A Natural Product Ligand of the Oxysterol Receptor, Liver X Receptor

Natural products are proven sources of biologically active molecules that have played a critical role in pharmacology. Many of these natural products have been demonstrated to be useful for medicinal purposes. A limited number of natural products appear to act as ligands of nuclear hormone receptors. One of the most well characterized natural product nuclear receptor ligands is genestein, a product of soy, which binds directly to estrogen receptors (NR3A1 and NR3A2) and is used as a dietary supplement to alleviate menopausal symptoms in women (Kuiper et al., 1998; Tham et al., 1998). Interestingly, genestein has been recently shown to also function as a low-affinity ligand for yet another nuclear receptor, peroxisome proliferator-activated receptor γ (NR1C3) (Dang et al., 2003). A related isoflavone soy product also used in dietary supplements, daidzein, is another estrogen receptor agonist, albeit weaker than genestein (Wiseman, 2000). The active compound within St. John's wort, an herbal remedy for depression, is a potent agonist of the xenobiotic nuclear receptor, pregnane X receptor (PXR, NR1I2) (Moore et al., 2000; Watkins et al., 2003). Although the antidepressent activity of St. John's wort does not appear to be mediated by its PXR activity, the induction of cytochrome P450 (CYP) 3A4 is mediated by activation of this receptor potentially leading to a significant increase in the metabolism of a variety of drugs taken by individuals using this supplement (Moore et al., 2000). The most recent report of a natural product ligand of a nuclear receptor is the discovery that guggulsterone, found in the resin of the guggul tree, is an efficacious antagonist of the farnesoid X receptor (FXR, NR1H4), the bile acid receptor (Urizar et al., 2002; Wu et al., 2002). Guggul tree extract has been suggested to lower low-density lipoprotein levels in animal models, has been successfully used in Ayurveda medicine to treat obesity and lipid disorders, and a modern antihyperlipoproteinemic drug based on the actions of guggulsterone is marketed in India (Satyavati, 1988; Singh et al., 1990; Dev, 1997). Beyond the use of natural products directly as dietary supplements, identification of novel natural product ligands for a given receptor often provides pharmacological tools and unique insight into drug design.

The liver X receptors [LXRα (NR1H3) and LXRβ (NR1H2)] are additional members of the nuclear receptor superfamily and were originally identified as orphan receptors (Shinar et al., 1994; Song et al., 1994; Seol et al., 1995; Teboul et al., 1995; Willy et al., 1995). Subsequently, oxysterol cholesterol metabolites were demonstrated to be physiological ligands for LXR (Janowski et al., 1996). These two receptors play a key role in the regulation of cholesterol metabolism and transport as well as glucose metabolism and inflammation (Repa and Mangelsdorf, 2002; Cao et al., 2003; Joseph et al., 2003). Modulation of the activity of these receptors may be useful in the treatment of a number of pathophysiological states including dyslipidemia, atherosclerosis, and diabetes (Joseph et al., 2002; Tangirala et al., 2002; Cao et al., 2003).

In this study, we describe the discovery of the first nonoxysterol natural product ligand of LXR. The indole alkoid fungal metabolite from Penicillium paxilli, paxilline, functions as a ligand for both LXRα and LXRβ. Paxilline binds directly to both receptors leading to recruitment of coactivators and activates the receptors in a cell-based context. In addition, paxilline efficaciously induces the expression of LXR target genes, ABCA1 and SREBP. Our data confirm that natural product libraries are a rich source of ligands for nuclear receptors and may provide pharmacological agents for investigation of the function of these receptors as well as potential drugs.

Materials and Methods

LXR Ligands. 22(R)-Hydroxycholesterol and paxilline were obtained from Sigma-Aldrich (St. Louis, MO).

Radioligand Binding Assay. The LXR radioligand binding assay was performed using scintillation proximity technology as previously described (Thomas et al., 2003). We used 800 ng of baculovirus-expressed, His-tagged LXRα-LBD protein (amino acids 162–447) or 600 ng of LXRβ-LBD protein (amino acids 202–461), 25 nM ³H-labeled 25-hydroxycholesterol (Amersham Biosciences, Inc., Piscataway, NJ), 0.05 mg of yttrium silicate polylysine-coated SPA beads (Amersham Biosciences, Inc.), and varying concentrations of competitor per well of a 96-well OptiPlate (PerkinElmer Life Sciences, Boston, MA). Protein, radioligand, and competitor were added to the plate. SPA beads were then added to the assay plate followed by 10 min of gentle shaking at room temperature and protected from light. The plates were incubated in the dark at room temperature for 2 h before reading in a TopCount plate reader (PerkinElmer Life Sciences).

Coactivator Interaction Assay. Interaction between LXRα/LXRβ and the coactivators SRC-1 or TIF-2 were assayed using AlphaScreen (amplified luminescent proximity homogenous assay) technology (PerkinElmer Life Sciences). The assay was performed in white, low volume, 384-well plates using a final volume of 15 μl containing final concentrations of 20 nM His-tagged baculovirus expressed LXRα-LBD or LXRβ-LBD protein, 5 nM GST-TIF-2 or GST-SRC-1 protein that contained the entire nuclear receptor interacting domain of the coactivator protein fused to GST and 10 μg/ml of both Ni²⁺ chelate donor beads and anti-GST acceptor beads (PerkinElmer Life Sciences). The assay buffer contained 25 mM HEPES (pH 7.0), 100 mM NaCl, 0.1% bovine serum albumin, and 2 mM dithiothreitol. All manipulations involving assay beads were done in ambient light. Assay plates were covered with a clear seal and incubated in the dark for 2 h after which the plates were read for 1 s per well in a PerkinElmer fusion microplate analyzer using the manufacturers standard AlphaScreen detection protocol.

Cell Culture and Transfections. HEK293 cells were cultured in 3:1 DMEM/F-12 containing 10% fetal bovine serum and supplemented with 1% penicillin and streptomycin, 1% l-glutamine, and 20 mM HEPES. Cells were seeded 48 h before transfection at 6 × 10⁶ cells/T225 flask in 30 ml of growth medium. Cells were transfected with Fugene transfection reagent (Roche Diagnostics, Indianapolis, IN) according to the Fugene protocol with 330 ng of pcDNA3-hRXR-α, 33 ng of pCMV6 LXRα or LXRβ, 660 ng of pGL3B-E1b-3XLXRE luc (Thomas et al., 2003), and 10 μl of Fugene per 10⁶ cells. Growth medium was replaced during transfection with 3:1 DMEM/F-12 containing 10% charcoal/dextran-treated, heat-inactivated fetal bovine serum and supplemented with 1% penicillin and streptomycin, 1% l-glutamine, and 20 mM HEPES. After 24 h, cells were harvested and plated into 96-well white plates at 50,000 cells per well in 90 μl of complete transfection medium, allowed to attach for 2 h, then treated with 10 μlof10× compound and dimethyl sulfoxide controls. After 24 h, cells were lysed and assayed for luciferase activity.

ABCA1 mRNA Quantitation. ABCA1 mRNA expression was measured in THP-1 macrophage cells using a bDNA assay (Quanti-Gene high volume kit; Bayer Corp.-Diagnostics Div., Tarrytown, NY). THP-1 cells were grown in suspension at 37°C 5%/95% CO₂/air incubator in growth medium (RPMI 1640 medium containing 0.05 mM 2-mercaptoethanol and 10% fetal bovine serum at a density of 250,000 cells per milliliter and allowed to reach a density of 1 million cells per milliliter. Growth medium was then changed to growth medium containing 10 nM phorphol 12-myristate 13-acetate and cells were plated in 96-well dishes at a density of 100,000 cells per well. After an overnight incubation, medium was changed to a growth medium containing 10% lipoprotein depleted fetal bovine serum. Cells were treated with various concentrations of compounds serially diluted to obtain a 10-point concentration curve from a final concentration of 20 to 0.001 μM. After a 24-h incubation with compound, cells were lysed using 50 μl/well bDNA assay kit lysis reagent. The kit reagents as well as ABCA1-specific primer sets were used to process the samples for the bDNA assay as previously described (Zhang et al., 2002). After a 15-min incubation at 37°C, 100 μl of the lysis buffer from each well were transferred to the corresponding wells of the capture plate. The capture plate was incubated overnight at 53°C. The capture plate was then washed twice with QuantiGene wash buffer followed by the addition of 100 μl/well QuantiGene amplifier working reagent. The plate was incubated for 60 min at 46°C followed by two washes. The mRNA to be measured was then labeled by the addition of 100 μl of QuantiGene label probe working buffer followed by a 60-min incubation at 46°C. The capture plate was then washed twice followed by the addition of 100 μl/well QuantiGene substrate plus QuantiGene enhancer reagent. The plates were incubated at 37°C for up to 30 min and then read on a luminometer to detect the luminescent signal. The induction of ABCA1 mRNA expression was calculated as a ratio of compound-treated luminescent levels compared with untreated control levels.

SRE Assay and SREBP mRNA Quantitation. As previously described (Thomas et al., 2003), HepG2 cells stably transfected with a 3XSRE thymidine kinase luciferase reporter construct were treated for 24 h with an LXR ligand to assess SREBP activity. SREBP mRNA was quantitated by Taqman real-time polymerase chain reaction as previously described (Thomas et al., 2003).

Data Analysis. Dose responses and displacement curves were analyzed in GraphPad Prism (GraphPad Software Inc., San Diego, CA) allowing calculation of both EC₅₀ and K_i values. Each point of data represents minimal triplicate wells, and the results shown are representative of at least three independent experiments.

Results

A screen of a natural product library for compounds with the ability to modulate the activity of LXR yielded the identification of the first nonoxysterol natural product ligand for this receptor, paxilline. Paxilline is an indole alkoid metabolite from the fungus, P. paxilli. The structure of paxilline is shown in Fig. 1 and compared with several LXR ligands including a natural ligand, 22(R)-hydroxycholesterol, as well as synthetic ligands GW3965, T0901317, and APD (Repa et al., 2000; Schultz et al., 2000; Collins et al., 2002; Sparrow et al., 2002).

We examined the ability of paxilline to bind directly to both LXRα and LXRβ using a radioligand binding assay. A scintillation proximity assay format was employed using tritiated 25-hydroxycholesterol as the radioligand. As illustrated in Fig. 2, paxilline displaced 25-hydroxycholesterol from both receptors (LXRα K_i = 660 nM; LXRβ K_i = 1,100 nM). In contrast a natural oxysterol ligand, 22(R)-hydroxycholesterol [22(R)-OHC] displayed higher affinity in this assay (LXRα K_i = 250 nM; LXRβ K_i = 490 nM) (Fig. 2). Paxilline did not bind and/or activate any other nuclear receptor examined (IC₅₀ and/or EC₅₀ >10 μM for ERα, ERβ, TR, RXR, FXR) (data not shown). Paxilline also induced LXR recruitment of both coactivators SRC-1 or TIF-2 in a cell-free AlphaScreen assay system (Fig. 3). Using purified recombinant LXRα or LXRβ and purified GST-SRC-1 or GST-TIF-2, we demonstrated that increasing amounts of paxilline resulted in dose-dependent recruitment of these coactivators to both LXRs. The EC₅₀ for paxilline-mediated LXRα recruitment of SRC-1 was 1,800 nM [22(R)-OHC = 2,600 nM] whereas TIF-2 was 660 nM [22(R)-OHC = 1,400 nM]. The EC₅₀ for paxilline-mediated LXRβ recruitment of SRC-1 was 930 nM [22(R)-OHC = 300 nM] whereas TIF-2 was 1,200 nM [22(R)-OHC = 780 nM]. The ability of paxilline to induce coactivator recruitment by LXR suggested that paxilline might function as an agonist. This was confirmed in a transfection experiment in which HEK293 cells were cotransfected with either LXRα or LXRβ and a reporter containing three copies of a DR4 element derived from the ABCA1 promoter (Fig. 4). Paxilline activated transcription of both LXRα and LXRβ with equivalent potency and efficacy as 22(R)-OHC (EC₅₀ ∼ 4,000 nM for both receptors and ligands). The apparent discrepancy between the affinity of 22(R)-OHC for LXR in the binding assay and the potency in a cotransfection assay has been previously described and is apparently a function of the physiochemical properties of this compound (Janowski et al., 1999). Interestingly, paxilline functions as a partial agonist in the coactivator interaction assay [60–90% 22(R)-OHC] but as a full agonist in the cotransfection assays. This suggests that either additional coactivators are recruited in the cell-based assay that allow for retention of full efficacy or that only a threshold of coactivator recruitment is required to reach full agonism. Consistent with its function as an agonist in both the coactivator recruitment and cotransfection assay, paxilline induced the expression of ABCA1 in THP-1 cells very efficaciously with a maximal induction of approximately 7-fold (EC₅₀ = 1,300 nM) (Fig. 5). In this paradigm, the dose response for 22(R)-OHC was limited by toxicity above 10 μM; however, the potency is greater than 5,000 nM and the maximal efficacy at 10 μM is approximately 6-fold. Paxilline also increased the expression of a second LXR target gene, SREBP, in HepG2 cells (Fig. 6A). We used a SREBP-responsive luciferase reporter stably transfected into HepG2 cells (Thomas et al., 2003) to investigate the dose-responsiveness of paxilline induction of SREBP expression. As illustrated in Fig. 6B, paxilline efficiently increased transcription from the SRE reporter with an EC₅₀ of 2,800 nM. These data demonstrate that paxilline binds directly to both LXRα and LXRβ, functions as an agonist with the ability to mediate recruitment of coactivators to the receptors, and activates transcription of a reporter gene as well as two natural LXR target genes.

Discussion

Since the identification of LXRs as orphan members of the nuclear hormone receptor superfamily in the mid-1990s, several ligands both natural and synthetic have been identified. Endogenous oxysterols were the first ligands identified for LXR, which suggested a role for this receptor in regulation of cholesterol homeostasis via action as a “cholesterol sensor” (Janowski et al., 1996). The first target gene identified for LXR, cholesterol 7α-hydroxylase (CYP7A), was consistent with this suggestion, illustrating a role for this receptor in regulation of the rate-limiting step in the conversion of cholesterol to bile acids (Lehmann et al., 1997), which was later confirmed in the LXR null mouse (Peet et al., 1998). In addition to its function in regulation of CYP7A, which appears to be rodent-specific, an array of additional target genes have been identified that establish LXR as a key regulator of cholesterol homeostasis. These genes include sterol transporters such as ABCA1, ABCG1, ABCG5, and ABCG8, as well as other genes demonstrated to be critical in lipid metabolism including apoE, apoCII, lipoprotein lipase, phospholipid transfer protein, and cholesterol ester transfer protein (Luo and Tall, 2000; Repa et al., 2000, 2002; Venkateswaran et al., 2000; Laffitte et al., 2001; Zhang et al., 2001; Cao et al., 2002; Mak et al., 2002). The manner in which LXR regulates these genes suggests that activation of this receptor may be antiatherogenic, which has been confirmed in mouse models (Joseph et al., 2002; Tangirala et al., 2002). The therapeutic potential of a LXR agonist has recently expanded by the demonstration that activation of this receptor results in both anti-inflammatory and antidiabetic activity by regulating an array of genes involved in either inflammatory processes or gluconeogenesis, respectively (Cao et al., 2003; Fowler et al., 2003; Joseph et al., 2003).

Development of selective, high-affinity ligands for LXR such as T1317 and GW3965 has proven to provide essential tools in characterization of the physiological and pathophysiological roles of LXR. In this study, we identified and characterized an additional ligand of LXR, which represents the first nonoxysterol natural product ligand for this receptor. Natural products have proven to be an abundant source of agents for pharmacological characterization of biomolecules as well as for medicinal purposes. Although natural product libraries have not been as profitable for identification of ligands for nuclear hormone receptors as they have been for other fields, such as ion channel pharmacology, key ligands have been identified in the past that target receptors such as ER, FXR, PXR, and peroxisome proliferator-activated receptor γ. We identified paxilline, an indole alkoid fungal metabolite from P. paxilli, as an efficacious LXR agonist. Paxilline binds directly to both LXRα and LXRβ resulting in coactivator recruitment and activation of LXR-dependent gene transcription. Paxilline exhibits similar potency and efficacy as the natural ligand, 22(R)-OHC, in cotransfection assays and in terms of induction of expression of a natural target gene, ABCA1. Additional pharmacological activities of paxilline precluded examination of modulation of LXR activity in vivo due to toxicity. Paxilline is a tremorgenic mycotoxin that is a well characterized antagonist of high conductance calcium-activated K channels (BK channel) with potencies in the range of 100 nM, which is clearly greater than the potencies we detected for the LXRs in the low single digit micromolar range (Knaus et al., 1994).

Given the limitations for evaluation of paxilline in vivo, this compound still provides an additional tool for pharmacological characterization of LXR. In addition, the novel chemical structure provides insight into the diversity of chemical structures that can recognize the ligand binding pockets of both LXRα and LXRβ leading to coactivator recruitment and transcriptional activation. Furthermore, identification of an additional natural product ligand for a nuclear hormone receptor indicates that natural product libraries may be a rich source for ligands of additional nuclear hormone receptors including the orphans.