Each BBQ analogue was accessed by condensation of the required anhydride and diamine under acetic acid reflux conditions (Scheme 1). Reaction workup (see experimental) afforded the desired BBQ analogues (Figure 2). With analogues 6 - 11, in keeping with prior reports of analogues of this nature, a mixture of regioisomers was obtained (with the exception of analogue 6: due to steric constraints, only the 10,12-dichloro analogue formed). ¹H, ¹³C NMR and UPLC-MS analysis was consistent with the presence of a regioisomeric mixture in all cases identified (see Figure 2 for detail).

Each BBQ analogue, 2-14 was screened for their growth inhibition ability (GI₅₀) in a broad panel of cancer cell lines and a non-cancer breast cell line (MCF10A), using the MTT growth inhibition assay (Table 1) (Baker et al., 2021b). The cancer panel comprised HT29 (colorectal), U87 (brain), SJ-G2 (brain), A2780 (ovary), Du145 (prostate), BE2-C (neuronal), MIA (pancreas), H460 (lung, ER+), A431 (vulva, ER+ (estrogen receptor positive)), and a wide selection of breast cancer cell lines with varying receptor status including MCF-7 (ER + luminal A), T47D (ER/PR+ (progesterone receptor positive)luminal A), ZR-75-1 (ER/PR + luminal A), BT-474 (ER/PR/HER2+ luminal B), SKBR-3 (HER2+), MDA-MB-468, BT20, and MDA-MB-231 (basal, triple negative (TN) for ER, PR, and HER2).

Our previously identified breast cancer selective molecule 2 (NAP-6) (Gilbert et al., 2020) was inactive in U87, SJ-G2, DU145, BE2C, MIA, and MCF10A cells (defined here as GI₅₀ > 50 μM), more active in MDA-MB-231 A2780, HT29, H460 and BT20 (GI₅₀ 35–14 μM) and sub-micromolar potent in A431, MCF-7, T47D, ZR-75-1, SKBR3, MDA-MB-468, BT474 (GI₅₀ 0.1–0.7 μM). With A431, and the breast cancer cell lines T47D, ZR-75-1, SKBR3, MDA-MB-468 showing a 100 to 300-fold increase in growth inhibition potency compared with the MCF10A cells.

Dichloro analogue 6 was insoluble in the compound testing media, and thus not screened. The remaining BBQ analogues produced GI₅₀ values that varied between inactive (GI₅₀ > 50 μM) (3, MIA, BT474; 4, BT474; 7, SJ-G2, BT474; 9, BE2-C, BT474; 10, DU145; 12, U87, Du145, H460, BT474, MDA-MB-231; 13, HT29, U87, A2780, Du145, H460, BT20, BT474, MDA-MB-231; 14, MCF-7, BT474) to low nanomolar (5, MDA-MB-468; 10, MDA-MB-468; 11, MCF-7, ZR-75-1, MDA-MB-468, BT20, and MCF-10A) and picomolar (11, H460).

The introduction of a phenyl disposed Cl-moiety with 3 and 4, yielded analogues with a similar growth inhibition profile across the cell line panel examined, with good inhibition in the breast cancer cell lines ZR-75-1, SKBR3 and MDA-MB-468 cells (GI₅₀’s 0.26–0.88 μM) and moderate to no growth inhibition in all other cell lines (GI₅₀’s 3.9 to >50 μM). The greatest difference between the growth inhibition in these breast cancer cell lines and that of the normal breast cells was 16-fold (c.f. 4, GI₅₀ 0.26 μM SKBR3 vs. 4.1 μM MCF10A cells). These data suggest potentially two roles for the Cl-moiety. Firstly, as a potential metabolic blocking group inhibiting CYP activation of a particular oxidation site (c.f. 3 and 4), and secondly as an electron withdrawing moiety, activating sites adjacent (or distal sites dependent on system conjugation) to the Cl-substituent (Zhang and Tang, 2018; Fang et al., 2019). While di-Cl additions produced an insoluble compound 6, the chlorine free analogue 5, represents the core BBQ scaffold. Analogue 5 induced strong growth inhibition in H460, A431 (GI₅₀ 0.61 and 0.05 μM), and the breast cancer cell lines T47D, ZR-75-1, SKBR3, MDA-MB-468, and BT20 cells (GI₅₀ 0.001–0.41 μM), with only moderate inhibition in all other cell populations (GI₅₀ 1.5–7 μM). The growth inhibition in these cell lines compared with normal cells was 5 to 3200-fold (c.f. GI₅₀ 0.001 μM MDA-MD-468 vs. 3.2 μM MCF10A cells), representing a significant enhancement in potency and selectivity. Methyl (CH₃) or cyano (CN) additions to the phenyl ring produced 7 (10-methyl and 11-methyl), 8 (9-methyl and 12-methyl) and 9 (10-cyano and 11-cyano) which resulted in a potency reduction compared with 5, supporting the concept that blocking substituents on the phenyl ring potentially reduces oxidation by CYP at these and adjacent sites.

Repositioning of the Cl-moiety to the naphthyl rings with 10, with no substituent on the phenyl moiety, saw a re-introduction of potency across all cell lines (GI₅₀ 0.0056 μM MDA-MB-468 to 9.8 mM in MIA cells). Indeed, 10 induced growth inhibition in skin A431 (GI₅₀ 0.15 μM), and the breast cancer cell lines MCF-7, T47D, ZR-75-1, SKBR-3, MDA-MB-468, BT20, BT474 (GI₅₀ 0.0056–0.56 μM) with a selectivity relative to the normal cell line of 7.5 to 750-fold (c.f. GI₅₀ 0.0056 μM MDA-MD-468 vs. 4.2 μM in MCF10A cells), showing a clear re-establishment of selectivity and potency in a manner similar to 5. The addition of a NO₂ moiety to the naphthyl ring produced 11 (3-NO₂ and 4-NO₂), which induced a significant increase in growth inhibition across all cell lines (GI₅₀ 0.00019 μM H460 to 2.1 μM MDA-MB-231), with the highest activity noted in H460 cells (0.00019 μM), breast cancer cell lines MCF-7 (GI₅₀ 0.0065 μM), ZR-75-1 (GI₅₀ 0.0027 μM), MDA-MB-468 (GI₅₀ 0.0015 μM), and BT20 (GI₅₀ 0.0084 μM), and normal breast cells MCF10A (GI₅₀ 0.0031 μM). The profile of growth inhibition suggests that 11 is mediating its effects via a similar pathway to the other BBQ analogues, but with an additional broad off-target(s), culminating in a highly potent broad spectrum compound. The removal of one ring from the naphthyl system with 12 had a profound, detrimental impact on both overall potency and selectivity, consistent with the need for an extended conjugated system, as present in the BBQ polyaromatic scaffold (Baker et al., 2021a). This conjugation requirement is reinforced by the further potency and selectivity reduction evident with the methylene spaced 13. The commercially available 14, with a naphthyl disposed carboxylate displayed little potency and selectivity, presumably a combination of the, relatively, poor electron withdrawing nature and the anticipated lower cell permeability of the carboxylate moiety (Ballatore et al., 2013).

Our previous studies have shown that the activity of NAP-6 (2) and 10-Cl-BBQ (3) required the activation of the AhR pathway and the induction of CYP1 family of phase I metabolising enzymes (Gilbert et al., 2020; Baker et al., 2021a). Therefore, we sought to determine the activation of AhR using a stable AhR reporter assay in HT29-Lucia™ AhR cells, in the presence of each BBQ analogue (Figure 3). Activation of the AhR was indicated by enhanced luminescence caused by the translocation of the AhR to the nucleus of the cell and interaction with the xenobiotic response elements imbedded in the luciferase gene. The analysis showed that all BBQ analogues induced AhR translocation and binding to the xenobiotic response element of the target luciferase gene, but at varying levels. Indeed, analogues 3-5 and 7-10 induced a response comparable to that of the natural AhR ligand FICZ (6-formylindolo[3,2-b]carbazole) (2.7-fold activation of AhR), which was slightly stronger than the signal noted with NAP-6 analogue 2 (1.8-fold). Consistent with the enhanced cytotoxicity observed, analogue 11 induced the greatest level of AhR activation (10.2-fold), while 12, 13, and 14 (1.2, 1.8 and 1.5-fold), induced a response similar to that of 2 (1.8-fold).

While induction of AhR activity is required to stimulate CYP-induced metabolic activation of 1, 2 and 3 (Gilbert et al., 2017; Gilbert et al., 2020), the magnitude of AhR induction by the BBQ analogues did not directly correlate with the potency of growth inhibition or to tumour type selectivity. For example, compounds 3, 4, and 5 induced a similar effect on AhR induction (2.7, 2.9, 3.1-fold), however, compound 5 was substantially more potent at growth inhibition (GI₅₀ 0.001 µM, MDA-MB-468 cells), than 3 (GI₅₀ 0.56 μM, MDA-MB-468 cells) and 4 (GI₅₀ 0.88 μM, MDA-MB-468 cells). This suggests that additional factors are at play, such as the limited ability of 3, and 4 to undergo CYP induced metabolic conversion caused by the Cl-blocking moieties, rather than a lack of AhR stimulation. An exception to the lack of correlation between AhR activity and growth inhibition was analogue 11 which induced the greatest growth inhibition and the greatest induction of AhR activity.

To further explore the role of the AhR pathway in the biological response of our BBQ analogues we chose to investigate the expression of downstream genes in the most sensitive cell line MDA-MB-468 cells, in response to compounds 3 and 5. These analogues were specifically chosen because they were synthesised without the production of regioisomers, their structural differences are small while their biological differences are large, c.f. 3 possesses a phenyl Cl and presented with minimal growth inhibition, while 5 lacks a phenyl substituent and presented with strong growth inhibition. Thus, we sought to examine the effect of 3 and 5, on AhR, CYP1A1, CYP1B1 and SULT1A1 expression in MDA-MB-468 cells, at 1 μM and the GI₅₀ value (0.2 and 0.02 μM, respectively) (Figure 4). Not surprisingly, the expression of AhR was not significantly altered following treatment with either 3 or 5 (Figure 4A); however, CYP1A1 and CYP1B1 expression was significantly increased (p < 0.01) within 6 h of treatment (Figures 4B,C). Indeed, the expression of CYP1A1 increased by 18 and 20-fold for analogue 3, which was comparable to that of the endogenous AhR ligand, FICZ (19-fold). Analogue 5 also induced CYP1A1 expression, and the response at the GI₅₀ value of 0.02 μM (12-fold) was comparable to that at 1 μM (15-fold), suggesting that the CYP1A1 activity may be rate limiting, in that substantially higher concentrations do not induce a commensurate increase in expression. Both analogues also significantly (p < 0.01) induced CYP1B1 expression (3.7-fold for 3; 4.6-fold for 5 at 1 μM) to a comparable level to that of FICZ (3.7-fold, 1 μM). Neither analogue altered the expression of SULT1A1 (Figure 4D). Collectively, the gene expression analysis following treatment with 3 and 5 was equivalent, suggesting that the differences in growth inhibition induced by these compounds was not caused by an inability to induce the AhR pathway, but rather the ability of the analogues to be bio-activated.

To further clarify the role of phase I and phase II metabolic activation of the BBQ compounds herein, we examined the effect of CYP1 and SULT1A1 inhibition on the growth inhibitory effect of compound 5 in MDA-MB-468 cells (Figure 5) and MCF10A cells (Supplementary Material). The co-administration of the CYP1 inhibitor α-naphthoflavone (Khojasteh et al., 2011) (10 μM) with 5, significantly reduced the growth inhibition response as indicated by the shift in the growth inhibition curve to the right, from a GI₅₀ of 0.005 μM–0.18 µM (36-fold reduction) (Figure 5A). Similarly, the co-administration of the SULT1A1 inhibitor quercetin (Pacifici, 2004) (5 μM) with 5, also significantly reduced the growth inhibitory effect of 5, from GI₅₀ of 0.005 μM–0.023 μM (4.6-fold reduction) (Figure 5B). A similar effect was noted for 3 but at a reduced level (Supplementary Material). Neither 3, 5, αNF, or quercetin affected growth in normal MCF10A cells (Supplementary Material). This analysis confirms that both CYP1 and SULT1A1, activity is needed to induce the growth inhibitory effect of the BBQ analogue.

We, and others have noted that the activation of the AhR pathway by ligands such as 1-7 results in specific cytotoxicity towards breast cancer cell lines. This was noted in the cases of 1-6, but as with the level of AhR expression, 3-NO₂ 11 was an outlier. This BBQ analogue displayed high potency across most cell lines examined, while retaining a preference for ER-positive cell lines. This resulted in excellent levels of MTT cytotoxicity with GI₅₀ values: 0.00019 (H460), 0.0065 (MCF-7), 0.0013 (T47D), 0.0027 (ZR-75-1), 0.0360 (SKBR3), 0.0015 (MDAMB468) and 0.0084 (BT20) μM. Given the significant AhR amplification (Figure 2), logically this enhanced potency originates from the AhR/CYP1/SULT1A1 axis, and the magnitude of the effect (with no differences in the predicted binding poses) is most likely due to electronic effects.

The possibility that the enhanced AhR upregulation noted with 11 (relative to the other analogues screened) is due to enhanced binding to the AhR, was investigated in a molecular docking study using our in-house AhR homology model. Our AhR model is based on the published PAS-B domain proteins, PDB 4F3L, 3RTY and 2KDK (Baker et al., 2018; Baker et al., 2021a). AhR upregulation may be a function of increased engagement of 11 with the active site. As previously described (Baker et al., 2018; Baker et al., 2021a), analogues 3, 5, 10b, and 11b were docked within the AhR active site using the Molecular Operating Environment (MOE) software suite (Chemical Computing Group, Montreal, Canada). On docking of each analogue, 500-poses were generated, and their respective docking energies minimised with the top-10 (based on predicted binding energies) examined in greater detail. The outcome of this examination is shown in Figure 6.

Examination of the stacked pose snapshot of the top binding poses of all screened analogues (Figure 6) revealed limited deviation in docked binding poses, binding energy, and binding location of the analogues (3, 5, 10b and 11b) examined. There were no observable changes in analogue binding. The introduction of a C-10 disposed halogen had no effect on binding energies (3, −6.8610 kJ.mol^-1 vs. 5, -6.8478 kJ.mol^-1). C-4 derivatisation resulted in a slight increase in binding efficiency (10b, −7.0198 kJ.mol^-1 and 11b, −7.5122 kJ.mol^-1), however no further interactions were observed toward the -Cl or -NO₂ moieties. This suggests that it is highly unlikely that the observed increase to AhR upregulation is due to increased or altered binding to the AhR.

With molecular modelling revealing no key analogue-AhR active site docking differences that explained the increased AhR activation by 11, we conducted higher level (DFT) calculations to explore the possible impact of compound geometry and electronic effects. The endogenous AHR ligand, FICZ (6-formylindolo(3,2-b)carbazole) was also examined to highlight key regions that are conserved between species. DFT calculations were performed using the dispersion-corrected (D3BJ), spin-component scaled double hybrid functional DSD-PBEPB86 developed by Kozuch and Martin (Grimme et al., 2010; 2011; Goerigk and Grimme, 2011; Goerigk et al., 2017) in conjunction with the valence triple-zeta with additional polarisation and diffuse functions basis set def2-TZVPP of Ahlrichs (Weigend and Ahlrichs, 2005; Hellweg et al., 2007; Jensen, 2013). Functional choice was guided by literature, showing improved accuracy over other common hybrid functionals such as B3LYP or B2PLYP (Kozuch and Martin, 2011; Karton and Spackman, 2021). Geometry optimisation calculations were performed in ORCA version 5.1 (Neese, 2012; 2018) and relevant plots constructed in Avogadro version 1.20 (Hanwell et al., 2012; Avogadro, 2022). These studies commenced with the calculation and visualisation of electrostatic potential (ESP), electron density (ED), and electrostatic potential mapped density (MEP) plots for 3, 5, 10, 11, and FICZ (Figure 7, Electronic Supplementary Material).

Analysis of the data presented in Figure 7 reveals that the BBQ analogues (3, 5, 10b and 11b) share essentially identical geometries and electronic distributions governed by the parent BBQ scaffold (5). Modification of parent 5, with the C-4 substituted 10b and 11b is accompanied by a decrease to positive potential in the naphthyl region of these analogues, as indicated by a colour mapping shift (red to blue) (Figure 7). Conversely, the C-10 dispose chloro-3 results in activation of this naphthyl region towards π−π type interactions (blue to red) (Figure 7). The van der Waal radii remains similar in all analogues with a slight increase observed in the NO₂ containing analogue 11b. All analogues, including FICZ, show remarkably similar regions of positive potential as indicated by the electrostatic potential map (ESP) map, localised around the carbonyl and imine (surface shown in red) (Figure 7). Our initial docking studies suggested that the imide carbonyls were not required for activity (contrary to what we have shown in related studies) (Baker et al., 2021a), however their inclusion aligns well to the endogenous ligand FICZ. While detailing the nature of BBQ analogues geometry and electronic distribution, this analysis is not consistent with the observed activation of the AhR by 11. As the interplay between AhR, CYP and SULT1A1 is crucial for activity, we hypothesised that the 11-mediated AhR activity may be a function not of 11, but of a CYP activated 11-metabolite.

The cytotoxicity of BBQ analogues is dependent on CYP1A1 and subsequent SULT1A1 metabolism (Figures 4, 5) to form the sulfonic ester warhead. CYP1A1 is capable of nucleophilic, electrophilic and single electron transfer (SET) oxidation initiating a metabolic process designed to solubilise and ultimately excrete ligands translocated by the AHR. The CYP1A1 mediated oxidation of FICZ is well studied (Wei et al., 1998; Wincent et al., 2009; Wincent et al., 2012), with monosubstituted 2-, and 8- OH analogues as the primary metabolites. Sequential oxidation produces the di-substituted 2-/8-, 2-/10- and 4-/8- diOH metabolites. However, the exact CYP1A1-FICZ hydroxylation mechanism remains elusive (Meunier et al., 2004; Guengerich, 2018). To examine the potential sites of CYP1A1 oxidation, via DFT calculations, we looked to calculate Fukui functions (indices) for these BBQ analogues (Proft et al., 2002; Pucci and Angilella, 2022). Specifically, using condensed Fukui Functions, permits calculation of the change in electron density when an electron is removed (mimicking electrophilic attack at that atom), added (mimicking nucleophilic attack at that atom) to a molecule or if a site is likely to undergo a SET reaction. These are the known reactions of CYP1A1, and related cytochrome P450 enzymes (Beck, 2005). This allows determination of the pseudo-probability of attack at a specific atom within a molecule (Electronic Supplementary Material).

Commencing with FICZ, Fukui Functions were calculated to predict the potential CYP1A1 metabolite(s). In analysing the Fukui Function data output, transformations that are unlikely to occur (nucleophilic attack towards the carbonyl oxygen or nitrogen atom, or addition to a quaternary carbon or loss of aromaticity) were considered non-viable options for metabolite generation. Additional consideration was given to the size and shape of the CYP1A1 catalytic site (Electronic Supplementary Material), it is likely that FICZ and other analogues are too large to pass through with their largest vDW radii parallel to the metabolic site which invalidates them as reactive sites. Hirshfeld population analysis was utilised for generation of Fukui indices (for complete numerical Fukui indices, see Electronic Supplementary Material) due to it its lower reliance on basis set choice (North et al., 2023).

The primary CYP1A1 metabolites of FICZ are known to be the corresponding 2- and 8-OH analogues (atoms 9 and 18, A₉ and A₁₈ respectively). However initial analysis of the Fukui Function probability plots, Figure 8, highlights atom 13 (A₁₃) with the largest values for both f ^- and f ⁰ , indicating a propensity to undergo electrophilic or SET attack at this atom. However, the size and binding characteristics of the CYP1A1 active precludes this possibility. Within the remaining FICZ atoms, A₁₈ displays the next largest f ⁻ (0.06529) and f ⁰ (0.04334) function values (Figure 8A), and thus a high probability of electrophilic or SET attack at this atom. This corresponds to the 8-OH FICZ metabolite, mechanistically suggesting its generation from FICZ by CYP1A1 occurs via electrophilic or SET attack. With a known second CYP1A1 mediated oxidation of FICZ (and thus most likely 8-OH FICZ), we generated Fukui Functions arising from the potential second pass oxidation of 8-OH FIZC (Figure 8B). From this, A₈ has the highest probability of participating in electrophilic or SET attack. This, again, corresponds well with the known second generation FICZ metabolism products, the 2,8-di-OH FICZ analogues (Wincent et al., 2009). Within the limitations of the Fukui Functions, i.e., an ability to predict the likelihood of an atom to accept or donate an electron. This is not the ability to predict that the reaction will occur (Pucci and Angilella, 2022; Zaklika et al., 2022).

With the Fukui Functions predicting the known metabolites of FICZ, we recapitulated this study with BBQ analogues 3, 5, 10b and 11b (Figure 9) in an effort to correlate Fukui Function identified metabolites trends with the observed cytotoxicity (Table 1).

Analysis of the Fukui Function output (full details of the Fukui Functions of 3, 5, 10b and 11b first and second pass metabolites are supplied in the Electronic Supplementary Material) suggests that with analogue 3, A₆ is most likely to undergo electrophilic and SET attack to produce the 4-OH analogue; while 5, 10b and 11b all show high f ^- and f ⁰ values which predicts the production of the corresponding C10-OH analogues (A₁₉, A₂₀, and A₂₂ respectively, Figure 9)). This is consistent with the observed cytotoxicity of these analogues, with 5, 10 and 11 > 10 fold more potent than head group substituted analogue, c.f. 3, 4, 6, 7, 8, and 9 across the cell lines examined (Table 1). The sub nanomolar activity of 11b is likely a function of the in situ formation of the corresponding nitrenium ion, through a nitroso intermediate (Uetrecht et al., 1997; Sidorenko et al., 2014; Ergüç et al., 2023).

All reactions were performed using standard laboratory equipment and glassware. Reagents and solvents were purchased from Sigma Aldrich, or AK Scientific and used as received. Organic solvents were of reagent quality and used as received. Melting points were recorded in open capillaries on a Büchi Melting Point M-565. Where available, literature values are provided and appropriately referenced. Electrospray ionization (ESI) mass spectra and HPLC spectra were recorded on an Agilent Technologies 1,260 Infinity UPLC system with a 6120 Quadrupole LC/MS in ESI positive (ESI⁺) and negative (ESI⁻) modes. Zorbax SB-C₁₈ Rapid Resolution HT 2.1 × 50 mm 1.8-Micron column, using a ratio of 1:4 10% HPLC-grade acetonitrile (ACN)/milli-Q H₂O and HPLC-grade ACN (both with 0.1% formic acid) as carrier solvents. Thin-layer chromatography (TLC) was performed on Merck silica gel 60 F254 pre-coated aluminium plates with a thickness of 0.2 mm and retention factors (r_f) determined where required. Column chromatography was performed under ‘flash’ conditions on Merck silica gel 60 (230–400 mesh).

Nuclear magnetic resonance (NMR) spectroscopy was performed on a Brüker Avance III 400 MHz spectrometer, where proton NMR (¹H NMR) spectra and carbon NMR (¹³C NMR) spectra were acquired at 400 and 100 MHz respectively, or a Brüker Avance III 600 MHz spectrometer, where ¹H NMR spectra and ¹³C NMR spectra were acquired at 600 and 150 MHz respectively. All spectra were recorded in deuterated dimethyl sulfoxide (DMSO-d ₆ ); or deuterated trifluoroacetic acid (TFA-d) obtained from Cambridge Isotope Laboratories Inc. Chemical shifts (δ) were measured in parts per million (ppm) and referenced against the internal standard and solvent peaks. Coupling constants (J) were measured in Hertz (Hz). NMR assignments were determined through the interpretation of one- and two-dimensional spectra. Multiplicities are denoted as singlet (s), doublet (d), doublet of doublets (dd), and multiplet (m). Peaks are listed in decreasing chemical shift in the following format: chemical shift (multiplicity, coupling constant, integration). Infrared spectroscopy (IR) were recorded on a Perkin Elmer Spectrum 2 FT-IR spectrometer.

The structures of all ligands to be docked were constructed in MOE and their conformations energy-minimized using Molecular Mechanics in conjunction with the AMBER force field. Docking was performed using MOE’s default settings, using the triangle matcher method in combination with the London δG scoring function for the initial placement of the ligand, followed by a refinement using induced fit methods and the GBVI/WSA scoring function. Compounds were docked 500 times, and the top 15 poses subjected to energy minimisation after refinement. Analysis and visualization of the docking output, such as identification of hydrogen bonds, steric clashes, hydrophobic interactions, or π−π-interactions were performed in MOE. Protein crystal structures were protonated, and energy minimised prior to docking.

Stock solutions were prepared as follows and stored as 20 mM DMSO solutions at −20 °C. All cell lines were cultured in a humidified atmosphere 5% CO₂ at 37 °C. The cancer cell lines were maintained in Dulbecco’s modified Eagle’s medium (DMEM) (Trace Biosciences, Australia) supplemented with 10% foetal bovine serum (FBS), 10 mM sodium bicarbonate, penicillin (100 IU/mL), streptomycin (100 μg/mL), and glutamine (4 mM). The non-cancer MCF10A cell line was cultured in DMEM:F12 (1:1) cell culture media, 5% heat inactivated horse serum, supplemented with penicillin (50 IU/mL), streptomycin (50 μg/mL), 20 mM Hepes, L-glutamine (2 mM), epidermal growth factor (20 ng/mL), hydrocortisone (500 ng/mL), cholera toxin (100 ng/mL), and insulin (10 μg/mL). The AhR reporter cell line HT29-Lucia™ (Invivogen, United States) was maintained in DMEM supplemented with 10% heat inactivated FBS, L-glutamine (2 mM), penicillin (100 IU/mL), streptomycin (100 μg/mL), 100 μg/mL Normocin™, and 100 μg/mL of the selective antibiotic Zeocin™. All cell lines were purchased from the American Type Culture Collection (ATCC), except for A2780 and A431 which were from the European Collection of Authenticated Cell Cultures (ECACC), and SJ-G2 cell line which was from Dr Mary Danks, St Jude Children’s Research hospital, Memphis, TN.