High AHR expression in breast tumors correlates with expression of genes from several signaling pathways namely inflammation and endogenous tryptophan metabolism

Increasing epidemiological and animal experimental data provide substantial support for the role of aryl hydrocarbon receptor (AhR) in mammary tumorigenesis. The effects of AhR have been clearly demonstrated in rodent models of breast carcinogenesis and in several established human breast cancer cell lines following exposure to AhR ligands or AhR overexpression. However, relatively little is known about the role of AhR in human breast cancers. AhR has always been considered to be a regulator of toxic and carcinogenic responses to environmental contaminants such as TCDD (dioxin) and benzo[a]pyrene (BaP). The aim of this study was to identify the type of breast tumors (ERα-positive or ERα-negative) that express AHR and how AhR affects human tumorigenesis. The levels of AHR, AHR nuclear translocator (ARNT) and AHR repressor (AHRR) mRNA expression were analyzed in a cohort of 439 breast tumors, demonstrating a weak association between high AHR expression and age greater than fifty years and ERα-negative status, and HR-/ERBB2 breast cancer subtypes. AHRR mRNA expression was associated with metastasis-free survival, while AHR mRNA expression was not. Immunohistochemistry revealed the presence of AhR protein in both tumor cells (nucleus and/or cytoplasm) and the tumor microenvironment (including endothelial cells and lymphocytes). High AHR expression was correlated with high expression of several genes involved in signaling pathways related to inflammation (IL1B, IL6, TNF, IL8 and CXCR4), metabolism (IDO1 and TDO2 from the kynurenine pathway), invasion (MMP1, MMP2 and PLAU), and IGF signaling (IGF2R, IGF1R and TGFB1). Two well-known ligands for AHR (TCDD and BaP) induced mRNA expression of IL1B and IL6 in an ERα-negative breast tumor cell line. The breast cancer ER status likely influences AhR activity involved in these signaling pathways. The mechanisms involved in AhR activation and target gene expression in breast cancers are also discussed.


Introduction
Breast cancer is the most common cancer in women in Western countries. All of the established risk factors combined can only explain less than half of all cases of breast cancer [1][2]. Environmental factors, i.e. the world around us and the way we live today, are probably also involved. Of particular concern are pollutants that alter the endocrine system and may modify cancer risks [3][4][5][6]. Epidemiological studies conducted after the "Seveso accident", one of the best known industrial accidents, and studies on chemical workers exposed to 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD or dioxin) [6] have revealed increased risks of developing breast cancer following exposure to TCDD, a potent ligand of the aryl hydrocarbon nuclear receptor (AhR). In particular, breast cancer incidence increased even more than 15 and 30 years after the Seveso accident [7][8][9]. Growing evidence from animal studies, including those performed on TCDD-exposed rodents during gestation, and on a number of established human breast cancer cell lines, provides substantial support for the role of AhR in mammary tumorigenesis. However, relatively little is known about the role of AhR in human primary breast tumors.
Over recent decades, AhR has been characterized as a regulator of toxic and carcinogenic responses to environmental AhR xenobiotics, such as TCDD, polycyclic aromatic hydrocarbons and halogenated aromatic hydrocarbons [10]. These ligands are widespread environmental contaminants. The repertoire of AhR ligands has been considerably expanded and now includes many industrial compounds, several classes of chemoprotective phytochemicals, namely flavonoids (quercetin), several indoles [(indole-3-carbinol, FICZ (6-formylindolo [3,2b]carbazole)], and several pharmaceuticals (omeprazole, hydroxytamoxifen, tranilast, aminoflavone) [11]. No study has addressed the molecular consequences of a combination of environmental and dietary ligands (a situation that is likely to occur in the environment) with an endogenous tumor-promoting AhR ligand [12].
Our understanding of AhR function was initially derived from toxicology and pharmacology studies on the response of AhR to xenobiotics, including regulation of the expression of xenobiotic metabolizing enzymes such as Cyp1A1 and Cyp1B1 [13][14]. More recent studies have demonstrated that AhR may be involved in other important cellular and pathological processes such as control of proliferation, regulation of cell cycle and cell migration [15], angiogenesis [16] and tumorigenesis [17][18][19]. Importantly, AHR modulates the response of immune cells to the presence of environmental and endogenous compounds [20][21][22]. These functions are likely dependent upon ligand-mediated activation of the receptor. The molecular pathway leading to AhR activation by exogenous ligands has been extensively studied. Once activated by ligands, AhR translocates into the nucleus and forms heterodimers with the AhR nuclear translocator protein (ARNT) to activate downstream gene transcription through interactions with cognate dioxin-responsive elements (DREs) located on AhR-responsive gene promoters (reviewed in [11]). An AhR repressor (AhRR) has been identified that regulates AhR activity by binding and sequestering the nuclear translocator ARNT [23][24]. More recent findings suggest that AhR may also participate in cross-talk with other transcription factors such as estrogen receptor (ERα) and NF-KB [25][26][27].
Various animal experimental data have provided substantial support for an association between abnormal AhR expression/function and breast cancer [4,17,[28][29][30]. TCDD is the prototypical and most potent known environmental AhR ligand. Its effects on breast cancer were first reported in rodent models of tumorigenesis, including TCDD-induced breast tumors that express high levels of AHR [17,[29][30]. The effect of different concentrations and the timing of exposure to TCDD on tumor development have been reported [4,28]. Studies have also focused on using cancer cells as models to determine the mechanisms and pathways activated by TCDD and other AhR ligands. Upregulation of AHR expression and transcriptional activity plays an important role in several, if not all, stages of malignant transformation [31][32]. Knock-down of aberrantly upregulated aryl hydrocarbon receptor reduces tumor growth and metastasis of MDA-MB-231 [33]. However, activation of AhR inhibits invasive and metastatic features of human breast cancer cells and promotes breast cancer cell differentiation [34]. Immortalized mammary fibroblasts lacking AhR also have impaired tumorigenic potential in a subcutaneous mouse xenograft model [35]. Different responses in sensitive genes have been observed in different breast cancer cells [36]. Recent studies indicate that AhR could also play distinct roles in the survival, migration and invasion of ERα-positive as compared to ERα-negative breast tumor cell lines in the absence of environmental chemicals [12,37]. A high degree of complexity has therefore emerged concerning the role of AhR in mammary tumorigenesis in various in vitro systems.
Despite clear-cut demonstration of the role of AhR in rodent models of carcinogenesis and in a number of established human breast cancer cell lines, relatively little is known about the roles of AhR in human primary breast tumors [19,38]. In particular, it is not known whether the relative expression of AhR may be a determinant factor for its role in breast tumor development, and very few studies have been published on the nature of AhR-positive cells in breast tumors or the abundance and potency of AhR ligands within the tumor [39]. The primary objectives of this study were therefore to identify the types of breast tumors that express AHR and investigate whether variations in AHR gene expression are associated with classical pathological parameters and outcome and with a panel of other gene expressions in order to provide insight into the signaling pathways that correlate with AHR expression levels, and thus affect breast tumor growth and progression.

Patients and samples for analysis of AHR and AHRR expression
Samples of 439 primary unilateral invasive breast tumors excised from women managed at the Institut Curie-René Huguenin Hospital (Saint-Cloud, France) between 1978 and 2008 were analyzed. Each patient signed a written informed consent form and this study was approved by the Institut Curie-René Huguenin Hospital Ethics Committee. Patients were followed at the Institut Curie-René Huguenin Hospital.
Patients had a mean age of 61.8 years (range: 31-91 years) and had to meet the following inclusion criteria: a) primary unilateral nonmetastatic breast carcinoma for which complete clinical, histological and biological data were available; b) no radiotherapy or chemotherapy prior to surgery. Treatment consisted of radical mastectomy in 277 cases (63.5%) and breastconserving surgery plus locoregional radiotherapy in 159 cases (36.5%) (information available for 436 cases). Tumor samples were collected immediately after biopsy or surgery and were stored in liquid nitrogen until mRNA extraction.
Patient follow-up consisted of physical examination and routine chest radiography every 3 months for 2 years, and annually thereafter. Mammograms were performed annually. Adjuvant therapy was administered to 354 patients, consisting of chemotherapy alone in 86 cases, hormone therapy alone in 168 cases, and combined chemotherapy and hormone therapy in 100 cases.
The histological type and the number of positive axillary nodes were established at the time of surgery. The malignancy of infiltrating carcinomas was scored according to the Scarff-Bloom-Richardson (SBR) histoprognostic system. Hormone receptor (HR) [estrogen receptor alpha (ERα), progesterone receptor (PR)] and human epidermal growth factor receptor 2 (ERBB2) protein status was determined at by biochemical methods (dextran-coated charcoal method, enzyme immunoassay or immunohistochemistry) [40] and was confirmed by realtime quantitative RT-PCR assay.

Real-time qRT-PCR expression of AHR and AHRR genes, and genes involved in AHR signaling pathways
The conditions of total RNA extraction, complementary DNA synthesis and qRT-PCR were as described in protocols.io: http://dx.doi.org/10.17504/protocols.io.kehctb6.
Results, expressed as N-fold differences in target gene expressions relative to the TBP gene (and termed "NTARGET"), were determined as NTARGET = 2 ΔCtsample , where the ΔCt value of the sample was determined by subtracting the Ct value of the specific target gene from the Ct value of the TBP gene. NTARGET values of the samples were subsequently normalized so that the median of NTARGET values for normal breast tissues was 1 for AHR and AHRR expressions, and so that the value for "basal mRNA level" (smallest amount of quantifiable target gene mRNA, Ct = 35) was 1 for AHR signaling pathway genes.
The nucleotide sequences of the primers used were as follows: AHR-U

Immunohistochemistry
Immunohistochemical labeling was performed as previously described [5]. Immunohistochemical staining for AhR (rabbit polyclonal antibody clone H211, Santa Cruz, 1/50 dilution) was performed using Ventana Autostainer (USA). Sections of some tumors were also immunostained with antibodies directed against Cyp1B1 (Santa Cruz, dilution 1/200) and CD4 (clone Sp35, Roche Ventana, USA). The antigen-antibody complex was visualized using DAB as chromogen, as previously described [5]. AhR immunostaining was analyzed blindly in duplicate by two specialists including a certified pathologist [5].

Statistical analysis
The relative expression of each gene was characterized by the median and the range, as described above. Relationships between AHR or AHRR mRNA expression and clinical parameters were identified using non-parametric tests, the Kruskal-Wallis H test (relationship between one quantitative parameter and two or more qualitative parameters) and Spearman's rank correlation test (relationship between two quantitative parameters). Differences were considered significant at confidence levels greater than 95% (p<0.05). To visualize the efficacy of a molecular marker to discriminate between two populations (patients who developed/or who did not develop metastases) in the absence of an arbitrary cut-off value, data were summarized in a ROC (receiver operating characteristic) curve [41]. The AUC (area under the curve) was calculated as a single measure to discriminate efficacy. Survival distributions were estimated by the Kaplan-Meier method, and the significance of differences between survival rates was ascertained with the log-rank test. Metastasis-free survival (MFS) was determined as the interval between initial diagnosis and detection of the first metastasis. Cox's proportional hazards regression model was used to assess prognostic significance in multivariate analysis [42].

Relationships between AHR mRNA expression in breast tumors and classical clinical and pathological parameters and patient outcome
AHR mRNA expression level was assessed in the cohort of 439 tumor specimens and was compared to normal breast tissue samples ( Table 2). A fairly wide range of AHR mRNA expression was observed (0.0 to 5.77). AHR mRNA expression level was weakly associated only with one classical prognostic factor, i.e. age (p = 0.040). The cohort (n = 439) was classified into four breast cancer subtypes on the basis of hormone receptor (ERα and PR) and ERBB2 status: HR +/ERBB2+ (n = 50), HR+/ERBB2-(n = 281), HR-/ERBB2+ (n = 41) and HR-/ERBB2-(n = 67). AHR mRNA expression level was weakly associated with ERα and HR negative status (p = 0.039 and p = 0.018, respectively) and HR-/ERBB2-breast cancer subtypes (p = 0.047) ( Table 2). The possible relationship between AHR and MKI67 expression, EGFR expression or PIK3CA mutation status was also tested in the same tumors. No correlation was observed between AHR and MKI67 or EGFR mRNA levels (Spearman's rank correlation coefficient: r = +0.082, p = 0.082; r = +0.027, p = 0.58, respectively), and between AHR mRNA levels and PIK3CA mutation status (p = 0.32) ( Table 2). The impact of AHR mRNA levels on patient outcome was also assessed by studying MFS survival curves. AUC analyses were performed to identify a cut-off point, which divides the cohort into relevant AHR expression subgroups. No link was observed between AHR mRNA levels and MFS, suggesting that AHR expression is not a prognostic factor in breast cancer (data not shown).

Relationship between AHRR mRNA expression in breast tumors and classical clinical and pathological parameters and patient outcome
AHRR functions as a feedback modulator by repressing AHR-dependent gene expression, and may therefore have a major impact on the AHR signaling pathway [23,43]. Consequently, AHRR mRNA expression was also assessed in the cohort of 439 samples and was compared to normal breast tissue samples (Table 3). A very wide range of AHRR mRNA expression was also observed (0.0 to 19.8).
The level of AHRR mRNA expression was not associated with any classical clinical and pathological factors, i.e. age, SBR histological grade, lymph node status, macroscopic tumor size or breast cancer subtype. No correlation was observed between AHRR and EGFR mRNA expression, or between AHRR mRNA expression and PIK3CA mutation status (Table 3). AHRR mRNA expression was only positively correlated with MKI67 mRNA expression  (Spearman's rank correlation coefficient: r = +0.131, p = 0.0058) ( Table 3), suggesting a link between AHRR mRNA expression and cell proliferation. The level of AHRR mRNA expression was also correlated with AHR mRNA levels (Spearman's rank correlation coefficient: r = +0.115, p = 0.015) ( Table 3).
Tumors with the lowest levels of AHRR mRNA expression ( 0.49, n = 47, 10.7%) were significantly associated with poor MFS (p = 0.029; Fig 1), compared to tumors with higher levels of AHRR mRNA expression (>0.49, n = 392, 89.3%). Patients with the lowest levels of AHRR mRNA expression had a 5-year MFS of 68.1% +/-6.8% and a 10-year MFS of 47.7% +/-7.73%. Patients with the highest levels of AHRR mRNA expression had a 5-year MFS of 75.3% +/-2.2% and a 10-year MFS of 66.4% +/-2.49%. Multivariate analysis using a Cox proportional hazards model assessed the prognostic value for MFS of the parameters found to be significant on univariate analysis, i.e., SBR histological grade, lymph node status, macroscopic tumor size, PR status (Table 1) and AHRR mRNA level. The prognostic significance of lymph node status (p = 0.00025), macroscopic tumor size (p = 0.0062), SBR histological grade (p = 0.011) and AHRR mRNA level (p = 0.033) was maintained. This analysis confirmed that AHRR mRNA level is an independent prognostic factor in breast cancer.

mRNA expression analysis of genes involved in the AHR signaling pathway in the ERα-negative and ERα-positive subpopulations
The ER status of breast cancer cells and tumors is known to influence AHR transcriptional activity [36,44]. In order to study the possible implication of AHR expression on the expression of genes involved in the AHR signaling pathway regardless of ERα status, mRNA levels of 54 candidate genes were therefore analyzed in the two subpopulations (ERα-positive and ERαnegative) of this cohort of 439 breast cancers: 26 low AHR-expressing and 26 high AHRexpressing ERα-negative breast tumors, and 14 low AHR-expressing and 14 high AHRexpressing ERα-positive breast tumors. The cut-off to distinguish high AHR expression from low AHR expression in the two subpopulations (ERα-positive and ERα-negative) was defined so that the lowest value of the "high AHR" category was much higher than the highest value of the "low AHR" category. We chose a threshold of significance of p<0.01 to not find genes differentially expressed by chance, which can happen when a large number of variables are studied. AHR mRNA expression was significantly different between the low AHR-expressing and high AHR-expressing breast tumor groups in both the ERα-negative (p<0.0000001) and ERαpositive (p = 0.0000067) subpopulations (Table 4).
ARNT mRNA levels were also increased in the highest AHR-expressing groups independently of ERα status. AHRR mRNA level was significantly twofold higher in high AHRexpressing breast tumors compared to low AHR-expressing tumors, but only in the ERα-negative subpopulation (p = 0.0036) ( Table 4). Little or no correlations were observed between AHR mRNA expression level and nuclear receptors including ERα, PR and AR (Table 4).
In addition to ARNT, AHRR and genes encoding nuclear receptors (n = 6), genes reported in the literature to be involved in AHR signaling pathways, and various well known cancer pathways, including angiogenesis (n = 1), cell proliferation (n = 4), epithelial-mesenchymal transition (EMT) (n = 4), cell motility (n = 6), IGF pathway (n = 5), inflammation (n = 10), chromatin structure regulation (n = 5), DNA repair (n = 1), and upstream and downstream metabolism of AHR signaling (n = 3 and n = 8, respectively) were also analyzed ( Table 4). Little or no correlations were observed between AHR mRNA expression level and angiogenesis, cell proliferation, EMT, chromatin structure and, surprisingly, xenobiotic metabolism gene expression levels, which are well known AHR-inducible genes (Table 4). More interestingly, MMP1, MMP2 and PLAU mRNA expressions were significantly associated with the high AHR-expressing group in the ERα-negative subpopulation (p = 0.0024, 0.0022 and 0.0032, respectively) ( Table 4). A weaker correlation was also observed between AHR mRNA levels and MMP9 and MMP14. These results suggest that AHR plays an important role in cell motility, mainly in ERα-negative breast tumors. Among the five genes selected from the IGF pathway, IGF2R expression was significantly increased in the highest AHR-expressing groups, independently of ERα status (p = 0.00044 in ERα-positive tumors and p = 0.0082 in ERα-negative tumors, respectively) ( Table 4). IGF1R and TGFB1 were significantly overexpressed in the highest AHR-expressing group, but only in ERα-positive breast tumors (p = 0.0088 and p = 0.0035, respectively) ( Table 4).
IL1B, IL6, TNF, IL8 and CXCR4, involved in the inflammation pathway, were significantly overexpressed in the high AHR-expressing group compared to the low AHR-expressing group in the ERα-negative subpopulation (Table 4). Furthermore, these associations were not observed in the ERα-positive subgroup, except for IL6 and CSF1. The majority of these positive correlations (in particular for IL1B and IL6, p-value<0.01) were confirmed in the TCGA breast  cancer dataset (data not shown) [45]. Higher AHR mRNA expression levels therefore appear to be strongly involved in inflammation processes, mainly in ERα-negative breast tumors. Other genes that have recently been shown to be involved in tryptophan metabolism via the kynurenine pathway were also analyzed [46]. IDO1 mRNA levels were significantly increased in high AHR-expressing breast tumors relative to low AHR-expressing breast tumors in both ERα subpopulations (Table 4). TDO2 mRNA levels were also significantly increased in high AHR-expressing breast tumors, but only in the ERα-negative subpopulation (Table 4). These positive correlations, in particular for IDO1 (p-value<0.01), were also confirmed with the data of The Cancer Genome Atlas Breast invasive carcinoma project (data not shown) [45].
Finally, BRCA1 mRNA expression was strongly associated with high AHR-expressing breast tumors in the ERα-negative subpopulation (p = 0.0000023) ( Table 4). A similar, approximately fourfold difference in BRCA1 mRNA expression was observed between the high and low AHR mRNA expression groups in both ERα-positive and ERα-negative subpopulations. However, a weaker association was observed in the ERα-positive subpopulation (p = 0.012) compared to the ERα-negative subpopulation.

Effect of two different AHR ligands on mRNA expressions of AHR, AHRR and ARNT, and of genes involved in inflammation, in MDA-MB-436 breast cancer cell line
In order to confirm the implication of AHR in the regulation of inflammation genes, we examined the effect of two AHR ligands: TCDD (2,3,7,8-tetrachlorodibenzo-p-dioxin) and BaP (benzo[a]pyrene), on mRNA expression of several inflammation genes: IL1B, IL6, TNF, IL8 and CXCR4, in MDA-MB-436 ERα-negative breast cancer cells. First, to test the activity of the two ligands, we determined their effects on the expression of AHR, AHRR and ARNT. We did not detect significant effect of the ligands on AHR and ARNT expressions. However, the ligands strongly stimulated the expression of AHRR (AHR repressor). The two ligands induced therefore a negative feedback loop, indicating that they are active in our experimental conditions [24]. Moreover, we found that mRNA expression levels of IL1B and IL6 were significantly higher in cells treated with TCDD or BaP compared to control cells (Fig 2) confirming the positive effect of AHR on the regulation of inflammation genes. We were not able to study the effect of our ligands on the expressions of TNF and of genes of the endogenous tryptophan metabolism pathway (IDO1, IDO2 and TDO2), because of the absence of expression of these genes in this cell line.

AhR protein is present in breast cancer tissues
In line with the objective of this study, immunohistochemistry was performed on paraffin sections to assess the localization of AhR protein on a total of 30 ERα-positive or ERα-negative breast tumors. These tumors corresponded to a panel of freshly excised tumors ranging from grade 1 to grade 3. AhR immunostaining in peritumoral tissue ("normal" tissue adjacent to the tumor) was mainly observed in epithelial (glandular) cells and capillaries (Fig 3a). Tumor cells and intratumoral stroma were immunostained for AhR (Fig 3b). The sub-localization of AhR (nuclear or cytoplasmic) was analyzed in tumor cells from all samples. 100% of tumor samples were positively stained for cytoplasmic AhR in breast tumor cells. However, AhR immunostaining was also observed in both the cytoplasm and nuclei in 24/30 tumors (80%) (Fig 3b).
The intensity of AhR immunostaining in tumor cells varied from strong (Fig 3b) to low or moderate (Fig 4a-4c), depending on the individual tumor. AhR immunostaining was also observed in both nuclei and cytoplasm of tumor cells (Fig 3b). In addition, AhR was present in the intratumoral nonepithelial tissue (Figs 3b and 4a), including endothelial cells and immune cells including lymphocytes. The presence of AhR in tumor-infiltrating lymphocytes was confirmed by the use of CD4 antibodies (Fig 4d). The expression of Cyp1B1 protein, a known target of activated AhR, was also analyzed in tumors with high or low AhR protein levels. No correlation was observed between Cyp1B1 and high (or low) AhR protein levels in epithelial cells (data not shown), thereby confirming the mRNA expression results.

Discussion
This study suggests that the level of AHR expression could play an important role in breast tumorigenesis. Tumors with high levels of AHR show increased expression of genes involved in several pathways, including invasion, IGF signaling, inflammation, DNA repair and kynurenine metabolism. ER status influences the correlation between AHR expression and the transcription of most of these genes in breast cancers. We also show, for the first time, that AHRR status is associated with MFS and is an independent prognostic factor, while AHR is not an independent prognostic factor. In this study, the gene expression level was assessed using a quantitative real-time RT-PCR method. This single-step method has several advantages as compared to microarrays or RNAseq, including higher accuracy in the quantification procedure, and higher sensitivity.

Expression of AHR, AHRR and ARNT
In a series of 439 ERα-positive or ERα-negative breast tumors, representing the different molecular subtypes, large ranges of AHR (0-5.8-fold difference) and AHRR (0-19.8-fold difference) mRNA expression were observed, reflecting extensive heterogeneity of gene expression in tumor samples. When comparing AHR and AHRR, AHR was associated with age, ER and HR negative status, and HR-/ERRB2-subtype, whereas AHRR was independent of age, ER and HR negative status, and HR-/ERRB2-subtype. Higher basal AHR expression in triple-negative breast cancer has been previously reported [47][48][49]. The present study did not reveal any correlation between AHR and MKI67 or EGFR mRNA levels, or between AHR and PIK3CA mutation status, suggesting that high AHR mRNA levels are not implicated in cell proliferative activity. The strong correlation between high AHR and ARNT levels in both ERα-negative and ERα-positive tumors ( Table 4) also suggests that the heterodimer, AHR-ARNT (AHR nuclear translocator), could be active in breast tumors. The immunocytochemical results indicated a role for AhR in both tumor cells and the tumor microenvironment. Saito (2014) previously reported that AhR status was inversely correlated with the histological grade of invasive ductal carcinoma [38]. However, in their study, AhR status was based on a 10% immunocytochemical positivity threshold of carcinoma cells in breast tumors (n = 90) [38], while the present study analyzed both on AhR protein and mRNA. AhR protein was observed in both tumor cells and the tumor microenvironment (intratumoral stroma). AhR immunostaining was observed in the cytoplasm of tumor cells in the 30 tumors tested, and nuclear immunostaining was also observed in 24/30 tumors (80%). Nuclear AhR localization suggests AhR activation in breast cancers due to the presence of exogenous or endogenous ligands.
AHR mRNA levels in breast cancers were not a prognostic factor for patient survival, as also reported for colon, pancreas, stomach and thyroid cancers, which express high AHR mRNA levels [44]. Notably, our results revealed, for the first time, that breast tumors with high AHRR mRNA levels were significantly associated with good metastasis-free survival, compared to tumors expressing low AHRR mRNA levels. These results raise the novel suggestion that AHRR levels may represent an independent prognostic factor for breast cancer. This correlation has been previously reported for other tumors, including colon, lung, stomach and ovarian cancer [43]. AHRR acts as a tumor suppressor gene in several types of cancer cells. Knockout of AHRR in mammary epithelial cells enables them to grow in an anchorage-dependent manner [43]. AhRR is an evolutionarily conserved bHLH-PAS protein that inhibits both xenobiotic-induced and constitutively active AHR transcriptional activity in multiple species including humans [23][24]. AhRR functions as a feedback modulator by competing with the transcription factor for heterodimer formation with ARNT [23][24]. AHRR expression is induced by the AhR/ARNT heterodimer via binding to xenobiotic response elements (XREs) To identify genes that may be correlated with AHR levels and AhR signaling pathways in breast cancers, 54 candidate genes were selected from the two ERα subpopulations expressing high or low AHR levels. These genes were chosen on the basis of data of the literature for their involvement in AHR signaling pathways including cell motility, IGF pathway, inflammation, DNA repair, and upstream and downstream metabolism.

High AHR expression correlates with expression of genes involved in inflammation
IL1B, IL6, TNF, IL8 and CXCR4 mRNA levels were significantly increased in the high AHRexpressing ERα-negative breast tumor subpopulation, while these associations were only observed for IL6 and CSF1 in the ERα-positive tumor subgroup. Moreover, mRNA expression levels of IL1B and IL6 were significantly higher in MDA-MB-436 cells treated with TCDD or BaP compared to control cells confirming the positive effect of AHR on the regulation of inflammation genes. These results confirm and extend in vitro results showing that AhR activation promotes induction of IL6 [49][50][51]. Gene expression of IL6, a cytokine involved in immune cell homeostasis that elicits protumor and antitumor properties, has also been shown to be synergistically induced by stimulation of AhR activity in combination with IL1B or TNF in MCF-7 cells [50]. These results further confirm the important role of AhR in the regulation of inflammation.

High AHR expression correlates with several metalloproteases and genes involved in IGF signaling
Several proteases, including MMP1, MMP2, MMP9, MMP14 and u-PA, are involved in invasive breast tumor growth and metastasis [52]. Interestingly, high levels of MMP2 and PLAU that encodes uPA mRNAs were shown to be significantly associated with high AHR expression in addition to MMP1 in ERα-negative tumors. Other MMPs such as MMP9 and MMP14 were more weakly associated with AHR expression. These results also suggest an important role of AHR expression in breast tumor cell motility. A positive correlation was also demonstrated between AHR expression and IGF2R expression. In contrast, IGF1R expression was correlated with high AHR mRNA expression only in ERα-positive breast tumors, as previously described [53]. These results add support to the involvement of AHR, together with IGF1R and IGF2R, in the IGF signaling pathway, depending on the breast tumor group. Another novel finding of this study is that BRCA1 expression is strongly associated with the high AHR-expressing ERαnegative subpopulation, indicating a possible implication of AHR in DNA repair in ERα-negative breast cancers.

High AhR expression correlates with expression of genes involved in tryptophan metabolism
The levels of expression of IDO1, IDO2 and TDO2, involved in the early steps of tryptophan metabolism leading to kynurenine, an AhR ligand were analyzed [12,46]. Interestingly, IDO1 levels, but not IDO2 levels (not expressed), were significantly elevated in high AHR-expressing breast tumors compared to low AHR-expressing breast tumors in both ERα subpopulations. The IDO1 enzyme (indoleamine-2,3-dioxygenase) mediates the first rate-limiting step by converting tryptophan metabolites into L-kynurenine and is upregulated in an inflammatory microenvironment (e.g. in the presence of IL6) [54]. IDO1 enzyme activity may lead to a local "amino-acid starvation" response. By generating downstream metabolites, IDO1 enzyme activity may also affect immunity, including specific immunomodulatory or cytotoxic functions [55]. Recent studies have shown that tryptophan metabolites can alter the balance of Treg and Th17, two related populations of CD4+ T cells with opposing functions during immune responses [32]. IDO1 expression and differentiation of the common precursor of these immune cells may be governed by the presence of inflammatory cytokines [16]. L-kynurenine has also been reported to activate AhR [5], which positively regulates IDO1 expression by immune cells such as dendritic cells. TDO2 mRNA levels were also significantly increased in high AHR-expressing breast tumors, but only in the ERα-negative subpopulation. A strong correlation was therefore demonstrated between AHR, IDO1 and TDO2 expression in breast tumors. Whether the level of AHR is mainly due to kynurenine involving an autocrine loop, or due to the presence of an exogenous ligand has yet to be elucidated.
ER status is likely to influence the correlation between AHR expression and transcription of most of these genes in breast cancers. Following ligand binding in breast cancer cells, AhR can activate two pathways, an X/DRE-mediated DNA binding pathway and/or a non-X/DREmediated protein-protein interaction pathway, both of which can lead to changes in gene expression. The DRE-mediated pathway is the classical AhR pathway that leads to induction of dioxin-dependent genes, such as CYP1A1 and CYP1B1 and other genes including IL1B, IL6 and AHRR [10,56], as also observed in the present work, via direct binding of the AhR complex to the promoter containing a DRE (an AhR/ARN-T binding motif). In addition to this canonical AhR pathway, alternative pathways for AhR-mediated transcriptional regulation have also been described, in which ligand-bound AhR interacts with other signaling pathways (ERα, SP1 and NF-KB) to regulate gene transcription. This process involves protein-protein interaction. Many studies have suggested cross-talk between AhR and ERα in ERα-positive breast cancer cells [25,[57][58][59]. TGFB1, TNF, IL1B and IL6 are inhibited by ligand-bound AhR in ERα-positive breast cancer cells. Cross-talk between AhR and NF-KB pathways has also been implicated in the regulation of AhR-mediated gene transcription such as IL6 and IL8 in breast cancers [60][61].
Our findings documenting AhR expression levels may contribute to targeting AhR for breast cancer therapy. A complete overview of the role of AhR in breast tumor growth is not currently available, especially as most studies have been conducted on cell culture and rodent models. AhR plays a key role in driving normal mammary gland development, and in driving breast cancer progression [4,28]. AhR influences the major stages of tumorigenesis: initiation, promotion, progression and metastasis. Various classes of AhR ligands may influence tumorigenic outcome, especially in aggressive breast tumors [37,48,[62][63][64][65]. A major gap in our understanding of AhR activity in mammary tumors is the nature of the signals that drive AhR activation. In particular, the contribution of endogenous ligands vs exogenous ligands in various breast tumor types remains unknown. The presence of high-affinity AhR ligands can produce substantial AhR transcriptional activity, even in the presence of modest levels of AhR expression. The relative expression of AhR may be a determinant factor in the presence of low-affinity ligands. In the present study, none of the genes involved in xenobiotic metabolism (such as CYP1B1, the most extensively studied AhR-activated gene) were significantly increased with high AhR levels.
In conclusion, the results reported here represent progress compared to previous studies that focused on breast cancer cells as models for determining the mechanisms and pathways activated by TCDD, the main potent AhR ligand. The role of AhR in breast tumorigenesis has also been demonstrated in animal models. The present study documents the variable expression of AhR in tumors, both in tumor cells and in other cells present in the tumor microenvironment, confirming the complexity of AhR functions in breast cancer. Several genes whose expression is correlated with AHR expression and which are involved in various signaling pathways related to AhR activation were identified in this study. A major gap in our understanding of AhR activity in mammary tumors concerns the nature (exogenous or endogenous) of the signal that constitutively drives AhR activation. The present results suggest that AHR expression levels could also be discriminant in patients with hormone receptor positive and negative breast cancers and also indicate that AhR is a target for breast cancer therapy. The role of AhR in breast cancers, in the presence of endogenous or exogenous AhR ligands, merits further examination, by distinguishing ERα-positive from ERα-negative breast cancers. and reading the manuscript. We also thank P Balaguer (INSERM U1194, Montpellier), C Pim-