Dissociation of the AhR/ARNT complex by TGF-β/Smad signaling represses CYP1A1 gene expression and inhibits benze[a]pyrene-mediated cytotoxicity.

Cytochrome P450 1A1 (CYP1A1) catalyzes the metabolic activation of polycyclic aromatic hydrocarbons (PAHs) such as benzo[a]pyrene (B[a]P) and is transcriptionally regulated by the aryl hydrocarbon receptor (AhR)/AhR nuclear translocator (ARNT) complex upon exposure to PAHs. Accordingly, inhibition of CYP1A1 expression reduces production of carcinogens from PAHs. Although transcription of the CYP1A1 gene is known to be repressed by transforming growth factor-β (TGF-β), how TGF-β signaling is involved in the suppression of CYP1A1 gene expression has yet to be clarified. In this study, using mammalian cell lines, along with shRNA-mediated gene silencing, CRISPR/Cas9-based genome editing, and reporter gene and quantitative RT-PCR assays, we found that TGF-β signaling dissociates the B[a]P-mediated AhR/ARNT heteromeric complex. Among the examined Smads, Smad family member 3 (Smad3) strongly interacted with both AhR and ARNT via its MH2 domain. Moreover, hypoxia-inducible factor 1α (HIF-1α), which is stabilized upon TGF-β stimulation, also inhibited AhR/ARNT complex formation in the presence of B[a]P. Thus, TGF-β signaling negatively regulated the transcription of the CYP1A1 gene in at least two different ways. Of note, TGF-β abrogated DNA damage in B[a]P-exposed cells. We therefore conclude that TGF-β may protect cells against carcinogenesis because it inhibits CYP1A1-mediated metabolic activation of PAHs as part of its anti-tumorigenic activities.

The cytochrome P450 (CYP) superfamily, known as a family of heme-containing enzymes, can catalyze the monooxygenation of not only endogenous substrates but also xenobiotics. The human body has 57 different CYPs, which contribute to assorted metabolic and biosynthetic pathways (1). Among them, the CYP1A family, consisting of CYP1A1 and CYP1A2, is known to activate a variety of compounds with carcinogenic properties. CYP1A1 catalyzes epoxidation of an aromatic ring with a vacant position to initiate carcinogenesis via the formation of a highly reactive moiety that can promote oncogenic mutation in the genome. On the other hand, CYP1A1 is implicated in the metabolization of ingested natural aryl hydrocar-bon receptor (AhR) ligands in the intestine to fine-tune intestinal immunity (2).
Some of the environmental carcinogens that CYP1A1 can metabolize promote the activation of the CYP1A1 gene via xenobiotic-responsive elements (XREs) in its promoter as a ciselement (3). Both the AhR and the AhR nuclear translocator (ARNT), which possess an N-terminal bHLH (basic helix-loophelix) motif flanking a PAS (Per, ARNT, Sim) domain in their structures, play an essential role in the activation of the CYP1A1 gene through their binding to XREs (4,5). The inactive AhR that interacts with chaperon complex, including Hsp90, a co-chaperone p23, and hepatitis B virus X-associated protein 2 (XAP2), is located in cytosols. When AhR binds to polycyclic aromatic hydrocarbons (PAHs), AhR dissociates from the complex among Hsp90, p23, and XAP2 to translocate to the nucleus, where AhR forms a heteromeric complex with ARNT. Then the AhR/ARNT complex can bind to XREs within the CYP1A1 promoter to enhance its transcription, in cooperation with coactivators and chromatin-remodeling factors (6)(7)(8). On the other hand, the AhRR (AhR repressor), which is also induced by PAHs, forms a complex with ARNT to compete with AhR/ARNT for the binding to XRE, followed by recruitment of transcriptional corepressors such as ANKRA2, HDAC4, and HDAC5 to suppress the CYP1A1 gene (9). The activity of AhR can be regulated by interaction with other transcription factors or its phosphorylation. Estrogen receptor a or glucocorticoid receptor augments PAH-induced CYP1A1 mRNA expression through its physical interaction with AhR, whereas NF-kB or HIF-1a that is stabilized under hypoxia suppresses CYP1A1 expression (10,11). Phosphorylation of AhR by protein kinase C or c-Src further up-regulates the CYP1A1 gene, whereas that by protein kinase A attenuates CYP1A1 induction by 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) (12). Besides, the classical mitogen-activating protein kinase pathway increases the TCDD-mediated induction of CYP1A1 via AhR and simultaneously sensitizes AhR for ubiquitin-dependent degradation (13,14). In addition, AhR is rapidly degraded via cytoplasmic ubiquitin/proteasome pathway after AhR is exported from the nucleus (7,15).
TGF-b is known to possess the inhibitory action of the enzymatic activity of CYP1A1, which is induced by TCDD. When the mRNA expression of the CYP1A1 gene was examined, TGF-b directly down-regulated the transcription of the CYP1A1 gene and other AhR-regulated genes upon TCDD (26). As one of its inhibitory mechanisms, TGF-b repressed the transcription of the AhR gene through the Smad4-dependent pathway (27). However, an inhibitory mechanism(s) other than the TGF-b-mediated suppression of AhR expression might also contribute to the negative regulation of CYP1A1 gene expression by TGF-b.
In this current study, we found that Smad3 activated upon TGF-b stimulation attenuates B[a]P-induced CYP1A1 gene expression via its interaction with both AhR and ARNT. Furthermore, TGF-b retained AhR in the cytosol. Besides, HIF-1a stabilized by TGF-b could compete with AhR for binding to ARNT, which was followed by dissociation of the AhR/ARNT heteromeric complex. Therefore, TGF-b signaling possesses a dual role of inhibiting the transcription of the CYP1A1 gene. Consistently, TGF-b could reduce the DNA damage provoked by B[a]P owing to the decreased expression of CYP1A1.

Inhibition of CYP1A1 expression by TGF-b
B[a]P, one of the PAHs, is a well-known inducer of the CYP1A1 gene. When A549 cells were stimulated with B[a]P, we could see the induction of CYP1A1 mRNA with the maximum expression 6 h after the treatment with B[a]P (Fig. 1A). Like the CYP1A1 gene, B[a]P could also induce the expression of CYP1B1 mRNA (Fig. 1B), as described previously (26,28). On the other hand, AhR mRNA was not affected by B[a]P (Fig. 1B). Pretreatment of the cells with different doses of TGF-b for 2 h exhibited suppression of the CYP1A1 and CYP1B1 mRNA expression induced by B[a]P, although the inhibitory effect of TGF-b on B[a]P-induced CYP1B1 mRNA expression was marginal ( Fig. 1, C and D). Furthermore, we could observe TGF-bmediated suppression of B[a]P-induced CYP1B1 protein in mouse hepatocytes as well (Fig. 1E), whereas we could not succeed in detecting endogenous CYP1A1 protein using several commercially available antibodies (data not shown).
To determine whether the TGF-b/Smad pathway is implicated in the inhibitory action of B[a]P-mediated CYP1A1 expression, we prepared primary hepatocytes from both WT and Smad3-knockout (Smad3-KO) mice lacking TGF-b/Smad3 signaling (29,30). Although TGF-b perturbed B[a]P-induced CYP1A1 transcription in hepatocytes from the WT mice, CYP1A1 mRNA induced by B[a]P was only weakly affected by TGF-b in the Smad3-deficient hepatocytes ( Fig. 2A). Similarly, TGF-b did not reduce the B[a]P-induced transcription of the CYP1B1 gene in Smad3-deficient hepatocytes as much as in the WT hepatocytes (Fig. 2B). Curiously, high expression of both CYP1A1 and CYP1B1 mRNAs was observed without TGF-b stimulation in hepatocytes from Smad3-KO mice. We hypothesized that endogenous TGF-b secreted from hepatocytes might be involved in suppression of the basal expressions of CYP1A1 and CYP1B1 transcription in WT mice. To prove our hypothesis, we stimulated hepatocytes from WT mice with B[a]P 8 h after the addition of SB431542, a TGF-b type I receptor kinase inhibitor (Fig. 2, D and E). The B[a]P-induced mRNA expression of both CYP1A1 and CYP1B1 was further enhanced in the presence of SB431542. These results indicate that endogenous TGFb might be partially implicated in inhibition of CYP1A1 and CYP1B1 expression. Of note, similar inhibitory effects by TGFb were observed both in human and mouse cells even though the human AhR not only shares limited sequence identity to mouse AhR but also shows different affinity to TCDD or other AhR ligands to mouse AhR (31)(32)(33)(34).

Implication of XREs in TGF-b-mediated repression of CYP1A1 promoter activity
The promoter activity of the CYP1A1 gene by PAHs is regulated via XREs as cis-elements, where the AhR/ARNT complex binds (3). To seek the negative cis-element(s) through which TGF-b represses the B[a]P-induced promoter activity of the CYP1A1 gene, we cloned the fragment from 119 to 24150 of the human CYP1A1 gene and inserted it into the pGL3-basic vector (CYP1A1(24150)-luc). In parallel, we made one 59 deletion mutant of the CYP1A1 promoter, termed CYP1A1 (21249)-luc (Fig. 3A). When we investigated whether the human CYP1A1 promoter includes a TGF-b-mediated inhibitory cis-element(s), the B[a]P-induced activities of both CYP1A1(24150)-luc and CYP1A1(21249)-luc were significantly perturbed by TGF-b (Fig. 3B). To narrow down the negative cis-element(s) in the CYP1A1 promoter, we made three luciferase reporter constructs deleted from the 59 distal region of CYP1A1(21249)-luc, termed CYP1A1(2931)-luc, CYP1A1 (2620)-luc, and CYP1A1(2313)-luc (Fig. 3A). Then, after the cells were prestimulated with TGF-b, the B[a]P-induced activity of each reporter was evaluated (Fig. 3C). The TGF-b-mediated inhibition of the reporter activity by B[a]P declined, dependent on the decreased numbers of XREs in their promoter regions. In particular, the CYP1A1(2313)-luc reporter lacking a typical XRE in its promoter region was not affected by either B[a]P or TGF-b. Thus, we hypothesized that TGF-b2mediated repression of the CYP1A1 promoter activity induced by B[a]P involves XREs. To test this hypothesis, we inserted two, four, and six copies of the XRE in front of a minimal promoter (35), termed pGL3ti-(XRE) 2 -luc, pGL3ti-(XRE) 4 -luc, and pGL3ti-(XRE) 6 -luc, respectively (Fig. 3D), and tested the effect of TGFb on the B[a]P-induced activity of these constructs in A549 cells. In all three reporter constructs, TGF-b inhibited the B[a]P-induced reporter activities, although the inducibility of the reporter activity upon B[a]P stimulation depended on the numbers of XREs (Fig. 3E). On the other hand, the luciferase reporter possessing four mutated XREs, termed pGL3ti-(mXRE) 4 -luc, could respond against B[a]P nor TGF-b (Fig. S1, A-C).

Requirement of the canonical Smad pathway to inhibit the CYP1A1 promoter activity induced by B[a]P
To investigate whether the Smad pathway is indispensable for the inhibition of the CYP1A1 promoter activity induced by B[a]P, we introduced pGL3ti-(XRE) 2 -luc into mouse embryonic fibroblasts (MEFs) from either Smad3 heterozygous (Smad3-HET) or Smad3-KO mice (29,30), which was followed by pretreatment with TGF-b 2 h before the addition of B[a]P. As seen in Fig. 4A, deficiency of the Smad3 gene disabled TGF-b-mediated suppression of the CYP1A1 promoter activity induced by B[a]P. Although both Smad2 and Smad3 can transduce the intracellular canonical Smad pathway together with Smad4, Smad2 does not possess the ability to bind to the specific DNA element (termed SBE) because of the insertion of a 30-aminoacid length immediately behind the DNA-binding domain in the MH1 domain of Smad2 (36,37). When either Smad2 or Smad3 was transfected into MEFs from Smad3-KO mice together with pGL3ti-(XRE) 2 -luc, Smad2 and Smad3 improved TGF-b-mediated repression of the reporter activity induced by B[a]P, although the effect of Smad2 was weaker than that of Smad3 (Fig. 4B). We also observed that the introduction of Smad2Dex3 (38) could also recover TGF-b-mediated inhibition of B[a]P-induced reporter activity, like Smad3 (data not shown). To further confirm that TGF-b/Smad3 signaling perturbs B[a]P-induced CYP1A1 gene activation, we generated A549 cells lacking the Smad3 gene by means of the CRISPR/ Cas9 system (A549-S3KO cells) (Fig. S2). In addition to A549-S3KO cells, we established A549-S3KO cells overexpressing human Smad3, termed A549-S3KO(S3OE) cells (Fig. 4C). When each cell line was transfected with pGL3ti-(XRE) 2 -luc, enhancement of the B[a]P-induced reporter activity in A549-S3KO cells could be observed. Furthermore, TGF-b-mediated suppression of the B[a]P-induced reporter activity in A549-S3KO cells was lowered when compared with that in the control cells. On the other hand, introduction of human Smad3 into A549-S3KO cells improved the ability of TGF-b-mediated suppression of the B[a]P-induced reporter activity (Fig. 4D).
Because Smad proteins are known to interact with a number of transcriptional factors (37), we tried to examine their interaction with AhR and ARNT. As seen in Fig. 5 (A and B), Smad3 could associate with both AhR and ARNT most strikingly among the Smads examined. Thus, we tried to clarify the involvement of TGF-b/Smad3 signaling in the transcriptional control of the CYP1A1 gene in the following experiments. When we tested the endogenous interaction between Smad3 and AhR in A549 cells, we could see a TGF-b-dependent association between the two proteins ( Fig. 5C). We also tried to show the endogenous interaction between Smad3 and ARNT. However, Smad3 was never coimmunoprecipitated with ARNT using the anti-ARNT antibody that can detect endogenous ARNT protein in A549 cells. Instead of coimmunoprecipitation, we adopted a proximity ligation assay (PLA) to detect the endogenous interaction between Smad3 and ARNT in cells. When both rabbit anti-ARNT and mouse anti-Smad3 antibodies were simultaneously added to the fixed sample in which cells were pretreated with TGF-b for 2 h, a large number of red spots could be observed (Fig. 5D). On the other hand, no dots were detected when mouse anti-Myc9E10 and rabbit anti-ARNT antibodies were used for the PLA (Fig. 5E).

Disruption of the AhR/ARNT complex by TGF-b/Smad3 signaling
Having shown that Smad3 can interact not only with AhR but also with ARNT, we hypothesized that TGF-b/Smad3 signaling can dissociate the complex between AhR and ARNT. The treatment of A549 cells with B[a]P promoted the complex formation between AhR and ARNT, whereas the introduction of Smad3 reduced it. TGF-b stimulation obviously further dissociated the AhR/ARNT complex in the presence of Smad3 (Fig. 6A).
Smad3 is composed of three domains: the Mad homology 1 (MH1), linker, and MH2 domains (36). To confirm which domain(s) is required for Smad3 to interact with AhR and ARNT, we carried out a coimmunoprecipitation (co-IP) assay. We observed that both AhR and ARNT interact with the MH2 domain of Smad3, whereas the Linker 1 MH2 domain of Smad3 only marginally associated with AhR and ARNT (Fig. 6, B and C). Consistently, overexpression of the Smad3 MH2 domain partially disrupted the B[a]P-induced complex between AhR and ARNT (Fig. 6D). The AhR/ARNT complex can bind to XREs in the presence of B[a]P. Overexpression of the MH2 domain of Smad3 indeed inhibited the binding of the AhR/ARNT complex to XREs (Fig. 7, A and B), although Smad3 by itself did not have any ability to bind to XREs directly (Fig. S3). We were also convinced that TGF-b interrupted the binding of the AhR/ ARNT complex to XREs in the presence of B[a]P (Fig. 7C).

Contribution of HIF-1a to the inhibition of TGF-b-mediated transcriptional repression of the CYP1A1 gene induced by B[a]P
In addition to AhR, ARNT can form a heteromeric complex with HIF-1a, whose expression is known to be induced by TGF-b and contributes to the activation of hypoxia-related genes (39)(40)(41). Indeed, we could detect a prolonged t ½ of HIF-1a protein in the presence of TGF-b when HIF-1a was expressed in 911 cells (Fig. 8A). Next, we investigated whether HIF-1a stabilized by TGF-b interrupts the B[a]P-induced transcription of the CYP1A1 gene. To address this possibility, we used constitutively active HIF-1a (caHIF-1a) in which two proline residues targeted by prolyl hydroxylase (PHD) were replaced with alanine and glycine at positions 402 and 564, respectively (39), because the WT HIF-1a protein was unstable in the cells. Then caHIF-1a was transfected with pGL3ti-(XRE) 2 -luc in the presence or absence of B[a]P. As seen in Fig.  8B, caHIF-1a dose-dependently interfered with the reporter activity induced by B[a]P. When we transfected the fusion gene between AhR and the Gal4 DNA-binding domain (Gal4 DBD) together with ARNT, we could observe transactivation of the reporter gene including the Gal4 DNA-binding sequence as an enhancer, although B[a]P marginally enhanced its transactivation. caHIF-1a obviously blocked the basal and B[a]P-induced reporter activities (Fig. 8C). To further confirm the above results, we also tested whether decreased expression of HIF-1a promoted TGF-b-mediated suppression of B[a]P-induced CYP1A1 transcription in A549 cells. Thus, we generated two independent HIF-1a-knocked-down cell lines, termed A549-shHIF-1a-2 and A549-shHIF-1a-3 (Fig. 8D). When these cells were transfected with pGL3ti-(XRE) 2 -luc, their TGF-b-mediated suppression of B[a]P-induced CYP1A1 transcription was reduced when compared with that of A549-shCont (Fig. 8E). Therefore, HIF-1a stabilized by TGF-b might contribute to inhibition of B[a]P-induced CYP1A1 expression to some extent. From these results, we speculated that HIF-1a disrupted the heteromeric complex between AhR and ARNT. Expectedly, HIF-1a promoted the dissociation between AhR and ARNT.

Effect of TGF-b on the genomic instability mediated by B[a]P
Because TGF-b inhibits the expression of CYP1A1 induced by B[a]P, the CYP1A1-mediated metabolic activation of B[a]P might decrease in cells exposed to TGF-b. To evince this possibility, we examined the expression of phosphorylated histone H2AX (g-H2AX), which is an indicator of DNA double-strand breaks (42). Fluorescence microscopy showed that B[a]Pinduced expression of g-H2AX was reduced upon TGF-b stimulation (Fig. 9A). Consistently, Western blotting analysis also confirmed the reduction of B[a]P-induced g-H2AX protein upon TGF-b stimulation (Fig. 9B). On the other hand, loss of TGF-b/Smad3 signaling in hepatocytes potentiated expression of B[a]P-mediated g-H2AX protein even in the presence of TGF-b (Fig. 9C).

Discussion
Integration of several diverse signal transduction pathways is implicated in tumor development. Among them, TGF-b is known to possess biphasic effects during tumorigenicity: in the early phase of tumorigenicity, TGF-b suppresses tumor growth via induction of cyclin-dependent kinase inhibitors and repression of c-Myc expression, whereas in the late stage, it promotes cell motility, the epithelial mesenchymal transition, angiogenesis, and suppression of cytotoxic T cells and NK cells (43). Thus, the TGF-b signaling pathway is the so-called "Jekyll and Hyde" of cancer because it possesses the contradictory roles of tumor suppression and tumor promotion (44). PAHs, including B[a]P and TCDD, are involved in the induction of CYP1A1 metabolizing them. Hence, PAHs become potential toxicants and chemical carcinogens with a concomitant formation of epoxide by epoxide hydroxylase. Thus, inhibition of CYP1A1 expression might protect the cells from the genotoxic action of CYP1A1-activated PAHs. It has been reported that CYP1A1 is down-regulated by TGF-b. However, the molecular mechanism by which TGF-b interferes in the transcription of the CYP1A1 gene is not fully understood except for reports that TGF-b perturbs the expression of AhR mRNA (26,45). In this study, we hypothesized that TGF-b signaling possesses other inhibitory machinery to regulate transcription of the CYP1A1 gene besides repression of AhR expression. Interestingly, the inhibitory action of TGF-b depended on the presence of XREs in the CYP1A1 promoter because the luciferase activity of CYP1A1(-313)-luc, which has no XREs in its promoter, was not affected by TGF-b. To further confirm that TGF-b inhibits the transcription of the CYP1A1 promoter, we assessed the luciferase reporters in which the XREs are multimerized. Expectedly, TGF-b hampered B[a]P-induced luciferase reporters, including the XRE in the proximal region of the minimal promoter. Curiously, there were no inhibitory actions of TGF-b on B[a]Pinduced activity of pGL3ti-(XRE) 2 -luc in Smad3-deficient MEFs. On the other hand, the introduction of either Smad2 or Smad3 in Smad3-deficient MEFs improved the responsiveness to TGF-b, although the introduction of Smad2 had a smaller effect than that of Smad3. This finding triggered the idea that Smad3 might interfere in the B[a]P-mediated interaction between AhR and ARNT. As expected, among the examined Smads, Smad3 associated with both AhR and ARNT most strongly, although Smad2 and Smad4 could interact with them as well. In particular, the MH2 domain of Smad3 played a crucial role in the dissociation of AhR and ARNT (Fig. 6D). Consistently, AhR and ARNT lost the ability to bind to the XREs when the MH2 domain of Smad3 was overexpressed (Fig. 7, A  and B). This event might occur because heterodimerization between AhR and ARNT is necessary for the AhR/ARNT Figure 5. Interaction of Smads with AhR or ARNT. A, interaction of AhR with Smads. COS7 cells were transfected with the indicated plasmids and harvested for the co-IP experiments. The interaction between Smads and AhR is shown in the top panel. The total expressions of AhR and FLAG-Smads are indicated in the middle and bottom panels, respectively. The intensity of the band for immunoprecipitated AhR was normalized using the intensity of the band corresponding to total AhR. Relative intensity was calculated relative to the value for V5-AhR plus FLAG-Smad2. B, interaction of ARNT with Smads. COS7 cells were transfected with the indicated plasmids and harvested for the co-IP experiments. The interaction between Smads and ARNT is shown in the top panel. The total expressions of Myc-ARNT and FLAG-Smads are indicated in the middle and bottom panels, respectively. The intensity of the band for immunoprecipitated ARNT was normalized using the intensity of the band corresponding to total ARNT. Relative intensity was calculated relative to the value for Myc-ARNT plus FLAG-Smad2. C, endogenous interaction between Smad3 and AhR. A549 cells were stimulated with 5 ng/ml TGF-b for 2 h and harvested for the co-IP experiments. The anti-Myc9E10 antibody was used as the control antibody for immunoprecipitation. The interaction between Smad3 and AhR is shown in the top panel. The total expressions of AhR, Smad3, phosphorylated Smad2 (PS2), and b-actin are indicated in the second, third, fourth, and bottom panels, respectively. The bands corresponding to phosphorylated Smad2 indicate that the cells were responsive to TGF-b. The intensity of the band for immunoprecipitated AhR was normalized using the intensity of the band corresponding to total AhR. Relative intensity was calculated relative to the value for the band immunoprecipitated with anti-Smad3 antibody in the absence of TGF-b stimulation. D and E, endogenous interaction between Smad3 and ARNT. A549 cells were stimulated with 5 ng/ml TGF-b for 2 h. Then PLA was carried out. The red dots show the interaction between Smad3 and ARNT. The nuclei were stained with DAPI (D). As the negative control, anti-Myc9E10 antibody instead of anti-Smad3 antibody was used. No red dots were observed when the cells were stimulated with 5 ng/ ml TGF-b for 2 h (E). Bar, 10 mm. ND, not determined. complex to bind to the XREs (3). These lines of evidence might support the notion that the MH2 domain of Smad2 is also required for Smad2 to suppress the B[a]P-induced CYP1A1 expression by TGF-b because the MH2 domains between Smad2 and Smad3 are highly homologous.
Because HIF-1a, which has been reported to be induced by TGF-b, interacts with ARNT (46), we hypothesized that HIF-1a prompts dissociation between AhR and ARNT to perturb B[a]P-mediated CYP1A1 transcription upon TGF-b stimulation. We could detect obvious prolongation of the t ½ of HIF-1a protein when cells were stimulated with TGF-b. Furthermore, caHIF-1a inhibited B[a]P-induced reporter activities of both pGL3ti-(XRE) 2 -luc and Gal4-TATA-luc. Indeed, HIF-1a dissociated the complex formation between AhR and ARNT upon TGF-b stimulation, suggesting that stabilized HIF-1a might interfere with B[a]P-mediated association between AhR and ARNT to suppress the transcription of the CYP1A1 gene. Although McMahon et al. (39) revealed that TGF-b inhibits the expression of PHD2 to increase HIF-1a protein, we could not see any differences in mRNA expression among PHD1, PHD2, and PHD3 when cells were stimulated with TGF-b (data not shown). Thus, an unknown mechanism(s) other than decreased expression of PHD2 might lead HIF-1a to be stabilized in the presence of TGF-b. We plan to elucidate this unknown mechanism in the future. CYP1A1 can oxidatively metabolize PAHs including B[a]P to convert them into mutagens that covalently bind to DNAs (3). Thus, we postulated that TGF-b suppresses B[a]P-induced genetic instability because of a diminishment of CYP1A1 expression. Using the expression of g-H2AX as a biomarker for DNA double-strand breaks (47), we found that TGF-b obviously blocked B[a]P-induced genotoxicity. In the present study, we have not measured metabolites of B[a]P in cells using HPLC because we could evaluate TGF-b-mediated inhibition of B[a]Pinduced genotoxicity by detection of g-H2AX protein. To further investigate the effect of TGF-b on inhibition of CYP1A1 expression, the detection of metabolites for B[a]P using HPLC will be required.
Müller et al. (48) could also observe the inhibition of CYP1A1 enzymatic activity when cells were stimulated with TGF-b, although they did not know the exact mechanism by which TGF-b inhibited the expression of B[a]P-induced CYP1A1 transcription. Therefore, our current study proposes a novel mechanism in which TGF-b/Smad3 signaling hampers the B[a]P-induced transcription of the CYP1A1 gene. Additionally, TGF-b marginally maintained the cytoplasmic retention of AhR in the presence of B[a]P (data not shown), which was consistent with the results from Staršíchová et al. (27).
AhR has been reported to potentiate the TGF-b signaling pathway in glioma to promote its tumorigenicity (49,50). Therefore, B[a]P-induced AhR activation might affect the TGF-b/Smad signaling pathway, although we did not examine the effect of the AhR/ARNT complex on TGF-b/Smad3 signaling. Thus, we must also investigate the involved intricate signaling circuit in the future.
In conclusion, the TGF-b/Smad3 pathway promotes dissociation between AhR and ARNT even in the presence of B[a]P. Additionally, HIF-1a stabilized by TGF-b competes with AhR for binding to ARNT. Consequently, the metabolic activation of B[a]P by CYP1A1 might be lowered owing to decreased expression of CYP1A1 (Fig. 10).

Experimental procedures
Antibodies Antibodies were obtained from the following sources: anti-FLAG M5 mouse mAb (F4042-1MG) from Sigma-Aldrich; anti-DYKDDDDK tag mAb (040-30953) corresponding to anti-FLAG antibody from FUJIFILM Wako; anti-Myc9E10 mAb (sc-40) from Santa Cruz Biotechnology, Inc.; anti-Myc tag rabbit polyclonal antibody (pAb) (06-549) from Merck; anti-V5 tag mouse mAb (011-23594) from FUJIFILM Wako; anti-b-actin mouse mAb (AC-15) (sc-69879) from Santa Cruz Biotechnology; anti-AhR rabbit pAb (ABS522) from Merck; anti-ALK5 (V-22) rabbit pAb (sc-398) from Santa Cruz Biotechnology; anti-ARNT1 mouse mAb (sc-17812) from Santa Cruz Biotechnology; anti-HIF-1a rabbit mAb (ab51608) from Abcam; anti-phosphohistone H2A.X (Ser-139) mouse mAb (05-636) corresponding to anti-g-H2AX antibody from Merck; anti-Smad3 rabbit mAb (04-1035) from Merck; anti-phospho-Smad3 (Ser-423/425) (C25A9) rabbit mAb (9520T) (PS3) from Cell Signaling Technology; anti-mouse IgG HRP-linked sheep pAb (NA931-1ML) from GE Healthcare; and anti-rabbit IgG HRP-linked donkey F(ab) 2 fragment (NA9340-1ML) from GE Healthcare. Antiphosphorylated Smad2 rabbit pAb (PS2) antibody was homemade (51). Figure 6. Disruption of the AhR/ARNT complex by TGF-b. A, dissociation between AhR and ARNT upon TGF-b stimulation. 911 cells were transfected with the indicated plasmids. Forty hours later, the cells were pretreated with 5 ng/ml TGF-b 4 h before exposure to 1 mM B[a]P for 2 h. Then the cells were harvested for the co-IP experiments using an anti-V5 antibody. The heterodimer formation between AhR and ARNT is shown in the top panel. The total expressions of ARNT, AhR, Smad3, and phosphorylated Smad3 are indicated in the second, third, fourth, and bottom panels, respectively. The band corresponding to phosphorylated Smad3 indicates that the cells were responsive to TGF-b. The intensity of the band for immunoprecipitated ARNT was normalized using the intensity of the band corresponding to total ARNT. Relative intensity was calculated relative to the value for Myc-ARNT plus V5-AhR in the presence of B[a]P. B, determination of the interacting domain of Smad3 with AhR. COS7 cells were transfected with the indicated plasmids. Then the cells were harvested for the co-IP experiments using an anti-V5 antibody. The interaction of Smad3 mutants with AhR is shown in the top panel. The total expressions of 6xMyc-Smad3 mutants and V5-AhR are indicated in the middle and bottom panels, respectively. The intensity of the band for each Smad3 mutant precipitated with AhR was normalized using the intensity of the band corresponding to each of the total Smad3 mutants. Relative intensity was calculated relative to the value for each Smad3 mutant alone. C, determination of the interacting domain of Smad3 with ARNT. COS7 cells were transfected with the indicated plasmids. Then the cells were harvested for the co-IP experiments using an anti-V5 antibody. The interaction of Smad3 mutants with ARNT is shown in the top panel. The total expressions of 6xMyc-Smad3 mutants and V5-ARNT are indicated in the middle and bottom panels, respectively. The intensity of the band for each Smad3 mutant precipitated with ARNT was normalized using the intensity of the band corresponding to each of the total Smad3 mutants. Relative intensity was calculated relative to the value for each Smad3 mutant alone. D, dissociation of the AhR/ARNT complex by the Smad3 MH2 domain. 911 cells were transfected with the indicated plasmids. Then the cells were harvested for the co-IP experiments using an anti-V5 antibody. The interaction of V5-AhR with Myc-ARNT is shown in the top panel. The total expression of V5-AhR was detected with an anti-V5 antibody after immunoprecipitation using total lysates by an anti-V5 antibody (second panel). The total expressions of Myc-ARNT and 6xMyc-Smad3 MH2 are indicated in the third and bottom panels, respectively. The intensity of the band for immunoprecipitated ARNT was normalized using the intensity of the band corresponding to total ARNT. Relative intensity was calculated relative to the value for Myc-ARNT plus V5-AhR.
Cell culture 911, 293A, A549, and COS7 cells were cultured in Dulbecco's modified Eagle's medium (DMEM; Nacalai Tesque) containing 10% fetal calf serum (FCS; Invitrogen). MEFs established from Smad3-HET and Smad3-KO mice (29,30) were cultured in DMEM containing 10% FCS. Primary hepatocytes from WT and Smad3-KO mice were prepared and cultured according to the methods described by Nemoto et al. (52) and Sakuma et al. (53). Then the hepatocytes were cultured in collagen-coated dishes. All animal work was performed with institutional approval by the Animal Care and Use Committee of Showa Pharmaceutical University (approval P-2018-01-R1, dated April 19, 2018).
Genome editing using the CRISPR/Cas9 system A549 cells were cloned and used as a parental cell line (A549-P1), which was authenticated by short tandem repeat analysis. Knockout of SMAD3 was conducted using Double Nickase plasmid (catalog no. sc-4000069-NIC-2; Santa Cruz Biotechnology). Briefly, cells were grown until 70% confluence on a 6well plate and then transfected with 1 mg of plasmid DNA using Lipofectamine 2000 Transfection Reagent (Thermo Fisher Scientific). After 24 h of incubation, the cells were treated with 1 mg/ml puromycin (Sigma-Aldrich) to eliminate nontransfected cells within 48 h. Deletion/disruption of the target genes was confirmed by sequencing (Fig. S2). There are three SMAD3 loci in A549 cells.

Transcriptional reporter assays
One day before transfection, A549 cells and MEFs were seeded at 1.2 3 10 5 and 1.0 3 10 5 cells/well in 12-and 6-well plates, respectively. The cells were transfected with a reporter gene, pCH110 (GE Healthcare), and the indicated plasmids by use of polyethyleneimine (PEI; Polysciences). When 5 ng/ml TGF-b1 (Peprotech) was added to the wells 24 h after transfection, the cells were cultured in the absence of FCS for 18 h. In all the experiments, b-gal activity was measured to normalize for transfection efficiency. Each transfection was carried out in triplicate and repeated at least twice.

Immunoprecipitation and Western blotting analysis
To detect interactions among the proteins, plasmids were transfected into COS7 cells (5 3 10 5 cells/6-cm dish), 911 cells (1.5 3 10 6 cells/10-cm dish), or 293T cells (3.0 3 10 6 cells/10-cm dish) by use of PEI. Forty hours after the transfection, the cells were lysed in 500 ml of TNE buffer (10 mM Tris (pH 7.4), 150 mM NaCl, 1 mM EDTA, 1% Nonidet P-40, 1 mM phenylmethylsulfonyl fluoride, 5 mg/ml leupeptin, 100 units/ml aprotinin, 2 mM sodium vanadate, 40 mM NaF, and 20 mM b-glycerophosphate). The cell lysates were precleared with protein G-Sepharose beads (GE Healthcare) for 30 min at 4°C and then incubated with the indicated antibodies for 2 h at 4°C. The protein complexes were immunoprecipitated by incubation with protein G-Sepharose beads for 30 min at 4°C and then washed three times with TNE buffer. The immunoprecipitated proteins and aliquots of the total cell lysates were boiled for 5 min in sample buffer, separated by SDS-PAGE, and transferred to UltraCruz Nitrocellulose Pure Transfer membranes (Santa Cruz Biotechnology). The membranes were probed with the primary antibodies, which were detected with horseradish peroxidase-conjugated secondary antibodies and a chemiluminescent substrate (Western BLoT Quant HRP Substrate, Takara). The protein expression in the total cell lysates was evaluated by Western blotting analysis.

PLA
A549 cells on cover glasses coated with 0.1% gelatin (Sigma-Aldrich) were cultured with DMEM. After TGF-b stimulation, the cover glasses were washed once with PBS, fixed for 10 min with 4% paraformaldehyde, washed three times with PBS, permeabilized with 0.5% Triton X-100 in PBS for 10 min, and again washed three times with PBS. The following procedures were performed according to the manufacturer's instructions (Olink Bioscience). To visualize the fluorescence, a confocal microscope (Nikon, A1R) was used.

RNA preparation and quantitative PCR (qPCR) analysis
Total RNAs from A549 cells and hepatocytes from WT or Smad3-KO mice were extracted using a ReliaPrep RNA Cell Miniprep System (Promega). Reverse transcription was performed with a High-Capacity RNA-to-cDNA kit (Thermo Fisher Scientific). qPCR was performed using a KAPA SYBR Fast qPCR kit (KAPA). All reactions were carried out on a Thermal Cycler Dice (TAKARA). Each sample was analyzed in triplicate at least twice for each PCR measurement. The melting curves were checked to ensure specificity. Relative quantification of mRNA expression was calculated using the standard curve method with the b-actin level. Before qPCR, we used agarose gel electrophoresis to confirm that the DNA fragment amplified using each primer set was a single band with the correct size. The primer sets are listed in Table 1.
DNA affinity precipitation 911 cells were seeded at 1.5 3 10 6 cells/well in a 10-cm dish 1 day before transfection. The cells were transfected with the indicated plasmids by use of PEI. Forty hours after the transfection, the cells were pretreated with 50 mM MG-132 (FUJIFILM Wako) 2 h before stimulation of the cells with 1 mM B[a]P (Tokyo Chemical Industry) for 2 h. Then the cells were lysed in 1 ml of TNE buffer. The cell lysates were precleared with 12 mg/ ml poly(dI·dC) (GE Healthcare) and streptavidin agarose (Sigma-Aldrich) for 30 min and incubated with 24 mM biotinylated (XRE) 2 for 2 h at 4°C. Subsequently, streptavidin agarose was added to the reaction mixture and incubated for 30 min at 4°C. After the precipitates were washed with TNE buffer three times, the precipitates and aliquots of the total lysates were separated by SDS-PAGE. Then the proteins were transferred to the membrane for Western blotting analysis. The membrane was incubated with the indicated primary antibodies. The primary antibodies were detected as described above. The sequences of biotinylated (XRE) 2 are as follows: 59-biotinated-CTCTCACGCGAGCCGGGACTCTCACGCGAGCCG-GGACG-39/3'-CATGGAGAGTGCGCTCGGCCCTGAGA-GTGCGCTCGGCCCTGCGATC-59 (the underlined letters The intensity of the band for HIF-1a was normalized using the intensity of the band corresponding to b-actin. Relative intensity was calculated relative to the value for FLAG-HIF-1a without treatment with CHX. B, inhibition of B[a]P-induced reporter activity by HIF-1a. A549 cells were transfected with the indicated plasmids. Twenty-four hours later, the cells were exposed to 1 mM B[a]P for 18 h. Significant differences from the control with B[a]P in either the absence or the presence of AhR are indicated with asterisks. C, inhibition of B[a]P-induced reporter activity by HIF-1a using a Gal4 luciferase reporter. The expression plasmid for the fusion protein between Gal4 DBD and AhR, ARNT, Gal4-TATA-luc, and pCH110 was transfected with or without caHIF-1a in A549 cells. Twenty-four hours later, the cells were stimulated with or without 1 mM B[a]P for 18 h. Significant differences from the control without HIF-1a in either the absence or the presence of B[a]P are indicated with asterisks. D, expression of HIF-1a in A549 cells carrying the knocked-down vectors for HIF-1a. To detect endogenous HIF-1a protein, cells were cultured under hypoxia 28 h before cell lysis. The top and bottom panels indicate the expressions of HIF-1a and b-actin, respectively. A549-shCont, A549 cells carrying pLKO.1 alone; A549-shHIF-1a-2, A549 cells carrying pLKO.1-HIF-1a-sh2; A549-shHIF-1a-3, A549 cells carrying pLKO.1-HIF-1a-sh3. The intensity of the band for HIF-1a was normalized using the intensity of the band corresponding to b-actin. Relative intensity was calculated relative to the value for HIF-1a under hypoxia in A549-shCont. E, partial recovery of B[a]P-induced reporter activity in the presence of TGF-b. HIF-1aknocked-down cells were transfected with pGL3ti-(XRE) 2 -luc and CH110. Twenty-four hours later, the cells were exposed to 1 mM B[a]P for 6 h after a 4-h treatment with 5 ng/ml TGF-b. Significant differences between A549-shCont and either A549-shHIF-1a-2 or A549-shHIF-1a-3 are indicated with asterisks. F, HIF-1a-mediated dissociation between AhR and ARNT. 293T cells were transfected with the indicated plasmids. The interaction between AhR and ARNT is shown in the top panel. The total expressions of Myc-ARNT, V5-AhR, and FLAG-HIF-1a are indicated in the second, third, and bottom panels, respectively. The intensity of the band for ARNT precipitated with AhR was normalized using the intensity of the band corresponding to total ARNT. Relative intensity was calculated relative to the value for Myc-ARNT plus V5-AhR. ND, not determined.
indicate the XREs). When the endogenous proteins were used for DNA affinity precipitation, A549 cells were prestimulated with 5 ng/ml TGF-b 2 h before treatment with 1 mM B[a]P for 2 h.

Measurement of protein t ½
FLAG-HIF-1a was transfected into 911 cells by use of PEI. After 40 h, the cells were pretreated with 50 ng/ml cycloheximide (Sigma-Aldrich) 30 min before the addition of 5 ng/ml TGF-b for the indicated times. Then Western blotting analysis was performed as described above.
g-H2AX staining A549 cells were seeded at 3 3 10 4 cells/12-well plate on glass cover slides coated with 0.1% gelatin (in PBS). The cells were then pretreated with 5 ng/ml TGF-b 4 h before stimulation of the cells with 39.6 mM B[a]P for 24 h. Then the cover glasses were washed once with PBS, fixed for 10 min with 4% paraformaldehyde, washed three times with PBS, permeabilized with 0.5% Triton X-100 in PBS for 10 min, and again washed three times with PBS. The cover glasses were blocked with 5% BSA in PBS at 37°C for 1 h and incubated with 5% BSA (in PBS) containing g-H2AX antibody at 4°C overnight. The cover glasses were then washed three times with PBS, incubated with 5% BSA (in PBS) containing Alexa Fluor 555-conjugated goat anti-mouse IgG antibody (diluted 1:250) (A-21422, Thermo Fisher Scientific) at room temperature for 1 h, and then washed three times with PBS. The nuclei were stained with 49,6-amino-2-phenylidole (DAPI). A confocal microscope was used to visualize the fluorescence.

Data availability
All data generated during this study except for data described as "data not shown" are included in this article. The data described as "data not shown" can be obtained from the corresponding author (Susumu Itoh, sitoh@ac.shoyaku.ac.jp) upon request.  [a]P binds to AhR, the complex between B[a]P and AhR enters the nucleus, where AhR interacts with ARNT. Then AhR/ARNT complex recognizes XREs on the CYP1A1 promoter to activate the CYP1A1 gene. Consequently, the CYP1A1 protein metabolically activates B[a]P to make it an ultimate carcinogen. With TGF-b, activated Smad3 (and maybe activated Smad2) upon TGF-b stimulation interacts with AhR and/or ARNT to interfere in the heteromeric complex between AhR and ARNT. Additionally, HIF-1a stabilized by TGF-b promotes dissociation between the AhR and ARNT complex. Therefore, enough of the AhR/ARNT complex cannot bind to XREs in the CYP1A1 promoter even when exposed to B[a]P. Thus, TGF-b signaling might be able to protect cells from increased carcinogenic risk via metabolic activation of B[a]P by CYP1A1.