Activation of the Aryl Hydrocarbon Receptor Leads to Resistance to EGFR TKIs in Non-Small-Cell Lung Cancer by Activating Src-Mediated Bypass Signaling

Purpose: The aryl hydrocarbon receptor (AhR) has been generally recognized as a ligand-activated transcriptional factor that responds to xenobiotic chemicals. Recent studies have suggested that the expression of AhR varies widely across different cancer types and cancer cell lines, but its significance in cancer treatment has yet to be clarified. Experimental Design: AhR expression in non-small-cell lung cancer (NSCLC) was determined by Western blotting and IHC staining. In vitro and in vivo functional experiments were performed to determine the effect of AhR on sensitivity to targeted therapeutics. A panel of biochemical assays was used to elucidate the underlying mechanisms. Results: A high AhR protein level indicated an unfavorable prognosis for lung adenocarcinoma. Inhibition of AhR signaling sensitized epidermal growth factor receptor (EGFR) tyrosine kinase inhibitors (TKIs) in NSCLC cells that express high level of endogenous AhR protein. Notably, activation of AhR by pharmacological and molecular approaches rendered EGFR mutant cells resistant to TKIs by restoring PI3K/Akt and MEK/Erk signaling through activation of Src. In addition, we found that AhR acts as a protein adaptor to mediate Jak2-Src interaction, which does not require the canonical transcriptional activity of AhR. Conclusions: Our results reveal a transcription-independent function of AhR and indicate that AhR may act as a protein adaptor that recruits kinases bypassing EGFR and drives resistance to TKIs. Accordingly, targeting Src would be a strategy to overcome resistance to EGFR TKIs in AhR-activated NSCLC. Recent studies have suggested that AhR functions in multiple pathways outside of its canonical role in detoxification. Here, we described AhR as a novel therapeutic target for cases of NSCLC that expresses high levels of endogenous AhR. Activation of AhR promotes resistance to EGFR TKIs by activating Src-mediated bypass signaling. In this model of resistance, AhR acts as a protein adaptor to recruit Src and Jak2 kinases, rescuing PI3K/Akt and MEK/Erk signaling. Taken together, our results provide proof-of-principle insights into a novel resistance mechanism driven by AhR and suggests that targeting Src represents a new strategy to overcome resistance to targeted therapy. which bypasses EGFR to preserve PI3K/Akt and MEK/Erk signaling. We employed to assess the statistical significance of differences between different groups. A statistically significant difference was defined as * P < 0.05 or ** P 0.01. Y527F were treated with DMSO or 1 µmol/L gefitinib for 6 h and subjected to Western blot analysis. E. PC-9 AhR WT cells were treated with gefitinib and AhR ligands with or without the Src kinase inhibitors dasatinib and PP2 for 24 h. Cleavage of PARP and caspase-3 and phosphorylation of Src, Akt and Erk were determined by immunoblotting. F. Src expression in PC-9 AhR WT cells was knocked down by three independent shRNAs. The resulting cells were treated with DMSO or 1 µmol/L gefitinib for 6 h and subjected to Western blot analysis. phosphorylates Src to bypass mutant EGFR to restore PI3K/Akt and MEK/Erk signaling and elicits a TKIs resistant phenotype.


Introduction
For the past few decades, the aryl hydrocarbon receptor (AhR) has been extensively recognized as a ligand-activated basic helixloop-helix transcriptional factor that mediates metabolic response to environmental pollutants. In a ligand-free condition, AhR protein is primarily localized in the cytoplasm and complexed with two Hsp90 molecules, the Hsp90-interacting protein p23 and the AhR-interacting protein AIP (1). Ligand binding causes the receptor to dissociate from its partner proteins and leads to the release of the nuclear translocation signal (NLS) region, which triggers the shuttling of AhR protein from the cytoplasm to the nucleus and the formation of the AhR-ARNT (aryl hydrocarbon nuclear translocator) heterodimer. The AhR-ARNT complex binds to xenobiotic response elements (XRE) in dioxin-responsive gene promoters to induce transcriptional activation of phase I metabolic enzymes (2,3). However, recent studies have also suggested a transcription-independent function of AhR signaling. For example, AhR protein has been proposed to regulate the balance between Treg and Th17 cells, suppress dendritic cell immunogenicity and maintain immune cell homeostasis (4)(5)(6)(7). AhR protein also acts as a key component of the Cul-4B E3 ubiquitin ligase complex to mediate a nongenomic pathway regulating estrogen receptor destruction (8,9). The identification of kynurenine (Kyn) as an endogenous AhR ligand was a major impetus for in-depth investigation of AhR's previously unknown function in physiologic and pathologic events (10). Although recent studies have suggested a potential role of AhR in regulating cell proliferation, cell-cycle distribution, and cell apoptosis, the underlying mechanism is largely unknown (11).
The great success of targeted therapy in non-small cell lung cancer (NSCLC) is no doubt a model paradigm of precision medicine in cancer research and treatment. The clinical application of EGFR inhibitors, anaplastic lymphoma kinase (ALK) inhibitors, and ROS1 kinase inhibitors markedly extended the progression-free survival (PFS) of patients with genetically defined NSCLC (12)(13)(14). Unfortunately, targeted therapy failed to prolong the overall survival (OS) of these patients. This paradox may be attributed to the loss of responsiveness to the therapeutic drugs, known as secondary resistance or acquired resistance. Resistance mechanisms that have been identified include the emergence of resistance mutations (15) and bypass signaling (16,17). In the interest of overcoming resistance driven by these mechanisms, increasing efforts have been focused on the development of more potent tyrosine kinase inhibitors (TKIs) and combinational strategies, which ultimately suppress the major cell survival-related output of tyrosine kinases, such as PI3K/Akt and MEK/Erk signaling (18). However, the molecular mechanism of resistance is not fully understood; thus, elucidating resistance mechanisms would greatly help in guiding new treatment strategies to extend the clinical benefits of TKIs.
AhR is ubiquitously overexpressed in multiple cancer cell lines and tumor samples (19). In addition, the AhR repressor (AhRR) and miRNAs posttranscriptionally targeting AhR are predominantly suppressed in cancer (20)(21)(22). Direct evidence showing that AhR is an oncogenic driver gene was demonstrated by the finding that transgenic mice with constitutively activated AhR (AhR CA) developed gastric cancer and hepatocellular carcinoma (23,24). Although there is no direct evidence showing that the activation of AhR leads to NSCLC tumorigenesis, AhR has attracted increasing attention in this type of cancer. For example, overexpression of AhR has been detected in more than 50% of patients with NSCLC, and polymorphisms of AhR gene are implicated in the development of NSCLC (25)(26)(27)(28). Activation of AhR signaling upregulates several growth factors, and NSCLC cells require elevated AhR for sustained proliferation (29,30). Moreover, the AhR gene encodes several resistance proteins, drug metabolic enzymes and drug transporters, and it also negatively regulates the rate of apoptosis (31)(32)(33)(34). It is plausible that AhR is implicated in determining the sensitivity of NSCLC to apoptosis-inducing agents, such as targeted therapeutics. In this study, we demonstrated that activation of AhR signaling leads to resistance to EGFR TKIs. Once activated, the AhR protein acts as an adaptor to recruit Src, which bypasses EGFR to preserve PI3K/Akt and MEK/Erk signaling. We therefore propose AhR and Src kinase as novel therapeutic targets to overcome resistance to EGFR TKIs in NSCLC.

Constructs, mutants, and lentivirus production
pLKO.1-shRNAs against AhR, ARNT, cyp1a1, cyp1b1, Src, and Jak2 were obtained from the National Institute of Biological Sciences (NIBS, Beijing, China; Supplementary Table S1). shRNA lentiviruses were produced in HEK293 cells according to the standard protocol. Cells were infected with the indicated lentiviruses and selected with 2 mg/mL puromycin for 2 weeks. To generate doxycycline-inducible shRNAs, we cloned complementary DNA oligos for AhR into the pSingle-tTS-shRNA vector (Clontech). Cells were transfected with the pSingle-tTS-AhR shRNA (Dox-on/AhR shRNA) plasmid and selected with 100 mg/mL G418. To conditionally knock down AhR expression, we added doxycycline at a final concentration of 1 mg/mL to cell culture medium supplemented with 10% doxycycline systemapproved FBS (Clontech).
The AhR ORF with an N-terminal Flag tag was inserted into the PstI/XhoI restriction sites of the pcDNA4 vector. The AhR WT DNLS, AhR CA, AhR CA DNLS, AhR WT DSH2, DAA2-200, DAA201-400, DAA401-600, and DAA601-800 mutants were generated with a site-directed mutagenesis kit (Supplementary  Table S2). To generate an AhR lentivirus, we subcloned the Flagtagged AhR fragment into a pWPI vector (laboratory of Feng Shao at NIBS, Beijing, China) at the PmeI restriction site. Lentivirus production, cell infection, and selection were performed as described elsewhere. The Myc-tagged Jak2 plasmid was generated by standard PCR amplification and was sequenced in full. HA-tagged Src WT and Src Y527F mutants were obtained from Siyuan Zhang (University of Notre Dame, Notre Dame, IN).

RNA isolation and qPCR
RNA extraction and cDNA synthesis were performed following the standard protocol. qPCR was performed with SYBR Premix Ex Taq (Takara). The relative amount of each gene transcript was normalized to the amount of b-actin transcript. The qPCR primer sequences are listed in Supplementary Table S3.
Protein extraction, Western blot, kinase array, and immunoprecipitation Total cell protein was extracted using RIPA lysis buffer (Pierce) supplemented with a protease inhibitor cocktail (Roche). For the extraction of membrane protein, the cell pellet was treated with Mem-PER Membrane Protein Extraction Reagent (Pierce) following manufacturer's instructions. Protein concentration was

Translational Relevance
Recent studies have suggested that AhR functions in multiple pathways outside of its canonical role in detoxification.
Here, we described AhR as a novel therapeutic target for cases of NSCLC that express high levels of endogenous AhR. Activation of AhR promotes resistance to EGFR TKIs by activating Src-mediated bypass signaling. In this model of resistance, AhR acts as a protein adaptor to recruit Src and Jak2 kinases, rescuing PI3K/Akt and MEK/Erk signaling. Taken together, our results provide proof-of-principle insights into a novel resistance mechanism driven by AhR and suggest that targeting Src represents a new strategy to overcome resistance to targeted therapy. measured with a bicinchoninic acid kit (Clontech), and the protein was boiled in SDS-PAGE loading buffer. The kinase array was performed with a Phospho Kinase Array Kit (R&D Systems, #ARY003B) following the manufacturer's instructions. Briefly, a total quantity of 300 mg of whole-cell lysate (WCL) was incubated with nitrocellulose membranes containing 43 phosphorylated kinase antibodies printed in duplicate overnight at 4 C on a rocking platform shaker. The membranes were washed with Wash Buffer, and then incubated at room temperature with Detection Antibody Cocktail A/B for 2 hours and Streptavidin-HRP Reagent for 30 minutes at room temperature. The phosphorylated protein signal was developed with Chemi Reagent Mix.
An immunoprecipitation assay was performed as described elsewhere. Briefly, the lysates were precleared and incubated with the indicated primary antibodies overnight with gentle rotation at 4 C. The immunocomplex was pelleted with Protein A/G agarose beads (Santa Cruz Biotechnology), then resuspended, boiled in SDS-PAGE loading buffer, and subjected to Western blot analysis.
Equal amounts of all protein samples were loaded and run on SDS-PAGE gel, transferred onto a nitrocellulose membrane (Millipore), blocked with skim milk, and incubated with the indicated antibodies overnight. The membrane was washed with TBST and incubated with HRP-conjugated IgG. Protein bands were visualized by enhanced chemiluminescence (Millipore). The source and dilution of antibodies were listed in Supplementary Table S4.

Xenograft models
Approximately 5 Â 10 6 cells were suspended in 100 mL of Matrigel (BD Biosciences) and subcutaneously injected into the flanks of 6-week-old athymic nude mice (Laboratory Animal Center of Fourth Military Medical University, Xi'an, China; FMMU). When the tumors reached appropriate size, the indicated treatments were initiated. Tumor growth was monitored and recorded every three days. At the end of the experiment, the mice were humanely sacrificed, and the tumors were carefully isolated and processed for histologic studies. All the animal experiments were conducted in compliance with institutional guidelines and approved by the Animal Care and Use Committee of FMMU (Xi'an, China).

Bioinformatic analysis and patient samples
To assess the association between AhR mRNA expression and NSCLC grade, metastasis, and prognosis, we performed bioinformatics analysis following the guidance of the TCGA database.
Fifty-five patients who underwent surgical resection of lung adenocarcinoma from 2005 to 2013 were followed up and retrospectively screened for AhR protein expression. Four paired samples of primary lung adenocarcinoma (not overlapping with the cases used for the tissue array) and metastatic lymph nodes were obtained from patients undergoing lung cancer surgery at Tangdu Hospital (Fourth Military Medical University, Xi'an, China). Another five patients with EGFR-mutant NSCLC who relapsed on EGFR TKIs and underwent repeated biopsy were also enrolled. Expression of AhR protein was determined by IHC and Western blotting. The experiment using human specimens was approved by the Ethics Committee of FMMU. A written document illustrating the experimental design and purpose was sent to each participant so that informed consent could be obtained.

Immunofluorescent staining and IHC
After the indicated treatments, cells were fixed with 4% paraformaldehyde and permeabilized with Triton X-100. The cells were then incubated with primary antibodies overnight at 4 C, then with Cy3-conjugated IgG for 2 hours at room temperature. The cells were counterstained with DAPI, and fluorescent signal was detected by a confocal laser microscope (Nikon).
The IHC and TUNEL assays were performed as described previously. The tumors were cut into 5-mm-thick sections, which were mounted on slides and incubated with the indicated antibodies or TUNEL reagents (Roche). The slides were then thoroughly washed and visualized with DAB or FITC. The AhR IHC staining density was scored as negative (score 0), weakly positive (score 1), moderately positive (score 2), and strongly positive (score 3). The percentage of cancer cells positive for AhR expression was assigned a proportion score (<5%, score 0; 6%-25%, score 1; 26%-50%, score 2; 51%-75%, score 3; >75%, score 4). To quantitative evaluate AhR protein expression, we multiplied the intensity score by the proportion score to yield the H-score. The median H-score was set as the cut-off value to separate NSCLC specimens with low and high AhR expression.

Statistical analysis
Values were expressed as the mean AE SD. The software programs GraphPad Prism 5 and Origin 6.1 were used to analyze the data. A paired t-test and one-way ANOVA were employed to assess the statistical significance of differences between different groups. A statistically significant difference was defined as Ã , P < 0.05 or ÃÃ , P < 0.01.

AhR protein predicts poor prognosis and constitutes a therapeutic target for NSCLC
To investigate whether AhR correlates with NSCLC prognosis, we initially examined AhR expression at the mRNA level in the TCGA database. AhR gene amplification was more frequent in NSCLC, especially in lung adenocarcinoma, than in SCLC and reached approximately 4% overall frequency ( Supplementary  Fig. S1A). However, AhR mRNA was not associated with lung adenocarcinoma stage, lymph node infiltration, metastasis, or recurrence ( Supplementary Fig. S1B). Consistently, AhR mRNA did not correlate with lung adenocarcinoma patients' survival ( Supplementary Fig. S1C). To investigate whether AhR at the protein level carried prognostic value, we examined surgically resected lung adenocarcinoma samples from 57 patients for whom up to 8 years of follow-up data were available. IHC staining of normal lung tissues was performed to assess the baseline endogenous expression of AhR protein. As a positive control, human placenta tissues from two individuals were used. The patients were well balanced in terms of their characteristics, which are listed in Table 1. Twenty-six of the 57 cases (46%) showed a high level of AhR protein expression ( Fig. 1A; Supplementary  Table S5), which is consistent with previous IHC studies reporting AhR overexpression in 40% to 50% of NSCLC cases (25)(26)(27)30). Positive AhR staining was prominently observed in the nuclei of the NSCLC cells (Fig. 1A), indicating the activation of AhR signaling in these patients. Correlation analysis revealed that high AhR expression was associated with an aggressive tumor phenotype (Fig. 1B). Strikingly, seven of these 26 AhR high patients had disease recurrence with distant metastasis, whereas none of the 31 AhR low patients developed local recurrence or metastasis (Supplementary Table S5). Western blot analysis of another four paired primary lung adenocarcinomas and metastatic lymph nodes showed that AhR protein tended to increase more sharply in the metastatic region (Fig. 1C) than the primary region. Log-rank analysis further revealed that high expression of AhR protein was associated with reduced overall survival in this small cohort (Fig. 1D).
We next analyzed AhR expression in NSCLC cell lines (HCC827, PC-9, H358, H292, A549, SPC-A-1, and H1975). The human normal bronchoalveolar epithelial cell line Beas-2B was used. Western blot analysis showed a significant increase in AhR expression in SPC-A-1 and H1975 cells, whereas its expression in HCC827, PC-9, H358, H292, and A549 cells was comparable to that in Beas2B cells (Fig. 1E). These NSCLC cells were further divided into AhR high (SPC-A-1 and H1975 cells) and AhR low (HCC827, PC-9, H358, H292, and A549 cells) subtypes, according to their endogenous levels of AhR. Notably, the AhR high cells also exhibited elevated expression of AhR downstream genes, including cyp1a1 and cyp1b1, while the AhR low cells did not ( Supplementary Fig. S1D). To determine the biological significance of AhR in NSCLC cell proliferation and survival, we treated the AhR high and AhR low cells with the AhR antagonist a-naphthoflavone (a-NF). Interestingly, the AhR high cells were sensitive to a-NF, with an IC 50 less than 1 mmol/L. In contrast, the AhR low PC-9 cells were refractory to the same agent, with an IC 50 over 10 mmol/L ( Supplementary Fig. S1E). Moreover, activation of AhR signaling by its ligands tetrachlorodibenzo-p-dioxin (TCDD), b-naphthoflavone (b-NF), and Kyn promoted H1975 cell proliferation ( Supplementary Fig. S1F) and rendered growth independent of anchorage ( Supplementary Fig. S1G), indicating that AhR is a tumor-promoting signaling molecule in NSCLC.
Persistent PI3K/Akt and MEK/Erk signaling activation is prevalent in tumors and contributes to cancer cell survival and outgrowth. Indeed, inhibition of EGFR and ALK signaling by corresponding TKIs efficiently abolished the phosphorylation of Akt and Erk, thereby leading to cancer cell apoptosis. In contrast, selective inhibition of either pathway failed to elicit an apoptotic response ( Supplementary Fig. S1H). To characterize the underlying molecular alterations following AhR inhibition in NSCLC, we treated cells with increasing concentrations of a-NF and evaluated the phosphorylation levels of Akt and Erk by immunoblotting. The AhR low PC-9 cells were resistant to a-NF, and the treatment had no effect on the phosphorylation of Akt or Erk (Fig. 1F). However, a-NF exhibited single-agent efficacy in PI3K/Akt and MEK/Erk signaling inhibition and apoptosis induction in the AhR high SPC-A-1 and H1975 NSCLC cells. Consistent with the observation that H1975 cells had higher levels of endogenous AhR, H1975 cells tended to be more sensitive to a-NF than SPC-A-1 cells, as the treatment induced more cleaved form of PARP and caspase-3 (Fig. 1F). This pharmacologic effect may not rely on interference with EGFR, as a-NF exerted minimal effects on EGFR phosphorylation in the three tested cell lines. Taken together, these results suggested that AhR is a potential druggable target in NSCLC, in which inhibition of AhR signaling predominantly leads to AhR high cell apoptosis by reducing Akt and Erk phosphorylation.

Inhibition of AhR sensitizes H1975 cells to EGFR TKIs
To confirm the data obtained from pharmacologic experiments, we next inhibited AhR expression in H1975 cells by use of shRNAs. In agreement with the potent oncogenic function of AhR, two independent AhR shRNAs significantly reduced H1975 cell proliferation ( Supplementary Fig. S2A). Western blot analysis showed that knockdown of AhR readily inhibited phosphorylation of Akt and Erk but not EGFR ( Fig. 2A). Moreover, conditional knockdown of AhR in H1975 cells led to apoptosis (Supplementary Fig. S2B), highlighting the pivotal role of AhR in maintaining survival and proliferation in this AhR high NSCLC cell line.
Because the AhR high H1975 cells also harbored an EGFR L858R/ T790M mutation, it is plausible that AhR cooperates with mutant EGFR to induce oncogenic dependence. Thus, simultaneous inhibition of both pathways is expected to lead to a complete abrogation of PI3K/Akt and MEK/Erk signaling and induce more profound apoptosis. As shown in Fig. 2B, afatinib and a-NF exerted single-agent efficacy in apoptosis induction; however, neither drug alone completely blocked PI3K/Akt and MEK/Erk signaling. Strikingly, their combination led to a complete blockage of Akt and Erk phosphorylation and a sharp increase in apoptosis induction. In further support of this notion, afatinib in doses as low as 0.1 mmol/L efficiently blocked PI3K/Akt and MEK/Erk signaling in AhR knockdown cells, whereas afatinib at 1 mmol/L failed to do so in shRNA control cells (Fig. 2C), suggesting that inhibition of AhR signaling sensitized H1975 cells to afatinib. This effect was recapitulated in vivo: Although H1975 GFP shRNA xenograft tumors were responsive to afatinib, the AhR shRNA xenograft tumors exhibited a much more dramatic response as early as day 6 and achieved nearly complete regression on day 21 (n ¼ 6 in each group, Fig. 2D). The AhR shRNA tumors also underwent a higher prevalence of apoptosis, as judged by TUNEL immunofluorescence (Fig. 2E). Together, these data point to inhibition of AhR signaling as a potential strategy to sensitize NSCLC cells to EGFR TKIs.

Dysregulation of AhR signaling leads to EGFR TKI resistance
Having established that inhibition of AhR signaling impaired the phosphorylation of Akt and Erk and sensitized H1975 cells to afatinib, we next investigated whether activation of AhR signaling in EGFR-mutant NSCLC would render resistance to EGFR TKIs. IHC staining of AhR protein was performed in five pairs of  T790M mutation or c-Met amplification as potential causes of resistance, while patient No. 5 also had increased AhR expression in the repeated biopsy samples (Fig. 3A).
To further investigate the biological consequences of AhR overexpression, we established cell lines stably overexpressing AhR. In comparison with a control cell line overexpressing GFP (termed PC-9 GFP cells), a cell proliferation assay showed accelerated growth of PC-9 cells overexpressing AhR (termed PC-9 AhR WT cells, Supplementary Fig. S3A); thus, we proposed that AhR signaling may rescue EGFR-mutant PC-9 cells from EGFR inhibition. Indeed, the PC-9 GFP cells remained responsive to gefitinib at nanomole potency, whereas the PC-9 AhR WT cells were insensitive to the same drug, as judged by a short-term cell viability assay (Fig. 3B) and a long-term colony formation assay (Fig. 3C). Furthermore, resistance driven by AhR was specific to TKIs because stable overexpression of AhR did not alter sensitivity to chemotherapeutics, such as cisplatin and paclitaxel (Supplementary Fig. S3B). To our initial surprise, cells overexpressing the constitutively activated form of AhR (termed PC-9 AhR CA cells) failed to show resistance to gefitinib ( Fig. 3B and C). This paradox led us to evaluate EGFR and downstream PI3K/Akt and MEK/Erk signaling across the tested cell lines. It was noted that overexpression of AhR WT efficiently increased the baseline levels of phosphorylated Akt and Erk without affecting the phosphorylation level of EGFR. However, overexpression of AhR CA marginally enhanced PI3K/Akt and MEK/Erk signaling (Fig. 3D). Consistently, gefitinib at 0.1 mmol/L readily abolished Akt and Erk phosphorylation in PC-9 GFP and PC-9 AhR CA cells, whereas Akt and Erk phosphorylation persisted after gefitinib treatment in PC-9 AhR WT cells (Fig. 3E).
Apart from overexpression, the hyperactivation of AhR signaling also results from elevated content of AhR ligands. To determine whether AhR ligand treatment also elicits a TKI-resistant phenotype, we treated H1975 cells with b-NF and Kyn. Although the most potent EGFR TKI, AZD9291, is designed to overcome the T790M-resistant mutation, AhR ligand preconditioning rendered H1975 cells resistant to AZD9291, as determined by PARP and caspase-3 cleavage (Fig. 3F). A consensus finding was the increase in Akt and Erk phosphorylation, which could not be fully inhibited by AZD9291 (Fig. 3F). Collectively, these data highlighted activation of AhR signaling as a resistance mechanism in NSCLC, presumably through the restoration of PI3K/Akt and MEK/Erk signaling.

AhR activates Src kinase bypass signaling independent of its transcriptional activity
To eliminate any possible "off-target" effect of the AhR ligand, we tested whether AhR was required for the maintenance of  AhR activation confers resistance to EGFR TKIs in NSCLC. A, Representative images of AhR staining in five pairs of EGFR-mutant treatment-na€ ve and EGFR-TKI-relapsed NSCLC samples. Scale bar ¼ 200 mm. B and C, Evaluation of the effects of AhR WT and AhR CA on sensitivity to gefitinib by short-term cell viability assay and long-term colony formation assay, respectively. D, AhR WT and AhR CA mutant were stably overexpressed in PC-9 cells; the resulting cells were lysed and probed with indicated antibodies. PC-9 GFP cells were used as an overexpression control. E, PC-9 cells overexpressing GFP, AhR WT or AhR CA were treated with increasing concentrations of gefitinib (0, 0.1, 0.5, and 1 mmol/L) for 6 hours and analyzed for EGFR and downstream signaling. F, H1975 cells were treated with 0.1 mmol/L AZD9291 for 24 hours. Two hours prior to AZD9291 treatment, cells were treated with 10 mmol/L b-NF or 100 mmol/L Kyn to activate AhR signaling. At the end of these treatments, cells were evaluated for apoptosis and EGFR signaling by Western blot.
PI3K/Akt and MEK/Erk signaling. To maximally reduce background phosphorylation signal, we cultured the H1975 GFP shRNA and H1975 AhR shRNA cells in FBS-deprived medium overnight and treated them with 10 nmol/L TCDD for 15 and 30 minutes, respectively. As shown in Fig. 4A, phosphorylation of Akt and Erk increased sharply with as little as 15 minutes of TCDD treatment, while this effect was entirely abolished in AhR-deficient cells. As expected, cyp1a1 and cyp1b1 mRNA transcription levels at this early time point were unchanged (Supplementary Fig. S4A). Confocal laser scanning of H1975 GFP shRNA cells showed a cytoplasmic distribution pattern of AhR protein following 15 and 30 minutes of TCDD treatment ( Supplementary Fig. S4B), suggesting that the transcriptional machinery may not be dispensable for increased Akt and Erk phosphorylation. Consistently, abolishing AhR's transcriptional activity by removal of the AhR NLS region or knockdown of ARNT still facilitated TCDD-induced Akt and Erk phosphorylation (Fig. 4B). Furthermore, shRNA-mediated knockdown of AhR target genes, such as cyp1a1 and cyp1b1, had minimal effects on pAkt and pErk induction (Fig. 4C). These results Activation of AhR activates Src independently of its transcriptional activity. A, H1975 cells carrying indicated shRNAs were treated with 10 nmol/L TCDD for 15 and 30 minutes. Levels of phosphorylated Akt, Erk, and Src were determined by Western blot analysis. B, HEK293 cells were transfected with Flag-tagged AhR WT DNLS mutant or shRNA against ARNT, treated with 10 nmol/L TCDD for 30 minutes, and subjected to Western blotting. C, H1975 cells stably carrying cyp1a1 or cyp1b1 shRNA were treated with 10 nmol/L TCDD for 12 hours and analyzed for Akt and Erk phosphorylation. D, PC-9 cells carrying mCherry or Src Y527F were treated with DMSO or 1 mmol/L gefitinib for 6 hours and subjected to Western blot analysis. E, PC-9 AhR WT cells were treated with gefitinib and AhR ligands with or without the Src kinase inhibitors dasatinib and PP2 for 24 hours. Cleavage of PARP and caspase-3 and phosphorylation of Src, Akt and Erk were determined by immunoblotting. F, Src expression in PC-9 AhR WT cells was knocked down by three independent shRNAs. The resulting cells were treated with DMSO or 1 mmol/L gefitinib for 6 hours and subjected to Western blot analysis. strongly indicated that AhR is crucial for TCDD-induced Akt and Erk phosphorylation, independently of its canonical transcriptional activity.
The nonreceptor tyrosine kinase Src has been reported as a common downstream node of multiple resistance pathways, and concurrently targeting Src overcomes resistance to targeted therapy in breast cancer and chronic myeloid cell leukemia (35,36). Although there was no difference in total Src protein expression across the tested cell lines, an impressive finding was that the AhR high cells had a considerable amount of phosphorylated Src at residue Tyr416 (Y416) while the AhR low cells did not (Fig. 1E). In agreement with this, activation of the AhR pathway by overexpression assay or ligand treatment readily increased Src Y416 phosphorylation (Fig. 3D), whereas AhR silencing compromised this effect (Fig. 2A). These findings imply that Src may be involved in AhR-driven resistance to EGFR TKIs in NSCLC. To address whether Src activation is sufficient to confer TKI resistance, we stably overexpressed its constitutively activated Src Y527F mutant in three representative genetically defined NSCLC cell lines (PC-9 cells harboring an EGFR delE746_A750 mutation, H3122 cells harboring an EML4-ALK rearrangement and HCC78 cells harboring an SLC34A2-ROS1 rearrangement). Compared with control cells overexpressing mCherry, the Src Y527F cells exhibited hyperphosphorylation of Src Y416 ( Fig. 4D; Supplementary Fig. S4C). The Src Y527F cells responded poorly to TKIs (gefitinib and crizotinib in this study), and the treatment failed to abolish PI3K/Akt and MEK/Erk signaling ( Fig. 4D; Supplementary  Fig. S4D). Thus, enhanced Src phosphorylation renders NSCLC resistant to targeted therapy, and it is plausible that Src is the intermediate signaling hub linking AhR to TKI resistance. Indeed, pharmacologic inhibition of the Src pathway diminished Akt and Erk phosphorylation and overcame gefitinib resistance in PC-9 AhR WT cells (Fig. 4E). shRNA-mediated Src silencing also sensitized cells to gefitinib and facilitated apoptotic response (Fig. 4F). Collectively, these data strongly indicated that enhanced Src phosphorylation is a critical event following activation of AhR signaling, which bypasses EGFR to restore PI3K/Akt and MEK/Erk signaling and contributes to EGFR TKI resistance.

Jak2 phosphorylates Src upon AhR activation
We have established that AhR restores Akt and Erk phosphorylation through activation of Src kinase; however, how Src becomes phosphorylated in cells with active AhR signaling remains to be elucidated. Src kinase can bind to protein adaptors that present the Y-X-X-N motif (Supplementary Fig. S5A). For example, hepatocyte growth factor (HGF) binding to the c-Met receptor tyrosine kinase leads Src to bind to Gab1, Gab2, Grb2, and other adaptors, thus initiating full activation of the c-Met signaling cascade (37,38). Alignment analysis of the AhR protein sequence indicated a highly conserved putative SH2 domain binding motif within its C-terminal proportion that may provide docking site(s) for Src recruitment (Fig. 5A). Immunoprecipitation assays with both endogenous AhR (Fig. 5B) and ectopically expressed AhR (Fig. 5C) showed that binding between AhR and Src was enhanced by TCDD, whereas this interaction was dramatically diminished upon removal of the SH2 domain binding motif (Fig. 5C). While AhR is not a protein kinase, Src Y416 phosphorylation cannot be a direct biological consequence of its binding to AhR. To investigate how Src became phosphorylated, we used phospho-protein kinase array profiling in AhR ligandtreated cells and vehicle-treated control cells. This analysis revealed increased phosphorylation of 17 proteins (!2-fold change) in the ligand-treated cells ( Fig. 5D; Supplementary Fig.  S5B). Consistent with our previous observation, the kinase array revealed hyperactivation of Src, Akt, and Erk after AhR ligand treatment. Strikingly, we found a significant increase in the phosphorylation of all the Stat family proteins, including Stat2, Stat3, Stat5b, and Stat6, in the AhR ligand-treated cells (Supplementary Fig. S5B). On the basis of these findings, we proposed that protein kinases upstream of the Stat pathway are very likely to mediate Src phosphorylation.
Numerous studies have shown that Stat proteins are primarily phosphorylated by Janus kinase 2 (Jak2, not represented in the kinase array) (39)(40)(41). Thus, we determined whether Jak2 mediated Src phosphorylation. We found that activation of AhR signaling readily increased Jak2 Y1007/1008 phosphorylation ( Supplementary Fig. S5C). Inhibition of Jak2 phosphorylation by ruxolitinib, a selective orally available small-molecule inhibitor evaluated in phase I/II clinical trials, abolished Src Y416 phosphorylation (Fig. 5E). Moreover, knockdown of Jak2 impaired TCDD-induced Src-Akt/Erk signaling in H1975 and PC-9 AhR WT cells ( Supplementary Fig. S5D). These findings indicate a causal relationship between Jak2 and Src phosphorylation; therefore, we speculated that targeting Jak2 would block AhR-induced Src phosphorylation and overcome TKI resistance. In PC-9 AhR WT cells, the combination of gefitinib and ruxolitinib resulted in cell apoptosis, whereas neither drug exerted singleagent efficacy in apoptosis induction ( Supplementary Fig. S5E). Although ruxolitinib was capable of inhibiting Src bypass signaling, the PI3K/Akt and MEK/Erk signaling pathways persisted owing to mutant EGFR. The combination of gefitinib and ruxolitinib efficiently blocked "EGFR pathway signaling" and "Jak2-Src bypass signaling," leading to the blockade of Akt and Erk ( Supplementary Fig. S5E). Taken together, these results strongly indicated that Jak2 is one of the upstream kinases responsible for Src phosphorylation. Targeting Jak2 would overcome TKI resistance in NSCLC activated by Src signaling.
Finally, to address the significance of AhR in mediating the Src-Jak2 interaction, we expressed Flag-tagged AhR, HA-tagged Src, and Myc-tagged Jak2 in HEK293 cells. AhR activation by TCDD readily increased Src binding to Jak2, whereas this interaction was lost upon AhR depletion (Fig. 5F). To map the region responsible for Jak2 binding, we transfected HEK293 cells with expression vectors for Myc-tagged Jak2 together with Flag-tagged full-length, DAA2-200, DAA201-400, DAA401-600, or DAA601-800 truncated forms of AhR. Immunoprecipitation of cell lysates with anti-Flag antibody and probing of the resultant precipitates with anti-Myc antibody revealed that TCDD triggered Myc-tagged Jak2 in association with substantial amounts of Flag-tagged AhR constructs, with the exceptions of DAA2-200 and DAA201-400 (Fig. 5G), indicating that AhR was implicated as an intermediate protein linking Src to Jak2 through its SH2 domain-binding motif and N-terminal segment of AA2-400, respectively.

Discussion
This study describes a novel mechanism of resistance to targeted therapy. We showed that activation of AhR signaling drives resistance to EGFR TKIs in NSCLC independently of its canonical transcriptional activity. In this model, AhR acts as a protein adaptor to mediate the cross-talk between Src and Jak2 kinases, which bypass EGFR to restore PI3K/Akt and MEK/Erk signaling (Fig. 6). These findings serve as a new paradigm of nongenomic effects of AhR and indicate a new strategy to overcome resistance to targeted therapy.
In addition to mediating the cytotoxicity of TCDD and other environmental pollutants, AhR protein also facilitates tumorigenesis through multiple mechanisms involving disruption of the cell cycle, evasion of apoptosis, and suppression of immune surveillance. Therefore, inhibition of AhR signaling by directly targeting the AhR protein or targeting its ligands is expected to induce tumor regression. Indeed, the AhR antagonist a-NF (a pseudoligand that binds to the Ah receptor to form a nonfunctional complex) exhibits single-agent efficacy in inhibiting cell-cycle progression (42) and inducing apoptosis in cancer cells expressing high level of endogenous AhR (as shown in this study). In addition, the indoleamine 2,3-dioxygenase (IDO) and tryptophan-2,3-dioxygenase (TDO) inhibitors that inhibit Kyn, an endogenous AhR ligand, potentiate chemotherapy and immunotherapy, and clinical trials evaluating their safety and efficacy in cancers are currently underway (43)(44)(45). Interestingly, the AhR high cells tend to have a much higher magnitude of Kyn synthase expression than the AhR low cells ( Supplementary   Fig. S1D), enabling constitutive activation and amplification of the IDO/TDO-Kyn-AhR signaling loop. Testing AhR expression and Kyn synthase activity may help to predict the responsiveness of cancers to IDO/TDO/AhR inhibitors and chemotherapy.
Our study implies a compelling role of AhR in determining the sensitivity of NSCLC to EGFR TKIs. This is a surprising finding because AhR is not a protein kinase and is not associated with resistance mutations, while activation of AhR restores PI3K/Akt and MEK/Erk signaling despite inhibition of EGFR. This is not an entire recapitulation of the c-Met amplification resistance mechanism because AhR lacks kinase activity and aberrant phosphorylation of Akt and Erk may not be a direct consequence of AhR activation. Our data reveal a rational explanation of how the kinase activity-defective AhR protein preserves PI3K/Akt and MEK/Erk signaling, in which AhR and targeted therapy resistance converge at the nonreceptor tyrosine kinase Src. As highlighted by Zhang and colleagues, activation of Src is a common biological event downstream of multiple resistance pathways; targeted therapy resistance in ErbB2-positive breast cancer and BCR-ABL-positive chronic myeloid cell leukemia can be overcome by concurrently targeting Src (35,36). We have found that AhR activates Src through Jak2 kinase. A, Alignment analysis of the AhR protein SH2 domain binding motif. B, After H1975 cells were treated with 10 nmol/L TCDD for 30 minutes, endogenous AhR and Src protein were coprecipitated and their interaction was determined by immunoblotting. C, Flag-tagged AhR WT or DSH2 mutant and HA-tagged Src were coexpressed in HEK293 cells. Cells were treated with 10 nmol/L TCDD for 30 minutes. Immunoblotting analysis of anti-HA tag precipitate was performed accordingly. D, Representative image of phospho-protein kinase array profiling of AhR-ligand-treated and non-AhR-ligand-treated H1975 cells. Phospho-proteins with more than 2-fold changes are numbered. E, H1975 cells were treated with increasing concentrations of ruxolitinib (0, 0.1, 0.5, and 1 mmol/L), an orally available Jak2 inhibitor, for 6 hours, and subjected to Western blot analysis. F, HEK293 cells were transfected with HA-tagged Src, Myc-tagged Jak2 and Flag-tagged AhR plasmids and treated with 10 nmol/L TCDD for 30 minutes. WCL was precipitated with anti-HA tag antibody and probed with the indicated antibodies. G, Expression plasmids for Myc-tagged Jak2 and Flag-tagged AhR or its deletion mutants were introduced into HEK293 cells. Cells were treated with 10 nmol/L TCDD for 30 minutes, and WCL was subjected to immunoprecipitation with anti-Flag tag antibody.
Src phosphorylation markedly increases following the activation of AhR signaling and that Src inhibition resensitizes AhR WToverexpressing cells to gefitinib. Thus, Src also emerges as a therapeutic target to overcome resistance driven by AhR in NSCLC. Although others have also demonstrated that AhR ligands treatment enhance Src phosphorylation (46)(47)(48), our study provides a proof-of-principle insight into the molecular machinery responsible for Src activation. In our model, the binding of AhR protein to a ligand recruits cytosolic Src protein in a transcriptionally independent manner. The AhR-Src complex transiently translocates to the cell membrane, where AhR provides docking sites for Jak2 to phosphorylate Src ( Fig. 6; Supplementary  Fig. S6A). This result is consistent with findings from recent studies reporting that Jak2 directly phosphorylates Src and impacts the behavior of cancer cells (49,50). Importantly, we further demonstrated that the AhR protein adaptor is essential for Jak2-Src interaction in NSCLC. In contrast to AhR WT, which is prominently localized in the cytoplasm, the AhR CA mutant does not require ligand treatment for translocation to the nucleus ( Supplementary Fig. S6B), leading to the loss of its ability to recruit cytosolic Src kinase. The removal of NLS resulted in redistribution of the AhR CA protein into the cytoplasm, enabling its binding to Src and the restoration of PI3K/Akt and MEK/Erk signaling ( Supplementary Fig. S6B and S6C). These distribution patterns explain the differential effects of AhR WT and AhR CA on gefitinib sensitivity. However, AhR protein membrane translocation is an immediate and transient response to ligand-induced signaling activation, Src phosphorylation persists regardless of subsequent AhR protein nuclear translocation. One possible explanation for this discrepancy is the structure and properties of Src kinase: once Src Y416 phosphorylation is triggered by upstream kinases, it undergoes autophosphorylation of this tyrosine residue and persists in Src signaling activity (51). Thus, AhR-mediated Src phosphorylation by Jak2 is an initiation signal for constitutively active Src signaling that bypasses EGFR to restore PI3K/Akt and MEK/Erk signaling (Supplementary Fig. S6D). Again, this places Src at a common signaling node for targeted therapy resistance in NSCLC and may have important clinical implications. The availability of Src and Jak2 inhibitors makes it possible to overcome EGFR TKI resistance in patients whose tumors express high endogenous levels of AhR and depend on Jak2/Src kinase activity for persistent PI3K/Akt and MEK/Erk signaling activation.
Taken together, our findings highlighted a novel resistance mechanism and identified Src activation as the key resistance output of AhR signaling. More importantly, more treatment options may be discovered by identifying protein adaptors, such as AhR, that catalyze persistent PI3K/Akt and MEK/Erk signaling rather than solely focusing on the alteration of kinases. It will be interesting to determine whether AhR activation also contributes to resistance to ALK TKIs, ROS1 TKIs, and other targeted therapeutics in NSCLC.

Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.

Figure 6.
A schematic diagram of AhR protein adaptor-mediated Jak2-Src interaction. In the AhR OFF condition, the AhR protein is localized in the cytoplasm and forms an inactive complex with its partner proteins, including Hsp90, AIP, and p23. Cell-survival-related PI3K/Akt and MEK/Erk signaling in EGFR-mutant NSCLC could be blocked by the EGFR TKI gefitinib. In contrast, activation of AhR signaling (AhR ON) leads to conformation changes of the AhR protein, which results in enhanced binding with the nonreceptor Src kinase. The AhR protein transiently translocates to the cell membrane and acts as an adaptor to mediate the cross-talk between Jak2 and Src. Jak2 phosphorylates Src to bypass mutant EGFR to restore PI3K/Akt and MEK/Erk signaling and elicits a TKI-resistant phenotype.