A Mutation in the Aryl Hydrocarbon Receptor (AHR) in a Cultured Mammalian Cell Line Identifies a Novel Region of AHR That Affects DNA Binding*

Introduction of a retroviral expression vector for the aryl hydrocarbon receptor (AHR) restores CYP1A1 inducibility to a mutant derivative of the Hepa-1 cell line that is defective in induction of CYP1A1 by ligands for the receptor. An AHR protein with normal ligand binding activity is expressed in the mutant but ligand treatment of mutant cell extract fails to induce binding of the AHR·ARNT (aryl hydrocarbon receptor nuclear translocator) dimer to the xenobiotic responsive element (XRE). AHR cDNAs derived from the mutant encode a protein that is unimpaired in ligand-dependent dimerization with ARNT, but the AHR·ARNT dimer so formed is severely impaired in XRE binding activity. The mutant cDNAs contain a C to G mutation at base 648, causing a cysteine to tryptophan alteration at amino acid 216, located between the PER-ARNT-SIM homology region (PAS) A and PAS B repeats. Introduction of the same mutation in the wild-type AHR sequence by site-directed mutagenesis similarity impaired XRE binding activity. Substitution with the conservative amino acid, serine, had no effect on XRE binding. The tryptophan mutation, but not the wild-type allele, was detectable in genomic DNA of the mutant. The implication that an amino acid within the PAS region may be involved in DNA binding indicates that the DNA binding behavior of AHR may be more anomalous than previously suspected.

Introduction of a retroviral expression vector for the aryl hydrocarbon receptor (AHR) restores CYP1A1 inducibility to a mutant derivative of the Hepa-1 cell line that is defective in induction of CYP1A1 by ligands for the receptor. An AHR protein with normal ligand binding activity is expressed in the mutant but ligand treatment of mutant cell extract fails to induce binding of the AHR⅐ARNT (aryl hydrocarbon receptor nuclear translocator) dimer to the xenobiotic responsive element (XRE). AHR cDNAs derived from the mutant encode a protein that is unimpaired in ligand-dependent dimerization with ARNT, but the AHR⅐ARNT dimer so formed is severely impaired in XRE binding activity. The mutant cDNAs contain a C to G mutation at base 648, causing a cysteine to tryptophan alteration at amino acid 216, located between the PER-ARNT-SIM homology region (PAS) A and PAS B repeats. Introduction of the same mutation in the wild-type AHR sequence by sitedirected mutagenesis similarity impaired XRE binding activity. Substitution with the conservative amino acid, serine, had no effect on XRE binding. The tryptophan mutation, but not the wild-type allele, was detectable in genomic DNA of the mutant. The implication that an amino acid within the PAS region may be involved in DNA binding indicates that the DNA binding behavior of AHR may be more anomalous than previously suspected.
The AHR 1 mediates carcinogenesis by certain environmental pollutants, including the halogenated aromatic hydrocarbon TCDD, and polycyclic aromatic hydrocarbons (PAH) (reviewed in Ref. 1). Unliganded AHR is located in the cytoplasm associated with two molecules of the 90-kDa heat shock protein (HSP90) and possibly another protein(s) (2). After binding ligand, AHR dissociates from this complex and dimerizes with the ARNT protein. In the nucleus, this dimer acts as a transcription factor. Most if not all of the pathological effects mediated by AHR appear to depend on modulation of transcription by the receptor. The mechanism of transcriptional regulation is best understood for the CYP1A1 gene. Activation of transcription occurs through interaction of the AHR⅐ARNT dimer with several copies of short sequences, termed xenobiotic responsive elements (XREs), located in the 5Ј-flanking region of the gene (reviewed in Ref. 3).
The AHR and ARNT proteins both contain basic helix-loophelix motifs toward their amino termini. More centrally located in both proteins is an approximately 300 amino acid PAS homology region, containing two approximately 50 amino acid degenerate direct repeats, PAS A and PAS B. Analysis of deletion mutants has identified putative functional domains of both proteins. Ligand binding occurs in a region encompassing the PAS B repeat of AHR (4,5). One molecule of HSP90 appears to bind within the PAS region of AHR. The other molecule of HSP90 appears to require interaction at two sites: one over the basic helix-loop-helix region and the other located within the PAS region (6,7). Dimerization of AHR and ARNT requires the HLH regions of both proteins. The PAS regions of both proteins facilitate dimerization. The basic submotifs of AHR and ARNT are required for DNA binding but not for dimerization (7,8). Site-directed mutagenesis of individual amino acids confirmed this last observation and also indicated that an additional block of basic amino acids of AHR located amino-terminal to the above basic submotif also appears to contact DNA (9,10). The carboxyl-terminal half of AHR appears to contain several discrete transcriptional transactivation domains, while a single such domain occurs toward the extreme carboxyl-terminal region of ARNT (11)(12)(13)(14)(15).
The AHH activity of CYP1A1 metabolizes PAH, such as benzo(a)pyrene, to cytotoxic as well as carcinogenic products. The mouse hepatoma cell line, Hepa-1, is highly inducible for CYP1A1 by PAHs and TCDD. We isolated mutants of Hepa-1 cells that are resistant to the toxicity of benzo(a)pyrene. These arose spontaneously at the relatively low rate of 2 ϫ 10 Ϫ7 events per cell generation, and their frequency was increased markedly by mutagenesis. All of the clones had much reduced or undetectable CYP1A1-dependent AHH activities after treatment with concentrations of PAHs and HAHs that lead to maximal induction in wild-type Hepa-1 cells (16,17). Analysis of somatic cell hybrids between individual clones and the Hepa-1 parental line demonstrated that most of the clones are recessive to the wild-type cells, whereas a few of the clones are dominant. Somatic cell hybridization experiments performed between the recessive clones permitted their assignment to four complementation groups (18,19). These mutants have been used extensively for analyzing the mechanism of CYP1A1 induction and the mechanism of action of AHR and ARNT (20).
One complementation group corresponds to the CYP1A1 gene (21,22). Clones assigned to the other three complementation groups are all affected in functioning of the AHR⅐ARNT dimer. Clones in complementation group B express markedly reduced levels of AHR mRNA and appear to be defective in a factor required for transcription of the AHR gene (23). Clones in complementation group C are defective in activity of the ARNT protein (24). One clone was isolated in complementation group D. Analysis of the D mutant demonstrated that it has moderately reduced levels of ligand binding to AHR, but appeared to be even more severely deficient in ligand-dependent translocation of AHR to the nucleus (19). The original D mutant analyzed in these previous studies exhibited an unstable phenotype. In the studies reported here we utilize a subclone of the original mutant that possesses a stable phenotype, maintaining undetectable TCDD-inducible AHH activity over many months in culture. We demonstrate here that the stable subclone carries a point mutation in the Ahr gene that does not affect ligand binding or dimerization with ARNT, but markedly reduces the efficiency with which the encoded AHR protein binds the XRE. This mutation changes a cysteine residue located in the PAS region to a tryptophan residue. This region was previously not thought to be involved in DNA binding, and our results point to the possibility that this region represents a novel DNA binding domain in AHR.

EXPERIMENTAL PROCEDURES
Cell Culture-The mouse hepatoma cell line, Hepa1c1c7 (Hepa-1), and a subclone, c35-3, of the D mutant strain, c35, which was derived from Hepa-1 cells (19), were maintained in nucleoside-free ␣-minimal essential medium (Irvine Scientific) supplemented with 10% fetal calf serum in a 5% CO 2 incubator at 37°C. Retroviral Expression Vector for Mouse AHR-pSR␣(NotI)AHR and the preparation of retroviruses from this vector, have been described (23). In brief, the mouse AHR cDNA was ligated into pSR␣(NotI) (a generous gift of Dr. O. N. Witte, Department of Microbiology and Molecular Genetics, UCLA, Los Angeles, CA) 2 , using the XbaI and HindIII sites, to generate pSRa(NotI)AHR. Infectious retroviruses derived from pSRa(NotI)AHR was prepared by the rapid procedure of Muller et al. (25). Hepa-1 and D cells in 100-mm culture dishes were treated with the viral suspension in growth medium supplemented with 8 g/ml Polybrene. Two or three days later, the infected cells were trypsinized and plated in growth medium supplemented with G418 (8). The benzo[ghi]perylene plus near-UV reverse selection was carried out as described without prior treatment of cells with TCDD (26).
Protein Assays-Cell extracts were prepared as described (24). Protein concentrations were determined with the Bradford assay. Western blot analysis of cytosolic extracts of either Hepa-1 or D cells was performed as described previously using affinity-purified polyclonal antibodies to AHR (24). Immune complexes were detected using the enhanced chemiluminescence (ECL) detection system (Pierce) with a secondary antibody coupled to horseradish peroxidase. The AHH assay and the in vivo AHR ligand binding assay (using the photoaffinity ligand, 2-azido-3-[ 125 I]iodo-7,8-dibromodibenzo-p-dioxin) were performed as described (7).
RNA Analysis-Total RNA was generated from Hepa-1 and D cells using the method of Chomczynski and Sacchi (27). Messenger RNA was prepared directly from the cells using a FastTract mRNA isolation kit (Invitrogen). For quantification of steady-state messenger RNA levels by competitive RT-PCR, internal standards for mouse Ah receptor and ribosomal large protein 7 (ML7) (28) were generated as follows: modified (deleted) cDNA fragments for AHR and ML7 were derived from RT-PCR reactions with the primers designed to generate an 80-bp internal deletion located toward the 5Ј-end (see Table I) and then cloned into a TA cloning vector, PCRII (Invitrogen). Transcription into cRNA from the T7 promotor was performed with the MegaScript™ kit (Ambion). RT reactions were carried out using specific primers for AHR or ML7 from total cellular RNA in the presence of the corresponding cRNA internal standard in decreasing quantity. A fraction of each RT mixture was subjected to PCR reaction and the products analyzed by electrophoresis in 3% agarose gel containing ethidium bromide.
Northern blot analysis was performed as described previously (23). Briefly, cellular mRNA was fractionated on 1% formaldehyde-agarose gels and transferred to nylon membrane (Hybond-N, Amersham Corp.). The filters were probed with cDNA fragments labeled with [ 32 P]dCTP by random priming. Filters were then washed and subjected to autoradiography. Relative level of expression for each mRNA was then quantified with the aid of an radioanalytic imaging System (AMBIS Inc.), which is subsequently refered to as ␤-scanning.
DNA Analysis-Genomic DNA was isolated (29) from Hepa-1 and D cells. PCR reaction was carried out to amplify a 210-bp fragment of the Ahr gene using the indicated primers ("Upstream," see Table I) which extended from Ϫ170 bp through the first 40 bp of the coding sequence of the AHR cDNA. Another 1.2-kilobase pair genomic fragment of Ahr was also generated by PCR with the indicated primers ("Genomic," see Table I). PCR products were then cloned into the PCRII vector for NaeI restriction digestion or sequence analysis.
RT-PCR Cloning of the Ah Receptor cDNA-Total RNA from either Hepa-1 or D cells was used in RT-PCR reactions to amplify AHR cDNA 2 O. N. Witte, unpublished data.  TCCAC AATCC GCAGC ATG a Internal standard (I.S.) cDNA for AHR was generated by PCR with indicated forward and reverse primers from an oligo(dT)-primed RT mixture from Hepa-1 cells. The I.S. cDNA was then transcribed into cRNA from a T7 polymerase promotor included in the forward PCR primer. b I.S. cDNA for ML7 was generated by PCR with indicated forward and reverse primers from an oligo(dT)-primed RT mixture from Hepa-1 cells. The I.S. cDNA was then cloned into a TA cloning vector, pCRII, and transcribed into cRNA from the T7 polymerase promotor in the vector.

GAATT TCATC CTGGC ATGGG AG
spanning the coding region. (The "full-length forward primer" contained the first 20 nucleotides of the coding sequence, while the "full-length reverse primer" (see Table I) corresponded to a sequence in the 3Јuntranslated region (see Table I).) Superscript II reverse transcriptase (Life Technologies, Inc.) was used in the RT reactions. A mixture of the proofreading Pfu (Stratagene) and AmpliTaq (Perkin-Elmer) thermostable DNA polymerases was used in the PCR reactions, to ensure fidelity. AHR cDNAs from separate RT-PCR reactions were then cloned into the eukaryotic expression vector PCRIII using the TA cloning kit (Invitrogen). Plasmids containing the AHR cDNA insert were verified by restriction digestion.
Generation of AHR Mutants-Serine or tryptophan substitution at amino acid 216 of AHR was generated by the PCR methodology of overlap-extension (30) using Pfu polymerase (Strategene). Briefly, two primary PCR reactions were carried out using pcDNA3-AHR (9) as the template. One reaction amplified between an internal 3Ј-primer (containing serine or tryptophan mutation codons) and an external 5Јprimer while the other amplified between an internal 5Ј-primer (containing serine or tryptophan mutation codons) and an external 3Јprimer. The reactions thus created two overlapping AHR PCR fragments with serine or with tryptophan substitutions. In a secondary PCR reaction, the two overlapping fragments for each mutation were used as templates to generate serine or tryptophan mutation products using the external 5Ј-and 3Ј-primers. The internal primers for the serine mutant were: Ser-5Ј, 5Ј-AGCCGGCTGAGGTGC-3Ј; Ser-3Ј, 5Ј-GCACCTCAGCCGGCTCCTGAAGCACCTCTC-3Ј; and Trp-3Ј: 5Ј-GC-ACCTCAGCCGCCACCTGAAGCACCTCTC-3Ј.
The external 5Ј-primer corresponds to AHR bases 190 -211 and the external 3Ј-primer corresponds to AHR bases 878 -894. The secondary PCR products were digested with Bpu1102I and Eco47III and then ligated to similarly digested pcDNA3-AHR. The generated clones were sequenced to confirm their mutations.
In Vitro Transcription and Translation-Constructs were expressed in the TNT T7 coupled reticulocyte lysate system (Promega Biotech) in the presence or absence of [ 35 S]methionine (Amersham Corp.). Expression of each construct was assayed by SDS-polyacrylamide gel electrophoresis analysis of an aliquot of the reaction mixture performed in the presence of [ 35 S]methionine. Quantification of each construct's level of expression was then performed by ␤-scanning.
Dimerization of AHR Proteins with ARNT Proteins-Dimerization of AHR proteins with ARNT proteins was performed as described (10). Briefly, ARNT protein was synthesized as described above in the presence of [ 35 S]methionine, whereas all the AHR clones were transcribed/ translated into protein in the absence of the isotope. AHR proteins were mixed with an equimolar amount of ARNT in the presence or absence of 10 nM TCDD. The protein mixture was then incubated with affinity purified anti-AHR polyclonal antibody. The resultant immune complexes were then precipitated with protein A-Sepharose CL-4B beads, washed, and analyzed by SDS-polyacrylamide gel electrophoresis. ␤-Scanning of the gel was used to estimate the relative ARNT heterodimerization activity of the AHR clone in question (expressed as a percentage of the dimerization activity of the wild-type AHR cDNA). Mean values were calculated from three independent experiments.
XRE Binding Assay-This assay was performed either with nuclear extracts prepared from cells that had been incubated with 10 nM TCDD for 90 min or with in vitro generated AHR and ARNT proteins. In the former case, nuclear extracts were prepared as described previously (31) The XRE binding assay was performed as described (8). Briefly, each in vitro synthesized, unlabeled AHR protein was mixed with unlabeled ARNT in an equimolar ratio and the mixture incubated in the presence of 10 nM TCDD for 1.5 h. A poly(dI-dC)-containing binding buffer was then added. A 32 P-labeled double-stranded synthetic oligonucleotide containing mouse XRE1 was added to the mixture for a further incubation of 20 min at room temperature. In Vivo Functionality-PCRIII derivatives containing AHR cDNA inserts or the parental plasmid vector PCRIII (0.2-0.5 g) were cotransfected along with the CAT reporter plasmid, pMC6.3k (1-2 g) (32) into 2 ϫ 10 5 D cells/35-mm tissue culture well, using the lipo-fectAMINE method according to the provider (Life Technologies, Inc.). 5 h after transfection, cells were refed media with or without 10 nM TCDD. They were harvested 24 h later. In an attempt to minimize potential inducers of CYP1A1, the culture medium was supplemented with fetal calf serum that had been treated with dextran-coated charcoal and stored in light-proof bottles. Cell lysates were produced by three cycles of freezing and thawing. Endogenous acetyltransferases were inactivated by incubation at 65°C for 10 min. CAT assays were performed as described (33). Results were expressed as a percentage of the CAT activity obtained with the wild-type AHR cDNA.
Sequence Analyses-AHR cDNA clones generated by RT-PCR from D cells, Hepa-1 cells, and L5 as well as AHR genomic DNA clones generated by PCR were sequenced on an ABI PRISM cycle sequencing system (Perkin-Elmer) for the entire coding region of AHR. Results of the sequence analyses were compared with the AHR cDNA information from the GenBank™ (34) and any ambiguity reanalyzed by sequencing the complementary strand as well as sister clones.
Fluorescence in Situ Hybridization (FISH)-The fluorescence in situ hybridization procedure was carried out according to the method of Trask (35) with some modifications. In brief, cytogenetic preparations were made from the cell lines by usual cytogenetic means to perform FISH analysis. An Insert from a YAC clone (Genome System, Inc.) known to be located at the centromere of mouse chromosome 12 was prepared as probe by nick translation with digoxigenin-dUTP. Hybridization was performed on slides containing metaphase chromosomes derived from either Hepa-1 or D cells in a solution including 50% formamide, 10% dextran sulfate, 2 ϫ SSC (0.15 M NaCl, 0.015 M sodium citrate), and sheared mouse genomic DNA. Specific hybridization signals were then detected by incubating the hybridized slides in fluorescein-conjugated anti-digoxigenin antibodies, followed by counterstaining with 4,6-diamidino-2-phenylindole.

AHR Functionality in the D Mutant
Cells-Both the D mutant strain and the Hepa-1 cells were analyzed for the steadystate expression level of AHR mRNA by RT-PCR. The expression of AHR mRNA was substantial (8 ϫ 10 6 molecules/g mRNA) in the D cells (Fig. 1A, panel i), although significantly reduced compared with that in the Hepa-1 cells (3.2 ϫ 10 7 molecules/g of mRNA). The amount of the mRNA for the constitutively expressed ribosomal protein L7 (ML7) was measured in parallel from the same RNA samples (Fig. 1A, panel ii) to provide a control for possible differences in sample preparation.
Expression of AHR protein was then analyzed in cytosolic extracts prepared from the D mutant and the Hepa-1 cells. The AHR protein level in the D mutant cells varied between onethird and two-thirds of that in Hepa-1 cells in different experiments and extracts. Results of one experiment are presented in Fig. 1B.
The capacity of the AHR in the cytosol of the D mutant strain to bind the photoaffinity ligand 2-azido-3-[ 125 I]iodo-7,8-dibromodibenzo-p-dixoin was also evaluated. In comparison to the Hepa-1 cells, the D mutant cells exhibited approximately onehalf as much ligand binding capacity (Fig. 1C). Thus, the levels of expression of AHR mRNA, AHR protein, and ligand binding activity are all reduced by about three-quarters to one-half in the D strain. The AHR protein expressed by the D mutant cells, therefore, appears to have normal ligand binding activity.
To evaluate the overall in vivo functionality of AHR expressed by the D mutant cells, nuclear extracts were prepared from either vehicle (Me 2 SO)-or 10 nM TCDD-treated Hepa-1 and D cells. The binding capacity of the nuclear extracts to a 32 P-labeled double-stranded synthetic oligonucleotide containing mouse XRE1 was analyzed by electrophoretic mobility shift assay. As shown in Fig. 1D, formation of the AHR⅐ARNT⅐XRE complex was observed in the Hepa-1 cells after TCDD treatment, whereas no similar XRE complex was detectable in D cells, with or without ligand activation. (The band indicated as the AHR⅐ARNT⅐XRE complex in Fig. 1D has been unambiguously demonstrated to correspond to this complex in previous publications from this laboratory (36).) Thus, the D cells have lost ligand-inducible XRE binding activity.
Rescue of the D Mutant Phenotype by Infection with a Retroviral Vector Expressing AHR-Construction of an ecotropic recombinant retroviral vector containing both the neo gene and the AHR cDNA has been described previously (23). Hepa-1 and D cell cultures were infected with either pSR␣(NotI)AHR or the parental vector lacking the AHR insert, pSR␣(NotI). Infected cells were selected in G418. G418-resistant clones were obtained at frequency of 10 Ϫ3 in each case. Clones from each culture were pooled and expanded for further analysis. Successful introduction of the AHR cDNA was demonstrated by Northern blot analysis ( Fig. 2A). A 5.4-kilobase pair band was detected in markedly higher quantity in both D and Hepa-1 cells infected with the retroviral vector pSR␣(NotI)AHR than in the corresponding cells infected with the parental retroviral vector pSR␣(NotI). As reported previously, the mRNA encoding AHR that is expressed by the pSR␣(NotI)AHR provirus is indistinguishable in size from the endogenous AHR mRNA (23). Equivalent gel loading of all the mRNA samples were verified by the presence of similar amounts of mRNA corresponding to the constitutively expressed CHOb gene in each sample.
Two weeks after infection, pooled G418-resistant clones were subjected to the benzo[ghi]perylene plus near-UV reverse selection, which selects for cells that possess inducible CYP1A1 activity (27). pSR␣(NotI)AHR-infected D cells survived the reverse selection at a frequency of 86%, while the pSR␣(NotI)-infected D cells survived the reverse selection at a frequency less than 0.01% (Table II and Fig. 2B). Restoration of CYP1A1 inducibility was further demonstrated when both pSR␣(NotI)and pSR␣(NotI)AHR-infected D cells were analyzed for expression of CYP1A1-dependent AHH activity with or without TCDD induction (Table II). While TCDD-inducible AHH activity in the D cells infected with the parental retroviral vector pSR␣(NotI) was 1% of that in Hepa-1 cells, TCDD-inducible AHH activity was restored to 46% of wild-type levels in D cells infected with pSR␣(NotI)AHR. These data provide direct evidence that the CYP1A1-noninducible phenotype of the D strain is principally, if not completely, ascribable to loss of AHR activity.
RT-PCR Cloning of the Full-length Coding Region of AHR cDNA-To test whether the AHR protein encoded by the AHR gene in the D cells is defective, RT-PCR reactions were performed to generate the AHR cDNA covering the entire protein coding region (the forward primer contains the first 20 nucleotides of the coding sequence) from the total RNA of D cells. A mixture containing two different thermostable DNA polymerases was used in the PCR reactions to maximize fidelity. (The proofreading polymerase Pfu failed to amplify the AHR cDNA on its own (data not shown), probably due to its low template possessiveness.) AHR cDNA products from different RT-PCR reactions were cloned separately into the eukaryotic expression vector, PCRIII. Thus, any sequence artifacts intro- duced by the RT-PCR reactions could be detected by comparing independently derived RT-PCR products. The same procedure was followed to generate the full-length coding region of the AHR cDNA from the Hepa-1 cells. The cDNA fragments of AHR derived from either Hepa-1 cells or the D mutant cells were of the same size (data not shown).
Functionality of AHR Protein Derived from the D Mutant Cells-The ability of the AHR protein encoded by the D mutant cDNA to dimerize with ARNT was analyzed. The AHR cDNA clones derived from both the Hepa-1 (U1-5 and U2-6) and the D cells (D1-7 and D3-3), each through an independent RT-PCR reaction, were transcribed and translated into full-length AHR proteins in a transcription-translation-coupled reticulocyte expression system. Equimolar amounts of in vitro synthesized AHR proteins (unlabeled) were mixed with equimolar amounts of [ 35 S]methionine-labeled ARNT proteins in the presence or absence of TCDD and incubated at room temperature for 2 h to allow for heterodimerization. The amount of ARNT protein dimerized with AHR was estimated by co-immunoprecipitation with an excess amount of affinity-purified anti-AHR polyclonal antibody. Without TCDD induction, a minimal amount of ARNT protein co-immunoprecipitated with all the AHR proteins analyzed. ARNT heterodimerization activities were greatly enhanced by TCDD treatment, and the extent of induced dimerization was similar among all the AHR cDNA clones. This experiment was repeated three times with the same outcome. A representative result is presented in Fig. 3A. Therefore, the AHR protein derived from the D cells possessed normal capacity for TCDD-induced heterodimerization with ARNT. The specificity of the co-immunoprecipitation assay was evaluated using pre-immune IgG serum instead of AHR-antibody, to measure any background precipitation of the ARNT protein due to nonspecific protein interactions. As shown in Fig. 3A, the nonspecific AHR-ARNT association/coprecipitation was negligible.
The binding capacity to a 32 P-labeled XRE sequence was then analyzed for all the AHR proteins derived either from Hepa-1 (U1-5 and U2-6, obtained from two independent RT-PCR reactions) or D cells (D1-7, D1-14, D2-3, D2-8, D3-3, and D3-9, obtained from three independent RT-PCR reactions). In vitro synthesized AHR and ARNT proteins (both unlabeled) were mixed in equimolar ratio and incubated in the presence or absence of TCDD. Formation of the AHR⅐ARNT⅐XRE complex was then detected by gel mobility shift assay. None of the AHR cDNA clones derived from the D cells possessed XRE binding activity comparable with the wild-type AHR, L5, while those derived from the Hepa-1 cells behaved similarly to L5 (Fig. 3B). The greatly attenuated XRE binding activity of AHR proteins encoded by the cDNAs derived from the D cells was determined to be in all cases less than 10% of the activity expressed by the cDNAs derived from the Hepa-1 cells. The low XRE binding activity of AHR derived from the D cells remained ligandinducible and could be competed by excess unlabeled XRE.
To analyze the overall biological functionality of the AHR protein derived from the D cells, AHR cDNAs derived from the D cells (D1-7 and D1-14) were cotransfected into the D cells together with a plasmid containing a CAT reporter gene driven by the upstream regulatory region of the rat CYP1A1 gene. A wild-type AHR cDNA as well as the empty parental vector pCRIII were also included as positive and negative controls repectively (Fig. 3C). There was no detectable constitutive or TCDD-inducible CAT activity in the D cells cotransfected with the CAT reporter construct plus the parental vector lacking the AHR insert. In D cells transfected with all the AHR cDNAs, TCDD-inducible CAT activity was detected. However, the levels of CAT activity were markedly lower in cells transfected with the AHR cDNAs derived from the D cells than that in cells transfected with a cDNA derived from wild-type Hepa-1 cells. The CAT activity was influenced by the amount of AHR cDNA transfected as well as its ratio to the reporter construct. Nevertheless, AHR cDNAs derived from the D cells generated markedly lower CAT activities than the cDNA derived from Hepa-1 under all conditions tested (data not shown).

FIG. 2. Rescue of the D mutant phenotype by infection with a retroviral vector expressing AHR.
A, Northern blot analysis showing restoration of AHR mRNA expression to D cells infected with pSRa(NotI)⅐AHR. Cells were subjected to the reverse selection two weeks after infection. All infectants were assayed 8 weeks after infection. Upper arrow, AHR mRNA; lower arrow, CHOb mRNA; left, size markers (kilobase pairs). mRNA (5 g) was probed with a mixture of mouse AHR cDNA and CHOb cDNA. B, resistance of pSRa(NotI)AHRinfected D cells to the reverse selection. D cells infected with pSRa(NotI) (left) or pSRa(NotI)AHR (right) were selected with G418. Two weeks after infection, surviving cells were subjected to the reverse selection at 5 ϫ 10 5 cells/dish. The dishes were stained after a further 9 days of incubation. Infection with the AHR-expressing vector allowed the D cells to survive this selection procedure, which selects for cells expressing CYP1A1 inducibility. Sequence Analysis-Cycle sequencing analysis was performed on two AHR cDNA clones derived from the D cells (D1-7 and D2-8) encompassing the entire AHR coding region (although the forward PCR primer used to generate D1-7 and D2-8 contained the first 20 nucleotides of the coding sequence). AHR cDNA clones derived either from the wild-type Hepa-1 cells (U1-6) or from the normal AHR clone (L5) were also analyzed in parallel. Sequencing results obtained were compared with the normal AHR cDNA sequence in GenBank™ (34). All the noted sequence variations were further confirmed or eliminated by analyzing the complementary strand or sister clones.
A sequence variation was located at bases 221 and 222, which read GC rather than CG. This would change the amino acid coding from threonine to serine at position 74. Since the above GC sequence was observed in all the AHR cDNA analyzed (derived from both Hepa-1 and the D mutant), we concluded that the sequence in GenBank™ is inaccurate at that location. Our observation confirms other published AHR sequences (37,38) at this location. A sequence variation existed in the AHR cDNA clones derived from the D cells but not in any of the normal clones (Fig. 4). This single C to G mutation at base 648 would cause an amino acid alteration from cysteine to tryptophan at position 216. To rule out the possibility of the alteration being a cloning artifact, 10 other D mutant AHR cDNA clones derived from three different RT-PCR reactions were analyzed. All contained the above C to G point mutation.
To verify the existence of the altered sequence in the genome of the D cells, we also cloned a short 1.2-kilobase pair fragment of genomic DNA corresponding to the segment between bases 516

FIG. 3. Functionality of AHR protein derived from the Hepa-1 and the D mutant cells.
A, ARNT heterodimerization activity of AHR cDNA clones. Equal amounts of AHR proteins synthesized from cDNA clones derived from the D mutant (D1-7 and D3-3) and Hepa-1 cells (U1-5 and U2-6), as well as L5 (the wild-type AHR cDNA) were used in the analysis described. p ϭ preimmune IgG; I ϭ affinity-purified AHR antibodies. B, XRE binding activity of AHR cDNA clones. Equal amounts of AHR protein synthesized from cDNA clones derived from D mutant (from three separate PCR reactions: D1, D2, and D3) and Hepa-1 cells (two separate RT-PCR reactions: U1 and U2), as well as L5 (the wild-type AHR cDNA) were analyzed for their XRE binding activity. C, CAT activity of D cells cotransfected with pMC6.3K and AHR cDNA. 2 ϫ 10 5 /dish D cells were cotransfected with 2 g of pMC6.3k plus 0.5 g of empty vector (pCRIII) or pCRIII-AHR derived from either the D mutant (D1-7 and D1-14) or Hepa-1 cells (U1-5). CAT activity was assayed from cell extracts prepared from the transfectants and is expressed as a percentage of the specific activity of TCDD-treated cells cotransfected with pMC6.3k plus U1-5. and 680 of the cDNA, which spans intron E and encompasses the site of the mutation. All 17 D mutant genomic clones so derived from three different PCR reactions contained the C to G point mutation when analyzed by either restriction digestion (the C to G mutation leads to the loss of a NaeI restriction site) or sequence analysis. None of the 4 corresponding cloned genomic Ahr fragments from Hepa-1 cells contained the mutated sequence. There were two other possible sequence alterations observed in certain D derived AHR clones, but since they were each found in only 1 out of seven sister clones analyzed (Fig. 4), they were considered to be cloning artifacts.
Since the forward primer which we used for generating the AHR cDNAs corresponded to the first 20 nucleotides of the coding region, potential mutations in this region would be overlooked. Therefore, 210-bp fragments extending from Ϫ170 bp through the first 40 bp of the coding sequence of the AHR cDNA were generated by PCR from the genomic DNA of both the D and the Hepa-1 cells, and these were cloned and sequenced. No mutations were observed in the first 20 bp of the coding sequence of the D-derived AHR gene.
Substitution at Position 216 of AHR by Site-directed Mutagenesis-To confirm the affect of the above tryptophan mutation on AHR, in vitro mutagenesis was performed to substitute Cys with Trp at position 216 in the wild-type AHR (pcDNA3-AHR). In addition we substituted the same position with serine. The two mutant forms of AHR cDNA were transcribed and translated into proteins in vitro as described above. As expected, Trp 216 -AHR had about 5-10% of DNA binding activity compared with the wild-type AHR (Cys 216 -AHR) when analyzed in the XRE gel shift experiment (Fig. 5). This result confirmed our observation that the same single mutation in the AHR derived from the D cells was responsible for its loss of DNA binding ability. Interestingly, serine substitution at position 216 did not affect the DNA binding activity of AHR.
Determination of Chromosome Aneuploidity by FISH-To determine if the observed decreased expression of AHR in the D mutant is due to a reduced copy number of the Ahr gene, FISH analysis was conducted. Metaphase chromosomes were prepared from both Hepa-1 and D cells, then hybridized with a probe specific for the centromeric region of mouse chromosome 12, where the Ahr gene is located (39). Analysis of 80 metaphase cells from each cell strain indicates that Hepa-1 cells contain three chromosomal sites that hybridize with the probe (Fig. 6A) while the D cells contain only two (Fig. 6B). DISCUSSION Our previous studies demonstrated that the non-inducibility phenotype of the mutant is expressed in a recessive fashion in somatic cell hybrids formed between D cells and Hepa-1 cells, indicating that the mutant does not express a novel activity that eliminates inducibility. These studies also found that binding to AHR is moderately reduced in the D mutant, but that nuclear translocation of AHR is more severely affected. We proposed that the strain is either mutated in AHR or in a protein required for functionality of AHR (19). Our previous results were obtained with the original D mutant, c35, and several subclones of the mutant that exhibit an unstable phenotype in that they slowly reacquired TCDD-inducible AHR activity as they were maintained in culture. The current results were obtained with a subclone (c35-3) of the original D mutant, that possesses a stable phenotype, maintaining low TCDD-inducible AHH activity (less than 1% of that of wild-type Hepa-1 cells) over many months in culture (Ref. 19 and data not shown). In the current studies we thereby avoided potential confounding effects of an unstable phenotype. In this paper we FIG. 5. DNA binding activity of AHR with in vitro generated Ser or Trp substitution at position 216. Equal amounts of AHR protein synthesized from mutated cDNA clones: pcDNA3-Ser 216 -AHR, pcDNA3-Trp 216 -AHR, as well as the wild-type pcDNA3-AHR (Cys 216 ,L5) were analyzed for their XRE binding activity as described. All lanes are from the same gel.
FIG. 4. Sequence analysis of AHR cDNA clones. The entire coding region of AHR cDNA generated from Hepa-1 cells and the D mutant or L5 was sequenced with an Applied Biosystems PRISM cycle sequencing system. As the upstream PCR primer for RT-PCR cloning included the first 20 bp of the normal coding sequence, a 210-bp genomic fragment spanning the above region was generated from both Hepa-1 and D cells. Sequence analysis did not show any additional mutations in the first 20 bp of the coding region of AHR from the D cells.
confirm that ligand binding is moderately reduced in the mutant. Furthermore, we show that the D mutant possesses levels of AHR mRNA and AHR protein equivalent to its ligand binding activity (in each case about one third to one half of the magnitude in Hepa-1 cells). Thus, the AHR protein that is expressed in the D mutant appears to have normal ligand binding activity, indicating that the defect in the mutant does not reside in a protein required for ligand binding (such as HSP90) or in a segment of AHR required for this function. Reversal of the mutant phenotype by a retroviral expression vector for AHR indicated that loss of CYP1A1 inducibility in the mutant is principally, if not totally, ascribable to its loss of AHR function.
XRE binding in nuclear extracts obtained from D mutant cells was found to be severely reduced. Three possible explanations for the reduced XRE binding in the D mutant were envisaged: viz. that the D strain is mutated in (i) a portion of AHR that affects DNA binding or (ii) in a protein required for a necessary step preceeding DNA binding or (iii) in a protein directly required for AHR to bind DNA. (This last protein could be a protein kinase, for example, since AHR appears to require phosphorylation to bind the XRE (but not to dimerize with ARNT) (40), or it could be a protein required to maintain AHR in the reduced state required for DNA binding (41).) To distinguish between these possibilities, we analyzed AHR cDNAs derived from the D mutant. The AHR proteins expressed from the D mutant-derived cDNAs exhibited normal dimerization activity toward ARNT. However, seven different cDNAs, derived from three independent RT-PCR reactions performed on the D mutant, exhibited markedly reduced XRE binding activities, demonstrating that the D mutant carries a mutation in the AHR gene, and in particular, one that affects XRE binding. Subsequent sequencing of the cDNAs identified this mutation as being a C to G transversion, leading to a Trp from Cys mutation located between the PAS A and PAS B domains. Generation of the Trp mutation by site-directed mutagenesis of the wild-type AHR confirmed that this mutation leads to a dramatic decrease in XRE binding activity. Interestingly, conservative substitution of cysteine with serine had no effect on XRE binding.
The D mutant cDNAs were also markedly deficient in their ability to transactivate transcription from an AHR-dependent reporter gene. This deficiency is probably secondary to the defect in DNA binding, since no evidence indicates that the PAS region possesses transcriptional transactivation activity (12,13). The fact that the activity of the D mutant cDNAs in the transactivation assay are less markedly reduced than their activities in the XRE binding assay may reflect differences in the ratios or absolute amounts of AHR, ARNT, and/or DNA in these two assays. We have observed a similar discrepancy with regard to the results obtained with these two assays for in vitro generated mutants of AHR that are affected in XRE binding (9). The nearly complete absence of TCDD-inducible AHH activity in D mutant cells (1% of that in Hepa-1 cells) is probably ascribable to a combination of both reduced AHR expression (one-quarter to one-half of wild-type) as well as reduced XRE binding activity of the expressed protein (less than 10% of wild-type).
Fluorescence in situ hybridization analysis demonstrated that the D mutant cells contain two chromosomal segments that hybridize to a probe that maps to the centromeric region of mouse chromosome 12, very close to where the Ahr gene is located (39), while Hepa-1 cells contain three copies. It is likely therefore that D and Hepa-1 cells contain two and three copies of the Ahr gene, respectively. However, only the mutant AHR sequence (Trp 216 ) was detected in genomic DNA from the D mutant, while only the wild-type sequence was detected in the genomic DNA of Hepa-1 cells. Furthermore, all 12 cDNAs derived from the D mutant contained the mutation, while neither of those derived from the Hepa-1 strain contained it. Thus the mutation must have arisen de novo in the D mutant. A possible scenario for the origin of the c35-3 mutant strain is that it arose from Hepa-1 cells by loss of two copies of the Ahr gene and mutation of one copy (not necessarily in that order), followed by duplication of the mutant allele. This suggestion is compatible with the observation that the strain expresses reduced levels of AHR compared with Hepa-1 cells. The proposed mode of origin of the D strain is consistent with the known mechanism of action of gamma irradiation, which was used to induce the D mutant (19). Gamma irradiation is known to cause G to C transversions as well as gross chromosomal deletions, including loss of whole chromosomes, in mammalian cells (42)(43)(44). We previously observed that the originally isolated D mutant strain, c35, and certain subclones derived from it exhibited an unstable phenotype, is that they slowly increased in TCDDinducible CYP1A1 activity as they were maintained in culture (19). It is possible that each of these strains consists of a heterogenous population of cells, some possessing and some lacking the wild-type Ahr allele, and that instability is due to an increase in the proportion of cells containing the wild-type allele. The clone used in the present paper, c35-3, exhibited a stable phenotype.
The DNA binding behavior of AHR is very different from that of other basic helix-loop-helix PAS proteins, including ARNT. DNA contacts within the basic domain of AHR are restricted to its carboxyl-terminal half. An amino-terminal basic segment of AHR separated from the above basic domain may also contact the XRE (although this latter point is controversial) (9,10,45,46). The mutated cysteine residue, Cys 216 , is embedded in an arginine and cysteine-rich region of AHR: 212 RCFRCRLRC 220 , that is conserved in rat and human AHR (except for Arg 215 , which is Ile in human AHR). This region is highly basic like the above two regions of AHR believed to contact DNA. It is thus possible that DNA binding by AHR is even more anomolous than hitherto suspected, with DNA contacts also occurring within the PAS region where Cys 216 is located. Several large deletions in the PAS region of AHR have previously been characterized by several investigators. Large deletions encompassing Cys 216 eliminate XRE binding (6,7,47), consistent with the notion that Cys 216 contacts DNA, although these deletion mutations may compromise other activities preceeding DNA binding (such as dimerization), rather than DNA binding per se. Another explanation for the phenotype of the D mutant is that Cys 216 does not contact DNA, but that mutation of Cys 216 to Trp 216 alters the tertiary structure of AHR in such a way as to eliminate DNA binding. If indeed the loss of DNA binding by the D mutant protein is due to an alteration in tertiary structure rather than a specific effect on DNA binding, then this alteration must be relatively minor, since the above mutation does not affect either ligand binding or dimerization with ARNT.
Dougherty and co-workers (41)demonstrated that binding of the AHR⅐ARNT heterodimer to the XRE in vitro could be reversibly activated or inhibited by agents that reduce or oxidize cysteine residues, respectively, and that dimerization of AHR with ARNT was not affected by these agents. A number of considerations suggested to us that the (or a) target for redox regulation of AHR⅐ARNT DNA binding might be Cys 216 of AHR: (i) no cysteine residues exist in the first 100 residues of AHR, that encompasses all previously identified domains involved in DNA binding, and no cysteine residues occur in the basic region of ARNT, which is its only known DNA binding region (8,10). (ii) Basic residues are located either side of Cys 216 . The presence of basic residues on both sides of a cysteine residue dramatically enhances its reactivity (48), potentially making it particularly susceptible to oxidation. (iii) Redox regulation of DNA binding by Jun and Fos is mediated by single cysteine residues in the DNA binding domains of these proteins. Like Cys 216 of AHR, these cysteine residues are each flanked on both sides by basic residues (49). To investigate this possibility we tested the effect of oxidation/reduction on the XRE binding activities of in vitro transcription/translationgenerated, dialyzed, and TCDD-treated AHR⅐ARNT dimers. Diamide treatment reduced XRE binding of the wild-type AHR⅐ARNT dimer by 2-fold, and this effect was reversed by dithiothreitol. 3 The Ser 216 mutant AHR⅐ARNT dimer responded to these agents in exactly the same manner (data not shown). Thus Cys 216 of AHR does not appear to mediate redox regulation of XRE binding.
We previously reported that the D mutant is severely affected in ligand-dependent nuclear translocation of AHR, as assessed by conventional subcellular fractionation analysis. One possible explanation for these results is that amino acids 212-220 represent part or all of a nuclear translocation signal (this region is rich in basic amino acids like other known nuclear translocation signals) and that the Trp 216 mutation negates activity of this signal. An alternative explanation is that, because of its defect in DNA binding, the AHR mutant binds with reduced avidity to DNA and is therefore more readily extracted from nuclei during the subcellular functionality procedure. Immunocytochemical analysis will be required to determine whether nuclear translocation of AHR is genuinely defective in the mutant.
We now possess mutants affected in the major proteins involved in induction of CYP1A1. Certain of our mutants (originally called A mutants) are mutated in the Cyp1a1 gene, others (originally called C mutants) are defective in ARNT function and are probably mutated in the Arnt gene, while we identify here a mutation (in the D mutant) in the Ahr gene. These mutants, which are all derived from the same cell line, together provide a powerful experimental system for investigating the role of this induction mechanism in any cellular process that is expressed in these cells. AHR knockout mice have previously been reported by two research groups (50,51). AHR null mutant cell lines that are potentially obtainable from these mice strains would constitute a complementary system for studying the role of AHR in cellular processes.
It should be noted that another class of mutants of Hepa-1 cells, the B mutants, have previously been used to investigate the role of AHR in particular cellular functions (52,53). However, we recently demonstrated that the B mutants are deficient in a factor required for expression of the Ahr gene, rather than being mutated in the Ahr gene itself, and may potentially be defective in expression of many other genes besides Ahr (23). Our demonstration that the D mutant is specifically mutated in the Ahr gene makes this mutant much preferable to the B mutants for investigating the role of AHR in cellular physiological processes. Our observation that the D strain is mutated in the Ahr gene also sheds light on certain previous observations. For example, we previously studied differentiated and dedifferentiated derivatives of the rat hepatoma line, H4IIE-C3, and found that dedifferentiated variants express CYP1A1 inducibility while differentiated variants generally do not. Treatment of the differentiated variants with 5-azacytidine or sodium butyrate, which are known to be able to reactivate expression of silenced genes, restored inducibility and AHR activity to these cells. Furthermore, the non-inducibility phenotype of the differentiated variants was complemented in somatic cell hybrids between these cells and the A, B, and C mutant classes of Hepa-1 cells, but was not complemented in somatic cells hybrids formed between these cells and the D mutant (54). We can now conclude that lack of CYP1A1 inducibility in the differentiated variants is due to silencing of the AHR gene in these cells.