Activation of the nuclear receptor LXR by oxysterols defines a new hormone response pathway.

Accumulation of cholesterol causes both repression of genes controlling cholesterol biosynthesis and cellular uptake and induction of cholesterol 7α-hydroxylase, which leads to the removal of cholesterol by increased metabolism to bile acids. Here, we report that LXRα and LXRβ, two orphan members of the nuclear receptor superfamily, are activated by 24(S),25-epoxycholesterol and 24(S)-hydroxycholesterol at physiologic concentrations. In addition, we have identified an LXR response element in the promoter region of the rat cholesterol 7α-hydroxylase gene. Our data provide evidence for a new hormonal signaling pathway that activates transcription in response to oxysterols and suggest that LXRs play a critical role in the regulation of cholesterol homeostasis.

Cholesterol (CH) 1 is a major structural constituent of cellular membranes and serves as the biosynthetic precursor for bile acids and steroid hormones. Animal cells can obtain CH endogenously through de novo synthesis from acetyl-CoA or exogenously through receptor-mediated endocytosis of low density lipoproteins. Cells must balance the internal and external sources of CH so as to maintain mevalonate biosynthesis while at the same time avoiding the accumulation of excess CH, which can result in diseases such as atherosclerosis, gallstones, and several lipid storage disorders (1).
CH homeostasis is maintained in part through feedback regulation of the low density lipoprotein receptor gene and at least two genes encoding enzymes in the CH biosynthetic pathway, 3-hydroxy-3-methylglutaryl coenzyme A synthase and 3-hydroxy-3-methylglutaryl coenzyme A reductase (1). Although increases in dietary CH lead to the inhibition of expression of these genes in vivo, it remains unclear whether CH or CH metabolites are responsible for this inhibition (2). Experiments performed in vitro using several different cell lines have indicated that derivatives of CH that are oxygenated on the CH side chain are significantly more potent in the suppression of sterol biosynthesis than CH (3). These oxysterols are produced through the actions of P450 enzymes in various metabolic pathways including bile acid synthesis in the liver and sex hormone synthesis in the adrenal glands. The in vitro activities of oxysterols together with their presence in vivo suggests that oxysterols may serve in metabolic feedback loops to regulate CH homeostasis.
Although CH and its oxysterol metabolites can repress gene transcription, in at least one instance dietary CH has been shown to stimulate gene expression. Expression of the cholesterol 7␣-hydroxylase (CYP7A) gene, which encodes the enzyme responsible for the initial and rate-limiting step in the conversion of CH to bile acids (4), is up-regulated in rats fed a CH-rich diet (5)(6)(7). This stimulatory effect provides a regulatory mechanism whereby excess dietary CH can be converted to more polar bile acids for subsequent removal from the body. Although the molecular mechanism is unknown, induction of CYP7A expression in the presence of CH occurs at the level of gene transcription (8,9).
A variety of CH derivatives, including steroid hormones and vitamin D, exert effects on gene expression through interactions with members of the nuclear receptor superfamily (10). Members of this family function as ligand-activated transcription factors by binding to short stretches of DNA, termed hormone response elements, present in the regulatory regions of target genes. In addition to the nuclear receptors with known ligands, this superfamily includes a large number of structurally related members that contain DNA binding domains and putative ligand binding domains but lack identified ligands, the so-called "orphan receptors." In this report, we have identified a binding site for the orphan nuclear receptor LXR␣ in the proximal promoter region of the rat CYP7A gene. LXR␣ and the closely related orphan receptor LXR␤ are broadly expressed and bind to DNA as heterodimers with the retinoid X receptors (RXRs) (11)(12)(13)(14). As part of an effort to identify natural LXR ligands, we have found that the oxysterols 24(S),25-epoxycholesterol and 24(S)-hydroxycholesterol activate both LXR␣ and LXR␤ at concentrations consistent with those found in tissue extracts. Our results provide evidence for a novel oxysterol signaling pathway regulating hepatic CH homeostasis that may also be important in mediating other cellular and developmental effects of CH.
Plasmids-To generate the plasmids pSG5-mLXR␣, pSG5-hLXR␤, and pSG5-hRXR␣, the cDNAs encoding the murine LXR␣, human LXR␤, or human RXR␣ were inserted into the expression vector pSG5 (Stratagene). The GAL4-LXR constructs contain in the pSG5 expression vector the translation initiation sequence and amino acids 1-76 of the glucocorticoid receptor fused to the DNA-binding domain of the yeast * This research was supported in part by Grant HL52069 from the National Institutes of Health (to T. A. S.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Cotransfection Assay-CV-1 cells were plated in 24-well plates in Dulbecco's modified Eagle's medium supplemented with 10% charcoalstripped fetal calf serum at a density of 1.2 ϫ 10 5 cells/well. In general, transfection mixes contained 33 ng of receptor expression vector, 100 ng of reporter plasmid, 200 ng of ␤-galactosidase expression vector (pCH110, Pharmacia Biotech Inc.), and 166 ng of carrier plasmid. Cells were transfected overnight by lipofection using Lipofectamine (Life Technologies Inc.) according to the manufacturer's instructions. The medium was changed to Dulbecco's modified Eagle's medium supplemented with 10% delipidated calf serum (Sigma), and compound was added for 24 h. Cell extracts were prepared and assayed for chloramphenicol acetyltransferase (CAT) and ␤-galactosidase activities as described previously (30).
For the antibody supershift assays, a pool of monoclonal antibodies was used. The antibodies were generated by fusing spleenocytes from an SJL mouse (male, 8 weeks old, Jackson, Bar Harbor, ME) immunized with histidine-tagged recombinant human LXR␣ protein (amino acids 164 -447) with mouse myeloma cells as previously described (32).

RESULTS AND DISCUSSION
The CYP7A Promotor Contains an LXR-RXR Heterodimer Binding Site-Previous work had identified a region of the CYP7A promoter that interacts with nuclear proteins and contains a motif that closely resembles known binding sites for members of the nuclear receptor family (15,16). This putative response element is composed of a nearly perfect tandem repeat of the nuclear receptor half-site recognition sequence AG(G/T)TCA separated by four nucleotides, a so-called DR-4 motif (17). This type of DR-4 response element has been shown to function as a high affinity binding site for heterodimers between RXR and the orphan nuclear receptors LXR␣ and LXR␤ (11)(12)(13)(14).
To investigate whether the DR-4 motif in the CYP7A promoter could function as an LXR response element, we performed a series of gel retardation assays. First, to verify that the in vitro synthesized receptor proteins were functional, we used an oligonucleotide in which two nucleotides in the 5Ј half-site of the CYP7A direct repeat sequence were mutated to obtain an idealized, tandem repeat of the AGTTCA motif (Fig.  1A, DR4-LXRE). Consistent with previous findings (11-14), a strong complex was formed when either LXR␣ or LXR␤ was added together with RXR, indicating that high affinity DNA binding required the formation of LXR-RXR heterodimers (Fig.  1B, lane 3). Interestingly, a weaker complex migrating slightly faster than the LXR-RXR heterodimer was observed when LXR␣ was used in the absence of RXR (Fig. 1B, lane 1), suggesting the weak binding of an LXR homodimeric complex. In experiments performed with an oligonucleotide containing the CYP7A DR-4 element (Fig. 1A), strong binding was observed with both the LXR␣-RXR and LXR␤-RXR heterodimers (Fig.  1B, lane 6). This binding was sequence-specific because an excess of either the unlabeled CYP7 oligonucleotide or the idealized DR-4-LXRE oligonucleotide (Fig. 1B, lanes 7-10) competed efficiently for binding to the probe, but no competition for binding was seen with an oligonucleotide in which the 3Ј halfsite had been mutated to a consensus glucocorticoid receptor half-site (mCYP7-LXRE) (Fig. 1, A and B, lanes 11 and 12). As observed with the DR-4-LXRE, a weak and slightly faster migrating complex of a potential LXR␣ homodimer was seen when LXR␣ was added alone to the CYP7A direct repeat element (Fig. 1B, lane 4). Based on these data, we refer to the DR-4 of the CYP7A promoter as the CYP7-LXRE.
LXR␣ is abundantly expressed in the liver (11,12), the site of CYP7A expression. To determine whether endogenously expressed LXR␣ bound to the CYP7-LXRE, we prepared nuclear extracts from rat liver for use in gel retardation assays. A strong, shifted complex was observed in assays performed with the CYP7-LXRE and liver extracts (Fig. 2B, lane 3) that migrated at the same position as the one obtained with in vitro synthesized LXR and RXR protein (Fig. 2B, lane 1). A portion of this complex was supershifted (Fig. 2B, lane 4) upon the addition of a pool of monoclonal antibodies that specifically recognizes the LXR␣ but not the LXR␤ ligand binding domain ( Fig. 2A). An analogous supershifted complex was seen in experiments performed with these antibodies and in vitro syn- thesized LXR␣ and RXR (Fig. 2B, lane 2). These data show that LXR␣ is one of the components present in liver extracts that binds to the CYP7-LXRE.
Oxysterols Activate LXR␣ and LXR␤-The presence of a binding site for LXR in the promoter of the CYP7A gene, which encodes the rate-limiting enzyme responsible for the conversion of CH to bile acids, prompted us to test a comprehensive set of CH precursors and metabolites including bile acids, oxysterols, and steroids for their ability to activate LXR␣ or LXR␤ in cotransfection experiments (Fig. 3A). The compounds were initially tested using chimeric receptor proteins containing the ligand binding domains of either LXR␣ or LXR␤ fused to the DNA binding domain of the yeast transcription factor GAL4 (18). The use of chimeric receptors allowed us to avoid complications in the interpretation of the results caused by endogenous nuclear receptors as well as to identify compounds that mediated their effects through the ligand binding domains of these orphan receptors. Expression plasmids encoding the chimeric receptors were cotransfected into CV-1 cells together with a reporter plasmid containing five copies of the GAL4binding site upstream of the minimal tk promoter driving the expression of CAT. All compounds were tested on the chimeric receptors at a concentration of 10 M.
As shown in Fig. 3A, neither of the LXR chimeras was activated in response to CH or its precursors lanosterol and desmosterol. Interestingly, however, significant activation of both the LXR␣ and LXR␤ chimeras was observed in transfected cells treated with the oxysterols 20(S)-, 22(R)-, and 24(S)-hydroxycholesterol (Fig. 3A). 25-and 27-hydroxycholesterol had relatively little or no effect on either LXR chimera (Fig. 3A). Notably, activation of the LXRs was more efficient with the naturally occurring 22(R)-and 24(S)-hydroxycholesterol enantiomers than with the synthetic 22(S)-and 24(R)-hydroxycholesterol stereoisomers. We note that while this manuscript was in preparation, several of these oxysterols were shown to activate LXR␣ (19). 24-Keto-cholesterol was also an efficacious activator of both LXRs (Fig. 3A). In contrast, no activation of either LXR␣ or LXR␤ was seen in the presence of additional, downstream intermediates in the steroid hormone biosynthetic pathway or in the presence of testosterone or progesterone (Fig.  3A). 7␣-Hydroxycholesterol, its bile acid derivatives chenodeoxycholic acid and cholic acid, or bile salts also failed to activate the LXR chimeras (data not shown). In addition, steroids with the 3␣-hydroxyl configuration were inactive (data not shown).
In addition to the various hydroxycholesterol derivatives, we also examined the activity of 24(S),25-epoxycholesterol (Fig.   3B). 24(S),25-Epoxycholesterol is derived from a shunt in the mevalonate pathway in which squalene epoxide, rather than undergoing cyclization to lanosterol, is instead metabolized to 24(S),25-epoxycholesterol (20). 24(S),25-epoxycholesterol has been identified in cultured cells (21) and in tissue homogenates prepared from human and rat livers (22,23,28). Because 24(S),25-epoxycholesterol is derived from a biosynthetic pathway different from that of any of the other oxysterols, it has been proposed that this epoxide might serve as a signaling molecule in the feedback regulation of CH biosynthesis (23). Indeed, we found that 24(S),25-epoxycholesterol was an efficacious activator of both LXRs (Fig. 3A). Taken together, the results of these transfecton studies indicate that the optimal structural requirement for LXR activation is a 3␤-sterol with an oxygen substituent at C-24 capable of forming a hydrogen bond.

24(S),25-Epoxycholesterol and 24(S)-Hydroxycholesterol
Are Potent Activators of LXR␣ and LXR␤-We next wished to determine whether the potencies of these oxysterols in the activation of the LXRs were consistent with their known abundance in vivo. Three of these compounds are particularly abundant in tissue extracts. 24(S)-hydroxycholesterol is present in brain extracts at concentrations of roughly 100 M. Although its physiological relevance is not known, these high concentrations have led to 24(S)-hydroxycholesterol being termed "cerebrosterol." Although circulating levels of 24(S)hydroxycholesterol are low in adults (ϳ50 nM), this concentration can be elevated as much as 10-fold in children (24). 24(S),25-Epoxycholesterol has been isolated from human or rat livers at concentrations estimated to be 1-5 M (22, 23, 28). Finally, 22(R)-hydroxycholesterol is present in adrenal extracts at micromolar concentrations. Other oxysterols are present in the circulation or in tissue extracts at concentrations that are nanomolar or lower (25).
We performed dose response analysis using expression plasmids for full-length LXR␣ or LXR␤. Reporter constructs were generated that contained four copies of either the CYP7-LXRE or the idealized DR-4-LXRE driving CAT gene expression. In preliminary experiments performed with 10 M 24(S),25-epoxycholesterol, we found that LXR␣ induced expression of both the CYP7A-LXRE-CAT and DR-4-LXRE-CAT reporter constructs but that LXR␤ only efficiently induced expression of the DR-4-LXRE-CAT reporter construct (data not shown). We speculate that LXR␤ may not bind to the CYP7A-LXRE with sufficient affinity to activate reporter expression. Subsequent dose response analyses with LXR␣ and LXR␤ were performed with the CYP7A-LXRE-CAT and DR-4-LXRE-CAT reporter constructs, respectively. LXR␣ and LXR␤ responded to 24(S),25epoxycholesterol with EC 50 values of 7.5 and 1.5 M, respectively (Fig. 4A). 24(R),25-Epoxycholesterol, which has not been found to occur naturally, was significantly less active with an EC greater than 10 M on both LXRs. 24(S)-Hydroxycholesterol had an activation profile similar to that of 24(S),25-epoxycholesterol on the LXRs, activating LXR␣ and LXR␤ with EC 50 values of 7 and 1.5 M, respectively (Fig. 4B). The synthetic isomer 24(R)-hydroxycholesterol was at least 1 order of magnitude less active. 20(R)-Hydroxycholesterol, 22(R)-hydroxycholesterol, and 24-ketocholesterol displayed EC 50 values for LXR activation that were greater than 10 M (data not shown).
Based upon these data, we suggest that 24(S),25-epoxycholesterol, which is abundant in liver, and 24(S)-hydroxycholesterol, which is abundant in brain, may function as endogenous activators of LXR␣ and LXR␤. We note that both LXRs are expressed in the liver. Interestingly, Northern analyses have demonstrated that LXR␤ is expressed in the brain (14). More detailed in situ studies have shown that LXR␤ is broadly expressed in fetal brain and that its expression becomes more restricted in postnatal and adult brains (26). Taken together, these findings support the suggestion that 24(S)-hydroxycholesterol may serve in the development and function of the brain via interactions with LXR␤ (22).
In summary, we have demonstrated that the LXRs are activated by the oxysterols 24(S),25-epoxycholesterol and 24(S)hydroxycholesterol at concentrations consistent with those found in tissues. It is well known that oxysterols suppress de novo CH biosynthesis as well as cellular uptake of CH (1). However, oxysterols have a number of other biological effects including marked effects on DNA synthesis and cell growth and proliferation (25). Our data suggest that some of these effects may be mediated through the LXRs. It is interesting that the nuclear receptor family member most closely related to the LXRs is the insect ecdysone receptor (12,13,15). Ecdysone is a sterol hydroxylated at the 22 and 25 positions that functions as a temporal signal to coordinate tissue-specific morphogenetic changes during insect development (27). The use of nuclear receptor pathways as a means to transduce the effects of oxygenated sterols thus appears to have been conserved throughout a large part of evolution, from insects to man.