Transcriptional activation function of the mouse Ah receptor nuclear translocator.

We cloned from mouse hepatoma cells a cDNA which encodes the Ah receptor nuclear translocator (Arnt). Sequence comparisons reveal 89% nucleotide and 92% amino acid identity between mouse and human Arnt. Transfection of the cDNA into Arnt-defective mouse hepatoma cells fully restores their responsiveness to 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD), indicating that the cDNA encodes a functional Arnt protein. Transfection of the cDNA into wild type mouse hepatoma cells increases the magnitude, but not the sensitivity, of the transcriptional response to TCDD. Analyses of mutants indicate that Arnt has a modular organization. The unit that mediates both heterodimerization with the liganded Ah receptor and DNA recognition is functionally distinct from the unit that mediates transcriptional activation. A 96-amino acid, C-terminal domain of Arnt, which includes a glutamine-rich region, confers transcriptional activation capability upon the protein.

W e cloned from mouse hepatoma cells a cDNA which encodes the Ah receptor nuclear translocator (Arnt). Sequence comparisons reveal 8% nucleotide and 92% amino acid identity between mouse and human Arnt. Transfection of the cDNA into Arnt-defective mouse hepatoma cells fully restores their responsiveness to 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD), indicating that the cDNA encodes a functional Arnt protein. Transfection of the cDNA into wild type mouse hepatoma cells increases the magnitude, but not the sensitivity, of the transcriptional response to TCDD. Analyses of mutants indicate that Arnt has a modular organization. The unit that mediates both heterodimerization with the liganded Ah receptor and DNA recognition is functionally distinct from the unit that mediates transcriptional activation. A 96-amino acid, C-terminal domain of Arnt, which includes a glutamine-rich region, confers transcriptional activation capability upon the protein.
The environmental contaminant 2,3,7,8-tetrachlorodibenzop-dioxin (TCDD,' dioxin) is the prototype for a class of potentially toxic halogenated aromatic hydrocarbons, which share a similar mechanism of action but differ in potency. In animals, TCDD produces a spectrum of adverse metabolic, reproductive, immunologic, and neoplastic effects (Poland and Knutson, 1982;Safe, 1986). TCDDs risk to human health is uncertain; its potential reproductive and carcinogenic effects are of particular concern (Bailar, 1991;Johnson, 1992;Peterson et al., 1993).
TCDD-responsive cells contain an intracellular protein, designated as the aromatic bydrocarbon (Ah) receptor, which binds the compound saturably and with high affinity (Poland et al., 1976). Biochemical and genetic studies implicate the Ah receptor (AhR) in TCDD's biological effects (Poland and Knutson, 1982;Okey et aZ., 1993;Swanson and Bradfield, 1993). Mechanistic studies, using the induction of cytochrome P4501A1 (CYPlAl) as a model response, indicate the liganded AhR activates gene transcription . Cloning of its cDNA reveals that the AhR appears to be a basic helix-loophelix (bHLH) type of transcription factor (Burbach et al., 1992;Ema et al.. 1992).
* This research was supported by Outstanding Investigator Grant CA 53887 from the National Cancer Institute. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTMIEMBL Data Bank with accession numbeds) U14333.
Genetic analyses of mouse hepatoma cells that respond poorly to TCDD implicate a second protein in dioxin action (Hankinson, 1983;Miller et al., 1983). This second protein has been termed the receptor nuclear franslocator (Arnt); cloning of human Arnt cDNA reveals that it contains a bHLH motif (Hoffman et al., 1991). Neither the liganded AhR nor Arnt binds to DNA by itself Whitelaw et al., 1993;Matsushita et al., 1993). I n vitro studies imply that the liganded AhR and Arnt heterodimerize via their bHLH domains to generate a DNA-binding species (Whitelaw et al., 1993). The AhWArnt heteromer activates transcription by binding to specific recognition sites designated as dioxin-responsive elements (DRE), which are located within an enhancer upstream of the CYPlAl gene (Denison et al., 1989;Fisher et al., 1990).
AhR and Arnt contain regions with sequence similarity to the Drosophila Per and Sim proteins; by analogy with findings for Per (Huang et al., 1993), these "PAS domains may contribute to protein-protein interactions between AhR and Amt. In addition, AhR and Arnt contain regions with multiple glutamine residues; by analogy with findings for the transcription factor S p l (Courey and Tjian, 19881, these glutamine-rich regions might represent transcription activation domains. To analyze the functional organization of Arnt in greater detail, we have isolated mouse Arnt cDNA, determined its nucleotide and deduced amino acid sequences, and verified by transfection that it encodes a functional Arnt protein. We demonstrate that Arnt has a modular organization; the 96 amino acids comprising the C terminus of Arnt confers transcription activation capability upon the protein and functions independently of the domains responsible for heterodimerization and DNA binding.  (Cleveland, OH). Reagents for chloramphenicol acetyltransferase and P-galactosidase assays, as well as in vitro transcriptiodtranslation kit, were from Promega (Madison, WI). The reagent for measuring protein concentration was from Bio-Rad. Cell culture materials were from Life Technologies Inc. 2,3,7,8-Tetrachlorodibenzo-p-dioxin was from the National Cancer Institute Chemical Carcinogen Reference Standard Repository. Polybrene (hexadimethrine bromide) was from Aldrich.

EXPERIMENTAL PROCEDURES
Cell Culture-Wild type (Hepa lclc71, Arnt-defective (BP'cl), and AhR-defective (TAOclBP'cl) mouse hepatoma cells were cultured in a-minimal essential medium containing 10% fetal calf serum, as described previously (Miller et al., 1983). M r n t n t :   ATG GCG GCG ACT ACA GCT AAC CCA GAA ATG AcA TcA GAT GTA CCA TCG  CTG GGT CCC ACC ATT GCT TCT GGA AAC  7 5 M r n t aa: Met A l a A l a T h r T h r A l a A s n P r o G l u Met T h r Ser A s p V a l P r o Ser L e u G l y P r o Thr I l e A l a Ser G l y A s n 2 5 h A r n t a a : .  GGT GGA GGA GCT GTT GTA CAG AGG GCT ATT AAG CGA CGG TCA GGG CTG GAT TTT GAT 150 P r o G l y P r o G l y I l e G l n G l y G l y G l y A l a V a l V a l G l n A r g A l a I l e L y s A r g A r g Ser G l y L e u A s p P h e A S P A s p G l u V a l G l u V a l A s n T h r L y s P h e L e u A r g Cys A s p A s p A s p G l n Met Cys A s n A s p L y s G l u A r g P h e A l a 7 5 Gln I l e Ser Arg His Ser A s n P r o A l a G l n Gly Ser Ala P r o T h r T r p T h r Ser S e r S e r Arg P r o Gly P h e A l a 6 3 5

2537
Construction of a cDNA Library-Poly(A)+ mRNA was prepared after lysis of wild type mouse hepatoma cells in guanidinium isothiocyanate.
A cDNA library was synthesized using a ZAP-cDNA Synthesis Kit and the ZAP Express Vector. The mRNA was annealed to an XhoI-oligo(dT) linker-primer and reverse transcribed with Maloney murine leukemia virus reverse transcriptase in the presence of 5-methyl dCTP. The RNA-DNA hybrid was digested with RNase H, and double-stranded cDNA was synthesized using DNA polymerase I, using recommended conditions (Stratagene). EcoRI adaptors were attached using T4 DNA ligase, and the cDNA was digested withXho1 and EcoRl, cloned into the EcoRI! B o 1 Site of ZAP Express Vector, and packaged using Glgapack I1 gold extract.
Screening of the &HA Library-A cDNA corresponding to a portion ofArnt was synthesized from wild type mouse hepatoma cell total RNA using an RT-PCR kit (Perkin-Elmer Cetus) and degenerate oligonucleotide primers based upon the human Arnt sequence (HoEman et al.,   1991). containing 1 pg of total RNA and 2.5 p~ oligo(dT), primer, using conditions suggested by the manufacturer. PCR was performed in 100 pl containing 200 pmol of each primer, 20 nmol of dNTPs, and 2.5 units of Taq polymerase, under the following conditions: 94 "C, 2 min; 55 "C, 0.5 min; 72 "C, 1 min; 1 cycle and then 94 "C, 1 min; 55 "C, 0.5 min; 72 "C, 1 min plus 5 additional seconds in each subsequent cycle; 34 cycles.
Labeled hybridization probes were generated using [CZ-~~PI~ATP, random primers, and exonuclease-deficient Klenow DNA polymerase and were used to screen 1 x lo6 phage plaques, using recommended conditions (Strategene). Positive clones, contained within the pBK-CMVphagemid (a eukaryotic expression vector), were excised in v i m from the ZAP Express vector using the ExAssist-SOLR System, as recommended by the manufacturer. For sequencing, unidirectional nested deletions were generated using exonuclease 111 and mung bean nuclease (New England Biolabs, Bedford, MA) and were sequenced using the dideoxynucleotide method and Sequenase 2.0 (USB). used a PCR-based method (Higuchi et al., 1988). The same 5'-primer Generation of Deletion Mutants-% generate deletion mutants, we was used in each case, containing a BamHI restriction site and nucleotides 1-17 of mouse Arnt (underlined): 5'-CCCGGGGATCCTCaG-GCGGCGACTACAGC-3'. To generate the mutant designated "-Q" (see Fig. 4A), the 3"primer was 5'-AATCTAGACCGCGGCCGCGGCAGCAGTACCAGATGAGGC-3', which contains a XbaI restriction site and nucleotides 2020-2040 of mouse Arnt (underlined).
Construction of GALA-Amt Fusion cDNAs-A GAL4 DNA-binding domain (amino acids 1-147, see Carey et al., 1990) was generated by PCR, digested with HindIII and XbaI, and ligated between the HindIII and XbaI sites of pBK-CMV. PCR was used to generate fragments of Arnt, using the following primer sets. Full-length (FL): 5'-GTAGGATC-CAAGCn"rCTAGAATGGCGGCGACTACA-3', which contains an XbaI restriction site and nucleotides 1-15 of mouse Arnt (underlined); 5'-AT-ACCCGGGCTACTCGAGTTCGGAL~AAGGGGGGA~"~', which contains anApa1 restriction site and nucleotides 2311-2328 of mouse Arnt  Mouse Arnt cDNA (mArntlpBk-CMV, 5 pg) was transfected into wild type mouse hepatoma cells along with the reporter plasmid pMcat5.9 (2 pg) and the P-galactosidase expression vector pCHllO (1 pg). After 30 h, cells were induced with TCDD at the indicated concentration. After an additional 18 h, CAT-specific activity was measured in cell extracts (100 pl). The data points represent the mean of duplicate measurements within a single experiment; brackets indicate the range of values. Similar results were obtained in three additional experiments.
The Arnt segments generated by PCR were inserted into the pBK-CMV expression vector, immediately downstream of the GAL4 DNAbinding domain.
Dansfection and CAT Assay-Transfection was carried out using Polybrene, as described previously (Fisher et al., 1990). The cloned mouse Arnt cDNA or its mutants were cotransfected with a eukaryotic P-galactosidase expression vector, pCHllO (Promega), and a reporter gene, either pMCat5.9, which contains a CAT gene controlled by a TCDD-responsive enhancer , or pG5ETCAT, which contains a CAT gene downstream of five copies of the GAL4 DNAbinding site . The transfected cells were treated with 1 I" TCDD as indicated, for 18 h prior to harvest. 48 h after transfection, cells were harvested in Reporter Lysis Buffer (Promega). CAT activity was measured using a differential extractiodiquid scintillation assay, according to the manufacturer's suggested protocol (Promega). @-Galactosidase activity was measured to control for differences in transfection efficiency.
In Vitro DanscriptionlDanslation-Full-length and mutant mouse Arnt cDNAs in the pBK-CMV vector were used for expression from the T3 promoter. Full-length, functional mouse AhR cDNAwas cloned using PCR' and was inserted into the pRc-CMV vector (Invitrogen, San Diego, Q. Ma and J. P. Whitlock, Jr., unpublished observations. , CA) for expression from the T7 promoter; a synthetic Kozak consensus sequence (Kozak, 1987) was inserted immediately upstream of the initiation methionine codon of the AhR.
In uitro transcription and translation were performed at 30 "C for 90 min, using TN'P-coupled reticulocyte lysate (Promega), as recommended by the manufacturer. Expression of the cDNAs was verified by including [35Slmethionine in the incubations and analyzing the proteins by SDS-polyacrylamide gel electrophoresis and autoradiography. The expressed proteins were analyzed using an electrophoretic mobility shift assay, as described below.
Electrophoretic Mobility Shift Assay (EMSA)-The EMSA was performed as described previously, using as a DNA probe the recognition sequence for the AhWArnt heteromer designated DRE D (Lusaka et al., 1993). Preparation of hepatoma cell nuclear extract, 32P-labeling of DNA using T4 polynucleotide kinase, electrophoresis, and autoradiography were as described previously (Denison et al., 1989). The in vitro translated proteins were incubated with poly(d1-dC) for 15 min at room temperature (-23 "C), followed by the addition of radiolabeled probe, further incubation for 20 min, and electrophoresis.

RESULTS
Cloning of Mouse Arnt cDNA-We generated a cDNA fragment of mouse Arnt using RT-PCR and primers based upon the human cDNA sequence. The fragment was used to isolate a 2537-base pair cDNAfrom a mouse hepatoma cell library. Comparison of its deduced amino acid sequence with that of human Arnt reveals that the cDNA encodes the shorter form of Arnt, which lacks a 15 amino acid segment, compared to the longer form (Hoffman et al., 1991). Additional RT-PCR and sequencing experiments indicate that the 15 amino acid segment is identical in mouse and human (data not shown). The mouse Arnt cDNA open reading frame contains 2328 base pairs, which encode 776 amino acids and a protein of 85.4 kDa (Fig. IA). There is 89% nucleotide and 92% amino acid sequence identity be- two regions of homology with PAS, and a glutamine-rich region (Fig. lB). At positions 493 and 514, the mouse Arnt contains 2 additional amino acids that are not present in human Arnt. These differences do not appear to affect Arnt function as described below.
Functional Analyses of Arnt cDNA-We used transient transfection to analyze the function of the cloned cDNA. The plasmid pMcat5.9, which contains the chloramphenicol acetyltransferase gene under the control of a TCDD-responsive, AhRdependent, Arnt-dependent transcriptional enhancer, was cotransfected as a reporter system. Transfections into Arntdefective cells indicate that the cloned cDNA restores responsiveness to TCDD, complementing the defect in these cells; these findings indicate that the cDNA encodes a functional Arnt protein (Fig. 2). In contrast, transfection of the cDNA into Ah receptor-defective cells produces no change in their responsiveness to TCDD (data not shown). Thus, the complementation is specific for the Arnt-defective cells.
Transfection of the cDNA into wild type cells increases the maximal level of TCDD-inducible CAT expression 2-3-fold (Fig.  2). Thus, an increase in the intracellular Arnt concentration is associated with an increase in the extent to which a target gene responds to TCDD. This finding implies that the concentration ofArnt limits the magnitude of TCDD-induced gene expression in this cell system. This observation led us to ask whether the Arnt concentration also influences the sensitivity of a target gene to TCDD. Dose-response experiments (Fig. 3) indicate that increased Arnt expression in wild type cells does not alter the concentration of TCDD a t which induction of CAT activity is half-maximal. Thus, Arnt influences the magnitude of the induction response but not the sensitivity of the response to the inducer.
Dunscription Activation Function of Amt-Transcription factors are often modular and contain domains that are functionally distinct (Ptashne, 1988;Tjian and Maniatis, 1993). The bHLH domain of Arnt appears to mediate heterodimerization and DNA recognition (Whitelaw, 1993). Here, we have asked whether Arnt also has a transcriptional activation domain and, if so, whether it is functionally distinct from the domain that mediates heterodimerization and DNA binding.
To address this issue, we first constructed several mutants, which contained progressively larger deletions from the C terminus ofArnt (Fig. 4A). We tested the mutants for function by cotransfection into Arnt-defective cells, together with the CAT reporter plasmid. Our findings (Fig. 4 B ) reveal that deletion of 96 amino acids from the C terminus of Arnt (generating the mutant designated as -Q, because the deleted segment contains a glutamine-rich region) is associated with a 70-80% decrease in TCDD-inducible, Arnt-dependent CAT activity. This observation indicates that the deleted region contains a domain(s) that makes a substantial contribution to Arnt function and TCDD responsiveness. Deletion of an additional 210 amino acids from the C terminus (generating the mutant designated as -CT) produces no further loss of TCDD-inducible, Arntdependent CAT expression. Therefore, this region makes no obvious contribution to Arnt function. Deletion of still another 320 amino acids (generating the mutant designated as bHLH) results in the loss of essentially all Arnt-dependent CAT activity. This observation implies that Arnt contains a second functional domain, which is located within the deleted region. Notably, the deleted region contains the two PAS homologies.
We performed several additional analyses to characterize the deletion mutants in greater detail. First, we used an in vitro transcriptiodtranslation system, combined with an electrophoretic mobility shift assay, to determine whether the mutant Arnt proteins interact with AhR to form a complex with the correct DNA recognition characteristics. Control experiments (Fig. 5A) indicate that both Arnt and AhR are required to generate a TCDD-inducible protein.DNA complex at a dioxinresponsive element. Thus, this assay displays the Arnt dependence, AhR dependence, TCDD dependence, and DNA sequence dependence that is characteristic of the binding of the AhW Arnt heteromer to DNA.
Analyses of the full-length and mutant Arnt cDNAs reveal that removal of 96 amino acids from the C-terminal of Arnt does not impair the ability of the truncated protein to participate in the formation of a TCDD-inducible protein.DNA complex with the DRE (Fig. 5B). This observation implies that the mutant protein retains its capacity both to heterodimerize with the liganded AhR and to specifically recognize the DRE. Therefore, the failure of the mutant protein to fully complement the Arnt-defective cells (Fig. 4 B ) must reflect the loss of some other function. Deletion of an additional 210 amino acids generates a noticeably smaller protein.DNA complex in the EMSA, however, the deletion has no apparent effect on the ability of Arnt to heterodimerize or to bind DNA (Fig. 5B). This observation is consistent with the transfection studies, which revealed no further loss of Arnt function associated with this deletion (Fig.  4B). The deletion of still another 320 amino acids abolishes the ability of Arnt to form a TCDD-inducible proteinaDNA complex (Fig. 5B). This finding suggests that the deleted region is required for heterodimerization and/or DNA binding. Loss of either capability can account for the absence of function in the transfection experiments (Fig. 4B).
These findings led us to hypothesize that the 96 amino acid segment comprising the C terminus of Arnt contained a transcriptional activation domain. To test this idea directly, we analyzed several fragments of Arnt, using an approach designed to measure its transcriptional activation function independent of its DNA binding function. Our findings (Fig. 6) reveal that the full-length Arnt protein exhibits substantial activity as a transcriptional activator; the activity is independent of TCDD, implying that it is inherent to Arnt and does not involve the AhR. Deletion of the C-terminal 96 amino acids  " " " " " " " " " " " " " " " " " "

776
! Z E B " " " " " " " " " _" " from Arnt is associated with the loss of 90-95% of the transcriptional activation function (Fig. 6). Furthermore, the 96 amino acid fragment by itself (designated as Q ) exhibits activity similar to that of full-length Arnt. A larger C-terminal fragment of Arnt (designated as CT) has activity comparable to that of the 96 amino acid fragment. Again, these activities are independent of TCDD and, therefore, presumably do not involve the AhR. These findings indicate that Arnt has a transcriptional activation domain, which is located within 96 amino acids of its C terminus. Furthermore, the transcriptional activation domain is functionally distinct from the domain(s) that mediates heterodimerization and DNA binding. DISCUSSION We have cloned Arnt cDNA from mouse hepatoma cells. Sequence comparisons imply that the mouse and human Arnt proteins are very similar in structure. In particular, the bHLH and PAS domains ofArnt are essentially identical in mouse and human. Their conservation of primary structure among species implies that these domains are functionally important. This conclusion is consistent with the finding that deletion of the bHLH domain ofArnt is associated with loss of responsiveness to TCDD, probably because the mutant cannot heterodimerize with the liganded AhR (Whitelaw et al., 1993). Here, we find that a C-terminal deletion that removes the PAS domains of Arnt abolishes the TCDD-inducible signal in the EMSA. Thus, the deletion affects Arnt's heterodimerization andor DNA binding capability. Because PAS domains appear to participate in protein-protein interactions (Huang et aZ., 1993), our observa-

GAL GALIFL GAL/-Q GAUQ GALJCT
tion suggests that the PAS domains of Arnt are required for heterodimerization with AhR. However, this hypothesis requires additional study. Expression of Arnt cDNA in Arnt-defective cells restores their responsiveness to TCDD, indicating that the cDNA encodes a functional protein. Expression of Arnt cDNA in wild type cells increases the magnitude of the response to TCDD. This finding implies that the intracellular concentration of Arnt is a variable that can limit the extent to which a target gene responds to TCDD. Because Arnt functions in partnership with AhR, we infer that the concentration of Arnt in wild type cells does not vastly exceed that of AhR and that AhR-Arnt heterodimerization obeys the law of mass action. Our findings also imply that conditions which alter the effective Arnt concentration have the potential to affect the magnitude of the response to TCDD. For example, the level of Arnt gene expression or the presence of dominant inhibitory protein partners for Arnt would influence its effective concentration within the cell. Differences among tissues in such parameters may contribute to tissue-specific variation in the responses that TCDD elicits.
Our analyses imply that Arnt, like other transcription factors, has a modular structure and is organized into distinct units, which can function independently of each other. For example, our analyses of transcriptional activation indicate that a 96-amino-acid, C-terminal segment of Arnt functions as an independent unit and exhibits full activity even when separated from the heterodimerizatiodDNA-binding segmentb) of the protein. We speculate that additional proteins may use these same independent functional units in different combina-tions and in other transcriptional regulatory contexts.
The relationship between protein structure and transcriptional activation is poorly understood; several different protein motifs are capable of activating transcription (Hahn, 1993). The 96 amino acid C terminus of Arnt contains a glutaminerich region and, in this respect, resembles the transcriptional activation domain of Spl (Courey and Tjian, 1988). However, the contribution(s) that the glutamine residues make to transcriptional activation is unclear, and it is possible that other amino acids play more important roles in the process (Gill et al., 1994). Deletion of 96 amino acids from the C terminus of Arnt decreases its transcriptional activation capability by >go%; however, our transfection studies indicate that the same deletion decreases the TCDD responsiveness of the AhRf' Arnt system by only 7040%. Therefore, we hypothesize that the AhR contains a domaids) that can partially compensate for the effect of the Arnt deletion. Transcriptional activation presumably reflects a protein-prok i n interaction between Arnt and another component of the transcriptional apparatus. However, the protein(s) that Arnt contacts is unknown. By analogy with findings for Spl (Emili et al., 19941, Arnt may touch the TATA-binding protein directly and stabilize its binding to the CYPlAl promoter. A second possibility (again, by analogy with findings for Spl (Pugh and Tjian, 1990)) is that Arnt contacts a coactivator protein that functions as an intermediary between Arnt and TATA-binding protein. With respect to the latter possibility, we note that DNA.protein cross-linking studies reveal a 110-kDa protein associated with the DNA-bound AhlUArnt heteromer (Elferink et al., 1990); furthermore, a 110-kDa protein copurifies with AhR and Arnt during DNA recognition site chromatography (Elferink and Whitlock, 1994). We hypothesize that this 110-kDa protein is a coactivator, which Arnt contacts in the process of activating transcription.
The dioxin-responsive enhancer upstream of the CYPlAl gene contains multiple binding sites for the AhlUArnt heteromer. Deletion analyses of the enhancer indicate that the multiplicity of binding sites produces a synergistic effect on TCDDinducible gene expression (Fisher et al., 1990). We hypothesize that the transcriptional activation function of Arnt accounts for this synergy, in that the binding of multiple AhWArnt heteromers to the enhancer could permit several Arnt molecules to simultaneously contact the transcriptional machinery at the CYF'lAl promoter and to stabilize it in an active configuration (for discussion, see Lin et al., 1990;Carey et al., 1990). An analogous mechanism may account for the synergy between AhlUArnt and Spl (Fisher et al., 1990). These hypotheses can be tested once the additional protein(s1 that Arnt contacts are known.