The Emerging Role of SOX2 in Cell Proliferation and Survival and Its Crosstalk with Oncogenic Signaling in Lung Cancer

Authors


  • Author contributions: Y.-T.C.: conception and design, collection and assembly of data, data analysis and interpretation, and manuscript writing; C.-C.L.: collection and assembly of data; S.-H.H. and S.-E.L.: provision of study patients and clinical data analysis; S.-C.L., C.-Hu.C., C.-Hs.C., and Y.-R.K.: collection of data; Y.-H.W.: clinical data analysis; C.-T.C. and Y.-H.W.: oxygen consumption analysis; C.-W.W.: integration of research resources, including direction, financial support, manuscript writing, and final approval of manuscript.

Correspondence: Cheng-Wen Wu, M.D., Ph.D., National Yang-Ming University, No. 155, Sec. 2, Linong Street, Taipei 112, Taiwan. Telephone: 886-2-2826-7919; Fax: 886-2-2823-6518; e-mail: cwwu@ym.edu.tw

Abstract

Tumor cells have long been observed to share several biological characteristics with normal stem/progenitor cells; however, the oncogenic mechanisms underlying the lung stem/progenitor cell signaling remain elusive. Here, we report that SOX2, a self-renewal factor in lung stem/progenitor cells, is highly expressed in a subclass of lung cancer cells, the proliferation, survival, and chemoresistance of which are dependent on SOX2 signaling. Overexpression of SOX2 promotes oncogenic phenotypes in lung cancer cells; knockdown of SOX2 attenuated cell proliferation. We observed that SOX2 increased the expression of epidermal growth factor receptor (EGFR), and EGFR activation further upregulated SOX2 levels, forming a positive feedback loop. SOX2 expression promoted chemoresistance, and silencing of SOX2 perturbed mitochondrial function, causing marked apoptosis and autophagy. SOX2 induced BCL2L1, the ectopic expression of which rescued the effects of SOX2 silencing on apoptosis, autophagy, and mitochondrial function. SOX2 promoted tumor formation, along with increased cell proliferation in a xenograft mouse model. SOX2 expression is associated with poor prognosis in lung cancer patients; moreover, SOX2, EGFR, and BCL2L1 expression levels were significantly correlated in lung tumors. Our findings support the emerging role of SOX2 in cell proliferation and survival by eliciting oncogenic EGFR and BCL2L1 signaling with potential applications as a prognosis marker and a therapeutic target in lung cancer. Stem Cells 2013;31:2607–2619

Introduction

The transcription factor SOX2 belongs to the SOX (Sry-related HMG Box) family of proteins and is essential for maintenance of self-renewal and pluripotency in embryonic stem cells (ESCs) as well as adult tissue progenitor cells [1, 2]. Proliferation of cells lining the respiratory tract is initiated by SOX2 signaling in lung stem/progenitor cells; SOX2-expressing lung cells are critical for tissue homeostasis and cell proliferation during injury repair [3, 4]. In addition, SOX2 expression has been detected in human breast and lung cancer [5, 6], and genomic amplification of SOX2 has been observed in human lung squamous cell carcinomas [7]. SOX2 overexpression in the lung epithelium of adult mice causes lung tumor formation [8], suggesting that SOX2 may play an oncogenic role in lung tumorigenesis.

During the repair of bronchiolar injury, expression of epidermal growth factor receptor (EGFR) is induced in proliferating cells of the injury target zone, indicating a critical requirement for EGFR signaling in stimulating cell proliferation during bronchiolar injury repair [9]. Aberrant expression of EGFR has been detected in more than 50% of non-small-cell lung cancers (NSCLC) and has been suggested to cause abnormal cell proliferation and tumorigenesis [10]. Transforming growth factor-α (TGF-α), a cognate ligand of EGFR, is frequently overexpressed in lung cancer, causing receptor hyperactivity in autocrine feedback loops [11].

BCL2L1, also known as BCL-XL, is a member of the BCL2 family, which is highly expressed in NSCLC and is associated with cancer progression [12]. BCL2L1 is located in the outer mitochondrial membrane and regulates mitochondrial membrane channel opening, functioning as a prosurvival regulator to protect cells from apoptosis [13, 14]. BCL2L1 also inhibits autophagy, a global catabolic process involving the recycling of intracellular components, via an inhibitory interaction with Beclin-1 [15]. Decreased BCL2L1 expression sensitizes chemoresistant lung cancer cells to chemotherapeutics [16].

Although it has been proposed that stem cell signaling is associated with tumor malignancy [17], the oncogenic mechanism of stem cell signaling needs to be further clarified. In this study, we found that SOX2 promotes malignancy by enhancing EGFR-mediated proliferation and BCL2L1-induced survival signaling, linking the SOX2 stem/progenitor cell pathway to the oncogenic phenotypes observed in lung cancer cells.

Materials andMethods

Reagents

Recombinant human TGF-α was purchased from PeproTech EC (London, United Kingdom). Doxycycline (doxycycline hyclate), cisplatin (cis-Diammineplatinum(II) dichloride), paclitaxel, oligomycin, carbonyl cyanide m-chloro phenyl hydrazone (CCCP), and antimycin-A were ordered from Sigma-Aldrich (St. Louis, MO, http://www.sigmaaldrich.com).

Cell Culture

CL1-0 and PL1 cells were established in our laboratory [18]. A549, H2170, H358, and BEAS-2B cells were obtained from the American Type Culture Collection. HCC827 cells were kindly provided by Dr. Jeff Wang of the Development Center for Biotechnology, Taiwan. NL-20 and H1975 cells were obtained from Dr. Wayne Chang of the National Health Research Institutes, Taiwan. All lung cancer cell lines were grown in RPMI-1640 medium with 10% fetal bovine serum (FBS). BEAS-2B and NL-20 cells were maintained in Ham's F-12 medium with 5 μg/ml insulin, 10 ng/ml epidermal growth factor, 1 μg/ml transferrin, 500 ng/ml hydrocortisone, and 4% FBS.

Plasmid Construction

Detailed procedures for construction of SOX2-, MYC-, or EGFR-expressing lentiviral vectors, SOX2 or EGFR promoter reporters, and shRNA lentiviral vectors are described in Supporting Information.

Quantitative RT-PCR

Quantitative RT-PCR (Q-PCR) was performed as described [19]. Relative gene expression was analyzed using the math formula method using the LightCycler 480 Real-Time PCR System (Roche Applied Science) combined with the Universal Probe Library (Roche Applied Science, Indianapolis, IN, https://www.roche-applied-science.com). 18S rRNA was used as a reference transcript. Primer sequences designed to detect specific genes are listed in Supporting Information Table S2.

Clonogenic and WST-1 Cell Viability Assays

A clonogenic assay was performed using 500 cells per well in a six-well plate. Colonies were fixed and stained with 3.7% formaldehyde/80% methanol/0.25% crystal violet (Fisher Scientific, Waltham, MA, http://www.fishersci.com) and counted 2 weeks after plating. Infected cells (3 × 103 cells per well) were seeded on 96-well plates. After 24 hours, cells were cultivated with different concentrations of cisplatin or paclitaxel for 48 hours. Cell viability was measured by WST-1 reagent (Roche Applied Science, Indianapolis, IN, https://www.roche-applied-science.com). The absorbance at 450 nm was measured by a microplate reader, Model 680 (Bio-Rad).

Sphere Assay

The sphere assay for in vitro analysis and passaging was performed as in Xin et al. [20], with modifications. In brief, 1,000 CL1-0 cells were resuspended in 100 μl of a 50:50 mixture of Matrigel (BD Biosciences, San Jose, CA, http://www.bdbiosciences.com) and F-12, 0.5% FBS medium and plated around the rim of a well in a 12-well tissue culture plate. Matrigel mix was allowed to solidify at 37°C, and 800 μl of F-12 serum-free medium supplemented with EGF and insulin was then added. The medium was changed every 3 days. To recover the spheres, cells in each well were treated with 1 mg/ml dispase solution (Gibco-BRL, Grand Island, NY, http://www.invitrogen.com). Spheres were digested with trypsin in 0.05% EDTA, triturated through a 27-gauge needle, and filtered through a 40-μm filter.

Oxygen Consumption Assay

The mitochondrial oxygen consumption rate (OCR, pmol/minute) was analyzed using an XF24 extracellular flux analyzer (Bucher Biotec AG, Basel, Switzerland, http://www.bucher.ch/) as described [21]. Oligomycin-A (10 μM), CCCP (10 μM), and antimycin-A (5 μM) were added to cells sequentially to inhibit ATP synthase, uncouple the proton gradient, and inhibit the electron transport chain, respectively. The reserve capacity is calculated by dividing the CCCP-induced maximal OCR by the pre-oligomycin rate [22].

Chromatin Immunoprecipitation

The chromatin immunoprecipitation (ChIP) assay was performed according to the protocol for the LowCell# ChIP kit (Diagenode, Liège, Belgium, http://www.diagenode.com). Detailed procedures for ChIP are described in Supporting Information. Supporting Information Table SI lists primer sequences designed to detect specific promoters.

Xenograft Tumorigenicity Assay

All animal experiments were performed in accordance with the animal guidelines of the Academia Sinica Institutional Animal Care and approved by Animal Care Committee. Detailed procedures for the xenograft tumorigenicity assay are described in Supporting Information.

Patients

Most of the specimens used in immunohistochemistry were obtained from biopsy at Taipei Medical University Hospital. The specimens subjected to Q-PCR were collected from surgery. Approval for this study was granted by the Institutional Review Board protocol number CRC-04–11-05.

Immunohistochemistry

Detailed procedures for immunohistochemistry are described in Supporting Information. The expression levels of SOX2 were examined in lung cancer obtained from patients. The primary SOX2 antibody used in the study was mouse anti-SOX2 (MAB4343, Millipore, Billerica, MA, http://www.millipore.com) clone 6F1.2 with a 1:100 dilution. The positivity of the SOX2 expression was recorded with Olympus photo system (Olympus DP20 microscope camera Tokyo, Japan, http://www.olympus.co.jp/) and scored by a pathologist. The staining intensity was classified by nuclear stain: SOX2-negative expression included no stain in nuclei of tumor cells, and SOX2-postive expression included focal faint nuclear stain, diffuse to moderate nuclear stain, and diffuse strong nuclear stain.

Statistical Analysis

Differences in the immunohistochemical staining of human lung cancer specimens were assessed with the Fisher's exact test. The associations among the levels of various markers were calculated using Spearman's rho correlation coefficient. Data were analyzed using either Student's t test or Mann-Whitney U analysis. p values <.05 were considered to be statistically significant. All statistical analyses were performed using SPSS software version 12.0 (SPSS Inc., Chicago, IL).

Results

SOX2 Signaling Regulates Oncogenic Phenotypes in Lung Cancer Cells

Self-renewal of lung stem/progenitor cells is tightly regulated by SOX2 [3, 4]; thus, to investigate the possible involvement of SOX2-mediated stem/progenitor cell signaling in the development of oncogenesis, we investigated SOX2 mRNA expression in lung bronchial epithelial cells and NSCLC cell lines. Using Q-PCR, we observed that SOX2 was expressed in lung cancer cells such as CL1-0 (adenocarcinoma), H2170 (squamous), H358 (adenocarcinoma), and A549 (adenocarcinoma) but not in noncancerous lung bronchial epithelial cells such as BEAS-2B and NL-20 (Fig. 1A). Immunoblotting analysis showed that SOX2 was highly expressed in CL1-0 cells and moderately expressed in A549 cells but not in NL-20 (Fig. 1A). To monitor the status of the SOX2 signaling in lung cancer cells, we used the EOS-S(4+) reporter, which contains four copies of SOX2's regulatory region 2 (SRR2) and has been used previously to monitor the status of the endogenous SOX2 signaling in induced pluripotent stem cells [23]. We observed that the EOS-S(4+) reporter was highly activated in CL1-0, mildly in A549, but not in NL-20 cells (Fig. 1B), indicating that SOX2 signaling is active in some of lung cancer cells. We investigated the role of SOX2 signaling in lung cancer cells. WST-1 analysis showed that ectopic SOX2 expression enhanced A549 cell proliferation (Fig. 1C). Clonogenic analysis demonstrated that SOX2 expression promoted cell growth (Fig. 1D, left); moreover, soft-agar analysis indicated that SOX2 increased the anchorage-independent growth of A549 cells (Fig. 1D, right). We further examined the function of SOX2 in sphere formation and self-renewal of lung cancer cells using EOS-S(4+) to monitor SOX2 signaling. While SOX2 expression was enhanced during sphere formation (Supporting Information Fig. S1), shRNA-mediated silencing of SOX2 attenuated sphere formation in CL1-0 cells and also blocked the self-renewal of CL1-0 cells in subsequent generations (Fig. 1E). These results supported the notion that SOX2 signaling promotes oncogenic phenotypes in lung cancer cells.

Figure 1.

SOX2 signaling in lung cancer cells. (A): SOX2 expression in lung cancer cell lines. Quantitative RT-PCR analysis of the mRNA expression of SOX2 in immortalized bronchial epithelial (BEAS-2B and NL-20) and non-small-cell lung cancer cell lines (left). Immunoblotting of SOX2 with β-actin as control in NL-20, CL1-0, and A549 cells (right). (B): Analysis of SOX2 signaling in lung cancer cells. NL-20, CL1-0, and A549 cells were infected with the lentiviral reporter EOS-S(4+) or EOS-S(0), in which four copies of SRR2 were removed from EOS-S(4+). Green fluorescent protein expression driven by EOS-S(4+) or EOS-S(0) was detected by fluorescence microscopy in infected cells (top). Cell morphology was monitored under a phase-contrast microscope (bottom). Scale bars = 10 μm. (C): WST-1 cell proliferation analysis of A549 cells infected with a lentiviral vector expressing SOX2 or an empty control vector. Results are representative of three independent experiments and are expressed as the means ± SD. *, p < .05. (D): Clonogenic assays in A549 cells infected with a lentiviral vector expressing SOX2 or an empty control vector (top left) and stained with crystal violet (bottom left). Results are representative of three independent experiments. The relative colony number of Ctrl cells was set to 100%, and the results are expressed as the means ± SD. *, p < .05. Soft agar colony formation assay in A549 cells infected with a lentiviral vector expressing SOX2 or an empty control vector (top right) and stained with crystal violet (bottom right). The relative colony number of Ctrl cells was set to 100%, and the results are expressed as the means ± SD. *, p < .05. (E): SOX2-mediated self-renewal in lung cancer cells. CL1-0 cells were infected with scrambled control (Sc) or SOX2 shRNA (shSOX2 and shSOX2-2) lentiviral vectors, followed by a sphere formation assay. shSOX2 and shSOX2-2 target different regions of the SOX2 mRNA. Infected CL1-0 spheres were dissociated, and equal numbers of cells were passaged for three generations. Sphere counts were normalized to the result for the first-generation scrambled shRNA spheres (left). Representative sphere images from the first generation are shown to the right. Fluorescence images (upper right) show EOS-S(4+) activity, and phase-contrast pictures (lower right) demonstrate the sphere phenotype. Scale bars = 50 μm. Results are representative of three independent experiments and are expressed as the means ± SD. ***, p < .001.

SOX2 Induces EGFR Expression, Forming a Positive Feedback Loop

Clonogenic assays showed that knockdown of SOX2 inhibits cell growth in SOX2-expressing lung cancer cells, but not in SOX2-negative bronchial epithelial cells (Supporting Information Fig. S2). To investigate the effect of SOX2 on cell proliferation in lung cancer cells, SOX2 was knocked down in lung cancer cells using shRNA. BrdU incorporation analysis indicated that SOX2 silencing attenuated DNA synthesis in A549 cells (Supporting Information Fig. S3). Because both SOX2 and EGFR induce proliferation in lung epithelial cells and contribute to oncogenesis, we further investigated whether SOX2 signaling regulates EGFR expression. Knockdown of EGFR as well as SOX2 decreased cell growth in A549 cells (Fig. 2A). Q-PCR and immunoblotting demonstrated that SOX2 silencing attenuated EGFR expression in CL1-0 cells (Fig. 2B). Moreover, ectopic SOX2 expression enhanced EGFR levels in A549 cells (Fig. 2C, left). Analysis of the EGFR promoter sequence via Transcription Element Search Software (TESS) identified a possible SOX2 binding site in the EGFR promoter region 389–383 bp upstream of the transcriptional start site. To study the effect of SOX2 on EGFR transcription, the 1.8-kb EGFR promoter, which includes the putative SOX2 binding site, was cloned into a luciferase reporter and subjected to a reporter assay. The results indicate that SOX2 activated EGFR transcription (Fig. 2C, right). To examine whether SOX2 binds to the EGFR promoter, we performed ChIP of the EGFR promoter with antibodies against SOX2. The result indicates that SOX2 bound to the EGFR promoter and that SOX2 silencing attenuated SOX2 binding to the promoter (Fig. 2D). Thus, in lung cancer cells, SOX2 appears to activate EGFR transcription by binding to its promoter.

Figure 2.

SOX2 promotes EGFR expression in a positive-feedback manner. (A): Clonogenic assay of A549 infected with scrambled control (Sc), shEGFR, or shSOX2 lentiviral vectors (right) and stained with crystal violet (left). shSOX2 and shSOX2-2 target different regions of the SOX2 mRNA. (B): Decrease in EGFR expression elicited by SOX2 silencing. Quantitative RT-PCR (Q-PCR) of EGFR expression (left) and immunoblotting for EGFR and SOX2 expression (right) in CL1-0 cells infected with scrambled control (Sc) or shSOX2 lentiviral vectors. (C): Increased EGFR expression elicited by SOX2 overexpression. Immunoblotting of EGFR and SOX2 expression in A549 cells infected with a lentiviral vector encoding SOX2 or an empty control vector (left). Luciferase reporter assay in A549 cells transfected with pGL3-EGFR promoter reporter together with pcDNA3-SOX2 (SOX2) or control pcDNA3 (right). Results are expressed as the means ± SD. ***, p < .001. (D): Chromatin immunoprecipitation (ChIP) analysis of SOX2 binding to the EGFR promoter. Binding of SOX2 to the EGFR promoter was assessed by ChIP in A549 cells infected with scrambled control (Sc) or shSOX2 lentiviral vectors. Results are expressed as the means ± SD. ***, p < .001. (E): SOX2 regulation by EGFR signaling. Q-PCR for SOX2 expression (left) and immunoblotting for EGFR and SOX2 expression (right) in A549 cells infected with a lentiviral vector encoding EGFR or an empty control vector. (F): Effect of MYC on SOX2 expression. Immunoblotting of MYC and SOX2 expression in A549 cells infected with scrambled control or shMYC lentiviral vectors (upper left). Immunoblotting of MYC and SOX2 expression in A549 cells infected with a lentiviral vector encoding MYC or an empty control vector (lower left). Luciferase reporter assay in A549 cells transfected with pGL3-SOX2 promoter reporter together with cDNA encoding MYC (MYC) or control nonspecific cDNA (right). Results are expressed as the means ± SD. *, p < .05. (G): SOX2 regulation by TGF-α/EGFR signaling. Serum-starved A549 cells infected with scrambled control (Sc) or shMYC lentiviral vectors were treated with or without TGF-α (50 ng/ml) for 3 hours. Whole-cell lysates were collected and subjected to immunoblotting with antibodies against MYC and SOX2. (H): MYC binding to SOX2 promoter upon EGFR stimulation. MYC binding to the SOX2 promoter was determined by ChIP in A549 cells infected with scrambled control, shEGFR, or shMYC lentiviral vectors. Results are expressed as the means ± SD. **, p < .01; ***, p < .001. Abbreviation: EGFR, epidermal growth factor receptor.

Using Q-PCR and immunoblotting, we found that EGFR silencing attenuated SOX2 expression (Fig. 2E). TESS analysis of the SOX2 promoter sequence identified potential MYC binding sites at the promoter region 691–686 and 680–674 bp upstream of the transcriptional start site. Knockdown of MYC inhibited SOX2 expression (Fig. 2F, upper left). Ectopic MYC expression increased the levels of SOX2 in A549 cells (Fig. 2F, lower left). To study the effect of MYC on SOX2 transcription, the 1.6-kb SOX2 promoter was cloned into a luciferase reporter and subjected to a reporter assay. The results indicate that ectopic MYC expression activated SOX2 transcription (Fig. 2F, right). Immunoblotting showed that TGF-α, a cognate EGFR ligand, caused an increase in the expression of MYC as well as SOX2; furthermore, MYC silencing attenuated TGF-α mediated upregulation of SOX2 (Fig. 2G), demonstrating that TGF-α/EGFR signaling induces SOX2 expression through a MYC dependent pathway. To examine whether EGFR induces MYC binding to the SOX2 promoter, we performed ChIP of the SOX2 promoter with antibodies against MYC. The data indicated that MYC bound to the promoter; moreover, EGFR knockdown inhibited MYC binding to the promoter (Fig. 2H), suggesting that EGFR induces MYC recruitment to the SOX2 promoter. These findings demonstrated the reciprocal feedback regulation between SOX2 and EGFR in lung cancer cells.

SOX2 Knockdown Induces Autophagy, Apoptosis, and Mitochondrial Abnormalities

During the course of study, we observed that SOX2 silencing induced the accumulation of abundant autophagic vacuoles in the cytoplasm of A549 cells, which were stained with acridine orange (Fig. 3A). These vacuoles were also observed in EGFR-silenced A549 cells, albeit to a less extent (Fig. 3A). Green fluorescent protein (GFP)-labeled LC3, a cytoplasmic protein, is conjugated with phosphatidylethanolamine to form LC3-II during autophagy and then aggregates to form punctate dots, which can be visualized by fluorescence [24]. To confirm the involvement of SOX2 in autophagy, we silenced SOX2 in GFP-LC3-overexpressing A549 cells; formation of GFP-LC3-II puncta surged upon SOX2 knockdown (Fig. 3B). Various stages of autophagosomes and the formation of autophagolysosomes in SOX2 knockdown cells were also observed under transmission electron microscope (Fig. 3C). Because autophagy is associated with mitochondrial integrity, we further assayed the effect of SOX2 silencing on mitochondrial function. Oxygen consumption analysis indicated that SOX2 silencing decreased the mitochondrial reserve capacity (Fig. 3D), suggesting that SOX2 is essential for the maintenance of mitochondrial integrity. Flow cytometry with annexin V and propidium iodide staining demonstrated that SOX2 knockdown caused a significant amount of apoptosis in A549 lung cancer cells (Fig. 3E). Immunoblotting demonstrated that SOX2 silencing induced expression of the autophagy marker LC3-II and the apoptotic markers cleaved PARP and caspase-3 (Fig. 3F). These data indicated that SOX2 is necessary for the survival of SOX2-expressing lung cancer cells.

Figure 3.

SOX2 regulates apoptosis, autophagy, and mitochondrial integrity. (A): Fluorescence microscopy of acridine orange staining of A549 cells infected with scrambled control, shSOX2, shSOX2-2, or shEGFR vectors. shSOX2 and shSOX2-2 target different regions of the SOX2 mRNA. Scale bars = 10 μm. (B): Fluorescence microscopy of GFP-LC3-labeled A549 cells infected with a scrambled control or shSOX2 lentiviral vector (left). GFP-LC3 puncta-positive cells were quantified (right). Scale bars = 5 μm. Results are representative of three independent experiments and are expressed as the means ± SD. ***, p < .001. (C): Transmission electron microscopy of A549 cells infected with scrambled control or shSOX2 lentiviral vectors. Organelles: nucleus (N), mitochondria (M), isolation membrane (IM), autophagosome (A), and autophagolysosome (AL). Scale is indicated in the image. (D): Oxygen consumption assay in A549 cells infected with a scrambled control or shSOX2 lentiviral vectors, as assessed by sequential treatment with three mitochondrial inhibitors, (A) 10 μM oligomycin-A, (B) 10 μM carbonyl cyanide m-chloro phenyl hydrazone (CCCP), and (C) 5 μM antimycin-A (left). Oligomycin-A, CCCP, and antimycin-A were used for ATP synthase inhibition, proton gradient uncoupling, and electron transport chain inhibition, respectively. The reserve capacity is calculated by dividing the CCCP-induced maximal oxygen consumption rate by the pre-oligomycin rate (right). Results are expressed as the means ± SD. **, p < .01. (E): Annexin V-FITC/PI staining and flow cytometry analysis of apoptosis in A549 cells infected with a scrambled control (Sc) or shSOX2 lentiviral vector. (F): Immunoblotting of SOX2, autophygic (LC3-II), and apoptotic (Cleaved PARP and CASP3) markers with β-actin as control in A549 cells infected with scrambled control or shSOX2 lentiviral vectors. Abbreviation: GFP, green fluorescent protein.

SOX2 Regulates BCL2L1 Prosurvival Signaling

Because BCL2L1 is a mitochondrial protein that regulates autophagic and apoptotic pathways, both of which are affected by SOX2 knockdown, we tested whether SOX2 regulates the expression of BCL2L1 in lung cancer cells. Q-PCR and immunoblotting analyses demonstrated that BCL2L1 was downregulated in SOX2-silenced CL1-0 cells (Fig. 4A). Ectopic SOX2 expression increased BCL2L1 levels in A549 cells (Fig. 4A). TESS analysis of the BCL2L1 promoter sequence identified a possible SOX2 binding site in the BCL2L1 promoter region 94–101 bp downstream of the transcriptional start site. To examine whether SOX2 binds to the BCL2L1 promoter, we performed ChIP on the BCL2L1 promoter with antibodies against SOX2 (Fig. 4B); the results indicate that SOX2 bound to the BCL2L1 promoter and that SOX2 silencing decreased promoter occupancy by SOX2 (Fig. 4B). These data suggested that SOX2 regulates BCL2L1expression by binding to its promoter. In addition, ectopic BCL2L1 expression prevented SOX2 silencing-induced production of LC3-II, PARP, and caspase-3 (Fig. 4C), indicating that loss of BCL2L1 mediates SOX2 silencing-induced cell death. Consistent with this result, acridine orange staining indicated that ectopic BCL2L1 expression attenuated SOX2 silencing-induced autophagy (Fig. 4D, top). Flow cytometry with annexin V and propidium iodide staining also demonstrated that ectopic expression of BCL2L1 rescued the increase in apoptosis induced by SOX2 silencing (Fig. 4D, bottom). Furthermore, an oxygen consumption assay demonstrated that ectopic BCL2L1 expression partially rescued mitochondrial reserve capacity in SOX2 knockdown cells, suggesting a critical role for SOX2-BCL2L1 signaling in the maintenance of mitochondrial integrity (Fig. 4E). These data indicate that SOX2 signals through BCL2L1 to regulate cell survival by preventing apoptosis and autophagy as well as maintaining mitochondrial integrity.

Figure 4.

SOX2 regulates BCL2L1 prosurvival signaling in lung cancer. (A): Regulation of BCL2L1 by SOX2. Quantitative RT-PCR of BCL2L1 expression (left) and immunoblotting for BCL2L1 and SOX2 expression (upper right) in CL1-0 cells infected with scrambled control (Sc) or shSOX2 lentiviral vectors. Immunoblotting for BCL2L1 and SOX2 expression in A549 cells infected with a lentiviral vector expressing SOX2 or an empty control vector (lower right). (B): Chromatin immunoprecipitation (ChIP) analysis of SOX2 binding to the BCL2L1 promoter. SOX2 binding to the BCL2L1 promoter was determined by ChIP of A549 cells infected with a scrambled control or shSOX2 lentiviral vector. Results are expressed as the means ± SD. **, p < .01. (C): Immunoblotting of whole-cell lysates derived from A549 cells infected first with a lentiviral vector expressing BCL2L1 or an empty control vector and subsequently infected with scrambled control or shSOX2 lentiviral particles. (D): Fluorescence microscopy of acridine orange staining of A549 cells infected with shSOX2 and/or BCL2L1 lentiviral vectors (top). Scale bars = 10 μm. Annexin V-FITC/PI staining and flow cytometry analysis of apoptosis in A549 cells infected with shSOX2 and/or BCL2L1 lentiviral vectors (bottom). (E): Oxygen consumption assay in A549 cells infected with scrambled control (Sc) shRNA plus a BCL2L1 or control (Ctrl) lentiviral vector, or with shSOX2 plus BCL2L1, as assessed using sequential treatment with (A) 10 μM oligomycin-A, (B) 10 μM carbonyl cyanide m-chloro phenyl hydrazone (CCCP), and (C) 5 μM antimycin-A (left). The reserve capacity is calculated by dividing the CCCP-induced maximal oxygen consumption rate by the pre-oligomycin rate (right). Results are expressed as the mean ± SD. *, p < .05; ***, p < .001.

SOX2-BCL2L1 Signaling Mediates Chemoresistance by Lung Cancer Cells

Because SOX2 regulates BCL2L1, which helps determine the level of chemoresistance in various cancers, we further examined whether SOX2 regulates chemoresistance in lung cancer cells. Under cisplatin or paclitaxel treatment, two chemotherapeutics commonly used to treat lung cancer, the cell viability of SOX2-expressing and/or BCL2L1-scilencing cells was determined (Fig. 5A). We found SOX2 expression increased the cell viability, which can be further attenuated by BCL2L1-silencing (Fig. 5A). Consistently, clonogenic assays indicated ectopic expression of SOX2 or BCL2L1 prevented cell death in the presence of cisplatin or paclitaxel (Fig. 5B, 5C). BCL2L1 knockdown attenuated SOX2-induced chemoresistance in A549 cells (Fig. 5D). These results supported the notion that SOX2 promotes BCL2L1 expression, leading to increased chemoresistance in lung cancer cells.

Figure 5.

SOX2 promotes chemoresistance in lung cancer cells. (A): Cell viability evaluated by WST-1 assay in SOX2-expressing (SOX2) or vector-only (Ctrl) A549 cells infected with scrambled control or shBCL2L1 lentiviral particles under treatment with indicated concentrations of cisplatin (left) or paclitaxel (right) for 48 hours. (B): Clonogenic assay in A549 cells infected with HR′-puro (Ctrl) or HR′-puro-SOX2 (SOX2) under treatment with 5 μM cisplatin (top) or 1 nM paclitaxel (bottom) for 4 weeks. Results are representative of three independent experiments and are expressed as the means ± SD. ***, p < .001. (C): Clonogenic assay in A549 cells infected with HR′-puro (Ctrl) or HR′-puro-BCL2L1 (BCL2L1) under treatment with 5 μM cisplatin (top) or 1 nM paclitaxel (bottom) for 4 weeks. Results are representative of three independent experiments and are expressed as the means ± SD. ***, p < .001. (D): Clonogenic assay of SOX2-expressing A549 cells infected with scrambled control or shBCL2L1 lentiviral particles under treatment with 5 μM cisplatin (top) or 1 nM paclitaxel for 4 weeks (bottom). Results are representative of three independent experiments and are expressed as the means ± SD. ***, p < .001.

SOX2 Promotes Lung Tumorigenesis

To investigate the role of SOX2 in tumorigenesis in vivo, we injected SOX2-overexpressing or vector-only control A549 cells subcutaneously into nude mice (Fig. 6A). We examined tumor volume over time and observed that SOX2 overexpression significantly increased the tumor growth rate (Fig. 6A, left). Ectopic SOX2 expression also enhanced the weight of tumors harvested 39 days after injection (Fig. 6A, right). To examine the expression of SOX2, EGFR, and BCL2L1 during lung tumorigenesis, lung tumors generated in the SOX2-expressing xenograft mouse model were excised and analyzed by immunohistochemistry for SOX2, EGFR, BCL2L1, and Ki67, a marker for cell proliferation. SOX2, EGFR, BCL2L1, and Ki67 were highly expressed in SOX2-expressing xenograft tumors (Fig. 6B), suggesting that SOX2 promotes cell proliferation in this model. These data support the notion that SOX2 promotes the expression of EGFR and BCL2L1, leading to increased tumorigenesis.

Figure 6.

SOX2 enhances lung tumor growth in a xenograft model. (A): A549 cells infected with HR′-puro (Ctrl) or HR′-puro-SOX2 (SOX2) were injected subcutaneously into the flank region of nude mice. Tumor volumes were monitored over time (left). Tumor weights were measured 39 days after injection (right). Error bars indicate SEM (n = 10). *, p < .05. (B): Excised xenografted tumors from the mice in (A) were subjected to immunohistochemical staining for SOX2, EGFR, BCL2L1, and Ki67. Scale bars = 200 μm. (C): A549 cells were stably transfected with pLKO-tet-on-shSOX2, which encodes a doxycycline (Dox)-inducible shSOX2, to generate A549-pLKO-tet-on-shSOX2 cells in which endogenous SOX2 levels could be downregulated by treatment with doxycycline. Quantitative RT-PCR and immunoblotting were used to measure SOX2 mRNA and protein expression, respectively, in A549-pLKO-tet-on-shSOX2 cells treated with or without doxycycline for 24 hours (left). A549-pLKO-tet-on-shSOX2 cells were injected subcutaneously into the flank region of nude mice for 3 weeks, after which the mice were given drinking water containing or lacking doxycycline. Tumor volumes were monitored over time (middle). Tumor weights were measured 63 days after injection (right). Error bars indicate SEM (n = 10). *, p < .05; **, p < .01.

To evaluate the role of endogenous SOX2 in tumor formation, we subcutaneously implanted nude mice with A549 cells expressing doxycycline-inducible shRNA specific for SOX2 (Fig. 6C). When palpable tumor bulges were observed in the xenografted mice 3 weeks after cell injection, half of mice were given doxycycline in their drinking water. Doxycycline treatment decreased SOX2 expression and attenuated tumor growth (Fig. 6C). These data demonstrate that SOX2 expression promotes tumorigenesis in vivo.

SOX2 Signaling Correlates with Poor Prognosis in Patients with NSCLC

To investigate the possible involvement of SOX2 signaling in the development of lung cancer malignancy, we investigated SOX2 expression by immunohistochemistry using a panel of specimens from 175 individuals with NSCLC (Fig. 7A, a). Patient characteristics are summarized in Supporting Information Table S1. Kaplan-Meier survival analysis was then conducted to determine the prognostic significance of SOX2 expression in these patients. We found that patients in the SOX2-positive group had lower overall survival rates than those in the SOX2-negative group (Fig. 7A, b). The median survival times for the SOX2-positive and SOX2-negative groups were 11.5 and 29.1 months, respectively (Fig. 7B, top). Furthermore, SOX2 expression correlated with advanced stages (Fig. 7B, bottom). These data suggest that SOX2 expression may encourage malignancy in NSCLC.

Figure 7.

SOX2 signaling in primary lung cancer. (A): Correlation of SOX2 expression with poor prognosis in patients with NSCLC. (a) Representative pictures of the immunohistochemical analysis of SOX2 expression in NSCLC tumor specimens. A SOX2-negative adenocarcinoma case (left), a SOX2-positive adenocarcinoma case (middle), and a SOX2-positive squamous cell carcinoma case (right). Scale bars = 100 μm. (b) Kaplan-Meier analysis of the overall survival of 175 patients with NSCLC based on SOX2 expression. (B): The survival rate of the patients in (A) based on SOX2 expression (upper) and correlation analysis between tumor stages with SOX2 expression (lower). (C): Correlation analysis of SOX2, EGFR, and BCL2L1 transcript levels in 71 primary human lung tumors. (D): A model for SOX2 mediated EGFR/BCL2L1 signaling in lung tumorigenesis. In a subset of lung cancer cells, SOX2 binds to the EGFR promoter to induce EGFR expression, and EGFR stimulates MYC to bind to the SOX2 promoter, generating a positive feedback loop and enhancing tumor cell proliferation. Moreover, SOX2 binds to the BCL2L1 promoter to induce BCL2L1 expression, leading to increased cell survival and chemoresistance. Abbreviations: EGFR, epidermal growth factor receptor; NSCLC, non-small-cell lung cancers.

The above data indicate that SOX2 regulates EGFR and BCL2L1 transcription in lung cancer cells. We further examined the mRNA expression levels of SOX2, EGFR, and BCL2L1 in 71 lung tumors. The relative expression of SOX2, EGFR, and BCL2L1 in tumor was subjected to Spearman's rho correlation analysis. SOX2, EGFR, and BCL2L1 expression levels were significantly correlated (Fig. 7C). These data support the in vitro observation that SOX2 regulates EGFR and BCL2L1 expression to contribute to lung tumorigenesis.

Discussion

The intriguing observation that cancer cells have biological features similar to stem cells suggests that cancer and stem cells may share the same regulatory signaling. In this study, we observed that expression of SOX2 is associated with poor prognosis in NSCLC patients. We further demonstrated that SOX2 is expressed in a subclass of lung cancer cells, the proliferation, survival, and chemoresistance of which are dependent on SOX2 expression. Because SOX2 is essential for maintaining stemness properties such as self-renewal and injury repair in lung stem/progenitor cells and is associated with tumor progression, SOX2 signaling in lung tumors may represent cancer stemness signaling, which not only elicits stem cell features but also encourages malignancy in lung cancer.

Homeostasis in normal lung tissues is maintained by SOX2 and EGFR signaling [3, 9]. During the repair of bronchiolar injury, both SOX2 and EGFR participate in the proliferation of lung progenitor cells, which replenishes damaged tissues [3, 9]. A recent study reported that both SOX2 and EGFR are required for self-renewal of neural progenitor cells and that their effects are enhanced through a positive EGFR-SOX2 feedback loop [25]. In this study, we found that SOX2 induced EGFR expression and EGFR stimulated SOX2, thus forming a mutual induction circuit in lung cancer cells. It is widely recognized that lung cancer often originates via EGFR-mediated oncogenesis [10]. Crosstalk between the SOX2 stemness signaling and the EGFR oncogenic pathway provides a molecular mechanism for SOX2-mediated oncogenesis. Although the mechanism of how the SOX2-EGFR signaling in normal stem cells is altered toward a deregulated oncogenic pathway in a subset of lung cancer cells may require further study, the presence of the SOX2-EGFR feedback loop in lung cancer cells suggests that SOX2 signaling used by adult lung stem/progenitor cells is adopted by lung cancer cells to promote tumor progression.

Using the EOS-S(4+) reporter, which contains four copies of SRR2, we observed that SOX2 signaling is highly active in CL1-0 but not in NL-20 cells. Because epigenetic regulation of SRR2 is involved in controlling SOX2 expression during differentiation of ESCs and adult progenitor cells [26, 27], it is possible that the methylation status of SRR2 in part regulates SOX2 expression in lung (cancer) cells. Indeed, we observed that methylation of SRR2 was higher in NL-20 than in CL1-0 cells (Supporting Information Fig. S4). The expression of SOX2 in NL-20 cells was boosted by epigenetic modulating agents (Supporting Information Fig. S4), indicating the involvement of epigenetic regulation of SOX2 in lung cancer cells. In addition, we found that MYC binds to the SOX2 promoter to activate its expression upon EGFR stimulation. MYC is necessary to maintain self-renewal in ESCs as well as adult progenitor cells [28, 29] and greatly enhances the formation of induced pluripotent stem cells, in which the SOX2-mediated stemness circuit is generated by transcriptional regulation and epigenetic reprogramming [30]. Our findings that MYC regulates SOX2 expression in lung cancer cells further provide a novel transcriptional regulatory mechanism of SOX2, highlighting the critical role of MYC in the establishment of SOX2 signaling.

It has been suggested that the stem cell signaling is associated with chemoresistance during cancer progression [31]; however, it is not clear how chemoresistance is induced by the stem cell signaling. Here, we observed that SOX2 expression promoted chemoresistance in lung cancer cells. SOX2 increased BCL2L1 expression, which protected cells from apoptosis and autophagy. Our data suggest that SOX2-BCL2L1 signaling maintains mitochondrial integrity and enhances cell survival, leading to increased chemoresistance. Because BCL2L1 is a prosurvival regulator and plays a critical role in crosstalk between the apoptotic and autophagic pathways [32], loss of BCL2L1 may affect both apoptosis and autophagy. Indeed, we observed that both apoptosis and autophagy were induced upon SOX2 silencing and that both could be rescued by ectopic BCL2L1 expression, supporting our notion that SOX2 induces BCL2L1-mediated survival signaling.

SOX2 regulates EGFR in not only lung adenocarcinoma cell lines (e.g., CL1-0, A549) but also H2170 squamous cell carcinoma cells (Supporting Information Fig. S5). Although gene copy number amplifications of EGFR and SOX2 have been reported in NSCLC [7], no amplification of either of these two genes was observed in CL1-0, A549, or H2170 cells (Supporting Information Fig. S6). Interestingly, cell lines with EGFR mutations (e.g., H1975, PL1, and HCC827) have extremely low SOX2 levels (Fig. 1A), implying that the mechanism of tumor initiation in EGFR-mutated tumors differs from that in SOX2-expressing lung cancers. Consistent with this observation, the correlation between expression levels of EGFR and SOX2 can be observed among NSCLC cell lines used in Figure 1A only if EGFR-mutant cell lines such as H1975, PL1, and HCC827 are excluded (data not shown). Activating mutations of EGFR in the form of deletions in exon 19 (del19) or the missense mutation L858R in the tyrosine kinase domain impart an increased affinity for EGFR tyrosine kinase inhibitors (TKIs) and reduced affinity for ATP as compared to wild-type receptor, thus contributing to the therapeutic effect of EGFR-TKIs, such as gefitinib or erlotinib, on patients with EGFR mutation [33]. As SOX2 induces wild-type EGFR in lung cancer cells, which is resistant to EGFR-TKI, the therapeutic approach may differ between SOX2-expressing and EGFR-mutated lung tumors. Characterization of differential clinical, pathological, and molecular features between SOX2-expressing and EGFR-mutated NSCLC is currently underway.

SOX2 knockdown caused significant cell death in some lung cancer cells but not in noncancerous cells such as BEAS-2B and NL-20 (Supporting Information Fig. S2), indicating that SOX2 signaling is essential for a subclass of lung cancer cells but not for normal bronchial epithelial cells. Consistent with these results, we observed that SOX2 knockdown attenuated tumor growth in a xenograft mouse model. These data suggest that SOX2 may serve as a potential therapeutic target for lung cancer treatment.

Conclusion

In conclusion, we have discovered that SOX2 expression is associated with poor prognosis in NSCLC patients. SOX2 promotes cell proliferation and activates EGFR expression in a positive feedback loop in lung cancer cells; moreover, SOX2 induces BCL2L1 prosurvival signaling, eliciting chemoresistance (Fig. 7D). Our discovery provides the first evidence that SOX2 induces EGFR and BCL2L1 to endow cancer cells with proliferation and survival advantages, supporting the involvement of SOX2 signaling in lung cancer development.

Acknowledgments

This research was supported by the Institute of Biomedical Sciences, Academia Sinica, National Yang-Ming University, National Tsing Hua University and the National Science Council (NSC100-2325-B-010-011, NSC100–3112-B-010-004 and NSC102-2320-B-007-002), Executive Yuan, Taiwan, R.O.C.

Disclosure of Potential Conflicts of Interest

The authors indicate that they have no conflict of interest.

Ancillary