|Home | About | Journals | Submit | Contact Us | Français|
Hops extracts are used to alleviate menopausal symptoms and as an alternative to hormone replacement therapy, but can produce potentially harmful drug-drug interactions. The nuclear xenobiotic receptor PXR is promiscuously activated by a range of structurally distinct chemicals. It has a key role in the transcriptional regulation of genes that encode xenobiotic metabolism enzymes. In this study, hops extracts are shown to induce the expression of numerous drug metabolism and excretion proteins. The β-bitter acid colupulone is demonstrated to be a bioactive component and direct activator of human PXR. The 2.8 Å resolution crystal structure of the ligand binding domain of human PXR in complex with colupulone was elucidated, and colupulone was observed to bind in a single orientation stabilized by both Van der Waals and hydrogen bonding contacts. The crystal structure also indicates that related α- and β-bitter acids have the capacity to serve as PXR agonists, as well. Taken together, these results reveal the structural basis for drug-drug interactions mediated by colupulone and related constituents of hops extracts.
While more than 1,500 botanical-derived products are currently available in the U.S., herbal formulations are not subject to FDA approval, and there is often a lack of clinical data regarding efficacy and potential side effects (Burka, 2003). The flowers of the hops plant (Humulus lupulus) were historically used as a preservative and flavoring agent in beer. Hops extracts are currently marketed as a source of phytoestrogens to alleviate menopausal symptoms and as an alternative to hormone replacement therapy (Bowe et al., 2006; Milligan et al., 2002; Milligan et al., 1999; Milligan et al., 2000), and have also been applied to treating insomnia and anxiety (Hoffmann, 2003). In addition to plant fibrous material and proteins, hops contain a number of small molecules including volatile oils, flavonoids, and, primarily, the so-called bitter resins or acids, which comprise 12–15% of all components (Stevens, 1967). Bitter acids have exhibited several antineoplastic properties, including inhibition of tumor transition to malignancy (Chen and Lin, 2004). Bitter resins are classified as α-(e.g., humulone) or β-acids (e.g., lupulone) (Figure 1). The β-acid colupulone has been reported to have antibacterial properties and to inhibit tumor cell proliferation (Manering GJ, 1993). In addition, colupulone was shown to stimulate expression of hepatic CYP3A enzymes in rats and mice (Mannering et al., 1992).
The pregnane X receptor (PXR), a member of the nuclear receptor superfamily of proteins, modulates the expression of genes involved in the metabolism and clearance of a wide array of structurally diverse endogenous and exogenous compounds (Chrencik et al., 2005; Goodwin et al., 2002; Huang et al., 2007; Noble et al., 2006; Orans et al., 2005; Wang et al., 2007; Watkins et al., 2003a; Watkins et al., 2003b; Watkins et al., 2002; Watkins et al., 2001; Xue et al., 2007a; Xue et al., 2007b). Ligand-regulated nuclear receptors contain cannonical DNA binding and ligand binding domains (DBD and LBD, respectively), the latter of which maintains a surface activation function region (AF-2) groove that binds to transcriptional coregulator proteins. Genes regulated by PXR include those encoding cytochrome P450s, gluthathione S transferases, UDP-glucuronosyltransferases, sulfotransferases and the multidrug resistance efflux pumps (Orans et al., 2005).
The PXR LBD has been reported to bind to drugs such as phenobarbital (Guzelian et al., 1988a; Guzelian et al., 1988b), dexamethasone (Maurel, 1996), avasimibe (Sahi et al., 2003; Sahi et al., 2004) and hyperforin, a bioactive compound in the herbal anti-depressant St. John’s wort (Watkins et al., 2003b). PXR activation by these compounds leads to the expression of drug metabolism enzymes, which can lead to dangerous drug-drug interactions. For example, the presence of hyperforin has been shown to reduce the serum concentration and the efficacy of oral contraceptives, immunosuppressants, HIV protease inhibitors, and anticancer chemotherapeutics (Ernst et al., 1998a; Mathijssen et al., 2002; Piscitelli et al., 2000).
In addition to its potential for mediating drug-drug interactions, PXR plays a major role in protecting tissues from xenobiotic and endobiotic stress. For example, PXR activation has been shown to decrease the severity of ulcerative colitis and Crohn’s disease by suppressing pro-inflammatory mediators (Shah et al., 2007). PXR offers hepatoprotection from the toxic accumulation of bile acids by inducing their clearance (Teng and Piquette-Miller, 2007). Neuroprotective effects are also mediated by PXR against neurodegenerative diseases such as Niemann-Pick C by clearing excess lipids and cholesterol (Langmade et al., 2006). In this study, the ability of human PXR to be activated by hops extracts is examined both structurally and functionally.
Colupulone was a gift from KALCEK, Inc. (Kalamazoo, MI). St. John’s wort and gugulipid were purchased from General Nutrition Companies, Inc. (Pittsburgh, PA), and hops were purchased from Nature’s Way Products, Inc. (Springville, UT). Prior to extraction, lyophilized hops and gugulipid were removed from their gelatin capsules, and St. John’s wort tablets were ground into a fine powder with a mortar and pestle. The resultant powders were extracted by vortexing for 2 min in the presence of ethanol (1 g of herbal product/10 ml). A 1 ml aliquot of the mixture was transferred into a microcentrifuge tube and centrifuged for 15 min at 1500 rpm to remove the particulate material. The supernatant was transferred to a fresh microfuge tube and recentrifuged for 15 min at 1500 rpm. The resulting ethanol extracts were dried, weighed and the residue redissolved in DMSO.
Human primary hepatocytes were obtained from the Liver Tissue Procurement and Distribution System (LTPADS) as attached cells in 6-well plates in Human Hepatocyte Maintenance Medium (Cambrex Bio Science Walkersville Inc., Walkersville, MD) supplemented with 100 nM dexamethasone, 100 nM insulin, 100 U/mL penicillin G and 100 μg/mL streptomycin. Twelve hours after changing the culture medium to serum-free William’s E medium, cells were treated with herbs, colupulone, rifampicin or vehicle (0.1% DMSO) for 24 hr.
Total RNA was isolated using Trizol reagent (Invitrogen) according to the manufacturer’s instructions. Real-time quantitative PCR (RTQ-PCR) was performed using an ABI PRISM 7000 Sequence Detection System instrument and software (Applied Biosystems, Inc., Foster City, CA). Samples were assayed in triplicate 25-μl reactions using 25 ng of RNA per reaction. Primers were designed using Primer Express Version 2.0.0 (Applied Biosystems) and synthesized by Integrated DNA Technologies (Coralville, IA). All primers and probes were entered into the NCBI Blast program to ensure specificity. Fold induction values were calculated by subtracting the mean threshold cycle number for each treatment group from the mean threshold cycle number for the vehicle group and raising 2 to the power of this difference. RTQ-PCR primers: CYP2B6, forward: AAGCGGATTTGTCTTGGTGAA, reverse: TGGAGGATGGTGGTGAAGAAG; CYP3A4, forward: CAGGAGGAAATTGATGCAGTTTT, reverse: GTCAAGATACTCCATCTGTAGCACAGT; MDR1, forward: GTCCCAGGAGCCCATCCT, reverse: CCCGGCTGTTGTCTCCAT.
Transfection assays were performed in CV-1 cells plated in 96-well plates at a density of 20,000 cells/well in Dulbecco’s modified Eagle’s medium high glucose medium supplemented with 10% charcoal/dextran treated fetal bovine serum (HyClone, Logan, UT). Transfection mixes included 5 ng of receptor expression vector, 20 ng of reporter plasmid, 12 ng of β-actin secreted placental alkaline phosphatase as an internal control, and 43 ng of carrier plasmid. Human PXR expression plasmids and the CYP3A4/XREM-luciferase reporter, containing the enhancer and promoter of CYP3A4 driving luciferase expression, were used as described previously (Watkins et al., 2003b). Transfections were performed with LipofectAMINE (Life Technologies, Inc.) according to the manufacturer’s instructions. Luciferase activity was normalized to secreted placental alkaline phosphatase expression.
PXR LBD (residues 140–434) was expressed in the N-terminal His-tagged expression vector, pRSET-A (Invitrogen). As described previously, residue Cys-284 was mutated to a serine using the QuikChange mutagenesis kit (Stratagene) to prevent formation of covalent complexes in the presence of DTT (Chrencik et al., 2005; Xue et al., 2007a; Xue et al., 2007b). An 88 amino acid construct of the human SRC-1 gene (residues 623–710) in the pACYC184 vector was co-transformed with the PXR/pRSET-A plasmid into BL21(DE3) E. coli cells. 15 L of cell culture in LB broth supplemented with ampicillin and chloramphenicol were inoculated with PXR/SRC-1 and grown overnight at 22 °C. Harvested cells were centrifuged (20 minutes, 3500 g, 4 °C) and the resulting pellet was resuspended in nickel buffer A (50 mM Tris-Cl pH 7.8, 250 mM NaCl, 50 mM imidazole pH 7.5 and 5% Glycerol). Cells were sonicated on ice for 20 minutes and centrifuged at 20,000 g for 90 minutes at 4 °C. The supernatant was loaded onto a 50 mL nickel column (ProBond, Invitrogen). The column was washed with 200 mL each of nickel buffer A and nickel buffer B (50 mM Tris-Cl pH 7.8, 250 mM NaCl, 75 mM Imidazole pH 7.5, 5% Glycerol). On-column buffer exchange was achieved by washing the column with nickel buffer C (50 mM Tris-Cl pH 7.8, 75 mM Imidazole, 5% Glycerol and 50 mM NaCl) to prepare the sample for subsequent ion-exchange chromatography. Protein was eluted off using nickel buffer D (50 mM Tris-Cl pH 7.8, 250 mM Imidazole, 5% Glycerol and 50 mM NaCl). Column fractions were pooled and immediately loaded onto a SP-cation exchange column (BioRad) pre-equilibriated with SP buffer A (50 mM Tris-Cl pH 7.8, 50 mM NaCl, 5mM DTT, 2.5 mM EDTA pH 8.0 and 5% Glycerol). The protein sample was washed with 200 mL SP buffer A and eluted with SP buffer B (50 mM Tris-Cl pH 7.8, 400 mM NaCl, 5mM DTT, 2.5 mM EDTA pH 8.0 and 5% Glycerol). Pooled fractions were diluted to double the volume, and concentrated to 10 mg/ml using the Centri-prep 30K units (Amicon) in the presence of 25-fold molar excess colupulone and 2-fold molar excess SRC-1 peptide.
PXR-LBD was crystallized using hanging-drop vapor diffusion methods at room temperature against a crystallant containing 50 mM imidazole at pH 8.0, 10% (v/v) isopropanol and 50 mM DTT. Crystals were cryoprotected by serial dipping into 15%, 25% and 35% ethylene glycol. Data collection was conducted at SER-CAT at the Advanced Photon Source in Argonne National Labs (Beamline 19-ID). Diffraction data were indexed, scaled and integrated using HKL2000 (Minor, 1997). Using the apo structure of PXR-LBD (PDB ID: 1ILG) as a search model, molecular replacement was conducted with the MolRep module of CCP4(Collaborative et al., 1994; Winn et al., 2002). Clear molecular replacement solutions were obtained in the spacegroup of P43212. The structure was manually adjusted using a combination of O (Jones, 1991) and WinCoot 3.1 (Emsley and Cowtan, 2004), and was refined using CNS (A.T. Brunger, 1998) and CCP4 (Murshudov GN, 1997). Molecular graphics figures were created using Pymol (Delano, 2002).
We sought to determine the effects of hops on metabolic gene regulation in hepatic tissues using real-time quantitative PCR (RTQ-PCR) methods. St. John’s wort extracts and rifampicin, two established PXR activators, were used as positive controls. Hyperforin from St John’s wort has been shown to have nanomolar affinity for PXR (Moore et al., 2000), while rifampicin is a micromolar affinity ligand (Kliewer et al., 1998; Watkins et al., 2003b; Watkins et al., 2001). RTQ-PCR methods indicate that hops extracts increase mRNA levels for CYP3A4, CYP2B6 and MDR1 in a concentration-dependent manner (Figure 2). The efficacy of hops in inducing these genes was comparable to that exhibited by rifampicin at 10 μM. Comparison of hops and St. John’s wort results indicates that both herbal extracts affect CYP3A4, CYP2B6 and MDR1 levels. Activation of CYP3A4 is noteworthy because this gene product is the most abundant of all the cytochrome P450s, clearing over half of all prescription drugs (Kliewer, 2003; Kliewer et al., 2002).
A transient transfection assay was used to determine whether hops activated PXR. Gugulipid, a herbal extract from the guggul tree (Commiphora mukul) that reduces hyperlipidemia in humans (Brobst et al., 2004), was used as an additional positive control. The biotransformation of gugulipid has been linked to CYP3A4 oxidation in both rodent and human hepatocytes, via a PXR-regulated pathway (Brobst et al., 2004). Hops, gugulipid and St. John’s wort all activated PXR with comparable efficacy. Our data indicate that hops induces CYP3A4 and other drug-metabolizing genes by activating PXR.
Since the hops constituent colupulone is known to activate the transcription of CYP3A genes in mice (Mannering et al., 1992), we hypothesized that it serves as the PXR agonist in hops extracts. Cotransfection data from CV-1 cells validated this hypothesis, and demonstrate a dose-dependent transcriptional activation 2.0- to 2.5-fold above basal levels with only nanomolar (3–10 nM) concentrations of colupulone (Figure 3B). Addition of 30 nM colupulone drops activation levels, possibly due to cell death. Indeed, α- and β-acids have been shown to activate the death receptor Fas, causing apoptosis (Chen and Lin, 2004). The efficacy of colupulone alone, however, was less than that of hops extract, suggesting that other bitter acids in hops may be binding to PXR to induce full transcriptional activity (as discussed below).
The crystal structure of the PXR-LBD in complex with colupulone was determined in space group P43212 using molecular replacement and refined to a resolution of 2.8 Å (Table 1, Figure 4A). Root-mean square deviations (RMSD) between the PXR-colupulone complex and previously reported PXR structures are small, ranging from 0.27–0.54 Å over Cα positions (Chrencik et al., 2005; Watkins et al., 2003a; Watkins et al., 2003b; Watkins et al., 2001; Xue et al., 2007a; Xue et al., 2007b). Low RMSD values were observed regardless of whether the space group of the previously reported structure was P43212 with one complex per asymmetric unit, like the PXR-colupulone structure reported here (Chrencik et al., 2005; Watkins et al., 2003b; Watkins et al., 2001; Xue et al., 2007b), or P212121 with two complexes per asymmetric unit (Watkins et al., 2003a; Xue et al., 2007a). The PXR LBD maintains the canonical nuclear receptor ligand binding fold with a seven membered α-helical sandwich arranged in three layers (α1/α3, α4/α5 and α7/α8). The PXR-colupulone complex structure also contains this five-stranded anti-parallel β-sheet unique to PXR (Noble et al., 2006). The surface AF-2 groove maintains a conformation consistent with the agonist bound form for nuclear receptors, wherein the αAF helix remains immobilized against the groove formed by α3, α3′ and α4.
The main core of the PXR LBD (comprised of the bottom half of α1, α3, bottom half of α4, α5, α7, α8, top half of α10, β2 and β3) exhibits low thermal displacement parameters (B-factors; ~20 Å2). However, as seen previously (Chrencik et al., 2005), higher degrees of thermal motion (thermal displacement parameters approaching 90 Å2) are observed for the AF-2 region (α3, α3′, α4 and αAF), the bottom half of the ligand pocket (β1 and β1′) and other solvent exposed areas (top half of α1, top of half α4 α9, and bottom half α10) (Figure 4A). Average thermal displacement parameters for ligand-binding pocket residues of reported PXR complex crystal structures yield the following ranking, from highest to lowest: rifampicin (51.3 Å2) > colupulone (47.3 Å2) > SR12813 (47.2 Å2) > T0901317 (42.2 Å2) > hyperforin (35.1 Å2) > SR12813 with SRC-1 peptide (24.3 Å2) (Chrencik et al., 2005; Watkins et al., 2003a; Watkins et al., 2003b; Watkins et al., 2001; Xue et al., 2007a; Xue et al., 2007b). In both the rifampicin- and colupulone-PXR complexes structures, α2 and the loops connecting β3 and β4 are disordered (Figure 4A; Figure 5). Thus, while portions of the ligand binding pocket of PXR remain relatively fixed regardless of the ligand bound, other elements (e.g., α2) are capable of a high degree of flexibility even when the LBD is complexed to established agonists (e.g., colupulone, rifampicin). In this way, PXR shares both similarities (ability to change the ligand binding pocket to adapt to ligands) and distinctions (one hemisphere of PXR’s pocket is fixed, while the other is highly mobile) with other members of the nuclear receptor superfamily.
Difference electron density facilitated the positioning of central region of colupulone in the PXR ligand binding pocket, and subsequent refinement allowed the building of the remaining atoms of the isoprene units. Thirteen hydrophobic residues (Met425, Met323, Phe281, Phe288, Trp299, Tyr306, Val211, Leu209, Met243, Ala244, Phe420, Ile414, Leu411) and two polar residues (Arg410, His327) contact carbon atoms of colupulone (Table 2; Figure 4B). Note that residues Met425 and Phe420 are on αAF of the AF-2 region of the receptor. In addition, a direct hydrogen bond is formed between a colupulone hydroxyl and His407, and a water-mediated hydrogen bond is observed between another colupulone hydroxyl group and Gln285 (Figure 4B).
The pocket of the PXR-colupulone complex was compared to other reported PXR crystal structures (Chrencik et al., 2005; Watkins et al., 2003a; Watkins et al., 2003b; Watkins et al., 2001; Xue et al., 2007a; Xue et al., 2007b), and it was noted that the hyperforin (537 Da) and colupulone (400 Da) ligands exhibit some structural similarities. Both contain a cyclic core with isoprene extensions that are significantly involved contacting PXR. However, the hyperforin-PXR complex exhibits interactions in the ligand binding pocket that more closely resembles the PXR-rifampicin complex than the receptor with colupulone (Figures 5, ,6;6; Table 2). Hyperforin contacts the same residues as colupulone, but requires further stabilization provided by seven additional hydrophobic amino acids (Leu240, Leu206, Cys284, Met250, Met246, Phe251 and Leu324) that are also found in the rifampicin structure. Thus, although residues in the colupulone pocket have been observed to contact other ligands in previous structures, it appears difficult to predict the exact identity of residues that may contact a ligand.
Our functional data indicate that additional hops compounds beyond colupulone likely contribute to PXR activation (Figures 2,,3).3). Thus, because only purified colupulone was readily available, we superimposed the other bitter α- and β-acids found in hops onto the ligand in the PXR-colupulone structure and found that these compounds appear capable of binding to human PXR in an analogous manner (Figure 7A,B). Docking of the largest and most substituted member of the bitter acids family, lupulone (414.5 Da), indicates the potential for improved hydrophobic packing with PXR (e.g., the distance lupulone carbons and decreased from 5 Å to 3 Å) but no new polar or non-polar contacts (Figure 7B). Taken together, these modeling observations suggest that both the bitter α- and β-acids from hops have the potential to act as activators of human PXR.
The use of herbal remedies and supplements together with prescribed medications increases the risk of potentially dangerous drug-herb interactions (Burka, 2003). Altered drug clearance due to changes in CYP450 expression profiles have been observed for cardiovascular drugs (e.g., digoxin), immunosuppressants (e.g., cyclosporine, tacrolimus) and anticancer drugs (e.g., imatinib, irinotecan) (Yang et al., 2006). Herbal therapies can also impact laboratory test results and interfere with proper diagnoses (Dasgupta and Bernard, 2006). Thus, we investigated the ability of hops extracts, which are used as herbal supplements, to induce gene transcription in primary human hepatocytes. We found that extracts activated the expression of drug metabolism and clearance genes (Figure 2) in a manner similar to that of St. John’s wort, an established mediator of herb-drug interactions (Ernst et al., 1998a; Ernst et al., 1998b; Piscitelli et al., 2000; Watkins et al., 2003b). We also establish that the human xenobiotic receptor PXR was activated by the hops β-bitter acid colupulone, which has been shown to up-regulate rodent CYP3A expression (Mannering et al., 1992) (Figure 3). The human PXR LBD-colupulone complex crystal structure then facilitated a molecular understanding of the ability of other hops bitter acids to activate PXR (Figures 4–7).
While they may contribute to drug-drug interactions, activators of PXR have the potential to serve as therapeutic leads. For example, PXR agonists have been shown to attenuate inflammatory bowel disease (IBD) through reducing NF-κB target gene expression (e.g., IL-1β, IL-10, iNOS and TNF-α) that mediate colonic inflammation. PXR activators may provide new avenues for the treatment of IBD (Shah et al., 2007). PXR agonists are also hepatoprotective by promoting the elimination of toxic bile acids (Teng and Piquette-Miller, 2007). Similarly, PXR activation has been shown to be neuroprotective in Niemann-Pick Type C disease, characterized by cholesterol and lipid accumulation in the brain (Langmade et al., 2006). Concomitant use of the neurosteroid allopregnanolone and the PXR agonist T0901317 delays symptom onset and prolongs neural cell survival (Langmade et al., 2006; Mellon et al., 2008). PXR activation induces cerebellar CYP3A13 expression, increasing cholesterol clearance and attenuating neuronal injury (Ghoumari et al., 2003; Langmade et al., 2006).
For related nuclear receptors, peroxisome proliferator-activated receptor-γ(PPARγ) agonists suppress inflammation by disrupting nuclear factor κB (NFκB) function (Trifilieff et al., 2003). Nonsteroidal anti-inflammatory drugs (NSAIDs) were reported to reduce the risk of developing Alzheimer’s disease by as much as 80% through mechanisms dependent on PPARγ activation (in t’ Veld et al., 2001). Induction of PPARγ also reduced inflammation related to multiple sclerosis (Schmidt et al., 2004) and are now being used to treat CNS disorders (Heneka et al., 2007). Similarly, there is potential for developing new therapies that exploit PXR’s protective functions in various tissues, in addition to its role in xenobiotic metabolism and drug-drug interactions (Ekins et al., 2002; Kliewer, 2003; Kliewer et al., 2002; Moore and Kliewer, 2000; Sahi et al., 2003; Staudinger et al., 2001; Watkins et al., 2003b; Watkins et al., 2002; Watkins et al., 2001; Willson and Kliewer, 2002; Xie et al., 2000a; Xie et al., 2000b; Xie and Evans, 2002; You, 2004). Several recent reports examine the possibility of PXR agonists as therapeutics (Langmade et al., 2006; Shah et al., 2007; Stedman et al., 2005; Teng and Piquette-Miller, 2007; Xue et al., 2007a). Thus, an expanded understanding of chemical scaffolds capable of activating PXR may facilitate the design of PXR-directed lead compounds.
This work was supported by NIH grants DK62229, DK92310, and the Robert A. Welch Foundation.
We thank Jill Orans, Eric Ortlund, and members of the Redinbo Group for experimental assistance, and Dr. Stephen Strom (University of Pittsburgh) and the Liver Tissue Procurement and Distribution System for human hepatocytes.