An Aryl Hydrocarbon Receptor Odyssey to the Shores of Toxicology: The Deichmann Lecture, International Congress of Toxicology-XI
Department of Pharmacology, Medical Sciences Building, University of Toronto, 1 King's College Circle, Toronto, Ontario, Canada M5S 1A8
1 To whom correspondence should be addressed. Fax: (416) 978-6395. E-mail: allan.okey{at}utoronto.ca.
Received February 22, 2007; accepted April 17, 2007
 |
ABSTRACT |
|---|
 TOP ABSTRACT PRELUDE AND DISCLAIMERARYL HYDROCARBON RECEPTOR: THE...AHR: THE ESSENTIAL MEDIATOR...Conclusion: Return on investmentREFERENCES |
|---|
The science of toxicology is devoted, in large part, to understanding mechanisms of toxicity so that we can more accurately assess the risk posed by exposure to xenobiotic agents and, perhaps, intervene in the toxicologic process to mitigate harm. Dioxin-like chemicals continue to be of great concern as environmental toxicants. About 30 years ago the aryl hydrocarbon receptor (AHR) was discovered as a specific binding site for 2,3,7,8-tetrachlorodibenzo-p-dioxin. This giant step led to our current view that essentially all toxic effects of dioxins are AHR-mediated. The AHR serves as the archetype for understanding toxicity mediated by other soluble receptors. The fact that toxicity is receptor-mediated has important implications, especially for dose–response relationships. In laboratory animals genetic differences in AHR gene structure lead to profound differences in responsiveness to dioxin-like chemicals. Humans, however, exhibit relatively few AHR polymorphisms and these seem to exert only modest effects on downstream events. Dioxin toxicity is fundamentally due to AHR-mediated dysregulation of gene expression. Our current challenging goal is to determine which dysregulated genes underlie specific forms of dioxin toxicity. Mapping AHR-mediated gene expression in a variety of biological systems may help explain why dramatic differences in susceptibility to dioxin toxicity exist among laboratory species and why humans appear to be relatively resistant to adverse effects of dioxins.
Key Words: aryl hydrocarbon receptor; dioxin; 2,3,7,8-tetrachlorodibenzo-p-dioxin; gene regulation.
|
PRELUDE AND DISCLAIMER |
|---|
|
TOPABSTRACTPRELUDE AND DISCLAIMERARYL HYDROCARBON RECEPTOR: THE...AHR: THE ESSENTIAL MEDIATOR...Conclusion: Return on investmentREFERENCES |
|---|
This story is a chronicle of the conception, birth, and growth of aryl hydrocarbon receptor (AHR) research in relation to toxicology with a focus on evolution of AHR investigations in my own laboratory. I ask forbearance from readers as well as from fellow toilers in the AHR field if work from my own laboratory seems overemphasized or overrepresented. This article is intended to give a flavor of various aspects of AHR research rather than a definitive and comprehensive review of each topic and to be prospective as well as retrospective. Due to space constraints it won't be possible to cite the bulk of the important original contributions made by the numerous laboratories who study the AHR. Please see the excellent reviews by other investigators, cited in this paper, which provide additional perspectives on the multitudinous facets of AHR structure and function.
Before we consider recent developments in the AHR arena, it may be useful for newcomers to have an overview of how the AHR field arrived at its current state, illustrated, mainly, by my own research journey. Journeys are more enjoyable when they are taken with affable companions. It's been my good fortune during the AHR voyage to have been accompanied by many talented and congenial trainees and collaborators. This narrative is intended as a tribute to members of my laboratory who contributed so significantly to understanding this intriguing receptor as well as to the international community of AHR scholars with whom I've been privileged to interact over the past 30 years.
Research and research careers do not always proceed in a straightforward, logical, linear fashion. They evolve, just as do biological systems, via the force of natural selection picking from among the range of variants. Along the way, there are serendipitous or fortuitous events (akin to mutations—some good, some not-so-good) that strongly influence the direction and the success of our research (Jensen, 2004
; Rothstein, 1986). In my case, two lines of research (which on the surface seemed independent) coalesced, partly by chance, into a 30-year infatuation with the AHR.
Providential Connections: Estrogens and Polycyclic Aromatic Hydrocarbons
My original research interest was induction of mammary cancer by exogenous estrogens. In my Ph.D. thesis research, directed by Prof. George Gass, we found that continuous low-dose administration of the potent synthetic estrogen, diethylstilbestrol (DES), induced mammary carcinoma in more than 95% of mice if they carry the mouse mammary tumor virus. If mice lack either the virus or the estrogen stimulus, tumor incidence is very low (Gass et al., 1974; Okey and Gass, 1968).
DES was synthesized in the 1930s from precursors obtained from coal tar (Dodds et al., 1938). Coal tar (not so coincidentally) happens also to be the original source from which another notorious group of carcinogens was isolated, the polycyclic aromatic hydrocarbons (PAHs), typified by benzo[a]pyrene (BP).
Although DES can be carcinogenic in animal models (and in humans exposed prenatally to high doses—"DES daughters") (Herbst et al., 1971), DES also found its way into cancer therapy. Charles Huggins was awarded the Nobel Prize in Physiology or Medicine in 1966 for his discovery that DES and other hormonal therapies have value in some cases of advanced prostate cancer. Huggins also explored hormonal therapy for breast cancer. To this end, his laboratory was a leader in developing a remarkable animal model by showing that PAHs such as 7,12-dimethylbenz[a]anthracene (DMBA) and 3-methylcholanthrene (3-MC) are superb mammary carcinogens. These PAHs induce adenocarcinomas in young female rats within only a few weeks after a single dose (Dao and Sunderland, 1959; Huggins et al., 1961).
When I took up my first independent research position at the University of Windsor (Ontario, Canada), I pursued my interest in the respective roles of the estrogen and the PAH in the dramatic Huggins model of PAH-induced mammary cancer. It was bemusing to me, as a new investigator, to find that there were two (apparently diametrically opposed) mechanistic explanations being floated about for mammary carcinogenesis. One school held that PAHs were carcinogenic because they mimic the action of estrogens on mammary epithelium. A concurrent and competing view flipped this around to propose that estrogens are carcinogenic because they mimic PAHs—that is, estrogens are bioactivated into mutagenic metabolites. Later it would become established that there is a core of truth in each of these views (see Belous et al., 2007).
Steroids and carcinogenic PAHs bear a passing structural similarity (Fig. 1). The first known steroid receptor, the estrogen receptor (ER), was discovered through pioneering work by Elwood Jensen in the late 1960s (see Jensen, 2004). David Keightley, my first Ph.D. student, tested the supremely potent mammary carcinogen, DMBA, to see if it could interact with the ER and found that DMBA did not compete with estradiol-17ß for binding to the ER (Keightley and Okey, 1973). DMBA was not a good estrogen but we found that DMBA could interfere with some ER functions in vivo (Ianicello and Okey, 1976; Keightley and Okey, 1974). At that time we had not foreseen that a receptor which behaves very much like a steroid receptor might exist for the PAHs themselves. We will revisit the estrogen/PAH/AHR story near the end of this review.
|
Our research on estrogens in relation to mammary carcinogenesis took on a new dimension when the environmental movement was spurred in the late 1960s by the disturbing discovery that pesticides such as dichloro-diphenyl-trichloroethane (DDT) interfere with reproduction in birds (Bitman et al., 1968
; Peakall, 1967) and mammals (Bitman et al., 1968). Since we were an "estrogen lab," John Clement, a graduate student, tested DDT for estrogenic activity in standard bioassays and found that the o,p'-DDT isomer has estrogenic activity but that the commercial mixture (which predominantly is the p,p'-DDT isomer) exerts "antiestrogenic" effects and interferes with uterotrophic activity of the natural estrogen, estradiol-17ß (Clement and Okey, 1972).
Because DDT exerts both estrogenic and antiestrogenic effects, I wanted to find out how this ubiquitous pesticide would affect development of estrogen-dependent mammary cancer in rats treated with a PAH. Perhaps this "real-world" exposure to both a pesticide and a PAH carcinogen would be disastrous to the recipient. In fact, my graduate student, Charles Silinskas, found that brief pretreatment of female rats with DDT, at doses as low as 10 ppm, in the diet confers dramatic protection from DMBA-induced mammary cancer and leukemia (Silinskas and Okey, 1975). This protection appeared to be due to the ability of DDT to enhance metabolism and elimination of DMBA (Okey, 1972).
I became intrigued with the phenomenon of induction of "drug-metabolizing enzymes." I wanted to understand the induction mechanism and how induction might relate to protection from environmental carcinogens. It was serendipitous that I was eligible for a sabbatical leave and I was fortunate that Daniel Nebert was willing to accept this unknown investigator from a small provincial university into his laboratory, then at the National Institute for Child Health and Human Development, National Institutes of Health (NIH).
Before I continue the tale of my own work during the nascent days of AHR, it's necessary to jump back to events beginning in the 1950s that paved the way for discovery of this captivating receptor.
|
ARYL HYDROCARBON RECEPTOR: THE EARLY YEARS—FROM CONCEPT TO CLONING |
|---|
|
TOPABSTRACTPRELUDE AND DISCLAIMERARYL HYDROCARBON RECEPTOR: THE...AHR: THE ESSENTIAL MEDIATOR...Conclusion: Return on investmentREFERENCES |
|---|
Discovery of "MC-type" Induction Lights the Path
Many important nuclear receptors were discovered by "reverse endocrinology." That is, clones that harbor sequences similar to those of known nuclear receptors were retrieved from complementary DNA (cDNA) libraries. However, the protein products of these novel clones remained "orphan receptors" until their ligands and functions eventually were deciphered. This genetic pathway to discovery yielded several nuclear receptors whose acronyms now are widely known in pharmacology and toxicology: RAR, RXR, LXR, PPAR, CAR, PXR, and FXR (reviewed in Evans, 2004
; Giguere, 1999; Gustafsson, 1999; Kliewer et al., 1999).
In contrast to the strategy of "clone first, find function later," the AHR's discovery preceded the era of receptor cloning and resulted from efforts to understand the mechanism by which polycyclic hydrocarbons induce their own metabolism. Allan Conney, working in the Millers' laboratory at the University of Wisconsin in the 1950s, discovered that BP and 3-MC induce what then was called "BP hydroxylase." (Later "BP hydroxylase" was designated aryl hydrocarbon hydroxylase [AHH] in recognition of the fact that many PAHs in addition to BP are substrates (Nebert and Bausserman, 1970b). Molecular investigations eventually linked AHH activity to CYP1 enzymes.) The phenomenon of "MC-type" induction by PAHs was an essential antecedent to discovery of the AHR. See the engaging autobiographical sketch by Conney (2003b) for a full account of the circumstances which led to his discovery of MC-type induction.
Genetic Models Plant a Seed that Will Yield a Bountiful Harvest
Although Prof. Werner Kalow at the University of Toronto had published his landmark monograph Pharmacogenetics in 1962 (Kalow, 1962), the possibility that genetic factors might actually matter in drug metabolism and drug response was still not on the radar screen for most pharmacologists and toxicologists by 1970. Today pharmacogenetics and pharmacogenomics have fully penetrated biomedical research as well as the pharmaceutical industry; their importance may seem obvious to younger investigators but this is a relatively recent enlightenment.
Cell models of AHH induction.
Daniel Nebert was an "early adopter" who, at the start of the 1970s, had developed a well-honed appreciation for the potential power of a genetic approach to pharmacology and toxicology. Working initially in Harry Gelboin's laboratory at the National Cancer Institute-NIH, Dan Nebert found that induction of BP hydroxylase/AHH is not confined to liver of intact animals. He developed a very informative induction model in hamster fetal cell cultures where many fundamental characteristics of the induction process were worked out (Nebert and Bausserman, 1970a; Nebert and Gelboin, 1968a,b).
A previous in vivo survey pointed to substantial differences between mouse strains in AHH inducibility by 3-MC (Nebert and Gelboin, 1969). As a new independent investigator at the National Institute of Child Health and Human Development, Dan demonstrated that AHH activity was much more highly inducible in fetal cells derived from C57BL/6 mice than cells from DBA/2 mice (Nebert and Bausserman, 1970b), establishing the utility of cell models for exploring pharmacogenetic aspects of AHH regulation.
In the late 1970s, Oliver Hankinson (University of California) brought a powerful new tool—somatic cell genetics—to bear on mechanisms regulating induction of AHH. (AHH, by then was becoming associated with CYP1A1.) The Hankinson laboratory exposed Hepa-1 mouse hepatoma cells to BP in culture. BP induces AHH activity in wildtype Hepa-1 cells, thereby causing them to self-destruct because the induced enzyme bioactivates BP into cytotoxic metabolites. The rare mutant cells in the population that are not AHH-inducible survive; these resistant cells then were selected for further study to determine the basis of their nonresponsiveness. The BP-selection process (Hankinson, 1979) and subsequent genetic complementation analyses revealed that, in addition to the Cyp1a1 gene itself, products of at least two other genes are required to induce CYP1A1. One of these genes encodes the AHR (Legraverend et al., 1982).
The biggest payoff from their somatic cell genetic strategy in the Hankinson laboratory was identification of the other key regulatory gene product. That is, discovery of a novel protein, ARNT (aryl hydrocarbon receptor nuclear translocator), which would turn out to be the essential dimerization partner for the AHR (Hoffman et al., 1991; Reyes et al., 1992) (see below). Discovery of ARNT triggered explosive growth in the AHR field per se as well as in areas such as hypoxic signaling where ARNT (also known as HIF-1ß) plays a vital role (Fryer and Simon, 2006) and extending into such diverse areas as vascular tumorigenesis (Rankin et al., 2005) and type 2 diabetes (Gunton et al., 2005). Cloning of ARNT and subsequent cloning of the AHR (see below) were instrumental in unveiling an entire family of regulatory proteins containing bHLH/PAS domains (Gu et al., 2000).
James Whitlock Jr's laboratory (Stanford University) used a fluorescence-activated cell sorter to select cells that are unresponsive to induction by BP and derived AHR-deficient mutant cell lines akin to those produced by the Hankinson laboratory (Miller and Whitlock, 1981; Miller et al., 1983). The AHR-deficient mutant Hepa-1 cells produced by the Hankinson laboratory and the Whitlock laboratory have been invaluable to other investigators who use them to determine if a particular response requires the AHR.
In vivo model—the mouse Ah locus.
The original Nebert discovery of a strain difference in AHH induction laid the foundation for a classic genetic approach to inheritance of AHH regulation in vivo. Breeding studies in the Nebert laboratory showed that inheritance of inducibility essentially is an autosomal dominant trait. The genetic locus controlling induction was defined as Ah for aromatic hydrocarbon responsiveness (reviewed in Nebert, 1988, 1989; Nebert et al., 1981). C57BL/6 mice (Ahrb1 allele in current nomenclature) constitute the prototype "responsive" strain and DBA/2 mice (Ahrd allele) the prototype "nonresponsive" strain. The Nebert laboratory and many other laboratories would go on to show that genetic differences at the Ah locus (now termed the Ahr locus) influence sensitivity of mice to a very broad range of responses to xenobiotic chemicals including mutagenesis, carcinogenesis, teratogenesis, and dioxin toxicity (summarized in Nebert, 1989; Okey et al., 2005b).
The Induction-Receptor Hypothesis Arises
Marshall McLuhan, media guru at the University of Toronto, inverted the dictum "seeing is believing" to: "If I hadn't believed it, I wouldn't have seen it." In other words, for some discoveries, having the conviction that a particular phenomenon exists is the precondition that permits us to recognize evidence which supports that phenomenon. For the AHR, the "belief" that there was an induction receptor was based on a combination of genetic findings along with the precedent of receptors for other small hydrophobic molecules (i.e., steroids) which was burgeoning in the 1970s.
The first hint that AHH induction might be mediated by a receptor goes back to the statement by Nebert and Bausserman (1970a) who proposed that: "the process of hydroxylase induction involves a rate-limiting step, which may be the saturation of ‘inducer-binding’ sites in the cell."
2,3,7,8-Tetrachlorodibenzo-p-dioxin, a Super-potent AHH Inducer, Becomes the Ideal Radioligand for the Receptor Search
The plausibility of the "induction-receptor" hypothesis took a great leap forward because of 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD). "Nonresponsive" mouse strains are so called because they do not exhibit AHH induction when treated with nonhalogenated PAHs such as 3-MC, even at very high doses. TCDD shifted the focus of investigation away from the enzyme and toward pathways that regulate AHH expression. TCDD would soon establish that the "Ah locus" is a regulatory locus that encodes the AHR.
Alan Poland's laboratory (then at the University of Rochester) was devoted to determining mechanisms of toxicity of halogenated aromatic compounds. They found that TCDD, which they previously showed to be a potent inducer of 
-aminolevulinic acid synthase (Poland and Glover, 1973a), also was a potent AHH inducer in chick embryo liver (Poland and Glover, 1973b). Moving to mammals, TCDD proved to be 30,000 times more potent than 3-MC at inducing AHH in rat liver (Poland and Glover, 1974). Even more revealing was the ability of TCDD to overcome the "nonresponsive" phenotype in mice. In collaboration with Dan Nebert's group, the Poland–Nebert team found that TCDD was able to induce hepatic AHH activity in five mouse strains that are nonresponsive to 3-MC (Poland et al., 1974). The fact that TCDD induced AHH in "nonresponsive" mice led to the conclusion that the P450 gene encoding the enzyme was normal and to the hypothesis that nonresponsiveness was due to a mutation which leads to production of an "inducer-binding receptor" with a reduced affinity for nonhalogenated aromatic hydrocarbons. Breeding studies in mice supported this hypothesis (Nebert et al., 1975; Poland and Glover, 1975), setting the stage for use of radiolabeled TCDD in the search for the induction receptor.
Alan Poland's chemistry collaborator at the University of Rochester, Andrew Kende, prepared [3H]TCDD as the quintessential bait for the receptor fishing expedition. In addition to its great potency as an AHH inducer, TCDD has the virtue of being chemically stable and highly resistant to metabolism in most biological systems in vitro or in vivo. Armed with [3H]TCDD, Poland, Glover, and Kende brought forth the eagerly sought first experimental evidence for an induction receptor in their landmark JBC paper in 1976 (Poland et al., 1976b). The hypothetical receptor had become real.
The Okey Lab Enters the Induction-Receptor Arena
Adventures with PAH radioligands.
In 1976, prior to my sabbatical leave in the Nebert lab at NIH, [3H]TCDD was not generally available. My laboratory, therefore, made our initial foray into the induction-receptor field with tritiated versions of the nonhalogenated AHH inducers, BP and 3-MC. We attempted to identify a [3H]BP-binding component in C57BL/6 hepatic cytosol that could be saturated at reasonable radioligand concentrations and that would show specificity when nonradioactive AHH inducers were introduced as competitors. Binding profiles, after separating radiolabeled cytosol by velocity sedimentation on sucrose gradients, revealed a very large radiolabeled peak with a sedimentation coefficient of about 4S. There also was a small peak at about 9S but we could not get "clean" competition by other AHH inducers for the 9S component and did not attempt to publish these data.
In retrospect, the binding component that sedimented at 9S does, in fact, represent binding of the PAH radioligands to AHR as we (Okey and Vella, 1982, 1984; Okey et al., 1984) and Poellinger et al. (1983) would demonstrate in the 1980s by direct binding studies with [3H]BP, [3H]3-MC and [3H]dibenzo[a,h]anthracene. The identity and function of the 4S component that becomes labeled by [3H]BP and [3H]3-MC remains mysterious. The abundant 4S component was tentatively identified as glycine N-methyltransferase and it was proposed that the 4S binder mediates CYP1A1 induction (Bhat and Bresnick, 1997; Raha et al., 1995). However, other studies do not support a role for the 4S binding component in P450 regulation (Harris et al., 1988; Kamps and Safe, 1987). Most recently it was reported that ß-naphthoflavone, a well-known CYP1A1 inducer, binds a 4S component but the functional significance of this binding for regulating gene expression remains unclear (Brauze, 2004; Brauze and Malejka-Giganti, 2000).
[3H]TCDD at the NIH.
Gregory Bondy, an exceptionally talented graduate student from my laboratory at the University of Windsor, joined me at the Nebert NIH lab for a few months in the summer of 1978 prior to his entry into medical school. We applied techniques we previously had used to study ER (Okey and Bondy, 1977, 1978a,b) to the study of [3H]TCDD binding in the Nebert Ah-locus mouse model. It may be difficult for current students to envision this Paleozoic era in receptor research when the only method available to identify and characterize soluble receptors was by reversibly tagging them with a dissociable radioligand. Nevertheless, rapid progress was made, riding on the back of the splendid radioligand, [3H]TCDD.
In addition to the important receptor properties of Kd, Bmax and specificity that were revealed in Alan Poland's milestone 1976 paper, we wanted to understand macromolecular structure of the binding protein. To that end we employed velocity sedimentation on sucrose density gradients (SDG) along with ion exchange chromatography. The chromatography experiments were greatly facilitated by the collaboration of the late Howard Eisen who was a glucocorticoid receptor specialist and part of the Nebert laboratory. Both the SDG technique and column chromatography were widely used in the 1970s in studies of steroid hormone receptors; our use of these methods reflected our presumption that the dioxin-binding protein was a type of steroid receptor.
With data gathered primarily by SDGs, we confirmed Alan Poland's findings. Namely, that hepatic cytosol from C57BL/6 mice contains a saturable, high-affinity [3H]TCDD binding site that is selective for compounds known to be AHH inducers and that this specific binding component was not detectable in cytosol from genetically nonresponsive DBA/2 mice. It was Dan Nebert's wisdom and logic to designate this binding component the "AH receptor," since it is the product of the Ah locus; we introduced the "AH receptor" terminology in our first paper (Okey et al., 1979).
In the following sections I will concentrate on findings made by my laboratory but attempt to do justice to the many other AHR investigators by placing our discoveries in the context of overall developments in the field.
Nuclear Translocation—Behaving Like a Steroid Receptor
Cloning in the 1990s would reveal that the AHR's primary structure does not qualify it for bona fide membership in the formal nuclear receptor superfamily, notably because the AHR lacks the zinc-finger domain that typifies steroid receptors. However, despite belonging to a different gene family, the AHR behaves very much like a steroid receptor. Because we had a preconception in the 1970s (based on the steroid receptor precedent) that liganded AHR should translocate from cytoplasm into nucleus, we prepared both cytosol and nuclear extract from livers of mice injected with [3H]TCDD. We were rewarded with beautifully symmetrical [3H]TCDD-binding peaks in both cell fractions (Figs. 2 and 3). Clearly, however, the cytosolic and nuclear forms of AHR had different sedimentation velocities (Fig. 3). Further experiments would be required to find out why.
|
|
Although nuclear uptake of the [3H]TCDD • AHR complex could be demonstrated in vivo, (as shown by our experiments and by William Greenlee's studies in Alan Poland's laboratory at about the same time; Greenlee and Poland, 1979
), in vivo studies are cumbersome. Thus, after I completed my sabbatical leave in the Nebert laboratory, we collaborated with Michael Dufresne, a cell biologist at the University of Windsor, to refine our understanding of the cytosol-to-nucleus translocation process in a more tractable system, Hepa-1 cells in culture (Okey et al., 1980). Cell culture experiments confirmed that sedimentation properties are quite different between cytosolic and nuclear forms of the AHR. Moreover, we found that translocation into the nucleus is a temperature-dependent process. The nuclear compartment is devoid of AHR until cells are exposed to a ligand at physiologic temperature. Recent work by Kawajiri and colleagues shows that the AHR protein contains a motif for nuclear localization as well as a motif for nuclear export and that phosphorylation of the nuclear-localization motif inhibits nuclear uptake of the AHR (Ikuta et al., 1998, 2004a,b). However, the exact nature of the temperature-dependent step which is required for nuclear uptake still is not resolved.
Physicochemical Characterization: Appreciating the AHR as a Macromolecule
Kinetics of ligand binding (Kd and Bmax) can be determined without knowing anything about the structure of the binding protein. However, in order to truly understand receptor function it is imperative to know the receptor's macromolecular properties.
When I returned to Canada after completion of my sabbatical year in the Nebert laboratory I was offered a position in the Division of Clinical Pharmacology in the Research Institute at The Hospital for Sick Children, Toronto, where we continued to investigate multiple facets of the AHR. One limitation in our research was that the SDG assay we initially used required overnight centrifugation in a swinging-bucket rotor that held only six samples. In Toronto, to increase our analytical capacity, my postdoctoral fellow, Hing Wo Tsui, developed a 2-h vertical-tube-rotor SDG assay (Tsui and Okey, 1981) that became a mainstay in our research program for several years and was widely adopted by other laboratories.
Devotees of the author George Orwell might have anticipated the dawn of the year 1984 with trepidation. But, in fact, 1984 was a very good year for my laboratory because it saw the arrival of two exceptionally productive postdoctoral fellows, Michael Denison and Patricia Harper, and an excellent Ph.D. student, Rebecca Prokipcak. Mike Denison and Becky Prokipcak immersed themselves in physicochemical characterization of the AHR while Patricia Harper (a cell biologist by training) spearheaded our transition toward a molecular approach to the AHR.
Mike Denison's hydrodynamic experiments showed that cytosolic AHRs from Sprague–Dawley rat liver and C57BL/6 mouse liver exist as macromolecular complexes of 250–280 kDa that can be dissociated, under conditions of high ionic strength, into smaller ligand-binding subunits of about 120 kDa for rat and 105 kDa for C57BL/6 mouse (Denison et al., 1986c). These experiments provided the first evidence that the AHRs from rats and mice are similar but not identical molecular species, a finding that later would be confirmed and extended in the eras of immunoblotting and cloning.
As described above, our initial experiments in vivo and in cell culture showed that cytosolic and nuclear forms of AHR have different sedimentation properties (Okey et al., 1979, 1980). One possibility was that the nuclear receptor is simply a monomeric ligand-binding subunit contained within a multimeric cytosolic AHR complex and that the ligand-binding subunit generated when we exposed labeled cytosol to high salt is the same macromolecule as the nuclear receptor. Further hydrodynamic analysis by Becky Prokipcak showed that this simple scenario is not true. Exposure of cytosol to high salt yields a [3H]TCDD-binding component with a mass of 105 kDa (as seen in the Denison experiments), whereas the form of receptor recovered by high-salt extraction of nuclei from cells treated with [3H]TCDD has a mass of about 176 kDa (Prokipcak and Okey, 1988). Clearly, the nuclear AHR was not simply the monomeric ligand-binding subunit. Becky went on to show, by photoaffinity labeling with [3H]TCDD and electrophoresis under denaturing conditions, that the size of the ligand-binding component in nuclear AHR is the same as the ligand-binding component in cytosol (Prokipcak and Okey, 1990). Some additional component would need to be identified to account for the extra mass of the nuclear form of AHR detected in our laboratory and others (Elferink et al., 1990; Gasiewicz et al., 1991). The "missing piece" of the nuclear complex turned out to be the ARNT protein, later identified by Oliver Hankinson's laboratory (Hoffman et al., 1991; Reyes et al., 1992). The identities of the multiple constituents of the cytosolic AHR complex would eventually be identified by several laboratories (see below).
Of course my laboratory was not the only group involved in this early euphoric phase of AHR characterization. Jan-Åke Gustafsson's laboratory at the Karolinska Institute, Stockholm, had great expertise and experience with the glucocorticoid receptor and turned some attention to the "TCDD receptor" or "dioxin receptor" (aka, AH receptor). They first used isoelectric focusing of [3H]TCDD-labeled rat liver cytosol to identify a specific binding component that had high affinity and selectivity for CYP1A inducers (Carlstedt-Duke et al., 1978). However, partial proteolysis with trypsin was required in order to focus the specific band; thus it was not possible to determine physicochemical properties of the native receptor protein. Lorenz Poellinger (initially in the Gustafsson laboratory) and Thomas Gasiewicz (University of Rochester) also performed extensive physicochemical analysis of the AHR and found that its overall properties were reminiscent of steroid receptors but with some distinct differences (Gasiewicz and Bauman, 1987; Gasiewicz and Rucci, 1984; Nemoto et al., 1990; Poellinger et al., 1982, 1983).
Since the AHR has many physicochemical properties in common with steroid receptors, both Lorenz Poellinger and Tom Gasiewicz adapted a steroid receptor technique based on adsorption to hydroxylapatite to measure [3H]TCDD binding to AHR (Gasiewicz and Neal, 1982; Poellinger et al., 1985). The "HAP" method became widely used by many laboratories as a rapid assay in AHR binding studies.
The ultimate physicochemical characterization for a protein is to derive a 3D crystallographic structure that will reveal how the protein functions. The AHR has not yet been crystallized. However, some insight into AHR 3D structure has been obtained by recent homology modeling based on similarities in primary structure between the AHR and related proteins (Pandini et al., 2007). Modeling and site-directed mutagenesis reveal several structural features that are important to the ligand-binding function.
Lonely No More: Multiple AHR Partner-Proteins are Identified
As shown in Figure 4, the AHR resides in cytoplasm until binding of ligand triggers transformation of the receptor and its translocation into the cell nucleus. Our early physicochemical characterization indicated that both the cytosolic and the nuclear forms of AHR are oligomeric complexes composed of the AHR protein in association with other macromolecules. Identities of AHR-interacting proteins were resolved, beginning in the late 1980s, through the efforts of many laboratories, particularly those of Gary Perdew, Christopher Bradfield, and Lorenz Poellinger.
|
The heat-shock protein, hsp90, is a major constituent of the cytosolic complex; its presence was sought in the AHR complex because hsp90 already was well-known to be a vital chaperone for steroid receptors. In addition to hsp90 the cytosolic AHR complex contains at least two other proteins, ARA9 (also known as AIP or XAP2) which assists in stabilizing the AHR and retaining it in the cytoplasmic compartment and p23 which appears to aid release of hsp90 from the AHR after a ligand binds (reviewed in Harper et al., 2006
; Petrulis and Perdew, 2002).
At first glance things seem simpler in the nucleus since the nuclear AHR complex contains only the AHR itself tightly bound to its dimerization partner, ARNT. However, the nuclear ligand • AHR • ARNT complex undergoes a host of protein–protein interactions with coactivators, coreppressors, chromatin remodeling proteins, and basal transcription factors (reviewed in Hankinson, 2005; Kewley et al., 2004; Rowlands et al., 1996; Swanson, 2002).
The formal family of nuclear receptors (including steroid receptors) and the bHLH/PAS family (to which the AHR belongs) are structurally unalike, even though both gene families encode ligand-dependent transcription factors. It should not come as a surprise that many of the chaperones, coactivators, and corepressor proteins which interact with the AHR also interact with other receptors. The evolutionary tool-kit contains component parts that frequently are shared by multiple cellular pathways.
Identification of Response Elements: the AHR Finds its Home on DNA
Successful biological regulation requires a degree of specific recognition at multiple levels in signaling pathways such as "specificity" of a receptor for its ligands. Within the nucleus, the ligand • AHR • ARNT complex also must be recognized by specific sites in order to regulate gene expression in an orderly fashion. The specific nucleotide sequence to which the nuclear AHR complex binds was first identified by Mike Denison who, after a very productive stint as a post-doc in my laboratory, joined Jim Whitlock's laboratory at Stanford where they located and sequenced a "dioxin-responsive element" in the 5'-flanking sequence of the highly inducible mouse Cyp1a1 gene (Denison et al., 1988a,b). This subsequently was corroborated by Yoshiaki Fujii-Kuriyama's laboratory (then at Tohoku University, Japan) who termed the enhancer element "xenobiotic-responsive element" (XRE) (Fujisawa-Sehara et al., 1988). (In keeping with terminology for the receptor itself, we prefer the term "AHRE" [AH response element].) The Whitlock laboratory went on to extensively describe how the AHR affects the CYP1A1 promoter and chromatin structure to alter gene expression (reviewed in Swanson, 2002; Whitlock, 1999).
In collaboration with Mike Denison, Patricia Harper and I showed that human AHR can be activated by ligand to bind to the same nucleotide sequence that comprises the mouse AHRE (Harper et al., 1992). The fundamental mechanism of gene regulation by the AHR is well-conserved across mammalian species. The core pentanucleotide AHRE sequence (GCGTG) occurs frequently within mammalian genomes (Lee et al., 2006; Sun et al., 2004; Tijet et al., 2006). Recently Oliver Hankinson's laboratory reported that the mouse Cyp2s1 gene contains three overlapping AHRE sequences upstream of the promoter and three overlapping hypoxia response elements (HREs) embedded within the region containing the AHREs (Rivera et al., 2007). The potential complexity of gene regulation by the AHR is illustrated by the fact that not only does the AHR • ARNT dimer bind to this regulatory region, the region also binds the HIF-1
• ARNT dimer which is a powerful regulator of genes that respond to hypoxia. As we will see later, the architecture of receptors and their response elements provide ample opportunity for cross-talk in a potentially very complex combinatorial fashion.
The AHRE sequence originally identified in the mouse Cyp1a1 gene probably is the response element for the majority of AHR-regulated genes. However, induction of CYP1A2 has perennially been a more complex problem than induction of CYP1A1. Sogawa et al. (2004) identified a novel enhancer element in the rat CYP1A2 gene which seems to be the site of action of the TCDD • AHR • ARNT complex and termed this response element "XRE-II." As a twist on the "standard model" of AHR signaling, the TCDD • AHR • ARNT complex does not bind directly to the XRE-II response element; rather, the complex appears to couple to XRE-II through binding to an unidentified adapter protein which itself binds XRE-II. My laboratory wondered whether the XRE-II motif (which we term AHRE-II) was unique to the rat CYP1A2 gene or whether this element might be involved in other AHR-mediated gene responses. Paul Boutros, an insightful bioinformaticist in my laboratory, used phylogenetic footprinting to show that the AHRE-II motif is conserved in at least 36 genes across the genomes of mouse, rat and human. By gene expression array analyses we found that about 15 genes which contain conserved AHRE-II motifs respond to TCDD. Rather surprisingly, many of these genes that appear to respond through the AHRE-II element encode ion-channel proteins and transporters rather than enzymes related to metabolism of xenobiotic chemicals (Boutros et al., 2004).
AHR Downregulation by its Ligands
Our initial AHR studies in cell culture hinted that treatment with TCDD causes the total cellular AHR content to decrease rapidly (Okey et al., 1980). My Ph.D. student, Becky Prokipcak, performed a thorough accounting of cytosolic and nuclear forms of AHR in Hepa-1 cells and established, for the first time, the phenomenon of ligand-induced downregulation of AHR (Prokipcak and Okey, 1991).
A skilled technical assistant in my laboratory, John Giannone, then showed that if protein synthesis is blocked with actinomycin D or cycloheximide, nuclear AHR levels do not decrease after cells are exposed to TCDD. These data were the first evidence that ligand-dependent downregulation of the AHR likely results from protein degradation involving a short-lived protease (Giannone et al., 1995). Subsequently, we confirmed that downregulation of cellular AHR content is not due to a decrease in AHR messenger RNA (mRNA) but, rather, via loss of AHR protein while sparing its dimerization partner, ARNT (Giannone et al., 1998). The laboratories of Richard Pollenz, Murray Whitelaw, and Qiang Ma then independently discovered that the mechanism of ligand-induced AHR downregulation in cell culture is predominately through the ubiquitin-proteasome pathway (Ma and Baldwin, 2000; Pollenz, 2002; Roberts and Whitelaw, 1999).
As described above, downregulation was first discovered in cell culture. Downregulation also can occur in TCDD-exposed tissues in vivo (Pollenz et al., 1998; Sommer et al., 1999). However, a Ph.D. student in my laboratory, Monique Franc, found that downregulation in rodent liver is transient following a single TCDD dose in vivo and that after a few days TCDD actually causes a slight upregulation in AHR mRNA and AHR protein (Franc et al., 2001a). She also mimicked real-world environmental/dietary exposure to TCDD and found that AHR levels remain relatively constant in the face persistent, low-dose TCDD intake (Franc et al., 2001b).
AHR downregulation presumably represents a cell's method of desensitizing itself and preventing excessive stimulation from potent agonists. Downregulation is dramatic in cell culture but its transient nature in vivo suggests that most tissues will not be desensitized following persistent, low-dose TCDD challenge. So far as we know, resistance to TCDD toxicity cannot be attributed to sustained downregulation of AHR levels in any animal species. AHR levels in vivo and in cell culture are affected by a bewildering variety of factors, including the receptor's own ligands; for a recent summary please see Harper et al. (2006). Our understanding of the "what regulates the regulator?" remains rudimentary. New factors continually are being discovered such as the recent reports that Erk kinase participates in AHR degradation (Chen et al., 2005) and that NS1BP, a protein which contains a "kelch" domain, may regulate functional levels of AHR in cells both by tethering AHR to the cytoskeleton and by influencing proteasomal degradation (Dunham et al., 2006).
Ubiquitous Expression of the AHR
The original discovery of a specific [3H]TCDD binding site in mouse liver (Poland et al., 1976b) naturally led to the question of how widely this new receptor is distributed across animal species and tissues.
Ontogeny and tissue distribution.
As described above, fetal cells were a valuable early model system to study induction mechanisms for CYP1 enzymes (Nebert and Gelboin, 1968b). In addition, TCDD is one of the most potent teratogens known in rodents. With these motivations, multiple laboratories mapped AHR expression during development.
The developing mouse kidney is exceptionally sensitive to teratogenesis by TCDD which induces hydronephrosis in an AHR-dependent fashion (Lin et al., 2001; Mimura et al., 1997; Peters et al., 1999). Sharon Choi, a Ph.D. student working with my colleague Patricia Harper in Toronto, found that AHR mRNA is expressed as early as gestational day-14 in ureter from C57BL/6 mice (Choi et al., 2006).
There was a surprise when we studied AHR expression and function in embryonic tissue and cells derived from "nonresponsive" mouse strains. Recall that livers of adult nonresponsive strains are completely refractory to CYP1A induction by nonhalogenated ligands. In cytosol from tissues of embryos at 15–19 days gestation, not only were we able to detect specific binding of the potent agonist, [3H]TCDD, we also were able to detect some specific binding of the nonhalogenated inducer, [3H]3-MC, to AHR. Moreover, in primary cultures derived from embryos we found that dose–response curves for AHH induction by another PAH inducer, benz[a]anthracene, were essentially the same for cells from "nonresponsive" mice as they were for "responsive" mice (Harper et al., 1991a). Ying Huang, a graduate student working with Patricia Harper and me, found that the AHR in embryonic cells was indistinguishable from the receptor expressed in adult liver (Huang et al., 1995). The mechanistic explanation remains elusive for why embryonic cells from "nonresponsive" strains can, in fact, respond to PAHs, whereas adult tissues in vivo show strong separation into "responsive" and "nonresponsive" phenotypes. These experiments in embryonic cell cultures suggest that the cellular context has a significant influence on AHR function, possibly because the particular repertoire of proteins that interact with the AHR differs between embryonic cells versus cells from adult animals.
During postnatal development, Carlstedt-Duke et al. (1979) discovered that rat liver AHR levels are highest around weaning, then wane as animals age. This pattern of highest hepatic AHR prior to weaning was confirmed for rat by Kahl et al. (1980) and also found to hold true for mouse and rabbit. Tom Gasiewicz's laboratory at the University of Rochester extended the ontogenic findings by showing that although AHR levels drop in liver and lung after weaning, levels in thymus remain elevated for a longer period (Gasiewicz et al., 1984). In rat prostate, AHR levels are high at birth but undergo a steep decline even before weaning (Sommer et al., 1999). The general pattern holds for most rodent tissues: AHR levels are highest in the younger animals and decline with age. This pattern suggests that the main biological role of the AHR plays out during development. Later, the tools of molecular biology would further illuminate the fundamental biology of the AHR in relation to development (see below).
After the AHR was discovered in rodent livers, we and other laboratories surveyed a wide range of mammalian tissues to determine how broadly this new regulatory protein is distributed. By radioligand binding and SDG assays, Michelle Mason, a research assistant in my laboratory, detected AHR in liver, lung, kidney, intestine, thymus, and prostate of C57BL/6 mice and Sprague–Dawley rats. By treating mice in vivo with [3H]TCDD she also found that the [3H]TCDD x AHR complex could be recovered from nuclei of liver, lung, kidney not only in responsive C57BL/6 mice but also in "nonresponsive" DBA/2 mice (Mason and Okey, 1982). This was the first definitive evidence that nonresponsive mice do, in fact, possess an AHR that is competent to bind ligand and translocate into the nucleus.
Space does not permit a full accounting of each mammalian tissue that subsequently has been shown to express the AHR. Suffice it to say that methods ranging from ligand binding to mRNA expression profiling reveal that the AHR can be detected in virtually all mammalian cells and tissues, albeit at widely varying levels.
A few words on phylogeny.
From the earliest days of AHR research there has been the conundrum of why animals are endowed with a receptor whose main function appears to be binding of notoriously toxic and carcinogenic xenobiotic chemicals. In the precloning era, mapping the phylogenetic distribution of AHR was one avenue to trying to understand the receptor's "purpose" and evolutionary history.
Most toxicology is "mammalocentric" but we and other laboratories wanted to find out if the AHR was present in nonmammalian species as well as in a broad spectrum of mammals (Denison and Wilkinson, 1985). Chick embryo has a noble history as an excellent model system for studying AHH induction and other biochemical/toxic effects of halogenated aromatic hydrocarbons (Hamilton et al., 1983; Poland and Glover, 1973a,b; Rifkind et al., 1990). In collaboration with Christopher Wilkinson's laboratory (Cornell University), Mike Denison and I found that AHR is expressed in chick embryo as early as at 5 days of incubation and that levels in chick liver drop rapidly after hatching, reminiscent of the postnatal decline in AHR levels in rodents (Denison et al., 1986a). Gail Bellward's laboratory (University of British Columbia) found, using our assay methods, that AHR is not confined to domestic fowl but also is detectable in feral bird species such as pigeon, heron, and cormorant (Sanderson and Bellward, 1995).
We also were interested in AHR expression in fish because trout are highly sensitive to biochemical and toxic effects of dioxin-like chemicals and PAHs. Our attempts to detect AHR with our standard [3H]TCDD binding assay in trout liver proved fruitless, possibly because fish livers contain proteolytic enzymes (Hahn et al., 1994) that are adapted to low temperatures and happily degrade the AHR, even at the 0–4° conditions of the binding assay. Angela Lorenzen, a graduate student in my laboratory, was first to gain solid evidence that the AHR exists not only in homeothermic vertebrate species but also in poikilothermic animals. She used a trout hepatoma cell line in culture to demonstrate binding of [3H]TCDD to cytosolic AHR, translocation of the [3H]TCDD • AHR complex into the nucleus and subsequent induction of AHH activity (Lorenzen and Okey, 1990). As we would find with recalcitrant AHRs in other systems, adding molybdate to the buffer was essential to stabilize trout AHR so that specific [3H]TCDD binding could be detected.
Research over the past three decades indicates that AHR structure and function are remarkably diverse among vertebrates and invertebrates. Early physicochemical analyses suggested that molecular properties of the AHR are similar among laboratory mammals such as rat, mouse, guinea pig, and hamster (Gasiewicz and Rucci, 1984; Poellinger et al., 1983); however, our physicochemical characterization indicated that rat and mouse AHRs are not identical (Denison et al., 1986c).
Structural differences became even more apparent when both photoaffinity labeling and the development of anti-AHR antibodies allowed electrophoretic separation of AHRs from different animal species under denaturing conditions. These experiments indicate apparent molecular masses ranging from 95 kDa for the product of the mouse Ahrb1 allele to 146 kDa in trout (Hahn et al., 1994; Landers et al., 1989; Poland and Glover, 1987, 1990; Poland et al., 1991; Prokipcak and Okey, 1990). Cloning and sequencing of AHR genes, beginning in the 1990s, confirmed the diversity of molecular masses among vertebrate AHRs and also revealed that most of the variation in AHR primary structure resides near the carboxy terminus of the protein (Gu et al., 2000; Korkalainen et al., 2001).
[3H]TCDD binds with specificity and high affinity to AHR proteins from a wide range of vertebrate species. However, in invertebrates, specific [3H]TCDD binding has not been detectable in any species out of the many tested (Denison et al., 1985, 1986d; Hahn et al., 1994). Mark Hahn's laboratory at the Woods Hole Oceanographic Institution has taken a leading role in demystifying phylogeny and evolution of the AHR (Hahn, 1998, 2002; Hahn et al., 2006). He proposes that during the AHR's evolutionary history it has changed from a protein that does not bind ligand (invertebrates) to a protein that is a ligand-activated transcription factor (Hahn et al., 2006). Among the Hahn laboratory's other key discoveries is the unexpected finding, from comparative genomics, that mammals seem impoverished in regard to how many AHR genes exist within the genome of an individual species. Genomic sequencing indicates that mammals have but a single AHR gene, whereas in certain fish or bird species there may be as many as two to five genomic sequences that are predicted to encode AHRs. For a full appreciation of AHR phylogeny and evolution see the comprehensive and authoritative reviews by Mark Hahn (Hahn, 1998, 2002; Hahn et al., 2006).
The nearly ubiquitous occurrence of AHR in vertebrate tissues implies that this receptor has important biological functions. However, as we will see below, AHR knockout, at least in mice, is not lethal.
Humans, Too, Have AHR
After the discovery and initial characterization of AHR in rodent tissues, we and other laboratories were eager to determine if humans possess a similar receptor. Such knowledge would be valuable when attempts are made to incorporate mechanistic data on dioxin toxicity into human risk assessment. But the human AHR obstinately refused to cooperate. When we used assay methods that worked well in rodent cells and tissues, AHR abundance appeared to be very low or absent in clinical samples such as lung (Roberts et al., 1986).
We screened many human cell lines and tissues with disappointing results. With human tissues, there is the challenge of obtaining tissues of good quality while adhering to ethical requirements. Fortunately, human placenta is highly responsive to induction of CYP1A enzymes by cigarette smoke (Manchester et al., 1984; Nebert et al., 1969; Welch et al., 1968). David Manchester (a pediatrician and clinical geneticist at the University of Colorado Health Sciences Center) had a long-standing interest in regulation of CYP1 enzyme induction in placenta and had set up a very effective protocol for obtaining placental tissues of high quality from smoking mothers and nonsmokers, so we collaborated with David's laboratory to investigate AHR in this responsive and available tissue.
We knew that molybdate was a very helpful ingredient in homogenizing buffers to stabilize various steroid hormone receptors but we had found that molybdate was not really necessary to stabilize AHR in rodent livers (Denison et al., 1986b). Nevertheless, when we modified our procedures with human placenta by including molybdate in the homogenizing buffer, we found that molybdate was an elixir that finally permitted us to detect human AHR (Manchester et al., 1987). As it turns out, placenta is the human tissue that is perhaps the most richly endowed with AHR (Manchester et al., 1987; Okey et al., 1997).
By adding molybdate to the buffer and making other adjustments, we were able to routinely detect AHR in a wide variety of human tissues. My graduate student, Angela Lorenzen (who also discovered AHR in trout), used the improved assay to demonstrate that human tonsils express significant AHR levels (Lorenzen and Okey, 1991); this is of potential relevance to human health since atrophy of immune organs such as the thymus is one of the most sensitive toxic responses to dioxin-like chemicals in rodents.
Human tissues are useful for obtaining a snapshot of AHR abundance in diverse tissues from different donors. However, tissue samples cannot provide much information on the function of AHR pathways. Thus, we applied our improved receptor assay to several human cell lines. My Toronto colleague, Patricia Harper, a genuine cell biologist, directed most of our studies in cell model systems.
In humans, skin is the most notable target for dioxin toxicity which manifests as chloracne (Geusau et al., 2001; Panteleyev and Bickers, 2006), the disfiguring condition that came to world-wide attention with the poisoning of presidential candidate, Victor Yushchenko, in Ukraine in 2005 (Schecter et al., 2006). As a surrogate for skin cells, we tested the human squamous cell carcinoma line, A431 and detected good levels of AHR. In the A431 cells (just as in our previous research with the mouse Hepa-1 cell line) exposure to TCDD provoked nuclear translocation of the AHR. Both TCDD and benz[a]anthracene induced AHH activity in a classic sigmoidal dose–response fashion (Harper et al., 1988).
Eve Roberts, a clinician hepatologist, undertook a period of basic research training in my laboratory and found that HepG2 cells, a human hepatoma cell line widely used as a model for human hepatic drug metabolism, express AHR and induction of CYP1A1 (Roberts et al., 1990). In collaboration with William Waithe and Alan Anderson (L' Hotel dieu de Quebec), we found that human peripheral blood lymphocytes, after being immortalized for growth in culture, displayed the complete AHR-dependent regulatory mechanism for CYP1A1 induction (Waithe et al., 1991)
The highest AHR concentration that we've detected in any human cell line is in LS180 colon carcinoma cells (Harper et al., 1991b). A postdoctoral fellow in my laboratory, Wei Li, found that LS180 cells exhibit CYP1A2 induction by TCDD or 3-MC (Li et al., 1998). Although CYP1A1 is highly inducible in a wide variety of mammalian cell lines, CYP1A2 expression in immortalized cells lines usually is silenced for reasons that are unclear. Thus, the LS180 cell line constitutes an opportunity to clarify factors that regulate basal CYP1A2 expression and its AHR-dependent induction.
After early frustrations in our search for human AHR it has been rewarding to see dozens of reports on AHR expression in a wide range of human tissues and cell types (summarized in Okey et al., 1994a). Unquestionably, the AHR is available in humans to carry out many of the same functions (for better or worse) that it does in laboratory animals.
Astonishing Range of AHR Ligands
As we have seen, [3H]TCDD became the ideal ligand for detection and characterization of the AHR. Quite clearly, however, the binding site is not just a "dioxin receptor" or "TCDD receptor." Beginning with the first studies with [3H]TCDD as radioligand, a multitude of chemicals has been tested to see if they can compete with [3H]TCDD for specific AHR-binding sites. Competition studies are a mainstay in determining if a new chemical is a receptor ligand because the test chemical does not need to be radiolabeled. Through competition studies, the catalog of AHR ligands has been greatly expanded over the past three decades. However, because [3H]TCDD binds with very high affinity and dissociates exceedingly slowly from the AHR, methods used to test for competition need to be carefully designed to prevent false negative conclusions, especially when weak ligands are tested (Denison and Nagy, 2003).
Based on the original experiments with TCDD and closely related compounds it appeared that the AHR ligand-binding site had rigid dimensions and that it could accommodate ligands only if they were highly planar. This view has changed dramatically as the catalog of ligands expanded via competition studies and high-throughput screening assays. Now the AHR is viewed as one of the "promiscuous" receptors; that is, a receptor that can effectively bind compounds of diverse shape and chemical properties. Among the ligands from exogenous sources: halogenated dioxins, dibenzofurans, and polychlorinated biphenyls (PCBs); nonhalogenated PAHs; flavones and carbinols of plant/dietary origin; therapeutic agents such as omeprazole (for an excellent review see Denison and Nagy, 2003).
As I mentioned in the Prelude, my original interest in enzyme induction was motivated by effects of the pesticide, DDT. Although we found that commercial grade p,p'-DDT has some ability to inhibit binding of [3H]TCDD to the AHR (Okey et al., 1979), it later would turn out that P450 induction by DDT in vivo most likely is due to binding of its metabolite, dichlorodiphenyldichloroethylene to the nuclear receptors CAR and PXR rather than to the AHR (Coumoul et al., 2002; Wyde et al., 2003). Thus, although the AHR is "promiscuous," it is not the universal receptor for all environmental contaminants.
Endogenous Ligands (or Why did Evolution Endow us with a Receptor for Toxic Dioxins?)
The question of why a "dioxin receptor" arose in evolution is inextricably tied to the question of whether there is an endogenous AHR ligand that regulates "normal" physiologic functions. From the earliest days of AHR research there has been a keen interest in identifying the ever-elusive endogenous ligand. It might be more judicious to say endogenous ligands because there is no a priori reason why multiple endogenous agents from different chemical classes might not exist, given the AHR's promiscuous reputation for binding structurally diverse exogenous compounds (Denison and Nagy, 2003).
Progress in identifying candidate endogenous ligands was slow during the first two decades of AHR research. However, in recent years multiple endogenous agents have been shown to activate AHR pathways. For example, arachidonic acid (AA) metabolites are released in response to TCDD and it is possible that some AA metabolites, such as prostaglandins, may act as AHR agonists (Denison and Nagy, 2003; Rifkind, 2006). Bilirubin and related tetrapyroles, at high concentrations, can activate AHR, perhaps serving to induce glucuronosyltransferase enzymes that conjugate and remove the potentially toxic products of heme degradation (Denison and Nagy, 2003; Sinal and Bend, 1997). Ultraviolet (UV) irradiation photo-converts tryptophan into products that have high affinity for the AHR and are potent inducers of CYP1A1 (Denison and Nagy, 2003; Rannug and Fritsche, 2006). Tryptophan also can be converted by the enzyme, aspartate aminotransferase, into indole-3-pyruvate which spontaneously generates multiple compounds that can act as AHR agonists (Bittinger et al., 2003).
One approach to identifying endogenous ligands is to prepare tissue extracts and assay their ability to activate an AHR-mediated reporter gene system. Song et al. (2002) employed this approach and isolated, from porcine lung, a compound whose structure was identified as 2-(1'H-indole-3'-carbonyl)-thiazole-4-carboxylic acid methyl ester (ITE). Very recently Henry et al. (2006) confirmed that synthetic ITE is a potent AHR agonist which induces CYP1A1 in cell culture and in vivo; however, ITE does not produce dioxin-like teratogenic effects in mice. This is another instance which reminds us that high-affinity AHR ligands are not necessarily toxic. We need to find out why this is true for specific chemical cases and what features of a chemical ligand are necessary to elicit severe dioxin-like toxicity.
Of course, an obvious explanation for why a high-affinity ligand might lack toxicity is that the ligand is an antagonist rather than an agonist. Savouret et al. (2000) identified 7-ketocholesterol (7-KC) as an endogenous compound that competitively binds AHR and inhibits CYP1A1 induction by TCDD; they propose that 7-KC is a "protective modulator" of AHR function.
Another contender for the role of endogenous AHR agonist is modified low-density lipoprotein (LDL). In the 1990s hydrodynamic shear stress was found to induce CYP1A1 in cell culture and initially it was thought that induction was due to release of AA metabolites (Mufti and Shuler, 1996). However, very recently the Bradfield laboratory reported that hydrodynamic shear stress (mimicking blood flow in the vasculature) modifies an LDL in blood serum such that the LDL becomes an AHR activator (McMillan and Bradfield, 2007). This may provide a mechanism for the vital role which the AHR plays in vascular development (Lahvis et al., 2005).
This brief overview of the range of exogenous and endogenous AHR ligands is intended simply to remind us of how little we understand about the diversity of AHR ligands and AHR functions. Chris Bradfield's laboratory divides AHR-mediated responses into three pathways: (1) adaptive responses (such as changes in xenobiotic metabolism); (2) toxic pathway; (3) developmental pathway (Walisser et al., 2004b). Different ligands may selectively act upon one or more of these pathways for good or ill (Denison and Nagy, 2003).
To conclude this section, let's consider what happens if the AHR is activated in the absence of any ligand. Lorenz Poellinger's laboratory deleted the ligand-binding domain and thereby created an AHR that is constitutively active, i.e., it stimulates gene expression without any ligand (Kohle et al., 2002; McGuire et al., 2001). This led to the disturbing discovery that mice whose AHR is locked into the "on" state have a shortened life-span and frequently develop stomach cancer. However, stomach cancer is not commonly seen in laboratory animals exposed to dioxin-like chemicals and it is unclear why the constitutively active AHR causes stomach tumors, whereas the persistent AHR ligand, TCDD, is not an efficient gastric carcinogen. The constitutively active