Possible Involvement of the Upregulation of ΔNp63 Expression Mediated by HER2-Activated Aryl Hydrocarbon Receptor in Mammosphere Maintenance

Cancer stem cells (CSCs) contribute to the drug resistance, recurrence, and metastasis of breast cancers. Recently, we demonstrated that HER2 overexpression increases mammosphere formation via the activation of aryl hydrocarbon receptor (AHR). In this study, the objective was to identify the mechanism underlying mammosphere maintenance mediated by HER2 signaling-activated AHR. We compared the chromatin structure of AHR-knockout (AHRKO) HER2-overexpressing MCF-7 (HER2-5) cells with that of wild-type HER2-5 cells; subsequently, we identified TP63, a stemness factor, as a potential target gene of AHR. ΔNp63 mRNA and protein levels were higher in HER2-5 cells than in HER2-5/AHRKO cells. Activation of HER2/HER3 signaling by heregulin treatment increased ΔNp63 mRNA levels, and its induction was decreased by AHR knockdown in HER2-5 cells. The results of the chromatin immunoprecipitation assay revealed an interaction between AHR and the intronic region of TP63, which encodes ΔNp63. A luciferase reporter gene assay with the intronic region of TP63 showed that AHR expression increased reporter activity. Collectively, our findings suggest that HER2-activated AHR upregulates ΔNp63 expression and that this signaling cascade is involved in CSC maintenance in HER2-expressing breast cancers.


Introduction
Breast cancer (BC) is the most commonly diagnosed neoplasm and the leading cause of cancer-related death in women worldwide [1,2]. The cancer stem cell (CSC) theory provides new insights into cancer therapies [3,4]. CSCs account for a relatively small population of tumor cells that are mainly identified by their properties of self-renewal, slow cell division, and tumor-and metastasis-initiating capacity [5,6]. Consequently, CSCs contribute to the drug resistance, recurrence, and metastasis of BCs. CSCs are commonly identified as a CD44 high /CD24 low or aldehyde dehydrogenase-high population or a population with tumorsphere formation capacity [7]. Although the mechanisms underlying the self-renewal and tumor-initiation capacities of breast CSCs remain unclear, CSCs have been suggested as potential targets for radical cancer treatment.
HER2 belongs to the HER family of receptor tyrosine kinases, including HER1, HER3, and HER4 [8]. Upon ligand-HER binding via the extracellular ligand-binding domain of HER, the receptor undergoes homodimerization or heterodimerization, which activates its downstream signaling pathways, such as the mitogen-activated protein kinase pathway and phosphoinositide-3-kinase/Akt signaling pathway [8]. Although ligands for HER2 are not known, HER2 activation via heterodimerization with other HER receptors and/or

Knockout of AHR Downregulates Mammosphere Formation and ∆Np63 Expression in HER2-Overexpressing Breast Cancer Cells
First, to investigate the role of AHR in the function of CSCs from HER2-overexpressing breast cancer cells, we compared the mammosphere formation efficiency of HER2-5 cells with that of HER2-5/AHRKO cells. HER2-5 is a previously established HER2-expressing MCF-7-derived stable cell line, and HER2-5/AHRKO is a HER2-5-derived cell line in which AHR has been knocked out using the CRISPR/Cas9 technique [17,18]. As expected, AHR was not observed in HER2-5/AHRKO cells ( Figure 1A). AHR is highly expressed and constitutively activated in HER2-5 cells compared to levels in MCF-7 cells [18]. We previously showed that the mammosphere formation efficiency of HER2-5 cells is better than that of MCF-7 cells. As shown in Figure 1B, the mammosphere formation efficiency of HER2-5/AHRKO cells was worse than that of HER2-5 cells. These observations indicate that AHR is involved in mammosphere formation in HER2-overexpressing BC cells, as previously  (10 µg) were resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, and proteins were detected by immunoblo analysis using antibodies against AHR and α-tubulin. (B) HER2-5 cells and HER2-5/AHRKO cells were cultured under non-adherent conditions for 7 days. The mammospheres were observed unde a light microscope with 2× magnification (scale bar is 2000 µm) and 20× magnification (scale bar is 200 µm). The percentage of cells contributing to mammosphere formation was calculated as the number of spheres divided by the original number of seeded cells [22]. Each column represents the mean of six wells. The depicted results are representative of two independent biological experi ments. **, p < 0.01 (Student's t test). (C) ATAC-seq was performed using HER2-5 cells and HER2 5/AHRKO cells. The overlap peaks are presented in a VENN diagram. The diagram shows the num ber of merged regions that contain peaks from the samples in each of the overlapping categories (D) Upregulated and downregulated GO terms for biological processes enriched among differen tially expressed genes were annotated using the DAVID tool.
ΔNp63 is known as a master regulator of HER2-subtype breast CSCs [19,20]. Since "downregulated" peak areas corresponding to the TP63 gene were observed based on the results shown in Figure 1C, we targeted the ΔNp63-encoding gene in this study. First, we compared the ΔNp63 expression level in HER2-5 cells with that in HER2-5/AHRKO cells As expected, ∆Np63 mRNA levels were higher in HER2-5 cells than in HER2-5/AHRKO cells ( Figure 2A). Moreover, ΔNp63 protein levels were higher in HER2-5 cells than in HER2-5/AHRKO cells ( Figure 2B). These results suggest that AHR upregulates ∆Np63 expression in HER2-overexpressing BC cells.  (10 µg) were resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, and proteins were detected by immunoblot analysis using antibodies against AHR and α-tubulin. (B) HER2-5 cells and HER2-5/AHRKO cells were cultured under non-adherent conditions for 7 days. The mammospheres were observed under a light microscope with 2× magnification (scale bar is 2000 µm) and 20× magnification (scale bar is 200 µm). The percentage of cells contributing to mammosphere formation was calculated as the number of spheres divided by the original number of seeded cells [22]. Each column represents the mean of six wells. The depicted results are representative of two independent biological experiments. ** p < 0.01 (Student's t test). (C) ATAC-seq was performed using HER2-5 cells and HER2-5/AHRKO cells. The overlap peaks are presented in a VENN diagram. The diagram shows the number of merged regions that contain peaks from the samples in each of the overlapping categories. (D) Upregulated and downregulated GO terms for biological processes enriched among differentially expressed genes were annotated using the DAVID tool.
Next, we investigated whether AHR engages in the chromatin relaxation of gene enhancers associated with the self-renewal of HER2-5 cells. We compared chromatin structures in HER2-5 and HER2-5/AHRKO cells by performing an assay for transposaseaccessible chromatin using sequencing (ATAC-seq). ATAC-seq revealed 80,077 and 57,612 enriched peaks (i.e., open chromatin regions) in HER2-5 and HER2-5/AHRKO cells, respectively, and the chromatin positions of 50,265 peaks for HER2-5 cells matched those in HER2-5/AHRKO cells ( Figure 1C); thus, unmatched peaks were changed by knocking out AHR in HER2-5 cells. These results indicated that HER2-5 cells had more regions of increased chromatin accessibility than HER2-5/AHRKO cells, suggesting that AHR engages in chromatin relaxation in HER2-5 cells.
The peak sizes of the matching regions between these cells were then compared using the DESeq2 algorithm. The numbers of regions with increased ("upregulated") and decreased ("downregulated") peak areas after AHR knockout were 1519 and 8593, respectively. We annotated the genes close to the differentially regulated peaks using DA-VID annotation databases with criteria of 1.5-fold for upregulation and 0.5-fold for downregulation ( Figure 1D). The term "pathways in cancer" was most frequently enriched for downregulated chromatin regions. Pathways related to xenobiotic metabolism, which is a classical function of AHR, were enriched with both upregulated and downregulated peaks.
∆Np63 is known as a master regulator of HER2-subtype breast CSCs [19,20]. Since "downregulated" peak areas corresponding to the TP63 gene were observed based on the results shown in Figure 1C, we targeted the ∆Np63-encoding gene in this study. First, we compared the ∆Np63 expression level in HER2-5 cells with that in HER2-5/AHRKO cells. As expected, ∆Np63 mRNA levels were higher in HER2-5 cells than in HER2-5/AHRKO cells ( Figure 2A). Moreover, ∆Np63 protein levels were higher in HER2-5 cells than in HER2-5/AHRKO cells ( Figure 2B). These results suggest that AHR upregulates ∆Np63 expression in HER2-overexpressing BC cells. (B) Whole-cell lysates (10 µg) were resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, and proteins were detected by immunoblot analysis using antibodies against ΔNp63 and α-tubulin. The intensity of ΔNp63 and α-tubulin protein bands was measured using ImageJ software. ΔNp63 levels were normalized to those of α-tubulin and are expressed as relative values (those in HER2-5 cells were set at 1) based on the mean ± S.D. (n = 3). Each dot represents independent biological data points of triplicate measurements. *, p < 0.05 (Student's t test).
HER2-5 cells were cultured in low-serum medium (2% charcoal-stripped fetal bovine serum, 2% csFBS). Under this condition, phosphorylation of the HER2/HER3 downstream proteins ERK1/2 and Akt was not detected by Western blotting; however, phosphorylation was detected after stimulation with HRG (50 ng/mL) for 15 min ( Supplementary Figure S1A). These observations suggest that HER2 activation does not occur under the lowserum conditions. Next, the cells cultured in low-serum medium were treated with HRG for 3, 6, 12, or 24 h. ∆Np63 mRNA levels increased following HRG treatment, with the highest level at 3 h ( Figure 3A). Furthermore, ∆Np63 protein levels also increased following HRG treatment ( Figure 3B). Subsequently, we knocked down AHR using siRNA to investigate the role of AHR in the HRG-induced increase in ∆Np63 mRNA levels. As shown in Figure 3C, the HRG-mediated increase in ∆Np63 mRNA levels was suppressed by AHR knockdown. Similarly, the HRG-mediated increase in ∆Np63 mRNA levels was Values are expressed as relative values (those in HER2-5 cells were set at 1) and the mean ± S.D. (n = 4). ** p < 0.01 (Student's t test). Each dot represents individual data points of quadruplicate measurements. The depicted results are representative of two independent biological experiments. (B) Whole-cell lysates (10 µg) were resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, and proteins were detected by immunoblot analysis using antibodies against ∆Np63 and α-tubulin. The intensity of ∆Np63 and α-tubulin protein bands was measured using ImageJ software. ∆Np63 levels were normalized to those of α-tubulin and are expressed as relative values (those in HER2-5 cells were set at 1) based on the mean ± S.D. (n = 3). Each dot represents independent biological data points of triplicate measurements. * p < 0.05 (Student's t test).
HER2-5 cells were cultured in low-serum medium (2% charcoal-stripped fetal bovine serum, 2% csFBS). Under this condition, phosphorylation of the HER2/HER3 downstream proteins ERK1/2 and Akt was not detected by Western blotting; however, phosphorylation was detected after stimulation with HRG (50 ng/mL) for 15 min (Supplementary Figure S1A). These observations suggest that HER2 activation does not occur under the low-serum conditions. Next, the cells cultured in low-serum medium were treated with HRG for 3, 6, 12, or 24 h. ∆Np63 mRNA levels increased following HRG treatment, with the highest level at 3 h ( Figure 3A). Furthermore, ∆Np63 protein levels also increased following HRG treatment ( Figure 3B). Subsequently, we knocked down AHR using siRNA to investigate the role of AHR in the HRG-induced increase in ∆Np63 mRNA levels. As shown in Figure 3C, the HRG-mediated increase in ∆Np63 mRNA levels was suppressed by AHR knockdown. Similarly, the HRG-mediated increase in ∆Np63 mRNA levels was suppressed by AHR antagonist treatment (Supplementary Figure S2A). These results suggest that AHR participates in the HER2/HER3-mediated expression of ∆Np63. However, the HRG-mediated increase in ∆Np63 mRNA levels was not completely suppressed in HER2-5/AHRKO cells (Supplementary Figure S2B). Previously, we reported that the signaling activation of AHR via HER2 is mediated by the MEK cascade. To identify the mechanism underlying ∆Np63 mRNA expression induced by HER2 signaling, HER2-5 cells were pretreated with specific inhibitors of MEK and AKT, namely PD0325901 and MK-2206, respectively (Supplementary Figure S1B). Following HRG treatment, HRG-induced mRNA expression of ∆Np63 was completely inhibited by pre-treatment with PD0325901, but not MK-2206 ( Figure 3D). These results suggest that ∆Np63 induction via HER2 is regulated by AHR and other molecules through MEK.

HRG-Activated AHR Binds to the Enhancer Region of TP63
We next investigated the mechanism by which AHR increased ∆Np63 expression. Figure 4A shows the tracks of ATAC-seq analyses of HER2-5 and HER2-5/AHRKO cells. TP63 encodes two major isoforms of p63, TAp63 and ∆Np63. ∆Np63 regulates CSC selfrenewal in HER2-overexpressing BC cells [19,20]. Therefore, we investigated whether AHR regulates ∆Np63 expression. First, to investigate the recruitment of AHR to the promoter region of TP63, we performed chromatin immunoprecipitation (ChIP) analysis for regions R1, R2, R3, and R4 ( Figure 4A), which contained differentially enriched peaks between HER2-5 and HER2-5/AHRKO cells based on the ATAC-seq analysis. R1 and R3 are regions distal and proximal to the transcription start site of TAp63-and ∆Np63-encoding mRNAs, respectively. R2 and R4 are the intronic regions of TP63.
The ChIP assay demonstrated that AHR was recruited to R4 in HER2-5 cells but not in negative control HER2-5/AHRKO cells. Additionally, AHR was not recruited to R1, R2, or R3 in either cell line ( Figure 4B). As expected, AHR was recruited to the CYP1B1 enhancer region only in HER2-5 cells ( Figure 4B).
Subsequently, we investigated whether AHR recruitment to R4 is induced by HER2 signaling in HER2-5 cells. Under low serum conditions, AHR did not bind to the R4 region as well as it did to the R3 region; however, HRG treatment for 3 h increased AHR recruitment to R4 but not to R3 ( Figure 4C). These results suggest that HER2 signaling promotes the recruitment of AHR to the R4 region of TP63.
Finally, we investigated whether AHR acts as a transcriptional activator by binding to the R4 region of TP63. We constructed a luciferase reporter plasmid containing the R4 region, pGL4.24-TP63 R4 ( Figure 4D). Our analysis using the JASPAR database [24] identified three putative AHREs in the R4 region; therefore, we also prepared a reporter plasmid, pGL4.24-TP63 R4 AHREm, with mutated AHREs ( Figure 4D).
HER2-5/AHRKO cells were transfected with pGL4.24-TP63 R4 and AHR and/or ARNT expression plasmids, and luciferase reporter activity was measured. As shown in Figure 4E, reporter activity was increased by the expression of AHR and ARNT. This increase was suppressed by treatment with the AHR antagonist StemRegenin 1 ( Figure 4F). Moreover, the increase in the luciferase reporter activity due to AHR and ARNT overexpression was not observed in the AHRE-mutated plasmid ( Figure 4G). These results suggest that the AHREs in the R4 enhancer region are responsible for AHR-mediated ∆Np63 expression in HER2-5 cells.
were pretreated with specific inhibitors of MEK and AKT, namely PD0325901 and MK-2206, respectively (Supplementary Figure S1B). Following HRG treatment, HRG-induced mRNA expression of ∆Np63 was completely inhibited by pre-treatment with PD0325901, but not MK-2206 ( Figure 3D). These results suggest that ∆Np63 induction via HER2 is regulated by AHR and other molecules through MEK. Figure 3. Upregulation of ∆Np63 expression via AHR mediated by HRG. (A) HER2-5 cells were cultured in medium with low serum (2% charcoal-stripped fetal bovine serum, csFBS DMEM) for 24 h and treated with HRG (50 ng/mL) for 3-24 h. Subsequently, the cells were harvested, and ∆Np63 mRNA levels were determined by RT-qPCR analysis and normalized to those of B2M mRNA. The relative mRNA levels are expressed as the average of the fold-induction relative to that in the control (0 h) (mean ± S.D., n = 4). **, p < 0.01; *, p < 0.05 (Dunnett's test vs. 0 h). The depicted results are representative of two independent biological experiments. (B) HER2-5 cells were cultured in medium with low serum (2% csFBS DMEM) for 24 h and treated with HRG (50 ng/mL) for 1-24 h. Subsequently, the cells were harvested, whole-cell lysates (10 µg) were resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, and proteins were detected by immunoblot analysis using antibodies against ΔNp63 and α-tubulin. (C) HER2-5 cells were transfected with siRNA targeting AHR (siAHR, 10 nM) or control siRNA (siNC, 10 nM). After 24 h, the culture medium containing 10% FBS was changed to medium containing low serum (2% csFBS), and the cells were incubated for 21 h. The cells were stimulated with HRG (50 ng/mL) or vehicle (water) for 3 h, and relative mRNA levels were determined (mean ± S.D., n = 3). **, p < 0.01 (Student's t test). The results are representative of two independent biological experiments. (D) HER2-5 cells were cultured in medium with low serum (2% csFBS DMEM) for 24 h and pre-treated with PD0325901 (1 Figure 3. Upregulation of ∆Np63 expression via AHR mediated by HRG. (A) HER2-5 cells were cultured in medium with low serum (2% charcoal-stripped fetal bovine serum, csFBS DMEM) for 24 h and treated with HRG (50 ng/mL) for 3-24 h. Subsequently, the cells were harvested, and ∆Np63 mRNA levels were determined by RT-qPCR analysis and normalized to those of B2M mRNA.
The relative mRNA levels are expressed as the average of the fold-induction relative to that in the control (0 h) (mean ± S.D., n = 4). ** p < 0.01; * p < 0.05 (Dunnett's test vs. 0 h). The depicted results are representative of two independent biological experiments. (B) HER2-5 cells were cultured in medium with low serum (2% csFBS DMEM) for 24 h and treated with HRG (50 ng/mL) for 1-24 h. Subsequently, the cells were harvested, whole-cell lysates (10 µg) were resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, and proteins were detected by immunoblot analysis using antibodies against ∆Np63 and α-tubulin. (C) HER2-5 cells were transfected with siRNA targeting AHR (siAHR, 10 nM) or control siRNA (siNC, 10 nM). After 24 h, the culture medium containing 10% FBS was changed to medium containing low serum (2% csFBS), and the cells were incubated for 21 h. The cells were stimulated with HRG (50 ng/mL) or vehicle (water) for 3 h, and relative mRNA levels were determined (mean ± S.D., n = 3). ** p < 0.

Discussion
AHR overexpression is observed in BC, and increased nuclear localization of AHR is positively correlated with poor prognosis [25][26][27][28]. AHR is reportedly overexpressed in BC cells relative to expression in normal breast tissue and has been negatively associated with the histological type and p53 protein expression levels [29]. Furthermore, increasing evidence has shown that the repression of AHR inhibits the proliferation, invasion, and migration of BC cells [30,31]. These reports suggest that constitutively activated AHR contributes to cancer progression. We previously reported that HER2 overexpression, which is associated with a high degree of malignancy and poor prognosis for BC, enhances the CSC properties of BC cells via AHR [17,18].
An analysis of the role of AHR in HER2-overexpressing BC cells by performing ATACseq using HER2-5 and HER2-5/AHRKO cells suggested that HER2-activated AHR upregulates the expression of genes annotated as "pathways in cancers" ( Figure 1D). AHR activation modulates the invasive properties of cancer cells. In melanoma cells, the activation of AHR promotes TNFα-dependent inflammation and metastasis [32]. Moreover, AHR activation by benzo[a]pyrene promotes cell migration, invasion, and epithelial-mesenchymal transition by upregulating the expression of long non-coding RNA in lung cancer [33].
In this study, we demonstrated that AHR binds to the R4 enhancer region of TP63 in HER2-5 cells (Figure 4B), in which AHR was constitutively activated and accumulated in the nucleus [17,18]. Moreover, we showed that HRG-activated AHR was recruited to the R4 region ( Figure 4C). Thus, growth factors such as HRG might activate AHR. Previously, we showed that mitogen-activated protein kinase/extracellular signal-regulated kinase kinase signaling enhances AHR expression and induces the accumulation of AHR in the nucleus [17]. These results suggest that HER2 signal-activated AHR binds to the R4 enhancer region of TP63 in HER2-5 cells.
Estrogen receptor α (ERα) is a critical transcriptional regulator in BC cells. ∆Np63 mRNA transcription is regulated by various promoter regions. ERα binds to the estrogen response element located between −2,858 and −2,839 bp at the translation start site within the ∆Np63 promoter and induces the transcription of ∆Np63 mRNA [34]. ERα is known to exhibit crosstalk with AHR via various mechanisms [35][36][37]. For example, ligand-activated AHR forms a complex with ERα on the estrogen response element and potentiates the transactivation function of 17β-estradiol-unbound ERα. However, AHR represses 17βestradiol-bound ERα [38]. Furthermore, AHR has ubiquitin ligase activity and pro-motes ERα degradation [36]. These facts suggest the involvement of ERα in the AHR-dependent expression of ∆Np63 via the R4 region. However, ERα was not found to bind to the R4 region based on in silico analysis using JASPAR (data not shown). In addition, mutations in the AHREs of the R4 region led to reduced AHR-dependent reporter gene expression ( Figure 4G). Therefore, these results suggest that HRG-mediated ∆Np63 expression is independent of AHR-ERα crosstalk and is mediated by the direct binding of AHR to the R4 region.
∆Np63 maintains the self-renewal capacity of mammary stem cells and the stemness of BC cells [19,39,40]. ∆Np63 enhances stemness through various stemness signaling pathways such as WNT, Hedgehog, and NOTCH. In addition, ∆Np63 directly controls the expression of stemness factors such as FZD7, SHH, GLI2, PTCH1, and NOTCH1 [20,21,41]. Based on these roles for ∆Np63, the upregulation of its expression via AHR could enhance CSC proliferation in BC cells. HER2 signaling can activate AHR and lead to constitutive ∆Np63 expression in HER2-overexpressing BC cells. This suggests that AHR controls mammosphere formation via ∆Np63 in HER2-overexpressing BC cells. As described previously herein, AHR expression might correlate with the poor prognosis and malignancy of BC. Because this study using HER2-overexpressing BC cells is limited, further study is required to determine whether AHR can increase ∆Np63 expression in BC cells of other subtypes.
AHR activation mediated by the endogenous ligand FICZ induces the expression of genes associated with migration, invasion, and stemness in triple-negative BC cells [42]. Furthermore, functional studies of AHR in CSCs have shown that it contributes to the chemoresistance of CSCs [43][44][45]. The expression of chemotherapy resistance-related transporter (such as ABCG2)-and enzyme (such as aldo-keto reductase 1C3 and ALDH1A1)-encoding genes, which are AHR-target genes, is increased in AHR-overexpressing CSCs [43][44][45]. Previous reports, including our studies, have established the anti-tumorigenic ability of agonist-activated AHR [46]. Agonist-mediated AHR activation represses mammosphere formation and reduces the ALDH-high population, suggesting the repressive effects of AHR on breast CSCs [47][48][49][50]. Moreover, we have previously reported the suppressive effect of AHR (depending on the agonist used) on mammosphere formation in BC cells [22,51]. These results suggest that AHR has both pro-and anti-tumorigenic activities in cancers and that an agonist can determine its activity. Thus, understanding how the ligand selectivity of AHR contributes to these pro-and anti-tumorigenic functions might provide information on AHR as a novel therapeutic target for cancer. In summary, the present results suggest that HER2 signaling-activated AHR maintains mammosphere formation by breast CSCs by inducing ∆Np63 expression. Therefore, AHR could be a potential therapeutic target for BC stem cell therapy.

Chemicals
StemRegenin 1 was purchased from Selleck Chemicals (Houston, TX, USA). Dimethyl sulfoxide was purchased from Wako Pure Chemical Industries (Osaka, Japan). HRG (EGF Domain) was purchased from Sigma-Aldrich (St. Louis, MO, USA) and dissolved in water.
For the knockdown experiment, HER2-5 cells were transfected with AHR siRNA (Stealth™ RNAi, AHR-HSS 100338; Thermo Fisher Scientific) or control siRNA (Stealth™ RNAi siRNA Negative Control, Thermo Fisher Scientific) using Lipofectamine™ RNAiMAX transfection reagent (Thermo Fisher Scientific), according to the manufacturer's instructions. After 24 h, the culture medium was changed to low-serum medium, and the cells were cultured for 21 h. Cells were stimulated with HRG (50 ng/mL) or vehicle (water) for 3 h.

Mammosphere Formation Assay
The mammosphere formation assay was performed as previously described [22]. Cells were seeded into 96-well ultra-low adherent plates at 1000 cells/well in MammoCult™ Medium (StemCell Technologies, Vancouver, Canada) and cultured for 7 days. The mammospheres were observed using the EVOS ® FL Cell Imaging System (Invitrogen, Thermo Fisher Scientific). Mammosphere formation efficiency was calculated as the number of spheres divided by the original number of seeded cells.

ATAC-Seq
ATAC-seq using HER2-5 and HER2-5/AHRKO cells was performed at Active Motif, Carlsbad, CA, USA. Only reads that were mapped to the human genome as matched pairs were used for analysis. Data from the HER2-5 and HER2-5/AHRKO cell samples were normalized to the same number of unique alignments (by downsampling to 28 million). Peaks were identified using MACS2 with a p-value cutoff of 1.0 × 10 −7 . For each possible pair-wise comparison, the shrunken log 2 fold-change was calculated using the DESeq2 algorithm.

Luciferase Reporter Gene Assay
HER2-5/AHRKO cells were transfected with the reporter and expression plasmids and the Renilla luciferase pGL4.74[hRluc/TK] plasmid (Promega; used as an internal standard) using the reverse transfection method with PEI Max ® (Polysciences, Warrington, PA, USA). After transfected HER2-5/AHRKO cells were incubated overnight at 37 • C, the cells were treated with the AHR antagonist StemRegenin 1 (10 µM) for 24 h. Subsequently, the cells were harvested, and luciferase activity was measured using the Dual-Luciferase ® Reporter Assay System (Promega). The activity of firefly luciferase was normalized to that of Renilla luciferase.

Statistical Analysis
Data of two groups were compared using a Student's t test, whereas those of three or more groups were compared using one-way analysis of variance followed by Dunnett's or Tukey's test. All statistical analyses were conducted using KaleidaGraph (version 4.1.1; Synergy Software, Eden Prairie, MN, USA); p-values < 0.05 were considered statistically significant.