Identification of residues in the PXR ligand binding domain critical for species specific and constitutive activation Tove O ¨ stberg 1, *, Go¨ ran Bertilsson 1, * ,† , Lena Jendeberg 2 , Anders Berkenstam 2,‡ and Jonas Uppenberg 3 1 Department of Cell and Molecular Biology, Medical Nobel Institute, Karolinska Institute, Stockholm, Sweden; 2 Departments of Biology, and 3 Structural Chemistry, Biovitrum, Stockholm, Sweden The cytochrome P450 family of enzymes has long been known to metabolize a wide range of compounds, including many of today’s most common drugs. A novel nuclear receptor called PXR has been established as an activator of several of the cytochrome P450 genes, including CYP3A4. This enzyme is believed to account for the metabolism of more than 50% of all prescription drugs. PXR is therefore used as a negative selector target and discriminatory filter in preclinical drug development. In this paper we describe the design, construction and characterization by transient transfection of mutant recep- tors of the human and mouse PXR ligand binding domains. By modeling the human PXR ligand binding domain we have identified and mutated two polar residues in the puta- tive ligand binding pocket which differ between the human and the mouse receptor. The first residue (Q285 in human/ I282 in mouse) was mutated between the two species with the corresponding amino acids. These mutants showed that this residue is important for the species specific activation ofPXR by the ligand pregnenolone-16a-carbonitrile (PCN), while having a less pronounced role in receptor activation by rif- ampicin. The second residue to be mutated (H407 in human/ Q404 in mouse) unexpectedly proved to be important for the basal level of activation of PXR. The H407A mutant of the human receptor showed a high level of constitutive activity, while the Q404H mutant of the mouse receptor demonstra- ted a sharply decreased basal activity compared to wild-type. Keywords: PXR, NR1I2, VDR, ligand binding domain, mutagenesis. The nuclear receptor PXR (NR1I2, PAR, SXR) has been demonstrated to be a key determinant for the transcrip- tional regulation of the drug metabolizing enzyme family of heme-containing monooxygenases P450 CYP3A [1–4]. Consequently this nuclear receptor is likely to play a role in the molecular mechanisms behind common drug inter- actions. PXR is coexpressed in tissues where CYP3A is induced and expressed [5]. The key role of PXRs in CYP3A induction has been further corroborated by targeted disruption of the mouse PXR [6]. These genetically modified animals not only become more sensitive to xenobiotics but also fail to induce CYP3A by known PXR activators [6]. PXR heterodimerizes with 9-cis-retinoic acid receptors (RXR, NR1B1-3) and binds and induces gene expression through a specific genomic response element in the promo- ter region of CYP3A4 and CYP3A7 [1–3,7–9]. PXR is closely related to the constitutive androstane receptor (CAR, NR1I3), which is believed to have a complementary role to PXR in the genetic regulation of cytochrome P450 expression. CAR has been established as a CYP2B gene regulator [10–12], but also activates the same genomic response elements in CYP3A4 and CYP3A7 as PXR [9,13]. PXR has been shown to bind phenobarbital response elements in the CYP2B gene promoter and to be a regulator of CYP2B10 [14] and CYP2B6 gene transcription [15]. The PXR receptor exhibit a promiscuous ligand dependent activation profile and a broad range of synthetic xenobiotics are known to activate the receptor [1–4]. In addition to the activation of PXR by exogenous xenobiotics, it was recently shown that also the endogenously produced, but highly hepatotoxic cholesterol derivative litocholic acid is a potent activator of PXR [16,17]. Accordingly, PXR is involved not only in the detoxification of exogenous xenobiotics, but also of endogenously produced substances. Cloning of PXR orthologs from human, rabbit, rat and mouse [18] has shown that the ligand-binding domain has diverged considerably between the different species. The species divergence and specific activation profile of the orthologous PXRs have also been shown to reflect species specific differences in CYP3A gene induction. For example, the antibiotic compound rifampicin induces human and rabbit, but not rodent CYP3A. It is also a ligand and activator of the human and rabbit PXR but not the rodent PXRs. Pregnenolone-16a-carbonitrile (PCN) on the other hand induces rodent but not human CYP3A and likewise is a ligand for rodent but not human PXR [13]. To date the most potent endogenously produced PXR activator is 5b-pregnane-3,20-dione [1,2,13,18]. This nonplanar steroid activates both the human and the rodent PXR at super- physiological concentrations, but has a preferential affinity Correspondence to J. Uppenberg, Structural Chemistry, Biovitrum, Lindhagensgatan 133, S-112 76, Stockholm, Sweden. Fax: + 46 86972320, Tel.: + 46 86973136, E-mail: jonas.uppenberg@biovitrum.com Abbreviations: PCN, pregnenolone-16a-carbonitrile; LBD, ligand binding domain. *Note: these authors contributed equally to this work. Present address: Neuronova AB, Fiskartorpsva ¨ gen 15 A, S-114 33, Stockholm, Sweden. àPresent address: KaroBio AB, Novum, S-141 57 Huddinge, Sweden. (Received 19 June 2002, revised 8 August 2002, accepted 28 August 2002) Eur. J. Biochem. 269, 4896–4904 (2002) Ó FEBS 2002 doi:10.1046/j.1432-1033.2002.03207.x for the rodent receptors [13]. The species-specific induction pattern of PXR is possibly an adaptive response to the environment and a need to adjust toxicological responses to endogenously produced substances. A transgenic mouse over-expressing the human PXR has been developed with the potential to predict species differences in response to xenobiotics [6]. Structural insights into the molecular mechanism of PXR activation will increase the understand- ing of these species differences and may be used in structure based drug design to avoid PXR activation with its potentially linked side-effects, such as drug-interactions, drug-induced hepatomegaly and decreased bile acid excre- tion [16]. The aim of this study was to explore the molecular mechanism of ligand binding and activation of PXR by modeling and site-directed mutation of the PXR ligand binding domain (LBD). In particular, we wanted to identify residues responsible for the observed differences between rodents and man in order to construct human PXR mutants with mouse like properties and vice versa. In this study we focused on identifying polar amino acids involved in ligand binding. Transient transfection in combination with site directed mutagenesis of the PXR LBDs have enabled us to identify one amino-acid residue involved in the species specific response to activators. An intriguing and more unexpected result was the identification of an amino-acid position in the PXR structure that dramatically affects the basal activity of both the human and mouse receptors. MATERIALS AND METHODS Plasmid constructs, human PXR The full length cloning of human nuclear receptor hPXR (hPAR-2) and the expression vector construct (pcDNA3, Invitrogen) of hPXR have been described previously [2]. Mutants of the human nuclear receptor hPXR (PAR-2) were obtained by Transformer Site-directed mutagenesis Kit (Clontech). The following primers were used: 5¢-TCGAGCTGTGTATACTGAGATTCA-3¢ for Q285I, 5¢-TCAATGCTCAGCAGACCCAGCGGC-3¢ for H407Q, 5¢-TCAATGCTCAGGCCACCCAGCG GC-3¢ for H407A. The selection restriction site mutation was created by primer 5¢-GTAGCTGACTGGAGCATG CAT-3¢ mutating a unique XhoIsite. Plasmid constructs, mouse PXR The full-length mouse PXR (mPXR-2) expression vector was generated by RT-PCR using mouse liver polyA + RNA (BalB/c, Clontech). After PCR amplification the fragment was subcloned into the pcDNA3 vector (Invitro- gen). Oligonucleotides carrying the amino-acid substitutions corresponding to the hPXR mutations were designed: 5¢-TGAGATGTGCCAGCTGAGGTTCA-3¢ for I282Q (forward), 5¢-CAACGCCCAGCATACCCAGCAGT-3¢ for Q404H (forward), 5¢-CAACGCCCAGGCAACCCAG CAGT-3¢ for Q404A (forward), 5¢-TGAACCTCAGCT GGCACATCTA-3¢ for I282Q (reverse), 5¢-ACTGCTG GGTATGCTGGGCGT-3¢ for Q404H (reverse), 5¢-ACT GCTGGGTTGCCTGGGCGT-3¢ for Q404A (reverse). Mutations were introduced by PCR mutagenesis in a two step reaction. The pCDNA3 vector primers used were na614 5¢-CTGCTTACTGGCTTATCGAA-3¢ (forward) and na1106 5¢-GGGTCAAGGAAGGCACGG-3¢ (reverse). The mutants were subcloned into the pCDNA3 vector (Invitrogen). General plasmid constructs The CYP3A4 luciferase reporter plasmid ()10466 to +53) has been described previously [9]. The pRSV-AF control plasmid for transfection normalization was previously described [2]. All constructs were verified by sequence analysis. Reporter gene assay All transient transfection experiments were performed in C3A cells (ATCC, CRL-10741, lot i 1414101) in 6-well plates. C3A cells were seeded at a concentration of 5 · 10 5 cells in each well and incubated for 24 h at 37 °Cin2mLgrowth medium containing minimal essential medium (MEM), 10% fetal bovine serum, nonessential amino acids and sodium pyruvate (Life Technologies). The medium was replaced with 2 mL transfection medium (MEM, 10% charcoal/dextran treated fetal bovine serum (Hyclone), nonessential amino acids, sodium pyruvate) and the cells were cotransfected with 2 lg CYP3A4-luciferase reporter, 0.05 lg hPXR/mPXR/mutant plasmid and 0.1 lgRSV-AF plasmid (alkaline phosphatase activity was used for nor- malization of transfection efficiency) using FuGENE-6 (Roche) according to the manufacturer’s instructions. After 20–24 h, medium was replaced and cells were induced with rifampicin (Sigma), SR12813 (synthesized by Biovitrum) or Pregnenolone-16a-carbonitrile (PCN) (Sigma) in optimized serial dilutions as indicated in the figures. DMSO was used as vehicle. Following 48 h incubation, the medium was analyzed for alkaline phosphatase activity according to the manufacturer’s recommendations (Great EscAPe SEAP, Promega). Cells were harvested and the cell lysates were analyzed for luciferase activity. All experiments were performed at least three times in duplicates and luciferase activity was normalized for alkaline phosphatase activity. For curve fitting and EC50 calculations, XLFIT version 2.0.3 was used. Western blot analysis C3A cells were seeded into 75 cm 2 flasks at a density of 3.75 · 10 6 cells per flask and incubated at 37 °Covernight. Co-transfections were performed as described earlier and the cells were transfected with 15 lg CYP3A4-luciferase reporter, 0.75 lg RSV-AF plasmid and 0.375 lg plasmid containing hPXR/mPXR or a mutant variant thereof. After 24 h the cells were washed and scrapeloaded in NaCl/P i . Cell pellets were collected by centrifugation and resus- pended in Lysis buffer A (10 m M Hepes/KOH pH 7.6, 1.5 m M MgCl 2 ,10m M KCl, 0.5 m M dithiothreitol, 1 m M EDTA, 1 m M EGTA, 1% Triton X-100, protease inhibi- tors). Nuclear pellets were collected by centrifugation at 4000 r.p.m. for 10 min (4 °C). The supernatants were cleared by centrifugation at 140 00 r.p.m. for 10 min (4 °C) and saved as cytoplasmic fractions. The nuclear pellets were resuspended in Lysis buffer B (20 m M Hepes/ KOH pH 7.6, 1.5 m M MgCl 2 ,420 m M NaCl, 1 m M EDTA, Ó FEBS 2002 Mutagenesis of PXR ligand binding domain (Eur. J. Biochem. 269) 4897 1m M EGTA, 20% glycerol and protease inhibitors) and gently mixed for 20 min at 4 °C. The insoluble fractions were removed by centrifugation at 140 00 r.p.m. for 10 min (4 °C) and the supernatants were saved as nuclear fractions. The protein content of the nuclear and cytoplasmic fractions were determined by amino-acid analysis. 250 lL of 6 M HCl with 0.5% Phenol was added to each of the samples (5 lL) and the hydrolysis was carried out at a temperature of 155 °C for 45 min. For the amino-acid analysis an AminoQuant II/M High Sensitivity Instrument (Hewlett Packard, Waldbronn, Germany) was used. The AminoQuant amino-acid analyzer combines the OPA and FMOC as derivatization reagents for the complete detection of all residues. BSA was used as a standard protein to calculate the amount of protein in the samples. Western blot analysis was performed using 20 lL(hPXR/mPXR, respectively) of the nuclear and cytoplasmic fractions mixed with NuPAGE Sample buffer supplemented with reducing agent. The mixes were heated at 70 °C for 10 min and the samples were applied on a 10% NuPAGE Bis/Tris Gel (Invitrogen). The protein was transferred to a Hybond-C extra membrane (Amersham Life Science) and blocked in NaCl/P i /Tween supplemented with 5% dry milk overnight. ThemembranewaswashedwithNaCl/P i /Tween and incubated for 1 h at RT with the hPXR/mPXR specific antibodies PXR (N-16): sc-9690/PXR (R-14): sc-7739 (Santa Cruz), diluted 1/100 in NaCl/P i /Tween supplemented with 5% dry milk. The membrane was washed with NaCl/ P i /Tween and subsequently incubated for 45 min with Peroxidase-conjugated rabbit anti-(goat IgG) Ig (DAKO), diluted 1/2000 in NaCl/P i /Tween supplemented with 5% dry milk. A final wash was made with NaCl/P i /Tween. All NaCl/P i /Tween used in Western blot analysis detected by the mPXR specific antibody was supplemented with both 5% dry milk and 5% fetal bovine serum. The Western blot was visualized using ECL Western blotting detection reagent RPN 2106 (Amersham Pharmacia Biotech) and Hyperfilm ECL (Amersham Pharmacia Biotech). Modeling The structure of the vitamin D receptor ligand binding domain [19] was used as template for modeling human PXR (PDB entry 1DB1). Modeling was performed with the program O [20]. The conserved residues in VDR and PXR were kept intact in the PXR model. Substituted amino acids were modeled as the most likely conformer from the O structural database. In cases where the side chain modeling gave rise to close contacts, other energetically favorable conformations were chosen. The VDR crystal structure [19] lacks a region of 50 amino acids in the omega-loop that were deleted in the expression construct to obtain suitable protein for crystallization. It was suggested that this region lacked stable structure and therefore interfered with crystallization. We have consequently not modeled this region of PXR. The two receptor sequences are furthermore most dissimilar in this part of the structure. In addition there are four deletions of one or two amino acids in the LBD of the PXR sequence as compared to VDR. These are found in surface and loop regions in the structure and were modeled manually in O , followed by geometric regularization using the refine_zone command. The model was finally subjected to 50 cycles of conjugate gradient energy minimization with the program CNS [21]. The minimized structure was examined for large structural changes and none were observed. The ligand binding pocket was identified with the program VOIDOO , using a probe diameter of 1.4 A ˚ [22]. RESULTS Our homology model of human PXR LBD suggested the presence of an elongated and closed ligand binding pocket with an approximate size of 15 · 5 · 5A ˚ . The binding pocket as found by the program Voidoo was delimited by atoms from the following residues: Leu240, Met243, Ala244, Met246, Ser247, Phe251, Phe281, Cys284, Gln285, Phe288, Trp299, Tyr306, Thr311, Gly314, Phe315, Leu319, Met323, His407, Leu411, Ile414, Gln415, Ile417, His418, Phe420, Ala421, Met425, Gln426 and Phe429. Of these amino-acid residues we identified two polar residues, Gln285 and His407, which were not conserved between the mouse and human receptors and where the side chains lined the ligand binding pocket (Fig. 1). We proceeded to construct mutants of these two residues based on the hypothesis that they were involved in the species specific activator response. Three single point mutations were made for human PXR: Q285I, H407Q and H407A. The first two replaced the human amino-acid residue with its mouse counterpart. The third mutant was made in order to create a more pronounced change than the spatially and electrostatically moderate change of a histidine to a glutamine and thereby give additional information into its potential role in ligand binding. We also made the three analogous mutants of mouse PXR: I282Q, Q404H and Q404A. The wild-type and mutant receptors were tested in a transient cotransfection assay, using expression vectors for Fig. 1. The ligand binding pocket of human PXR LBD (coordinates from the crystal structure [23] with PDB code: 1ILH). A cavity surface was generated with the program VOIDOO [22] and represents the surface accessible by the center of a 1.4-A ˚ probe. The side chains of the two mutated residues, His407 and Gln285, are both adjacent to the ligand binding pocket. The crystal structure of human PXR has shown that these residues are also involved in hydrogen bonding interactions with the synthetic ligand SR12813 [23]. 4898 T. O ¨ stberg et al. (Eur. J. Biochem. 269) Ó FEBS 2002 the full length mouse and human PXR variants, in combination with a reporter vector containing the CYP3A4 promoter ()10466 to +53) fused to a luciferase reporter gene. Luciferase activity was measured as read-out after induction with rifampicin, Pregnenolone-16a-carbonitrile (PCN) and SR12813 (Fig. 2). All compounds are well characterized ligands for human and mouse PXR, where rifampicin and SR12813 are potent activators of the human receptor and PCN primarily activates the mouse receptor [1–4,13,18]. During this study a crystal structure of PXR was published by Watkins et al. [23], which led us to compare our model with the experimental coordinates (PDB entry 1ILH was used). A total of 204 carbon-alpha atoms with an rms deviation of 1.50 A ˚ were aligned with the lsq_explicit option in the program O [20]. A few regions of the model were not properly aligned due to large differences. Most of these nonaligned regions were located in the omega-loop and beta-sheet of the protein and contained the following residues: 175–236, 302 and 308–320. Two short additional loop regions were poorly modeled: residues 385–387 between helices 9 and 10 and residues 416–421 between helices 10 and 12. Wild-type human and mouse PXR PCN was a strong activator of mouse PXR, while it was a poor activator of human PXR (Fig. 3). Rifampicin and SR12813 on the other hand showed strong activation of the human receptor, while only weak activation of the mouse PXR could be detected (Fig. 3). Human Q285I and mouse I282Q The basal reporter gene activities (i.e. in the absence of activator) of the human Q285I and mouse I282Q receptor Fig. 2. The structures of ligands tested for PXR activation: (A) rif- ampicin, (B) SR12813, (C) pregnenolone-16a-carbonitrile (PCN). Fig. 3. Diagrams of transciptional activation, as determined by luciferase reporter assay, at two ligand concentrations for (A) human and (B) mouse wild-type PXR. Ligand concentrations chosen were 5, 10 and/or 20 l M . The values have been corrected for alkaline phosphatase activity and normalized against a DMSO control. The human receptor was strongly activated by rifampicin (RIF) and SR12813 (SR), while mouse PXR was primarily activated by PCN. Ó FEBS 2002 Mutagenesis of PXR ligand binding domain (Eur. J. Biochem. 269) 4899 variants are similar in levels to their corresponding wild-type human and mouse receptor (Fig. 4). The mutant Q285I in the human receptor is activated by PCN at lower concen- trations compared to the wild-type human receptor, with calculated EC50s of 4 l M and 14 l M , respectively (Fig. 5d,F). For rifampicin (Fig. 5A) and SR12813 (Fig. 5c) we found an approximately twofold decrease in fold induction by Q285I compared to human wild-type receptor. The corresponding mutant of the mouse receptor, I282Q, shows a decreased activation by PCN both in terms of EC50 and fold induction (Fig. 5E). As observed with the wild-type mouse receptor the I282Q mutant was neither activated by rifampicin nor SR12813 (data not shown). Human H407Q and mouse Q404H The basal activity of the human PXR mutant H407Q is similar to the wild-type receptor (Fig. 4A). The correspond- ing mutant of mouse Q404H shows a marked decrease in basal activity as compared to both the wild-type and the other mutants of the mouse receptor (Fig. 4B). The human H407Q mutant and the wild-type receptor are activated to a similar degree by PCN (Fig. 5D). H407Q is still activated by rifampicin with a slightly lower EC50, but also with a lower fold induction (Fig. 5a,F). The SR12813 compound simi- larly activates H407Q, but with a lower fold induction (Fig. 5C). The mouse mutant Q404H is strongly activated by PCN and in terms of fold induction surpasses the wild- type (Fig. 5E). Neither rifampicin nor SR12813 activated Q404H (data not shown). Human H407A and mouse Q404A The human mutant receptor H407A showed nearly a four- fold increase in basal activity compared to wild-type and the other mutants of the human receptor (Fig. 4A). This was not observed for the corresponding mouse receptor mutant Q404A, where basal activity was similar to the wild-type (Fig. 4B). Although H407A displayed a high basal activity it could still be activated further by rifampicin (Fig. 5B). Also SR12813 could activate this mutant although to a lesser extent than rifampicin. PCN however, had no effect on this mutant (data not shown). The mouse receptor Q404A resembled the wild-type receptor in its activation by PCN (Fig. 5e), while showing no activation by rifampicin or SR12813 (data not shown). Western blots To compare the expression levels of wild-type hPXR/ mPXR, mutant hPXR/mPXR and the endogenous expres- sion of hPXR in C3A cells, Western blot analysis was performed on the nuclear fractions of the cell lysates. In cells transfected with wild-type or mutant hPXR, two bands of similar strength were detected (Fig. 6). The band corres- ponding to the larger protein product (approximately 54 kDa) agrees in size with the PXR isoform hPAR-2 [2]. The second band (approximately 50 kDa) corresponds in size to hPXR-1 [1]. The amount of overexpressed protein was similar for all four constructs. In the untransfected cells andcellstransfectedwithemptyvector,pcDNA3(Fig. 6),a single weak band was observed corresponding in size to hPXR-1. In cells transfected with wild-type or mutant mPXR a band of similar strength (approximately 50 kDa) was detected (data not shown, see Discussion). Cytosolic fractions were also analyzed and only very weak bands could be detected on a Western blot (data not shown). DISCUSSION The PXR nuclear receptor has become a new focus of nuclear receptor research after the discoveries of its central role in drug metabolism and xenobiotic signaling. In this study we have used mutated receptors to investigate the role of specific residues in receptor activation and in particular address the different activation profiles observed for human and mouse PXR. For that purpose we have built a homology model of human PXR in order to identify residues that were likely to be involved in ligand binding. Since the initiation of this study the crystal structure of the human receptor has been published [23]. This has allowed us to compare our model with the crystal structure and validate our choice of mutations. To a large extent our model corresponds to the crystal structure and our choice of candidate residues for mutation reflects well the questions we wanted to address. These residues are also located in regions where our model agrees closely with the crystal structure. There are other parts of the model that do not correlate with the crystal structure, in particular the region Fig. 4. A diagram of basal transcriptional activity in wild-type and mutated receptors as determined by a luciferase reporter assay. Prior to measurements 2 lL DMSO was added to each well. (a) The human PXR constructs showed similar basal levels with exception for H407A, which was strongly activated without addition of ligand. (b) The mouse PXR mutants I282Q and Q404A displayed basal activites that were close to that of the wild-type receptor, while Q404A showed a distinctly lower level of activation. 4900 T. O ¨ stberg et al. (Eur. J. Biochem. 269) Ó FEBS 2002 neighboring the beta sheet and what is usually referred to as the omega loop. This could not be accurately modeled, as the corresponding region of the template structure was not present. As a consequence the full extent of the ligand binding pocket was not fully modeled. We will therefore refer to the crystal structure rather than our model in the molecular interpretation of our results. Western blot analysis of protein expression levels A Western blot analysis of cell lysates containing the human PXR constructs shows the presence of a protein of expected size, approximately 54 kDa. However, another band of equal strength also appears for all constructs. This band corresponds to a protein of lower molecular mass, approximately 50 kDa, which is comparable to a band seen in the empty plasmid and untransfected cell control experiments. The bands seen in the control experiments are considerably weaker however. Some endogenous human PXR is likely to be present in all experiments and should be taken into account in the interpretation of the results. However we believe that the background activity that stems from endogenous hPXR-1 is low in comparison with that from the transfected constructs. The second band seen in the lanes of the transfected constructs are much stronger than in the control experiment, suggesting instead the presence of a truncated protein of a molecular mass similar to endogenous PXR. This is likely due to an alternative translation initiation site by a non-AUG codon [33], which is present in PXR [2]. Any substantial Fig. 5. Activation curves for different ligands and receptor constructs used in the luciferase reporter assay. Human PXR wild-type and mutant receptors activated by (a) rifampicin (b) rifampicin (mutant H407A) (c) SR12813 and (d) PCN; (e) mouse PXR wild-type and mutant receptors activated by PCN. (f) Table of EC50 values as calculated by the program Xlfit. Ó FEBS 2002 Mutagenesis of PXR ligand binding domain (Eur. J. Biochem. 269) 4901 contribution from the presence of endogenous human PXR in our experiments is expected to result in a strong induction by rifampicin or SR12813 in cells transfected with the mouse PXR. This has not been observed (Fig. 3b). We also made a Western blot analysis of the mouse receptor constructs. Although many attempts were made only weak detection of mouse PXR could be performed with the antibody at our disposal and we have been unable to obtain a blot clear enough to print. However we estimated that the levels of expression are roughly equal for all constructs. Mutation of human Gln285 and mouse Ile282 The residue Q285/I282 is located in the ligand binding pocket on helix 5 with the side chain easily accessible for potential ligands (Fig. 1). The crystal structure of human PXR in complex with SR12813 shows how this residue is involved in hydrogen bonding to the ligand in one of the three modes that this ligand can bind to the receptor [23]. This is consistent with our mutant Q285I, which has a slightly decreased ability to be activated by SR12813. This suggests that at least one of the binding modes of SR12813 have been altered. The activation of Q285I by PCN has been improved compared to human wild-type PXR, while the reverse mutant I282Q of the mouse receptor shows a decreased activation by PCN. This suggests that PCN also binds in close proximity to this residue and that a hydrophobic interaction may be more favorable. Given the fact that PCN is a better activator of mouse than human PXR, we believe that this mutation is central to making the human receptor more like the mouse receptor. This is supported by the fact that this is the only clear example where a hydrophilic side chain has been replaced by a hydrophobic one in the core of the ligand binding pocket. The Q285I mutant also shows decreased propensity for activation by rifampicin, which indicates that this large molecule may also come in contact with this residue. The binding mode of rifampicin however, is unclear as it is too large to be accommodated into the binding pocket described by the crystal structure. The reverse mutation I282Q does not impose enough human like properties to the mouse receptor to make it susceptible to activation by either rifampicin or SR12813. This suggests that while some species specific properties may be changed by single point mutations, others are more subtle and requires multiple substitutions to mimic. Mutation of human His407 and mouse Gln404 Our model suggested that this residue was located at one end of an elongated ligand binding pocket. The crystal structure confirmed its accessibility to ligands and the histidine residue makes a hydrogen bond to SR12813 in one of its binding modes. The mutation of this residue gave a number of surprising results suggesting that this residue play a key role in receptor activation. The basal activity in particular seems to be sensitive to the nature of this residue. This was evident from the human H407A mutant, where the basal activity increased dramatically, and the mouse Q404H mutant, where the basal activity decreased by more than 50% (Fig. 4). The basal activity of H407Q and Q404A on the other hand remained close to wild-type levels. The structural reasons for the observed changes in basal activity are not obvious, although one can speculate on rearrange- ments of the region around helix 12 (Fig. 1), which is known to be critical for coactivator binding and thereby activation. Replacing the histidine with an alanine in the human receptor creates a void, which is surrounded by the hydrophobic side chains of Phe281, Met323, Leu411, Phe420 and Phe429. It is possible that the mutation causes these side chains to reorient themselves to partly fill this void. Phe429 is of special interest as it belongs to helix 12 and even a small movement or stabilization of this residue could be of importance for receptor activation. It is noteworthy that His407 takes on a different conformation in the ligand bound structure of PXR, with a side chain movement away from helix 12, as compared to the apo- structure. It is interesting to note that a similar mutation in this area, R410A also creates a constitutively active receptor [23]. This residue lies side by side with His407 one helical turn away on helix 11. The replacement of these two large side chains with the beta-carbon of alanine could introduce more flexibility to helix 11 itself. Although one cannot predict exactly what effect this has on the structure, the proximity to helix 12 both sequentially and geometrically could have an influence on coactivator binding. Helix 11 is also part of the dimerization interface and one cannot exclude an impact on the conformation of the heterodimer that PXR forms with RXR. It is surprising that the analogous mutation in mouse PXR, Q404A, does not affect basal activity, while Q404H shows a dramatic decrease of the same. The only correlation seems to be that a histidine in this position has a negative relative effect on basal activity. While the effect on basal activity is striking for mutations in this position, the ligand dependent activation is less dramatically affected and there is little evidence to show that this residue is important for species specific activation. H407Q is still strongly activated by rifampicin and SR12813, while Q404H is strongly activated by PCN. The Fig. 6. Western blot analysis of nuclear fractions showing hPXR expression in cells transfected with empty vector (lane 1), hPXR wild- type (lane 2), Q285I (lane 3), H407Q (lane 4), H407A (lane 5). The amino-acid analyses determined the protein contents loaded on the gel as follows: empty vector (lane 1) 31 lg, hPXR wild-type (lane 2) 26 lg, Q285I (lane 3) 34 lg, H407Q (lane 4) 35 lg, H407A (lane 5) 25 lg. Aweakbanddetectedinthecontrolexperiment(lane1)couldbe attributed to endogenous expression of hPXR-1. The overexpression of hPXR-2 wild-type and mutant proteins (lanes 2–5) resulted in two strong bands with little difference observed between the four con- structs. The largest band corresponds to the molecular mass of hPXR- 2 (approximately 54 kDa), while the second band agrees with the molecular mass of hPXR-1 (approximately 50 kDa). The appearance of two gene products is most likely due to alternative translational initiation by a non-AUG codon [33], one of which is present in the PXR sequence [2]. 4902 T. O ¨ stberg et al. (Eur. J. Biochem. 269) Ó FEBS 2002 strong effect of PCN on Q404H should be viewed in perspective of the basal activity. The full activation of the mutant is similar in level to the wild-type in absolute terms, but as the basal activity is lower for the mutant, the number of fold activation is higher. One can see Q404H as a sensitized receptor, where the negative effect on the basal activity is countered and neutralized by the ligand. No improvement is seen in activation of H407Q by PCN over wild-type, nor Q404H by rifampicin or SR12813. Although the mutant H407A shows a high basal activity, it can be further activated by the potent activator rifampicin. If the mutation triggers specific conformational changes that facilitate receptor activation, the binding of ligand may still improve activation by a general stabilization of the receptor. This phenomenon has earlier been observed in NMR studies for the PPARc receptor [24]. No clear increase in activation by PCN or SR12813 was observed. SR12813 could be expected to lose some affinity for this mutant as one of the hydrogen bond partners has been removed. The mouse mutant Q404A was similar to the wild-type in its ability to be activated by PCN, while being unresponsive to rifampicin and SR12813. The corresponding residue to His407 of human PXR is remarkably conserved across a wide variety of nuclear receptors, including the PPARs, TRs, VDRs and RORs [25]. The crystal structures of these receptors show that this histidine side chain is interacting directly with ligands and/or helix 12 through hydrogen bonds [19,26–29]. The discovery that His407/Gln404 plays a crucial structural role in the activation process of PXR, could be applicable also to other receptors and further mutational and structural studies would be of great interest to further elucidate the dynamics of this part of the ligand binding domain. There are other examples where a single mutation has yielded constitutively active nuclear receptors, such as RXR [30] and the estrogen receptor [31,32]. In the case of RXR a mutation of Phe318 into an alanine in helix 5 causes a destabilization in a network of hydrophobic interactions in the apo-receptor core. In the estrogen receptor Tyr571 was mutated to an aspartic acid in the vicinity of helix 12. This produced a constitutively active receptor, which interacted with some but not all coactivator proteins tested. With more structural data on mutated nuclear receptors we may anticipate a more detailed dynamic picture of the transition from a silent to an activated nuclear receptor and the role of heterodimer formation and coactivators in the relay of the transcrip- tional signal. ACKNOWLEDGEMENTS We would like to thank Kristina Zachrisson for performing the amino- acid analysis and Sven-A ˚ ke Franze ´ n, Andrea Varadi and Marianne Israelsson for DNA sequence analysis. REFERENCES 1. Kliewer, S.A., Moore, J.T., Wade, L., Staudinger, J.L., Watson, M.A., Jones, S.A., McKee, D.D., Oliver, B.B., Willson, T.M., Zetterstrom, R.H., Perlmann, T. & Lehmann, J.M. (1998) An orphan nuclear receptor activated by pregnanes defines a novel steroid signaling pathway. Cell 92, 73–82. 2. Bertilsson, G., Heidrich, J., Svensson, K., Asman, M., Jendeberg, L., Sydow-Backman, M., Ohlsson, R., Postlind, H., Blomquist, P. & Berkenstam, A. (1998) Identification of a human nuclear receptor defines a new signaling pathway for CYP3A induction. Proc. Natl Acad. Sci. USA 95, 12208–12213. 3. Blumberg, B., Sabbagh, W. Jr, Juguilon, H., Bolado, J. Jr, van Meter, C.M., Ong, E.S. & Evans, R.M. (1998) SXR, a novel steroid and xenobiotic-sensing nuclear receptor. Genes Dev. 12, 3195–3205. 4. Lehmann, J.M., McKee, D.D., Watson, M.A., Willson, T.M., Moore, J.T. & Kliewer, S.A. (1998) The human orphan nuclear receptor PXR is activated by compounds that regulate CYP3A4 gene expression and cause drug interactions. J. Clin. Invest. 102, 1016–1023. 5. de Wildt, S.N., Kearns, G.L., Leeder, J.S. & van den Anker, J.N. (1999) Cytochrome P450, 3A: ontogeny and drug disposition. Clin. Pharmacokinet. 37, 485–505. 6. Xie, W., Barwick, J.L., Downes, M., Blumberg, B., Simon, C.M., Nelson, M.C., Neuschwander-Tetri, B.A., Brunt, E.M., Guzelian, P.S. & Evans, R.M. (2000) Humanized xenobiotic response in mice expressing nuclear receptor SXR. Nature 406, 435–439. 7. Goodwin, B., Hodgson, E. & Liddle, C. (1999) The orphan human pregnane X receptor mediates the transcriptional activation of CYP3A4 by rifampicin through a distal enhancer module. Mol. Pharmacol. 56, 1329–1339. 8. Pascussi, J.M., Jounaidi, Y., Drocourt, L., Domergue, J., Bala- baud, C., Maurel, P. & Vilarem, M.J. (1999) Evidence for the presence of a functional pregnane X receptor response element in the CYP3A7 promoter gene. Biochem. Biophys. Res. Comm. 260, 377–381. 9. Bertilsson, G., Berkenstam, A. & Blomquist, P. (2001) Function- ally conserved xenobiotic responsive enhancer in cytochrome P450 3A7. Biochem. Biophys. Res. Commun. 280, 139–144. 10. Honkakoski, P., Zelko, I., Sueyoshi, T. & Negishi, M. (1998) The nuclear orphan receptor CAR-retinoid X receptor heterodimer activates the phenobarbital-responsive enhancer module of the CYP2B gene. Mol. Cell. Biol. 18, 5652–5658. 11. Sueyoshi, T., Kawamoto, T., Zelko, I., Honkakoski, P. & Negishi, M. (1999) The repressed nuclear receptor CAR responds to phe- nobarbital in activating the human CYP2B6 gene. J. Biol. Chem. 274, 6043–6046. 12. Wei, P., Zhang, J., EganHafley, M., Liang, S. & Moore, D.D. (2000) The nuclear receptor CAR mediates specific xenobiotic induction of drug metabolism. Nature 407, 920–923. 13. Moore, L.B., Parks, D.J., Jones, S.A., Bledsoe, R.K., Consler, T.G., Stimmel, J.B., Goodwin, B., Liddle, C., Blanchard, S.G., Willson, T.M., Collins, J.L. & Kliewer, S.A. (2000) Orphan nuclear receptors constitutive androstane receptor and pregnane X receptor share xenobiotic and steroid ligands. J. Biol. Chem. 275, 15122–15127. 14. Xie, W., Barwick, J.L., Simon, C.M., Pierce, A.M., Safe, S., Blumberg, B., Guzelian, P.S. & M.E.R. (2000) Reciprocal acti- vation of xenobiotic response genes by nuclear receptors SXR/ PXR and CAR. Genes Dev. 14, 3014–3023. 15. Goodwin, B., Moore, L.B., Stoltz, C.M., McKee, D.D. & Kliewer, S.A. (2001) Regulation of the human CYP2B6 gene by the nuclear pregnane X receptor. Mol. Pharmacol. 60, 427–431. 16. Staudinger, J.L., Goodwin, B., Jones, S.A., Hawkins-Brown, D., MacKenzie, K.I., Latour, A., Liu, Y.P., Klaassen, C.D., Brown, K.K., Reinhard, J., Willson, T.N., Koller, B.H. & Kliewer, S.A. (2001) The nuclear receptor PXR is a lithocholic acid sensor that protects against liver toxicity. Proc. Natl Acad. Sci. USA 98, 3369– 3374. 17. Xie, W., Radominska-Pandya, A., Shi, Y.H., Simon, C.M., Nelson, M.C., Ong, E.S., Waxman, D.J. & Evans, R.M. (2001) An essential role for nuclear receptors SXR/PXR in detoxification of cholestatic bile acids. Proc. Natl Acad. Sci. USA 98, 3375–3380. 18. Jones, S.A., Moore, L.B., Shenk, J.L., Wisely, G.B., Hamilton, G.A., McKee, D.D., Tomkinson, N.C., LeCluyse, E.L., Lambert, Ó FEBS 2002 Mutagenesis of PXR ligand binding domain (Eur. J. Biochem. 269) 4903 M.H., Willson, T.M., Kliewer, S.A. & Moore, J.T. (2000) The pregnane X receptor: a promiscuous xenobiotic receptor that has diverged during evolution. Mol. Endocrinol. 14, 27–39. 19. Rochel, N., Wurtz, J.M., Mitschler, A., Klaholz, B. & Moras, D. (2000) The crystal structure of the nuclear receptor for vitamin D bound to its natural ligand. Mol. Cell. 5, 173–179. 20. Jones, T.A., Zou, J.Y., Cowan, S.W. & Kjeldgaard, M. (1991) Improved methods for building protein models in electron density maps and the location of errors in these models. Acta Crystallogr. A47, 110–119. 21. Bru ¨ nger, A.T., Adams, P.D., Clore, G.M., DeLano, W.L., Gros, P., Grosse-Kunstleve, R.W., Jiang, J.S., Kuszewski, J., Nilges, M., Pannu, N.S., Read, R.J., Rice, L.M., Simonson, T. & Warren, G.L. (1998) Crystallography and NMR system: a new software suite for macromolecular structure determination. Acta Crystal- logr. D54, 905–921. 22. Kleywegt, G.J. & Jones, T.A. (1994) Detection, delineation, measurement and display of cavities in macromolecular structures. Acta Crystallogr. D50, 178–185. 23. Watkins, R.E., Wisely, G.B., Moore, L.B., Collins, J.L., Lambert, M.H., Williams, S.P., Willson, T.M., Kliewer, S.A. & Redinbo, M.R. (2001) The human nuclear xenobiotic receptor PXR: Structural determinants of directed promiscuity. Science 292, 2329–2333. 24. Johnson, B.A., Wilson, E.M., Li, Y., Moller, D.E., Smith, R.G. & Zhou, G. (2000) Ligand-induced stabilization of PPARgamma monitored by NMR spectroscopy: implications for nuclear receptor activation. J. Mol. Biol. 298, 187–194. 25. Wurtz,J.M.,Bourguet,W.,Renaud,J.P.,Vivat,V.,Chambon,P., Moras, D. & Gronemeyer, H. (1996) A canonical structure for the ligand-binding domain of nuclear receptors. Nat. Struct. Biol. 3, 87–94. 26. Nolte, R.T., Wisely, G.B., Westin, S., Cobb, J.E., Lambert, M.H., Kurokawa, R., Rosenfeld, M.G., Willson, T.M., Glass, C.K. & Milburn, M.V. (1998) Ligand binding and co-activator assembly of the peroxisome proliferator-activated receptor-gamma. Nature 395, 137–143. 27. Uppenberg, J., Svensson, C., Jaki, M., Bertilsson, G., Jendeberg, L. & Berkenstam, A. (1998) Crystal structure of the ligand binding domain of the human nuclear receptor PPARgamma. J. Biol. Chem. 273, 31108–31112. 28. Wagner, R.L., Apriletti, J.W., McGrath, M.E., West, B.L., Baxter, J.D. & Fletterick, R.J. (1995) A structural role for hormone in the thyroid hormone receptor. Nature 378, 690–697. 29. Stehlin, C., Wurtz, J M., Steinmetz, A., Greiner, E., Schu ¨ le, R., Moras, D. & Renaud, J P. (2001) X-ray structure of the orphan nuclear receptor RORb ligand-binding domain in the active con- formation. EMBO J. 20, 5822–5831. 30. Vivat, V., Zechel, C., Wurtz, J.M., Bourguet, W., Kagechika, H., Umemiya, H., Shudo, K., Moras, D., Gronemeyer, H. & Cham- bon, P. (1997) A mutation mimicking ligand-induced conforma- tional change yields a constitutive RXR that senses allosteric effects in heterodimers. EMBO J. 16, 5697–5709. 31. Weis, K.E., Ekena, K., Thomas, J.A., Lazennec, G. & Katze- nellenbogen, B.S. (1996) Constitutively active human estrogen receptors containing amino acid substitutions for tyrosine 537 in the receptor protein. Mol. Endocrinol. 10, 1388–1398. 32. White, R., Sjoberg, M., Kalkhoven, E. & Parker, M.G. (1997) Ligand-independent activation of the oestrogen receptor by mutation of a conserved tyrosine. EMBO J. 16, 1427–1435. 33. Kozak, M. (1997) Recognition of AUG and alternative initiator codons is augmented by G in position +4 but is not generally affected by the nucleotides in positions +5 and +6. EMBO J. 16, 2482–2492. 4904 T. O ¨ stberg et al. (Eur. J. Biochem. 269) Ó FEBS 2002 . transfection of mutant recep- tors of the human and mouse PXR ligand binding domains. By modeling the human PXR ligand binding domain we have identified and mutated. Identification of residues in the PXR ligand binding domain critical for species specific and constitutive activation Tove O ¨ stberg 1, *,