Identification of a Cell-Specific Transcription Activation Domain within the Human Ah Receptor Nuclear Translocator

TOXICOLOGY AND APPLIED PHARMACOLOGY ARTICLE NO.

139, 272–280 (1996)

0166

Identification of a Cell-Specific Transcription Activation Domain within the Human Ah Receptor Nuclear Translocator J. CHRISTOPHER CORTON, EVELYN S. MORENO, SCOTT M. HOVIS, LINDA S. LEONARD, KEVIN W. GAIDO, MARIANNE M. JOYCE, AND SARAH B. KENNETT Chemical Industry Institute of Toxicology, 6 Davis Drive, P.O. Box 12137, Research Triangle Park, North Carolina 27709 Received June 30, 1995; accepted March 25, 1996

Identification of a Cell-Specific Transcription Activation Domain within the Human Ah Receptor Nuclear Translocator. CORTON, J. C., MORENO, E. S., HOVIS, S. M., LEONARD, L. S., GAIDO, K. W., JOYCE, M. M., AND KENNETT, S. B. (1996). Toxicol. Appl. Pharmacol. 139, 272–280. In the presence of 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) and related chemicals, the Ah receptor nuclear translocator (Arnt) forms a heterodimeric complex with the ligand-bound Ah receptor, leading to recognition of dioxin-responsive elements within the enhancer of the CYP1A1 gene and transcription activation by an unknown mechanism. To understand the role of Arnt in transcription activation by the Ah receptor–Arnt heterodimer, we performed a deletion analysis of Arnt to locate domains that are directly involved in transcription activation. We showed that the C-terminal 34 amino acids of Arnt encode a transcription activation domain (TAD) that functions independently of other sequences in the Ah receptor complex when attached to the heterologous Gal4 DNA binding domain. Deletion of the C-terminal acidic-rich 14 amino acids completely abolishes activity. Sequences important in Arnt TAD function were independent of the glutamine-rich region which is an important structural feature in the TAD of other transcription factors. The strength of the Arnt TAD when compared with the strong TAD from the herpes simplex virus VP16 protein was cell-type specific. Both the Arnt and VP16 TAD were equally strong in COS-1 cells, but the Arnt TAD had weak activity in an Arnt-deficient mouse hepatoma cell line and was not needed for restoration of CYP1A1 activation. These results imply that for CYP1A1 activation the Ah receptor provides the dominant activation function for the heterodimer in hepatoma cells. The potential of the Arnt TAD to contribute to activation by the Ah receptor complex is likely determined by availability or activity of cell-specific factors with which the TAD interacts. q 1996 Academic Press, Inc.

Exposure of experimental animals to the environmental contaminant 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD or dioxin) and related halogenated aromatic hydrocarbons (HAH) elicits a diverse set of tissue- and species-specific toxic responses, including teratogenesis, immunosuppression, epithelial metaplasia, thymic involution, enzyme induc-

272

0041-008X/96 $18.00 Copyright q 1996 by Academic Press, Inc. All rights of reproduction in any form reserved.

AID

TOX 7846

/

6h0f$$$$$1

tion, and tumor promotion (Vanden Heuval and Lucier, 1993). In humans, the most dramatic effect of exposure is a hyperkeratotic and metaplastic response of the hair follicles and interfollicular epidermis that leads to persistent acnelike lesions (Silbergeld, 1991). Despite the potent nature of TCDD as a tumor promoter in animals, the epidemiological evidence from studies of occupational or accidental exposure to TCDD is only suggestive of an increased cancer risk (Bertazzi et al., 1993; Fingerhut et al., 1991). Most, if not all, of the toxic and biological effects elicited by TCDD are mediated by the intracellular aryl hydrocarbon receptor (AHR), which binds to TCDD and related compounds with high affinity. The AHR undergoes a series of molecular alterations after binding ligand, collectively referred to as activation. This process includes the dissociation of one or more molecules of the molecular chaperone HSP90, alterations in phosphorylation content, translocation to the nucleus, heterodimerization with the Ah receptor nuclear translocator (Arnt), and interaction with specific DNA sequences called dioxin or xenobiotic response elements (DRE or XRE) (reviewed in Okey et al., 1994). Cloning of the AHR complex components has enabled a detailed analysis of the molecular mechanisms by which HAH binding to AHR leads to alterations in gene expression. The predicted amino acid sequences of AHR (Burbach et al., 1992) and Arnt (Hoffman et al., 1991) reveal that they belong to a large family of basic helix–loop–helix (bHLH) DNA binding factors (Murre et al., 1994). Similar to other members of this family, heterodimerization of AHR and Arnt is mediated by the HLH region (Reisz-Porszasz et al., 1994; Whitelaw et al., 1993a,b). In addition, AHR and Arnt encode an additional domain, the PAS domain which contributes to dimerization (Dolwick et al., 1993b; Reisz-Porszasz et al., 1994). The PAS domain is also found in two Drosophila proteins, the CNS midline developmental protein Sim (Nanbu et al., 1991) and the circadian oscillator protein Per (Huang et al., 1993). This unique juxtaposition of helix– loop–helix and PAS dimerization motifs most likely dictate specific heterodimer formation between AHR and Arnt from among the bHLH family of factors.

07-08-96 09:36:14

toxa

AP: Tox

Arnt ACTIVATION DOMAIN

In the absence of ligand, the AHR is found in the cytoplasmic compartment bound to two molecules of HSP90 (Chen and Perdew, 1994). A 190-amino-acid region of AHR that overlaps with part of the PAS domain has been identified as the minimal ligand binding domain (Whitelaw et al., 1993a), a region also encompassed by cross-linking and ligand binding studies (Nanbu et al., 1991; Poland et al., 1994). Interestingly, the ligand binding domain corresponds exactly to one of the domains mediating interaction with HSP90 (Antonsson et al., 1995; Whitelaw et al., 1993a), supporting the idea that HSP90 is required for ligand recognition by AHR. After release by HSP90, the nuclear translocation of AHR was originally thought to require the Arnt protein (Hoffman et al., 1991). However, immunofluorescence of intact cells using antibodies against AHR and Arnt demonstrates that nuclear translocation of AHR occurs in a cell line lacking functional Arnt (Pollenz et al., 1993). Once bound to DRE, the AHR–Arnt heterodimer complex increases transcription initiation of a number of genes by either direct interaction and recruitment of basal transcription factors to the promoter and/or by facilitating nucleosome displacement and thus accessibility of promoters to specific and general transcription factors (Wu and Whitlock, 1993; Robertson et al., 1994). Both AHR and Arnt were predicted to contain transcription activation domains (TAD) near their C-termini based on the presence of high concentrations of glutamine residues that are also found in the TAD of other transcription factors (Mitchell and Tjian, 1989). However, deletion of the C-terminal 318 amino acids of Arnt containing this Gln-rich region failed to have a significant effect on activation of a reporter gene under control of the CYP1A1 enhancer and promoter, leading to the conclusion that Arnt does not, in fact, encode an activation domain (Reisz-Porszasz et al., 1994). In contrast, a number of groups (Jain et al., 1994; Li et al., 1994; Whitelaw et al., 1994) have recently mapped an activation domain to the C-terminal Ç100 amino acids of Arnt as well as to the C-terminal Ç120–200 amino acids of AHR (Jain et al., 1994; Whitelaw et al., 1994). In reconstitution experiments using mutants of AHR and Arnt in which the activation domains of one or both had been deleted, Arnt provided the dominant activation domain in the AHR–Arnt heterodimer complex (Whitelaw et al., 1994). To further understand the contribution of Arnt in transcription activation by the AHR complex, we define in detail the sequences within Arnt required for transcription activation. Using chimeric proteins composed of the Gal4 DNA binding domain fused to Arnt sequences, we mapped the human Arnt TAD to the C-terminal 34 amino acids and show that the strength of this TAD is dependent upon the cellular context in which it is analyzed. MATERIALS AND METHODS Materials. Restriction endonucleases and DNA-modifying enzymes (T4 DNA ligase, Klenow fragment of DNA polymerase, and calf intestinal

AID

TOX 7846

/

6h0f$$$$$2

07-08-96 09:36:14

273

phosphatase) were purchased from Promega (Madison, WI). [14C]Chloramphenicol was from Amersham Corp. (Arlington Heights, IL). The sequenase 2.0 kit was from U.S. Biochemical Corp. (Cleveland, OH). Reagents for the chloramphenicol acetyltransferase assay were from Pharmacia (Piscataway, NJ). The reagents for the b-galactosidase assay were from Sigma (St. Louis, MO). The bicinchoninic acid (BCA) reagent for measuring protein concentration was from Pierce (Rockford, IL). Cell culture materials were from Gibco/BRL (Gaithersburg, MD). 2,3,7,8-Tetrachlorodibenzo-p-dioxin was from KOR Isotopes, Inc. (Cambridge, MA). Cell culture. COS-1 cells were cultured in Delbucco’s minimal essential medium containing 5% fetal calf serum, as described by America Type Culture Collection. Wild-type (Hepa1c1c7) and Arnt-defective mouse hepatoma cells (c4) were cultured in a-minimal essential medium containing 10% fetal calf serum (Hoffman et al., 1991). Construction of Arnt expression vectors. Fragments of the ARNT cDNA were cloned into the polylinker of pSG424 (Sadowski and Ptashne, 1989), placing these fragments in frame with the Gal4 DNA binding domain. Fragments of ARNT originated from the plasmid ARNT/pTZ18U (gift of O. Hankinson), which does not contain the 15-amino-acid exon in the amino terminus. Absence of this exon is inconsequential to Arnt function (Hoffman et al., 1991; Reisz-Porszasz et al., 1994). Plasmid Gal4DBD/Arnt1–774 was constructed by cloning the BamHI fragment containing the full-length Arnt into the BamHI site of pSG424. Gal4DBD/Arnt1–581 was constructed by cutting the plasmid Gal4DBD/Arnt1–774 with XmaIII and SalI, filling in the ends with Klenow fragment, and allowing self-ligation of the plasmid. Gal4DBD/Arnt1–498 and Gal4DBD/Arnt1–101/499–774 were constructed by partial digestion of Gal4DBD/Arnt1–774 with Kpnl and purification of the 4.8- and 5.0-kb fragments, allowing self-ligation of the plasmids. Gal4DBD/Arnt1–484 was constructed by cloning the BamHI–PvuII fragment from ARNT/pTZ18U into the BamHI and SalI (filled-in) sites of pSG424. Gal4DBD/Arnt1–331 was constructed by cutting Gal4DBD/ Arnt1–484 with SpeI and XbaI and allowing self-ligation of the plasmid. Gal4DBD/Arnt1–198 was constructed by cutting Gal4DBD/Arnt1–484 with XbaI followed by an ApaLI partial digest, purifying the 4.0-kb fragment, filling in the ends with Klenow fragment, and allowing self-ligation of the plasmid. Gal4DBD/Arnt1–101 was constructed by cloning the 0.26kb BamHI–KpnI fragment from ARNT/pTZ18U into the BamHI and KpnI sites of pSG424. Gal4DBD/Arnt499–581 was constructed by cloning the PvuII–NaeI fragment from ARNT/pTZ18U into the filled-in XbaI site of pSG424. The open reading frame was corrected by filling in the only EcoRI site. Gal4DBD/Arnt582–774 was constructed by isolating the NaeI–BamHI fragment from ARNT/pTZ18U, filling in the BamHI end with Klenow fragment, and cloning into the two SmaI sites of pSG424. Gal4DBD/ Arnt582–603/697–774 was constructed by performing a partial PstI digestion of Gal4DBD/Arnt582–774, purifying the 3.9-kb fragment, and allowing self-ligation of the plasmid. Gal4DBD/Arnt582–603/697–712 was constructed by cutting the plasmid Gal4DBD/Arnt582–603/697–774 with BalI and then allowing the plasmid to self-ligate. Arnt portions of the constructs Gal4DBD/Arnt699–774, Gal4DBD/ Arnt699–761, Gal4DBD/Arnt699–741, Gal4DBD/Arnt712–774,Gal4DBD/ Arnt712–761, Gal4DBD/Arnt712–741, and Gal4DBD/Arnt741–747 were amplified using the Expand High Fidelity PCR system (Boehringer-Mannheim, Indianapolis, IN). Reactions contained the full-length Arnt as a template and one of three oligonucleotides at the 5* end which incorporates a BamHI site immediately upstream of the first amino acid pictured in Fig. 2 and one of three oligonucleotides at the 3* end which incorporates a stop codon and an XbaI site immediately after the last amino acid. Amplified fragments were cut with BamHI and XbaI and cloned into pSG424. The 5* oligonucleotides were AR1 (5*-AGTTCGCGGATCCAATTCCAGACACGGACA-3*), AR2 (5*AGTTCGCGGATCCCACAGTGGCAGGGCCAG-3*), and AR3 (5*-AGTTCGCGGATCCCTGAGGTCTTCCAGGAG). The 3* oligonucleotides were AR4 (5*-ATCTAGTCTAGACTATTCTGAAAAGGGGGGAAA-3*), AR5 (5*-ATCTAGTCTAGACTAATTGTTGTAGCTGTTGCT-3*), and AR6 (5*ATCTAGTCTAGACTAAGGCTGGCCAGGTTGCTG-3*).

toxa

AP: Tox

274

CORTON ET AL.

Arnt1–484, Arnt1–331, and pSG424DGAL4 were constructed by cutting the plasmids Gal4DBD/Arnt1–484, Gal4DBD/Arnt1–331, and pSG424 respectively with BamHI and BglII to remove the Gal4DBD and allowing the plasmids to self-ligate. Arnt1–774 was made by cloning the BamHI fragment from ARNT/pTZ18U into the BglII and BamHI sites of pSG424. The orientation and correct open reading frames of all plasmids were confirmed by restriction enzyme analysis and sequencing of 5* and 3* ends of inserts. In addition the correct sequence of all fragments generated by PCR was confirmed by sequencing the entire fragment. A number of constructions generated non-Arnt, non-Gal4 sequences at their 3* ends. These constructions were Gal4DBD/Arnt1–498 and Gal4DBD/Arnt1–101, RAIRALDK; Gal4DBD/Arnt1–484 and Arnt1–484, STVPRGIRALDK; Gal4DBD/Arnt1–331 and Arnt1–331, R; and Gal4DBD/Arnt582–603/ 697–712, A. Transfection and chloramphenicol acetyltransferase (CAT) assay. Transfections were carried out using the calcium phosphate transfection kit according to manufacturer’s instructions (GIBCO/BRL, Gaithersburg, MD). Plasmids encoding the cloned human ARNT cDNA or its deletion mutants were cotransfected with a eukaryotic b-galactosidase expression vector, pCH110 (Promega), and a reporter plasmid, G5BCAT, which contains five binding sites for Gal4 upstream of the E1b TATA box (Gorman et al., 1982), or pRNH241c encoding 01140 to /59 of the enhancer and promoter region of the human CYP1A1 (gift of R. Hines). Transfections of COS-1 and c4 hepatoma cells contained 1 mg of expression plasmid encoding the Gal4–Arnt fusion protein, 1 mg of G5BCAT, and 3 mg of pCH110. Transfections of wild-type and c4 mutant hepatoma cells contained 2 mg of expression plasmid encoding the Arnt deletion mutants lacking Gal4 sequences, 3 mg of pRNH241c, and 3 mg of pCH110. Salmon sperm DNA was used as carrier to give a final concentration of 10 mg DNA per 100mm plate. The transfected cells were treated with 10 nM TCDD or DMSO vehicle alone or were not treated as indicated. The cells were harvested 24 hr after treatment, and extracts were used in CAT assays as previously outlined (Gorman et al., 1982). b-Galactosidase activity or protein concentration was used to normalize CAT activity with similar results. The standard error was never greater than 25%. Immunoblotting. COS-1 cells (in 60 1 10-mm dishes) were transiently transfected with the expression vectors (5 mg). After a period of 36–48 hr, whole cell extracts were obtained as previously described (Whitelaw et al., 1993a), and proteins were resolved by SDS–15% PAGE. Gal4–Arnt hybrid proteins were detected with mouse polyclonal antisera raised against the Gal4 sequence 1–147 (Santa Cruz Biotechnology, Santa Cruz, CA) followed by reaction with alkaline phosphatase- or horseradish peroxidaseconjugated goat anti-mouse IgG according to the supplier’s instructions (Dako, Glostrup, Denmark). Arnt proteins lacking Gal4 sequences were detected using a rabbit polyclonal antisera directed against a N-terminal peptide of human Arnt as described (McGuire et al., 1994). The sizes of the visualized proteins were compared to rainbow high-range MW markers (Amersham) or stained protein standards (Stratagene, La Jolla, CA). The calculated molecular weights of the proteins were always within 5 kDa of the weight detected by Western analysis and were calculated by multiplying the number of amino acids in the hybrid protein by 110, the average molecular weight of an amino acid. The calculated molecular weights were as follows: Gal4DBD/Arnt1–774, 101.3 kDa; Gal4DBD/Arnt1–581, 80.1 kDa; Gal4DBD/Arnt1–498, 71.8 kDa; Gal4DBD/Arnt1–484, 70.4 kDa; Gal4DBD/Arnt1–331, 52.7 kDa; Gal4DBD/Arnt1–198, 38.0 kDa; Gal4DBD/Arnt1–101, 28.2 kDa; Gal4DBD/Arnt499–581, 25.2 kDa; Gal4DBD/Arnt582–774, 37.3 kDa; Gal4DBD/Arnt582–603/697–774, 27.0 kDa; Gal4DBD/Arnt582–603/697–711, 20.1 kDa; Gal4DBD/Arnt1– 101/499–774, 57.5 kDa; Gal4DBD/VP16, which encodes the 78 carboxyterminal amino acids of VP16 (Sadowski et al., 1988), 24.8 kDa; Gal4DBD/ Arnt699–774, 25.2 kDa; Gal4DBD/Arnt699–760, 23.7 kDa; Gal4DBD/ Arnt699–741, 21.6 kDa; Gal4DBD/Arnt712–774, 23.8 kDa; Gal4DBD/ Arnt712–760, 22.2 kDa; Gal4DBD/Arnt712–741, 20.1 kDa; Gal4DBD/

AID

TOX 7846

/

6h0f$$$$$2

07-08-96 09:36:14

Arnt741–774, 20.6 kDa; Arnt1–331, 36.4 kDa; Arnt1–774, 85.1 kDa; Arnt1–484, 53.2 kDa.

RESULTS AND DISCUSSION

Identification of the Arnt TAD. The role of Arnt amino acids in transactivation was examined by using the wellcharacterized Gal4 hybrid protein approach (Sadowski and Ptashne, 1989). In this technique, cDNA fragments encoding the amino acids to be tested are fused in frame to the first 147 amino acids of the yeast Gal4 transcriptional activator. The Gal4 sequences encode a zinc-cluster DNA binding motif, which directs sequence-specific binding of the Gal4 dimer to an upstream activating sequence (UASG), and a signal sequence, which localizes Gal4 hybrid proteins to the nucleus in yeast and mammalian cells. The ability of the hybrid protein to activate is monitored by measuring the expression of the CAT gene, which is under control of a minimal promoter containing multiple UASG. Only when the cDNA fragment encodes a TAD can the hybrid protein activate. This technique allows the activation functions of Arnt to be assayed independently of disruptions in other functions of the AHR–Arnt heterodimer. The Gal4–Arnt hybrid proteins were tested for their ability to activate CAT gene expression by cotransfecting into COS-1 cells each expression plasmid encoding the Gal4– Arnt hybrid and the G5BCAT reporter plasmid (Lillie and Green, 1989). The full-length Arnt protein, when fused to the Gal4DBD (Fig. 1A, Gal4DBD/Arnt1–774), gave an 18fold increase in the activation of CAT over that by Gal4DBD alone. This activation was especially weak when compared with the activation by the Gal4DBD/VP16 hybrid encoding the potent TAD from the herpes simplex virus VP16 protein. A C-terminal deletion of 193 amino acids to give Gal4DBD/ Arnt1–581 completely abolished the activation, demonstrating that this region encodes a TAD. Because some transcriptional activators encode multiple TADs (Mitchell and Tjian, 1989) or encode TADs that can be repressed by protein sequences in cis (Godowski et al., 1987; Ma and Ptashne, 1987), we made five additional Cterminal deletion mutants of Arnt to uncover repressed or weak TADs. When tested in the transcription assay, these hybrid proteins (with C-terminal amino acids of 498, 484, 331, 198, and 101) were not able to transactivate (Fig. 1A). The lack of activity was not due to the absence of expression since all the hybrid proteins tested in this study could be detected by Western blot analysis using anti-Gal4 antibodies (Fig. 1B). Thus, the Arnt protein in COS-1 cells contains only one TAD, which lies in the C-terminal 193 amino acids. To further define the location of the TAD, we made additional hybrids using cDNA encoding the ARNT C-terminal region (Fig. 1A). As expected, the hybrid Gal4DBD/ Arnt499–581 had no appreciable activity. In contrast, Gal4DBD/Arnt582–774 gave strong activation comparable

toxa

AP: Tox

Arnt ACTIVATION DOMAIN

275

FIG. 1. Activation by Gal4–Arnt hybrids in COS-1 and Arnt-deficient c4 hepatoma cells. (A, Left) Structure of the Gal4–Arnt deletion mutants. The rectangle marked Gal4DBD represents Gal4 sequences 1–147; vertical bars represent the basic helix–loop–helix DNA binding domain; box with left-to-right diagonal lines represents the PAS dimerization domain with the A and B 50-amino-acid repeats represented by black boxes; box with both

AID

TOX 7846

/

6h0f$$7846

07-08-96 09:36:14

toxa

AP: Tox

276

CORTON ET AL.

to the activation by Gal4DBD/VP16. An internal deletion of Gal4DBD/Arnt582–774 to give Gal4DBD/Arnt582–603/ 697–774 did not appreciably affect the high level of activation. A deletion of the C-terminal 63 amino acids from Gal4DBD/Arnt582–593/697–774 to give Gal4DBD/ Arnt582–603/697–711 completely abolished CAT activation. These results demonstrate that the strong TAD of Arnt is located within the C-terminal 63 amino acids of the protein. The C-terminal 63 amino acids of Arnt are rich in amino acids that are found in three types of TAD in other transcription factors, including glutamines (Q-rich), prolines, serines, and threonines (P/S/T-rich) and acidic amino acids (D/Erich) (Mitchell and Tijan, 1989). To distinguish between these different types of activation domains playing a role in Arnt activation, we further subdivided the 63 amino acids into three regions: a Q/S/P-rich region between amino acids 712 and 741; a 19-amino-acid region between amino acids 742 and 760 which contains 3 acidic amino acids, 2 glutamines, and 3 serines; and a 14-amino-acid region between amino acids 761 and 774 containing 4 acidic amino acids, 3 prolines, and 2 serines or threonines but no glutamines. The hybrid protein Gal4DBD/Arnt699–774 (Fig. 2A) exhibited activation about equal to Gal4DBD/VP16 as seen in Fig. 1A. Deletion of amino acids 699–711 to give the construct Gal4DBD/Arnt712–774 resulted in Ç50% loss of activity indicating that these amino acids may provide flexibility for the TAD to interact with the transcription machinery or they act as an independent TAD. This latter possibility is unlikely in light of the complete lack of activity in the construct Gal4DBD/Arnt582–603/697–711 (Fig. 1A) which contains this region. Deletions of the C-terminal 14 amino acids (Gal4DBD/Arnt699–760 and Gal4DBD/Arnt712–760) completely abolished activation. Further deletion of amino acids 742–760 to give Gal4DBD/Arnt699–741 and Gal4DBD/Arnt712–741 did not restore activation. Thus, the region between amino acids 712 and 741 which is rich in amino acids, especially glutamines found in the TAD of other transcription factors, does not function as a TAD in Arnt. The lack of activity of the hybrid proteins which do not contain the 14-amino-acid C-terminal region was not due to lack of expression as all hybrid proteins tested were expressed (Fig. 2B).

Activation by Arnt TAD in Arnt-deficient hepatoma cells. We next wished to determine if the Arnt TAD exhibits the same or different activation behavior in another cellular context. We chose the Arnt-defective mouse mutant hepatoma cell line c4 because it allows the analysis of Arnt sequences in the absence of functional Arnt. We measured the ability of selected Gal4–Arnt hybrids or Arnt deletion mutants lacking Gal4 sequences to activate UASG or DRE-linked CAT genes. The following studies reveal four notable features of transactivation by Arnt sequences in these cells. First, activation by the Gal4–Arnt hybrids qualitatively paralleled that seen in COS-1 cells. The hybrid proteins that contained the C-terminal 63 amino acids (Gal4DBD/ Arnt582–774 and Gal4DBD/Arnt582–603/697–774) were able to activate the G5BCAT reporter plasmid (Fig. 1A) or another reporter plasmid in which four UASGs were inserted in the MMTV promoter (data not shown). The exception was the full-length Arnt, which lacked the ability to activate (discussed below). Hybrids that lacked the C-terminal TADexhibited no significant activation above background (Fig. 1A) (i.e., Gal4DBD/Arnt1-581 and Gal4DBD/Arnt582– 603/697–711). These results demonstrate that the C-terminal 63 amino acids encode the principal TAD of Arnt in these cells. Second, when compared with the strong activation by Gal4DBD/VP16, activation by TAD-containing Arnt hybrids (Gal4DBD/Arnt582–774 Arnt or Gal4DBD/Arnt582– 603/697–774) was 8- to 10-fold weaker. Thus, the Arnt and VP16 TAD have different strengths in this cell line, indicating that they activate transcription by different mechanisms. Third, activation by full-length Arnt in c4 cells was not significantly greater than that by the vector alone. Activation by full-length Arnt in COS-1 cells was also much lower than that by the N-terminal deletion mutants Gal4DBD/Arnt582– 774 and Gal4DBD/Arnt582–603/697–774 (Fig. 1A). These results imply that full-length Arnt harbors sequences that allow repression of activation in the hybrid protein. To begin to understand the nature of this repression, we made an additional Gal4–Arnt hybrid that lacked the helix–loop– helix and PAS domains responsible for heterodimerization with the AHR (Reisz-Porszasz et al., 1994; Whitelaw et al., 1993b). When expressed in COS-1 and c4 cells, the Gal4DBD/Arnt1–101/499–774 activated the CAT gene

left-to-right and right-to-left diagonal lines represents the glutamine-rich region (Q-rich); wavy lines represent the herpes simplex virus VP16 activation domain (VP16). The positions of the amino acids that flank the domains are shown by the numbers under the top line. (Right) Activation by Gal4–Arnt hybrids. COS-1 or hepatoma c4 cells were cotransfected with the reporter plasmid G5BCAT and the plasmids encoding the Gal4–Arnt hybrids as detailed under Materials and Methods. The normalized CAT activity is represented as fold induction relative to the control plasmid pSG424 (Gal4DBD). The values reported are the averages of duplicate measurements in a typical experiment. The standard error of the mean did not vary by more than 25%. Transfections of the plasmids were repeated two to four times with similar results. (B) Expression of Gal4–Arnt hybrids. Extracts from COS-1 cells transfected with the plasmids encoding the indicated proteins or mock-transfected (Mock) followed by treatment with DMSO (0T), 10 nM TCDD (/T), or untreated (all others) were separated on SDS–15% PAGE, transferred to nitrocellulose, and probed with an antibody specific for Gal4 sequences 1– 147. The arrowheads represent the positions of the expressed proteins. Numbers to the side are in kilodaltons and represent the positions of molecular weight markers. Analysis of protein expression was performed on extracts from two separate COS-1 cell transfections with similar results.

AID

TOX 7846

/

6h0f$$$$$3

07-08-96 09:36:14

toxa

AP: Tox

Arnt ACTIVATION DOMAIN

277

FIG. 2. Mapping the activation domain of Arnt. (A, Left) Structure of the Gal4–Arnt hybrids. (Right) Activation by Gal4–Arnt hybrids. COS-1 cells were cotransfected with the reporter plasmid G5BCAT and the plasmids encoding the Gal4–Arnt hybrids as detailed under Materials and Methods. The normalized CAT activity is represented as fold induction relative to the control plasmid pSG424 (Gal4DBD). The values reported are the averages of duplicate measurements in two experiments. The standard error of the mean did not vary by more than 25%. (B) Expression of the Gal4–Arnt hybrids. Extracts from COS-1 cells transfected with the expression plasmids were separated by SDS–15% PAGE, transferred to nitrocellulose, and probed with an antibody specific for Gal4 sequences 1–147. Numbers to the side are in kilodaltons and represent the positions of the molecular weight markers. Lane 1, Gal4DBD/Arnt699–774; lane 2, Gal4DBD/Arnt699–760; lane 3, Gal4DBD/Arnt699–741; lane 4, Gal4DBD/Arnt712–774; lane 5, Gal4DBD/ Arnt712–760; lane 6, Gal4DBD/Arnt712–741; lane 7, Gal4DBD/Arnt741–774.

about as well as or somewhat better than the Gal4DBD/ Arnt582–774 in both cell lines (Fig. 1A), demonstrating that the sequences mediating repression in the full-length Arnt had been removed. Differences in activation in COS-1 cells cannot be attributed to decreased expression of the fulllength Gal4DBD/Arnt1–774 since all hybrid proteins were expressed to similar levels (Fig. 1B). We also tested the effect of TCDD treatment on the repression of the full-length Gal4–Arnt, but concentrations of 10 nM TCDD were without effect on transcriptional activation (data not shown) as well as protein levels (Fig. 1B, compare lanes 2 and 3), indicating that AHR is not involved in mediating the repression. The data suggest that there is a factor or factors that interact with the Gal4–Arnt hybrid through the helix–loop–helix and/or PAS domains and negatively regulate the Arnt TAD function. Precedence for this type of regulation comes from studies of the Drosophila Per protein in which negative regulation of nuclear localization was shown to be mediated by

AID

TOX 7846

/

6h0f$$$$$3

07-08-96 09:36:14

the Per PAS domain (Vosshall et al., 1994). Additional experiments will be required to determine the significance of this negative regulation of Arnt and the manner in which it takes place. Finally, we wished to determine the consequences of deletion of the Arnt TAD on ligand-mediated transcriptional activation by the reconstituted AHR complex. We first tested the ability of a transfected full-length Arnt protein lacking Gal4 sequences (Arnt1–774) (Fig. 3A) to functionally replace the endogenous mutant Arnt in activating a CAT gene under control of the enhancer and promoter of the human CYP1A1 gene. This region of the CYP1A1 gene was previously shown to contain at least two DRE-like elements that direct HAH-specific gene expression (Hines et al., 1988). Transfection of the Arnt1–774 expression plasmid into Arntdeficient cells restored TCDD-inducible CAT expression to a level comparable to that seen in wild-type cells in the absence of exogenous Arnt expression (Fig. 3B). Despite

toxa

AP: Tox

278

CORTON ET AL.

FIG. 3. Effect of Arnt C-terminal trucations on activation of a CYP1A1-linked reporter gene. (A) Structure of Arnt and deletion mutants. Refer to Fig. 1A legend for description of the domains. (B) Activation of the human CYP1A1 enhancer/promoter by full-length Arnt and Arnt deletion mutants. The Arnt-deficient c4 cells were cotransfected with the plasmid pRNH241c encoding the CAT gene under control of the enhancer/promoter of the human CYP1A1 gene and the indicated expression plasmid. The control (ctrl) plasmid used was pSG424DGal4. The cells were treated with DMSO (white bars) or 10 nM TCDD (black bars) for 18 hr. Normalized activity is represented in arbitrary units. Wild-type hepatoma cells (Hepa1) (WT) were also transfected with the pRNH241c and pSG424DGal4 plasmids and treated similarly. The results presented represent the means of duplicate measurements within a single experiment. The standard error of the mean did not vary by more than 25%. Similar results were obtained in three additional experiments. (C) Expression of Arnt and Arnt deletion mutants. Extracts from cells transfected with plasmids encoding the indicated proteins or mock transfected (Mock) were analyzed as described in the legend to Fig. 1B but were probed with a polyclonal antibody specific for the N-terminus of Arnt. The arrowheads represent the positions of the expressed proteins. Numbers to the side are in kilodaltons and represent the positions of molecular weight markers. Analysis of protein expression was performed on extracts from two separate transfections with similar results.

deletion of the C-terminal TAD, Arnt1–484 still retained about the same ability as wild-type Arnt to activate the reporter gene. Thus, the Arnt TAD is not necessary for restoration of the Arnt defect. In contrast, the Arnt1–331 mutant lacking the TAD and the PAS B domain completely lost the ability to activate the reporter gene. Consistent with this, deletion of the PAS B domain in Arnt drastically reduced dimerization with AHR as well as DRE binding (ReiszPorszasz et al., 1994). Although the Arnt TAD may contribute to the overall strength of the activation by the AHR complex in other contexts, in hepatoma c4 cells the AHR provides the dominant TAD for ligand-dependent activation of CYP1A1. In addition, Arnt1–484 and to a lesser extent the wild-type protein gave increased levels of expression in the absence of TCDD as seen earlier (Whitelaw et al., 1994). Increases in constitutive reporter gene expression may be due to the ability of increased levels of Arnt to remove HSP90 from AHR, allowing AHR translocation to the nucleus. Increased levels of Arnt may explain in part why, in the absence of TCDD

AID

TOX 7846

/

6h0f$$$$$3

07-08-96 09:36:14

treatment, AHR can be found in the nuclei of a number of cell types from mouse tissues (Abbott et al., 1994). Nature and function of the Arnt TAD. Our deletion analysis of Arnt demonstrated the existence of a TAD within the 34 C-terminal amino acids. Deletion of the C-terminal 14 amino acids completely abolished activity. We show that the Q-rich region between amino acids 712 and 741, a notable feature in the defined TAD in other studies (Reisz-Porszasz et al., 1994; Li et al., 1994; Whitelaw et al., 1994; Jain et al., 1994), does not exhibit any transcription activation activity either by itself or juxtaposed to the 13 N-terminal or the 19 C-terminal adjacent amino acids. It is still possible that this Q-rich region may act as a TAD in another promoter context or in another cell type. The 34-amino-acid region which retains activity in COS-1 cells contains 7 acidic residues, 4 serines or threonines, 4 prolines, and 2 glutamines. The activity of the Arnt TAD is most likely not determined by the number of prolines, glutamines, or serines/threonines in this region because there are a number of properties which distinguish the TAD of Arnt from the Q-rich TAD of Sp1,

toxa

AP: Tox

279

Arnt ACTIVATION DOMAIN

Oct-1, Oct-2, and CREB and the P/S/T-rich TAD typified by AP-2 and CTF. First, the Arnt TAD lacks the arrangement of hydrophilic and hydrophobic residues (the hydrophobic patch) within the Q-rich region of Sp1 and CREB important for transcription activation and interaction with the intermediary transcription factor dTAFII110 (Gill et al., 1994). Second, Sp1 sites are usually found within promoters, whereas DRE are found distal to promoters, implying that these proteins and their TAD work by different mechanisms. In support of this, P/S/T-rich and Q-rich TAD have little if any activity at remote enhancer-like positions but stimulate transcription from promoter-like positions (Seipel et al., 1992). In this regard, the Arnt TAD behaves more like the D/Erich TAD, which can activate at both promoter-like and enhancer-like positions (Seipel et al., 1992). Finally, the individual components of the AHR–Arnt complex have been shown to function as activators of transcription in yeast (Carver et al., 1994; Whitelaw et al., 1995). In contrast, P/S/T-rich TAD from AP-2 and CTF as well as Q-rich TAD from Sp1 have little if any activity in yeast (Ku¨nzler et al., 1994; Ponticelli et al., 1995). The activity of the Arnt TAD may be due to the high concentration of acidic residues which gives the C-terminal 34 amino acids a net charge of 07. Inspection of the arrangement of acidic and hydrophobic residues in this region did not reveal any obvious similarities to the AF-2 TAD of nuclear receptors which contain an invariant glutamic acid (Beato et al., 1995). An intriguing possibility is that the activity of the Arnt TAD resides in the 14 C-terminal amino acids which have an arrangement similar to that in the Gal4 activation domain (Leuther et al., 1993) including a predicted b-turn flanked by a b-sheet. Mutagenesis experiments are required to determine if, as in the Gal4 TAD, the acidic residues are dispensible and if activity requires a b-sheet structure. Our studies and those of others demonstrated that the strength of the Arnt TAD is dependent on both the cell type and promoter architecture. When compared with the activity of the D/E-rich VP16 TAD, we showed that the Arnt TAD had the same strength in COS-1 cells but was Ç10-fold weaker in hepatoma c4 cells (Fig. 1A). When compared with the AHR TAD, the Arnt TAD exhibited lower activity in human hepatoma HepG2 cells but similar activity in three other cell types (Whitelaw et al., 1994). Differences in the ability of the Arnt TAD to activate transcription at the same promoter in different cellular contexts indicates that the TAD functions by contacting cell-specific intermediary factors or components of the transcription complex assembled at the promoter. Furthermore, within a particular cell type, the arrangement of the DRE relative to the promoter or the nature of the promoter itself is likely to influence the strength of the Arnt TAD. In c4 hepatoma cells, an Arnt mutant lacking the C-terminal TAD was able to completely restore Arnt activity at the human CYP1A1 enhancer/promoter (Fig. 2B),

AID

TOX 7846

/

6h0f$$$$$3

07-08-96 09:36:14

indicating that Arnt TAD is silent and the AHR within the heterodimer provides the dominant TAD. At other promoters however, including the rat CYP1A1 enhancer/promoter (Reisz-Porszasz et al., 1994) or the MMTV promoter artificially juxtaposed to a DRE (Whitelaw et al., 1994), the heterodimer containing the TAD-less Arnt mutant retained only 50 or 10% of wild-type activity, respectively. These studies indicate that the Arnt or AHR TAD can play dominant roles in transcription activation depending on the promoter structure as well as the cell type. Because there is structural and functional conservation of most if not all domains within the mouse and human AHR and Arnt (Burbach et al., 1992; Dolwick et al., 1993a,b; Reisz-Porszasz et al., 1994; Whitelaw et al., 1994; Li et al., 1994), the strikingly dissimilar species- and tissue-specific toxic responses observed in humans and mice exposed to dioxins and HAH may be due in part to the availability or functional activity of the transcription factors with which these TAD interact. ACKNOWLEDGMENTS The authors thank Drs. Tony Fox, Julian Preston, and Craig Rowlands for comments concerning this work; Dr. Ivan Sadowski for G5BCAT, pSG424, and pSGVP plasmids; Dr. Oliver Hankinson for the plasmid Arnt/ pTZ18U; Dr. Ronald Hines for the plasmid pRNH241c; Dr. William Greenlee for the mouse hepatoma cell lines; and Drs. Lorenz Poellinger and Murray Whitelaw for anti-Arnt antibody and helpful discussions. We also thank Sadie Leak and Linda Smith for preparation of the manuscript and Stan Piestrak and Robert Stallings for art work. This work was presented in part at the conference ‘‘Receptor-Mediated Mechanisms of Chemical Carcinogenesis,’’ Barton Creek, Texas, December 1992. It has been brought to our attention that in a recent paper Sogawa et al. (1995), using methods similar to ours, have also concluded that the Arnt TAD resides in the Cterminal 34 amino acids.

REFERENCES Abbott, B. D., Perdew, G. H., and Birnbaum, S. L. (1994). Ah receptor in embryonic mouse palate and effects of TCDD on receptor expression. Toxicol. Appl. Pharmacol. 126, 16–25. ˚ ., and PoelAntonsson, C., Whitelaw, M. L., McGuire, J., Gustafsson, J.-A linger, L. (1995). Distinct roles of the molecular chaperone hsp90 in modulating dioxin receptor function via the basic helix–loop–helix and PAS domains. Mol. Cell. Biol. 15, 756–765. Beato, M., Herrlich, P., and Schu¨tz, G. (1995). Steroid hormone receptors: Many actors in search of a plot. Cell 83, 851–857. Bertazzi, P. A., Pesatori, A. C., Consonni, D., Tironi, A., Landi, M. T., and Zocchetti, C. (1993). Cancer incidence in a population accidentally exposed to 2,3,7,8-tetrachlorodibenzo-p-dioxin. Epidemiology 4, 398– 406. Burbach, K. M., Poland, A., and Bradfield, C. A. (1992). Cloning of the Ah-receptor cDNA reveals a distinctive ligand-activated transcription factor. Proc. Natl. Acad. Sci. USA 89, 8185–8189. Carver, L. A., Jackiw, V., and Bradfield, C. A. (1994). The 90-kDa heat shock protein is essential for Ah receptor signaling in a yeast expression system. J. Biol. Chem. 269, 30109–30112. Chen, H.-S., and Perdew, G. H. (1994). Subunit composition of the hetero-

toxa

AP: Tox

280

CORTON ET AL.

meric cytosolic aryl hydrocarbon receptor complex. J. Biol. Chem. 253, 27551–27558. Dolwick, K. M., Schmidt, J. V., Carver, L. A., Swanson, H. I., and Bradfield, C. A. (1993a). Cloning and expression of a human Ah receptor cDNA. Mol. Pharmacol. 44, 911–917. Dolwick, K. M., Swanson, H. I., and Bradfield, C. A. (1993b). In vitro analysis of Ah receptor domains involved in ligand-activated DNA recognition. Proc. Natl. Acad. Sci. USA 90, 8566–8570. Fingerhut, M. A., Halperin, W. E., Marlow, D. A., Piagitelli, L. A., Honchar, P. A., Sweeney, M. H., Greife, A. L., Dill, P. A., Steeland, K., and Suruda, A. J. (1991). Cancer mortality in workers exposed to 2,3,7,8tetrachlorodibenzo-p-dioxin. N. Engl. J. Med. 324, 212–218. Gill, G., Pascal, E., Tseng, Z. H., and Tjian, R. (1994). A glutatmine-rich hydrophobic patch in transcription factor Sp1 contacts the dTAFII110 component of the Drosophila TFIID complex and mediates transcriptional activation. Proc. Natl. Acad. Sci. USA 91, 192–196. Godowski, P. J., Rusconi, S., Miesfeld, R., and Yamamoto, K. R. (1987). Glucocorticoid receptor mutants that are constitutive activators of transcriptional enhancement. Nature 325, 365–368. Gorman, C. M., Moffatt, L. F., and Howard, B. H. (1982). Recombinant genomes which express chloramphenicol acetyltransferase in mammalian cells. Mol. Cell. Biol. 2, 1044–1051. Hines, R. N., Mathis, J. M., and Jacob, C. S. (1988). Identification of multiple regulatory elements on the human cytochrome P4501A1 gene. Carcinogenesis 9, 1599–1605. Hoffman, E. C., Reyes, H., Chu, F.-F., Sander, F., Conley, L. H., Brooks, B. A., and Hankinson, O. (1991). Cloning of a factor required for activity of the Ah (dioxin) receptor. Science 242, 954–958. Huang, Z. J., Edery, I., and Rosbash, M. (1993). PAS is a dimerization domain common to Drosophila period and several transcription factors. Nature 364, 259–262. Jain, S., Dolwick, K. M., Schmidt, J. V., and Bradfield, C. A. (1994). Potent transactivation domains of the Ah receptor and the Ah receptor nuclear translocator map to their carboxyl termini. J. Biol. Chem. 269, 31518– 31524. Ku¨nzler, M., Braus, G. H., Georgiev, O., Seipel, K., and Schaffner, W. (1994). Functional differences between mammalian transcription activation domains at the yeast GAL1 promoter. EMBO J. 13, 641–645. Leuther, K. K., Salmeron, J. M., and Johnston, S. A. (1993). Genetic evidence that an activation domain of GAL4 does not require acidity and may form a b sheet. Cell 72, 575–585. Li, H., Dong, L., and Whitlock, J. P., Jr. (1994). Transcriptional activation function of the mouse Ah receptor nuclear translocator. J. Biol. Chem. 269, 28098–28105. Lillie, J. W., and Green, M. R. (1989). Transcription activation by the adenovirus Ela protein. Nature 338, 39–44. Ma, J., and Ptashne, M. (1987). Deletion analysis of GAL4 defines two transcriptional activating segments. Cell 48, 847–853. ˚ ., and PoelMcGuire, J., Whitelaw, M. L., Pongratz, I., Gustafsson, J.-A linger, L. (1994). A cellular factor stimulates ligand-dependent release of hsp90 from the basic helix-loop-helix dioxin receptor. Mol. Cell. Biol. 14, 2438–2446. Mitchell, P. J., and Tjian, R. (1989). Transcriptional regulation in mammalian cells by sequence-specific DNA binding proteins. Science 245, 371– 378. Murre, C., Bain, G., van Dijk, M. A., Engel, I., Furnari, B. A., Massari, M. E., Matthews, J. R., Quong, M. W., Rivera, R. R., and Stuiver, M. H. (1994). Structure and function of helix-loop-helix proteins. Biochim. Biophys. Acta 1218, 129–135. Nambu, J. R., Lewis, J. O., Wharton, K. A., Jr., and Crews, S. T. (1991).

AID

TOX 7846

/

6h0f$$$$$4

07-08-96 09:36:14

The Drosophila single-minded gene encodes a helix-loop-helix protein that acts as a master regulator of CNS midline development. Cell 67, 1157–1167. Okey, A. B., Riddick, D. S., and Harper, P. A. (1994). The Ah receptor: Mediator of the toxicity of 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) and related compounds. Toxicol. Lett. 70, 1–22. Poland, A., Palen, D., and Glover, E. (1994). Analysis of the four alleles of the murine aryl hydrocarbon receptor. Mol. Pharmacol. 46, 915–921. Pollenz, R. S., Sattler, C. A., and Poland, A. (1993). The aryl hydrocarbon receptor and aryl hydrocarbon receptor nuclear translocator protein show distinct subcellular localizations in hepa 1c1c7 cells by immunofluorescence microscopy. Mol. Pharmacol. 45, 428–438. Ponticelli, A. S., Pardee, T. S., and Struhl, K. (1995). The glutaminerich activation domains of human Sp1 do not stimulate transcription in Sacharomyces cerevisiae. Mol. Cell. Biol. 15, 983–988. Reisz-Porszasz, S., Probst, M. R., Fukunaga, B. N., and Hankinson, O. (1994). Identification of functional domains of the aryl hydrocarbon receptor nuclear translocator protein (ARNT). Mol. Cell. Biol. 14, 6075– 6086. Robertson, R. W., Zhang, L., Pasco, D. S., and Fagan, J. B. (1994). Aryl hydrocarbon-induced interactions at multiple DNA elements of diverse sequence—A multicomponent mechanism for activation of cytochrome P4501A1 (CYP1A1) gene transcription. Nucleic Acids Res. 22, 1741– 1749. Sadowski, I., and Ptashne, M. (1989). A vector for expressing GAL4(1– 147) fusions in mammalian cells. Nucleic Acids Res. 17, 7539. Sadowski, I., Ma, J., Triezenberg, S., and Ptashne, M. (1988). GAL4-VP16 is an unusually potent transcriptional activator. Nature 335, 563–564. Seipel, K., Georgiev, O., and Schaffer, W. (1992). Different activation domains stimulate transcription from remote (‘‘enhancer’’) and proximal (‘‘promoter’’) positions. EMBO J. 11, 4961–4968. Silbergeld, E. K. (1991). Dioxin: A case study in chloracne. In Dermatology (F. M. Marzulli and H. I. Maibach, Eds.), pp. 667–686, Hemisphere, Washington DC/New York. Sogawa, K., Iwabuchi, K., Abe, H., and Fujii-Kuriyama, Y. (1995). Transcriptional activation domains of the Ah receptor and Ah receptor nuclear translocator. J. Cancer Clin. Oncol. 121, 612–620. Vanden Heuval, J. P., and Lucier, G. (1993). Environmental toxicology of polychlorinated dibenzo-p-dioxins and polychlorinated dibenzofurans. Environ. Health Perspect. 100, 189–200. Vosshall, L. B., Price, J. L., Sehgal, A., Saez, L., and Young, M. W. (1994). Block in nuclear localization of period protein by a second clock mutation, timeless. Science 263, 1606–1609. ˚ ., and Poellinger, L. Whitelaw, M. L., Go¨ttlicher, M., Gustafsson, J.-A (1993a). Definition of a novel ligand binding domain of a nuclear bHLH receptor: Co-localization of ligand and hsp90 binding activities within the regulatable inactivation domain of the dioxin receptor. EMBO J. 12, 4169–4169. ˚ ., and PoelWhitelaw, M., Pongratz, I., Wilhelmsson, A., Gustaffson, J.-A linger, L. (1993b). Ligand-dependent recruitment of the Arnt coregulator determines DNA recognition by the dioxin receptor. Mol. Cell. Biol. 13, 2504–2514. ˚ ., and PoelWhitelaw, M. L., McGuire, J., Picardo, D., Gustafsson, J.-A linger, L. (1995). Heat shock protein hsp90 regulates dioxin receptor function in vivo. Proc. Natl. Acad. Sci. USA 92, 4437–4441. ˚ ., and Poellinger, L. (1994). Identification Whitelaw, M., Gustafsson, J.-A of transactivation and repression functions of the dioxin receptor and its basic helix-loop-helix/PAS partner factor Arnt: Inducible versus constitutive modes of regulation. Mol. Cell. Biol. 14, 8343–8355. Wu, L., and Whitlock, J. P., Jr. (1993). Mechanism of dioxin action: Receptor-enhancer interactions in intact cells. Nucleic Acids Res. 21, 119–125.

toxa

AP: Tox