JAK1/STAT3 Activation through a Proinflammatory Cytokine Pathway Leads to Resistance to Molecularly Targeted Therapy in Non–Small Cell Lung Cancer

This article is featured in Highlights of This Issue, p. 2045

The emergence of molecularly targeted therapies against driver mutations and gene fusions has resulted in significant therapeutic advances for a subset of non–small cell lung cancer (NSCLC) patients (1). However, even when targeted therapy is effective, patients eventually develop resistance; thus more effective therapies that overcome this acquired resistance are being sought (2). A variety of mechanisms are believed to contribute to intrinsic and acquired resistance to anticancer drugs, especially molecularly targeted drugs. In oncogene-driven cancers, nongenetic mechanisms have been widely reported in addition to the genetic mechanisms such as gatekeeper mutations. For example, growth factors secreted from the tumor microenvironment promote proliferative or antiapoptotic signaling and protect cancer cells from targeted drugs (3). Furthermore, cancer cells produce several factors, including heregulin, amphiregulin, and hepatocyte growth factor, that contribute to intrinsic and acquired resistance to EGFR inhibitors in NSCLC (4–6).

Several molecularly targeted drugs, including EGFR tyrosine kinase inhibitors (TKI) and MET inhibitors, promote activation of STAT3 signaling (7, 8). Lee and colleagues reported the relation between proinflammatory cytokine IL6 and resistance to targeted drugs (9). They showed that inhibition of MEK functioning downstream of various receptor tyrosine kinases, including EGFR, MET, ALK, and HER2, triggers the feedback activation of STAT3 through IL6 secretion, significantly contributing to resistance to pathway-targeted drugs. The family of IL6 cytokines comprises IL6, IL11, oncostatin-M (OSM), leukemia-inhibitory factor (LIF), ciliary neurotrophic factor, cardiotrophin-1, and cardiotrophin-like cytokine. These cytokines activate target genes associated with cell differentiation, survival, apoptosis, and proliferation (10). Among this family, IL6, OSM, and LIF are the most widely expressed in different organs and are associated with cancer progression (11). These proinflammatory cytokines have individual receptors (e.g., IL6R, OSMR, and LIFR), which generally work as heterodimers with IL6ST (gp130; refs. 12, 13). These cytokines are reported to be robust stimulators of STAT3 and to be strong promoters of epithelial-to-mesenchymal transition (EMT), cancer metastasis, and cancer stem cells (CSC) in several types of cancer (14).

In this study, we describe for the first time that activation of the other members of IL6 family proinflammatory cytokine pathway, in particular the OSM–OSMR duo, can contribute to resistance to molecularly targeted drugs in oncogene-driven NSCLC cells. In addition, an inhibitor of Janus kinase 1 (JAK1), a key mediator of IL6 cytokine pathway activation in NSCLC cells, effectively reversed resistance to targeted drugs in these cells. Our data suggest that the OSMRs/JAK1/STAT3 axis contributes to resistance to targeted drugs in oncogene-driven NSCLC cells, implying that this pathway could be a therapeutic target.

We used 6 human oncogene-driven NSCLC cell lines, all provided by Dr. Adi F. Gazdar (The University of Texas Southwestern Medical Center, Dallas, TX) and coauthor (J.D. Minna). For a control, we utilized the nonmalignant human bronchial epithelial cell line HBEC3KT (15). The identities of all cell lines were confirmed by short tandem repeat (STR) DNA fingerprinting using the Promega 16 High Sensitivity STR Kit. We used normal lung fibroblasts (NLF) obtained from a normal lung specimen and cancer-associated fibroblasts (CAF) obtained from a lung cancer specimen in cocultures to mimic the tumor microenvironment. These fibroblasts were used in passages 5 through 7.

Targeted inhibitors selumetinib (AZD6244; ref. 16), erlotinib, crizotinib, filgotinib (GLP0634; ref. 17), momelotinib (CYT387) (18), and Stattic (19) were purchased from Selleckchem. Recombinant human (rh) IL6, OSM, and LIF proteins were purchased from EMD Millipore.

Cell proliferation was quantified by a modified MTS assay with CellTiter 96 AQueous One Solution Reagent (Promega) as previously reported (20). For experiments testing the effect of proinflammatory cytokines, JAK1, or STAT3 on cell proliferation, the crystal violet staining or MTT dye reduction method (Sigma) was used. The percentage growth is shown relative to untreated controls. Each experiment was performed at least in triplicate, and three times independently.

PCR amplification was conducted on an ABI Real-Time PCR 7900 HT (Applied Biosystems) and gene expression was calculated by the comparative C_t method. Three replicates per sample were assayed for each gene. To quantify the relative changes in gene expression, the 2(-ΔΔC_t) method was used and reactions were normalized to endogenous control gene GAPDH expression level for mRNA expression analysis and to miR-374 expression level for miRNA expression analysis.

The primary antibodies were as follows: anti-STAT3, p-STAT3 (Tyr705), IL6ST (Gp130), E-cadherin, Vimentin (R28), ZEB1, Snail, Axl, AKT, p-AKT (Ser473), p44/p42 mitogen-activated protein kinase (ERK), p-p44/p42 MAPK (p-ERK1/2), JAK1, p-JAK1, JAK2, p-JAK2, JAK3, Tyk2 (a JAK family menber), PARP, and c-PARP (Cell Signaling Technology) and IL6R, OSMR, and LIFR (Santa Cruz Biotechnology). Monoclonal anti-actin antibody, used as an equal loading control, was purchased from EMD Millipore. Secondary antibodies were goat anti-rabbit and goat anti-mouse horseradish peroxidase–conjugated immunoglobulin G (Santa Cruz Biotechnology). To detect specific signals, the membranes were examined by Western Lightning Plus-ECL (Perkin-Elmer).

A Human Phospho-Kinase Array Kit (R&D Systems) was used to measure the relative level of phosphorylation of 46 distinct intracellular kinases according to the manufacturer's instructions.

For RNA interference assays, cells were transfected with Silencer Select siRNA against OSMR, LIFR, JAK1, JAK2, JAK3, Tyk2, or scrambled negative control siRNA. The cells were placed in 60-mm dishes and incubated for 72 hours and then treated with rhIL6, rhOSM, or rhLIF (1 ng/mL) for 1 hour. Cell lysates were analyzed by Western blotting for levels of the indicated proteins.

HBEC3K, NLF, and CAF cells were seeded at 2 × 10⁵ cells per 60-mm cell culture dish and incubated for 72 hours. For measurement of IL6 concentration, cell culture media were collected and subjected to an IL6 ELISA kit (R&D Systems) according to the manufacturer's instructions. For measurement of OSM, cells were subjected to lysis in cell lysis buffer (Sigma) analyzed with an OSM ELISA kit (R&D Systems).

Tumor cells were treated or not treated with targeted drugs and placed in the bottom chambers of transwell polycarbonate membrane cell culture inserts (2 × 10⁴ cells per chamber). NLFs or CAFs (2 × 10⁵) were placed in the top chambers; the bottom and top chambers were separated by 0.4-μm pore size filters (Corning).

A total of 855 patients from four clinical datasets, including The Cancer Genome Atlas (TCGA, n = 515), Profiling of Resistance patterns and Oncogenic Signaling Pathways in Evaluation of Cancers of the Thorax (PROSPECT, n = 180), and the Biomarker integrated Approaches of Targeted Therapy for Lung Cancer Elimination [BATTLE-1 (n = 57) and BATTLE-2 (n = 103)], were analyzed as independent datasets. Clinical and demographic information for 515 mostly early-stage, surgically resected lung adenocarcinomas included in TCGA dataset was obtained from the TCGA portal (https://gdc.cancer.gov/). The PROSPECT dataset includes tumor tissue collected from patients who underwent surgical resection of lung adenocarcinoma with curative intent between 1996 and 2008 at MD Anderson Cancer Center. BATTLE-1 was a randomized phase II, single-center, biomarker-based, biopsy-mandated clinical trial of targeted therapy in patients with advanced chemorefractory NSCLC (NCT00409968) (21). BATTLE-2 was a randomized phase II, multi-center, biomarker-integrated, biopsy-mandated trial of targeted therapy in patients with advanced chemorefractory NSCLC (NCT01248247) (22). Clinical characteristics of these samples are described in Supplementary Table S1.

Statistical analyses were conducted by using GraphPad Prism 6 (GraphPad Software) or the R system for statistical computing. Differences between groups were calculated by the t test. Univariate Cox proportional hazards regression analysis was utilized to assess the association of OSMR on overall survival (OS), recurrent-free survival (RFS), and progression-free survival (PFS). On the basis of univariate analysis, variables with P value less than 0.2 were initially included in a multivariable model in PROSPECT cohort. Backward elimination of multivariable Cox regression model was conducted to eliminate variables that were insignificant. Survival curves were plotted by using the Kaplan–Meier method and analyzed by using the log-rank test. Spearman test was used to evaluate a correlation between each cytokine and receptor expression level. A P < 0.05 was considered significant.

From three of the oncogene-driven NSCLC cell lines, we established resistant cell lines by long-term exposure to a specific targeted drug according to a previously reported method (20). Cell line H23-SR was established by exposing H23 KRAS-mutant (-mut) cells to MEK inhibitor selumetinib; HCC4006-ER by exposing HCC4006 EGFR-mut cells to EGFR-TKI erlotinib; and H3122-CR by exposing H3122 EML4-ALK fusion cells to ALK inhibitor crizotinib. After more than 4 months of drug exposure, each of these cell lines showed resistance to the pertinent drug (Fig. 1A). IC₅₀ values against each targeted drug were as follows: H23 and resistant H23-SR, 31.2 μmol/L and >100 μmol/L; HCC4006 and resistant HCC4006-ER, 0.034 μmol/L and > 10 μmol/L; and H3122 and resistant H3122-CR, 0.47 μmol/L and 9.9 μmol/L, respectively. After acquisition of resistance to each drug, all 3 cell lines showed transcriptional upregulation of IL6 family cytokine receptor genes IL6R, IL6ST, OSMR, and LIFR (Fig. 1B). Western blotting showed upregulation of basal IL6 family cytokine receptors, except for IL6ST in H3122 cells, and their downstream protein phospho-STAT3 (Y705) (p-STAT3) in each acquired resistance cell line compared with the parental cell line (Fig. 1C). While the reason was unclear, we could not detect the IL6ST protein in both H3122 and H3122-CR cells. As already reported, selumetinib and erlotinib treatment did not suppress p-STAT3 expression in the parental cells, while crizotinib did suppress p-STAT3 expression in H3122 cell (9, 23). Interestingly, 72 hours after targeted drug treatment, all acquired resistance cells showed further overexpression of p-STAT3 (Fig. 1C).

We next analyzed expression levels of the IL6 family cytokines after treatment with a targeted drug in the 6 parental NSCLC cell lines. We treated KRAS-mut cell lines H23 and H2009 with selumetinib, EGFR-mut cell lines HCC4006 and H3255 with erlotinib, and EML4-ALK fusion cell line H3122 and SLC34A2-ROS1 fusion cell line HCC78 with crizotinib for 24 hours, and analyzed IL6 family cytokine gene expression levels by qRT-PCR. Crizotinib is a MET and ALK kinase inhibitor that is also reported to have activity against ROS1 (24, 25). IL6 and OSM gene expression levels were upregulated by each drug treatment, while LIF modest increased expression varied (Fig. 1D), suggesting that exposure to these targeted drugs caused an activation of the proinflammatory cytokine pathway preponderantly mediated by IL6 and OSM in these cells.

Interestingly, all three acquired resistance cell lines showed changes suggestive of EMT, including downregulation of E-cadherin and upregulation of vimentin, ZEB1, Snail, and Axl, on Western blotting (Fig. 1E). These acquired resistance cells also showed morphologic changes to spindle shapes (Supplementary Fig. S1). Because we previously reported that long-term EGFR-TKI exposure caused EMT with downregulation of the miR-200 family in EGFR-mut NSCLC cells, we analyzed the expression status of the miR-200 family by qRT-PCR (20) and showed that miR-200a/b/c were downregulated in all acquired resistance cells with EMT features (Fig. 1F).

To confirm the functional effect of IL6 family cytokines on their downstream pathway in NSCLC cells, we stimulated NSCLC cells with each of three IL6 family cytokines (rhIL6, rhOSM, or rhLIF) and analyzed phospho-kinase activation. Whole-cell extracts from H23 cells treated or not treated with rhIL6, rhOSM, or rhLIF for 48 hours were incubated with a human phospho-kinase array to compare kinase's phosphorylation induced by each of the IL6 family cytokines (Fig. 2A). Phosphorylation of several kinases, including AKT (S473), ERK1/2, STAT3 (Y705 and S727), p53, and c-JUN (S7), was observed after cytokine treatment. Among the three cytokines, rhOSM was the most potent activator of these oncogenic molecules. Furthermore, the 6 parental NSCLC cell lines were cultured with each cytokine for various time intervals under serum-starved condition and analyzed by Western blotting (Fig. 2B; Supplementary Fig. S2). The rhOSM treatment induced marked STAT3 (Y705) phosphorylation in all cell lines. AKT (S473) and ERK1/2 (T202/Y204) were phosphorylated by rhOSM in all 6 cell lines even under serum-starved conditions.

To determine which molecule preferentially mediates STAT3 phosphorylation after treatment with these proinflammatory cytokines, we knocked down OSMR, LIFR, and JAK family members with siRNA in 23 cells and treated the knockdown cells with each cytokine (Fig. 2C). Sufficient knockdown efficacy by each siRNA was confirmed by Western blotting (Supplementary Fig. S3). In rhIL6-stimulated cells, STAT3 phosphorylation was blocked by JAK1 knockdown. In rhOSM-stimulated cells, STAT3 phosphorylation was blocked by coknockdown of OSMR and LIFR or by knockdown of JAK1. In rhLIF-stimulated cells, STAT3 phosphorylation was blocked by knockdown of LIFR or JAK1. Taken together, knockdown of JAK1 blocked the STAT3 phosphorylation stimulated by each of the three cytokines, while knockdown of the other JAK members (JAK2, JAK3, and TYK2) could not block STAT3 phosphorylation. In addition, STAT3 phosphorylation by rhOSM was mediated by both OSMR and LIFR, as previously reported (13). These results suggest that JAK1 is the critical kinase that phosphorylates STAT3 via stimulation by IL6 family cytokines in NSCLC cells. Interestingly, when we knocked down JAK1, expression of OSMR was also downregulated (Supplementary Fig. S3), suggesting the possibility of feedback activation of OSMR by JAK1 kinase.

OSM has been reported to enhance migration of several kinds of cancer cells (26–32). We also confirmed that rhOSM treatment resulted in EMT-associated changes in H23 cells, as evinced by repression of the epithelial marker E-cadherin and induction of the mesenchymal markers vimentin, ZEB1, Snail, and Axl, on Western blotting (Fig. 2D). rhOSM treatment caused morphologic change to a spindle shape (Supplementary Fig. S4). In addition, rhOSM treatment caused EMT-related epigenetic changes, with suppression of the miR-200 family and let-7 family members as previously reported (Fig. 2E; ref. 28). On the other hand, rhOSM did not cause major effect on cell viability (Supplementary Fig. S5). These EMT-related changes induced by rhOSM treatment were prevented by siRNA-mediated codownregulation of OSMR and LIFR or downregulation of STAT3 (Fig. 2F; Supplementary Fig. S6). In contrast, downregulation of JUN, previously reported to be an EMT-promoting factor in cancer cells (33), could not protect cells from rhOSM-induced EMT-related changes. These results suggest that rhOSM-induced EMT and cell survival are mediated by OSMR/LIFR-STAT3 activation (Fig. 2G).

To analyze the effect of OSM/STAT3 pathway activation on sensitivity to molecularly targeted drugs in oncogene-driven NSCLCs, we treated the 6 NSCLC cell lines with molecularly targeted drugs with or without 2 weeks of pretreatment with rhOSM in cell culture medium. Two cell lines harboring KRAS-mut (H23 and H2009) were treated with selumetinib, 2 cell lines harboring EGFR-mut (H3255 and HCC4006) were treated with erlotinib, and 2 cell lines harboring EML4-ALK fusion H3122 and SLC34A2-ROS1 fusion HCC78 were treated with crizotinib. In all cells, rhOSM treatment protected from targeted drug-induced cell death (Fig. 3A).

We then hypothesized that inhibition of the OSM receptors (OSMR)/JAK1/STAT3 pathway with each oncogenic pathway-targeted treatment could prevent drug resistance caused by OSM pathway activation. To explore JAK/STAT3 inhibitor activities, we treated H2009 cells with novel selective JAK1 inhibitor filgotinib, JAK1/2, and TBK1/IKKϵ inhibitor momelotinib, or STAT3 inhibitor Stattic (Fig. 3B). Filgotinb and momelotinib potently suppressed JAK1 and STAT3 phosphorylation in a dose-dependent manner. Interestingly, these JAK/STAT3 inhibitors suppressed OSMR expression as well.

To determine whether cotargeting of a major oncogenic pathway and OSMRs/JAK1/STAT3 pathway molecules could efficiently inhibit drug resistance caused by rhOSM, we treated cells with agents targeting each pathway in combination with siRNA against OSMR/LIFR, JAK1, and STAT3, and selective JAK1 inhibitor filgotinib. The reason we chose to use filgotinib, which is in an ongoing clinical trial for the treatment of rheumatoid arthritis and Crohn disease (17, 34), over momelotinib, which is in an ongoing clinical trial in combination with trametinib (NCT02206763) or erlotinib (NCT02206763) for the treatment of NSCLC, is that momelotinib would have had off-target activities (35). In other words, by using filgotinib instead of momelotinib, we were able to ensure more complete inhibition of JAK1 activation at doses that did not exhibit off-target effects. As shown in Fig. 3C, all 6 cell lines treated with rhOSM showed resistance to targeted drugs, and siRNA-mediated downregulation of OSMR/LIFR, JAK1, and STAT3 prevented the targeted drug resistance caused by OSM treatment. In particular, siRNA-mediated JAK1 inhibition showed a potent antitumor effect in combination with targeted drugs. Filgotinib also inhibited the resistance in combination with each targeted drug. Western blotting showed that, after 48 hours of treatment, expression of apoptosis marker cleaved (c)-PARP was suppressed by rhOSM treatment, while cotargeting of the oncogenic pathway by a targeted drug and of JAK1 by filgotinib inhibited p-STAT3 and OSMR and induced strong expression of c-PARP (Fig. 3D).

Because CAFs reportedly influence tumor progression and are associated with resistance to molecularly targeted drugs via the secretion of extracellular matrix proteins, proteases, cytokines, and growth factors, we analyzed the effect of cross-talk of cancer cells and fibroblasts on resistance to targeted drugs (3, 36, 37). Both NLFs and CAFs showed high expression levels of IL6 and OSM mRNAs compared with normal bronchial epithelial HBEC3KT cells by qRT-PCR, and high concentration of the IL6 protein in cell culture media or OSM in whole-cell lysates by ELISA (Fig. 4A). Cancer cells cocultured with fibroblasts in cell culture inserts showed upregulation of p-STAT3, OSMR, and LIFR, and downregulation of E-cadherin, suggesting that the OSMRs/STAT3 pathway was activated by the fibroblasts (Fig. 4B). In coculture with CAFs, cancer cells were highly resistant to each targeted drug (Fig. 4C).

In cancer cells treated with a targeted drug in combination with filgotinib (Fig. 4D), a single treatment with filgotinib had a weak to moderate antitumor effect, while cotargeting of each oncogenic pathway and JAK1 by targeted drugs and filgotinib prevented drug resistance caused by coculturing with CAFs. Targeted drug monotherapy suppressed apoptosis marker c-PARP in cancer cells cocultured with CAFs, while cotreatment with a targeted drug and filgotinib reversed this c-PARP suppression while including suppression of p-STAT3 and OSMR (Fig. 4E).

To explore whether cotargeting the oncogenic pathway and OSMRs/JAK1/STAT3 pathway molecules could be effective in ameliorating acquired resistance to targeted drugs, we treated cells with each molecularly targeted drug in combination with siRNA against OSMR/LIFR, JAK1, or STAT3, or in combination with filgotinib. As shown in Fig. 5A, siRNA-mediated downregulation of OSMR/LIFR, JAK1, or STAT3 in combination with a targeted drug was effective in both parental and acquired resistance cell lines. Importantly, filgotinib also reduced resistance to targeted drugs in combination with a molecularly targeted drug. After 48 hours of treatment, expression of p-STAT3 and OSMR was suppressed in both parental and acquired resistance cells that received cotargeted treatment, and expression of c-PARP was restored in acquired resistance cells (Fig. 5B).

To explore the correlation of each IL6 family cytokine and receptor in clinical lung adenocarcinoma samples, we analyzed gene expression levels of clinical samples in 515 TCGA lung adenocarcinoma samples. Figure 6A shows the correlation of gene expression level between each IL6 family cytokine and receptor in TCGA samples. The gene expression levels of each IL6 family cytokines (IL6, OSM, and LIF) have positive correlation, in particular, between IL6 and OSM (Spearman ρ = 0.56).

Kaplan–Meier curves analysis of surgically resected 180 patients in PROSPECT with lung adenocarcinoma show that prognosis in the OSMR-high (upper median) patients was significantly poorer than that in the OSMR-low (lower median) patients (P = 0.0096 for recurrent-free survival; Fig. 6B). Univariate Cox proportional hazards regression analysis in surgically resected lung adenocarcinoma samples (TCGA and PROSPECT) and chemorefractory advanced lung adenocarcinoma samples (BATTLE-1 and BATTLE-2) was shown in Supplementary Table S2. Interestingly, OSMR gene expression may be a prognostic factor in patients with surgically resected lung adenocarcinoma: TCGA [OS, hazard ratio (HR): 1.18, 95% confidence interval (CI): 1.00–1.40; P = 0.050], and PROSPECT (RFS, HR: 1.25; 95% CI, 1.01–1.55; P = 0.039), whereas it was not significant in patients with chemorefractory advanced lung adenocarcinoma. In multivariable Cox regression model of PROSPECT patients adjusting for age, tumor size, neoadjuvant treatment, pathologic stage, and taking into consideration of interaction between neoadjuvant treatment and pathologic stage, OSMR expression is an independent prognostic factor for RFS (HR: 1.25; 95% CI, 1.01–1.55; P = 0.045; Supplementary Table S3).

In this study, we found that activation of the IL6 family proinflammatory cytokine pathway in the cancer microenvironment can contribute to resistance to molecularly targeted drugs in oncogene-driven NSCLC cells. Our findings suggest that inhibition of MEK, functioning downstream of various receptor tyrosine kinases, triggers the feedback activation of STAT3 through not only IL6, as shown by Lee and colleagues, but also OSM upregulation in cancer cells, significantly contributing to resistance to pathway-targeted therapies (9). STAT3 has been reported to be a major contributor to inflammation and also has been shown to promote the acquisition of treatment resistance (38). In most cells, STAT3 activation can suppress apoptosis through regulation of the expression of various cell survival gene products, including BCL-2, BCL-xl, MCL-1, survivin, and cIAP2 (38). However, effective targeted therapies generally induce apoptosis in oncogene-addicted cancers. This process involves the downstream BCL-2 family of proteins, which act as guardians of mitochondria-mediated apoptosis (39). Therefore, activation of STAT3 directly affects sensitivity to targeted drugs in oncogene-addicted cancers. These findings are compatible with our findings that activation of STAT3 by rhOSM treatment or coculture with CAFs suppressed expression of apoptosis marker c-PARP induced by targeted drugs and that coinhibition of major oncogenic pathways by targeted drugs and STAT3 by a JAK1 inhibitor restored c-PARP expression.

There seem to be several possibilities for how targeted drugs upregulate IL6 or OSM expression in cancer cells. Ishiguro and colleagues reported that IL6 promoter contains an activator protein-1 (AP-1) binding site and that EGFR-TKI treatment activated IL6 promoter transcription via regulation of AP-1 activation in cancer cells (40). Interestingly, OSM promoter also contains an AP-1 binding site (41). Thus, AP-1 activation by targeted drugs might be one of the mechanisms for IL6 or OSM upregulation. Indeed, the gene expression levels of IL6 and OSM has positive correlation in TCGA samples.

In addition to autocrine activation of STAT3 in NSCLC cells triggered by targeted drugs, tumor stromal CAFs secreted IL6 and OSM and also contributed to resistance to targeted drugs in cancer cells. Lapeire and colleagues reported that cancer-associated adipose tissue promotes breast cancer progression by paracrine OSM secretion and STAT3 activation (26). Caffarel and colleagues reported that the main source of OSM in cancers seems to be the tumor stroma, suggesting the existence of paracrine signaling between tumor stroma and cancer cells (42). At the same time, they mentioned that OSM production has been observed in several kinds of cancers, including breast and prostate cancers, indicating the potential for additional autocrine signaling (42). The stromal cells along with CAFs may give rise to local inflammation in the cancer microenvironment through IL6 and OSM secretion, thus contributing to resistance to targeted drugs. Therefore, reduction of inflammatory pathway activation in cancer cells by JAK1/STAT3 inhibition may improve the cells' sensitivity to pathway-targeted drugs.

We found that JAK1 is a critical mediator of STAT3 activation by IL6 family cytokines in NSCLC cells. Many cytokines relying on JAK family members for intracellular signal transduction have key pathophysiologic roles in autoimmune and inflammatory diseases (43). Our findings indicate that STAT3 activation by IL6 family cytokines could be effectively blocked by JAK1 inhibition in NSCLC cells. Therefore, we investigated the combination efficacy of combining pathway-targeted drugs and selective JAK1 inhibitor filgotinib in STAT3-activated cells resistant to targeted drugs. In clinical trial, the combination of JAK1/2 inhibitor ruxolitinib with erlotinib did not demonstrate dramatic effect in patients with acquired erlotinib resistance in recent study (44), while the sample size was small and the results were still controversial. On the other hand, filgotinib is an oral selective JAK1 inhibitor, and showed encouraging safety and efficacy in clinical studies in patients with rheumatoid arthritis (17, 34, 45). Selective inhibition of JAK1 also may offer favorable safety and clinical efficacy in cancer treatment. Interestingly, JAK1/STAT3 pathway inhibition by siRNA or filgotinib suppressed not only p-STAT3 but upstream OSMR protein expression. Although the mechanisms that cause this suppression are still unclear, inhibition of the JAK1/STAT3 pathway could block the feedback activation of OSMR.

We found that OSMR expression in surgically resected samples appeared to portend worse prognosis for patients with NSCLC. The OSM/OSMR pathway closely relates to inflammation, hematopoiesis, and development (46). Activation of this pathway is increasingly recognized as an important contributor to progression of many types of cancer, including lung cancer (27, 29, 32, 47, 48). In breast cancer, high OSMR expression correlated with shorter recurrence-free and overall survival duration and with chemotherapy resistance in estrogen receptor–negative tumors (27). To our knowledge, the current study is the first report that OSMR could be a possible prognostic marker in patient with surgically resected NSCLC. In contrast, OSMR expression did not correlate with prognosis significantly in patients with chemorefractory advanced stage NSCLC. Although the detailed mechanism remains unclear, one possible explanation is that most of heavily treated chemorefractory NSCLC samples could harbor basal upregulation of inflammatory-related pathways, including OSM/OSMR pathway. The previous treatment caused inflammation in the cancer microenvironment and upregulation of OSMR in a subset of tumors, which could become resistant to targeted drugs.

Our study has some limitations. One is the inability to confirm the JAK1/STAT3 pathway activation associated with IL6 family cytokines upregulation in clinical samples that showed acquired resistance to targeted drugs. To confirm these pathway activations and to obtain a possible new biological profile that can be a treatment target, the comparative analysis of samples before and after acquired resistance to targeted therapies is required.

In summary, our findings indicate that activation of the OSMRs/JAK1/STAT3 pathway contributes to resistance to targeted drugs in oncogene-driven NSCLC cells. These findings suggest that cotreatment targeting the OSMRs/JAK1/STAT3 pathway could further improve the antitumor efficacy of oncogenic pathway inhibition by targeted therapies.

No potential conflicts of interest were disclosed.

Conception and design: K. Shien, R.S. Herbst, I.I. Wistuba, J.G. Izzo

Development of methodology: K. Shien, L. Shen, R.S. Herbst, J.D. Minna, I.I. Wistuba, J.G. Izzo

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K. Shien, V.A. Papadimitrakopoulou, C. Behrens, N. Kalhor, X. Tang, R.S. Herbst, J.D. Minna, J.M. Kurie

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K. Shien, V.A. Papadimitrakopoulou, D. Ruder, C. Behrens, L. Shen, N. Kalhor, J. Song, J.J. Lee, J. Wang, X. Tang, R.S. Herbst, L. Girard, I.I. Wistuba, J.G. Izzo

Writing, review, and/or revision of the manuscript: K. Shien, V.A. Papadimitrakopoulou, D. Ruder, J. Song, R.S. Herbst, S. Toyooka, J.D. Minna, I.I. Wistuba

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): I.I. Wistuba

Study supervision: J.G. Izzo

The authors would like to thank the Department of Scientific Publications, the University of Texas MD Anderson Cancer Center, for the English language review.

This work was supported by the NIH NCI grants R01 CA155196 (to V.A. Papadimitrakopoulou, R.S. Herbst, and I.I. Wistuba) and 2P50CA070907-16A1 (to J.D. Minna and I.I. Wistuba). In addition, this work was partially supported by the UT Lung Specialized Programs of Research Excellence grant P50CA70907, and MD Anderson's Institutional Tissue Bank (ITB) Award Number 2P30CA016672 from the NIH NCI. K. Shien is supported by the Japan Society for the Promotion of Science (JSPS) Postdoctoral Fellowship for Research Abroad.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.