Background
Despite growing insights into the mutational events that drive the genesis of NSCLC and the development of novel therapeutic strategies, lung cancer remains the leading cause of cancer related deaths. Lung cancer accounts for more deaths than breast, prostate, and colon cancers combined [
1,
2]. Nicotine is the major addictive component of tobacco smoke; while it is not a carcinogen and cannot initiate tumors itself, nicotine has been shown to possess a number of tumor promoting properties in multiple tumor types, both in vitro and in vivo [
3‐
9]. Nicotine exerts its tumor promoting functions through the activation of nicotinic acetylcholine receptors (nAChRs), which are typically expressed on neuronal cells; they are also expressed on cells of endothelial and epithelial origin, including tumor cells [
10‐
12]. Our lab and others have shown that nicotine can promote proliferation, angiogenesis, epithelial-to-mesenchymal transition (EMT), migration, invasion, and survival of cultured non-small cell lung cancer cells. In addition, nicotine could also promote the growth and metastasis of lung and pancreatic cancers in mouse xenograft models, primarily through the α7 subunit of nAChRs [
4,
5,
9,
13‐
15]. More recently, we have reported that nicotine can enhance the self-renewal of a subset of lung adenocarcinoma cells enriched in stem-like cell populations, through the induction of c-Kit ligand/Stem Cell Factor (SCF). SCF is known to promote self-renewal and differentiation of multiple stem cell types through the binding of its receptor, c-Kit [
16‐
19], and this finding reveals a novel mechanism by which nicotine might be promoting tumor progression.
Tumors were traditionally thought to be a disease of clonal origin where a single transformed cell has the ability to give rise to heterogeneous tumor cell populations, with each daughter cell having the same capacity to give rise to more tumor cells. More recently, growing evidence supports the cancer stem cell model, which indicates that cancer stem-like cells (CSCs) arise through reprogramming of adult stem cells or progenitor cells, and these cells are responsible for tumor initiation, maintenance, progression and metastasis [
20]; in addition, the tumor stem-like cells have also been shown to contribute to drug resistance, dormancy, recurrence, and metastasis [
21,
22]. The model proposes that only CSCs are able to initiate tumors; these stem-like cells resemble traditional stem cells in that they are able to self-renew, divide asymmetrically, and are slow cycling [
21]. Given these properties, understanding and targeting CSCs has become an important area of cancer research.
CSCs have been characterized and isolated using cell surface markers, which are differentially expressed on CSCs compared to non-stem-like, differentiated cancer cells. Various markers such as aldehyde dehydrogenase 1 (ALDH1) and CD133 positivity are effectively used for cancers such as breast, colon, brain, pancreas, head and neck [
23‐
26]; however, there is no single marker ubiquitously expressed and used to identify lung cancer CSCs. A subset of tumor cells enriched in CSCs can be isolated based on their ability to efflux Hoechst 33342 dye out of their nuclei through the ABCG2 drug transporter expressed on the cell membrane of the stem-like cells, and have been termed as side-population cells, based on their distribution in flow cytometric sorting [
27‐
29]. Our lab and others have shown that non-small cell lung cancer CSCs can be isolated using SP phenotype from cell lines as well as human tumor xenografts. Such cells were highly tumorigenic and produced highly invasive tumors in mice compared to the non-SP cells, and displayed stem-like properties such as the ability to self-renew, expression of epithelial-to-mesenchymal transition (EMT) markers as well as the classic embryonic stem cell transcription factors Sox2, Oct4, and Nanog [
27,
30]. Sox2 itself was critical to maintain self-renewal of SP cells from NSCLC cell lines, compared to Oct4 and Nanog [
27]. Since we find that Sox2 transcription factor is required to maintain NSCLC CSC stemness and that nicotine acts to enhance stemness, we sought to determine whether this nicotine-mediated promotion of stemness occurs through the induction of Sox2. Here we report that nicotine can induce Sox2 through a Yap1/E2F1/Oct4 signaling axis.
More recently, electronic cigarettes or e-cigarettes have been marketed as a healthy alternative to traditional cigarette smoking, as they do not contain tobacco, which contains multiple carcinogens such as polycyclic hydrocarbons, tobacco specific nitrosamines, and aldehydes [
31,
32]. While e-cigarettes do not contain tobacco carcinogens, they contain nicotine in addition to other components such as propylene glycol, glycerol, and flavorings [
32]. These devices are typically used by pressing a button which activates an internal heating coil which brings the e-liquid containing nicotine to a boil, which is then delivered as a vapor to the user [
32]. The concentration of nicotine present in e-cigarettes varies by brand and container, but is typically represented as percent nicotine by volume (NBV). How nicotine present in e-cigarettes impacts the pathophysiology and health of users remains unclear, and whether the additional components of the e-liquid might abrogate or amplify the effects of nicotine in the context of cancer has not been determined. Here we report that e-cigarette extracts can promote self-renewal in a manner similar to nicotine; further, e-cigarette extracts could induce Sox2 expression, suggesting that exposure to nicotine, either through tobacco smoke or through the use of e-cigarettes, might have deleterious effects.
Methods
Cell lines
Human non-small cell lung adenocarcinoma cell lines A549 and H1650 were obtained from the American Type Culture Collection (ATCC). A549 cells were maintained in Ham’s F12K medium (Cellgro, Mediatech, Inc.) supplemented with 10% fetal bovine serum (Atlas Biologicals), and H1650 cells were maintained in RPMI 1640 (Gibco, Life Technologies, Thermo Fisher Scientific Inc.) containing 10% fetal bovine serum. Normal human bone marrow derived mesenchymal stem cells (hMSCs) were purchased from Lonza and maintained in their mesenchymal stem cell basal growth medium (MSCGM) designed to maintain these cells in a proliferative but not differentiated state. A549 and H1650 cell lines have been validated by ATCC and were validated again on May 25, 2016. hMSCs were pre-validated by Lonza, and only used upto passage 10.
Generation of stable cell lines
A549 cell line was used for generating stable overexpression cells. The Sox2-core-luc (Bora-Singhal et al., 2015) and YAP1 (Addgene #15682), Oct4 (Addgene #17964) [
33] and E2F1 expression vectors [
13] were transfected using FugeneHD reagent (Promega) per manufacturer’s protocol. The transfected cells were selected using G418 and puromycin and maintained in Ham’s F12 K medium; single colonies were selected and expanded for use in experiments.
Nicotine, E-cigarettes, and inhibitor studies
(−)-nicotine (N3876; Sigma-Aldrich) or e-cigarettes (local stores) were used in these studies. A549 or H1650 cells were rendered quiescent by serum starvation in media containing 0.1% fetal bovine serum for 24 h, following which cells were stimulated with 2 μM nicotine or e-cigarette extracts for the indicated time points. For studies using nicotine or e-cigarette extracts in hMSCs, cells were maintained in stem cell media and stimulated with 2 μM nicotine or e-cigarette extracts 24 h after plating, for indicated time points. For studies using signal transduction inhibitors/anti-cancer drugs, cells were rendered quiescent by serum starvation for 24 h, were treated with inhibitors for 30 min prior to stimulation with 2 μM nicotine; the cells were maintained in the serum free medium during nicotine stimulation. The inhibitors used were AZD0530/Saracatinib (Sellekchem) at 10 μM, NVP-BKM120/Buparsilib (Chemietek) at 20 μM, GSK1120212/Trametinib (Chemietek) at 10 μM, LEE001/Ribociclib (Chemietek) at 20 μM, RRD251 at 10 μM, α-bungarotoxin (Sigma) at 10 μM, or visudyne (Sigma) at 2 μM.
Three different brands of e-cigarettes were used to demonstrate the effects; these included Fin, Njoy, and Mistic (which are referred to as E-cig 1, E-cig2, or E-cig3, respectively). E-cigarette liquid was obtained through extraction of an internal liquid-soaked sponge within the devices for E-cig 1 and 2, or by syringe extraction for E-cig 3. E-cig 1, 2, and extracts were 1.6% nicotine by volume (NBV) or 16 mg/ml, 1.5% NBV or 15 mg/ml, and 1.8% NBV or 18 mg/ml respectively as indicated on the manufacturer’s packaging. Molarity of extracts from each brand was calculated based on the molecular weight of nicotine of 162.23, and the working concentration of 2 μM was achieved by serial dilutions of 1:10, 1:9, or 1:11 for E-cig 1, 2, or 3 respectively, to achieve 10 mM, then diluted 1:50 for a final concentration of 2 μM.
siRNAs and antibodies
siRNAs used were purchased from Santa Cruz Biotechnology including Oct3/4 (sc36123), TEF4/Tead2 (sc45232), α7 nAChR (sc42532), E2F1 (sc29297), c-Src (sc44250), Sox2 (sc38408), Yap1 (sc38637), c-Yes (sc29860), and β-arr-1 (sc29741). Antibodies used for western blot against Sox2 (3579s), p-Src (2101s), p-AKT (9018p), pan-AKT (C67E7), Oct4 (2750s), p-ERK1/2 (9101s), and total ERK1/2 (9102s) were purchased from Cell Signaling Technologies; against c-Src (05–184) from EMD Millipore; against E2F1 (sc251) from Santa Cruz Biotechnology; against Yap1 (53–161) from Abnova, α7 nAChR (ab23832 and ab10096) from Abcam; and Actin from Sigma Aldrich (A1978).
Antibodies used for chromatin immunoprecipitation (ChIP) assays included E2F1 (sc193), E2F2 (sc633), E2F3 (sc879), E2F4 (sc1082), E2F5 (sc999), Rb (sc50), from Santa Cruz Biotechnology; Yap1 (ab56701), H327Kme3 (ab9045), and H327Kme1 (ab6002) from Abcam; a rabbit anti-mouse secondary antibody from Pierce was used as a negative control.
Antibodies used for immunofluorescence included Sox2 (3579s) from Cell Signaling Technologies; ZO-1 (339100) from Invitrogen; E2F1 (sc193), Crm1 (sc5595), and E-cadherin (sc8426) from Santa Cruz Biotechnology; α7 nAChR (ab10096) Yap1 (ab56701) from Abcam.
ChIP-PCR experiments
Chromatin immunoprecipitation assays were conducted using previously described protocols [
34,
35]. Interactions of the proteins with specific regions of the Sox2 promoter were detected by PCR amplification using the following primer sequences:
F1–5’-GAAAAGGCGTGTGGTGTGAC3–3′;
R1–5′- CGCTGATTGGTCGCTAGAAAC -3′;
F2–5’-GGGAGTGCTGTGGATGAGC-3′;
R2–5’-GTGGGTAAACAGCACTAAGACTACGTG-3′;
F3–5’-TGTGCGCTGCCTGCACCTGTG-3′;
R3–5’-ACTCCAGCAGAACCAGCCCTG-3′;
F4–5’-ACGTGCTGCCATTGCCCTC-3′;
R4–5’-CGGGTTAGAGGAGGATGAGA-3′.
Transient transfections and luciferase assays
Cells transfected in Opti-MEM medium (Gibco, Life Technologies) using Fugene HD (Promega) transfection reagent following the manufacturer’s protocol. The mutSox2-core-luc construct containing a mutated Oct4 binding site was previously generated using Quikchange Lightening multi-site-directed mutagenesis kit (Agilent Technologies), as previously reported from our lab [
30]
. To confirm the role of E2Fs in regulation of the Sox2 promoter, the 7 E2F consensus binding sites present in the 500 bp region upstream of TSS (where TSS = 0) were mutated. This was done by mutating each of the 4 bp CGCG consensus sites to AATT within the Sox2-core-promoter at the following positions: -37 bp through -40 bp, − 50 bp through -53 bp, − 107 bp though -110 bp, − 119 bp through -122 bp, − 336 bp through -339 bp, − 361 bp through -364 bp, and -476 bp through -479 bp. Mutation of the E2F sites was outsourced to Genscript USA, Inc., and the E2F-mutant Sox2-core promoter was then cloned into pGL3 expression vector by our lab, for use in transient transfection experiments. The expression vectors used were pcDNA3-HA-E2F1, pcDNA3-E2F2, pcDNA3-E2F3, pcDNA3-E2F4, pcDNA3-E2F5, and Yap1 (Addgene #18978). Empty vector pcDNA3 was used as a control. Luciferase assays were conducted 24 to 48 h after transfection per manufacturer’s protocol using the Dual Luciferase Assay system (Promega). Results are reported as relative luciferase activity (RLA) based on the ratio of RLUs1 (firefly luciferase) to RLUs2 (Renilla luciferase: normalization control) values as measured on a Turner Biosystems luminometer.
siRNA transfections and quantitative real-time PCR
Cells were transfected in Opti-MEM (Gibco Life Technologies) with 100 pmol of siRNAs using Oligofectamine reagent (Invitrogen) as per the manufacturer’s protocol. Media was replaced by complete medium containing 10% FBS 4–6 h after transfection. RNA was isolated using Qiagen RNEasy miniprep kit (Hilden, Germany) according to manufacturer’s protocol. First strand cDNA was synthesized using Bio-Rad iScript cDNA synthesis kit (Hercules, CA). mRNA expression was assessed using qRT-PCR (Bio-Rad CFX96 Real Time System) and data were analyzed using the CFX96 software. RT-primers used were as follows:
GAPDH(F): 5’-GGTGGTCTCCTCTGACTTCAACA-3′;
GAPDH(R): 5’-GTTGCTGTAGCCAAATTCGTTGT-3′.
Vimentin(F): 5’-GGACCAGCTAACCAACGACA-3′;
Vimentin(R): 5’-AAGGTCAAGACGTGCCAGAG-3′;
Fibronectin(F): 5’-TAGATGTACAGGCTGACAGA-3′;
Fibronectin(R): 5’-TCTTTCTTAAGCCCTTTGCT-3′;
Yap1(F): 5′- CCCAAGACGGCCAACGTGCC-3′;
Yap1(R): 5′- ACTGGCCTGTCGGGAGTGGG-3′;
Sox2(F): 5′ – GGGAAATGGGAGGGGTGCAAAAGA-3′;
Sox2(R): 5′- TTGCGTGAGTGTGGATGGGATTGG-3′;
ZEB1(F): 5’-AGCAGTGAAAGAGAAGGGAATGC-3′;
ZEB1(R): 5’-GGTCCTCTTCAGGTGCCTCAG-3′;
ZEB2(F): 5’-ATCTGCTCAGAGTCCAATGCAGCAC-3′;
ZEB2(R): 5’-AACAGTATTGTCCACAATCTGTAG-3′.
Data was normalized using GAPDH as an internal control, and fold change was determined using the 2-ΔΔCT method.
Lysate preparation and IP/Western blotting
Cell lysates were prepared and processed for western blotting as described in our previous work [
30,
35]. Protein was detected using ECL reagent from GE Healthcare or Pierce Biotechnology according to standard protocols; actin was used as a control.
For co-immunoprecipitation assays, 200 μg of total protein lysate from A549 and H1650 cells were incubated with 4 μg of indicated antibodies. An equal amount of non-specific IgG from rabbit or mouse serum (Sigma-Aldrich) was used as a negative control. The interacting proteins were detected by western blotting.
Immunofluorescence analysis and confocal microscopy
Immunofluorescence assays were conducted as previously described [
30,
36]. Cells were visualized with a DM16000 inverted Leica TCS SP5 tandem scanning confocal microscope at 630× or 1890× magnification.
Proximity ligation assays (PLA)
PLA studies were conducted as previously described using Duolink assay system (Sigma-Aldrich) [
14,
30,
36]. The images were taken using Leica TCS SP5 confocal microscope (Leica Microsystems) at 630x and 1890x magnification.
Isolation of side-population (SP) cells and self-renewal
SP cells were isolated from heterogenous cell populations using flow cytometry based on Hoechst 33342 dye efflux, and were then plated for self-renewal assays on low-adherence plates in stem-cell selective media, using protocols described in detail earlier [
27,
37]. For experiments involving nicotine or e-cigarette extracts, these were added directly into stem-cell media at the time of plating. For depletion experiments, cells were transfected using siRNA and SP cells were isolated 48 h later.
Wound healing assays
Wound healing or scratch assays were conducted as previously described [
4,
38]. Images were taken every 24 h for 48 h, using EVOS FL microscope system (Life Technologies) at 10× magnification.
Statistical analysis
All data have been statistically analyzed using Microsoft Office Excel 2010 (Microsoft Corporation, Redmond, WA). The data presented here is with ± standard deviation (SD) values derived from three independent experiments unless otherwise stated. The statistical comparisons between the groups were carried out by unpaired two tailed Student’s t-test or one-way ANOVA to calculate the p value for statistical significance. *p < 0.05, **p < 0.01 and ***p < 0.001.
Discussion
CSCs represent a subpopulation of tumor cells with increased tumor-initiating capability. They can divide asymmetrically to replenish the heterogenous tumor bulk, and are highly efficient in initiating tumors upon implantation in animal models [
22]. CSCs are resistant to various treatment modalities in part due to their enhanced ability to efflux drugs; additional reasons include the fact that they are slower cycling, and they express higher levels of anti-apoptotic proteins. At the same time, complete mechanisms underlying the drug resistance of these cells are not fully understood [
22]. Additionally, these cells are thought to remain dormant and facilitate tumor recurrence and metastasis [
22,
25]. Not surprisingly, based on these properties of CSCs, efforts are being made to elucidate mechanisms underlying the biology of CSCs in order to target this subpopulation. CSCs have different gene regulatory programs, including epigenetic changes, than the bulk tumor cells; understanding what these differences are, how the programs are regulated, will open up new opportunities for therapeutic targeting.
We have previously reported that nicotine could enhance self-renewal of NSCLC SP cells [
16]. The schematic In Fig.
6e represents the proposed mechanism of nicotine mediated induction of Sox2 and possibly stemness [
16]. Our earlier studies [
12,
13,
35,
56] as well as the current study showed that nicotine binding to the α7 nicotinic acetyl choline receptor recruits β-arr-1 and Src (Fig.
6e). This results in the activation of Yap1, which has been shown to be a target of Src and Src family members [
7,
53,
54]. α7 nAChR-mediated activation of Src leads to the phosphorylation of Rb and its dissociation from E2F1, enhancing the transcriptional activity of E2F1 (Fig.
6e) [
12,
13,
15,
35,
41,
56]. As mentioned earlier, Yap1 has been found to interact with E2F1 and promote the expression of its downstream targets. Our study suggests a potentially new mechanism by which nicotine induces Sox2 expression in NSCLC cells through Yap1 and its interaction with transcription factors like E2F1 or Oct4 (Fig.
6e)
. We also find that nicotine induces expression of Yap1 itself, and that the nicotine-mediated induction of Sox2 and Yap1 is not just specific to lung cancer cells but is also observed in human mesenchymal stem cells. One previous report has demonstrated the ability of nicotine to induce Yap1 in esophageal squamous cell carcinoma (ESCC), and this occurred through nAChRs [
42]. Interestingly, they find that Yap1 physically interacts with nAChRs and stimulation with nicotine could induce nuclear translocation and activation of Yap1 by disrupting its association with a negative regulatory complex in the cytoplasm composed of α-catenin, β-catenin, and 14–3-3 proteins [
42]. The molecular mechanisms regulating this process are not completely understood.
Our prior studies have shown that Yap1 regulates Sox2 through the binding to Oct4 transcription factor, facilitating self-renewal and vascular mimicry [
30]. Here we report that E2F1 transcription factor can regulate the Sox2 promoter, and that Yap1 binds to E2F1 likely modulating this effect. Further, we also find that nicotine or e-cigarette extracts can increase the binding of Yap1 to both E2F1 and Oct4. Nicotine has been shown to induce E2F1 transcriptional activity through a sequence of signaling events mediated downstream of nAChRs [
35]. Upon nicotine binding, β-arrestin-1 scaffolding protein is recruited to the receptor and activates Src kinase, which subsequently activates Raf-1. Raf-1 then acts to phosphorylate the Rb tumor suppressor protein, which is typically bound to E2F1 during cellular quiescence; but dissociation of hyperphosphorylated Rb from E2F1 allows it to turn on a number of promoters involved in proliferation and survival [
57]. We now find that this pathway might contribute to the induction of stemness, by facilitating the expression of Sox2 (Fig.
6e). The downregulation of Sox2 expression after 72 h of nicotine treatment is intriguing; the possibility exists that the cells undergo a transition to a more differentiated state, which might not require the presence of Sox2 by that time point. Alternately, the cells might have acquired sufficient levels of downstream targets of Sox2 to maintain stemness and self-renewal and my not require Sox2 per se by that later time point. It is also likely that the cells might have undergone metabolic changes that allows them to survive in the absence of Sox2.
Our studies also suggest that Yap1 is induced by a non-canonical signaling mechanism in response to nicotine. The Hippo signaling pathway has been demonstrated to have tumor suppressive roles, but is aberrantly altered in multiple cancers including those of the lung [
58]. Typically the activation of this pathway by upstream mediators Mst1/2 and Lats1/2 results in the inactivation of Yap1 through its phosphorylation, leading to cytoplasmic sequestration and degradation by 14–3-3 protein [
58]. Our results in NSCLC cells suggest that Yap1 is activated through Src and Yes kinases in response to nicotine; the role of the canonical Hippo signaling pathway in the induction remains unclear.
Overall these studies suggest that upon nicotine binding to α7 nAChR, Src is activated and subsequently leads to Yap1 binding to E2F1 and/or Oct4, upregulating Sox2 expression, thereby enhancing self-renewal of CSCs. However, the role of Oct4 in this process is not fully clear. When endogenous expression of Oct4 is knocked down, nicotine could still induce Sox2; in contrast, in cell lines stably expressing a Sox2 promoter containing a mutation of the Oct4 site prevented nicotine-mediated induction of Sox2-luciferase. The molecular basis for the difference in the induction of endogenous Sox2 versus artificially induced Sox2-luciferase remains elusive at this time. It could be that other proteins are forming complexes with Oct4 or E2F1 to regulate Sox2, and these are disrupted by mutation of the Oct4 binding site. It is also possible that post-translational modifications of the proteins involved or histone modifications on Sox2 promoter around the Oct4 binding site play a role.
Our lab had shown that nicotine induces the translocation of β-arrestin-1 scaffolding protein to the nucleus where it binds to E2F1 transcription factors to enhance transcription of E2F target genes [
35]. This was found to occur through the formation of an oligomeric complex consisting of β-arrestin-1, E2F1, and p300 histone acetyltransferase proteins which facilitated the acetylation of histones and E2F1, acting to induce transcription of genes involved in proliferation and survival [
35]. Our initial experiments show that depletion of β-arrestin-1 reduced endogenous levels of Sox2; this raises the possibility that Yap1 is recruited to a complex with β-arrestin-1 and E2F1 on the Sox2 promoter. Alternately, β-arrestin-1 might be recruiting p300 to E2F1 independent of Yap1. It is additionally worth noting that other E2F family transcription factors may be involved, and their role is worth investigating. These are novel findings that might have a significant impact on our understanding of how nicotine promotes self-renewal of stem-like cells from non-small cell lung cancer. Full elucidation of these mechanisms will shed light on the pathophysiology of smoking-related cancers, and reveal new pathways involved in promotion of CSC populations that can potentially be therapeutically exploited.