FGFR1-ERK1/2-SOX2 axis promotes cell proliferation, epithelial–mesenchymal transition, and metastasis in FGFR1- amplified lung cancer

 

Kaixuan Wang ● Wenxiang Ji ● Yongfeng Yu ● Ziming Li ● Xiaomin Niu ● Weiliang Xia ● Shun Lu

1 Shanghai Lung Cancer Center, Shanghai Chest Hospital, Shanghai Jiao Tong University, West Huaihai Road 241, Shanghai 200030, China

2 School of Biomedical Engineering and Med-X Research Institute, Shanghai Jiao Tong University, Huashan Road 1954, Shanghai 200030, China

 

 

Abstract

Epithelial–mesenchymal transition (EMT) is an important process for cancer metastasis, drug resistance, and cancer stem cells. Activation of fibroblast growth factor receptor 1 (FGFR1) was found to promote EMT and metastasis in prostate and breast cancers, but the effects and mechanisms in lung cancer was unclear. In this study, we aimed to explore whether and how activation of FGFR1 promotes EMT and metastasis in FGFR1-amplified lung cancer. We show that activation of FGFR1 by its ligand fibroblast growth factor 2 (FGF2) promoted proliferation, EMT, migration, and invasion in FGFR1- amplified lung cancer cell lines H1581 and DMS114, whereas inhibition of FGFR1 suppressed these processes. FGFR1 activation upregulated expression of Sry-related HMG box 2 (SOX2) by downstream phosphorylated ERK1/2; moreover, the upregulation of SOX2 by autophosphorylation variant ERK2_R67S plasmid transfection was not suppressed by FGFR1 inhibitor AZD4547 or MEK/ERK inhibitor AZD6244 in vitro. And SOX2 expression was also significantly upregulated in ERK2_R67S lentivirus-transfected stable cell lines in vivo. Overexpression of SOX2 promoted cell proliferation, EMT, migration, and invasion. Importantly, activation of FGFR1 could not promote these processes in SOX2-silenced stable cell lines. In orthotopic and subcutaneous lung cancer xenograft models, inhibition of FGFR1 suppressed tumor growth, SOX2 expression, EMT, and metastasis in vivo; however, these processes caused by SOX2-overexpressing stable cell lines were not suppressed by FGFR1 inhibition. Higher expression of FGFR1 and SOX2 were positively correlated, and both were associated with shorter survival in lung cancer patients. In conclusion, our findings reveal that activation of FGFR1 promotes cell proliferation, EMT, and metastasis by the newly defined FGFR1-ERK1/2-SOX2 axis in FGFR1-amplified lung cancer.

 

 

Introduction

Lung cancer is the leading cause of cancer morbidity and mortality, with a 5-year survival rate of mere 18% [1]. More than 85% of lung cancer is non-small cell lung cancer (NSCLC), which is subdivided into adenocarcinoma (LADC), squamous cell carcinoma (LSQCC), and large cell carcinoma [2]. The discovery of genomic alterations in kinase genes, such as EGFR mutation and ALK and ROS1 rearrangements, has changed the clinical care of LADC [3], and highlights molecular and genomic profiling and subtyping of lung tumors [4–8]. However, compared to LADC, only a few therapeutically druggable genome alterations have been found in LSQCC and small cell lung cancer (SCLC). Of these, fibroblast growth factor receptor 1 (FGFR1) amplification caused by the 8p12 amplicon appears in 20% of LSQCC, 5–7% of SCLC, and 1–3% of LADC [9–11]. FGFR1-amplified lung cancer cell lines have shown promising pre- clinical sensitivity to kinase inhibition [12–15]. Unfortunately, in phase I and II clinical trials, the overall response rates of selective pan-FGFR inhibitors AZD4547, BGJ398, and non- selective multitargeted FGFR inhibitor dovitinib are modest 8–11.5% [16–18]. Therefore, further characterizing the signaling paradigm and biological function of FGFR1 activation in FGFR1-amplified lung cancer is important.

Epithelial–mesenchymal transition (EMT) is a transition process of cell states from epithelial to mesenchymal, which is observed in embryonic development, tissue regeneration, organ fibrosis, and cancer progression and metastasis [19]. During EMT, cancer cells change the adhesion molecules and cell skeleton protein, resulting in migration and inva- sion, and tumor metastasis [20]. EMT is induced by pleio- tropic signaling factors and is regulated mainly by core transcription factors called EMT-TFs, such as SNAIL1, ZEB1, and TWIST [21]. Complex regulatory networks involving receptor tyrosine kinases, non-coding RNAs, and epigenetic and post-translational factors have been found in the modulation of EMT [20–22]. The EMT process has been linked to tumor metastasis, drug resistance, cancer stem cells, and immune escape [22–24].

The receptor tyrosine kinase family FGFRs are activated by different fibroblast growth factor (FGF) ligands, result- ing in phosphorylation and activation of extracellular signal-regulated kinase (ERK) and other parallel down- stream signaling pathways, modulating many physiological processes, including embryogenesis, tissue homeostasis, and angiogenesis [25]. Abnormal activation of FGFR sig- naling pathway can be caused by translocation, point mutation, and amplification of FGFR1 gene in a broad range of human malignancies, leading to tumorigenesis transformation and tumor progression [25–27]. Activation of FGFR1 was reported to initiate carcinogenesis and EMT of prostate cancer [28], and to participate in the EMT and metastasis processes of breast cancer [29]. Most recently, EMT and upregulation of FGFR1 have been observed in primary or secondary drug resistance to tyrosine kinase inhibitors in lung cancer [30]. However, whether and how FGFR1 promotes EMT in FGFR1-amplified lung cancer is still elusive.

Our recent studies have shown that activation of FGFR1 played essential roles in the maintenance of cancer stem cells, and in the resistance to apoptosis in FGFR1-amplified lung cancer [31, 32]. Moreover, Sry-related HMG box 2 (SOX2), an essential transcription factor of stem cells [33, 34], was unregulated after activation of FGFR1 [31].

Recently, Wang’s study found that SOX2 promoted TGFβ- induced EMT in NSCLC [35], and Kim reported that VEGFA induced expression of SOX2 and promoted EMT and metastasis in breast cancer [36]. On the basis of these results, we hypothesized that FGFR1 activation might promote EMT and metastasis through upregulation of SOX2 in FGFR1-amplified lung cancer.

Here, we identified a novel signaling pathway that acti- vation of FGFR1 upregulated expression of SOX2 by downstream phosphorylated ERK1/2. The newly defined FGFR1-ERK1/2-SOX2 axis promotes cell proliferation, EMT, migration, and invasion, and tumor metastasis in FGFR1-amplified lung cancer cell lines H1581 and DMS114 in vitro and in vivo.

 

 

Results

Activation of FGFR1 promotes proliferation, EMT, migration and invasion in FGFR1-amplified lung cancer cells, which can be suppressed by FGFR1 inhibition FGFR1-amplified lung cancer cell lines H1581 and DMS114 were used to explore the effects of FGFR1 activity on proliferation, EMT, migration, and invasion. Cell line authentication was conducted by short tandem repeat (STR) analysis and compared to DSMZ STR database (Supple- mentary Table S1). Knockdown effect of siFGFR1-1/2 was confirmed by western blot and qPCR, compared to control siRNA (Fig. 1a). Activation of FGFR1 by FGF2 (20 ng/mL) promoted cell proliferation in control siRNA group, but not in siFGFR1 group (Fig. 1b). Stimulation of FGF2 caused increment of pFGFR1 and downstream pERK1/2 in a time- dependent manner (Fig. 1c). Meanwhile, the epithelial marker E-cadherin was downregulated, and the mesenchy- mal markers N-cadherin and Vimentin were upregulated in line with FGFR1 activation (Fig. 1c). Furthermore, pFGFR1 and pERK1/2 induced by FGF2 was suppressed by pre- treatment of siFGFR1 (Fig. 1d) or FGFR1 inhibitor AZD4547 (1 μM; Supplementary Figure S1a). The transi- tion of EMT markers induced by FGFR1 activation was also reversed when FGFR1 signaling was inhibited (Fig. 1d, Supplementary Figure S1a). In the scratch, migration, and invasion Transwell assays, the cell migration and invasion promoted by FGF2 stimulation was also suppressed by FGFR1 inhibition (Fig. 1e and f, Supplementary Figure S1b and c). These results demonstrated that activation of FGFR1 promotes proliferation, EMT, migration, and invasion in FGFR1-amplified lung cancer cells.

 

FGFR1 activation upregulates SOX2 by FGFR1-ERK1/ 2-SOX2 axis in FGFR1-amplified lung cancer cells in vitro and in vivo

We further explored the modulating mechanism by which FGFR1 regulates SOX2. Activation of FGFR1 upregulated expression of SOX2 (Fig. 2a). Silencing of FGFR1 by siRNA downregulated expression of SOX2 (Fig. 2b).

AZD6244 is a potent, highly selective MEK1/2 inhibitor, and thus inhibits phosphorylation of ERK1/2 [37]. Fur- thermore, inhibition of FGFR1 by FGFR1 inhibitor AZD4547 or inhibition of downstream pERK1/2 by MEK/ ERK inhibitor AZD6244 both blocked the upregulation of SOX2 induced by FGF2 stimulation (Fig. 2c and d). The ERK2_R67S plasmid (Supplementary Figure S2c) gen- erates sustained autophosphorylation of ERK2 in a MEK1/ 2-independent manner [38]. To confirm the direct connec- tion between pERK1/2 and SOX2, H1581 and DMS114 cells were transfected with ERK2_R67S plasmid. Com- pared with the wild-type ERK2 plasmid and negative con- trol plasmid, ERK2_R67S plasmid improved the expression of pERK1/2 and SOX2 markedly (Fig. 2e). Meanwhile, the upregulation of SOX2 caused by ERK2_R67S plasmid transfection was not suppressed by FGFR1 inhibitor AZD4547 or MEK/ERK inhibitor AZD6244 (Fig. 2e). Furthermore, we used ERK2_R67S lentivirus to generate stable cell lines and established subcutaneous xenograft models (Fig. 2f), and the expression of SOX2 was also significantly upregulated by ERK2_R67S variant in vivo (Fig. 2g). Thus, activation of FGFR1 promotes phosphor- ylation of ERK1/2, through which the expression of SOX2 was upregulated. These results demonstrated the existence of FGFR1-ERK1/2-SOX2 axis in FGFR1-amplified lung cancer cells.

 

Overexpression of SOX2 promotes proliferation, EMT, migration, and invasion, and attenuates oncogenic addiction to FGFR1 in FGFR1-amplified

To examine the influence of SOX2 on proliferation, EMT, migration and invasion, H1581, and DMS114 cells were transduced with SOX2 lentivirus (LV-SOX2) or negative control lentivirus (LV-NC). Cell proliferation was promoted by LV-SOX2 compared to LV-NC (Fig. 3a). Over- expression of SOX2 decreased the expression of E-cadherin and increased the expression of N-cadherin and Vimentin (Fig. 3b). Cell migration and invasion was promoted after transduction of LV-SOX2 (Fig. 3c and d). Both H1581 and DMS114  cells  were  sensitive  to  FGFR1  inhibitor

AZD4547, with IC50 less than 1 μM (Supplementary Figure S3a). In order to further explore whether overexpression of downstream SOX2 influences the sensitivity to FGFR1 inhibition or not, SOX2-overexpressing stable cells were treated with increasing concentrations of AZD4547. The IC50 of AZD4547 was markedly increased after transduction of LV-SOX2, compared to LV-NC (Supplementary Figure S3b). Collectively, these results confirmed that overexpression of SOX2 promotes cell proliferation, EMT, migration, and invasion, and attenuates oncogenic addiction to FGFR1 in FGFR1-amplified lung cancer cells.

 

SOX2 is required for FGFR1-induced proliferation, EMT, migration, and invasion in FGFR1-amplified lung cancer cells

To determine whether FGFR1-induced proliferation, EMT, migration, and invasion was mediated by SOX2, SOX2-silencing stable cells were generated by SOX2 shRNA lentivirus (LV-shSOX2). SOX2 was knocked down in H1581 and DMS114 cells by LV-shSOX2, compared to negative control shRNA lentivirus (LV-shRNA-NC; Fig. 4a). Treatment of FGF2 promoted cell proliferation in LV-shRNA-NC group, but not in LV-shSOX2 group (Fig. 4b). Stimulation of FGF2 increased phosphorylation of FGFR1 and ERK1/2, and induced transition of EMT markers in LV-shRNA-NC group. However, knockdown of SOX2 by LV-shSOX2 did not affect the phosphorylation status of FGFR1 and ERK1/2, but blocked the transition of EMT markers induced by FGF2 stimulation (Fig. 4c). Scratch healing, migration, and invasion induced by FGFR1 activation was not observed in LV-shSOX2 groups (Fig. 4d and e). Taken together, these results suggested that SOX2 is imperative in FGFR1-induced proliferation, EMT, migration, and invasion in FGFR1-amplified lung cancer cells or DMS114 (3 × 106 cells) in a total volume of 50 μL mixed with Matrigel (PBS:Matrigel = 4:1) were injected into the left lung of 4-week-old male BCLB/C nude mice. One week after injection, the mice were randomly grouped and treated with FGFR1 inhibitor AZD4547 (12.5 mg/kg/d) or vehicle for 2 weeks. a Comparison of orthotopic lung cancer models. Left and right lungs, and heart were shown. Apparent orthotopic left lung tumor was observed in vehicle group (arrowheads) but not in AZD4547 group. b Orthotopic left lung tumor volume was calculated by microCT scan and three-dimensional region of interest analysis. The left lung tumor decreased in AZD4547 group, compared with the vehicle group. c, d, f, h H1581 and DMS114 was transfected with LV-NC or LV-SOX2, and orthotopic lung cancer models were established by stable cell lines and treated with AZD4547 or vehicle as before. c Comparison of orthotopic lung cancer models. Orthotopic left lung tumors were labeled with arrowheads. d Com- parison of orthotopic left lung tumor volume. The left lung tumor volume increased in LV-SOX2 vehicle group, compared with LV-NC vehicle group, and was equivalent between LV-SOX2 vehicle group and LV-SOX2 AZD4547 group. e, f HE and immunofluorescence staining of orthotopic left lung tumors of (e) wild-type H1581 and DMS14 and (f) LV-NC or LV-SOX2 stable transfected H1581 and DMS114. SOX2 and E-cadherin were red. N-cadherin and Vimentin were green. g, h HE staining of all the right and left lungs of orthotopic mouse models were performed. Metastatic nodules in the contralateral lungs (right lungs) were found out and counted under microscope.

 

Inhibition of FGFR1 suppresses tumor growth, SOX2 expression, EMT, and metastasis in vivo using orthotopic and subcutaneous lung cancer xenograft models, but cannot suppress these processes in SOX2-overexpressing stable cell lines

Orthotopic and subcutaneous lung cancer mouse models were established with H1581 and DMS114 or SOX2- overexpressing stable cell lines, and were treated with FGFR1 inhibitor AZD4547 (12.5 mg/kg/d) or vehicle as control. The tumor volume of orthotopic left lung cancer lesion was calculated by microCT scan and three- dimensional region of interest analysis. The subcutaneous tumor was measured by caliper every 3 days and tumor volume was calculated and plotted.

In H1581 and DMS114, tumor growth was markedly suppressed both in orthotopic (Fig. 5a and b) and subcutaneous (Supplementary Figure S4a and b) lung cancer models, compared to the vehicle group. Hematoxylin and eosin (HE) and immunofluorescence staining of the orthotopic left lung cancer lesion and subcutaneous tumor nodule was conducted. In the AZD4547 group, expressions  of  SOX2  and  mesenchymal  markers N-cadherin and Vimentin were suppressed, while the epithelial marker E-cadherin was highly expressed, com- pared to the vehicle group (Fig. 5e and Supplementary Figure S4c). Importantly, HE staining of all the left and right lungs of orthotopic lung cancer models was per- formed, and metastatic nodules in the contralateral lungs (right lungs) were found out and counted under microscope. Inhibition of FGFR1 suppressed formation of metastatic nodules in the contralateral lungs significantly (Fig. 5g). Namely, inhibition of FGFR1 inhibited tumor growth, SOX2 expression, EMT, and metastasis of FGFR1- amplified lung cancer in vivo.

In SOX2-overexpressing stable cell lines, the tumor growth (Fig. 5c and d), change of EMT markers (Fig. 5f), and numbers of metastatic nodules in the contralateral lungs (Fig. 5h) were promoted compared to the LV-NC group. However, these processes were not sup- pressed by FGFR1 inhibitor in orthotopic lung cancer models.

Taken together, all these data demonstrated that inhibition of FGFR1 signaling suppresses tumor growth, SOX2 expression, EMT, and metastasis of FGFR1-amplified lung cancer in vivo, but cannot suppress these processes in SOX2-overexpressing stable cell lines.

 

Higher expressions of FGFR1 and SOX2 are positively correlated, and both are associated with shorter OS and PFS in NSCLC

To evaluate the prognostic value of FGFR1 and SOX2 in lung cancer patients, we first analyzed the copy number and expression features of FGFR1 and SOX2 in our LSQCC cohort [39] and the TCGA provisional LSQCC and LADC cohort [4, 5]. Remarkably, 10 patients in our LSQCC cohort and 48 patients in the TCGA provisional LSQCC cohort are co-amplified with FGFR1 and SOX2. As a result, co-amplification of SOX2 is common in FGFR1-amplified LSQCC patients, that is, 10/14 (71.4%) in our LSQCC cohort and 48/92 (52.2%) in TCGA provisional LSQCC cohort (Fig. 6a). Positive correlation between expression of FGFR1 and SOX2 was observed both in our LSQCC cohort and the TCGA provisional LSQCC and LADC cohort (Fig. 6b). Higher FGFR1, SOX2, or total mRNA level are all associated with shorter overall survival (OS) and progression-free survival (PFS) in NSCLC (Fig. 6c). The total mRNA level is sum of FGFR1 and SOX2 mRNA levels at a 1:1 ratio. Collectively, these results confirmed that higher expression of FGFR1 and SOX2 are positively correlated, and both are associated with shorter survival in lung cancer patients. And co- amplification of SOX2 is common in FGFR1-amplified LSQCC.

 

 

Discussion

The modest response rate of FGFR1 inhibition in clinical trials was unparalleled with the promising preclinical sen- sitivity [12–18]. These controversial results make the amplified FGFR1 as a difficult target. Currently, to improve targeted therapy efficacy in FGFR1-amplified lung cancer, researches are trying to explore the signaling paradigm and the interaction between FGFR1 and other co-existing genomic alterations [15, 40–42]. Here, for the first time, we reported that FGFR1 regulates expression of SOX2 through the MAPK pathway. The novel FGFR1-ERK1/2- SOX2 axis promotes cell proliferation, EMT, and metastasis of FGFR1-amplified lung cancer in vitro and in vivo.

More than 90% of deaths from solid tumors are mainly attributed to metastasis [43]. Tumor metastasis is a com- plicated serial processes of metastatic cascades, including acquiring of motility and invasion, neovascularization and intravasation, survival and anti-anoikis in blood circulation, extravasation, and colonization [22, 23, 43, 44]. EMT plays essential roles in cancer metastasis by promoting each core procedure of the metastatic cascades. However, the effect and mechanism of FGFR1 activation in the EMT and metastasis of lung cancer has not been well characterized. Hu has reported that brachyury, an EMT transcription fac- tor, is activated by the FGFR1/MAPK pathway in lung cancer [45]. Maehara observed that FGFR1 promotes stem cell phenotype of esophageal cancer with transitions of EMT markers [46]. Here, we further explored and con- firmed that FGFR1 promotes EMT and metastasis by upregulation of SOX2 in vitro and in vivo. The study and observation of EMT and metastasis in vivo has always been difficult, for the apparent shortcomings of usual animal models. The prevalence of metastasis is low in sub- cutaneous tumor xenograft model. The tail vein injection model cannot mimic the whole natural process of metastasis because tumor cells are injected into blood circulation directly by artificial extravasation. In this study, we designed and modified the orthotopic lung cancer model to observe lung cancer in situ with microCT scan, and to simulate the natural process of lung cancer metastasis [47, 48]. Metastatic lung cancer nodules developed in the con- tralateral lungs (right lungs) of the orthotopic mouse model, which is a common metastasis pattern for lung cancer patients. The metastasis ability was accessed and compared by the number of metastatic nodules in the contralateral lungs. And this metastasis course was suppressed by FGFR1 inhibition in vivo. Therefore, these data demon- strated the effect and mechanism of FGFR1-induced EMT and metastasis, and suggested use of FGFR1 inhibitor in anti-EMT and anti-metastasis therapy.

Previous study showed that SOX2 is upregulated by EGFR activation and downstream AKT/mTOR and JAK/STAT3 pathway [34]. As a stem cell marker, SOX2 was upregulated after FGFR1 activation in our prior study [31]. However, the mechanism by which FGFR1 regulates SOX2 and the clinical implications has not been characterized. Activation of FGFR1 by different FGF ligands results in phosphorylation of FGFR substrate 2 and further phos- phorylates a number of downstream signaling cascades. MAPK signaling is the most common downstream of FGFR1, which mediates plenty of biological functions in embryo development and disease [25, 49]. Inhibition of FGFR1 causes downregulation of phosphorylated ERK in a dose-dependent manner [12, 50, 51]. Here, stimulation of FGF2 induced phosphorylation of FGFR1 and downstream ERK1/2 and expression of SOX2. In order to testify whe- ther the expression of SOX2 was upregulated by pERK1/2, MEK/ERK inhibitor AZD6244 was added to block the MAPK pathway [37, 52]. As a result, stimulation of FGF2 still induced augmention of pFGFR1, but not pERK1/ 2 or SOX2. Furthermore, we designed a rescue experiment to confirm the FGFR1-ERK1/2-SOX2 axis. Transfection of variant ERK2_R67S plasmid or lentivirus generated sus- tained autophosphorylation of ERK2 in a MEK1/2-inde- pendent manner and promoted expression of SOX2, which was not inhibited by FGFR1 inhibitor AZD4547 or MEK/ ERK AZD6244 in vitro and in vivo. Hence, these data indicated a new FGFR1-ERK1/2-SOX2 axis, adding the signaling paradigm of the FGFR1 pathway.

SOX2 is an essential transcription factor of stemness. By cooperating with different cofactors, SOX2 is able to bind with promoters of versatile genes and modulate their tran- scription [33]. Thus, SOX2 is found to participate in lots of biological processes, including maintenance of embryo and adult stem cells, differentiation of bronchi and lungs, tumorigenesis, cell survival, and so on [53]. Impressive studies have demonstrated that SOX2 is a lineage differ- entiation and survival gene of squamous cell carcinoma of the lung, esophagus, and skin [53–56]. Recent researches indicated that SOX2 also plays essential roles in EMT of NSCLC [35, 36], cancer stem cell phenotype of LSQCC [57],  and  chemoresistance  of  SCLC  [58].  In  fact, amplification and overexpression of SOX2 is a common genomic alteration of LSQCC and SCLC [4, 59], and is another molecular target of interest for lung cancer. Here, we demonstrate that SOX2 is necessary and indispensable for FGFR1-induced EMT. The EMT-promoting effect of SOX2 revealing in this study is in concordance with a recent impressive research that SOX2 promotes cell quies- cence, immune escape, and latent metastasis of lung and breast cancers [60]. Taken together, these studies and our present research highlight that SOX2 is a core operator of stemness, EMT, and metastasis in lung cancer. Researches, aiming to demonstrate the mechanisms of these versatile function of SOX2, will have great significance for clinical treatment of lung cancer. Most recently, the first DNA- binding inhibitor of SOX2, PIP-S2, was synthesized and reported [61]. The anti-EMT and anti-metastasis effect of PIP-S2 in cancer is eagerly anticipated.

Recent studies about FGFR1 signaling in lung cancer are focusing on the mechanisms of primary or secondary resistance to FGFR1 kinase inhibitor [42, 62, 63], and the reasons why FGFR1 kinase inhibitors have an overall response rate of merely ~10% [16–18]. Predictive bio- markers for efficacy of FGFR1-targeted therapy are eagerly looking for [64, 65]. Here we demonstrate that SOX2 is upregulated by FGFR1 activation and mediates cell pro- liferation, EMT, migration, and invasion. Importantly, SOX2 amplification is also common in LSQCC patients, and the prevalence of SOX2 co-amplification in FGFR1- amplified LSQCC is high, that is, 52.2–71.4%. And the copy number gains of FGFR1 and SOX2 were reported strongly correlated in NSCLC patients [66]. Both H1581 and DMS114 cell lines are sensitive to FGFR1 inhibitors [15, 42, 51]. However, here we show that the IC50 of FGFR1 inhibitor were markedly increased in SOX2- overexpressing stable cell lines of H1581 and DMS114. And inhibition of FGFR1 cannot suppress the improvement of tumor growth, EMT, and metastasis caused by SOX2- overexpressing in vivo. Thus, amplification and over- expression of downstream SOX2 might cause resistance to FGFR1 inhibitor. This pattern of target therapy resistance, caused by amplification and overexpression of alternative signaling molecules, has been investigated recently. For example, amplification of MET caused resistance to EGFR inhibition [30], and amplification of MET or NRAS caused resistance to FGFR inhibition [42, 63]. Collectively, our findings demonstrated that amplification and overexpression of SOX2 might be a predictive biomarker of resistance against FGFR1 inhibition.

In summary, our study revealed a novel pathway that activation of FGFR1 upregulates expression of SOX2 by downstream phosphorylated ERK1/2. The newly defined FGFR1-ERK1/2-SOX2 axis promotes cell proliferation, EMT, and metastasis in FGFR1-amplified lung cancer in vitro and in vivo. Higher expression of FGFR1 and SOX2 are positively correlated, and both are associated with poor prognosis in lung cancer. These findings impli- cated that co-amplification and overexpression of down- stream SOX2 might contribute to resistance against FGFR1- targeted therapy, especially in LSQCC, where co- amplification and overexpression of FGFR1 and SOX2 is common.

 

 

Materials and methods

Cell culture and reagents

H1581 and DMS114 were purchased from ATCC and cultured using RPMI 1640 (HyClone) with 10% fetal bovine serum (Gibco), 100 U/mL penicillin, and 100 μg/mL streptomycin (Gibco). FGF2 (PEPROTECH, 100-18B) and 0.2% heparin sodium salt in PBS (STEMCELL, 07980) were used to activate FGFR1. FGFR1 inhibitor AZD4547 was kindly offered by AstraZeneca Pharmaceutical Com- pany [50]. MEK/ERK inhibitor AZD6244 was bought from Selleck Chemicals [37]. They were dissolved in DMSO, aliquoted, and stored as 10 mM stocks at −20 °C for in vitro studies. And for in vivo studies, AZD4547 was diluted in a 1% (v/v) solution of Tween-80 in deionized water.

 

RNA interference and plasmids transfection

Specific short interfering RNAs targeting FGFR1 (siFGFR1- 1/2) and negative control scrambled siRNAs (siRNA-NC) were purchased from HanBio (Shanghai, China). Sequences of siRNAs are as follows: siFGFR1-1, sense 5′-CGGUCAUCGUCUACAAGAUdTdT-3′, antisense 5′-AUCUUGUAGACGAUGACCGdTdT-3′; siFGFR1-2, sense 5′-GAUGGUCCCUUGUAUGUCAdTdT-3′, antisense 5′-UGAC AUACAAGGGACCAUCdTdT-3′; siRNA-NC, sense 5′-UUCUCCGAACGUGUCACGUdTdT-3′, antisense 5′-AC GUGACACGUUCGGAGAAdTdT-3′. Wild-type ERK2 and autophosphorylation variant ERK2 p.R67S (ERK2_R67S) [38] plasmids were constructed by GENE- CHEM (Shanghai, China). siRNAs and plasmids were transfected into H1581 and DMS114 cells by Lipofectamine 3000 (Invitrogen) with or without P3000 (Invitrogen) fol- lowing the manufacturers’ instructions.

 

Lentivirus transduction and generation of stable cell lines

Human SOX2 lentivirus (LV-SOX2), negative control (LV- NC), shRNA lentivirus targeting SOX2 (LV-shSOX2-1/2), and scrambled shRNA lentivirus (LV-shRNA-NC) were purchased from HanBio. ERK2 lentivirus (LV-ERK2), ERK2_R67S lentivirus (LV- ERK2_R67S), and negative control (LV-NC) were constructed by GENECHEM (Shanghai, China). All the lentiviruses were labeled with enhanced green fluorescent protein. The target sequences of SOX2 were 5′-CGCTCATGAAGAAGGATAAGT-3′ for shSOX2-1 and 5′-TGGACAGTTACGCGCACATGA-3′ for shSOX2-2, 5′-TTCTCCGAACGTGTCACGTAA-3′ for shRNA-NC, respectively. Lentivirus transductions and stable cell line selection were performed as previously described [31].

 

Western blot, RNA extraction, cDNA synthesis, and qPCR

Cells were lysed for protein or RNA extraction, and sub- jected to western blot or cDNA synthesis and qPCR as previously described by us [31]. Antibodies for western blot were as follows: mouse anti-β-Tubulin (Abmart, M20005), rabbit anti-FGFR1 (CST, 9740), rabbit anti-pFGFR1 (CST, 3471), rabbit anti-ERK1/2 (CST, 4695), rabbit anti pERK1/2 (CST, 4370), rabbit anti-SOX2 (Abcam, ab59776), rabbit anti-E-cadherin (CST, 3195), rabbit anti-N-cadherin (CST, 13116), rabbit anti-Vimentin (CST, 5741), goat anti-mouse IgG-HRP (Santa Cruz, sc-2005), and goat anti-rabbit IgG-HRP (CST, 7074). Primers for qPCR were as follows: FGFR1, forward 5′-TAATGGACTCTGTGGTGCCCTC-3′, reverse 5′-ATGTGTGGTTGATGCTGCCG-3′; E-cadherin, forward 5′-CACCTGGAGAGAGGCCATGT-3′, reverse 5′-TGGGAAACATGAGCAGCTCT-3′; N-cadherin, for- ward 5′-ATGTGCCGGATAGCGGGAGC-3′, reverse 5′-TACACCGTGCCGTCCTCGTC-3′; Vimentin, forward 5′-CTTGAACGGAAAGTGGAATCCT-3′, reverse 5′-GTC AGGCTTGGAAACGTCC-3′; SOX2, forward 5′-GTATCA GGAGTTGTCAAGGC-3′, reverse 5′-AGTCCTAGTCTT AAAGAGG-3′.

 

Immunofluorescence microscopy

Cells were passed on coverslips in a 24-well plate and treated with corresponding reagents for indicated durations. Then, immunofluorescence staining was conducted as pre- viously described [31]. The primary antibody is rabbit anti- SOX2 (Abcam, ab59776). The secondary antibody is don- key anti-rabbit IgG conjugated with Alexa Fluor 594 (ThermoFisher, A-21207). Stained cells were observed and imaged under an immunofluorescence microscope (Leica DFC420C).

 

Scratch assay

Confluent monolayer cells in six-well plate were scratched and cultured with RPMI 1640 medium containing 1% FBS and different reagents. Photomicrographs were taken at 0 and 24 h after scratching using an Olympus CKX41 microscope. Scratch healing ratio was calculated as follow: (width of 0 h−width of 24 h)/width of 0 h, representing the healing degree of scratches.

 

Transwell migration and invasion assays

In all, 5 × 104 cells with 1% FBS medium were seeded into an 8-μm pore membrane or Matrigel-coated (CORNING, 356231) membrane Transwell chamber (CORNING, 3422) placed in a 24-well plate. After cell attachment, 10% FBS medium with FGF2 (20 ng/mL) and heparin (10 μg/mL), or AZD4547 (1 μM) were added to the lower chamber of the 24-well plate. After 36 h, migrated or invaded cells were stained as described before [35]. The stained cells were photographed and three microscopic fields were counted.

 

CCK8 assay and IC50 calculation

In all, 2.5 × 103 cells with 100 μL medium were seeded into each well (N = 5) of 96-well plate, and changed to medium with different concentrations of FGFR1 inhibitor AZD4547 on the second day. After 72 h, CCK8 assay was performed as described before [31]. The IC50 was calculated by Prism.

 

In vivo subcutaneous lung cancer model and tissue staining

Cell suspension of H1581 (1 × 106 cells) or DMS114 (2 × 106 cells) in a volume of 50 μL were injected sub- cutaneously into the right flanks of BCLB/C nude mice.

Subcutaneous tumor volumes were measured by caliper every 3 days, and calculated as 0.5 × length × width2. After the tumor reached 0.2 cm3, the mice were randomly grouped (N = 5, per group) and treated with AZD4547 (12.5 mg/kg/d) or equal volume of vehicle once daily by oral gavage for 3 weeks. After 3 weeks of treatment, tumors were collected and photographed, and then fixed in 4% PFA and embedded in paraffin. Sections were deparaffinized, rehydrated, and subjected to HE staining. For immuno- fluorescent stains, sections were subjected to citrate antigen retrieval solution (Beyotime, P0081) at 95 °C for 25 min, and then blocked for 1 h using 10% bovine serum albumin and incubated with primary antibodies against SOX2 (Abcam, ab59776), E-cadherin (CST, 3195), N-cadherin (CST, 13116), and Vimentin (CST, 5741) for overnight. After washing steps, sections were incubated with donkey anti-rabbit IgG conjugated with Alexa Fluor 594 (Ther- moFisher, A-21207) or goat anti-rabbit IgG conjugated with Alexa Fluor 488 (ThermoFisher, A-11034). Finally, sec- tions were nucleus-stained with DAPI and mounted with anti-fade mounting medium. Sections were observed and imaged with a laser scanning confocal microscopy (Leica TCS SP8).

 

In vivo orthotopic lung cancer model and microCT scan

Orthotopic lung cancer model was modified from our and Peng′s previous studies [47, 48]. In brief, 4-week-old male BCLB/C nude mice were anesthetized by 3% tri- bromoethanol in PBS, 100 μL per 15 g weight, and intra- peritoneal injection. A 5 mm incision was sheared on the dorsal side over left lung, ~0.5 cm below the scapula. Subcutaneous tissue and muscles are separated to visualize lung movement. Cell suspension of H1581 (2 × 106 cells) or DMS114 (3 × 106 cells) in a total volume of 50 μL (PBS: Matrigel = 4:1) were injected directly into the left lung with insulin injection syringes (29 G*12.7 mm, BD, 328421). One week after injection, the mice were randomly grouped (N = 5, per group) and treated with AZD4547 (12.5 mg/kg/ d) or equal volume of vehicle once daily by oral gavage for 2 weeks. Then, the mice were anesthetized and subjected to microCT analysis. MicroCT scan was performed on an Inveon MM Platform (Siemens, USA). The tumor volume in left lung was accessed by three-dimensional region of interest analysis. Then, all the right and left lungs were excised and photographed, followed by HE staining. Finally, numbers of metastatic tumor nodules in the con- tralateral (right) lungs were counted.

All animal experiments were carried out in full accor- dance with the protocols approved by Institutional Ethics Committee of Shanghai Jiao Tong University.

 

Analysis of public data sets from TCGA and Kaplan–Meier Plotter

Relative copy number and mRNA levels of FGFR1 and SOX2 of TCGA provisional LSQCC and LADC cohorts were downloaded from cBioPortal (http://www.cbioportal. org/index.do) [67]. Linear regression and Spearman corre- lations between mRNA levels of FGFR1 and SOX2 were conducted.

Prognostic values of AZD6244 or SOX2 mRNA levels were analyzed by Kaplan–Meier survival curves of NSCLC patients, using Kaplan–Meier Plotter (www.kmplot.com/a nalysis) [68]. Log-rank test was used for statistical analysis.

 

Statistical analysis

All statistical analyses were performed using Prism (version 7.0a, GraphPad Soft Inc.). Data were expressed as mean ± SD, and the paired or unpaired Student’s t-test was chosen to analyze the statistical significance between two groups. P < 0.05 was considered statistically significant.