.1 mM KH2PO4, 1 mM MgCl2, 1.8 mM CaCl2, 10 mM D-glucose, 10 mM Hepes and 136 mM NaCl (Na+-containing buffer) or 136 mM NMDG (N-methyl-D-glucamine; Na+-free buffer) and adjusted to pH 7.4 using Tris/HCl]. Cells were incubated at 37°C with uptake buffer containing 20 μM taurocholate supplemented with 0.2 μM [3H]taurocholate for 2 min. Cells were washed twice with ice-cold PBS and lysed in 0.05% SDS. Accumulation of radiolabelled substrates was determined by scintillation counting.
Cross-linking procedure of stably transfected U2OS cells and rat liver membranes
Rat liver membranes were isolated as described previously . Liver membranes were incubated for 60 min at 4°C with 1 mM EGS [ethylene glycol bis(succinimidyl succinate)] in DMSO/PBS, DSP [dithiobis(succinimidyl) propionate] in DMSO/PBS, or DTBP (dimethyl dithiobispropionimidate) (Pierce) in PBS, whereas stably transfected U2OS cells were treated with 0.25, 1 or 5 mM EGS, DSP or DTBP. Cells were lysed in lysis buffer [20 mM Tris/HCl (pH 7.4), 5 mM EDTA, 135 mM NaCl, 1% (v/v) Nonidet P40 and 10% (w/v) sucrose], scraped on ice, centrifuged for 10 min at 16000 g followed by NTCP detection on immunoblot using an anti-HA antibody for U2OS cells and rabbit anti-(rat Ntcp) for liver membranes (provided by Professor Bruno Stieger, University Hospital Zurich, Zurich, Switzerland ). Samples were treated with PNGase F (peptide N-glycosidase F) according to the manufacturer's instructions (New England Biolabs) and analysed by immunoblotting.
U2OS cells were transiently transfected with HA–NTCP and FLAG–NTCP. U2OS cells stably expressing HA–NTCP were transiently transfected with FLAG–NTCP. Cells were harvested 48–72 h thereafter, washed with PBS and lysed as described above. Supernatants were incubated with monoclonal anti-HA or anti-FLAG antibodies (Sigma) immobilized on Protein A–agarose beads (Sigma) for 2 h at 4°C. Immunoprecipitated proteins were analysed by immunoblot analysis using rabbit anti-FLAG or HRP (horseradish peroxidase)-conjugated mouse anti-HA antibodies (Sigma).
Acceptor bleaching for FRET (fluorescence resonance energy transfer) determination
U2OS cells transiently expressing ACP-tagged NTCP were labelled with CoA (coenzyme A)–Alexa Fluor™ 488 (donor) and CoA–DY547 (acceptor) at a 1:1 ratio according to the manufacturer's instructions (New England Biolabs). The cells were fixed using 4% (w/v) paraformaldehyde/PBS. FRET was ascertained using acceptor bleaching by ~90% bleaching of the DY547 fluorophore with 561 nm wavelength laser excitation using a Zeiss LSM 710 confocal microscope. FRET efficiency was calculated as described previously .
U2OS cells were transiently transfected with HA- or FLAG-tagged NTCP-encoding plasmids. After 2 days, the cells were washed with PBS, fixed at room temperature (20°C) with 4% (w/v) paraformaldehyde/PBS, and permeabilized with 0.2% Triton X-100/PBS. Subsequently, the cells were washed with 50 mM NH4Cl/PBS and blocked with 2% (w/v) BSA (Sigma) in PBS. HA–NTCP or FLAG–NTCP proteins were visualized using rabbit anti-FLAG antibody (F7425, Sigma), mouse anti-HA antibody (H9658, Sigma), Alexa Fluor™ 488-conjugated goat anti-mouse IgG (Invitrogen) and Alexa Fluor™ 568-conjugated goat anti-rabbit IgG (Invitrogen). For SLC10A3–7–V5, fluorescein isothiocyanate-conjugated mouse anti-V5 antibody was used (Invitrogen). As a plasma membrane marker we used ACP–NKI which was visualized by ACP labelling. The cells were incubated with labelling mix containing CoA–Alexa Fluor™ 488 substrate and ACP synthase. Since the ACP tag is located extracellularly, and both CoA–Alexa Fluor™ 488 and ACP synthase are cell impermeable, this labelling specifically visualized the plasma membrane. ER localization was determined using the ER marker rabbit anti-calreticulin antibody (Alexis Biochemicals). Fluorescent imaging was performed using a Zeiss LSM 710 Meta confocal laser-scanning microscope equipped with a Plan APOCHROMAT 63× 1.4 NA (numerical; aperture) objective. ImageJ (http://rsbweb.nih.gov/ij/) was used for data analysis.
NTCP-transfected U2OS cells were washed with PBS supplemented with 0.5 mM CaCl2 and 1.0 mM MgCl2 48 h post-transfection. Cell-surface proteins were biotinylated using sulfo-NHS-SS-biotin [sulfo succinimidyl-2-(biotinamido)ethyl-1,3-dithiopropionate]. Biotinylated proteins were precipitated and analysed by immunoblot analysis as described previously  using HRP-conjugated mouse anti-HA antibody (Sigma). Mouse anti-(transferrin receptor) antibody (Invitrogen) was used as a loading control. Hsp70 (heat-shock protein 70) was used for the detection of intracellular protein in protein lysates and biotinylated fractions.
All results are expressed as means±S.D., except for the FRET results which are means±S.E.M. Statistical significance was determined by unpaired Student's t test (SPSS; http://www-01.ibm.com/software/analytics/spss/). P values ≤0.05 were considered significant.
Epitope tags do not interfere with NTCP function
We generated constructs encoding human NTCP with various epitope tags (HA, FLAG or ACP) at the N-terminus (Figure 1A). Taurocholate uptake was strongly induced upon NTCP expression and was abolished upon replacing Na+ with NMDG (results not shown). Addition of the epitope tag did not affect NTCP taurocholate uptake activity (Figure 1B) or subcellular localization (Figures 3D and 6).
Epitope-tagged human NTCP is functional
Cross-linking subunits show NTCP dimeric architecture
Chemical cross-linking analysis was performed to determine whether NTCP is able to adopt an oligomeric structure. To this end, we treated U2OS cells stably expressing HA–NTCP with DSP, DTBP and EGS, chemical cross-linking agents with spacer arms ranging from 11.9 to 16.1 Å (1 Å=0.1 nm). Cross-linker treatment resulted in the appearance of an additional NTCP-specific band at ~100 kDa, approximately twice the size of the monomer (Figure 2A). After EGS or DSP treatment, NTCP bands >150 kDa were also observed. To test whether dimerization also occurred endogenously, rat liver membranes were examined. NTCP signal was detected at ~50 kDa. Treatment with EGS or DSP yielded an additional band at >250 kDa and a strong reduction of the monomeric NTCP signal, suggesting that NTCP is present in a complex also containing (un)known proteins. Membranes treated with DTBP showed an additional band at ~100 kDa, twice the size of the monomer (Figure 2B). To test whether the ~100 kDa band corresponds with a NTCP dimer, samples were treated with PNGase F, which cleaves N-glycan chains from glycoproteins (Supplementary Figure S1A at http://www.BiochemJ.org/bj/441/bj4411007add.htm). This procedure resulted in a near complete shift of NTCP signal in non-cross-linked samples to ~30 kDa (Supplementary Figure S1A, right-hand panel). Cross-linker treatment resulted in the appearance of a ~60 kDa band, exactly twice the size of the deglycosylated monomer. Similar results were obtained after PNGase F treatment of DSP- and EGS-treated samples (results not shown). Cross-linked and PNGase F-treated HA–NTCP was found at ~30 and ~60 kDa. Upon co-expression of ACP–NTCP, an additional band was observed migrating at a higher molecular mass, in line with the extra molecular mass of the ACP tag (Supplementary Figure S1B). ACP–NTCP which migrated at ~45 kDa, was also detected, suggesting that not all ACP–NTCP that was co-precipitated with HA–NTCP was efficiently cross-linked (Supplementary Figure S1B).
Cross-linking demonstrates the dimeric structure of NTCP
NTCP subunits physically interact
Since the cross-linking experiments suggest that NTCP can form dimers, we determined the physical interaction between NTCP subunits by co-immunoprecipitation. The HA–NTCP and FLAG–NTCP constructs were co-expressed in U2OS cells. HA–NTCP co-precipitated with FLAG–NTCP using the anti-FLAG antibody (Figure 3A). Omission of FLAG–NTCP prevented precipitation of HA–NTCP, demonstrating the specificity of the antibodies. Similarly, FLAG–NTCP co-precipitated with HA–NTCP using the anti-HA antibody (Figure 3B). In protein lysates, complex glycosylated (~50 kDa) NTCP was detected, whereas in co-immunoprecipitation experiments the high mannose glycosylated (~37 kDa) and unglycosylated NTCP (~30 kDa) were mostly detected, possibly due to less-accessible epitopes in the glycosylated proteins. To investigate whether complex-glycosylated NTCP is also able to self-interact, U2OS cells stably expressing HA–NTCP were used, which show more efficient NTCP glycosylation. Complex glycosylated NTCP is co-precipitated (Figure 3C), demonstrating that fully glycosylated NTCP is able to dimerize.
Interaction of NTCP subunits occurs at the plasma membrane
NTCP dimers are present at the plasma membrane
ACP–NTCP is properly targeted to the plasma membrane in U2OS cells, where it is functional, as taurocholate uptake is unaltered (Figure 1B). ACP-tagged proteins can be labelled with fluorescent CoA derivates using ACP synthase forming a covalent bond . Using cell-impermeable fluorophores and ACP synthase we selectively labelled the extracellular portion of ACP–NTCP. ACP–NTCP-expressing cells were covalently labelled simultaneously with CoA–Alexa Fluor™ 488 (donor) and CoA–Alexa Fluor™ (acceptor) at a 1:1 ratio. Each ACP peptide reacts with just one fluorescent molecule. To investigate whether NTCP dimerizes at the plasma membrane, we determined FRET between labelled NTCP proteins . Energy transfer from donor to acceptor was determined by acceptor bleaching, resulting in a ~10% increase in donor intensity (0.17–23%, n=49 cells; Figure 3D), demonstrating significant FRET efficiency. In unbleached regions of the cell, no difference in donor signal was observed (−3.7–4.2%).
NTCP dimerization takes place early in the secretory pathway
FLAG–NTCP lacking its C-terminus (FLAG–Y307X) co-precipitated with full-length HA–NTCP, indicating that these proteins interact (Figure 4A). The reverse experiment confirmed this interaction as FLAG–NTCP co-precipitated with HA–Y307X (Figure 4B). Also, HA–Y307X and FLAG–Y307X subunits co-precipitated. NTCP Y307X is mainly retained in the ER (Figure 4C, upper middle panel), whereas wtNTCP (wild-type NTCP) predominantly resides at the plasma membrane (Figure 4C, top panel). Interestingly, co-expression of wtNTCP and Y307X resulted in a dominant retention of the wild-type protein in the ER (Figure 4C). Dominant-negative effects of NTCP Y307X on the targeting of wtNTCP to the plasma membrane were confirmed functionally as heteromerization between wtNTCP and Y307X diminished NTCP-mediated taurocholate uptake (Figure 4D). Take together, these results suggested that dimerization occurs already early in the secretory pathway in a C-terminus-independent manner.
NTCP Y307X retains wtNTCP in the ER
Individual NTCP subunits operate as bile salt transporting units
To investigate whether oligomeric NTCP structures represent the functional unit for transport or that subunits within a complex are individually active, we determined taurocholate uptake upon mixing of active and inactive NTCP proteins. If multiple subunits combined form a single functional unit one would expect that only wild-type homodimers would contribute to taurocholate uptake. Replacing half of active wtNTCP with inactive subunits would strongly inhibit transport (Figure 5A). However, when 50% of the active wtNTCP was replaced by NTCP E257N, a mutant unable to transport bile salts, only a ~50% reduction in taurocholate uptake was observed (Figure 5B). No difference in taurocholate uptake was observed between U2OS cells only expressing wtNTCP and cells co-expressing wtNTCP and NTCP E275N, when the amount of wtNTCP was kept equal (Figure 5C). The E257N mutation did not affect plasma membrane targeting (Figure 5D). Furthermore, subunit interaction between FLAG–NTCP and HA–E257N was shown by co-immunoprecipitation (Figure 5E). These results indicate that individual subunits of NTCP dimers are functional.
Individual NTCP subunits operate as bile salt transporting units
NTCP co-localizes and associates with SLC10A4 and SLC10A6 in U2OS cells
NTCP has considerable sequence similarity with its family members [6,11–13,22]. In vivo, NTCP is exclusively expressed in hepatocytes. SLC10A3, SLC10A5, SLC10A6 and SLC10A7 are also detected in the liver. In particular, SLC10A5 is highly expressed in hepatocytes. We determined the subcellular co-localization of NTCP with SLC10A3–7. SLC10A3, SLC10A5 and SLC10A7 showed virtually no co-localization with NTCP, which predominantly localized to the plasma membrane, whereas they showed a Golgi-like distribution upon expression in U2OS cells (Figure 6). Therefore it seems unlikely that NTCP forms heteromeric complexes with these family members. However, SLC10A6 completely co-localized with NTCP at the plasma membrane. Furthermore, SLC10A4 localized at the plasma membrane and in intracellular vesicles and showed considerable co-localization with NTCP (Figure 6). Subsequently, co-immunoprecipitations were performed to investigate possible heteromeric interactions between NTCP and SLC10A4 or SLC10A6. SLC10A4–V5 (predicted molecular mass ~47 kDa) showed multiple bands when immunoblotted (Figure 7A). Importantly, SLC10A4 co-precipitated with FLAG–NTCP. No SLC10A4 was precipitated in the absence of FLAG–NTCP, indicating the specificity of the procedure (Figure 7A). The interaction between SLC10A4 and NTCP was confirmed in a reversed experiment (Figure 7B). SLC10A6–FLAG co-precipitated with HA–NTCP in a similar manner as FLAG–NTCP (Figure 7C). Omission of HA–NTCP prevented the precipitation of SLC10A6–FLAG. FLAG–NTCP was also co-precipitated with SLC10A6–V5. In addition, SLC10A6–FLAG co-precipitated with SLC10A6–V5, demonstrating that dimerization of SLC10A family members is not confined to NTCP (Figure 7D). These data suggest that wtNTCP co-localizes and physically interacts with SLC10A4 and SLC10A6 in vitro.
NTCP co-localizes with family members SLC10A4 and SLC10A6
NTCP associates with SLC10A4 and SLC10A6
Heteromerization with SLC10A4 decreased NTCP-mediated bile salt uptake
Individually expressed NTCP and SLC10A6 are almost exclusively present at the plasma membrane, whereas SLC10A4 is also present intracellularly (results not shown). Therefore we hypothesized that SLC10A4 could reduce the amount of NTCP at the plasma membrane. Co-expression with SLC10A6 did not affect NTCP localization and both proteins reside at the plasma membrane (Figure 8A). However, co-expression with SLC10A4 results in a partial shift in NTCP localization from the plasma membrane to intracellular compartments (Figure 8A). This was quantified by cell-surface biotinylation (Figure 8B). In cell lysates, NTCP-specific signal was detected as multiple bands, reflecting differently glycosylated products. The NTCP signal in the biotinylated fraction was strongly enriched for the ~50 kDa fully glycosylated band. SLC10A4 co-expression resulted in 60% reduction of NTCP abundance at the plasma membrane and in cell lysates, whereas SLC10A6 did not alter NTCP expression. Levels of biotinylated transferrin receptor were similar in all conditions. No signal was detected when biotin was omitted from the procedure (no biotin lane), and no Hsp70 intracellular protein was detected in the biotinylated fraction (Supplementary Figure S3 at http://www.BiochemJ.org/bj/441/bj4411007add.htm). The effect of co-expression with SLC10A4 or SLC10A6 on NTCP-mediated taurocholate uptake was determined in transiently transfected U2OS cells. NTCP expression resulted in a significant increase in Na+-dependent taurocholate uptake. In contrast, neither SLC10A4 nor SLC10A6 expression resulted in taurocholate uptake above background levels. Co-expression of SLC10A4 with NTCP resulted in a significantly decreased taurocholate uptake, whereas SLC10A6 co-expression did not affect NTCP-mediated uptake in vitro (Figure 8C).
Functional consequences of heteromerization with SLC10A4
In the present study, we show that human NTCP forms a dimeric complex and individual NTCP subunits can operate as a bile salt transporting unit. Furthermore, we demonstrate that heteromerization affects bile salt uptake, but only when this influences the NTCP abundance at the plasma membrane. Proper functioning of NTCP is necessary for bile salt absorption from the portal circulation. The present study provides four lines of evidence for the dimeric architecture of NTCP, based on chemical cross-linking, co-immunoprecipitation, FRET and immunocytochemistry. Cross-linker treatment of living cells stably transfected with NTCP results in the appearance of an NTCP band at a size consistent with dimerization. Theoretically this band could also originate from an unknown ~50 kDa protein binding to NTCP. We therefore deglycosylated NTCP, which strongly reduced the monomeric size. As the band of the cross-linked product shifted correspondingly to a size of twice the monomer, it is likely to be a NTCP dimer. The NTCP signal also shifted to a higher molecular mass in cross-linked rat liver membranes, indicating that NTCP dimerization also occurs at physiological expression levels. The relative abundance of monomer, dimer and higher-molecular-mass bands depends considerably on type of cross-linking reagent, concentration and temperature (results not shown). However, the results suggest that the majority of NTCP is present in a dimeric complex.
Significant FRET is observed between fluorescently labelled, ACP-tagged NTCP subunits, demonstrating close proximity, in line with subunit interaction. As the subunits were labelled extracellularly, this demonstrates that NTCP dimers are present at the plasma membrane. Furthermore, NTCP lacking its C-terminus (NTCP Y307X) retains wtNTCP in the ER. Since the C-terminus of NTCP is not required for dimerization, this suggests that subunit interaction occurs in the ER. If the half-life of wtNTCP proteins would largely exceed the time each subunit was interacting with the Y307X mutant protein, the majority of wtNTCP would either eventually find another wild-type partner subunit and target to the plasma membrane, or travel as a monomer. The strong dominant-negative effect of NTCP Y307X on plasma membrane targeting of wtNTCP, therefore, is in line with a durable interaction. Similar effects were obtained upon overexpression of SLC10A4. SLC10A4 expression is very low in the liver, suggesting that heteromerization with NTCP will be limited in vivo. Nevertheless, the in vitro effects of SLC10A4 co-expression on NTCP function support our observations with NTCP Y307X that dimerization can affect NTCP trafficking to the plasma membrane.
Although these lines of evidence demonstrate that NTCP adopts a dimeric structure, it remained unclear whether a dimeric NTCP complex constitutes a single bile salt translocation pathway, or that subunits individually form functional units for bile salt uptake. The present study aimed to answer this question by co-expression of active and inactive NTCP subunits. When the amount of active NTCP subunits was kept equal, the addition of inactive subunits did not affect cellular taurocholate uptake. This was observed at all taurocholate concentrations tested, suggesting that both the Km and Vmax values of the complex are determined by the active subunits. This was evident for co-expression with NTCP E257N, a mutant incapable of transporting bile salts. Similarly, co-expression with SLC10A6, which is also unable to transport taurocholate, does not affect cellular bile salt uptake. Therefore individual NTCP molecules form functional units within the dimeric complex. The next question to be answered is what the functional consequence is of NTCP dimerization. The advantage of multiple translocation pathways within an oligomeric complex generally remains unclear, although broadly present in Nature . Several Na+-coupled transporters are present in oligomeric complexes when each subunit is functional, including NKCC2 (Na/K/Cl co-transporter 2; SLC12A1 , NBCe1-A (SLC4A4)  and EAAT2 (excitatory amino acid transporter 2; SLC1A2) . It is believed that oligomerization in such instances might be beneficial for protein stability or allow efficient trafficking. For a number of membrane proteins, oligomerization has been described to change the relative permeation of various substrates and ions [26,27] to generate novel permeation pathways  or determine affinity for specific ligands . We found that dimerization of NTCP with variant NTCP E257N or with SLC10A6 did not change plasma membrane abundance or taurocholate uptake. Whether (hetero)dimerization affects transport of other NTCP substrates, or generates permeation pathways for novel substrates requires further studies. We demonstrate that NTCP function was affected upon in vitro co-expression with NTCP Y307X or SLC10A4, due to reduced plasma membrane expression of NTCP. Other examples exist where transporter trafficking is affected by oligomerization [30–32]. This is particularly of interest in view of consequences of deleterious gene mutations, affecting targeting or protein stability. A number of variations in the human NTCP gene have been described with severe functional consequences in vitro [7,33]. On the basis of the results from the present study, we expect that such heterozygous gene alterations (e.g. S267F substitution ) do not reduce taurocholate uptake more than 50% in humans, as the wtNTCP subunits will operate normally, also in complex with inactive subunits. Reductions in Na+-dependent hepatic bile salt uptake can be expected when the heteromeric complex is mistargeted or more rapidly degraded. On the basis of our results seen upon co-expression of NTCP with SLC10A4 or NTCP Y307X, some cellular bile salt uptake probably remains, mostly depending on the relative abundance of the trafficking-impaired subunit. Mutations in other dimeric Na+-driven transporters, including NCCT and NKCC2, lead to human disease in a recessive manner. Therefore defects in Na+-dependent bile salt uptake are likely also recessively inherited or will be compensated by Na+-independent pathways. The results of the present study on NTCP and SLC10A6 suggest that homodimerization could be a common phenomenon in the SLC10A protein family. This hypothesis is supported by previous papers in which immunoblotting results point to dimeric complexes at high molecular mass for ASBT [34–39], SLC10A4 , SLC10A5 , SLC10A6  and SLC10A7 , compatible with dimeric complexes. Therefore the novel structural and functional information on NTCP provided in the present study probably extrapolates to other members of this family. The contribution of heteromerization to the function and regulation of other SLC10A family members requires further investigation. In summary, NTCP can form homo- and hetero-dimeric complexes where each subunit operates as a bile salt transport unit. Bile salt uptake is influenced by heterodimerization when this impairs NTCP trafficking to the plasma membrane.
acyl carrier protein
apical Na+-dependent bile acid transporter
ethylene glycol bis(succinimidyl succinate)
fluorescence resonance energy transfer
heat-shock protein 70
Na/K/Cl co-transporter 2
Na+–taurocholate co-transporting protein
- PNGase F
peptide N-glycosidase F
solute carrier family 10 (sodium/bile acid cotransporter family)
Ingrid Bijsmans and Stan van der Graaf conceived the project and designed the experixments. Ingrid Bijsmans, Rianne Bouwmeester and Stan van der Graaf carried out the experiments and performed data analysis. Joachim Geyer and Klass Nico Faber provided materials. All authors were involved in writing the paper and approved the final version before submission.
We thank Janette Heegsma for technical assistance with isolation of the rat liver membranes, Dr Nils Johnsson (Center for Molecular Biology of Inflammation, Münster, Germany) for providing the ACP-encoding construct, Dr Barbara Döring and Stephanie Schmidt (Institute of Pharmacology and Toxicology, Justus Liebig University of Giessen, Germany) for providing the V5-tagged constructs, and Dr Leo Klomp for assistance and helpful discussions.
This work was supported by the Dutch Organization of Scientific Research (Veni-grant SFJ vdG) [grant number 016.096.108] and the Wilhelmina Children's Hospital Research Fund.