The Aryl Hydrocarbon Receptor Complex and the Control of Gene Expression

Abstract

The aryl hydrocarbon receptor (AhR) is a ligand-activated transcription factor that controls the expression of a diverse set of genes. The toxicity of the potent AhR ligand 2,3,7,8-tetrachlorodibenzo-p-dioxin is almost exclusively mediated through this receptor. However, the key alterations in gene expression that mediate toxicity are poorly understood. It has been established through characterization of AhR-null mice that the AhR has a required physiological function, yet how endogenous mediators regulate this orphan receptor remains to be established. A picture as to how the AhR/ARNT heterodimer actually mediates gene transcription is starting to emerge. The AhR/ARNT complex can alter transcription both by binding to its cognate response element and through tethering to other transcription factors. In addition, many of the coregulatory proteins necessary for AhR-mediated transcription have been identified. Cross talk between the estrogen receptor and the AhR at the promoter of target genes appears to be an important mode of regulation. Inflammatory signaling pathways and the AhR also appear to be another important site of cross talk at the level of transcription. A major focus of this review is to highlight experimental efforts to characterize nonclassical mechanisms of AhR-mediated modulation of gene transcription.

I. INTRODUCTION

Most transcription factors, by and large, have been defined for their ability to modulate or control a particular biology or physiological response. For example, the estrogen receptor has been associated with reproduction, thyroid, and glucocorticoid receptors with growth and inflammation, and NF-κB with apoptosis and inflammation. The aryl hydrocarbon receptor (AhR) along with its heterodimeric partner the aryl hydrocarbon receptor nuclear translocator (ARNT), together referred to as the aryl hydrocarbon receptor complex (AhRC), have long been associated with an organism’s response to environmental contaminants, most of which are man-made and irrelevant to normal eukaryotic biology. Indeed, for much of the last 40 years, the bulk of experimental effort invested into understanding the function of the AhR has focused on its activation by a wide variety of exogenous compounds, offering little or no insight into the role the AhR might play in normal physiological homeostasis. However, the advent of genetically engineered mice and the identification and emerging importance of nonclassical mechanisms of receptor function offer new avenues of research and may not only help us better define how AhR mediates its toxic effects, but may also shed light on the true physiological role of the receptor.

This review aims to provide a brief overview of nearly six decades of experimental investigation into the nature of aryl hydrocarbon receptor activity and the toxicities associated with the exogenous chemicals that activate it. In doing so, we hope to provide some historical perspective and outline the motivating factors that have driven research in this area in the past. Additionally, we hope to illuminate emerging areas of interest, the potential for discovery, and how new technologies offer an opportunity to explore the AhR function. However, any attempt to summarize data derived from decades of research presents many challenges. As those of us who have been in the field for many years understand, we often struggle with conflicting data derived from experimental systems that were the best available at the time. Yet, we increasingly find these systems rather inadequate to describe the complex interactions that exist between multiple signaling pathways as they converge on transcription. To our colleagues, we would like to apologize in advance if some work is not cited or if our conclusions seem critical. To those new to the field, we hope this serves as something of a road map. With the mapping of the human genome and the advent of DNA array technology, we are able to record changes in gene expression profiles in response to environmental cues. Applications such as chromatin immuno-precipitation (ChIP), fluorescence resonance energy transfer (FRET), small inhibitory RNAs (siRNAs) and real-time PCR allow us to study transcription factor function within the cell and measure associated changes in target gene expression. These advances afford us the opportunity to study the mechanistic determinants of AhR function and, potentially, to uncover the nature of AhR’s true physiological role. Future work may also aid in the discovery of a therapeutic use for AhR activation.

II. BACKGROUND AND HISTORICAL PERSPECTIVE

Organisms are challenged incessantly by potentially hazardous substances of anthropogenic origin. Xenobiotics, such as pesticides, solvents, and many other industrial products, are a major source of environmental pollution and public health concern. Throughout history several events of occupational and accidental exposure to halogenated and nonhalogenated aromatic hydrocarbons (HAHs and PAHs, respectively) revealed many of the health risks associated with exposure to these chemicals.¹

In 1949, the accidental release of HAHs, including TCDD, from Monsanto’s chemical plant in Nitro, West Virginia, resulted in several medical cases of chloracne, liver disease, blood diseases, tumors, and alleged exposure-associated deaths.² In 1957, Sandermann described the discovery and synthesis of 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD), one of the most potent non-genotoxic tumor-promoting substances known. Unfortunately, Sandermann’s laboratory assistant “mysteriously” suffered from chloracne, a painful and disfiguring skin condition, only days after an accidental exposure to the TCDD being synthesized in the laboratory. Despite the earlier association of chloracne with arochlor intoxication,³ it was not until later that it could also be associated with toxic exposure to TCDD. TCDD toxicity is primarily mediated through the aryl hydrocarbon receptor (AhR).⁴ In the post-TCDD discovery era, from 1962 through 1967, the United States introduced the term “chemical warfare” when the compound Agent Orange, containing residual TCDD, was utilized to defoliate large forested areas in Vietnam during Operation Trail Dust and Operation Ranch Hand. Surprisingly, despite the large number of experimental animal data suggesting TCDD as a potent carcinogen, a direct link to human cancer remains a topic of active research.⁵

TCDD is now known to be a trace by-product in the synthesis of the phenoxy-herbicide 2,4,5-trichlorophenoxyacetic acid and chlorophenol.⁶ In 1964 Dow Industrials released the first report on the possible dangers of TCDD exposure after several employees who worked in the production of Agent Orange developed chloracne. In 1976 a severe industrial accident took place at the Industrie Chimiche Meda Societa Azionaria in Seveso, Italy.⁷ Roughly up to kilogram quantities of TCDD and additional TCDD-like chemicals were dispersed into the air with adverse health consequences for the local population.⁸ Thousands of animals died and others were sacrificed to prevent further contamination of the food chain. Thanks to the foresight of Mocaralli (see Ref. 7) blood samples were collected and frozen from each patient for future analysis. A high proportion of human females were apparently born to couples exposed to highly contaminated areas in Seveso. However, this assertion continues to be debated.^9,11 Nevertheless, a convincing role for AhR ligands and the receptor itself in estrogen receptor (ER) function, degradation, and abnormal sex hormone metabolism has been suggested by recent research.^12–16 The increased awareness about the dangers of TCDD and related compounds prompted the industry to improve their production guidelines to protect their employees from chemical exposure and minimize the release of harmful substances into the environment.

In recent history, the Ukraine president Viktor Yushchenko became a widely recognized victim of TCDD poisoning.¹⁷ Unfortunately, an effective treatment for TCDD poisoning and from similar substances has yet to be established, although the use of potato chips fried in the indigestible synthetic fat Olestra™ (Procter & Gamble Company, Cincinnati, OH) has shown promise in this regard.¹⁸ Olestra cannot be absorbed in the gut and it is therefore believed to serve as a carrier of TCDD excreted from the body through feces.¹⁹ The half-life of TCDD in humans is estimated to be around seven years, although genetic and environmental factors are expected to minimize or prolong its clearance from human tissues.

III. IDENTIFICATION AND CLONING OF THE AHR

During the late 1950s it became clear that the efficacy or adverse effects of therapeutic drugs often could be associated with simple inheritable genetic differences (polymorphisms) between individuals. Despite hints dating as far back as Garrod’s 1906 publication entitled Inborn Errors of Metabolism,²⁰ what apparently delayed this realization was the confounding influence of the environment (e.g., lifestyle) on the phenotypic manifestation of genetic traits, as well as the fact that most responses to xenobiotics involved several nonlinked genetic loci. Nevertheless, through advances in statistical and genetic analyses, it became widely recognized that at least some of these metabolic differences between individuals had simple traceable genetic origins that segregated in Mendelian frequencies.

Similarly, differences in susceptibility to HAHs, PAHs, and other xenobiotics, even between organisms of the same species, also could be traceable to their genetic makeup. These polymorphic genetic loci were recognized to play a role in the controlled expression and/or stability of xenobiotic metabolizing enzymes, receptors, transporters, and perhaps unidentified accessory/regulatory factors. Preceding the identification and cloning of the aryl hydrocarbon receptor the upregulation of aryl hydrocarbon hydroxylase (AHH) activities, measured in vitro by the formation of 3-hydroxybenzo[a]pyrene from the parent compound benzo[a]pyrene (B[a]P), could only be readily detected in liver extracts and some extra-hepatic tissues from so-called “responsive” mice.²¹ The crossing and backcrossing of multiple inbred strains of mice further led to the identification of the Ah (aryl hydrocarbon) genetic locus, suspected to control the inducible expression of the spectrally distinct cytochrome P₁-450 (today known as CYP1A1)^22,23 as well as other AHH activities. This genetic locus segregated primarily as an autosomal dominant trait in specific crosses (e.g., C57BL × DBA mice) where the dominant responsive allele was denoted as Ah^b (in C57BL/6 and B6 strains) and the recessive nonresponsive as Ah^d (in DBA/2 strains). As expected, the complex inheritance pattern of AHH responsiveness observed between a wider screen of wild-type and inbred mice strains strongly suggested the additional contribution of nonlinked loci to the inducibility of AHH/P₁-450 activity.²⁴ Regardless, the rescue of “unresponsive” DBA mice with the potent AHH inducer TCDD, for which AHH activity approached that of 3-MC-treated C57BL/6 mice, sparked a new theory, namely, perhaps the Ah locus encoded a gene product, such as a receptor protein, central to the downstream induction of AHH/P₁-450.²¹ Thus, a mutated AhR in the DBA mice strain was suspected as the culprit for a failure to recognize the 3-MC signal.^25–27 This hypothesis was validated only after the AhR was cloned. A comparison between the AhR from C57BL and DBA strains of mice indicated that an alanine to valine substitution at position 375 (381 in human AhR) in C57BL AhR and a mutation at the stop codon resulted in an elongated carboxyl terminus in the AhR^d allelic variant with reduced affinity for [³H]TCDD.²⁸

Several research observations reinforced the AhR theory. For example, the peritoneal injection of C57BL/6 mice with [³H]TCDD resulted in high levels of radioligand retention in the liver, especially when compared to the nearly absent levels in the nonresponsive DBA/2 strain. Crosses between C57BL/6J × DBA/2J also resulted in offspring (B6D2F₁/J) displaying intermediate binding/responses, an observation that correlated with the segregation of a simple autosomal dominant trait controlling AHH inducibility.²⁷ Extensive in vitro and comparative animal toxicology studies emphasizing structure-activity relationships further suggested that a soluble cytosolic receptor protein with varying affinities for TCDD-like compounds was likely responsible for the induction of AHH/P₁-450 activity.^27,29–32 A model began to emerge resembling that of the already established steroid receptor pathway.³³ In brief, the cytosolic AhR, upon binding of TCDD, would translocate into the nucleus and activate genes controlled by the Ah genetic locus. However, until better methods (e.g., immunofluorescence^34–36) were available, the location of the AhR complex proteins in nonstimulated cells remained a controversial issue for several years.^37,38

The AHH/P₁-450 “inducer/receptor-complex” was later confirmed to undergo translocation into the nucleus of responsive cells following the binding of radioligand, an event that clearly preceded the induction of AHH/P₁-450 activity.^39,40 A comparative study of the nonresponsive VERO and HTC mammalian cell culture lines and the responsive H-4-II-E and Hepa-1 cell lines also indicated that the translocation of the AhR into the nucleus, as in HTC cells, does not guarantee that AHH/P₁-450 induction would proceed.⁴¹ These results further highlighted that both structural and regulatory genes other than the proposed AhR were likely controlled by the Ah locus. Work conducted by Okey and colleagues,⁴⁰ and later revisited by Hanna and coworkers,⁴² identified the presence of ~ 6 S (at high ionic strength) and ~ 9 S (at low ionic strength) receptor complexes, respectively. Both complexes had a measurable [³H]TCDD binding capacity that was virtually absent in extracts from DBA/2N mice. These complexes also displayed sensitivity to proteases but not to DNases and RNases. [³H]TCDD binding could be competitively displaced with known P₁-450 inducers but not with established steroid hormones or other classic P-450 activity inducers (e.g., phenobarbital). In addition, only known inducers of P₁-450 (CYP1A1) and unlabeled TCDD could effectively compete for binding to the receptor complex.⁴⁰

The lessons learned about the glucocorticoid receptor family were being progressively applied to the rapidly growing AhR field. For example, the use of 20–30 mM molybdate in sucrose gradient fractionation of the glucocorticoid receptor (GR) was found to stabilize the ligand-binding conformation of the receptor.⁴³ However, the murine and rat AhRs appeared to benefit only partially from this treatment by showing increased thermostability, as well as retaining their ligand-binding activity under high ionic strength.⁴⁴ However, the human AhR did appear to greatly benefit from enhanced stability in vitro by molybdate treatment.⁴⁵

A key defining moment toward the future cloning and characterization of the AhR was the synthesis of 2-azido-3-iodo-7,8-dibromodibenzo-p-dioxin,⁴⁶ a photoaffinity ligand capable of covalently binding the AhR when excited in the ultraviolet frequency. At first, two polypeptides with a mass of 70 kDa and a 95 kDa were suspected to represent the Ah receptor, given that the photoaffinity ligand could be displaced with known Ah agonists from both polypeptides (e.g., TCDD). However, the 70 kDa polypeptide was found to be no more than a proteolytic fragment of the 95 kDa polypeptide in Hepa-1 cells.⁴⁷ The suspected protease had similar characteristics to the Ca⁺²-dependent calpain II and was ruled as the culprit of this isolation “artifact.” Hence, the addition of EDTA was recommended to stabilize the AhR during purifications. With the advent of a new molecular photoaffinity probe, the AhR field exploded with numerous research findings. For instance, the characterization of several inbred mice strains with the new photoaffinity radioligand led to the identification of two AhR allelic forms such as the 95 kDa (Ah^b−1 allele) from the C57, C58, and MA/MyJ mice strains, and the 104 kDa (Ah^b−2 allele) from the C3H/HeJ, BALB/cByJ, and A/J strains.⁴⁸ Ligand competition-binding protocols with the photoaffinity ligand were also being established, which helped estimate the binding affinity and characteristics of other suspected AhR ligands.^29,49 In 1988, Perdew and Poland described the partial purification of the AhR from C57BL/6J mice.⁵⁰ This event led to the eventual purification, N-terminal sequencing, and production of antibodies against the AhR.^51,52 The generation of antibodies capable of immunoprecipitating the glucocorticoid receptor-associated hsp90 ^53–55 and the AhR helped identify the hsp90 as part of the unliganded AhR complex with two different approaches.^56,57

Before the cloning of the AhR, a subunit of the liganded complex was identified and thought to be required for the translocation of the AhR into the nucleus.^58,59 Briefly, a cell line expressing a “functional” AhR but uninducible for P₁-450 (CYP1A1) was systematically transfected with a cDNA library. The expression of one of these cDNA constructs restored the translocation of the AhR and the expression of CYP1A1. The product of this gene was therefore named ARNT (Ah receptor nuclear translocator), although the name itself is a misnomer, for ARNT is not directly involved in the AhR “translocation” event per se. ARNT was found to share homology with other known Drosophila proteins such as Per and Sim and contained a suspected DNA-binding/dimerizing domain termed basic helix-loop-helix.⁵⁸ Together the Per, ARNT, and Sim proteins are the representative members of the PAS domain superfamily. PAS domain proteins have been associated with important functions in transcription by relaying environmental signals (e.g., light, oxygen, and xenobiotics) to the cell⁶⁰ (for an excellent recent review see Ref. 61). The AhR appeared as a unique member in the PAS family because it could be activated by exogenous ligands. However, a putative high-affinity endogeous ligand for the AhR has yet to be identified and thus remains classified as an orphan receptor.

A distinction between the cytosolic 9 S and nuclear 6 S forms of the AhR was later established through chemical cross-linking studies.^62,63 The AhR complex was found as a tetrameric complex containing the AhR, two molecules of hsp90, and an unidentified ~ 43 kDa protein that sedimented at 9 S in sucrose gradients. This tetrameric complex could be transformed into a dimeric 6 S complex in the presence of ligand.⁶² The ligand-activated 6 S AhR complex was also paradoxically composed of a heterodimer containing the signature 95 kDa AhR and an “unknown” 85 kDa polypeptide (ARNT). Often, the 9 S form of the AhR is referred to as the “latent” form and the 6 S form as the “transformed” form of the receptor.

Numerous inbred strains of mice were further characterized with the photoaffinity ligand. This work led to the identification of four polymorphic alleles, namely, the 95-kDa (Ah^b−1), 104-kDa (Ah^b−2), 105-kDa (Ah^b−3), and the 104-kDa (Ah^d) allelic variants. Given the availability of antibodies and a preestablished AhR purification protocol, the N-terminal sequence of the AhR^b−1 was determined. This aided the creation of degenerate primers used in the cloning of the AhR^b−1 gene.⁶⁴ The AhR^b−1 was found to share many features with the previously cloned ARNT. For instance, it contained a bHLH region followed by two 51-amino acid Per-ARNT-Sim (PAS) A-B repeats⁶⁰ and an N-terminal glutamine-rich region.^58,65 The similarities between the AhR and ARNT, the presence of a bHLH, and previous observations on their interaction with enhancer DNA regions at the CYP1A1 promoter led to the hypothesis that the AhR and ARNT could be heterodimeric partners. In addition, the covalent binding of the photoaffinity ligand to the A–B region suggested that this region could be structurally part of a ligand-binding pocket that serves as a switch for the transformation of the AhR into a DNA-binding conformation.⁶⁴ Finally, the glutamine-rich region was suspected to serve as a transcription activation domain given its presence in other transcription factors.^66,67 The human AhR was subsequently cloned by the screening of a cDNA library generated from the hepatoma cell line Hep-G2.⁶⁸ In comparison with the murine AhR^b−1, the human receptor was 6 kDa larger. Most of the differences between the murine and human receptors were found in the carboxyl terminus. Sequence comparison revealed < 60% conservation between species, whereas the basic region, the helix-loop-helix, and PAS domains displayed about 100%, 98%, and 87% levels of conservation, respectively.

The creation of a photoaffinity ligand and antibodies for the AhR facilitated the use of numerous biochemical approaches that accelerated our understanding of AhR biology. At the same time, differences between responsive and nonresponsive animals to xenobiotics and the low degree of conservation between the mouse and human AhR were quickly being recognized as major obstacles, especially for the extrapolation of animal toxicological data to humans.

IV. CHARACTERIZATION OF AHR AND ARNT FUNCTIONAL DOMAINS

Following the cloning of the AhR and ARNT, an extensive effort to map their functional and interaction domains was initiated.^68–73 Detailed structure-function relationships were carried out in the laboratory of Oliver Hankinson.^74,75 Briefly, through immunoprecipitations of the murine AhR, deletion mutants of [³⁵S]ARNT were monitored for their ability to interact with the AhR in the presence or absence of TCDD.⁷⁵ The bHLH together with the PAS-A and PAS-B domains (amino acids 70–474 of the murine ARNT) were all required for optimal heterodimerization to occur in the presence of TCDD. Deletion of the basic domain of ARNT had virtually no effect on the dimerization process. The ability of the AhR:ARNT heterodimer to bind double-stranded DNA oligos containing the consensus XRE sequence also required the basic region of ARNT. Interestingly, an ARNT construct containing only the region from the bHLH through the PAS-A domain was unable to bind its XRE while still being able to heterodimerize with the AhR. Conversely, a construct bearing the bHLH, PAS-A, and PAS-B domains (bHLHAB) was able to restore XRE binding. It was therefore hypothesized that the PAS-B domain may help the ARNT protein fold in a way to maximize the interaction between the basic region of ARNT and DNA. The presence of an oligo containing a xenobiotic responsive element alone did not catalyze the heterodimerization of the AhR with ARNT. Thus, a ligand seemed to be required for AhR:ARNT heterodimerization to occur. Finally, a construct containing only the bHLH, PAS-A, and PAS-B domains could not restore in vivo function of ARNT. Therefore, other regions of ARNT were suspected to contribute to its transcriptional activity, namely, the region containing amino acids 474–627 of the murine ARNT.⁷⁵

To characterize the transactivation domains of the AhR and ARNT, chimeras were initially generated with the DNA binding domain of the glucocorticoid receptor.⁷⁶ Since the bHLH region of the AhR and ARNT are required for their heterodimerization, this fusion-protein approach seemed like a proper approach to study the individual transactivation potential of the AhR and ARNT. In this study the glucocorticoid DNA binding domain was fused to ARNT and the AhR, generating GRDBD-ARNT and GRDBD-AhR chimeras. In general, GRDBD-ARNT chimeras were more transcriptionally active than the GRDBD-AhR variety. Furthermore, in the case of ARNT, its transactivation domain potency varied between cell lines, suggesting that cell- and promoter- specific activities are possible for this transcription factor. However, an AhR chimeric construct containing residues 83–593 reduced the ligand-induced activity while a chimeric construct based off residues 594–805 led to a constitutively active AhR that was as potent as the GRDBD-ARNT on an MMTV promoter. Therefore, a region close to the PAS homology/ligand-binding region was suspected to impart regulatory control to the AhR. For instance, the construction of a GRDBD-AhR containing residues 340–805 restored the repressive effect and the interaction between the AhR and the hsp90. Based on this and previous observations indicating that an AhR stripped from hsp90 could not bind ligand,⁷⁷ a possible role for hsp90 in repressing constitutive AhR activity was suggested. In this model, the hsp90 protein would maintain the AhR in a latent state in the absence of a stimulus. Upon binding of a ligand, a conformational change would cause the concerted dissociation of AhR from the hsp90 complex, its heterodimerization with ARNT, and subsequent activation of target genes.

The use of chimeric AhR and ARNT containing the basic DNA binding domain of GR to study the transactivation domains was later criticized. ⁷⁴ As a result, a complementary analysis of the AhR domains was performed by Fukunaga and colleagues,⁷⁴ similar to previous domain-deletion studies with ARNT.⁷⁵ The ligand-binding region of the AhR was mapped to the PAS B domain. Earlier, through the use of a photoaffinity ligand, the actual ligand-binding pocket was found to encompass the PAS B domain.⁶⁴ Both the bHLH and PAS B domains of the AhR were required for hsp90 binding. Although ARNT could enhance the dissociation of the AhR from hsp90, ligand alone was sufficient to facilitate the dissociation of the AhR from hsp90 in vitro.⁷⁴ Whether the release of AhR from hsp90 in the presence of ligand happens in vivo has yet to be fully established, although evidence suggests that dissociation of the AhR from hsp90 may only be required for its heterodimerization with ARNT.⁷⁸ Indeed, the presence of monomeric AhR may not stably exist in the cell. The regions of the AhR responsible for its interaction with the hsp90 were further mapped to amino acids 1–166 and 289–347,⁷⁹ whereas the use of previously published deletion constructs of the hsp90 ⁸⁰ suggested that amino acids 272–617 (“the middle portion”) of the hsp90 were involved in AhR binding.⁸¹

V. BASIC ASPECTS OF AHR-MEDIATED GENE REGULATION

The nuclear uptake of the AhR was a phenomenon observed after the treatment of C57BL/6J mice and responsive cell cultures with [³H]TCDD.^39,40,82 It seemed evident that ligand binding was required for the AhR nuclear translocation event,⁸³ whereas limited proteolysis of the receptor complex suggested the presence of a DNA binding component of the AhR complex.⁸⁴ Indeed, subsequent studies involving mutant mice harboring an expressing nuclear localization-deficient form of the receptor were found to be resistant to the toxic effects of TCDD.⁸⁵ However, these nonspecific DNA binding studies only provided limited information about the DNA motif(s) or response element(s) recognized by the activated AhR. Hence, another group took a genomic approach to determine the regions of the P₁-450 (CYP1A1) promoter that contained TCDD-responsive elements. HAV (highly expressing variant) cells expressing high levels of the P₁-450 gene were selected for a genomic screen. Several clones containing P₁-450 DNA were screened and one containing a 2.58 kb fragment was subcloned upstream of a bacterial chloramphenicol acetyl-transferase gene (CAT) reporter.⁸⁶ In this manner the 5′-end of the P₁-450 gene was found to contain at least two elements that were responsive to several AhR ligands in responsive cell lines and contained a suspected cycloheximide-sensitive repressor binding site. Through a deletion analysis of the P₁-450 promoter, an AhR/TCDD-dependent enhancer region was identified.^87,88 This enhancer region was later documented as the dioxin responsive element (DRE),⁸⁹ although the term XRE (xenobiotic responsive element) or AhRE (aryl hydrocarbon responsive element) has also been used to describe the DNA elements recognized by the activated AhR. A third DRE was later identified and the consensus DNA sequence for the binding of the AhR to its DRE was suggested. ^90,91 This DRE sequence was suggested to be 5′-TA/TGCGTG-3′, a sequence that was present in all 3 DREs described to date. The asymmetrical nature of the DRE also differed from the typical palindromic responsive elements recognized by steroid receptors. The DRE flanking sequences in the enhancer/promoter appeared to be equally important because the subcloning of oligos containing the DRE sequence alone was not sufficient to restore inducible expression of a reporter gene.^90,91 The core DRE sequence recognized by the AhR complex was further refined through methylation protection and interference studies to 5′-T/GCGTG-3′, and four copies of this element were mapped to the P₁-450 promoter.^92,93

The AhR was initially suspected to bind as a monomer to XRE sequences on double-stranded DNA. This was based on the notion that the response elements for steroid receptors were symmetrical (palindromic) and the steroid receptors bound DNA as homodimers. However, several lines of evidence suggested a heterodimeric structure for the DNA-binding AhR complex. Among these were that the transformation of the AhR into a DNA binding complex required a non–ligand-binding component.⁹⁴ Through cross-linking studies and limited proteolysis studies, a ligand binding component and an accessory component of the activated AhR were identified.⁹⁵ Finally, following the cloning of the protein ARNT and the AhR itself, the Hankinson group identified ARNT as a component of the activated AhR in nuclear extracts.⁹⁶ In vivo DNA footprinting demonstrated that the AhR bound a DNA element with the sequence 5′ CACGCNA/T 3′ and another component interacted with a G-rich region.⁹⁷ ARNT was found to bind a 5′-GTG-3′ half-site that resembled that of the E-box element utilized by other members of the helix-loop-helix family of transcription factors. Since nucleotides flanking the core DRE sequence had an impact on AhR complex affinity for its DNA element, the putative XRE was suggested to be 5′-TTGCGTGAGAA-3′, where the AhR binds the second 5′ thymine and ARNT the third 5′ thymine.⁹⁸ With the identification of these elements and the cloning of the AhR and ARNT, efforts turned toward the identification of novel AhR-regulated genes. The list of genes directly regulated by the AhR continues to grow. Some examples of AhR-regulated genes are listed in Table 1.

TABLE 1.

AhR-Regulated Genes

Gene	References
CYP1A1	97
CYP1A2	99–101
CYP1B1	102, 103
aldehyde dehydrogenase 3	104, 105
UGT1A1	106
Ya subunit of GST
CYP2S1	107
NAD(P)H:Quinone-oxireductase	108
murine epiregulin	109
ecto-ATP	110
δ-aminolevulinic acid synthase	111
Prostaglandin endoperoxide H synthase 2	112
MDR1 and BRCP	113, 114
AhRR	115
p27kip1	116

VI. AHR ACTIVATION

The AhR is considered an orphan receptor, given that no putative high-affinity endogenous ligand has been identified to date. An endogenous role for the AhR has only been suspected from its ability to control the expression of drug-metabolizing enzymes when stimulated by synthetic substances belonging to the HAH and PAH families as well as numerous dietary substances. As such, a large number of AhR ligands can induce their own metabolism and clearance from the body by inducing drug-metabolizing enzymes of the phase I (oxidation) and phase II (conjugation), and transporters of the phase III (excretion) metabolic pathways.¹¹⁷ Unfortunately, the metabolism of some compounds (e.g., B[a]P) can lead to the formation of nucleophilic derivatives that bind covalently to DNA and proteins. Occasionally these adducts may form in a tissue-specific manner.¹¹⁸ The formation of DNA and protein chemical adducts has been associated with an increased risk for carcinogenesis.

Some widely recognized synthetic inducers of the AhR pathway include compounds such as B[a]P, 3-methyl cholanthrene, β-naphthoflavone, TCDD, 2,3,7,8-tetrachlorodibenzofuran, and 3,4,3′,4,′5-pentachlorobiphenyl.¹¹⁹ However, numerous studies have suggested that dietary substances can also readily activate AhR-regulated genes. Some examples include indole-3-carbinol,¹²⁰ curcumin,¹²¹ and quercetin.¹²² Some ligands can also exert both agonist and antagonist activities at different concentrations, such as resveratrol¹²³ and galangin.¹²⁴ It appears logical that the high ligand promiscuity of the AhR may be indicative of its proposed role in the adaptation of organisms to environmental chemical challenges. On the other hand, putative endogenous ligands for the AhR have been suspected to exist for numerous reasons. For example, hydrodynamic shearing of cells results in the formation of arachidonic acid metabolites suspected to induce CYP1A1 in an AhR-dependent manner.¹²⁵ A rat model of supplemental oxygen treatment for lung insufficiency revealed that hyperoxia treatment of rats led to AhR-dependent CYP1A1 induction.¹²⁶ The incubation of epidermal cells in a methylcellulose suspension activated CYP1A1 expression and this effect could be prevented by treatment with the AhR antagonist α-naphtoflavone.¹²⁷ UV irradiation could also possibly lead to an AhR-dependent CYP1A1 induction by the formation of active tryptophan oxidation products.¹²⁸ There are also examples of substances (e.g., omeprazole) that can readily activate the AhR, yet are not direct AhR ligands.^129,130 This leaves open the possibility that the AhR can be activated in a ligand-independent manner. In fact, AhR mutants that fail to bind ligand were used to demonstrate that they can still heterodimerize with ARNT and mediate transcription in the absence of ligand binding.¹³¹ This offers formal proof that the AhR does not require ligand binding to be transcriptionally active. Developmental studies of AhR “knockout” mice have revealed important roles for the AhR in the development of the liver, closure of the ductus venosus,¹³² the immune system,¹³³ and the control of many “ligand-independent” cellular activities. ¹³⁴ These observations strongly indicate a function for the AhR beyond the adaptive and toxic responses to xenobiotics. For instance, the affinity for exogenous ligands is not conserved between certain mouse strains and between species. Most notably there is a tenfold difference in affinity between C57BL/6 mice and human AhR.¹³⁵ This lack of conservation may suggest that either a ligand-independent activation of the AhR is important for endogenous AhR function, or a key endogenous ligand’s ability to activate the AhR is conserved.

VII. THE LATENT AHR MULTI-SUBUNIT COMPLEX

Prior to the cloning of the AhR and its designation as a bHLH-PAS protein,⁶⁴ it was considered possible that the AhR, due to its analogous biochemical properties, was a member of the steroid receptor family.^56,77,136 Although the AhR shares no sequence homology, it is functionally similar to some steroid receptors with respect to its regulation by chaperone proteins, principally HSP90. Glucocorticoid (GR) and progesterone receptor (PR) maturation has been well studied using purified proteins, and many of the conclusions obtained from these experimental systems have been extrapolated to pertain to the AhR. In the early 1990s, the Toft and Pratt laboratories demonstrated that rabbit reticulocyte lysate contained all the molecular components necessary to transform incompetent forms the PR and GR into competent hormone-binding receptors.^137,138 Subsequently, the presence of the chaperone heat shock proteins (HSPs) 40, 70, and 90, as well as the cochaper-ones HOP/p60 and p23 were determined to be sufficient for achieving optimal hormone-binding status, although only HSP70 and HSP90 are absolutely required to form a less efficient binding-competent receptor.^139,140 Of the five above-mentioned proteins required for GR and PR maturation, HSP90 and p23 are known components of the mature latent AhR complex. However, it is possible, if not likely, that the other three proteins are also involved in regulating the early steps of AhR maturation. The following sections will highlight several aspects of AhR regulation with respect to its association with HSP90, p23, and the immunophilin-like XAP2 protein. It is essential to recognize that activation of the AhR transcriptional complex does not stem simply from a bimolecular interaction of receptor and ligand, but rather is regulated through the interplay between the AhR and numerous ancillary factors. The chaperone and cochaperone components serve to maintain proper folding, ligand-binding competency, and overall transcriptional effectiveness of the AhR. An overview of our current understanding of the transformation of the latent AhR to an active transcriptional complex is outlined in Figure 1.

FIGURE 1.

Transformation of the latent AhR tetramer to an activated transcriptional complex.

VIII. 90 KDA HEAT SHOCK PROTEIN

As alluded to above, early biochemical characterization of the AhR complex was performed using the GR as a model due to the “virtually undistinguishable” physiochemical properties between the two receptors.¹⁴¹ As a result of these comparative studies, the inactive AhR, like the GR, was demonstrated to exist in a heterocomplex with HSP90.^56,57 The association of HSP90 with the AhR is representative of the transcriptionally inactive receptor since the AhR, when bound to HSP90, is unable to heterodimerize with its DNA-binding partner ARNT. HSP90 and ARNT both interact with the AhR via the receptors HLH and PAS domains, thereby making their association with the AhR mutually exclusive.^74,75,79,142 Similar to a number of other HSP90 substrates, the AhR interacts with HSP90 through the middle portion of the chaperone.⁸¹

Upon ligand activation, the human and mouse AhR translocate into the nucleus still associated with HSP90, demonstrating that the cytoplasmic AhR complex does not have to shed its chaperone machinery before nuclear uptake.^78,143 It is within the nucleus that ARNT heterodimerizes with the AhR. Whether ARNT actively displaces HSP90 from the AhR or other factors stimulate HSP90 dissociation is not yet understood. In vitro, however, HSP90 is released from the ligand-bound AhR when coincubated with cell extracts from ARNT-containing but not ARNT-deficient mouse hepatoma cells.¹⁴⁴ This study suggests that ARNT, or a component functionally tied to ARNT, is involved in the ligand-induced displacement of HSP90 from the AhR.

The GR and PR require the association with HSP90 in order to maintain hormone-binding competence ¹⁴⁵ and references within). Analogously, AhR-HSP90 complex formation is important to produce a receptor that has maximal ligand-binding capacity.^146,147 In a manner analogous to the GR, the monomeric AhR (isolated via high salt stripping of receptor-associated proteins) derived from mouse hepatoma cells has no ligand-binding capacity. However, the supplementation of the monomeric AhR with rabbit reticulocyte lysates, containing HSP90, partially restores ligand binding while HSP90-negative wheat germ lysate has no effect.⁷⁷ The AhRs dependence on HSP90 association for ligand binding appears to vary between species, however, because the rat AhR is more sensitive to HSP90 disruption than the guinea pig and rabbit forms of the receptor.¹⁴⁸ Additionally, disruption of the stable association between the AhR and HSP90 in living cells results in the rapid degradation of the receptor,¹⁴⁹ highlighting the importance of the AhR-HSP90 interaction in receptor signaling.

HSP90 is a known phosphoprotein, but it is unclear to what extent phosphorylation of HSP90 serves to regulate AhR function. A recent study suggests that phosphorylation of the constitutively expressed HSP90β isoform on two serine residues modulates the affinity of HSP90 for the AhR. The authors suggest that the unphosphorylated HSP90β isoform has a high affinity for the AhR and subsequently higher transcriptional activity. ¹⁵⁰ In yeast, it appears that the AhR may preferentially utilize the constitutive HSP90β isoform over the more inducible HSP90α chaperone.¹⁵¹ Further studies need to be performed to corroboratethe importance of these findings.

VIX. THE HSP90-ASSOCIATED CO-CHAPERONE

The phosphoprotein p23 is a small, acidic protein that is ubiquitously expressed in virtually all tissues. ¹⁵² It interacts with and stabilizes the ATP-bound conformation of HSP90. It is through this interaction that p23 incorporates into numerous steroid receptor complexes¹⁵³ and references within). Although the amino acid residues required for HSP90-p23 interaction have not been determined, it is known that it binds at the amino-terminal portion of HSP90 ¹⁵⁴, and this interaction can be disrupted by agents that compete at the ATP binding site, such as geldanamycin.¹⁵⁵ The entry of p23 into steroid receptor complexes occurs near the end of their maturation process, and it is likely that this is also true for the AhR. Experimental evidence shows that p23 is capable of modulating steroid receptor function by stabilizing the mature complex and altering its hormone binding capacity ¹⁵³ and references within). Additionally, p23 has been demonstrated to regulate receptor function by enhancing the hormone-stimulated substrate release from HSP90 in an ATP-dependent manner.¹⁵⁶ In vivo, p23 is essential for the perinatal survival of the developing mouse, and mortality associated with the loss of p23 expression appears to be due, at least in part, to defective GR signaling as evidenced by the disruption of glucocorticoid-stimulated lung development. ¹⁵⁷

To date, the number of studies performed to assess p23’s role in regulating AhR function are limited, but they universally find that p23 acts to stimulate AhR activity. The ability of p23 to enter into the AhR complex was first demonstrated in vitro by receptor complex reconstitution experiments using individually purified components.¹⁵⁸ Later studies using in vitro translated AhR demonstrated that p23 presence in the AhR complex is necessary to prevent the spontaneous formation of an AhR-ARNT heterodimer in the absence of ligand.¹⁵⁹ This spontaneous heterodimer formation could be blocked by the addition of molyb-date, which is known to stabilize, or mimic, the ATP-bound (i.e., p23-bound) conformation of HSP90 complexes.¹⁶⁰ Additionally, the increased formation of AhR-ARNT DNA-bound complexes in the presence of p23 has been reported, as determined by mobility gel shift assay.¹⁶¹ AhR activity is elevated in yeast reporter systems containing expression vectors for human p23 or the yeast p23 homolog SBA1 in strains null for the homolog.¹⁶² Finally, p23 has been hypothesized to enhance nuclear uptake of the AhR following ligand treatment by increasing the ability of the receptor to recognize the import protein importin β.¹⁶³ Taken together, the limited number of experiments performed suggests that p23 acts as a stimulatory factor in regulating AhR activity.

X. THE HEPATITIS B VIRUS X-ASSOCIATED PROTEIN 2

An additional class of proteins known to be highly involved in regulating steroid receptor function is the immunophilins. Immunophilins derive their name from their ability to bind immunosuppressive drugs. The large immunophilins, FK506 binding proteins (FKBP) 51 and 52, as well as cyclophilin-40 have all been established to exist in multiple steroid receptor complexes (Ref. ¹⁶⁴ and references within). Cochaperones, like p23, enter into steroid receptor complexes once the receptor achieves its mature folded state and therefore do not appear to be a component of the receptor folding machinery. The GR has been a model complex for a number of mechanistic studies performed to understand immunophilin-mediated regulation of steroid receptor function by FKBP51 and 52. Interestingly, although they are closely related proteins, FKBP51 acts as a repressor of the GR, whereas the presence of FKBP52 in the GR complex enhances receptor transcriptional activity.^165,166 Also, it has been suggested that hormone binding to the GR stimulates the switching of the binding status of receptor from an FKBP51-bound form to an FKBP52-bound form, thereby enhancing nuclear uptake of the GR.¹⁶⁷ The recent creation of the Fkbp52^−/− mouse for the first time has demonstrated that the immunophilin class of cochaper-one proteins play a significant role in regulating the functional properties of certain nuclear receptors, like the PR¹⁶⁸ and androgen receptor.¹⁶⁹ The loss of the FKBP52 protein in the null mouse line results in both male and female infertility due to severe deficiencies in androgen or progesterone responsiveness, respectively.

The latent AhR complex contains an immunophilin-like component termed the hepatitis B virus X-associated protein 2 (XAP2). XAP2 is a 37 kDa phosphorylated protein¹⁷⁰ that was first identified through its ability to interact with the hepatitis B virus X protein.¹⁷¹ Cross-linking studies performed by Perdew in 1992 showed that, in addition to two molecules of HSP90, the AhR tetrameric complex contains a ~ 46 kDa protein. ¹⁷² Five years later, three separate laboratories identified that the murine, human, and simian-forms of XAP2 (also known as AIP or ARA9) are components of the latent AhR complex.^173–175 XAP2 displays 28.9% amino acid identity to the immunophilin FKBP52 in its amino terminus (amino acids 9–90).¹⁷⁶ These residues of XAP2 overlap with the region in FKBP52 that harbors the peptidyl-prolyl isomerase activity associated with immunophilins. So far, however, XAP2 has not been demonstrated to possess this enzymatic activity. Also, unlike the FKBPs, XAP2 does not possess the ability to bind the immunosuppressant drugs FK506 or rapamycin,¹⁷⁷ and XAP2 is therefore by definition not an immunophilin but is referred to as immunophilin-like. Like the immunophilins, XAP2 interacts with the receptor complex through the utilization of tetratricopeptide repeat (TPR) domains. XAP2 has three TPR motifs in its carboxy-terminal half. Its carboxy-terminal–most TPR displays homology to the TPRs in FKBP52, and is important in mediating the direct interaction between XAP2-AhR and XAP2-HSP90.^81,177 In addition to the TPR motif in XAP2, amino acid residues in the extreme carboxy-terminus of XAP2 are also needed to mediate its interaction with the AhR. Alanine substitution in any of the last four amino acids in XAP2 results in the complete loss of interaction with the AhR, while HSP90 binding is still observed even when the last five amino acids are completely deleted.¹⁷⁸

Overexpression of the mouse AhR in a cell culture causes a shift in receptor localization from the cytoplasm to the nucleus. However, the receptor can be redistributed into the cytoplasm on co-expression of XAP2, implying that AhR cellular localization may rely on the presence of XAP2 in the complex.^179,180 When these cells are treated with the nuclear export inhibitor leptomycin B, the AhR remains in the cytoplasm, suggesting that XAP2 blocks the rapid nucleo-cytoplasmic shuttling of the AhR complex that normally occurs.¹⁸¹ A recent study by the Pollenz laboratory, however, has shown that silencing XAP2 expression in mouse hepatoma cells does not result in the nuclear accumulation of AhR, nor does it appear to alter AhR nucleo-cytoplasmic shuttling, confounding previous observations.¹⁸² To further complicate matters, overexpression of the human AhR in the absence of XAP2 coexpression does not result in nuclear accumulation of the receptor without XAP2 coexpression highlighting one of the numerous functional differences between the two species.¹⁸³

It has been observed that increasing amounts of XAP2 expression in a cell culture significantly raises the total cellular concentration of the unliganded AhR.^81,184 One report has attributed this effect to XAP2-mediated protection of the AhR against ubiquination, and subsequent reduced levels of proteosome-mediated degradation.¹⁸⁵ Ectopic expression of XAP2 has been demonstrated to enhance AhR transcriptional activity. The exact level of enhancement, however, can be difficult to assess. In the presence of high XAP2 levels, elevated constitutive and induced expression of reporter genes has been observed using chimeric AhR fusion constructs with alternate DNA binding motifs, such as GAL4 or LEXA.^177,179 In a yeast model system utilizing a β-galactosidase reporter gene to monitor AhR activity, the ED₅₀ of the dose-response curve is decreased by approximately fivefold when XAP2 is expressed along with a LEXA-AhR fusion protein.¹⁷⁷ When utilizing endogenous AhR levels in mouse hepatoma cells, however, XAP2 appears to cause a more modest (~ twofold) increase in ligand-induced AhR-driven luciferase activity, while the basal activity remains essentially unaltered.¹⁷⁵ XAP2-mediated enhancement of AhR activity is generally attributed to the elevated cytosolic AhR levels achieved in the presence of high amounts of XAP2, thereby increasing the available ligand-binding sites in a cell. Contrasting reports show that XAP2 can act as a transcriptional repressor or have only a slight effect on AhR transcriptional activity. The Perdew lab has demonstrated that coexpression of XAP2 with the AhR results in an overall repression of human and mouse AhR activity. ^181,183 It should be noted that under conditions in which coexpression of XAP2 results in repression of AhR signaling, little increase in overall AhR levels is observed.^183,186 Additionally, the mutation of a single amino acid residue (tyrosine 408) in the mouse AhR disrupts XAP2 binding while maintaining HSP90, p23, and the ligand-binding potential. This mutant AhR displays increased basal and ligand-induced transcriptional activity relative to wild-type AhR.¹⁸⁶ Furthermore, silencing of XAP2 expression in cell lines expressing different AhR alleles demonstrates that AhR stability is independent (Ah^b−2) of or only modestly dependent (Ah^b−1) on XAP2 expression to maintain AhR cellular levels.¹⁸⁷ These later studies suggest that enhancement of AhR signaling by XAP2 in cell culture may be an experimental artifact due to the use of overexpression systems. This idea is supported by the recent creation of transgenic mouse lines that express high levels of hepatocyte-specific XAP2. In these mice, no increase in cellular AhR levels is observed, nor is there any change in ligand-induced AhR activity compared to wild-type animals.¹⁸⁸ Clearly, future work must be performed to reconcile the published discrepancies with respect to AhR regulation by XAP2.

XI. TRANSCRIPTIONAL COACTIVATION OF THE AHRC

Shortly after the cloning of AhR and ARNT, several investigators set about the task of defining the functional domains of the receptor complex. In particular, attention was focused on the DNA-binding, dimerization, and transactivation capabilities of AhR and ARNT, with the latter involving studies utilizing heterologous reporter systems. These systems included chloramphenicol acetyl-transferase (CAT) and luciferase reporter plasmids serving as transcriptional readouts and, as such, represent unchromatinized templates. Gal4 fusions of AhR and ARNT and deletion mutation analysis allowed Jain and colleagues to map the basic transactivation function of both proteins to their carboxy-terminal regions.⁶⁹ For AhR, this region contained three modular domains associated with transcription, namely, an acidic region, a glutamine-rich region, and a region rich in proline/serine/threonine residues, all hallmarks of previously described transactivation domains (TADs). As would be expected, these domains seem to serve different functions. For instance, the glutamine-rich region of the human receptor is necessary for the recruitment of SRC-1.¹⁸⁹ If we take the examples of other modular transactivation domains such as those found in Sp1, Oct1, and myogenin,^190,191 it can be inferred that the complexity of the AhR TAD would allow for extreme flexibility in transactivation potential. This could include allowing for the formation of a functional transcription complex at elements varying in distance from the proximal promoter of a given gene and the ability to confer a degree of specificity with regard to activating transcription from the regulatory regions of different target genes. Similarly, the carboxy-terminal portion of ARNT harbored both glutamine- and proline/ serine/threonine-rich residues.^64,69 These observations were extended shortly thereafter by Sogawa and colleagues, who demonstrated the transactivation function of the three putative transactivation subdomains of AhR.¹⁹² Additionally, they suggested that the glutamine-rich region of ARNT had no transactivation function, but the 34 carboxy-terminal amino acids were essential for activity. Most importantly, the requirement of the carboxy-terminal transactivation function of ARNT, in the context of the AhR/ARNT heterodimer, was established with truncated mutants of ARNT using ARNT-negative mouse c4 Hepa1 cells.⁷⁵ The glutamine and acidic domains of other transcription factors, including Sp1 and MyoD,¹⁹³ have inherent transactivation functions that are capable of interacting directly with basal transcription factors.^194,195 Not surprisingly, it was later demonstrated that AhR could interact directly with TFIIB, TFIID/TBP, and TFIIF,^10,196 and that ARNT was capable of interacting with TFIIB and TFIIF. However, it has never been demonstrated that AhR and ARNT alone were sufficient to activate transcription in vitro in reconstituted systems with general transcription factors.

It is now well established that all transcription factors, irrespective of structure or function, appear to recruit protein complexes consisting of common pools of coactivator/corepressor proteins^197,198 often described as coregulators or master regulators of gene transcription. The intrinsic functions of transcription factors are therefore modulated by the recruitment of these ancillary factors. Furthermore, they appear to regulate a wide variety of functions including chromatin remodeling, stabilization of GTFs, RNA elongation and processing, translation, DNA break formation, and transcription termination.^199–202

Upon DNA binding, it is clear that the AhRC, like NRs, is capable of recruiting CBP/p300,^203–205 the p160/bHLH-PAS coactivators SRC-1,^189,206 NCoA2/GRIP1/TIF2, and p/CIP,²⁰⁶ as well as other transcriptional coactivators, including RIP140,^189,207 components of ATP-dependent chromatin remodeling complexes, such as Brahma-related gene 1 (BRG-1), components of the mediator complex,^208,209 and P-TEFb and RNA elongation factor.²¹⁰ These proteins are incorporated into multimeric complexes, which interact with and modulate the activity of the core transcriptional machinery, as well as modifying local chromatin structure.²¹¹ However, the identity of many factors, and the mechanisms in which they are recruited by the AhRC to its cognate response element, are largely unknown. A recent review by Hankinson provides a comprehensive overview of coactivator function and AhRC-dependent gene activation and we will not attempt to reiterate it here.²¹² However, there have been a few additions to the pantheon of putative coregulators for AhRC transcriptional activity. Table 2 provides a list of most, if not all, of the putative transcriptional coregulators known to interact with AhR and/or ARNT. It is important to note that most of these studies have been performed in the context of either synthetic DRE concatomers or the CYP1A1 promoter.

TABLE 2.

AhR- and ARNT-Interacting Proteins

Name	Function	Reference
CBP/p300	HAT	205
SRC-1/NCoA1		189, 206
NCoA-2/GRIP1/SRC-2		206
P/CIP,SRC-3/AIB1/ACTR		206
ERAP-140		213
Brg-1	Chromatin remodeling	209
Med220		208
CDK8		208
TRIP230	Coactivation/Unknown	214
CoCoA		215
GAC63		216
NcoA-4		217
BRCA1		218
Rb		219
Mybbp1a		220
PML		221
Nedd8		222
SMRT	Corepressor/coactivator	213, 223
RIP140		207
P-TEFb	Elongation factor	210
TFIIB	General transcription factor	196
TFIID/TBP		224
TAF4		224
TAF6		224
Sp1	Transcription factor	225
ERα/β		226–229
ERRα1		228
COUP-TF1		228
NF-κB		230
UBC9	Sumo E3 ligase	221
CUL4B		13

Because of its heterodimeric nature, the AhRC is capable of building multimeric complexes on multiple transcriptional activation platforms, namely, AhR and ARNT. Indeed, many coactivators are selectively recruited into this complex by ARNT, including CBP, BRCA1, and TRIP230.^205,214,218 The gene for thyroid hormone receptor (TR)/retinoblastoma protein (Rb) interacting protein 230 (TRIP230) was originally identified independently by two different labs based on its ability to interact with TR²³¹ and Rb,²³² respectively. In addition, it was demonstrated that Rb negatively regulated the TRIP230 coactivation function of TR-regulated transcription.²³² The manner in which Rb facilitates repression is less clear, although it may involve sequestration of TRIP230.²³² However, this does illustrate a salient point, which is that transcription is tightly regulated at many levels. It may be that Rb attenuates the coactivator function of TRIP230 in order to ensure that transcription of TRIP230-regulated genes proceeds at an appropriate rate and in response to appropriate temporal or contextual signals. As was the case with TR and Rb, TRIP230 was identified as an ARNT- interacting protein by the yeast two-hybrid assay.²¹⁴ Microinjection of Hepa-1 cells with affinity purified antibodies and transient transfection with siRNAs against TRIP230 demonstrated that this coregulator was essential for both TCDD and hypoxia responsive transcription. The dependency of TRIP230 on Rbs ability to modulate AhRC function has not yet been thoroughly investigated. However, a direct interaction between AhR and Rb has been demonstrated, ²¹⁹ and the ability of the AhRC to drive transcription in rat 5L hepatoma cells is somewhat dependent on Rb.²³³ For an overview of AhR/Rb function, particularly with regard to AhR cross talk with factors governing cell cycle, readers are referred to reviews by Puga and colleagues²³⁴ and Cornelius Elferink.²³⁵

Similarly, BRCA1 is recruited to the AhRC complex by virtue of its interaction with ARNT.²¹⁸ Thus, prompted by the observation that BRCA1 regulates numerous genes involved in xenobiotic stress responses,²³⁶ investigators in the laboratory of Insoo Bae carried out a series of insightful experiments to determine its effect on TCDD-responsive CYP1A1 and CYP1B1 transcription. Using conventional structure-activity approaches and siRNA, they demonstrated that BRCA1 is indispensable for AhRC transcriptional activity and that this coactivation function is mediated by ARNT.²¹⁸ Furthermore, this interaction is mediated through a domain of BRCA1 not associated with its own intrinsic transactivation function. Finally, the non-p160-related coactivator NcoA4/ARA70 has been demonstrated to enhance AhRC-driven transcription in a heterologous reporter system.²¹⁷ Interestingly, two splice variants of NcoA4 are expressed, with the shorter isoform being the dominant species in most cancer cell lines.^237,238 Furthermore, Kollara and colleagues demonstrate that the short isoform’s (NcoA4β) ability to coactivate AhRC-dependent transcription is significantly diminished.²¹⁷ The fact that BRCA-1, NcoA4, and Rb are deregulated, or functionally ablated in some fashion, in many cancers/cancer cell lines is intriguing.

As mentioned above, these early studies relied heavily on unchromatinized reporter plasmids as a readout of transcription. Therefore, it is unlikely that these systems would recognize interactions with proteins whose contribution to transactivation is based on the ability to modify chromatin structure or whose activity is dependent on a chromatin environment. The lone study regarding AhR on a chromatinized template focused solely on the acidic domain of AhR²³⁹ and did not address the requirement of cofactors in a chromatinized setting. Although all putative coactivators of AhR function enhance DRE-driven reporter constructs in mammalian systems, it is clear that not all coactivators mediate their effects purely in an AhR and ARNT-TAD dependent fashion. Studies to determine the interaction domain of the coiled-coiled coactivator, CoCoA, GAC63 and the p160 steroid receptor coactivator, SRC-1, revealed that both the b-HLH-PAS and Helix1 domain of AhR were essential for these interactions, respectively.^206,215,216 Furthermore, microinjection assays with antibodies to SRC-1 revealed that SRC-1 was absolutely essential for the transcriptional activity of the AhRC.²⁰⁶ Therefore, although CoCoA, GAC63, and SRC-1 are capable of forming complexes with other coactivators that are dependent on AhR’s TAD for protein-protein interactions (i.e., CBP, GRIP1, and p/CIP), it is apparent that transiently transfected heterologous reporter plasmids cannot adequately describe the contribution of all factors to a receptor’s transactivation potential.

The transcriptional activation domains of AhR and ARNT are structurally different from nuclear receptors and several other classes of transcription factors, with at least three distinct domains representing novel platforms on which transcriptional machinery can be assembled. In the case of nuclear receptors, separate amino and carboxy terminal TADs have been identified, termed AF-1 and AF-2.²⁴⁰ Furthermore, these domains are characterized by the presence of alpha-helical motifs. Presumably, these structural differences allow for the assembly of different complexes conferring different specificities, although one might note that these structurally divergent receptors share a common pool of coactivators. In fact, a coactivator that is recruited to the AhR and not to an NR has not been described. Therefore, it seems likely that coactivators can be recruited in different fashions to exploit different properties. The nature in which the coiled-coiled coactivator, CoCoA, is recruited to different classes of transcription factor may be a good illustration of this point. Nuclear receptor recruitment of CoCoA is mediated by the p160 coactivator, GRIP1, and this is by virtue of a direct interaction with the bHLH-PAS domain of GRIP1.²⁴¹ In addition, it appears that CoCoA is necessary for the synergy that exists between GRIP1, CARM1, and p300²⁴¹ in the context of nuclear receptor-mediated transcription. In direct contrast to this, the AhRC recruits CoCoA in a p160-independent process and the requirement for CoCoA and CARM1 function in this context is unclear.²¹⁵ In the case of β-catenin, CoCoA may serve as the scaffold on which GRIP1 is recruited to β-catenin–regulated genes.²⁴² Therefore, it is apparent that many of these master regulatory proteins are multifunctional and the diversity of transcription factor structure allows for the formation of transcriptional machinery utilizing these factors in multiple configurations. Thus, examination of the coactivator-AhR literature reveals that the AhRC has much in common with nuclear receptors.

XII. TRANSCRIPTION FACTORS AND OFF-TARGET EFFECTS

An intrinsic quality of many transcription factors’ function, whose importance is increasingly being recognized, is the ability to modulate transcription in a DNA-binding independent fashion. As with most transcription factors, nuclear receptors are thought to manifest their main biological functions by transducing the transcriptional information of their cognate response elements. This has been questioned over the past decade by several independent findings. Investigators in Gunther Schutz’s laboratory made the startling observation that DNA-binding/dimerization–deficient glucocorticoid receptor (GR) mutant mice were viable,²⁴³ whereas null mutations were lethal.²⁴⁴ These observations demonstrated unequivocally, for the first time, that the DNA-binding capability of a transcription factor is not essential for survival but that the non-DNA binding properties are.²⁴³ Furthermore, AP-1 regulation of interstitial collagenase was repressed by GR through a direct protein-protein interaction between GR and AP-1, a phenomenon termed “transrepression^245–248 GR-mediated transrepression of NF-κB signaling is also well documented.^249–251 Subsequently, other nuclear hormone receptors, including the thyroid hormone (TR) and estrogen receptor (ER), were shown to be able to repress AP-1 and NF-κB activity via direct protein-protein interactions.^252–254 The androgen receptor (AR), on the other hand, appears to transrepress Smad and Ets-mediated transactivation,^255,256 and to be transrepressed itself by AP-1.²⁵⁷ Therefore, it seems plausible that the intrinsic transrepression and coactivator functions of some transcription factors are as important as their respective DNA-binding functions. Furthermore, and certainly in the case of ER, the effects of endocrine disruptors and compounds that mimic the activities of endogenous hormones affect not only estrogen response element (ERE)-regulated genes, but also genes transrepressed/coactivated by ER. However, the relative importance of each phenomenon has yet to be established.

The nature of repression by tethering of ligand activated NRs is not well understood. The classic model of NR activation by ligand suggests that ligand facilitates an exchange of NCoR/SMRT/Sin3/HDAC corepressor complexes for coactivator complexes with ATP-dependent chromatin remodeling and histone acetyl-transferase activities.^197,198 It has been suggested that this happens only in the context of a receptor’s own cognate positive response element. Studies regarding the repression of thyrotropin b gene by TR suggest that this switch does not occur in the context of a negative TRE.²⁵⁸ Other studies have suggested that GR-mediated repression of the IL-8 gene occurs through a direct protein-protein interaction with the NF-κB heterodimeric transcription factor, and is independent of GR binding to its cognate DNA response element.²⁵⁹ They present evidence that GR recruits GRIP1 but a steric change in the tethered complex unmasks GRIP1’s intrinsic repressor function.²⁶⁰ This, in turn, may play a role in the GR-mediated inhibition of phosphorylation of the carboxy-terminal catalytic domain of RNA polymerase II.²⁶¹ Whether AhR exhibits similar properties is just beginning to be explored.

XIII. AHR AS A MEDIATOR OF ENDOCRINE DISRUPTION

Several indirect lines of evidence exist that would argue that AhR may mediate off-target or non-DNA binding dependent transcription. Arguably, the strongest support for this theory is provided by the ability of diverse classes of AhR ligands to perturb endocrine function. Even a cursory survey of the literature will make it readily apparent that different investigations and observations across species and tissues are often conflicting or paradoxical. This should serve to reinforce the reality that effects are contextual and highlight the need for investigations into the molecular nature of cross talk between AhR and other signaling pathways. Nevertheless, some of the effects mediated by AhR activators on nuclear receptor–regulated systems are well established.

TCDD and certain poly-chlorinated biphenyls are potent repressors of thyroid function. Rats administered single doses of TCDD were observed to have decreased thyroxine levels,²⁶² and this is apparently accompanied by a concomitant increase in levels of thyroid stimulating hormone (TSH).²⁶³ A study involving U.S. Air Force veterans who had been exposed to TCCD in Vietnam also revealed that they had abnormally high levels of TSH.²⁶⁴ TCDD and glucocorticoids at high doses are known teratogens, although they elicit their effects through different mechanisms and, in the case of cleft palate in mice, the defects are morphologically different (for review, see 265). Studies investigating the synergistic effects of glucocorticoids and TCDD in the formation of cleft palate in mice revealed that glucocorticoids upregulate AhR message and protein levels, while TCDD upregulates GR.²⁶⁶ This effect of TCDD has also been observed in the rat ovary.²⁶⁷

Peroxisome proliferator–activated receptors (PPARs) also are affected by dioxin and dioxin-like chemicals in an AhR-dependent fashion. Dioxin downregulates the expression of PPARγ in 3T3 cells during adipogenesis.²⁶⁸ Subsequently, researchers in the laboratory of Jefcoate, using wild-type and immortalized AhR^−/− mouse embryonic fibroblasts, determined that this is likely due to AhR-dependent repression of adipogenesis.²⁶⁹ Concentrations of the PPARα activator WY-14643 that do not increase AhR mRNA levels potentiate CYP1A1 expression by 3-MC,²⁷⁰ while, paradoxically, clofibrate represses AhR-dependent CYP1A1 and CYP1A2 expression in rat liver.²⁷¹ Certainly, multiple diverse mechanisms play a significant role in the effects described above, including alterations in ligand metabolizing or synthesizing enzymes, direct activation or repression of target genes, and alterations in mRNA stability and protein turnover. The remainder of this review will focus on attempts to describe the molecular events that occur in response to activation of AhR and how direct protein-protein interactions affect the activity of the AhRC and other transcription factors.

XIV. AHR AND ER

By far the most extensive studies involving cross talk between AhR and another transcription factor are those involving the estrogen receptor alpha (ERα). The antiestrogenic properties of TCDD have been documented repeatedly over the last 20 years, beginning with the observations that TCDD repressed estradiol function in rat uterus and liver.^272–276 Subsequently, HAHs and PAHs were implicated in the repression of ER-driven transcriptional activity. However, part of this phenomenon may be due to structural similarities that exist between PAHs and estrogenic compounds, and some overlap in the spectra of their respective activities may exist. Diethylstilbestrol (DES), one of the earliest clinically useful synthetic estrogens was originally synthesized from a polycyclic aromatic hydrocarbon²⁷⁷ and is linked to an increased incidence of vaginal and cervical adenocarcinoma.^278,279

The means by which AhR activation inhibits ER-regulated gene expression pathways have been the subject of much study in the past and are relatively well understood. TCDD inhibits the expression of several E2-inducible genes, including pS2,¹⁶ cathepsin D,^280–282 c-fos,²⁸³ and cyclin D1,²⁸⁴ among others. The first explanation proposed for these phenomena was that it was due to increased metabolism of E2 mediated by the TCDD-induced expression of CYP1A1 and 1B1.^285–287 Subsequently, evidence to support other theories has emerged. For example, TCDD-mediated inhibition of E2-induced pS2 expression in BG-1 cells was blocked by the presence of the protein translation inhibitor cycloheximide.²⁸⁸ This suggests that TCDD induces the expression of another inhibitory factor of E2-mediated gene induction, such as activation of proteasomes leading to increased ER degradation.¹⁵ A direct transcriptional mechanism for TCDD-induced repression of E2 signaling was proposed with the identification of an inhibitory DRE in the upstream region of the cathepsin D (catD) gene.²⁸⁹ More recently, a direct interaction between AhR and ER, or transrepression, was proposed.²²⁹ However, this was confounded by the observation that 3-methylcholanthrene at relatively high concentrations (10 µM) could induce both ERE-driven luciferase activity and BrdU incorporation in the glandular epithelium of mouse uterus.²²⁹ This observation was supported by other studies that identified several AhR agonists, including 3MC and B[a]P, as ligand activators of ERα.²⁹⁰ Finally, the most simplistic explanation for TCDD-mediated inhibition of E2 activity is the competition for a common pool of coactivators, although this hypothesis has not been pursued rigorously.

Pre-natal, peri-natal, and long-term exposure to estrogens, phyto-estrogens, and endocrine disruptors such as TCDD, put individuals at an increased risk of multiple forms of cancer and developmental defects of the urogenital tract.^291–294 Many testicular cancer etiologies implicate the activation of the AhR^294–296 or conversely, the disruption of the estrogen signaling pathways by these compounds and other endocrine disruptors such as diethylstilbestrol. Subsequent studies continue to reiterate these findings.^297,298 Furthermore, clinical applications of ER agonist and antagonist therapies are associated with many ER-mediated side effects, including breast and endometrial cancer, thrombosis, sexual dysfunction, CNS effects, and estrogen-independent tumors.²⁹⁹ As a result, cross talk between AhR and ER pathways have been implicated, but studies investigating the molecular mechanisms underlying AhR-ER cross talk are scarce. GST-pull-down and in vivo coimmunoprecipitation studies have demonstrated a direct interaction between AhR and ER,^226,228,229 as well as ARNT and ER.²²⁷ Furthermore, dominant negative studies have shown that the AhR TAD and the N-terminal region of ER can reciprocally repress transcription from ERE- and DRE-driven luciferase constructs, respectively.³⁰⁰ TCDD exposure causes a robust increase in CYP1A1 mRNA and DRE-driven reporter gene activity, which are almost completely abrogated by cotreatment with E2 in Hepa-1 and MCF-7 cells.³⁰¹ Results similar to these were obtained in the human endometrial carcinoma cell line, ECC-1.³⁰² A more recent study showed no effect of E2 on 3-methylcholanthrene-inducible AhR activity, but failed to investigate the effects of E2 on HAH-mediated AhRC activity.²²⁹ We have demonstrated that estradiol-activated ERα directly represses AhRC activity at the transcriptional level via tethering to the AhRC in MCF-7 cells.²²⁶ However, others have reported the E2-inducible enhancement of TCDD-dependent CYP1A1 expression in T47D human breast cancer and Huh7 liver cell lines, while having little or no effect in MCF-7 cells.³⁰³ This disparity could reflect differences in cell culture conditions or of relative abundance of essential cofactors in different cell lines. Most studies concerning endocrine disruption by AhR ligands have focused on the impact of AhR activation on ER target gene induction using transiently transfected cell lines, transcription reporter systems, and ex vivo approaches.^{227,228,283,288,300–306} However, a recent study using AhR “knockout” mice implicates AhR in the regulation of Cyp19 and ultimately in estrogenesis, suggesting that the disruptive effects of dioxin and other exogenous AhR ligands on endocrine physiology may be due to the untoward activation of AHR target genes, and may not be due to a direct effect on ER function.³⁰⁷ In addition, AhR knockout mice do not reproduce easily.³⁰⁸ Therefore, it seems imperative to elucidate the molecular mechanisms involved in AhR transcriptional processes and, in particular, ER-AhR cross talk. It is also worthy to note that most experimental efforts to characterize the biochemical pathways involved in ER-AhR cross talk have focused on the TCDD-inducible repression of ER target genes.

Other studies have placed ARNT^227–229 and AhR²²⁹ at the promoter elements of E2-responsive genes, again with disparate results. In one study ARNT was directed to the pS2 and c-fos promoters only in response to 3MC in MCF-7 cells,²²⁹ whereas in another study ARNT was implicated as an ER coactivator associating with the pS2 promoter in an E2-dependent fashion in T47D cells.²²⁷ These discrepancies may reflect differences in the cell lines used and the experimental techniques used to gauge each parameter. One possible explanation is that these studies relied heavily on synthetic response elements to drive heterologous reporters for transcriptional readout. Nevertheless, the in vitro and ex vivo observations of Ohtake and colleagues were, to a great extent, validated by their in vivo observations in wild-type and mutant mouse models.²²⁹ Their observation that 3-MC alone could activate ER target genes was confirmed by several other independent studies that established that 3-MC and other PAHs are weak agonists of ERα.^290,309,310 The Kato group later expanded this work with the report that AhR is part of a ligand activated E3 ubiquitin ligase complex.¹³ In turn, this cullin 4B-dependent process is responsible for the ultimate degradation of ERα.

Cotransfection of ERα with a CYP1A1 promoter–driven luciferase vector in Hepa1 and MCF-7 cells treated with TCDD, with or without E2, leads to an E2-mediated repression of TCDD-dependent luciferase activity.²²⁶ However, heterologous reporter systems are inadequate for studying the molecular mechanisms of receptor “cross talk” and transrepression. These systems cannot distinguish between direct transcriptional events at the target and secondary transcriptional events such as the transcriptional activation/repression of other regulatory genes. Therefore, in order to assess the direct transcriptional effect of ER activation on AhRC-dependent gene transcription, studies employed reverse-transcription/real-time PCR of the AhRC target genes CYP1A1 and CYP1B1 in the presence and absence of the protein translation inhibitor, cycloheximide. TCDD caused a significant increase in CYP1A1 gene transcription in the human breast cancer cell line, MCF-7, whereas 100 nM E2 did not.²²⁶ However, E2 (like TCDD alone) caused a significant increase in CYP1B1 gene transcription, consistent with reports that the 5′’ regulatory region of the CYP1B1 gene harbors a functional ERE.³¹¹ The addition of 100 nM estradiol significantly abrogated TCDD-mediated CYP1A1 gene induction in the presence or absence of cycloheximide.²²⁶ The observation that maximal CYP1A1 gene induction was repressed by E2 by approximately 50%, with or without cycloheximide, suggests that direct transcriptional repression of CYP1A1 is responsible for the decreased mRNA and protein levels observed, and is not due to another downstream posttranscriptional mechanism(s). Furthermore, E2 is capable of repressing TCDD-inducible CYP1A1 protein levels in a dose-dependent fashion (Beischlag and Perdew, unpublished data).²²⁶ Reciprocally, TCDD repressed E2-inducible pS2 and PR mRNA levels in MCF-7 cells (Beischlag et al., unpublished data), a phenomenon observed with other ER-regulated genes and reporter constructs.^302,306,310 These models are represented schematically in Figure 2.

FIGURE 2.

Proposed mechanisms of AhR/ARNT-mediated transcriptional regulation. (A) Ligand-activated AhR-ARNT heterodimers bind their cognate response element in the regulatory regions of their target genes, assemble coactivator machinery, and facilitate transcriptional activation. (B) In combination with other stimuli, transcription factor tethering (TF1) to AhR, including the combinatorial recruitment of other classes of transcription factors (TF2), or (C) AhR/ARNT tethering to transcription factors at their cognate response elements. (D) DNA bending allowing AhR/ARNT DRE-bound complexes to interact with other DNA-bound transcription factors.

Studies performed in vitro using glutathione-s-transferase pull-down assays were able to demonstrate an interaction of ERα with either AhR or ARNT.^226,228 Furthermore, one study demonstrated that the ERα interaction domain within AhR resides within the P/S/T region of AhR’s TAD.²²⁶ However, in this study, 100 nM E2 abrogated both wild-type AhR and AhRΔP/S/T TCDD-inducible reporter gene activity. Therefore, the ER-ARNT interaction may be more physiologically relevant to the transrepression phenomena than the ER-AhR interaction. In addition, transcriptional activity driven by a GAL4-ARNT chimera was repressed dramatically by overexpression of ERα in an E2-dependent fashion, further suggesting that ERα may transduce its repressor function via its interaction with ARNT. This functionality is further supported, if only tangentially, by the observation that ER may recruit ARNT for a coactivator function during E2-responsive transcription.²²⁷

Recently, two separate models involving ER and AhR function demonstrated the complexity of this interaction. Hockings and colleagues have identified a second level of complexity of AhR-ER interaction, which further expands our understanding of transcription factor function.³¹² Their model proposes that the unliganded AhRC would tether to ligand-activated ER, which in turn would be capable of being recruited to Fos/Jun heterodimers at AP1-responsive elements in a CBP/p300-dependent fashion. They also provide evidence that this scaffolding is XRE dependent; that is to say, DNA bending allows for the AhRC to bind its cognate response element and interact directly with ER tethered to AP1 (Fig. 2D). Studies using FRET and ChIP have demonstrated that ligand activated AhR disrupts the synergy between ERα and Sp1 at the CAD gene promoter.³¹³ Whether this is a promoter-specific phenomenon is unclear. We have observed Sp1 at the pS2 promoter in MCF-7 cells in response to E2, but this seems unaffected by cotreatment with TCDD (Beischlag et al., unpublished data). Whether either of these phenomena is mediated by the CUL4B or another E3 uquibitin ligase complex remains unknown, nor have other potential repressor mechanisms been explored. The interaction between the nuclear corepressor SMRT and AhR may offer some insights into potential mechanisms of AhR mediated cross talk with NRs. Widerak and colleagues have demonstrated that RXR-heterodimer–responsive genes (e.g., TR and RAR) are activated by TCDD by sequestration of SMRT.³¹⁴ However, nuclear receptors that form homodimers (i.e., ER and PR) are repressed by TCDD. The notion that AhR may recruit a repressor function to NR homodimers is as plausible as it is provocative. Clearly, additional studies are needed to determine the importance of AhR-ER tethering to the overall phenomenon of gene regulation.

XV. CROSS TALK BETWEEN AHR AND INFLAMMATORY SIGNALING PATHWAYS

Transcriptional control of inflammatory signaling is mediated by a variety of transcription factors, including NFκB, AP-1, STAT3, and C/EBP. Many genes that are activated during inflammation contain promoter regions with multiple regulatory elements that often are synergistically activated by a combination of regulatory proteins bound to their cognate response elements. Transcriptional regulation of these genes is further complicated by the combinatorial complexity of dimer formation by various members of the NFκB and AP-1 families of transcription factors, as well as by their posttranscriptional modification status. The NFκB family consists of p65 (RelA), c-Rel, Rel-B, p50, and p52.³¹⁵ Both p65 and p50 are maintained in the cytoplasm through the ability of IκB to sterically hinder importin binding to its nuclear localization sequence.³¹⁶ An increase in NFκB translocation into the nucleus occurs through an induction of toll-like receptor (TLR) signaling by bacterial products such as lipopolysacharide (LPS) or cytokine production. TLR activation leads to an increase in IκB kinase (IKK) activity that in turn causes phosphorylation of IκB on serines 32 and 36. These modifications mediate proteolytic turnover of IκB, release of p65/p50, and its subsequent translocation into the nucleus. The transcriptionally competent p65/p50 dimer binds to its cognate element, often displacing a p50 homodimer, which is in effect a switch from a repressive to a actively remodeled chromatin state on the proximal promoter of NFκB-regulated genes.³¹⁷

The AP-1 family of transcription factors are key mediators of oxidant stress and inflammatory signaling.³¹⁸ Members of this family include c-Fos, FosB, Fra1, c-Jun, JunB, JunD, and ATF2. These factors can form both homo- and heterodimers and bind to several related core DNA response elements. The C/EBP family of transcription factors also frequently participates in the regulation of genes targeted by inflammatory signaling. These basic-leucine zipper proteins bind to their cognate DNA response elements as homodimers. In particular, C/EBPβ and C/EBPδ are dramatically upregulated by IL-6 or TNF-α and appear to play an important role in sustaining the transcriptional response to inflammation.^319,320 An excellent example of combinatorial transcriptional regulation of inflammatory signaling can be observed on the enhancer/promoter of the IL-8 gene.³²¹ Within the proximal promoter, there are functional response elements for AP-1, C/EBPβ, and NFκB. These factors act together to synergistically activate IL-8 mRNA production. Another example of synergistic regulation of inflammation-mediated gene expression can be seen on the promoter of the COX2 gene. Both C/EBPβ and NFκB are required for high-level induction of COX2 mRNA expression.³²²

For several decades it has been shown that inflammation leads to repression of drug metabolism and, in particular, expression of a number of cytochrome P-450s is suppressed.³²³ The ability of IL-1β, IL-6, TNF-α, and LPS to suppress constitutive and AhR ligand-inducible CYP1A1 activity and protein levels in Hepa 1 cells has been firmly established.^324–326 Cytokines also reduce constitutive and AhR ligand-mediated CYP1A1 mRNA levels and enzymatic activity in human hepatocytes.^327,328 Reduction in CYP1A1 expression has also been observed in the brain after cytokine exposure.³²⁹ Whether these effects are due to direct transcriptional repression or changes in mRNA stability has been addressed in rat hepatocytes.³³⁰ Recombinant IL-1 suppressed the rate of transcription of both CYP1A1 and CYP1A2, as measured by nuclear run on assays. One possible mechanism for this repression was explored through studies examining the role of H₂O₂ and NF1 in CYP1A1 gene regulation.³³¹ Oxidation of NF1 decreases CYP1A1 activity, as determined using a CYP1A1 enhancer/promoter in reporter assays. Whether this mechanism occurs on short-term treatment with cytokines needs to be tested. Another possible mechanism would include the ability of cytokine-activated transcription factors to bind to the AhR/ARNT heterodimer at the CYP1A1 promoter. Support for this mechanism includes the ability of RelA to physically associate with the AhR in Hepa 1 cell lysates.²³⁰ ChIP assays have revealed that LPS and TNF-α treatment of Hepa 1 cells in the presence of TCDD leads to a decrease in histone H4 acetylation, yet does not alter the amount of AhR/ARNT bound to the promoter.³³² Expression of a mutant form of IκB, which acts as a potent repressor of NFκB, leads to rescue of TNF-α–induced repression of liganded AhR-mediated transcriptional activity. Taken together, these results would suggest that RelA/p50 tethers at the CYP1A1, leading to transcriptional repression. However, the presence of RelA at the CYP1A1 promoter has not yet been established. Thus, additional studies are needed to determine the exact transcriptional mechanism of cytokine-mediated repression of CYP1A1. Another question that also needs to be addressed is whether NFκB can repress other genes regulated by the AhR. Investigators are currently utilizing the combination of quantitative RT-PCR, siRNA, and ChIP assays to obtain additional mechanistic information to aid in our understanding of the transcriptional mechanisms involved.

Numerous reports have demonstrated that phorbol esters lead to repression of AhR-TCDD–mediated CYP1A1 expression. The phorbol ester 12-O-tetradecanoylphorbol acetate (PMA) has been utilized as a potent activator of protein kinase C (PKC), which ultimately leads to down-regulation of PKC. Studies both in cell culture and in mice have suggested that the repression of AhR ligand-induced CYP1A1 expression after PMA treatment is due to the loss of PKC activity.^333,334 Indeed, studies suggest that AhR transcriptional activity is repressed on inhibition of PKC activity as measured in cell-based DRE-driven luciferase reporter studies.³³⁵ However, there are also other possible mechanisms that may explain the ability of PMA to repress CYP1A1, and that is through its ability to induce an immediate early gene response. PMA is able to dramatically induce AP-1 activity both through the transcriptional upregulation of the Fos and Jun genes and direct phosphorylation of these transcription factors.³³⁶ In addition, oxidant stress levels are increased by PMA treatment, leading to NFκB activation.³³⁷ Thus, it is quite plausible that PMA-directed repression of CYP1A1 may be mediated through a mechanism similar to what has been observed with the cytokines discussed previously.

XVI. EVIDENCE FOR AHR-MEDIATED REPRESSION OF NFκB AND AP-1 TRANSCRIPTIONAL ACTIVITY

Sustained activation of the AhR by coplanar PCBs, TCDD, and PAH exposure has been linked to an enhancement of inflammatory signaling in numerous cell culture and in vivo models.^338–340 Much of this data is based on relatively high dose exposure to AhR ligands that would lead to highly elevated and prolonged cytochrome P4501A1 activity and other DRE-mediated gene expressions. Recently, a number of reports have challenged this view, and may actually indicate that conditions such as transient or constitutively low levels of AhR activity may actually inhibit cytokine production in certain cell types in the presence of an inflammatory signal. For example, the ability of AhR agonist to repress LPS-mediated IL-6 mRNA transcription has been demonstrated to occur in bone marrow stromal cells.³⁴¹ In the stromal cell system, a reduction in p65/p50 DNA-binding activity was observed on cotreatment with an AhR ligand and LPS, compared to LPS alone. Treatment of C6 glioma cells with dbcAMP/theophyline leads to enhanced IL-6 gene expression that appears to be mediated, at least in part, by STAT3 activation. Cotreatment with β-naphthoflavone, an AhR agonist, represses dbcAMP/theophyline-mediated activation of the IL-6 promoter.³⁴² However, this report did not firmly establish that the effect of β-naphthoflavone on IL-6 transcription was mediated by the AhR. Support for the ability of the AhR to inhibit NFκB activity can be found in cell-based reporter studies that demonstrate that activation of the AhR leads to repression of NFκB-mediated transcription.²³⁰ Furthermore, biochemistry studies have shown that the AhR can associate with Rel A in Hepa 1 and dendritic cells.^230,343 In the dendritic cell line DC2.4, the AhR inhibited Rel A/p50 nuclear translocation. However, it is important to note that neither study demonstrated that activation of the AhR repressed transcription of an NFκB target gene. Further support for a role of the AhR in repressing inflammatory/immune signaling can be seen in studies performed in AhR null mice. The lung of AhR-null mice exposed to cigarette smoke or endotoxin exhibit greater neutrophilia in vivo and enhanced TNF-α and IL-6 levels in bron-cholveolar lavage cells in vitro.³⁴⁴ Also, the absence of AhR expression can cause premature Rel B degradation, which may explain the heightened response to inflammatory stimuli observed in the lung. In another study, examination of in vitro differentiation of Th cells revealed heightened expression of IL-4 and IFN-γ.³⁴⁵ Thus, the expression of the AhR appears to be important in the regulation of immune signaling.

The possible interplay between the AhR and other transcription factors, such as AP-1 and C/EBP, has been explored to a limited extent. Dioxin has been shown to induce c-Jun, jun-B, jun-D, and c-Fos in both an AhR-dependent and AhR-independent manner in Hepa 1 cells maintained under basal cell culture conditions.³⁴⁶ However, studies in CH12LX B cells revealed that activation of the AhR resulted in a marked repression of c-jun mRNA expression and AP-1 DNA-binding activity.³⁴⁷ These results appear to indicate that the relationship between the AhR and AP-1 may be cell-type specific. There are no reports that reveal that the AhR can directly interact with AP-1 or its components. However, it is quite possible that in the context of a specific promoter they may then interact. For example, both AP-1 and the AhR have been shown to directly regulate CYP1A2 gene expression, although whether these factors interact at the promoter of the CYP1A2 gene has not been reported.^100,348 It has been established that the AhR can interact directly with the estrogen receptor.^226,228 This fact, coupled with the ability of ER to bind directly to AP-1, may indicate that the AhR could be present at occupied AP-1 response elements through its interaction with ER.³⁴⁹ In fact, this model has been proposed to occur at the BRCA-1 promoter in the presence of estrogen.³¹² Neither the AhR nor ARNT has been shown to directly interact with C/EBP. However, the presence and functional interplay between C/EBP and AhRC has been established at the promoter of glutathione S-transferase Ya (GST Ya) and CYP1A1 genes.^350,351 In the case of the GST Ya promoter, there are functionally over-lapping C/EBPα and AhR response elements, suggesting that C/EBPα and the AhR may physically interact. Interestingly, induction of prostaglandin endoperoxide H synthase-2 by TCDD in rat hepatocytes appears to be mediated by both the AhR and by the ability of TCDD to increase overall C/EBP levels.³⁵²

The ability of nuclear receptors and NFκB signaling to mutually repress transcription mediated by these factors has been firmly established ^353–355 To gain insight into how the AhR and NFκB may interact at gene promoters, it is useful to examine the nature of the cross talk between nuclear receptors and NFκB. Numerous nuclear receptors have been shown to interact with RelA/p50, including, GR, ER, PPARγ, AR, and PXR. Perhaps the most studied nuclear receptor in terms of repression of NFκB transcriptional activity is GR. Over the years a number of mechanisms have been proposed to explain GR-mediated repression of NFκB, and indeed probably multiple mechanisms mediate the overall repression. However, it has recently been recognized that transrepression or tethering of GR to NFκB at the promoter of its target genes may be the primary means of gene repression. The key question to address is whether and how the AhR interferes with transcription mediated by transcription factors activated by inflammatory signaling. Many genes regulated by NFκB are also regulated by CEBP/β, AP-1, and/or STAT3 on exposure to cytokines and other inflammatory signals. Several possible mechanisms that could lead to repression of inflammatory signaling are outlined in Figure 3. The ability of the liganded AhR to activate or repress genes involved in inflammatory signaling probably will be cell-type and promoter-context specific. Thus, the exact transcriptional mechanism(s) that mediate these effects warrants further investigation and may lead to new therapeutic approaches.

FIGURE 3.

Possible sites of AhR repression of inflammation-mediated transcription.

XVII. CONCLUSION

As with any field of research, certain discrepancies in the literature cannot be resolved immediately. Undoubtedly, this reflects the complexities of biological systems and the inadequacies of the experimental techniques that we continue to employ. Cloning of the human genome, the availability of new technologies that allow us to study protein function and gene expression in vivo, and examples of mechanistic phenomena in other biological systems have afforded us insights into AhR function that were previously unrecognized or underappreciated. Nonclassical mechanisms of gene expression modulation represents a relatively new avenue for transcription factor research in general, and may be the next appreciable step forward in our understanding of AhR function. Certainly, the field has suffered from a lack of attention given to the ability of the AhR to attenuate the activity of other transcription factors. Future challenges will include elucidating the mechanisms and associated enzymatic activities of cofactor recruitment and identification of target genes whose expression is modified in response to complex mixtures/combinations of xenobiotics and endogenous signals. Almost certainly, proteomics and DNA array technology will play a large role in these endeavors. The absolute number and the identity of genes whose expression is regulated by the AhR is only partially known and conventional DNA array technologies are likely not sensitive enough to detect these genes. However, the combination of sophisticated bioinformatics algorithms, along with chromatin I.P./DNA array strategies such as chromatin immunoprecipitation on tiled arrays (ChIPotle)³⁵⁶ and ChIP-DNA selection and ligation³⁵⁷ (DSL), provide researchers with powerful tools for identifying target genes whose expression is modified in response to transcription factor tethering.

The challenge that we continue to face is identifying the targets of AhR action that are vulnerable to regulatory perturbations and put an individual at risk to the development of cancer. Undoubtedly, attention must be given not just to genes whose expression is enhanced or repressed, but to other potential biomarkers as well, such as micro-RNAs or regions of chromatin susceptible to epigenetic modification in response to combinatorial stimuli. This would also include related factors such as imprinting that may further explain the maternal and generational effects observed in the progeny of exposed individuals. Certainly the identification of biomarkers could provide the impetus for the rational development of pharmaceuticals based on studies with selective aryl hydrocarbon modulators (SARMs), selective estrogen receptor modulators (SERMs), and other nonclassical ligands. Already, the benefits of some SERMs are being recognized for their potential to elicit nonclassical estrogenic effects without the often negative, direct target effects associated with ER target gene activation.^358–360 The potential benefits of AhR ligands that could be exploited for antiestrogenic properties, yet not produce the toxic side effects of PAHs and HAHs, would be enormous. The methyl-substituted diindolyl-methanes recently characterized in Stephen Safe’s laboratory are two such compounds.³⁶¹ In addition, the identification of an endogenous ligand(s) for AhR, if any, would prove a huge boon to the field and our understanding of AhR function.

It is likely that the AHR and ARNT Are integral components of endocrine function, cytokine signaling, and, undoubtedly, many other signal transduction pathways as well. therefore, in light of the evidence reviewed herein, it may be that AHR’S endogenous physiological function, at least in part, is mediated through a DNA-binding independent mechanism(s), such as tethering or some other function heretofore unrecognized. it is quite plausible that realization of the true impact of tethering may not be appreciated until the activation of the AhR is studied in combination with other stimuli. AhR target gene activation may be only one mode of function. in fact, activation by exogenous compounds may only serve to disrupt normal AhR function. as such, nonclassical receptor activities may be a key aspect of AhR’S true physiological role, as has been observed with GR.