Cross-talk between Thyroid Hormone Receptor and Liver X Receptor Regulatory Pathways Is Revealed in a Thyroid Hormone Resistance Mouse Model*

Hypercholesterolemia is found in patients with hypothyroidism and resistance to thyroid hormone. In this study, we examined cholesterol metabolism in a thyroid hormone receptor β (TR-β) mutant mouse model of resistance to thyroid hormone. Whereas studies of cholesterol metabolism have been reported in TR-β knock-out mice, generalized expression of a non-ligand binding TR-β protein in this knock-in model more fully recapitulates the hypothyroid state, because the hypothyroid effect of TRs is mediated by the unliganded receptor. In the hypothyroid state, a high cholesterol diet increased serum cholesterol levels in wild-type animals (WT) but either did not change or reduced levels in mutant (MUT) mice relative to hypothyroidism alone. 7α-Hydroxylase (CYP7A1) is the rate-limiting enzyme in cholesterol metabolism and mRNA levels were undetectable in the hypothyroid state in all animals. triiodothyronine replacement restored CYP7A1 mRNA levels in WT mice but had minimal effect in MUT mice. In contrast, a high cholesterol diet markedly induced CYP7A1 levels in MUT but not WT mice in the hypothyroid state. Elevation of CYP7A1 mRNA levels and reduced hepatic cholesterol content in MUT animals are likely because of cross-talk between TR-β and liver X receptor α (LXR-α), which both bind to a direct repeat + 4(DR+4) element in the CYP7A1 promoter. In transfection studies, WT but not MUT TR-β antagonized induction of this promoter by LXR-α. Electromobility shift analysis revealed that LXR/RXR heterodimers bound to the DR+4 element in the presence of MUT but not WT TR-β. A mechanism for cross-talk, and potential antagonism, between TR-β and LXR-α is proposed.

Thyroid hormone is an important physiological regulator of cholesterol metabolism (1,2). Hypercholesterolemia is found in patients with hypothyroidism and in those with resistance to thyroid hormone (2)(3)(4). Elevated serum cholesterol levels in hypothyroidism are restored to normal levels upon treatment with thyroid hormone (2,5). In contrast, thyroid hormone treatment occasionally has paradoxical effects on serum cholesterol levels in patients with resistance to thyroid hormone (4). Cholesterol homeostasis is maintained by coordinate regulation of three primary pathways in the liver ( Fig. 1) (6, 7). Two of these pathways maintain cholesterol supply by either de novo synthesis, which is largely regulated by hepatic 3-hydroxy-3-methylglutaryl coenzyme A reductase (HMGC-R), 2 or cellular uptake of plasma cholesterol via low density lipoprotein receptor (LDLR). A third pathway involves the elimination of cholesterol through the synthesis of bile acids; 7␣-hydroxylase (CYP7A1) is the rate-limiting enzyme in this pathway (8).
Thyroid hormone is known to regulate each of these pathways at the transcriptional level, and the predominate thyroid hormone receptor (TR) isoform in the liver is TR-␤1 (9 -12). Therefore, TR-␤ knock-out mice (Ϫ/Ϫ) have been analyzed in detail for changes in cholesterol metabolism (13,14). Both Weiss et al. (13) and Gullberg et al. (14) reported that serum cholesterol levels did not increase in response to hypothyroidism nor decrease after triiodothyronine (T 3 ) replacement in TR-␤ knock-out mice (Ϫ/Ϫ). Gullberg et al. (14) also reported that T 3 -deficient TR-␤ Ϫ/Ϫ mice showed an augmented CYP7A1 response to dietary cholesterol and did not develop hypercholesterolemia to the same extent as wild-type mice. Unfortunately, TR knock-out mice do not recapitulate the findings in either the hypothyroid state or in the syndrome of thyroid hormone resistance as other groups, including our own, have demonstrated that the TR possesses ligand-independent properties (15,16). We studied, therefore, a mouse model of resistance to thyroid hormone where a point mutation in the ligand-binding domain of TR-␤ abolished its ability to bind to T 3 (⌬337T, Ref. 15). Here, we characterize an interaction between liver X receptor (LXR) and TR, which bind to a similar DNA-response element, and show a unique role for the unliganded mutant TR-␤ in cholesterol metabolism.

EXPERIMENTAL PROCEDURES
Animals-TR-␤ knock-in mice have a mutation that deletes three nucleotides in exon 6 of the TR-␤ gene, resulting in loss of a threonine residue in the receptor, which prevents ligand binding (17)(18)(19)(20). The genetic background of the TR-␤ knock-in mice is a hybrid of two strains (129/Sv x C57/BL6). Four-week-old male mice were employed for the study. The number of mice for each study is indicated in the figure legends. All aspects of animal care were approved by the Institutional Animal Care and Use Committee of Gunma University Graduate School of Medicine (Maebashi, Gunma, Japan) and The University of Chicago (Chicago, IL). Animals were maintained on a 12-h light/12-h dark schedule (light on at 06:00 h) and fed laboratory chow as indicated and water ad libitum. A 2% cholesterol diet (21) was purchased from Oriental Bioservice (Tokyo, Japan). Mice were rendered hypothyroid by inclusion of 0.1% methimazole (MMI) in the drinking water and 1% (w/w) propylthiouracil (PTU) in chow for 21 days (22,23). Then, the mice were injected daily with 5 g/100 g body weight T 3 for an additional 5-day period to render mice euthyroid. A high cholesterol diet was begun during induction of hypothyroidism with MMI and PTU and continued through the T 3 dosing period. The total duration that animals were fed either chow or a 2% cholesterol diet was 2 weeks. Serumfree thyroxine and free T 3 levels were determined by double antibody radioimmunoassay kits (Ortho Diagnostics Co., Ltd., Tokyo, Japan). Mice were sacrificed in the morning to obtain either blood or tissue samples.
Lipid Determination-Serum total cholesterol levels were measured with a Determiner TC555 kit (Kyowa Medex, Tokyo, Japan). Serum triglyceride levels were determined with an SRL kit (SRL, Tokyo, Japan). Whole livers were removed, weighed, and sections were taken (100 mg) for analysis. The liver samples were homogenized in a Polytron in chloroform/methanol (2:1, v/v) and lipid was extracted as previously described (24) and analyzed for cholesterol content with a Determiner TC555 kit (Kyowa Medex).
Measurement of Bile Acid Pool Size-Mice were fed a 2% cholesterol diet for 2 weeks. Pool size was determined as the total bile acid content of small intestine, gallbladder, and the liver, combined as described (25). These organs were homogenized in 20 ml of ethanol (at 60°C), the extract was filtered and a 1-ml aliquot dried with a centrifuge concentrator at room temperature. The residue was dissolved in 1 ml of methanol and subjected to an enzymatic assay for total bile acid content (Kyokuto Pharmaceutical Co., Tokyo, Japan).
Northern Blot Analysis-Total RNA was extracted using ISOGEN (Nippon Gene, Tokyo, Japan) and 5-20 g of total RNA was subjected to Northern blot analysis as indicated in the figure legends. A rat cDNA probe for rat 5Ј deiodinase was a gift from Dr. C. N. Mariash. Rat cDNA templates for CYP7A1 (12), HMG-CoA reductase (11), and LDL receptor (11) were a gift from Dr. G. C. Ness. For CYP7A1, HMG-CoA reductase, and LDL receptor, Northern blot analysis was performed with the use of [␣-32 P]UTP-labeled antisense riboprobes. The hybridization bands were quantitatively measured using Adobe Photoshop 4.0 (Adobe Systems Corp., San Jose, CA) and NIH Image (Scion Corp., Frederick, MD) and standardized against cyclophilin controls. RNA samples from at least five individual mice treated in the same way were obtained for each condition and genotype. All Northern blots were repeated at least three times, using individual RNA samples, with similar results. Representative Northern blots are shown.
Cell Culture and Cotransfection Assays-Transient transfections in HepG2 cells were performed using a standard calcium phosphate method. The cells were maintained in 5% CO 2 in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. Receptor expression plasmids encoding mouse LXR-␣, and human TR-␤1, MUT TR-␤1 (⌬337T), and WT RXR-␣ in pSG5 vectors were cotransfected with a luciferase reporter plasmid (pGL3Luc) carrying mouse CYP7A1 promoter (Ϫ607 to ϩ26 bp). CYP7A1 mutant promoter was generated with PCR mutagenesis as it has the mutant DRϩ4 site (LXRE). The LXRE in the mouse CYP7A1 promoter (Ϫ73 bp, TGGTCACCCAAGT-TCA, Ϫ58 bp) was mutated (Ϫ73 bp, CTCGAGCCCACCATGG, Ϫ58 bp). 22(R)-Hydroxy(OH)cholesterol (Sigma) was dissolved in ethanol and added at a concentration of 10 mM to the cells after transfection in medium containing 10% resin charcoal double-stripped fetal bovine serum. Data are presented as mean Ϯ S.E. arbitrary light units from three experiments, each performed in triplicate.
Statistical Analyses-Values are expressed as mean Ϯ S.E. The significance of differences between mean values was evaluated using the unpaired Student's t test. Sample groups showing heterogeneity of variance were appropriately transformed before analysis.

Serum Cholesterol Levels in the TR-␤ Knock-in Mice-
To confirm the dominant negative effects of the mutant TR in the liver, we measured total cholesterol levels in wild-type (WT) and TR␤ knock-in mice. Because TR-␤ mutant (MUT) animals have significantly higher TH levels, we considered that excess TH (Table 1) may overcome some of the dominant negative effects of the mutant TR-␤. To eliminate this effect, TR-␤ WT and MUT mice were rendered hypothyroid with a MMI/PTU diet (22), then euthyroid by T 3 administration. As shown in Table 1, TR-␤ heterozygous (WT/MUT) and especially homozygous (MUT/MUT) mutant mice have extremely high baseline thyroxine levels, which were reduced to hypothyroid levels by MMI/PTU treatment. A MMI/PTU diet followed by T 3 administration returned free T 3 levels in all mice to euthyroid levels ( Table 2). We next measured serum cholesterol in hypothyroid animals before treatment or after administration of a high cholesterol diet, T 3 , or both. As shown in Fig. 2A, we found no significant differences in serum cholesterol levels in all genotypes in the hypothyroid state (first set of bars). When we administered a high cholesterol diet (2%) to hypothyroid animals (second set of bars), serum cholesterol levels were appropriately increased in WT animals. Paradoxically, serum cholesterol levels were decreased in MUT/MUT mice after cholesterol administration; WT/MUT mice had an intermediate value. This phenomenon was also observed when the animals were rendered euthyroid during a high cholesterol diet (third set of bars). Although euthyroid T 3 administration (fourth set of bars) reduced serum cholesterol levels somewhat, these differences did not achieve statistical significance versus the hypothyroid state.
Hepatic Cholesterol Levels in the TR-␤ Knock-in Mice-Hepatic cholesterol levels were then compared in WT and MUT animals after administration of a high cholesterol diet. Under hypothyroid condi-tions, hepatic cholesterol levels were significantly lower in WT/MUT and MUT/MUT versus WT mice after a high cholesterol diet (Fig. 2B, first set of bars). Correction of the hypothyroid state reduced hepatic cholesterol levels in WT animals but had no additional significant effect in MUT animals (Fig. 2B, second set of bars). In summary, we found a paradoxical reduction in both serum and hepatic cholesterol levels in mutant mice after administration of a high cholesterol diet. We also performed Oil Red O staining on liver sections from these mice. As shown in Fig. 2C, the liver histology of the WT animals showed increased hepatic Oil Red O staining versus the mutant animals. Measurement of bile acid pools in WT and MUT animals indicates that the decrease in hepatic cholesterol content and staining in MUT animals is associated with a significant increase in their bile acid pool (Fig. 2D).
The Expression of Thyroid Hormone-regulated Genes Involved in Cholesterol Metabolism in the TR-␤ Knock-in Mice-To understand the molecular mechanism by which cholesterol metabolism is altered in MUT mice, the expression of thyroid hormone-regulated genes involved in cholesterol metabolism was surveyed in MUT animals (Fig.  3). The 5Ј deiodinase type 1 (5ЈDI) gene is positively regulated by thyroid hormone in the liver (28) and is a good indicator of thyroid status in mice. During hypothyroidism, 5ЈDI mRNA levels were undetectable in all genotypes regardless of whether the animals were fed a high cholesterol diet. However, 5ЈDI mRNA levels were induced by T 3 in WT but induction was reduced in MUT animals. In fact, 5ЈDI induction was totally absent in homozygous mutant animals. These data indicate that the MUT TR-␤ has a significant dominant negative effect in the liver.
Cholesterol 7␣-hydroxylase, CYP7A1, the rate-limiting enzyme involved in cholesterol metabolism to bile acids, is a central regulator of cholesterol homeostasis (29,30). It is known that thyroid hormone increases CYP7A1 mRNA levels and that this effect is critically important for the hypocholesterolemic action of thyroid hormone (29,30). As shown in Fig. 3, mRNA levels for CYP7A1 were immeasurable in all genotypes during hypothyroidism. After T 3 replacement, CYP7A1 mRNA levels were increased in WT mice; however, this was not the case in mutant mice, where the maximal induction of mRNA levels was 30% in MUT/MUT animals as compared with WT mice. In contrast, administration of a cholesterol diet had a minimal effect on CYP7A1 mRNA levels in WT animals, whereas a large induction of CYP7A1 levels was observed in MUT animals. For example, in MUT/MUT mice, CYP7A1 mRNA expression was dramatically induced by cholesterol administration (280% compared with 15% in WT mice).
HMGC-R is the rate-limiting enzyme in cholesterol biosynthesis (31) and its expression is increased by thyroid hormone. We noted that HMGC-R mRNA levels were increased in WT animals after T 3 replacement (defined as 100%, after T 3 administration in WT animals, see Fig.  3). In contrast, HMGC-R mRNA levels were reduced by cholesterol treatment in T 3 -treated WT animals as expected. Interestingly, HMGC-R mRNA levels were increased in the hypothyroid state in MUT animals compared with WT animals, suggesting that the unliganded mutant TR may activate expression of certain genes in the liver. HMGC-R mRNA levels did increase after T 3 treatment in WT/MUT mice, but less so on a -fold basis (ϳ2-fold in WT/MUT versus 10-fold in

Cross-talk between TR and LXR Signaling Pathways
WT mice) because of the increase in expression in the hypothyroid state. The effect of T 3 treatment was further reduced in MUT/MUT animals (Ͻ1.5-fold increase versus the hypothyroid state). Like WT animals, cholesterol administration also decreased HMGC-R mRNA levels in MUT animals. Expression of LDLR is positively regulated by thyroid hormone (10 -12) and this receptor promotes the uptake of LDL into the cell. As shown in Fig. 3, WT mice demonstrate a significant increase in LDLR mRNA levels from the hypothyroid to the T 3 treated state (15-100%). Similar to HMGC-R mRNA levels, unliganded mutant TR increased expression of LDLR in the hypothyroid state. In addition, T 3 treatment in mutant mice resulted in less stimulation from the hypothyroid state, especially in the MUT/MUT animals. Cholesterol administration to T 3 -treated animals reduced LDLR mRNA expression in all genotypes, similar to the results observed with HMGC-R.
Finally, we measured LXR-␣ mRNA levels in these mice (Fig. 3). This receptor is essential for CYP7A1 expression and regulation by serum cholesterol levels (33). LXR-␣ mRNA levels in all genotypes displayed a similar pattern of gene expression. T 3 treatment resulted in a 2-fold increase in gene expression in both WT and MUT mice, suggesting a potential role for TR␣ or a non-TR mediated pathway in this process. Cholesterol treatment had no significant effect on gene expression.

Cross-talk of TR-␤ and LXR-␣ on the DRϩ4 Element in the CYP7A1
Gene Promoter-We next determined whether high cholesterol treatment could induce CYP7A1 expression in the presence of WT and MUT TR-␤. We employed a co-transfection study using HepG2 cells, a hepatocyte cell line. In rodents, the CYP7A1 gene promoter contains a single DRϩ4 element in the proximal promoter, which functions as a positive LXRE (34,35). We, therefore, investigated a possible interaction between LXR-␣ and TR-␤1 on a mouse CYP7A1 gene as illustrated in the model (Fig. 4A). As shown in Fig. 4B, LXR-␣ co-transfection stimulated the mouse CYP7A1 reporter activity about 2-fold after treatment with its ligand, 22(R)-hydroxycholesterol (36). Co-transfection of WT TR-␤1, in contrast, reduced both basal and ligand-stimulated reporter activity in a dose-dependent manner. Co-transfection of the MUT TR-␤, however, had no significant effect on LXR-mediated transactivation by 22(R)-hydroxycholesterol. In data not shown, the reporter was stimulated ϳ3-fold by 10 nM T 3 after cotransfection with the WT but not MUT TR-␤ and responses to both ligands were eliminated when the LXRE/thyroid hormone response element was mutated as described. These data indicate that WT antagonized LXR-␣ action, whereas the MUT TR-␤ did not.
Next, we performed a gel shift analysis using in vitro translated TR-␤, RXR-␣, and LXR-␣ proteins and radiolabeled mouse CYP7A1 and consensus DRϩ4 probes. As shown in Fig. 4C, the WT TR/RXR heterodimer migrated more rapidly than the LXR/RXR heterodimer (com- pare lanes 2 and 4). T 3 treatment dissociated the WT TR homodimer (compare lanes 2 and 3) but had no effect on the MUT TR homodimer (compare lanes 8 and 9). The LXR ligand, 22(R)-hydroxycholesterol, had no effect on LXR/RXR heterodimer binding (data not shown), indicating that DNA binding of the LXR/RXR heterodimer, like the TR/RXR heterodimer, was not dependent on ligand. When WT TR, LXR, and RXR were included in the binding reaction, only WT TR/RXR heterodimer formation was observed (lane 5), and this effect was not altered by the addition of T 3 (compare lanes 5 and 6) or by 22(R)hydroxycholesterol (data not shown). In contrast, when MUT TR, LXR, and RXR were included in the binding reaction, both MUT TR/RXR and LXR/RXR heterodimers were observed (lane 10). Western blot analysis for the TR-␤ protein confirmed that both the WT and MUT proteins were synthesized equally in an in vitro coupled transcription/translation system (data not shown).
Finally we evaluated the occupancy of the proximal LXRE in the CYP7A1 gene in WT and MUT mice using a ChIP assay. As shown in Fig. 4D, TR-␤1 binding to the CYP7A1 promoter in the liver of WT mice was significant (lane 4), whereas LXR-␣ binding to that same fragment was minimal (lane 7). In contrast, LXR-␣ binding to this fragment was significant in MUT/MUT mice (lane 9), whereas TR-␤1 binding to this fragment was minimal (lane 6). The WT/MUT animals displayed an intermediate pattern. As controls in this experiment, a nonspecific

. Expression of thyroid hormone-regulated genes involved in cholesterol metabolism is altered in TR-␤ knock-in mice.
Wild-type (WT/WT), heterozygous knock-in (WT/MUT), or homozygous knock-in (MUT/MUT) mice were rendered hypothyroid with MMI and PTU, and treated with or without a 2% cholesterol diet, and with or without T 3 replacement. Liver total RNA was isolated and subjected to Northern blot analysis. Representative Northern blots of type I 5Ј deiodinase (5ЈDI), CYP7A, HMGC-R, LDL-R, and LXR-␣ are shown. Relative OD was controlled for cyclophilin mRNA levels by NIH image software. OD levels in WT mice rendered hypothyroid and replaced with T 3 were assigned a value of 100% for each group. B.D., below detection.
IgG did not precipitate a protein-DNA complex in any mice (lanes 1-3) and a RXR-␣ specific antibody showed an equivalent pattern in all mouse genotypes (lanes 10 -12).

DISCUSSION
Thyroid hormone has important effects on cholesterol metabolism. Previous investigators have used TR knock-out animals to analyze cholesterol metabolism and have concluded that TR has both ligand-independent and ligand-dependent functions in cholesterol metabolism (13,37). Given this knowledge, analysis of cholesterol metabolism in animals lacking TR may lead to limited or perhaps erroneous conclusions. In this study, therefore, the TR-␤ allele was mutated in mice so that the resulting receptor was unable to bind to T 3 . This mutation (⌬337T) was found in a family with thyroid hormone resistance and acts as a dominant negative receptor in tissues (15). Because it constitutively binds to corepressor molecules, it closely mimics the TR in the hypothyroid state. Paradoxically, we found that animals bearing this mutation were protected against hypercholesterolemia induced by a high cholesterol diet because of increased hepatic cholesterol metabolism (Fig. 2).
To determine the mechanism of this effect, we chose to examine three major cholesterol metabolism-related genes whose gene expressions are regulated by thyroid hormone: CYP7A1, HMGC-R, and LDL-R. We confirmed that the mutant TR functioned as a dominant negative in the liver by first measuring 5ЈDI expression (Fig. 3). A progressive reduction in T 3 -stimulated 5ЈDI mRNA expression activity was found comparing WT to WT/MUT and MUT/MUT animals. Next, thyroid hormone is known to modulate serum cholesterol levels principally by regulating CYP7A1 mRNA levels (29,30). CYP7A1 mRNA levels were undetectable in the hypothyroid state in all mice. T 3 treatment induced CYP7A1 mRNA levels in WT mice, whereas a high cholesterol diet had minimal effect on CYP7A1 mRNA expression (Fig. 3). In contrast, a high cholesterol diet markedly induced CYP7A1 levels in MUT mice, and T 3 treatment had a minimal effect (Fig. 3). Gullberg et al. (14) also reported that hypothyroidism reduced CYP7A1 mRNA levels and activity in TR-␤ knock-out mice (14). The degree of suppression of CYP7A1 mRNA levels during hypothyroidism in the TR-␤ knock-out animals (ϳ70%) appears to be less than that observed in our MUT animals, which indicates that the mutant TR-␤ functioned as a TR-␤, and RXR␣ constructs were subjected to a coupled in vitro transcription/translation reaction in reticulocyte lysate using T7 polymerase. Proteins were incubated with a radiolabeled DRϩ4 probe and analyzed by electrophoretic mobility shift assays. LXR␣/RXR␣ heterodimerization is increased in the presence of the ⌬337T as compared with the presence of wild-type TR-␤. Four l of TR-␤, ⌬337T, or LXR-␣, and 2 l of RXR␣ were used. T 3 ϭ 100 nM. The consensus DRϩ4 probe is shown but equivalent results were obtained with the mouse CYP7A1 probe. Moreover, neither LXR-␣ nor TR-␤ complexes bound to the mutant CYP7A1 probe. D, ChIP assay of liver extract from all three animal groups using the indicated antibodies. The proximal promoter of the CYP7A1 gene was amplified using the indicated primers. dominant negative inhibitor on the CYP7A1 gene in our studies. In summary, our results support a dominant negative effect of the mutant TR-␤ on CYP7A1 gene expression.
On the other hand, HMGC-R and LDL-R mRNA levels, which are also positively regulated by thyroid hormone, were not reduced in MUT mice in the basal state when compared with WT mice. In fact, the mutant TR-␤ functioned as a ligand-independent activator on these genes (Fig. 3). In all animals, HMGC-R and LDL-R gene expression was inhibited by a high cholesterol diet after T 3 replacement, although expression from both genes was significantly higher after T 3 replacement in MUT versus WT mice (Fig. 3). These results point to differences in regulation of these gene products versus CYP7A1. Thus, the dominant negative effect of the mutant TR-␤ is variable and may depend on the nature of the regulatory pathway.
Clearly, hepatic cholesterol levels were reduced in MUT animals after treatment with a high cholesterol diet or after treatment with both T 3 and a high cholesterol diet (Fig. 2B). The most likely explanation for the decreased hepatic cholesterol levels is an increase in CYP7A1 activity. The increase in bile acid pool size in the MUT animals is consistent with this hypothesis (Fig. 2D). LXR-␣ is essential to CYP7A1 gene expression (33), and LXR-␣ and TR-␤ share the same DNA binding element, which is a DRϩ4 (33,38). The CYP7A1 promoter contains a DRϩ4 element and co-transfection of LXR-␣ activity mediated a response to 22(R)-OH-cholesterol (Fig. 4B), which was not observed in the absence of co-transfected receptor (data not shown). Interestingly, co-transfection of LXR-␣ and WT TR-␤, but not MUT TR-␤, antagonized LXR-␣ activity on the CYP7A1 promoter (Fig. 4B).
To determine the mechanism of this phenomenon, gel mobility shift assays were performed. As expected and as reported previously, the WT TR-␤ formed both TR homodimers and TR/RXR heterodimers and the former was dissociated by T 3 (Fig. 4C). In contrast, the mutant TR-␤ homodimer was not dissociated by T 3 as previously reported (Fig. 4C,  Ref. 15). LXR/RXR heterodimers were readily detected on the DRϩ4 element (Fig. 4C) regardless of the presence of its ligand (22(R)-OHcholesterol, data not shown). However, WT TR/RXR heterodimers were preferred over LXR/RXR heterodimers on the DRϩ4 element when tested in competition (Fig. 4C). In contrast, MUT TR-␤ was a much less effective competitor than WT TR-␤ for the heterodimerization with RXR-␣ (Fig. 4C) and LXR/RXR heterodimers were observed in this competition. Consistent with this finding, ChIP assay of liver tissue from WT mice indicates that the LXRE in the CYP7A1 promoter is primarily occupied by TR/RXR heterodimers, whereas the same element in MUT/MUT mice is primarily occupied by LXR/RXR heterodimers (Fig. 4D).
Based on these in vitro data, we offer an explanation for the mechanism of the marked increase in the CYP7A1 mRNA expression in MUT mice fed a high cholesterol diet. In these animals, LXR/RXR heterodimers would be more likely to bind to DRϩ4 elements making their target genes more responsive to cholesterol. Thus, at least two types of molecular mechanisms are involved in the dominant negative effect by MUT TR-␤: 1) thyroid hormone action is impaired by the mutant TR-␤ itself such as found on the 5ЈDI gene; and 2) MUT TR-␤ affects the function of other nuclear hormone receptors (LXR-␣) such as found on the CYP7A1 gene. Furthermore, our in vivo data reinforces the in vitro studies of Kawai et al. (39), who suggested that TR and RXR bind to certain DRϩ4 elements in common, and showed that unliganded TR represses LXR activation on LXREs derived from the dMTV-LTR and SREBP-1c promoters.
Thus, our data demonstrate a cross-talk between LXR-␣ and TR-␤ in vivo, which was uncovered through expression of a MUT TR-␤. It may not be possible to extend this finding beyond rodents given that the human CYP7A1 gene is reported to lack this proximal LXRE and be unresponsive to cholesterol-mediated stimulation (32,40). However, studies designed to evaluate human gene expression in a rodent background have limitations, which include the number and location of the human transgene in the mouse genome, the presence of the regulatory element at distances far from the human gene that are not included within the transgene, and unknown species-specific effects that could be critical for normal regulation. Thus it remains possible that the endogenous human gene is subjected to cross-talk regulation by LXR-␣ and TR-␤.
In the present study, we also found that the role of unliganded TR-␤ depends on each individual gene promoter. Dominant negative activity was observed on some but not all gene products. Moreover, it was revealed that the unliganded TR-␤ had considerable effects on cholesterol metabolism via a cross-talk between TR-␤ and LXR-␣ and this likely plays a pivotal role in the activation of CYP7A1 gene expression.