The AhR‐SRC axis as a therapeutic vulnerability in BRAFi‐resistant melanoma

Abstract The nongenetic mechanisms required to control tumor phenotypic plasticity and shape drug‐resistance remain unclear. We show here that the Aryl hydrocarbon Receptor (AhR) transcription factor directly regulates the gene expression program associated with the acquisition of resistance to BRAF inhibitor (BRAFi) in melanoma. In addition, we show in melanoma cells that canonical activation of AhR mediates the activation of the SRC pathway and promotes the acquisition of an invasive and aggressive resistant phenotype to front‐line BRAFi treatment in melanoma. This nongenetic reprogramming identifies a clinically compatible approach to reverse BRAFi resistance in melanoma. Using a preclinical BRAFi‐resistant PDX melanoma model, we demonstrate that SRC inhibition with dasatinib significantly re‐sensitizes melanoma cells to BRAFi. Together we identify the AhR/SRC axis as a new therapeutic vulnerability to trigger resistance and warrant the introduction of SRC inhibitors during the course of the treatment in combination with front‐line therapeutics to delay BRAFi resistance.


Introduction
Deciphering the genetic landscape of cancer led to a better understanding of tumor development, tumor annotation, and classification (Bailey et al, 2018). The identification of recurrent driver mutations underscored oncogenic addiction and designed new druggable targets revolutionizing patient care (Berger & Mardis, 2018). However, a major barrier to effective therapy is the capacity of cancer cells to resist. Melanoma represents a pioneering model to comprehend the multiple facets of resistance mechanisms.
The discovery of oncogenic BRAF mutations in about 50% of advanced melanomas has emerged as central, transforming melanoma therapy (Davies et al, 2002). The most common BRAF mutation consists of a T to A transition (T1799A), encoding a BRAF V600E oncogenic protein with constitutive kinase activity, leading to downstream MAPKinase signaling activation. Patient-tumors carrying such mutations are treated with BRAF inhibitors (BRAFi) namely vemurafenib (Bollag et al, 2010), dabrafenib (Hauschild et al, 2012), or encorafenib (Koelblinger et al, 2018), in combination with MEK inhibitors (MEKi) respectively cobimetinib (Larkin, 2014), trametinib (Salama & Kim, 2013;Robert et al, 2014;Daud et al, 2017), and binimetinib Shirley, 2018) to overcome BRAF paradoxical activation (Zhang et al, 2015) and maximize the therapeutic response. Under such front-line double blockade, patients show remarkable immediate responses. However, the response is transient, with median progression-free survival (PFS) of 15 months and a median overall survival up to 30 months (Michielin et al, 2020), followed by the development of resistance, leading to relapse and death Shirley, 2018).
Understanding the molecular mechanism of resistance to BRAFi/ MEKi double blockade is critical to maximize clinical response. Unlike other oncogenic addicted tumors, namely EGFR driven lung cancer (NSCLC), where the appearance of secondary mutation in the target gene (EGFR) is a common mechanism of resistance to EGFR inhibitors (Kobayashi et al, 2005), no BRAF secondary mutation has been so far reported in BRAFi-resistant melanomas.
In addition to these acquired genetic alterations, a new concept of resistance has emerged based on the capacity of melanoma cells to undergo transcriptomic reprogramming. Single cell transcriptomic analysis showed that the adaptive response to BRAFi is diverse, leading to the generation of a gradient of dedifferentiated cell states from melanocytic to neural crest state (Rambow et al, 2018;Tsoi et al, 2018). The plasticity of melanoma cells mediates a phenotype switching of the cells, which constitutes a robust escape route to therapy (Hoek et al, 2008;Kemper et al, 2014;Marin-Bejar et al, 2021). Under the control of the microenvironment or intrinsic cell factors, melanoma cells could switch from a proliferative to invasive state, acquiring resistance to targeted therapies. These phenotypic changes are mainly associated with a process of dedifferentiation similar to the epithelial-to-mesenchymal transition (EMTlike) that promotes metastatic spreading (Carreira et al, 2006;Hoek et al, 2008;Cheli et al, 2012;Verfaillie et al, 2015;Dilshat et al, 2021). Nonetheless, in some cases, melanoma cells still exhibit a differentiated state and are resistant to BRAFi (Tirosh et al, 2016;Rambow et al, 2018). Transcription factors, such as the master regulator of the melanocytic lineage, the microphthalmia-associated transcription factor (MITF) plays a critical and founding role in directing melanoma cell plasticity (Wellbrock & Marais, 2005;Mü ller et al, 2014;Noguchi et al, 2017;Goding & Arnheiter, 2019). While MITF High state is associated with melanocyte differentiation and drives melanoma proliferation (Hoek et al, 2008;Rambow et al, 2018), the MITF Low (Mü ller et al, 2014) state is associated with drug resistance, supporting the notion of transcriptional balance.
We demonstrated that the Aryl hydrocarbon Receptor (AhR) transcription factor is constitutively activated in a subset of melanoma cells, promoting the dedifferentiation of melanoma cells and the expression of BRAFi-resistant genes (Corre et al, 2018). Using two complementary genome-wide CRISPR/Cas9 screens (CRISPR-a and CRISPR-i), we and others further underscored the role of AhR in the acquisition of BRAFi resistance (Gautron et al, 2021;Goh et al, 2021).
AhR is a ligand-activated transcription factor belonging to the family of the basic-helix-loop-helix (bHLH) Per-Arnt-Sim (PAS) transcription factor. In its inactive state, AhR is part of a cytosolic multi-protein complex that includes heat-shock protein 90, p23, AhR-interacting protein (AIP) and SRC (Enan & Matsumura, 1996;Cox & Miller, 2004;Nukaya et al, 2010). Upon ligand binding, AhR dissociates from its chaperone complex and translocates into the nucleus, where it interacts with its partner the AhR nuclear translocator (ARNT). AhR-ARNT nuclear dimers regulate the expression of target genes through recognition and binding to xenobioticresponsive elements (XREs) located within the promoter of their target genes allowing cell specific gene expression programs. Remarkably, through ligand-binding interaction, AhR has the capacity to integrate environmental and cell-dependent signals (Denison et al, 2002) to shape and adapt the cell response, making AhR a very attractive candidate in regulating melanoma plasticity.
In addition, AhR activation has been shown to participate in the phosphorylation of the non-receptor tyrosine kinase SRC (Y416 residue) (Randi et al, 2008;Tomkiewicz et al, 2013;Fallahi-Sichani et al, 2017). SRC is known to be involved in many cellular functions, including the promotion of tumor-cell survival, motility, and invasion, through a rapid activation of focal adhesion kinase (FAK) contributing to cell migration and EMT (Nihal & Wood, 2016;Patel et al, 2016). However, the relation between AhR and SRC has not yet been explored in melanoma.
Herein, we aimed to delineate the role of AhR in orchestrating melanoma phenotypic switching during the acquisition of resistance through genomic and non-genomic routes. In particular, we pinpoint the crosstalk between AhR and SRC in reshaping cell fate and identify the AhR/SRC axis as a new therapeutic vulnerability for the treatment of BRAFi-resistant melanoma.

AhR controls acquisition of the invasive phenotype of melanoma
AhR is markedly expressed in highly dedifferentiated, resistant, and invasive melanoma cells, mediating resistance to BRAFi (Corre et al, 2018). Accordingly, genetic depletion of AhR in BRAFiresistant SKMel28 melanoma cells (SK28R) by CRISPR Cas9 technology (AhR KO; Fig EV1A) significantly reduced their resistance to various BRAFi (vemurafenib, dabrafenib, and encorafenib; Fig 1A). We specifically evaluated the role of AhR in controlling cell migration by performing wound-healing assays of BRAFi-sensitive or resistant melanoma cells (called SK28S and SK28R, respectively) after genetic depletion of AhR (SK28S/R KO; Fig 1B) or chemical inhibition using its specific antagonist (CH-223191, 10 μM; Fig 1C). The loss or inhibition of the AhR significantly reduced migration capacity of melanoma cells (Fig 1B and C). Next, we analyzed the invasive properties of SK28S and SK28R melanoma cell lines using tumor-spheroid assays, which mimic the 3D architecture of melanoma. BRAFi-resistant cells were far more invasive than BRAFisensitive cells ( Fig 1D) and the loss of the AhR transcription factor significantly reduced invasion of SK28R cells on a collagen matrix at 4 days ( Fig 1D). Comparable results were obtained by chemical inhibition (CH-223191, 10 μM) of AhR in SK28 cells and no effect was observed in AhR KO cells (Fig 1E), underscoring the specificity of the CH-223191 and AhR function.
The role of AhR transcription factor in governing the resistance and invasive capacity of melanoma cells was further highlighted after increasing the endogenous expression of AhR by CRISPR/SAM technology in melanoma cells expressing low levels of AhR protein (501Mel) or after rescuing SK28R AhR KO cell lines with a constitutively activated form of AhR, (CA-AhR;McGuire et al, 2001). CRISPR/SAM stable expression of endogenous AhR in 501Mel cells was obtained using two different single-guide RNAs (sgRNA) targeting the AhR promoter region, with a subsequent increase in the capacity to mediate the expression of AhR (Fig 2A). Increased AhR expression reduced slightly BRAFi sensitivity ( Fig 2B) and increased the invasive capacity of the 501Mel cells ( Fig 2C) according to the ability of the sgRNA to induce AhR expression. Comparable results were obtained with the stable SK28R KO cell line expressing the constitutively active form of AhR (CA-AhR; Fig 2D). Significant increase of BRAFi resistance ( Fig 2E) and invasive capacity were observed ( Fig 2F). Thus, both AhR expression and its activation control the phenotype of melanoma cells and their sensitivity to BRAFi.
The sensitivity of melanoma cells to BRAFi has been associated with a highly differentiated cell state under the control of the MITF transcription factor (i.e., MITF high or pigmentation signature) (Rose et al, 2016;Smith et al, 2016;Rambow et al, 2018). Conversely, we showed that AhR transcription factor participates in BRAFi resistance (Corre et al, 2018). To characterize the molecular role of AhR in such transcriptional reprogramming, we compared specific gene expression signatures (Invasion, Resistance, Proliferation, Melanocytic. . . corresponding to the median of gene expression of previously established gene-signatures associated with melanoma phenotype as described in Appendix Table S1). These comparisons were performed with the SK28S and SK28R melanoma cell lines before and after genetic depletion of AhR (SK28S/R KO) and its rescue with AhR constitutive active form (CA-AhR; from RNAseq data, GSE166617; Figs 2G, and EV1B and C). As we previously described and underscored here in Fig 2G, the β-signature (associated with BRAFi sensitivity; Corre et al, 2018) was highly represented in the proliferative, differentiated (Melanocytic and Transitory; Tsoi et al, 2018), and BRAFi-sensitive cell lines (SK28S). Conversely, the α-signature depicting canonical activation of AhR (Corre et al, 2018) was most prominent in dedifferentiated (neural crest-like and undifferentiated) BRAFi-resistant lines and co-occurred with the resistance signature (SK28R; Fig 2G). The absence of AhR expression (SK28R-KO) significantly decreased the expression of these genesignatures (Fig 2G), while the re-expression of the constitutively active form of AhR (SK28R-KO CA-AhR) led to their overexpression ( Fig 2G). Interestingly, these AhR associated signatures segregate BRAFi resistant melanoma cells from the sensitive ones (Cancer Cell Line Encyclopedia -CCLE RNA-seq data (Barretina et al, 2012)) and in Melanocytic-Transitory from Neural Crest like-Undiferentiated melanoma cells (GSE80824 (Tsoi et al, 2018); Appendix Fig S1A) and in invasive melanoma cell lines (Appendix Fig S1B; Verfaillie et al, 2015). Besides the role of AhR in the regulation of resistance gene expression, we underscored its role in the acquisition of the dedifferentiated/invasive/mesenchymal phenotype. Overall, these results report that AhR mediates specific gene signature controlling the phenotypic switch of melanoma cells.
AhR regulates the expression of genes associated with BRAFi resistance, invasion, and dedifferentiation phenotypes of melanoma To further decipher the direct role of AhR in the acquisition of the BRAFi-resistant associated phenotype we compared the previously established gene phenotype-signatures (Appendix Table S1) with RNAseq data from 501Mel cells exposed to BRAFi (Vem, 1 μM) or AhR ligand (TCDD, 10 nM) for 48 h (GSE104869 (Corre et al, 2018)), and with ChiP-Seq data identifying AhR target genes following exposure to TCDD (GSE90550 (Yang et al, 2018); Fig 3A  and Appendix Fig S2A). This led to the selection of 216 genes predicted to be regulated by AhR (Appendix Fig S2A). Among these genes, 92 were significantly enriched (GSEA) in the sensitive/differentiated phenotype and 75 in the resistant/dedifferentiated one (Appendix Fig S2B and C, and Dataset EV1). The 50 most highly enriched genes between these two states (25 sensitive/differentiation genes (green) and 25 resistant/dedifferentiation genes (red), Dataset EV1 and Fig 3B), segregated sensitive and resistant SK28 cells (GSE166617; Fig 3B). AhR knockout (SK28R KO) and expression of its constitutively active form confirmed the involvement of AhR in the regulation of these resistant genes (in red, Fig 3B and C). Comparable results were obtained in 501 Mel cells overexpressing endogenous AhR (CRISPR/SAM; Fig EV1D). Finally, these 50 AhRassociated genes segregated BRAFi-resistant melanoma cells from sensitive ones (CCLE RNA-Seq data (Barretina et al, 2012)) and Melanocytic-Transitory melanoma cells from Neural Crest like-Undifferentiated ones (GSE80824 (Tsoi et al, 2018); Fig 3B).
Interestingly, several of these AhR-associated genes have been involved in the aggressiveness of melanoma or other cancers (Appendix Table S2) and have been associated with a poor prognosis (ABCG2, COL1A1, COL6A1, COL6A2, TGFBI). CCL2, CRIM1, COL1A1, 6A1, 6A2 participate in cell migration, invasion, or EMT and ABCG2, ALDH1A1, NES are cancer stem-cell markers. Furthermore, AhR has been shown to directly regulate the expression of some of them (ABCG2, CCL2, STC2, etc.; Appendix Table S2), supporting the role of AhR in resistance.
We next explored the clinical relevance of this AhR-associated genes-signature by first examining melanoma samples from the TCGA cohort (Anaya, 2016). Among, the analyzed melanoma samples (n = 454), 17% of Patients strongly expressing AhR-associated resistance genes (red box) showed significantly lower overall survival than those highly expressing genes for sensitivity (blue box; Appendix Fig S3A). We investigated the expression of these genes in melanoma patients exposed to single drug-blockage (BRAFi) by classifying their melanoma biopsies during the course of medication and disease progression (baseline, early: DP1, intermediate: DP2, late: DP3; RNAseq dataset from Hugo et al (2015), GSE65185). Again, their expression level decreased at the beginning of the treatment (response to BRAFi phase: DP1) and slowly but significantly increased during the acquisition of BRAFi resistance ( Fig 3D). Using, additional RNAseq data from melanoma cells lines (M229 and M397; GSE110054 (Tsoi et al, 2018)), we confirmed that the acquisition of BRAFi resistance correlates with a late increase in the expression of AhR-associated gene-signature (Appendix Fig S3B). The BRAFi/MEKi double blockade led to similar reprogramming of gene expression (Appendix Fig S3C).
Canonical activation of AhR triggers the SRC pathway to promote the BRAFi-resistant/invasive phenotype of melanoma AhR is part of a cytosolic multiprotein complex with HSP90 and the SRC kinase (Enan & Matsumura, 1996; Rey-Barroso et al, 2013) ( Fig 4A). We investigated the potential cross-regulation between AhR and SRC signaling. We performed co-immunoprecipitation experiments to determine whether AhR and SRC are present in the same protein complex in melanoma cells (SKMel28). AhR was detected in the SRC immunoprecipitate ( Fig 4A) and this interaction was confirmed by proximity ligation assay ( Fig EV2A). We next tested whether AhR controls the activation of the SRC kinase in the context of BRAFi resistance. To this end, we examined their protein levels and the phosphorylated form (AhR, SRC, P-SRC, etc.) in four melanoma cell lines with increasing levels of BRAFi resistance (Figs 4B and EV2B). Concomitant to BRAFi resistance, we observed increased AhR protein levels and increased activation of SRC after phosphorylation on residue Tyr 416 (Y416). Activation of FAK (phosphorylation on tyrosine 576/577) followed SRC phosphorylation (Figs 4B and EV2B). We next examined SRC and FAK protein levels and phosphorylation state in SK28R cells expressing or lacking AhR (Figs 4C and EV2C). While constitutive loss of AhR did not significantly induce SRC or FAK activation, rescue experiments with AhR Sustained canonical activation of AhR mediates BRAFi resistance and the activation of the SRC/FAK athway. To delineate the contribution of AhR-SRC axis in resistance acquisition, we established the gene expression profile (RNA-seq) of SK28R cells exposed to AhR agonist ITE (5 μM, 24 h) or to dasatinib (Das), a specific inhibitor of SRC (1 μM, 24 h). Comparative analysis of differentially expressed genes in SK28R cells exposed to ITE (10 μM for 24 h; Appendix Fig  Together this underlines AhR-induced genomic and non-genomic reprogramming of melanoma cells.

Inhibition of SRC sensitizes melanoma cells to BRAFi treatment and disrupts the acquisition of an invasive phenotype
Having pinpointed the cellular role of AhR in directing BRAFi resistance, we explored new therapeutic opportunities. Using the Cell-Miner database (https://discover.nci.nih.gov/cellminercdb), we correlated the therapeutic efficacy of a library of 300 chemical compounds (IC50) according to AhR mRNA levels (Z-score) in various cancer cell lines (lung, brain_CNS, breast, skin; Appendix Fig S7). Volcano plots showed a significant correlation (P < 0.001) for several drugs and scored the SRC inhibitor (dasatinib, Das) to be potentially effective in the context of BRAFi resistance (Appendix Fig S7). Correlative analysis (Das efficacy/gene expression) further showed that the SRCi Das was more effective in cell lines with a high level of AhR mRNA and strongly expressing genes mediating resistance, invasion, and melanoma dedifferentiation (Appendix Fig S8A and Table S3). The expressions of genes associated with AhR and SRC/ FAK signatures were strongly correlated with the efficacy of Das (Appendix Fig S8B and C).
The two ATP-competitive protein tyrosine kinase inhibitors of SRC (bosutinib (Bos), dasatinib (Das)) were very effective resulting in the complete loss of the phospho-activated SRC form (P-Y416) in SK28R melanoma cells (WT, KO-AhR, CA-AhR; Fig EV3A). Concomitantly, we observed a decrease in the expression of SRC related genes (THBS1, MMP1; Fig EV3B). Treatment of SK28R melanoma cells with increasing doses of SRC inhibitors (Bos or Das) at different doses (10-500 nM) in combination with increasing doses of Vem significantly increased BRAFi sensitivity (Fig 5A). To analyze the effect of SCRi on cell viability, SK28R melanoma cells were treated alone with increasing doses of SRC inhibitors (Bos or Das, up to 0.5 μM) or in combination with BRAFi. Alone SCRi affected cell viability when used at relatively high concentration (> 0.125 μM for Bos and > 0.031 μM for Das). In contrast, when ◀ Figure 1. Loss of AhR reduces the invasive phenotype of BRAFi-resistant melanoma cells.

A
Vem sensitivity was established in BRAFi-resistant SK28 cells before and after knockout out of AhR by CRISPR/Cas9, by cell density measurements for 4 days after treatment (every 2 days), with an increasing concentration of BRAFi (vemurafenib, dabrafenib, or encorafenib). The IC50 (M) was calculated using GraphPad (PRISM9.0 ® ). B, C Wound healing assays were performed using IBIDI ® chambers to evaluate the role of the AhR on cell migration. Images of the wound were captured using an Axio Vert.A1 inverted microscope (Carl Zeiss ® ) at 5× magnification. The histogram represents the mean AE s.d. Wound closure was determined by measuring the distance between the edges of the wound at time 0 and 15 h (n = 3 independent technical experiments for each cell lines or conditions) and compared using unpaired t-tests with the Sidak-Bonferroni method. (B) Results obtained with BRAFi-sensitive or resistant SK28 cells KO for the AhR in the absence of treatment. (n = 3 independent technical experiments for each cell lines). (C) Results obtained for the migration assay (0-15 h) for SK28 R cells KO or not for the AhR after treatment or not with 10 μM CH-223191. (n = 3 independent technical experiments for each cell lines or conditions, mean AE s.d.). Statistical analysis using unpaired t-tests with the Sidak-Bonferroni method has been performed between the mean of the three independent experiments. D Three-dimensional spheroid growth of BRAFi-sensitive or resistant SK28 cells KO before or after knockout of AhR by CRISPR/Cas9 in the absence of treatment. Images were captured 4 days after implantation of the spheroids into collagen gel. (n = 4 independent technical experiments, mean AE s.d.). E Three-dimensional spheroid growth of BRAFi-resistant SK28 cells KO before or after knockout of AhR by CRISPR/Cas9 after daily treatment with the specific AhR inhibitor CH-223191 (5 μM) for 1 week or in the absence of treatment. Images were captured 4 days after implantation of the spheroids into collagen gel. (n = 3 independent technical experiments, mean AE s.d.). Statistical analysis using unpaired t-tests method has been performed with the Sidak-Bonferroni method, (P < 0.01 ##, **, P < 0.001 ###, ***).
Source data are available online for this figure.
Ó2022 Das also sensitized other resistant melanoma cell lines (M229R and M238R), to BRAFi ( Fig EV3C). In addition to their roles in sensitizing melanoma cells to BRAFi, the SRC inhibitors Bos and Das, even at low doses (1 μM), prevented the invasive capacity of wildtype AhR melanoma cells (red) in three-dimensional spheroid assays ( Fig 5B). Das was also able to reduce the low-invasive capacity of KO-AhR melanoma cells (green; Fig 5B). Remarkably, Das blocked invasion induced by AhR activation (ITE, TCDD; Fig 5C) without affecting AhR protein level (Fig 5D) but by significantly reducing AhR transcriptional activity, alone or after AhR activation by TCDD measured by Luciferase assay (Fig 5E). To further support this identified AhR/SRC cross-regulation, we performed RNAseq on SK28R cells before or after treatment with Bos or Das and characterized the effect of SRC inhibitors on gene-reprogramming signature and AhR-target genes. Both inhibitors significantly decreased the expression of genes associated with SRC, FAK, focal adhesion and invasive/dedifferentiation signatures and AhR-targets ( Fig 5F).
This in vitro evidence prompted us to examine the clinical relevance of using SRC inhibitors to resensitize BRAFi-resistant tumors to BRAFi. To this end, we used the Mel006R BRAFi-resistant patient-derived xenograft (PDX) mice model. The PDX line MEL006R is a BRAF V600E mutant cutaneous melanoma derived from MEL006 PDX lesions at relapse (Vendramin et al, 2021) upon acquisition of resistance to BRAFi/MEKi (Dabrafenib and Trametinib). Once tumors reached 200 mm 3 , grafted mice were treated with different treatment regimens: single-drug regimens (BRAFi or SRCi alone) or sequential administration (SRCi alone during the early growth phase of the tumor (16 days) followed by a BRAFi/SRCi double blockade). The growth of the tumor was monitored each 2 days until the tumor reached 1,500 mm 3 ( Fig 6A). As anticipated, BRAFi alone was largely ineffective in controlling the tumor growth of this BRAFi-resistant PDX (Figs 6B and EV4A). On the contrary, SRCi alone induced significant control of the tumor growth and the mice under SRCi survived significantly longer than the ones treated with BRAFi alone (Figs 6B and EV4A). Together this reinforces the role of SRC activation as an escape route to BRAFi. Remarkably, introducing BRAFi, 16 days post treatment with SRCi ( Fig 6A) or by combining BRAFi and dasatinib from the start of treatment (Appendix Fig S9) significantly diminished the tumor growth rate. As a consequence, those mice survived significantly longer, with an overall survival rate almost doubled compared to mice treated with BRAFi alone. In conclusion, dasatinib after specific inhibition of SRC phosphorylation (Fig EV4B) significantly resensitized resistant tumors to BRAFi treatment and significantly increased the overall survival ( Fig 6B-D). These results emphasize the therapeutical interest of SRCi for BRAFi-resistant patients.

Discussion
The acquisition of resistance to targeted therapy can be in part mediated by transcriptional reprogramming, eliciting a phenotypic ◀ Figure 2. The activated form of AhR rescues the invasive and BRAFi-resistant phenotype of melanoma cell lines.
A AhR Protein levels in 501Mel CTR cells and those transduced with sgRNA targeting AhR (#1, #2) were analyzed by western blotting. B BRAFi sensitivity was established in 501Mel cells or with endogenous overexpression of the constitutive form of the AhR (sgRNA, CRISPRi), by measuring cell density for 4 days after treatment (every 2 days), with an increasing concentration of BRAFi (vemurafenib). The IC50 (M) was calculated using GraphPad (PRISM9.0 ® ). Statistical analysis (two-way ANOVA) has been performed between the mean of four independent experiments (SK28R KO vs. SK28R) at different concentrations; P < 0.001 ***. C Three-dimensional spheroid growth of 501Mel cells. Images were captured 7 days after spheroid implantation (n = 3 independent technical experiments, mean AE s.d.). Each histogram represents the mean AE s.d. Comparisons with control were performed using unpaired t-tests with the Sidak-Bonferroni method, P < 0.001 ***. D AhR Protein levels in SK28R WT, KO, and KO CA-AhR cells were analyzed by western blotting. E BRAFi sensitivity was established in SK28R cells in the absence of AhR (SK28R KO (CRISPR-Cas9) or after overexpression of the constitutive form of AhR (CA-AhR), by measuring the cell density for 4 days after treatment (every 2 days), with increasing concentrations of BRAFi (vemurafenib). The IC50 (M) was calculated using GraphPad (PRISM9.0 ® ). Statistical analysis (two-way ANOVA) has been performed between the mean of four independent experiments (SK28R KO or SK28R KO CA-AhR vs. SK28R) at different concentrations; P < 0.01 **, P < 0.001 ***. F Three-dimensional spheroid growth of SK28 cells KO or not for AhR by CRISPR/Cas9 and rescued by the constitutive active form of AhR (KO CA-AhR). Images were captured 4 days after spheroid implantation (n = 3 independent technical experiments, mean AE s.d.). Comparisons were performed using unpaired t-tests with the Sidak-Bonferroni method. G Expression heatmap for various gene signatures (established by the median of expression for specific genes) (invasive vs. proliferative, alpha, beta, resistant, melanocytic, transitory, neural crest-like, and undifferentiated, see Appendix Table S1) in SK28 BRAFi-sensitive or resistant cell lines KO or not for AhR by CRISPR/Cas9 and rescue with the constitute active form of AhR (KO CA-AhR; GSE166617). Genes and clusters with similar expression profiles across the cohort are placed close to each other in the grid. The scale corresponds to the Z scores.
Source data are available online for this figure.
◀ Figure 3. AhR regulates the expression of genes associated with the BRAFi-resistant/dedifferentiated phenotype of melanoma.
A Workflow for the identification of AhR regulated genes among signatures. B Expression heatmap for the for the most highly enriched genes (n = 50; Appendix Fig S2B and C) from BRAFi-sensitive/proliferation/differentiation (green) or BRAFi resistance/invasion/dedifferentiation signatures (red) in SK28 BRAFi-sensitive or resistant cell lines KO or not for AhR by CRISPR/Cas9 and rescued by the constitute active form of AhR (KO CA-AhR; GSE166617), BRAFi-sensitive or BRAFi-resistant melanoma cell lines from the Cancer Cell Line Encyclopedia (CCLE; Barretina et al, 2012) and melanoma cell lines from the Graeber datasets (Tsoi et al, 2018). The enrichment rank for the signatures are available in Dataset EV1. C Expression heatmap for the most highly enriched genes (n = 50) in SK28R WT, KO, and KO CA-AhR cells. D Expression heatmap for the median expression of the most highly enriched genes after GSEA for 9 BRAFi-treated melanoma patients during melanoma progression (pre-treatment, during disease progression DP1 n = 9, DP2 n = 9, DP3 n = 5; GSE65185 (Hugo et al, 2015)). Clinical data are available in supplemental  A At the basal level, AhR is located at the membrane layer in a protein complex that includes HSP90, p23, XAP, and SRC. Immunoprecipitation of SRC was performed with specific antibodies and SRC/AhR interaction analyzed by western blotting.  Table S2). Interestingly, the phenotype of melanoma cells lacking Mitf transcription factor was similar to those observed while AhR is highly expressed and activated (Dilshat et al, 2021), allowing to consider a new balance between these two transcription factors for the control of melanoma plasticity.
Second, sustained activation of AhR mediates the activation of the SRC pathway following phosphorylation of the Tyr 416 (Y416). Together, AhR-dependent transcriptional reprogramming and SRC activation triggers the cell plasticity of BRAFi-resistant melanoma. The identification of an AhR/SRC regulation node fully supports the importance of non-genomic cell reprogramming. It also provides a strong rationale for the understanding of the role of the SRC-family in BRAFi treatment Close et al, 2020;Krayem et al, 2020) and allows to delineate the pathway that mediates the activation of SRC and elevated integrin/FAK observed in melanoma (Hirata et al, 2015).
Indeed, the crucial role of SRC in many aspects of tumor development including migration, invasion and survival has warranted the use of SRC inhibitors to disrupt these effects in several cancer types (Roskoski, 2015). In this respect, SRC inhibitors have been tested in melanoma. However, the anti-proliferative effect of SRC inhibitors alone, on melanoma cells, was minor to no effect. Importantly cytotoxicity was mainly observed in cells that did not carry BRAF oncogenic mutation (Eustace et al, 2008;Halaban et al, 2019). In accordance, clinical studies using SRCi as a single agent showed only minimal therapeutic activity in stage III/IV chemotherapy-naive unresectable melanoma (Kluger et al, 2011). These results contrast with those supporting the use of SRC inhibitors in resistance settings in line with the upregulation of members of the SRC-family kinases (Girotti et al, 2013) and downstream SRC-dependent effectors such as MCF2 and VAV1, two DBL family members identified through a genetic screen as candidate drug resistance in melanoma cells (Feddersen et al, 2019). SRC inhibitors were also reported to promote a differentiated state through the upregulation of Mitf expression and downstream melanocytic markers (TYR, TRP1. . .) via the MAPK and CREB pathways (Ku et al, 2019). This gives some hints of how SRC may participate in melanoma cell reprogramming. The identification herein of the AhR/SRC activation loop in BRAFi-resistant melanoma gives rationale to these studies filling an important gap to understand cell plasticity and propose innovative therapeutic regimens.
We previously showed in a preclinical PDX melanoma mice model that antagonizing AhR delayed the emergence of resistant cells (Corre et al, 2018). Here, using a BRAFi-resistant PDX melanoma model, we demonstrated that SRC inhibition (dasatinib) significantly controlled tumor growth and remarkably re-sensitize melanoma cells to BRAFi (dabrafenib), doubling the overall survival rate compared to BRAFi alone. This allows us to envision new therapeutic settings using SRC inhibitors to resensitize tumor cells to BRAFi and to improve therapeutic benefits with delayed relapses. The time to introduce SRCi could be determined by monitoring the presence of circulating tumor DNA (ctDNA) in liquid biopsies as an early marker of tumor progression (Calapre et al, 2017). The detection of the BRAF V600 mutation could serve as the starting point to initiate co-treatment with SRCi.
Our results also underscored that AhR-dependent activation of SRC in BRAFi-resistant cells leads to the activation of FAK kinase after phosphorylation. Marin-Bejar et al, 2021 have recently shown that gains activity of FAK signaling is associated with the emergence of neural crest stem cell (NCSC) subpopulation in BRAFi/MEKi drugtolerant cells known as minimal residual disease (MRD). This activation of FAK in melanoma cells is driven in part by a "paradoxical" activation of melanoma-associated fibroblasts and the induction of β1/FAK/SRC signaling (Hirata et al, 2015) but also after activation of GFRA2/GDNF expression and AKT activation (Marin-Bejar et al, 2021). Interestingly, FAK-inhibitors strongly decreased the emergence of the NCSCs in MRD lesions, and drastically delayed the onset of resistance to RAF/MEK inhibitors in preclinical PDX models. They also proposed to test combinations of both FAK and SRC inhibitors, such as dasatinib, as a more effective strategy to suppress the emergence of the NCSC population at MRD (Marin-Bejar et al, 2021).
In parallel, we underscored that AhR-dependent activation of SRC mediates the activation of the epidermal growth factor receptor ◀ Figure 5. Inhibition of the SRC pathway increases BRAFi sensitivity and prevents invasive phenotype of melanoma cells.

A Heatmap of the percentage of cell viability in SKR (left) or SKR KO or SKR KO CA-AhR (right) cells treated with a combination of increasing doses of vemurafenib (y)
and bosutinib or dasatinib (1 μM). IC50 values are represented by black squares. B Three-dimensional spheroid growth of a 50%/50% mix of SK28R WT (in red) and KO cells (in green) over 5 days. Cells were treated or not with SRC inhibitors: bosutinib (1 μM) or dasatinib (1 μM) every 2 days (n = 3 independent technical experiments, mean AE s.d.). Comparisons with control (DMSO) were performed using unpaired t-tests with the Sidak-Bonferroni method, P < 0.001 ***. C Three-dimensional spheroid growth of SK28R WT (in red) and KO cells (in green) over 5 days. Cells were treated or not with dasatinib (1 μM) in combination with ITE (5 μM) or TCDD (20 nM) every 2 days (n = 3 independent technical experiments, mean AE s.d.). Comparisons with control (DMSO) were performed using unpaired ttests with the Sidak-Bonferroni method, P < 0.05 *, P < 0.01 **. Comparisons were performed using unpaired t-tests with the Sidak-Bonferroni method. D Protein levels of AhR, p-SRC (Y416), and SRC in the SK28R cell line after 24 h of treatment with ITE (5 μM) or SRC inhibitor (dasatinib; 1 μM). E Evaluation of AhR transcriptional activity related to AhR/ARNT binding sites (XRE) using p3xXRE-luciferase constructs. HaCat keratinocytes cells were exposed or not to 10 nM TCDD alone or in combination with increasing concentrations of dasatinib O/N (n = 3 independent technical experiments, mean AE s.d.). Comparisons with control (0) were performed using unpaired t-tests with the Sidak-Bonferroni method, P < 0.05 *, P < 0.001 ***. F Expression heatmap of the median of gene expression for the various signatures (left) and AhR target genes (right) from RNAseq datasets for SK28R cells treated or not for 24 h with SRC inhibitors: bosutinib (Bos, 1 μM) or dasatinib (Das, 1 μM). The scale corresponds to the Z scores.
Source data are available online for this figure. (EGFR). Indeed, we showed that activation of SRC leads to the reactivation of the EGFR after its phosphorylation (Y845) in BRAFiresistant melanoma cell lines (Fig EV5A-C). Such SRC-AhR cross talk has been previously described to mediate EGFR phosphorylation in colon and lung cancer cells (Xie et al, 2012;Ye et al, 2018) and to contribute to an aggressive phenotype in multiple human tumors (Biscardi et al, 1999). Accordingly, BRAFi resistance commonly correlates with a high level of EGFR expression and a poor prognosis (Luebker & Koepsell, 2019). In addition, EGFR activation after phosphorylation has been shown to be more highly associated with resistance and EMT transition (Gross et al, 2015), notably after reactivation of the ERK pathway. The direct role of AhR in regulating the phosphorylation of SRC (Y416) and EGFR (Y845) may promote together the acquisition of the aggressive/invasive EMT like phenotype of BRAFi-resistant melanoma (Sato et al, 2003;Sato, 2013) (Synopsis). Consistent with these data, several therapeutic strategies using erlotinib or gefitinib have already been tested in preclinical studies to increase the sensitivity to BRAFi and decrease invasive abilities in melanoma (Sun et al, 2014;Notarangelo et al, 2017;Kenessey et al, 2018;Simiczyjew et al, 2019) (Fig EV5D). Together these results identify the central role of the AhR/SRC axis in supporting nongenetic cell reprogramming of melanoma cells exposed to targeted therapy. The AhR/SRC axis orchestrates cell plasticity, constituting an important therapeutic vulnerability. It warrants future clinical studies targeting the AhR-dependent SRC/ FAK/EGFR axis in combination with BRAFi/MEKi double blockade to re-sensitize melanoma to standard melanoma treatment and counteract resistance.

Cell culture and reagents
Human melanoma cell lines (SK28, 501Mel, M229, and M238) were grown in humidified air (37°C, 5% CO 2 ) in RPMI-1640 medium (Thermo Fisher Scientific, Invitrogen, Waltham, MA, USA) supplemented with 10% fetal bovine serum (Eurobio, Les Ulis, France) and 1% penicillin-streptomycin antibiotics (Thermo Fisher Scientific). SK28 (S + R) cells were obtained from J.C Marine at the VIB Center for Cancer Biology, VIB, Leuven, Belgium. M229 cells were obtained from Graeber's lab at the UCLA Molecular Biology Institute, Los Angeles, CA, USA. 501Mel cells (S) were obtained from the ATCC and 501Mel BRAFi-resistant cells (R) were obtained after 3 months of treatment with Vem (1 μM every 2 days). No difference of proliferation has been observed between resistant cells and parental ones.

CRISPR/Cas9 experiments
The AhR knockout was performed using CRISPR/Cas9 methodology. The guide sequence targeting AhR (Sigma-Genosys, St Louis, MO, USA) was cloned into the GeneArt CRISPR Nuclease vector according to the manufacturer's instructions (Life Technologies, Saint-Aubin, France). Next, 501Mel or SK28 cells were transfected with the vectors and the cells seeded 2 days later in 96-well plates at 0.5 cells/well for single-cell clonal expansion. The clones of interest were validated by DNA-sequencing, western blot analysis, and RT-qPCR.
A PDX model MEL006R (BRAFi resistant) was implanted in NMRI nude mice. Mice with tumors reaching 200 mm 3 were treated daily with vehicle (n = 6), dabrafenib alone (Dab, Biorbyt, 30 mg/kg, n = 5) dasatinib alone (Das, Selleckchem, 30 mg/kg, n = 9) or in combination dabrafenib + dasatinib (Das before combo, 30 mg/kg, n = 12). B PDX tumor volumes were measured every 2 days until reaching 1,500 mm 3 . Values correspond to the mean AE s.e.m. Statistical analysis (two-way ANOVA) has been performed between the different experiments (Das or Das before combo vs. Dab) at different times. Stars represent time from when tumor size is significantly lower than group treated with BRAFi alone (Dabrafenib) P < 0.05 *, P < 0.01 **. C Kaplan-Meier survival curve for MEL006R mice treated with the different drugs. Comparison of survival curves have been performed using the nonparametric Logrank (Mantel-Cox) test. D Number of days to reach max tumor volume (endpoint point size). Values correspond to the median AE s.d. Non-parametric Mann-Whitney t-test for the different treatments was performed compared with vehicle ***P < 0.001.

Cell density evaluation
Cell density was assessed using a methylene blue colorimetric assay. Briefly, cells were fixed for at least 30 min in 95% ethanol. Following ethanol removal, the fixed cells were dried and stained for 30 min with 1% methylene blue dye in borate buffer. After four washes with tap water, 100 μl 0.1 N HCl was added to each well. Plates were then analyzed with a spectrophotometer at 620 nm.

Wound healing migration assay
Briefly, cells were grown until confluent in 2-well silicone inserts (Ibidi ® , Germany) placed in 12-well tissue culture dishes. The cell culture inserts were removed after 1 day. Afterward, the plates were washed with PBS and incubated at 37°C in fresh RPMI-1640 medium (Gibco BRL, Invitrogen, Paisley, UK) supplemented with 10% fetal bovine serum (Eurobio) and 1% penicillin-streptomycin antibiotics (Gibco, Invitrogen), either naive or in the presence of vehicle (DMSO) or CH-223191 (5 μM). The wound was photographed with an inverted microscope at 5× magnification using an Axio Vert.A1 inverted microscope (Carl Zeiss). Wound closure was determined by measuring the distance between the edges of the wound at time 0 and 15 h using ImageJ (Fiji). Quantification of the distance migrated by the cells was performed as follows: D ¼ size of the woundatt ¼ 0hÀsize of the woundatt ¼ 15h ð Þ :

Spheroid formation assay
The spheroid formation assay was performed as previously described. Cells (20,000 cells/ml) were plated in 24-well plates coated with 1.5% agarose in complete RPMI medium and concentrated in the center by circular agitation. After 2 days, spheroids were recovered for inclusion in an extracellular matrix of collagen (100 μl; final concentration = 2 mg/ml in buffer (0.01 N acetic acid; neutralization buffer: 33 mM Hepes pH 7.4, 0.37% sodium bicarbonate, 0.03 N NaOH; 1× MEM)) in 24 well-plates coated with 1.5% agarose. Spheroids were maintained in complete medium with or without CH-223191 (5 μM) and images of the spheroids captured over several days (0-4 days) using an Axio Vert.A1 inverted microscope (Carl Zeiss) at 5× magnification. Invasion capacity was evaluated by determining the ratio between the maximum and initial diameter of the spheroid.

RNA extraction and RT-qPCR expression
RNA extraction & RT-qPCR expression was performed as previously described (Corre et al, 2018). The sequences of the primers used for the RT-qPCR experiments are available in Appendix Table S4.

Western blotting
Harvested cells were solubilized as previously described. Protein samples were denatured at 95°C, resolved by SDS-PAGE, and transferred onto Hybond™-C Extra nitrocellulose membranes (Amersham Biosciences, Bucks, UK). Membranes were probed with the appropriate antibodies Appendix Table S5 and the signals detected using a Fujifilm LAS-3000 Imager (Fuji Photo Film, Tokyo, Japan). Primary antibody information is available in Appendix Table S5. Horseradish-peroxidaseconjugated secondary antibodies were purchased from Jackson Immu-noResearch (Suffolk, UK) and used at a dilution of 1:10,000.

Immunoprecipitation
SKMel28-resistant cells were collected and lysed with cell lysis buffer (20 mM Tris-HCl, pH 8, 150 mM NaCl, 0.5 M EDTA) for 30 min on ice. The supernatant was incubated with 40 μl of protein G magnetic beads and 10 μl of SRC antibody (2108, Cell Signaling) or of normal rabbit IgG (sc2027, Santa Cruz Biotechnology) overnight at 4°C under rotation. Beads were washed with lysis buffer and eluated with 30 μl of Laemmli. Immunoprecipitates were analyzed by western blotting with appropriate antibody.

Luciferase activity
HaCat keratinocytes (2.10 5 ) were cultured in 12-well plates and transfected with the pGL3-XRE3-Luc construct carrying firefly luciferase. Transient transfection of cells was performed as previously described (Corre et al, 2018). After a 24-h period, cells were exposed to TCDD (10 nM) in the presence of increasing concentration of dasatinib (0.5-5 μM) overnight. Luciferase assays were then performed using a Promega kit according to the manufacturer's instructions. Data are expressed in arbitrary units relative to the value of luciferase activity levels found in DMSO-exposed cells, arbitrarily set to 1 arbitrary unit (a.u.). Firefly luciferase activity was normalized to protein content using the Bicinchoninic Acid Kit from Sigma-Aldrich ® and measured using a luminometer CLARIOStar (BMG Labtech).

Proximity ligation assay
The proximity ligation assay was used to visualize AhR/SRC complexes in SK28 cells. The cells, grown on glass coverslips, were fixed with 4% PFA in 0.1 M phosphate buffer (15735-60S, Electron Microscopy Sciences) for 15 min at RT and PLA performed using the Duolink ® in Situ detection Reagent Orange (DUO92007), Duolink ® in Situ PLA ® Probe Anti-Mouse PLUS (DUO92001), and Duolink ® in Situ PLA ® Probe Anti-Rabbit MINUS (DUO92005), SIGMA kits according to the manufacturer's protocol. After blocking, the reaction was performed with the primary antibodies: mouse anti-AhR (C20, 1/100) and rabbit anti-SRC (1C12, 1/100). Following the ligation and amplification steps, the coverslips were immobilized on microscopic slides using mounting medium containing DAPI. The ligation step was omitted in the control. Imaging analysis was carried out using a delta vision system (Applied Precision). The number of foci was quantified for at least 30 cells.

Patient-derived xenografts
In collaboration with TRACE and after approval by the University Hospital KU Leuven Medical Ethical Committee (S54185) and written informed consent from the patient, PDX model MEL006R (BRAFi resistant) was established from an in-transit metastasis resected as part of standard-of-care melanoma treatment at the University Hospital KU Leuven. The procedures involving mice were performed in accordance with the guidelines of the IACUC and KU Leuven and carried out within the context of approved project applications P147/2012, P038/2015 and P098/2015. Fresh tumor tissue was collected in transport medium (RPMI1640 medium supplemented with penicillin/streptomycin and amphotericin B). Tumor fragments were subsequently rinsed in phosphate-buffered saline supplemented with penicillin/streptomycin and amphotericin B and cut into small pieces of approximately 3 × 3 × 3 mm 3 . Tumor pieces were implanted in the interscapular fat pad of female SCID-beige mice (Taconic). After reaching generation 4 (F4), tumor fragments were implanted in the interscapular fat pad of female NMRI nude mice (8 weeks, Taconic). Ketamine, medetomidine and buprenorphine were used for anesthesia. Because tumor growth of the BRAFi-resistant PDX model (Mel006R) is very fast, this model is probably not appropriate to study the mechanisms of invasion and metastasis at least in the time window analyzed (20-60 days).

Pharmacologic treatment of mice
Mice with tumors reaching 200-300 mm 3 were treated via daily oral gavage. Dabrafenib (Biorbyt) and/or dasatinib (Selleckchem) were dissolved in DMSO at a concentration of 30 mg/ml respectively, aliquoted and stored at −80°C. Each day a new aliquot was diluted 1:10 with phosphate-buffered saline and mice were treated with a dose of 30 mg/kg for dabrafenib alone, with dasatinib alone (30 mg/kg) or with the combination dabrafenib + dasatinib (30 mg/kg each) after a pretreatment with dasatinib for 16 days. Tumor volume was monitored with a caliper and calculated using the following formula: V = (π/6)*length*width*height. The endpoint of the experiment corresponds when tumor volume reaches 1500 mm 3 according to ethical statements.

RNA-Seq
Total RNA was extracted from BRAFi-sensitive or resistant SK28, Mel501, and M229, cells before and after knockout out of AhR using the NucleoSpin RNA kit (Macherey Nagel, Dü ren, Germany). A complementary DNA library was prepared and sequencing performed according to the Illumina standard protocol by Beijing Novel Bioinformatics Co., Ltd. (https://en.novogene.com/). RNAseq was performed in collaboration with Novogene (Beijing, China). Libraries were generated from 500 ng total RNA using a Truseq Stranded mRNA kit (Illumina). The concentration of the library was first determined using a Qubit2.0 fluorimter and then diluted to 1 ng/μl. The size of the insert was checked using an Agilent bioanalyzer and further quantified by qPCR (library concentration > 2 nM). An aliquot (0.5 nM) of the pool was loaded on a high-output flow cell and sequenced on a NovaSeq 6000 instrument (Illumina) with 2 × 150 bp paired-end chemistry in two runs. Reads were aligned to human genome release hg38 using HISAT2 V2.0.5 with default parameters. Quantification of the expressed genes was performed using CUFFDIFF v2.2.1. The quality of the RNA-Seq count data was assessed using the Novogene standard protocol. The RNA-Seq data presented in this article was submitted to the Gene Expression Omnibus database (http://www.ncbi.nlm.nih.gov/geo/) under the accession number (GSE166617).

Data mining
TCGA/SKCM RNAseq data were analyzed using the OncoLnc portal [http://www.oncolnc.org] (Anaya, 2016). The raw data count matrix, composed of 454 samples (from SKCM melanoma cohort), was downloaded from the OncoLnc portal for the various transcriptional signatures. Expression heatmaps of differentially expressed genes between samples were obtained based on a log2 fold change using the ComplexHeatmap 2.0.0 (Gu et al, 2016) package in R/Bioconductor. Cluster-specific gene rankings were obtained by contrasting the samples with the rest. Cell density curves for the available melanoma cell lines were established using GraphPad PRISM 9.0 ® to establish the IC50 for the various treatments. The raw data count matrices from the RNA seq data were obtained in GEO database for previous experiments on melanoma The paper explained Problem Despite the considerable improvement made in the management of patients with metastatic BRAF V600 mutated melanoma, the vast majority of patients treated with BRAFi experience disease progression. Understanding the resistance mechanisms that support tumor progression is mandatory to overcome this process and to propose new therapeutic options.

Results
We show here that the ligand-activated transcription factor AhR drives cell plasticity, switching non-invasive and BRAFi-sensitive melanoma cells into invasive and resistant cells. AhR operates through genomic reprogramming and through the activation of the SRC kinase pathway. In vitro and in vivo use of SRC inhibitors in combination with BRAFi resensitize resistant melanoma cells to BRAFi treatment.

Impact
This study shows the AhR/SRC axis constitutes a therapeutic vulnerability in BRAFi-resistant melanoma, opening new therapeutic perspectives for BRAFi-resistant patients.
Analysis of the RNAseq dataset from the GDSC (Sanger/Massachusetts General Hospital Genomics of Drug Sensitivity in Cancer; Yang et al, 2013) was performed and recovered from the CellMi-nerCDB webtool (https://discover.nci.nih.gov/cellminercdb; Reinhold et al, 2012). CellMinerCDB is an interactive web application that simplifies access to and exploration of cancer cell line pharmacogenomic data from different sources. This webtool allows the comparison of molecular and/or drug response patterns across sets of cell lines to search for possible associations. Pearson's correlations with the reported p-values (not adjusted for multiple comparisons) between AhR expression (Appendix Fig S7) and drug activity (297 compounds) were recovered for various cancer cell lines (n = 1,080).

Statistics
Data are presented as the mean AE SD, unless otherwise specified, and differences were considered significant for a P value < 0.05. Comparisons between groups normalized to a control were carried out using a two-tailed t-test, with the Holm-Sidak multiple comparisons test when more than two groups are compared with the same control condition. Overall survival was estimated using the Kaplan-Meier method. Univariate analysis using the Cox regression model was performed to estimate the hazard ratios (HRs) and 95% confidence intervals (CI). All statistical analyses were performed using GraphPad (PRISM9.0 ® ; La Jolla, CA, USA).

Data availability
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request. RNA-Seq data: Gene Expression Omnibus GSE166617 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc= GSE166617).
Expanded View for this article is available online.