Regulation of Dioxin Receptor Function by Omeprazole*

The intracellular dioxin (aryl hydrocarbon) receptor mediates signal transduction by dioxin (2,3,7,8-tetra-chlorodibenzo- p -dioxin) and related environmental pol- lutants and functions as a ligand-activated transcription factor. In this study we have examined the effects on dioxin receptor function of a potentially novel ligand, omeprazole, which is widely clinically used as a gastric anti-ulcer drug. In primary human hepatocytes omeprazole potently induced cytochrome P4501A1 mRNA ex- pression, whereas this effect was not detected in mouse primary hepatocytes. In human hepatoma cells omepra- zole was found to induce transcription of reporter genes via the xenobiotic response element that is recognized by the ligand-activated dioxin receptor. In contrast, the human dioxin receptor was not activated by omeprazole upon expression in a receptor-deficient mouse hepatoma cell line. In a reconstituted yeast ( Saccharomyces cerevisiae ) model system, however, both the mouse and human dioxin receptors were potently activated by ome- prazole. Although omeprazole failed to displace dioxin in in vitro ligand binding assays,

The dioxin (also termed the aryl hydrocarbon) receptor is a ubiquitous intracellular protein that mediates signal transduction by the widespread and persistent environmental contaminant 2,3,7,8-tetrachlorodibenzo-p-dioxin (dioxin, TCCD 1 ; for recent reviews see Refs. 1 and 2). In various animal models, dioxins and structurally related compounds elicit a broad spectrum of biological effects ranging from hepato-, neuro-, and immunotoxic effects, reproductive and teratogenic disorders to carcinogenic effects (3,4). The dioxin receptor functions as a ligand-activated transcription factor that belongs to the basic helix-loop-helix (bHLH) family of DNA binding proteins (5,6). Within this large family of proteins the dioxin receptor is distinguished by a region of homology, the PAS domain, that is contiguous with the bHLH motif and is common to the Drosophila circadian rhythm regulatory protein period, the Drosophila developmental regulators trachealess and singleminded, the mammalian hypoxia-inducible factor-1␣, the constitutively active transcription factor Arnt, and a human putative Down syndrome-critical factor (7-10, and references therein).
Upon exposure to ligand the dioxin receptor acquires an increased affinity for xenobiotic responsive elements (XREs) that confer dioxin regulation on target genes, e.g. a battery of genes encoding drug metabolizing enzymes such as cytochrome P4501A1 (11). Activation of the receptor to an XRE binding form is accompanied by its apparent translocation from the cytoplasm to the nucleus of target cells (12, and references therein). The receptor activation pathway is not fully understood but is clearly a multistep process. In nonstimulated cells the receptor is recovered as a complex with the molecular chaperone hsp90. The ligand binding process induces release of hsp90 from the receptor (13) and concomitant recruitment of the bHLH/PAS factor Arnt, enabling both proteins to specifically recognize the XRE motif (14 -17). Individually neither the receptor nor Arnt show any detectable affinity for this target sequence (15).
Targeted disruption of the dioxin receptor gene in mice has indicated that the receptor may play a role in hepatic development (18,19). However, a physiological ligand of the dioxin receptor, if any, has not yet been identified. Among the known xenobiotic dioxin receptor ligands, all of which have a planar structure (3,20), there seems to exist a good correlation between the affinity of halogenated aromatic hydrocarbons, their in vitro dioxin receptor activating capacity (measured as in vitro induction of XRE binding activity of the receptor in cellular extracts), and their potency to induce transcription of the cytochrome P450IA1 gene in vivo (21). Other classes of dioxin receptor ligands include polycyclic aromatic hydrocarbons (e.g. benzo [a]pyrene and 3-methylcholanthrene), flavonoids, and indole derivatives, most notably indolo [3,2-b]carbazole, a dioxin receptor agonist of dietary origin that binds to the receptor with the same affinity as the prototypical dioxin TCDD (20,22,23, and references therein). Within the latter three classes of ligands, resistance to metabolism plays a very important role in the biological potency of the compound (23,24). Interestingly, omeprazole, a benzimidazole derivative clinically used as a gastric proton pump inhibitor (25)(26)(27) and references therein), has been demonstrated to induce expression of the dioxin receptor target gene cytochrome P4501A1 in both primary human hepatocytes (28), human hepatoma cells (29), tissues of the human alimentary tract (30), and human colon adenocarcinoma cells (31). In this context, the mechanism of action of omeprazole remains, however, unclear. Whereas omeprazole has been shown to induce XRE binding activity of the dioxin receptor in nuclear extracts of human hepatoma cells (29), the failure to detect any binding of omeprazole to the dioxin receptor has been reported (32). Thus, these data would argue against the simple model that omeprazole represents a dioxin receptor ligand.
In the present study we have investigated regulation of dioxin receptor function by omeprazole. We demonstrate that induction of cytochrome P4501A1 mRNA expression by this compound, and activation of a minimal XRE-driven reporter gene construct, was species-specific. Thus, the omeprazole-induced activation response was only observed in human but not mouse primary hepatocytes or hepatoma cell lines. In contrast, omeprazole potently activated both the mouse and human dioxin receptors in a reconstituted yeast model system. Strikingly, a specific point mutation that abrogates dioxin binding in vitro by the receptor inhibited the omeprazole activation response in yeast, strongly suggesting that receptor activation was mediated by the ligand binding domain. Moreover, omeprazole as well as dioxin responsiveness in yeast was totally impaired at low levels of expression of hsp90 that is a critical chaperone for ligand binding activity. We failed, however, to observe any binding by omeprazole in in vitro receptor ligand binding assays. In conclusion, dioxin receptor-mediated signal transduction by omeprazole appears to be regulated at two distinct levels. Although the mouse dioxin receptor is responsive to omeprazole treatment per se, the omeprazole activation response is repressed in mouse hepatocytes or hepatoma cells. Furthermore, omeprazole appears to require chemical conversion or metabolism in vivo to an active, planar principle of transient nature that regulates dioxin receptor function via the ligand binding domain, possibly as a novel class of receptor ligand.

MATERIALS AND METHODS
Cells, RNase Protection Assay, and Transient Transfection Experiments-Primary hepatocytes were prepared from human hepatic tissue fragments following lobectomy, or from C57BL/6 mouse liver as described (28,33,34), and maintained in culture in a 1:1 mixture of Ham's F12 and Williams' E media, supplemented as described (35). Mouse Hepa 1c1c7, Hepa 1c1c7 c12, and human HepG2 hepatoma cells were grown in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) as described (13). For RNase protection experiments cells were exposed to 100 M ␤-naphthoflavone or 100 M omeprazole for 15 h. Total RNA was isolated as described (36) and analyzed in RNase protection assays as described (37) using cytochrome P4501A1 antisense RNA probes in vitro synthesized from cDNA spanning either exon 2 of human cytochrome P4501A1 or exon 3 of mouse cytochrome P4501A1. The wildtype or point-mutated reporter gene constructs pXRE-MMTV-AP or pXMUT-MMTV-AP (15), respectively, were transiently transfected into 60 -80% confluent murine Hepa 1c1c7 or human HepG2 cells using cationic liposomes (N-[1-(2,3-dioleoyloxy)propyl]-N,N,N-trimethyl-ammonium methylsulfate, Boehringer Mannheim) according to the manufacturer's protocol. After a 12-h transfection period, cells were incubated with either 10 nM dioxin, 50 M omeprazole (Glaxo Research Group Ltd.), or vehicle alone (0.1% Me 2 SO) in fresh medium for 48 h. Secreted placental alkaline phosphatase levels produced by the reporter gene were determined by a colorimetric assay (38). Briefly, cell supernatants were heated at 65°C for 30 min and then incubated at 37°C in a mixture containing 200 mM Tris, pH 8.8, 275 mM NaCl, 0.5 mM MgCl 2 , and 5 mM p-nitrophenyl phosphate. Phosphatase levels were measured by absorbance at 405 nm. The dioxin receptor was transiently expressed in mutant hepatoma C12 cells (1) using pCMVDR (39), pTXDir (40) as an XRE-dependent luciferase reporter gene, and cationic liposomes (Lipofectin, Life Technologies, Inc.) as transfection agent.
Cell Extract Preparation and DNA Binding Assay-Cytosolic extracts were prepared from confluent Hepa 1c1c7 cells by homogenization of untreated cells in 1 volume of TEG buffer (1 mM EDTA, 20 mM Tris-HCl, pH 7.4, 10% (v/v) glycerol, 1 mM dithiothreitol), followed by centrifugation at 120,000 ϫ g for 45 min. The resulting supernatant was taken as the cytosolic fraction and used either immediately or frozen in small aliquots at Ϫ70°C. DNA binding activity of the dioxin receptor was monitored by a gel mobility shift assay performed essentially as described previously (41) using a double-stranded XRE oligonucleotide derived from the cytochrome P4501A1 gene as probe (42). Briefly, DNA binding reactions were assembled in the presence of 1 g of nonspecific poly(dI-dC) (Pharmacia Biotech Inc.) in 10 mM Hepes, pH 7.9, 5% (w/v) glycerol, 0.5 mM dithiothreitol, 2.5 mM MgCl 2 , 1 mM EDTA, 0.08 (w/v) Ficoll, in a final reaction volume of 20 l, and at a final concentration of 50 mM NaCl. The reactions were incubated for 30 min at 25°C, and protein-DNA complexes were resolved on 4% (acrylamide/ bis-acrylamide: 29/1) native polyacrylamide gels at 30 mA and 25°C using a Tris glycine EDTA buffer (41).
Ligand Binding Assay-pLBD-(230 -421) containing the minimal ligand binding domain of the mouse dioxin receptor (43) was used for in vitro translation in rabbit reticulocyte lysate (Promega) in the presence of either [ 35 S]methionine (Amersham Corp.) or unlabeled methionine, according to the manufacturer's protocol. In vitro translation mixtures (50 l) or Hepa 1c1c7 cell cytosolic extracts were incubated for 3 h at 0 -4°C with 50 l of TEG buffer containing 1 mM phenylmethylsulfonyl fluoride, 5 g/ml aprotinin, and [ 3 H]TCDD (40 Ci/mmol; Chemsyn, Lenexa, KS) at a final concentration of 2.5 nM in the absence or presence of a 400, 4,000, or 20,000-fold molar excess of competitor ligand. Ligand binding activity was analyzed by centrifugation on 5-20% (w/v) or 10 -40% (w/v) sucrose density gradients, respectively, prepared in TEG buffer containing 25 mM KCl and 20 mM sodium molybdate. Prior to sucrose gradient centrifugation, sodium molybdate was added to the samples to a final concentration of 20 mM, and the samples were treated with dextran-coated charcoal (13) for 5 min to remove unbound ligand. After rapid centrifugation the resulting supernatants were layered onto sucrose gradients that were centrifuged at 300,000 ϫ g for 15 h to a cumulative centrifugal force of 1.7 ϫ 10 12 rad 2 /s in a Beckman L8 -60 ultracentrifuge. Fractions were collected by gravity flow, starting from the bottom of the gradients, and analyzed for radioactivity by scintillation counting.
Yeast ␤-Galactosidase Assays-Single colonies were grown to saturation in selective medium containing either 2% galactose or 2% glucose at 25°C for the chimeric receptors and with 2% glucose at 30°C for the full-length dioxin receptor constructs. Cultures were diluted 1:6 in fresh medium and grown overnight in the absence or presence of either omeprazole or the high affinity dioxin receptor ligand indolo[3,2-b]carbazole, synthesized as described (48). Omeprazole analogues, timoprazole (TI), and desoxytimoprazole (D-TI), were prepared as described (49). Yeast cells were collected from the overnight culture by centrifugation and assayed for enzyme activity as described previously (50,51).
Yeast Extract Preparation and Immunoblot Analysis-Yeast cells expressing wild-type or point-mutated dioxin receptor forms were grown at 30°C in appropriate selection media containing 2% glucose. Extracts were prepared as described previously (52). For assaying expression levels of wild-type and mutant dioxin receptor forms, 20 g of total yeast extract protein was analyzed in immunoblot experiments using a dioxin receptor antiserum as described previously (15).

Regulation of Dioxin-responsive Reporter Genes by Omeprazole-
To investigate effects of omeprazole on dioxin receptor function in vivo, we transiently transfected mouse Hepa 1c1c7 and human HepG2 cells with the XRE-driven reporter gene construct pXRE-MMTV-AP. As schematically represented in Fig. 1A, this reporter gene contains a minimal mouse mammary tumor virus (MMTV) promoter in which endogenous glucocorticoid responsive elements have been substituted with a single XRE from the cytochrome P4501A1 gene (15). In Hepa 1c1c7 cells this reporter gene shows very low basal activity in the absence of ligand treatment and strong transcriptional activation following the addition of 10 nM dioxin (15; Fig. 1B). Interestingly, activation of the reporter gene was not detected upon addition of 50 M omeprazole to Hepa 1c1c7 cells (Fig.  1B). In contrast, transcription of the XRE-driven reporter gene was induced about 3-fold by omeprazole in human HepG2 cells. This induction response represented only a fraction of the maximal activation response (around 9 -10-fold induction) produced by dioxin treatment (Fig. 1B), possibly due to rapid metabolism of the inducing compound during the prolonged (24 -48 h) reporter gene assay. Consistent with the modest reporter gene activation response observed in the present study, omeprazole has been demonstrated to induce cytochrome P4501A1 mRNA levels in HepG2 cells to approximately 30% of the response observed with dioxin (29).

Activation of Reporter Gene Expression by Omeprazole Is Mediated by the Cytochrome P4501A1 XRE Sequence Motif-
Within the asymmetric P4501A1 XRE core motif TGCGTG (53; E box half-site shown in italics) a single point mutation (TGAGTG; mutation underlined) abrogates recognition of the XRE target sequence by the ligand-activated dioxin receptor in in vitro DNA binding assays (42) and dioxin receptor function in vivo (15). Given the XRE-driven reporter gene activation response that was produced by omeprazole treatment of human HepG2 cells, we also examined the effect of omeprazole on the activity of a reporter gene driven by a single copy of a mutated XRE motif, XRE mut (Fig. 1A), containing this single point mutation. In HepG2 cells, omeprazole failed to induce reporter gene activity from the XRE mut -driven promoter (Fig. 1B), demonstrating that the XRE sequence motif was the target for regulation and strongly implying the XRE binding dioxin receptor-Arnt complex in mediating the omeprazole induction response.
Regulation of Cytochrome P4501A1 Gene Expression by Omeprazole Is Species-specific-The failure to activate either the XRE-driven reporter gene construct (Fig. 1B) or cytochrome P4501A1 mRNA levels (data not shown) by omeprazole treatment of mouse Hepa 1c1c7 cells prompted us to examine if the absence of this response was due to the loss of a possible differentiation-sensitive cofactor. We therefore compared by a sensitive RNase protection assay (37) regulation of cytochrome P4501A1 mRNA expression by omeprazole in human primary hepatocytes and primary hepatocytes from the dioxin-and polycyclic aromatic hydrocarbon-responsive (3) mouse strain C57BL/6. In human or mouse primary hepatocytes treated with vehicle (Me 2 SO) alone, no basal expression levels of cytochrome P4501A1 mRNA were detected (Fig. 2, A and B, lane 2). As demonstrated earlier (28,33) exposure of human primary hepatocytes to 50 M omeprazole resulted in potent induction of cytochrome P4501A1 mRNA steady-state levels. A similar induction response was produced by an identical dose of the well characterized (3, 4) dioxin receptor ligand ␤-naphthoflavone ( Fig. 2A, compare lanes 2-4). In contrast, no induction of cytochrome P4501A1 mRNA levels was detected upon incubation of mouse primary C57BL/6 hepatocytes with increasing concentrations (up to 50 M) of omeprazole. In control experiments, cytochrome P4501A1 mRNA levels were maximally induced in the mouse primary hepatocytes by 10 M ␤-naphthoflavone (Fig. 2B, compare lanes 2-6). Thus, although the XRE reporter gene experiments strongly implicated the dioxin receptor in mediating the omeprazole activation response in HepG2 cells, the dioxin receptor endogenous to mouse primary hepatocytes or hepatoma cells was not activated by omeprazole, indicating striking species-specific differences in omeprazole sensitivity.

Regulation by Omeprazole of Human Dioxin Receptor Function upon Expression in Mouse Hepatoma
Cells-To further examine species-specific differences in omeprazole regulation of mouse versus human dioxin receptor function, we transiently expressed the human dioxin receptor in mutant mouse hepatoma C12 cells that express very low levels of dioxin receptor and are functionally dioxin-nonresponsive (1). Whereas this procedure reconstituted induction of XRE-dependent reporter gene activity by dioxin, no activation of the reporter gene was seen following omeprazole treatment for 24 h (Fig. 3). Since omeprazole is biologically unstable and rapidly and extensively metabolized in both rodents and humans (54, 55, and references therein), we also analyzed reporter gene activity after 8 and 12 h of omeprazole treatment but failed to observe any induction response (data not shown). These results demonstrate that activation of human dioxin receptor function by omeprazole is repressed in a mouse cell environment.

Regulation of Human and Mouse Dioxin Receptor Function by Omeprazole in a Reconstituted Yeast Model System-Given
the rapid metabolism of omeprazole, we next examined the effect of omeprazole on dioxin receptor function in a cellular environment that is metabolically more neutral than cells of hepatic origin. We first used a reconstituted yeast (Saccharomyces cerevisiae) model system where either the mouse or human full-length dioxin receptor was expressed in the presence of Arnt and a ␤-galactosidase reporter gene driven by the yeast CYC1 promoter and dimerized copies of the cytochrome P450 XRE sequence motif. 2 These reconstituted systems were highly inducible by the high affinity dioxin receptor ligand indolo[3,2-b]carbazole (ICZ) with an EC 50 value estimated to be approximately 0.5 and 5 nM, respectively, for both the mouse (Fig. 4B) and human (Fig. 4C) dioxin receptors. Thus, this yeast model system very faithfully reproduced the high affinity (K d Ϸ1 nM) with which ICZ binds the dioxin receptor in ligand binding assays in vitro (20, and references therein). In comparison, omeprazole also potently activated function of both the mouse and human dioxin receptor-Arnt complexes (Fig. 4A). This activation response was mediated by the dioxin receptor-Arnt complex since it was not observed in yeast cells only transformed with the reporter gene (Fig. 4A). Omeprazole yielded EC 50 values of 5-10 M for both the mouse (Fig. 4B) and human (Fig. 4C) dioxin receptors. Thus, in striking contrast to mouse primary hepatocytes or hepatoma cells, the mouse dioxin receptor was omeprazole-responsive following expression in yeast cells. The obtained EC 50 values are at least 1 order of magnitude lower than the EC 50 value reported for cytochrome P4501A1 promoter activation in HepG2 cells (ϳ100 M; 29), possibly due to a decreased rate of metabolism of omeprazole in yeast cells.
The Dioxin Receptor Is the Target for Regulation by Omeprazole-The yeast experiments described above demonstrated that both the mouse and human dioxin receptor-Arnt complexes were activated by omeprazole. However, this experimental system did not provide an answer as to whether the dioxin receptor, Arnt, or both transcription factors were regulated by omeprazole. To critically examine specific regulation of dioxin receptor function by omeprazole, we therefore used yeast strains that express chimeric dioxin receptor constructs and, in the presence of galactose, wild-type levels of the molecular chaperone hsp90 (44). As outlined in Fig. 5A, the chimeric receptors, GRDBD/mDR83-805 and GRDBD/hDR84-848, respectively, contain a heterologous DNA binding domain, the zinc finger motif of the human glucocorticoid receptor, fused to structures of the mouse or human dioxin receptors that lack the bHLH dimerization and DNA binding motif but span regions important for ligand binding and thus conditional regulation (56). These chimeric receptors are functionally uncoupled from the ubiquitous bHLH/PAS partner factor Arnt since they show ligand responsiveness in Arnt-deficient mutant hepatoma cells (56) and in yeast cells that do not express endogenous Arnt (44,57).
Exposure to omeprazole did not alter the low levels of constitutive functional activity produced by the glucocorticoid receptor DNA binding domain on a glucocorticoid response element-driven ␤-galactosidase reporter gene (Fig. 5B). However, when the yeast cells were grown in the presence of galactose and thus expressed wild-type levels of hsp90 (44), both the mouse and human dioxin-glucocorticoid receptor fusion proteins were activated by omeprazole (Fig. 5B). As a reference, ICZ treatment produced a strong activation response with both chimeras (Fig. 5B). These experiments demonstrate that structures spanning the ligand binding domain of both the mouse and human dioxin receptors mediate responsiveness to omeprazole.
Regulation of Dioxin Receptor Function by Omeprazole Requires hsp90 -The chimeric dioxin receptor constructs were next expressed in yeast cells in the presence of glucose, i.e. conditions yielding very low levels (about 5% of wild-type levels) of hsp90 expression (44,58). Under these conditions, activation of mouse and human dioxin receptor function by omeprazole was totally abrogated (Fig. 5B). In a similar fashion, the receptor was unresponsive to ICZ or ␤-naphthoflavone in the presence of low levels of hsp90 (44, 57, Fig. 5B). Since the molecular chaperone hsp90 has been demonstrated to be required for ligand binding activity of the dioxin receptor in vitro (59,60), these results imply that omeprazole regulates dioxin receptor function via a ligand binding mechanism.
Omeprazole Fails to Bind the Dioxin Receptor-It has previously been reported that omeprazole does not bind to the dioxin receptor in extracts from primary human hepatocytes in vitro (32). This apparent contradiction to the present results prompted us to further explore the possibility that omeprazole may bind to the dioxin receptor. We initially performed ligand binding studies with cytosolic extracts from mouse Hepa 1c1c7 cells. These extracts were labeled with [ 3 H]dioxin in the absence or presence of a 400-fold molar excess of the unlabeled high affinity receptor ligand indolo[3,2-b]carbazole, ICZ, or 4,000 to 20,000-fold molar excesses of omeprazole, i.e. concentrations producing maximal induction in the yeast model system. Upon sucrose density gradient centrifugation the [ 3 H]dioxin-labeled dioxin receptor was recovered in the ϳ9 S region of the gradient (Fig. 6A), consistent with the sedimentation properties of the ϳ300-kDa dioxin receptor-hsp90 complex (13). In contrast to results obtained with ICZ, omeprazole did not displace [ 3 H]dioxin from the dioxin receptor (Fig. 6A). We also used a fast hydroxylapatite adsorption assay to separate protein-bound [ 3 H]dioxin from free ligand. In this assay incubation

FIG. 4. Reconstitution of dioxin receptor function in yeast.
A, activation of full-length mouse or human dioxin receptor function in yeast by omeprazole. Arnt was co-expressed with either mouse or human dioxin receptor in yeast cells transformed with lacZ reporter gene (XRE-CYC1-lacZ) driven by tandem copies of the cytochrome P4501A1 XRE motif and the yeast CYC1 promoter. Yeast cultures were grown in medium containing 2% glucose and incubated with 10 M omeprazole (OM) or 10 nM ICZ and assayed for ␤-galactosidase activity. The effect of omeprazole and ICZ was also examined using yeast cells transformed with the reporter gene alone, as indicated. Dose-dependent activation of mouse (B) and human (C) dioxin receptors by omeprazole. Yeast cells transformed with reporter gene, Arnt, and mouse or human dioxin receptor were exposed to increasing concentrations of ICZ (squares) and omeprazole (diamonds) and assayed for ␤-galactosidase activity. Results of a representative experiment are shown. of the cytosolic extract from nontreated Hepa 1c1c7 cells with [ 3 H]dioxin in the absence or presence of an excess of TCCD or ICZ demonstrated specific binding of [ 3 H]dioxin, but we observed no competition for binding by omeprazole (data not shown).
In additional ligand binding experiments we expressed the minimal ligand binding domain of the dioxin receptor, DR230-421 (43, 56, schematically represented in Fig. 6B), by in vitro translation in reticulocyte lysate. Consistent with the results obtained in crude cytosolic extracts from Hepa1c1c7 cells, [ 3 H]dioxin binding to the minimal ligand binding domain was not displaced by an excess of omeprazole, as assessed by sucrose gradient sedimentation (Fig. 6B). Thus, we did not detect any affinity of omeprazole for the mouse dioxin receptor in a series of different ligand binding assays. In agreement with the present results, omeprazole has also been reported to fail to displace  ). B, omeprazole activates the chimeric dioxin receptors in an hsp90-dependent fashion. Yeast cells transformed with a glucocorticoid response element-driven ␤-galactosidase reporter gene together with either GRDBD or the chimeric receptors GRDBD/mDR83-805 or GRDBD/hDR84-848 were grown in a medium containing 2% galactose (yielding wild-type expression levels of hsp90) or 2% glucose (resulting in low expression levels of hsp90) and exposed to 50 M omeprazole, 10 nM ICZ, or vehicle alone (dimethyl sulfoxide, DMSO), as indicated, and assayed for ␤-galactosidase activity after 24 h of incubation. Results of a representative experiment are shown. represented in Fig. 7A, Val 381 within the ligand binding domain of the human dioxin receptor has been mutated to either Ala or Asp. The Val 3 Asp mutation totally abrogates ligand binding activity in vitro, whereas the Val 3 Ala mutation rather appears to result in a 2-5-fold increase in ligand binding affinity (45). By analogy the corresponding residue (375) in the C57BL/6 mouse dioxin receptor is Ala (45), and importantly, this receptor binds dioxin in vitro with about 1 order of magnitude higher affinity than the DBA/2 mouse dioxin receptor (61) which contains Val in position 375 (45). Consistent with these in vitro observations the Val 381 3 Ala mutation of the human dioxin receptor did not negatively affect the ICZ activation response in the yeast reconstituted model system (Fig.  7B). In contrast, this activation response was abolished when the dioxin receptor form containing the Val 381 3 Asp mutation was expressed in yeast (Fig. 7B). Surprisingly, the effect of these mutations on omeprazole responsiveness mirrored those observed with ICZ, suggesting that Val 381 is a critical residue not only for ICZ recognition but also for omeprazole signaling. In fact, the Val 381 3 Ala mutation appeared to increase omeprazole responsiveness, in agreement with its increased ligand binding affinity in vitro (45). Importantly, immunoblot analysis demonstrated very similar expression levels of the wild-type receptor relative to the responsive Val 381 3 Ala mutant, whereas the nonresponsive Val 381 3 Asp mutant receptor was expressed at slightly higher levels (Fig. 7C). Taken together these data strongly suggest that activation of dioxin receptor function by omeprazole is mediated by ligand recognition.

Structure-Activity Relationships in Dioxin Receptor Activation by
Omeprazole-Omeprazole has three important structural elements: a substituted benzimidazole ring, a substituted pyridine ring, and a sulfoxide-containing chain connecting the 2 positions of the two ring systems (Fig. 8A). A highly labile sulfenamide structure (Fig. 8A), generated by acid decomposition of omeprazole, has been proposed to be the active gastric acid secretion inhibitor formed in vivo (26). Interestingly, this conversion product shows planarity, a structural hallmark of dioxin receptor ligands (3,20). This compound is, however, very labile and of transient nature (26). Consistent with these properties, we have failed to induce the DNA binding activity of the cytosolic mouse dioxin receptor activity in a highly sensitive in vitro gel mobility shift assay using extracts from human HepG2 cells pretreated with 50 M omeprazole for 2 and 4 h, respectively (data not shown). To examine the importance of the hypothetical planar sulfenamide for dioxin receptor activation, we therefore tested the activity of omeprazole analogues containing either a sulfoxide (timoprazole, TI) or a sulfide (desoxytimoprazole, D-TI) group (Fig. 8A) but lacking substituents. To achieve ring closure omeprazole undergoes C-S bond cleavage and aromatization, and finally, the oxygen of the sulfoxide is protonized forming sulfenic acid. The subsequent formation of the sulfenamide is due to known reactions between sulfenic acids and amines (26, and references therein). Because of the loss of the oxygen atom the sulfide-containing form (D-TI) is unable to form the ring structure.
In the reconstituted yeast model system, omeprazole and TI showed very similar dose dependences in activating the dioxin receptor (Fig. 8B), demonstrating that the substituent groups were not important for generating this response. In excellent agreement with this observation, lansoprazole, a gastric antiulcer drug that chemically only differs from omeprazole with regard to substituents, potently activated dioxin receptor function in yeast cells (data not shown) and induced cytochrome P4501A1 mRNA expression in hepatocytes (33). Interestingly, however, D-TI was inert at the lowest tested doses (5 and 10 M) and showed activity only at the highest tested dose: 50 M (Fig. 8B). Thus, the sulfoxide group is critical for dioxin receptor activation. The ability of D-TI to activate the receptor at high doses is most probably due to metabolism generating the sulfoxide form, thereby enabling it to undergo the same reaction scheme as described for omeprazole. In strict correlation with these results, the sulfoxide chain has been reported to be essential for the antisecretory effect of omeprazole, whereas the sulfide group only has a low activity (62, 63). On the basis of these data we propose that the planar sulfenamide structure is a possible ligand of the dioxin receptor that is formed in vivo from omeprazole. DISCUSSION In the present report we have used a reconstituted yeast model system to demonstrate that omeprazole potently activated both mouse and human dioxin receptor function in a dose-dependent manner via a mechanism mediated by the ligand binding domain of the receptor. In agreement with these data, transfection experiments using minimal reporter gene constructs indicated that the omeprazole activation response in human hepatoma cells was mediated via the core XRE motif that is recognized by the ligand-activated dioxin receptor-Arnt complex. Strikingly, however, we failed to observe omeprazoleinduced reporter gene activation in mouse hepatoma cells or induction by omeprazole of cytochrome P4501A1 mRNA steady-state levels in either mouse primary hepatocytes or mouse hepatoma cells, suggesting that omeprazole responsiveness in liver cells may be species-specific. In strong support of this model, activation of the human dioxin receptor by omeprazole was repressed following expression of this receptor in a mutant mouse hepatoma cell line expressing very low levels of endogenous mouse dioxin receptor.
Is Omeprazole a Dioxin Receptor Ligand?-It has not been possible to demonstrate any affinity of omeprazole for either the human (31,32) or mouse (Fig. 6) dioxin receptors in a number of different in vitro assays including ligand binding experiments or assays monitoring induction of receptor-dependent DNA binding activity, and it has recently been proposed that benzimidazole derivatives such as omeprazole may induce cytochrome P4501A1 expression by ligand-independent mechanisms (64). In contrast to this model, we present here four independent lines of evidence based on in vivo experiments that, taken together, suggest that activation of the dioxin receptor by omeprazole is mediated via ligand recognition by the ligand binding domain. First, chimeric receptor experiments indicate that omeprazole responsiveness is mediated by the ligand binding domain of the receptor. Second, the omeprazole activation response was abrogated upon down-regulation in yeast cells of the expression levels of the molecular chaperone hsp90 which has previously been demonstrated to be required for folding of a high affinity ligand binding conformation of the dioxin receptor in vitro (59,60) and in vivo (44,57). Third, point mutation experiments demonstrated that omeprazole responsiveness was dependent on a residue, Val 381 , within the ligand binding domain that is essential for ligand binding activity in vitro (45). Whereas the Val 381 3 Asp mutation inhibits ligand binding activity, Val 381 3 Ala increases the ligand binding affinity in vitro (45). In excellent agreement with these in vitro observations, the same mutations either inhibited or enhanced in vivo responsiveness of the human dioxin receptor to the classical high affinity receptor ligand ICZ. Importantly, the effect of these mutations on ICZ responsiveness strictly correlated with omeprazole responsiveness of the receptor. Finally, structure-activity relationship experiments indicated that omeprazole responsiveness of the dioxin receptor was dependent on the ability of this compound to adopt planarity, a critical common property of dioxin receptor ligands (3,20).
These data strongly argue that omeprazole activates the dioxin receptor via a ligand binding mechanism in vivo. Since parental omeprazole is inert in binding to the receptor in vitro, a specific metabolic pathway or chemical rearrangement appears to be required to generate the active principle that regulates receptor function, most likely by occupation of the ligand binding domain of the receptor in vivo. The present experiments indicate that the putative ligand may be a planar sulfenamide structure, i.e. the theoretical active gastric acid secretion inhibitor formed in vivo from omeprazole (26,63). Given the highly reactive and transient nature of this compound, it has hitherto not been possible to synthesize it or to "trap" it in conditioned medium or cells. In summary, our results raise the possibility that benzimidazole derivatives such as omeprazole and lansoprazole represent precursors of a novel class of possibly biologically important dioxin receptor agonists. In view of the indicated role of the dioxin receptor in normal liver growth and development (18,19), it may be interesting to examine if these clinically widely used compounds modulate physiological regulatory pathways of the dioxin receptor in the liver and other target tissues, e.g. within the gastrointestinal tract.
Differences in Signal Transduction by Omeprazole between Mouse and Man-Interestingly, our results define two distinct levels of regulation of omeprazole signaling. In addition to omeprazole possibly representing a novel class of dioxin receptor ligands with the ability to activate both the mouse and human dioxin receptor forms, it is striking that neither dioxin receptor function nor expression of the dioxin receptor target gene cytochrome P4501A1 were induced by omeprazole in primary C57BL/6 hepatocytes and Hepa 1c1c7 cells. Moreover, omeprazole responsiveness of the human dioxin receptor was repressed in mouse Hepa-1 c12 cells. Thus, the omeprazole activation response is also determined by cellular factors other than the receptor. Since formation of the sulfenamide structure is critical for the activation response, it is tempting to speculate that cellular differences in the ability to chemically convert the parental compound to the planar one could determine the inability of dioxin receptor to respond to omeprazole in mouse hepatic cells. Alternatively, the very labile sulfenamide structure is so transient in mouse hepatoma cells that it fails to elict a response. In gastric parietal cells that represent one of the few examples of the cells with a low pH value, the sulfenamide structure is most likely generated by acid decomposition (26, and references therein). However, there is a paucity of data on other cellular processes generating this product. To understand the complex biology of both omeprazole and the dioxin receptor, it is therefore now important to not only identify the putative novel class of ligands derived from omeprazole and structurally related compounds but also to dissect the omeprazole signaling pathway to characterize molecular determinants of species specificity of omeprazole action in hepatic tissues.