Dephosphorylation of Threonine 38 Is Required for Nuclear Translocation and Activation of Human Xenobiotic Receptor CAR (NR1I3)

Abstract

Upon activation by therapeutics, the nuclear xenobiotic/ constitutive active/androstane receptor (CAR) regulates various liver functions ranging from drug metabolism and excretion to energy metabolism. CAR can also be a risk factor for developing liver diseases such as hepatocellular carcinoma. Here we have characterized the conserved threonine 38 of human CAR as the primary residue that regulates nuclear translocation and activation of CAR. Protein kinase C phosphorylates threonine 38 located on the α-helix spanning from residues 29–42 that constitutes a part of the first zinc finger and continues into the region between the zinc fingers. Molecular dynamics study has revealed that this phosphorylation may destabilize this helix, thereby inactivating CAR binding to DNA as well as sequestering it in the cytoplasm. We have found, in fact, that helix-stabilizing mutations reversed the effects of phosphorylation. Immunohistochemical study using an anti-phospho-threonine 38 peptide antibody has, in fact, demonstrated that the classic CAR activator phenobarbital dephosphorylates the corresponding threonine 48 of mouse CAR in the cytoplasm of mouse liver and translocates CAR into the nucleus. These results define CAR as a cell signal-regulated constitutive active nuclear receptor. These results also provide phosphorylation/dephosphorylation of the threonine as the primary drug target for CAR activation.

Introduction

The nuclear xenobiotic receptor CAR² (NR1I3), an orphan member of the nuclear receptor superfamily, was first characterized as a transcription factor that is activated by phenobarbital (PB), a common sedative used to treat diseases such as epilepsy and jaundice (1). CAR, upon activation by therapeutics such as PB, induces a large set of genes that encode for drug-metabolizing enzymes with the CYP2B enzymes being the classic CAR-regulated genes, thus increasing hepatic capability for drug metabolism and excretion (2–4). CAR plays the essential role in PB-induced promotion of hepatocellular carcinoma development (5). Additionally, CAR is critically involved in various liver functions including glucose, fatty acid, cholic acid, and bilirubin metabolism (6). The recent finding that CAR is the only PB-activated human nuclear receptor emphasizes the clinical and pharmacological importance of CAR (7). It still remains unknown as to how PB activates CAR, investigations for which have been hampered by the lack of a known residue in CAR to which PB transduces its signal.

CAR is sequestered in the cytoplasm of liver cells, making nuclear translocation the first step of CAR activation by drugs and xenobiotics (8). Two types of CAR activators can translocate CAR into the nucleus; one type, represented by PB, consists of a number of therapeutic drugs that indirectly translocate CAR into the nucleus without any direct binding. The other type includes various xenobiotics, acting as ligands, which directly bind to CAR, such as 1,4-bis-[2-(3,5-dichloropyridyloxy)]benzen (TCPOBOP) and 6-(4-chlorophenyl)imidazo[2,1-b][1,3]thiazole-5-carbaldehyde O-(3,4-dichlorobenzyl)oxime (CITCO) for mouse and human CARs, respectively (9, 10). Regardless of the types, the molecular mechanism of how these activators translocate CAR into the nucleus has been a mystery since CAR was first characterized as a xenobiotic-activated nuclear receptor (1). Available evidence, however, has suggested that the underlying mechanism should be a cell signal-mediated regulation; protein phosphatase 2A, for instance, may be involved in the PB-induced nuclear translocation in the mouse hepatocytes (11, 12). Moreover, growth factors, such as epidermal growth factor and hepatocyte growth factor, repress CAR nuclear translocation, and the U0126 inhibition of the MEK-ERK pathway spontaneously translocated CAR into the nucleus and activated the target gene Cyp2b10 without drug exposures (13). Neither the type of kinase that phosphorylates CAR nor the residue of CAR that can be phosphorylated by this kinase, however, has been identified to this time.

Here we present experimental results that identify protein kinase C and threonine 38 of human CAR as the key kinase and residue, respectively and, moreover, demonstrate that PB may dephosphorylate this threonine residue to translocate CAR into the nucleus in mouse liver. For the first time in our study of CAR biology, the target of CAR activation by PB is now directly linked to the phosphorylated threonine 38. Regardless of the direct or indirect type of CAR activators, the threonine 38 must be dephosphorylated in order for activation and translocation of CAR. This finding opens a new era for not only basic and mechanistic study but also clinical development.

EXPERIMENTAL PROCEDURES

Materials

CITCO was obtained from BIOMOL (Plymouth meeting, PA); PB and phorbol 12-myristate 13-acetate (PMA) were from Sigma; Complete mini protease inhibitor mixture tablets was from Roche Diagnostics GmbH. Antibody against human CAR was obtained from Perseus Proteomics Inc. (Tokyo, Japan). Antibody against the peptide KGFFRRpTVSKTSIGP (corresponding to residues 32–45 of hCAR with Thr-38 phosphorylated) was produced. The specificity of this antibody (named anti-P-Thr-38) was evaluated by enzyme-linked immunosorbent assay using the phosphorylated and non-phosphorylated peptides (AnaSpec Inc, San Jose, CA). For plasmids, hCAR in pCR3, pEYFP-c1, or pGEX4T-3 and (NR1)5-TK-luciferase reporter were previously constructed (14, 15). Using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) and appropriately mutated primers, mutations were introduced and verified by nucleotide sequencing.

In Vitro Phosphorylation Assays

Half a μg of bacterially expressed GST-hCAR and its mutants was incubated at 30 °C for 30 min in 30 μl of assay buffer containing 20 mm HEPES (pH 7.4), 1.67 mm CaCl₂, 1 mm dithiothreitol, 10 mm MgCl₂, 0.6 mg/ml phosphatidyl serine, and 100 μm ATP in the presence of 1 μl (1.6 units) of PKC (Promega, Madison, WI) and then subjected to Western blot analysis for phosphorylation using anti-P-Thr-38 antibody.

Cell Culture

HepG2 cells were cultured in minimal essential medium supplemented with 10% fetal bovine serum and antibiotics (100 units/ml penicillin) on plastic dishes at 37 °C for 24 h prior to transfection. At ∼75% confluence, the cells were transfected with plasmid using FuGENE 6 (Roche Diagnostics GmbH) according to the manufacturer's instructions.

Primary Hepatocytes

Mouse primary hepatocytes were isolated from 6–8-week-old CD1 male mice using a two-step collagenase perfusion and seeded on 6-well plates as described previously (1). One hour after seeding, the medium was changed to prewarmed Williams' E medium supplemented with dexamethasone (5 nm) with or without a given chemical and incubated for 6 h for preparation of nuclear extracts and total RNA.

Western Blot

Proteins were separated on a 10% SDS-polyacrylamide gel and transferred to polyvinylidene difluoride membrane (Invitrogen). The membrane was incubated with a given primary antibody: anti-GST antibody (1:5000), anti-FLAG horseradish peroxidase (1:5000), anti-P-Thr-38 antibody (0.1 μg/ml in Tris-buffered saline-Tween), or anti-CAR antibody, as described previously (16). Horseradish peroxidase-conjugated anti-mouse or anti-rabbit IgG (1:5000; Santa Cruz Biotechnology, Santa Cruz, CA) was used as the secondary antibody. Finally, protein bands were visualized using ECL Plus Western blotting detection reagent (GE Healthcare). Blots were subsequently stripped and restained for lamin β as a loading control.

Expression of YFP-tagged CAR in Mouse Liver

Plasmid was injected via the tail vein of CAR-null mice using the TransIT in vivo gene delivery system (Mirus, Madison, WI) according to the manufacturer's instructions. The mice were twice injected with PB (100 mg/kg of body weight) or phosphate-buffered saline at 3 and 6 h after the injection of plasmid and sacrificed at 2 h after the second drug administration. Liver sections (30-μm thickness) were prepared, and images were captured by confocal laser scanning microscopy as described previously (14). Using conventional fluorescent microscope, at least 100 of YFP fusion protein-expressing cells from each of four different sections were examined for each treatment to determine the pattern of intracellular localization of the receptor predominantly in the cytoplasm, present equally in the nuclear and cytoplasmic compartments, or primarily localized in the nucleus.

Luciferase Reporter Assays

Luciferase activity was measured in cell lysates using the Dual-Luciferase reporter assay system (Promega) as described previously (15). Promoter activity was determined in HepG2 cells from three independent transfections and calculated from firefly luciferase activity normalized against Renilla luciferase activity of an internal control pRL-SV40 plasmid.

Gel Shift Assays

Various hCAR proteins were produced using the in vitro transcription/translation system (TnT T7 quick-coupled system, Promega) and were coincubated with an in vitro translated retinoid X receptor and ³²P-labeled NR1 double strand DNA (40,000 cpm) probe in 10 μl of binding buffer containing 10 mm HEPES, pH 7.6, containing 0.5 mm dithiothreitol, 10% glycerol, 0.05% Nonidet P-40, 50 mm NaCl, and 0.5 μg of poly(dI-dC). The proteins were separated on a 5% acrylamide gel in 7 mm Tris acetate buffer, pH 7.5, containing 1 mm EDTA at 150 V for 2 h, and the gel was dried under vacuum and subjected to autoradiography at −70 °C.

Immunohistochemistry

Paraffin-embedded liver sections (5–7 μm thick) from phosphate-buffered saline or PB-treated mice were fixed with 4% paraformaldehyde for 6 h. Stain was performed on the Discovery XT system (Ventana Medical Systems, Inc., Tucson, AZ) using the OmniMap anti-rabbit kit. Tissues were incubated in anti-P-Thr-38 for 60 min without heat at a 1:10 dilution followed by a 16-min incubation time with label reagent. Visualization was performed with 3,3′-diaminobenzidine chromogen and then counterstained for 8 min with hematoxylin. For negative control tissue sections, normal rabbit serum (Jackson ImmunoResearch Laboratories, West Grove, PA) was used in place of the primary antibody.

Structural Models

Using molecular dynamics, solution structures of T38D and PO₃ Thr-38 mutants of hCAR (91-residue peptide, Arg-8–Arg-98) and wild type were derived. The initial structures were based on the DNA-bound structure of vitamin D3 receptor (VDR, PDB number 1KB2). Using this structure, mutations were made (SYBYL 8.0) to create a model of wild type hCAR in its DNA-bound form (two zinc ions, Cys coordinates), and these systems were energy-minimized using Amber 10 (17). The nucleotides were removed, and the wild type hCAR was solvated in a water box. Following the desired mutations, the resulting peptides were also solvated. Prior to equilibration, systems were subjected to 1) 100-ps belly dynamics runs with fixed peptide, 2) minimization, 3) low temperature constant pressure dynamics at fixed protein to assure a reasonable starting density, 4) minimization, 5) stepwise heating molecular dynamics at constant volume, and 6) constant volume molecular dynamics for a ns. Final unconstrained trajectories (see Fig. 2) were calculated at 300 K under constant pressure (11 ns, time step 1 fs) using PMEMD (Amber 10) to accommodate long range interactions. The parameters were taken from the FF03 force field. The charge on zinc ions is +2, whereas Cys coordinated with zinc carries −1 charge. The van der Waals parameters of zinc were adjusted to yield Zn-S distance to be between 2.25 and 2.50 Å during the simulations. No specific constraints were applied to maintain the Zn-S coordination.

FIGURE 2.

Functional characterization of the Thr-38 mutants. A, gel shift assays. The in vitro translated proteins of hCAR wild type (WT) and the T38A and T38D mutants were incubated with ³²P-labeled NR1 probe in the presence and absence of CITCO, electrophoresed on a polyacrylamide gel, and exposed to x-ray film as described under “Experimental Procedures.” Closed and open arrows point to CAR-NR1 complexes and free NR1, respectively. [³⁵S]Methionine was used for in vitro translation to show the equality of these proteins used for gel shift assay. The letter D above the lanes denotes DMSO, and the numbers indicate the μm concentrations of CITCO. B, cell-based reporter assays. Expression plasmids for hCAR wild type and the T38A and T38D mutants were co-transfected with pGL3-NR1-Luc reporter plasmid into HepG2 cells treated with DMSO (open bars) or CITCO (closed bars), from which cell extracts were prepared for luciferase assays. Empty pCR3 was co-transfected as control. Error bars indicate S.D.

RESULTS

Phosphorylation of Threonine 38

Empirically, we found that threonine 38 of hCAR can be phosphorylated by PKC in kinase assays in vitro. This prompted us to produce a specific antibody against a peptide of hCAR containing a phosphorylated threonine 38 to specifically detect phosphorylation of threonine 38; hereafter this antibody is called anti-P-Thr-38. The bacterially expressed GST-hCAR proteins were first incubated as the substrate with a purified PKC. Subsequent Western blot analysis, using anti-P-Thr-38, showed that PKC phosphorylates wild type hCAR but not the T38A mutant (Fig. 1A, top panel). Threonine 38 of hCAR is located in the region between the first and second zinc fingers within the DNA binding domain. It is conserved, as either threonine or serine, in not only a closely related VDR but also in the other members of the nuclear receptor superfamily (Fig. 1B). Noticeably, pregnane X receptor, another nuclear xenobiotic receptor that is nearest to CAR both structurally and functionally, has alanine instead at the corresponding position.

FIGURE 1.

In vitro phosphorylation of threonine 38 of hCAR by PKC. A, Western blots to show the phosphorylation. Purified GST-hCAR wild type (WT) and GST-T38A were incubated with or without PKC. The sample was divided into two fractions that were electrophoresed on SDS gels for Western blot using anti-P-Thr-38 (α-P-T38, top) and anti-GST (bottom). B, amino acid sequence alignments to show the conserved threonine residue (in red) on the helix that constitutes the region (in red) starting with glutamic acid 29 and ending with serine 42 of human CAR. mCAR, mouse CAR; hVDR, human VDR; hTRβ, human telomerase RNA component β; hPPARα, human peroxisome proliferator-activated receptor α; hRARα, human retinoic acid receptor α; hHNF4α, human HNF4α; hPXR, human pregnane X receptor; mPXR, mouse pregnane X receptor; Luc, luciferase.

Functional Alteration by Phosphorylation

Given the finding that threonine 38 can be phosphorylated, we mutated threonine 38 to aspartic acid and alanine to produce T38D and T38A mutants for studying the functional nature of this phosphorylation. First, wild type hCAR, T38A, and T38D were translated in vitro and subsequently incubated with a ³²P-labled NR1 oligonucleotide probe for gel shift assays. Although both wild type hCAR and mutant T38A strongly bound to the probe, the T38D mutant exhibited no gel shift band, thus confirming that the T38D mutant had greatly reduced DNA binding capability (Fig. 2A). The T38D did not show DNA binding even in the presence of the hCAR ligand CITCO. Next, these wild types and mutants were co-expressed with the NR-Luc reporter gene in HepG2 cells to measure their transactivation activity. Wild type CAR and the T38A mutant activated the NR1-Luc reporter gene, whereas the T38D mutant did not (Fig. 2B). In general, because of its high constitutive activity, CAR is not effectively activated by ligands in cell-based transfection assays. Consistent with this, CITCO was only able to slightly activate the wild type and the T38A mutant. The T38D mutant, on the other hand, did not respond to CITCO at all. These results indicated that phosphorylation of threonine 38 converts constitutively activated hCAR into an inactive receptor.

Structural Alteration by Phosphorylation

The three-dimensional structure of hCAR DNA binding domain was modeled by replacing the corresponding residues of a known x-ray crystal structure of human VDR DNA binding domain-DNA complex (1KB2) with those residues in hCAR. hCAR and human VDR share a 62% identity from arginine 8 to isoleucine 81 (a 73-amino acid residue segment) of hCAR. In the modeled hCAR DNA binding domain structure shown in supplemental Fig. 1A, threonine 38 is located on a C-terminal portion of the helix that contains the C-terminal part of the first zinc finger and continues into the region between the first and second zinc fingers. The side chain of threonine 38 is oriented opposite the CAR-DNA binding interface and does not directly interact with DNA. From this helix, the side chains of lysine 32, arginine 36, arginine 37, and lysine 41 interact with the DNA chain. Subsequently, threonine 38 was either phosphorylated or mutated to aspartic acid in the context of the hCAR DNA binding domain structure and was equilibrated 11 ns with a molecular dynamics run (PMEMD) in a realistic solvent environment, thereby deriving simulated solution structures for PO₃ Thr-38 and T38D, respectively. The simulated PO₃ Thr-38 structure revealed that phosphorylation distorts the C-terminal region of this helix, altering the side-chain orientation of lysine 41 so that this side chain is no longer able to bind to DNA (supplemental Fig. 1B). Resembling the simulated PO₃ Thr-38 structure, the T38D mutant also distorted a middle region of this helix and shifted the side chain orientation of Lys-41 (supplemental Fig. 1C). Thus, these structural simulations provided evidence supporting the conjecture that the T38D mutant mimics the structural alterations caused by phosphorylation of threonine 38.

What is seen in these models is that only one of the four hydrogen-bonding interactions between this helix of CAR and the DNA appears to be reduced significantly by the mutations. Struck by the severity of the functional consequence that occurred with phosphorylation of threonine 38, we questioned whether the phosphorylation might distort not only the C-terminal portion of the helix but also cause distortion of the entire helix. We decided to test this idea by selective mutations in the phosphorylated helix to see whether we could restore activity. There are two glycine residues near lysine 32 and cysteine 31 in the so-called P-box at the N terminus of the helix: glycine 30 and glycine 33. Cysteine 31 is one of the four residues that form the core of the first zinc finger. To answer the question of helix stability, these helix-destabilizing glycine residues were systematically substituted with a helix-stabilizing residue alanine in conjunction with the T38D mutation to form the G30A/T38D, G33A/T38D, and G30A/G33A/T38D mutants. First, the G30A/G33A/T38D structure was simulated, revealing that these alanine mutants retain the integrity of the helix (supplemental Fig. 1D).

Gel shift and cell-based transfection assays were employed to determine the DNA binding and transactivation activities of these double and triple mutants. For clarity, the simulated helix structures of the PO₃ Thr-38, T38D, and G30A/G33A/T38D mutants are depicted in supplemental Fig. 1 and are superimposed with the helix of wild type CAR (Fig. 3A, panels a, b, and c, respectively). All three glycine mutants recovered their DNA binding activity with G30A/T38D and G30A/G33A/T38D being more efficient than G33A/T38D (Fig. 3B). Reflecting the recovery of their DNA binding, these mutations enabled T38D to transactivate the NR1 reporter gene (Fig. 3C). G30A/T38D activated the reporter more efficiently than the G33A/T38D, whereas the G30A/G33A/T38D mutant fully recovered the transactivation activity. The same alanine substitutions were also introduced into the wild type and T38A mutant; both G30A/G33A/Thr-38 and G30A/G33A/T38A increased their DNA binding and retained the transactivation (supplemental Fig. 2). These results are consistent with our hypothesis that phosphorylation of threonine 38 destabilizes the entire helix, thereby abolishing DNA binding of CAR.

FIGURE 3.

Helix stability and CAR function. A, ribbon presentations of helix within the simulated PO₃ Thr-38 (red in panel a), T38D (cyan in panel b), and G30A/G33A/T38D (blue in panel c) structures are superimposed with that of wild type hCAR (gold). The program VMD was used to generate the pictures. B, gel shift assays. hCAR proteins of double and triple mutations were in vitro translated and were incubated with ³²P-labeled NR1 probe (top panel). Bottom panel, the equality of these fusion proteins used for assays by radioautography of [³⁵S]methionine-labeled proteins. WT, wild type. C, cell-based reporter assays. The double and triple mutants were co-transfected with (NR1)5-TK-luciferase reporter plasmid into HepG2 cells treated with DMSO (open bars) or CITCO (closed bars), from which cell extracts were prepared for a luciferase assay. Empty pCR3 was co-transfected as a control. Error bars indicate S.D.

Cytoplasmic Localization of the T38D Mutant in Mouse Liver

Expression plasmids of YFP-tagged hCAR, T38A, and T38D were injected via tail vein to directly express these CAR proteins in the mouse livers of Car^−/− mice. Wild type hCAR was retained in the cytoplasm of mouse liver and translocated into the nucleus following PB treatment (Fig. 4A). Over 70% of the hepatocytes examined expressed hCAR in the cytoplasm of the non-induced mouse livers; conversely 70% of PB-treated hepatocytes displayed hCAR expression in the nucleus (Fig. 4B). The T38A mutant, on the other hand, spontaneously accumulated in the nucleus of non-induced mouse livers. Nuclear localization did not significantly change even after PB treatment; approximately half of the hepatocytes examined showed the exclusive nuclear localization of the T38A mutant. In sharp contrast, the T38D mutant was found to be exclusively retained in the cytoplasm of mouse livers before and after PB treatment (Fig. 4, A and B). In addition, not only PB, which indirectly activates hCAR, but also CITCO, which directly binds to activate hCAR, could not translocate the T38D mutant into the nucleus. Thus, the T38D mutant lacked nuclear translocation ability, whereas the T38A mutant lost its capability to be retained in the cytoplasm. These results indicate that phosphorylation of threonine 38 is the primary determinant for retention of CAR in the cytoplasm and that the dephosphorylation is a prerequisite for CAR to undergo nuclear translocation.

FIGURE 4.

Intracellular localization in vivo in mouse liver. A, by injecting their expression plasmids via the tail veins of Car^−/− mice, YFP fusion proteins were directly expressed in liver cells. Phosphate-buffered saline (PBS), PB, DMSO (DM), or CITCO (CIT) was administered to these mice, from which liver sections were prepared and examined by microscopy for YFP expression. WT, wild type. B, from each of the three different sets of mice, over 100 cells, which expressed a given fusion protein, were examined to calculate the statistics of CAR localizations, predominantly in the cytoplasm (blue bars), equally in the cytoplasm and the nucleus (gray bars), and predominantly in the nucleus (red bars). Error bars indicate S.D. C, from each of the three (G30A/G33A/T38D) and two (wild type, G30A/G33A/Thr-38 and G30A/G33A/T38A), over 100 cells were examined and presented in the same way as those in B.

Given the observation that the alanine mutations of glycine 30 and glycine 33 restore DNA binding and transactivation activities to the T38D mutant (Fig. 3, B and C), YFP-tagged triple mutant G30A/G33A/T38D was directly expressed in mouse liver to examine its intracellular localization. This triple mutant was primarily retained in the cytoplasm of phosphate-buffered saline-treated mouse livers, as observed with the T38D mutant. Unlike the T38D mutant, however, this triple mutant accumulated in the nucleus of PB-induced mouse liver (Fig. 4C). The same alanine substitutions were also introduced into the wild type and T38A mutant, producing YFP-tagged G30A/G33A/Thr-38 and YFP-tagged G30A/G33A/T38A and directly expressed in mouse livers. The double G30A/G33A/Thr-38 mutant greatly increased its spontaneous nuclear localization (Fig. 4C). As expected, the triple G30A/G33A/T38A mutant spontaneously translocated into the nucleus, as observed with the T38A mutant (Fig. 4C). These results indicate that helix stabilization by the glycine mutations enables the T38D mutant to regain nuclear translocation capability.

Dephosphorylation of CAR by PB in Mouse Liver

After confirming that mouse CAR can also be phosphorylated by PKC (Fig. 5A), we employed Car^+/+ and Car^−/− mice to generate further experimental evidence to support the premise of CAR regulation by the PKC-dependent phosphorylation of threonine 38. TCPOBOP, acting as a ligand, bound directly to and activated mouse CAR so that mouse primary hepatocytes treated with TCPOBOP translocated CAR into the nucleus and exhibited increased CYP2B10 mRNA, as shown by Western blot (Fig. 5B) and real time-PCR (Fig. 5C). Co-treatment of mouse primary hepatocytes with TCPOBOP and the PKC activator PMA repressed both the nuclear accumulation of CAR and the increase of CYP2B10 mRNA, whereas treatment with PMA alone showed no effect (Fig. 5, B and C). In addition, PB was utilized for co-treatment, confirming that PMA also repressed PB-induced nuclear accumulation of CAR (supplemental Fig. 3A). The key question to be answered was whether CAR is phosphorylated in the cytoplasm of liver hepatocytes and then dephosphorylated following PB treatment. To this end, immunohistochemistry was performed using anti-P-Thr-38 antibody; liver sections were prepared from the non-induced and PB-induced Car^+/+ mice. Anti-P-Thr-38 immunostained the liver section from the non-induced Car^+/+ mice (Fig. 5D, panel a) but did not react with the section from the PB-induced Car^+/+ mice (Fig. 5D, panel b). Serving as controls for the specificity of this immunoreaction, normal IgG and anti-P-Thr-38 showed no staining on the liver sections from the non-induced Car^+/+ mice and Car^−/− mice, respectively (Fig. 5D, panels c and d). Another critical finding was that anti-P-Thr-38 did not stain CAR in the nucleus of the PB-induced mouse liver hepatocytes. In addition, Western blot analysis of the whole cell extracts from non-treated and PB-treated mouse primary hepatocytes also showed dephosphorylation of Thr-38 after PB treatment (supplemental Fig. 3, B and C). Thus, these results have demonstrated that threonine 38 of CAR is phosphorylated in the cytoplasm and that PB dephosphorylates this threonine to translocate CAR into the nucleus.

FIGURE 5.

Phosphorylation of CAR in vivo in mouse hepatocytes. A, a GST-fused wild type (WT) and a T38A mutant of mouse CAR (mCAR) were phosphorylated and analyzed as described in the legend for Fig. 1. αP-Thr-38, αP-Thr-38. B, repression of nuclear CAR accumulation by PMA. Mouse primary hepatocytes were pretreated with PMA (100 nm) for 30 min followed by treatment with DMSO, TCPOBOP (TC, 250 nm), or TCPOBOP plus PMA for 6 h, from which nuclear extracts were prepared for subsequent Western blot analysis using anti-CAR antibody as described under “Experimental Procedures.” An in vitro translated mouse CAR was also subjected to Western blots as CAR marker. The homogeneity of the nuclear extracts was confirmed by Western blots using anti-lamin β antibody (αLamin β). C, repression of TCPOBOP induction of CYP2B10 mRNA by PMA. RNAs were prepared from the same set of the mouse primary hepatocytes used in B, subjected to real time-PCR to measure the levels of CYP2B10 mRNA, which are expressed by using those in the DMSO-treated cells as one. Error bars indicate S.E. D, immunohistochemistry of liver sections. Liver sections were immunostained with anti-P-Thr-38 or normal IgG as described under “Experimental Procedures.” Sections from non-treated (panel a) and PB-treated (panel b); Car^+/+ mice were immunoreacted with anti-P-Thr-38, respectively; sections from non-treated Car^+/+ mouse were immunoreacted with normal IgG (panel c); sections from Car^−/− mouse were immunoreacted with anti-P-Thr-38 (panel d).

DISCUSSION

Because there is a broad spectrum of integrated roles for CAR in the regulation of various liver functions, it has been of critical importance to decipher the molecular mechanism by which therapeutics and xenobiotics activate this nuclear receptor. Threonine 38 is now characterized as the key residue conferring CAR with drug responsiveness for its activation. Dephosphorylation of threonine 38 enables CAR to translocate into the nucleus and to activate transcription of target genes. After treatment with the classic CAR activator PB, in fact, dephosphorylation of the corresponding threonine 48 of mouse CAR occurs. Thus, this dephosphorylation can be the primary mechanism by which therapeutics and xenobiotics translocate CAR into the nucleus as well as how cell signals regulate CAR activation.

Threonine 38 is conserved not only in mouse CAR but also in a variety of other members of the nuclear receptor superfamily. For example, the corresponding residue in HNF4α is threonine 78; mutating this residue to aspartic acid is known to abolish nuclear localization and DNA binding capability (18). Our dynamic simulation of the PO₃ Thr-38 and T38D solution structures has revealed that phosphorylation of threonine 38 distorts the C-terminal portion of the α-helix. Our additional simulation of the T38W structure revealed no distortion of the helix, substantiating the role of phosphorylation in helix destabilization (data not shown). Subsequent study with alanine mutations of Gly-30 and Gly-33, which are located in the N terminus of the helix, however, has indicated that the helix distortion may also extend toward the N terminus, destabilizing the entire helix as well as the first zinc finger. The helix in question (residues 29–42) appears to exhibit an intrinsic instability. Given the caveat that it is difficult to accurately estimate the relative stability of a helix in a protein, the stability score for this helix starting with glutamic acid 29 and ending by serine 42 of human CAR is 0.09 as compared with 4.48 and 0.00 for poly(A)₁₄ and poly(P)₁₄, respectively, when the AGADIR program was used to calculate these scores. This instability may allow phosphorylation to regulate the function of the helix to form a binding interface with DNA, providing CAR with an efficient mechanism to control its high constitutive activity as well as to retain it in the cytoplasm, thus converting CAR into a signal-regulated and drug-activated receptor.

What fascinates us most is that CAR has evolved to integrate phosphorylation and helix stabilization into the molecular mechanism that sequesters CAR in the cytoplasm. A body of previous observations has suggested that a small percentage of CAR population exists with threonine 38 that may be dephosphorylated in the absence of drug exposures. For example, our previous immunohistochemical study of the liver sections of normal mice had shown that some hepatocytes near the central vein of non-treated mice display nuclei stained with anti-CAR antibody (8). In addition, the basal expression of the Cyp2b10 gene is attenuated in normal Car^−/− mice (5). Thus, it is reasonable to expect that there should be a cellular signal mechanism built into normal liver physiology to regulate CAR in liver in a phosphorylation-dependent manner. PKC phosphorylation of threonine 38 is now found to be the primary determinant of cytoplasmic sequestration of CAR. Cytoplasmic localization of the T38D mutant in the mouse liver after treatment with PB or CITCO provided strong evidence indicating that threonine 38 must be dephosphorylated in order for CAR to undergo nuclear translocation regardless of the type of CAR activators, direct or indirect. We have now found that PB, which translocates mouse CAR into the nucleus, in fact dephosphorylates the corresponding threonine 48 in the cytoplasm of mouse liver cells. Thus, this conserved threonine (38 and 48 in human and mouse CARs, respectively) appears to be the primary determinant that regulates CAR activation. The S202D mutant of mouse CAR has previously shown retention in the cytoplasm of PB-treated liver, suggesting that serine 202 may regulate CAR nuclear translocation (16). Because the corresponding non-phospho-S202A mutant did not spontaneously accumulate in the nucleus of mouse liver, however, serine 202 was not thought to be the primary regulator of CAR activation. In addition, phosphorylation of serine 202 of endogenous CAR has not been shown in mouse liver.

The regulatory mechanism by which PB dephosphorylates threonine 48 (or threonine 38 in human CAR) remains of major interest for future investigations. Our previous study demonstrated that PB recruits PP2A to the cytoplasmic CAR-Hsp90 complex to translocate CAR into the nucleus (12). We now know that the target of PP2A dephosphorylation can be this threonine but that PP2A requires additional factors to dephosphorylate threonine 38 (data not shown). Thus, more work is needed to establish the molecular mechanism by which threonine 38 can be dephosphorylated after PB treatment. Recently, the growth factor-MEK-ERK1/2 pathway has been characterized as a cellular signal to sequester CAR in the cytoplasm; the inactivation of ERK1/2 by the MEK inhibitor U0126 spontaneously accumulates CAR in the nucleus of mouse primary hepatocytes (13). Our recent study has found that ERK1/2 specifically interacts with the phosphorylated form of CAR (data not shown), suggesting that ERK1/2 may be a signal that enables CAR to be phosphorylated. Given the notion that PKC phosphorylation and dephosphorylation of threonine 38 by PP2A are the primary determinant for CAR, ERK1/2 may be the signal that regulates the phosphorylation status of CAR.

Phosphorylation appears to be the primary regulatory mechanism that regulates CAR activation. Recent studies have shown that CAR can be naturally activated under certain disease states. For example, CAR appears to be the regulatory factor that increases serum triglyceride levels in diabetic ob/ob mice as this abnormality becomes normal when the Car^−/− genotype is introduced into the ob/ob mice (19). NADPH-cytochrome P450 oxidoreductase (POR) transfers electron to reduce cytochrome P450, which is essential for P450 to catalyze the oxygenase reaction. POR-null mice also exhibit elevated levels of hepatic steatosis, and the introduction of the Car^−/− genotype into the POR-null restores it to normal (20). The signal-mediated regulation via threonine 38 should be seriously considered as the CAR activation mechanism under pathophysiological conditions such as these.

In conclusion, we have defined CAR as the signal-regulated constitutive active receptor in which threonine 38 acts as the primary target of phosphorylation. PB dephosphorylates this threonine to activate and translocate CAR into the nucleus. CAR activation under pathophysiological conditions such as diabetes and high fat diets can also be regulated by this signal-mediated mechanism. Thus, the concept of CAR as a signal-regulated nuclear receptor provides us with a new manner in which to investigate CAR, including its activation mechanism, biological roles, and clinical implications.