Introduction
Despite significant therapeutic advances, lung cancer causes the maximum number of cancer related deaths worldwide [
1]. In the United States, ~85% of the patients diagnosed with NSCLCs, die within five years [
2‐
4], thus, highlight a need for better understanding of the cellular and molecular events underlying the genesis of this disease. Cancer stem cell model has emerged as a viable explanation for the initiation and progression of the aggressive cancers like NSCLCs.
Cancer stem cell model suggests that cancer stem-like cells (CSCs) are a subpopulation of cells within the tumor that have the deregulated properties of normal stem cells with sustained self-renewal, and can generate secondary tumors that recapitulate the heterogeneity and diversity of original tumor [
5‐
8]. CSCs are considered to be responsible for tumor initiation, propagation, recurrence and resistance to therapy [
9,
10]. Hoechst 33342 dye excluding cells, termed side-population (SP) cells, have been described as CSCs in a variety of tumor types, including NSCLCs [
11], where they have been shown to display increased tumorigenicity when transplanted into immunocompromised mice [
12,
13] as compared to major population (MP) cells. SP phenotype is dependent on the differential ability of cells to efflux the Hoechst 33342 dye via the ATP-binding cassette (ABC) family of transporter protein, mainly ABCG2 (breast cancer resistance protein, BRCP1) which is specifically expressed on the cell membrane of stem cell populations [
14]. Earlier studies have demonstrated the existence of SP cells in various established human NSCLC cell lines [
11] but their ability to generate tumors in lung microenvironment as well as the signaling pathways governing their stem-like properties remain to be elucidated.
The transcription factors Oct4, Sox2 and Nanog have been identified as core regulators that maintain the self-renewal of embryonic stem cells [
15]. These factors are overexpressed in various cancers and are associated with malignant progression and poor prognosis including NSCLCs [
16,
17], suggesting that the core regulators that govern normal stem cell self-renewal may also maintain the stem-like properties of CSCs in cancers. However, the influence of NSCLC specific oncogenic pathways on the expression of these factors remains relatively unknown. Alterations in EGFR-gene like copy number gains and/or mutant allele-specific amplifications are associated with NSCLC pathogenesis. In addition, activation of EGFR signaling increases the self-renewal capacity of neural precursor cells and brain tumor stem cells [
18‐
20]
. In this study, we provide biochemical and biological evidence that SP cells isolated from established human NSCLC cell lines and tumors are highly enriched in NSCLC-CSCs and EGFR-Src-Akt signaling axis contributes significantly to the self-renewal of SP cells. Interestingly, Sox2 transcription factor is the predominant downstream target of EGFR signaling in these cells and plays a major role in self-renewal growth and expansion of SP cells, independent of Oct4 and Nanog.
Discussion
In the current study, we used the SP phenotype to identify and enrich a subpopulation of NSCLCs with the properties ascribed to CSCs. The studies presented here demonstrates a specific and significant role for EGFR signaling cascade in facilitating the self-renewal growth and expansion of the side population cells from NSCLCs.
Our study, in accordance with earlier studies [
11], [
33], confirmed the presence of SP cells in established human NSCLC cell lines and in human tumor xenografts with the properties of CSCs. Comparing the self-renewal ability of SP and MP cells isolated from human tumor xenografts, we found that approximately 0.2% SP cells were able to self-renew and form spheres, whereas MP cells were unable to self-renew. Comparing the percentage of sphere forming cells in SP cells, we estimate that approximately 1-2% of SP cells from established cell lines may have stem-like properties; therefore, SP phenotype may not be the exclusive marker for CSCs, but can be used to enrich stem-like cells from NSCLCs.
SP cells were found to be more tumorigenic
in vivo, confirming the enrichment of tumor initiating cells in SP compartment. These cells were able to produce highly invasive disease upon implantation into the lungs. Also, the direct association of stem-like cells with generation of metastatic disease may be supported by our observation where a significant correlation was observed between high Sox2 expressions in the metastatic tumors of lung adenocarcinoma patients. Recent reports indicate that the normal epithelial cells acquire the CSCs properties upon induction of EMT governed by various cytokines and growth factors from stromal cells [
10,
25]. Our results demonstrate that SP-cells intrinsically exhibit loss of epithelial markers and/or the gain of mesenchymal markers as compared to MP cells and could be due to the higher expression of transcription factors Twist, Slug and Snail, which are known to be involved in maintaining the mesenchymal phenotype. Together with the expression of embryonic stem cell transcription factors like Oct4, Sox2, and Nanog along with the exhibition of EMT like features and orthotopic tumor-forming ability, collectively suggest that SP cells isolated from NSCLC cell lines and tumors have stem-like properties. The observation that EGFR signaling affects stem-like functions of SP cells is intriguing, given that several EGFR tyrosine-kinase inhibitors have efficacy against NSCLCs [
34,
35]. Interestingly, EGFR appears to regulate Sox2 levels, through the Src-Akt pathway; Sox2 has been shown to be regulated by Akt in ES cells, through the inhibition of proteasomal degradation [
36]. Consistent with these results, our observation suggest that inhibition of EGFR-Src-Akt signaling downregulates Sox2 levels along with a reduction in ABCG2 levels. This decrease in ABCG2 expression upon EGFR inhibition is probably a causal effect of Sox2 depletion-mediated differentiation of SP into MP cells.
The fact that EGFR-pathway inhibition resulted in specific depletion of Sox2 without any significant effect on Oct4 or Nanog expression suggests that their expression may be regulated through independent mechanisms in NSCLC SP cells. Our results as well as an earlier report [
37] suggest that Sox2 is expressed in both low as well as high stage adenocarcinomas irrespective of their grades. However, Oct4 or Nanog expression was found to be associated only with the high grade lung adenocarcinoma and not expressed in low grade tumors [
17,
37]. Therefore, we predict that the EGFR pathway inhibition may exert its favorable effects only for those tumors where Sox2 is the major determinant in controlling the self-renewal of CSCs. Interestingly, a recent study showed that the ectopic overexpression of Oct4 and Nanog increases the tumor initiating property of A549 cells [
17]. In agreement with these reports, we find that specific and independent depletion of Oct4 or Nanog also resulted in decrease in SP phenotype but in a cell type dependent fashion (Data not shown). Two recent reports demonstrate that ectopic expression of Sox2 increased the frequency of side population cells and tumor formation in mouse and human NSCLC cell lines [
33,
38]. These reports strongly suggest that Sox2 expressing cells harbor the stem cell-like properties. Our observation further strengthens this postulation where we demonstrate that Sox2 depletion was sufficient to inhibit the self-renewing property SP cells in all the three NSCLC cell lines.
In addition to the mutation in EGFR signaling, perturbation of p53 activity is another important event occurs in initiation and progression of NSCLCs [
39]. Recently, p53 is shown to have certain roles in promoting the differentiation of human embryonic stem cell through repression of factors like Oct4, Klf4, Lin28A, and Sox2 [
40]. However, there is not much information available on the direct role of p53 transcriptional activities in regulating Sox2 expression in stem-like cells in cancer, and would be interesting to explore in future.
Materials and methods
Cell lines and tumor samples
H1650, and H1975 cell lines were obtained from ATCC and maintained in RPMI or DMEM containing10% fetal bovine serum (FBS; Mediatech) in 5% CO2 at 37°C. Human tumor xenografts were obtained from SA laboratory.
Inhibitors, siRNAs and antibodies
Gefitinib, Erlotinib, BIBW2992 and Dasatinib were purchased from LC laboratories. PP2 and Fumitremorgin C (FTC) were purchased from Sigma Inc. In the present study, Gefitinib or erlotinib is used at 500 nM, dasatinib or BIBW2992 is used at 200 nM and PP2 is used at 1 μM dose. siRNA against EGFR, Src family kinases, Akt and Sox2, Oct4 and Nanog was purchased from Santa Cruz Biotechnology or OriGene Technology Inc. Primary antibodies against Sox2 (#3579), Oct4 (#2750), Nanog (#4903), Phos-Src-pY416 (#2101), pERK1/2 (#4376) and phospho-AKT-pS473 (#4058) were purchased from Cell Signaling Technology; Phos-EGFR-pY1068 (#44788G) from Invitrogen; EGFR neutralizing antibody (#05-101) from Milipore and isotype matched mouse IgG were purchased from Biolegend.
RNA preparation and qRT PCR analysis
RNA preparation and RT-PCR analysis was performed as described earlier [
41]. Fold inductions were calculated using the formula 2
–(ddCt) using GAPDH as internal control gene. The gene-specific primer pairs were as follows. ABCG2 (F) 5’-CAC AAG GAA ACA CCA ATG GCT-3’, ABCG2 (R) 5’-ACA GCT CCT TCA GTA AAT GCC TTC-3’; Oct4 (F) 5’-ACA TCA AAG CTC TGC AGA AAG AAC-3’, Oct4 (R) 5’-CTG AAT ACC TTC CCA AAT AGA ACC C-3’, Sox2 (F) 5’-GGG AAA TGG GAG GGG TGC AAA AGA-3’, Sox2 (R) 5’-TTG CGT GAG TGT GGA TGG GAT TGG-3’, Nanog (F) 5’-AGA AGG CCT CAG CAC CTA-3’, Nanog (R) 5’-GGC CTG ATT GTT CCA GGA TT-3’; Twist (F) 5’-CTC GGA CAA GCT GAG CAA GAT TCA GA-3’, Twist (R) 5’-CGT GAG CCA CAT AGC TGC AGC-3’, Slug (F) 5’- ACA CAT TAC CTT GTG TTT GCA AGA TCT-3’, Slug (R) 5’- TGT CTG CAA ATG CTC TGT TGC AGT G-3’, Snail (F) 5’- CCT CAA GAT GCA CAT CCG AAG CCA C-3’, Snail (R) 5’- CCG GAC ATG GCC TTG TAG CAG C-3’, GAPDH (F) 5’-GGT GGT CTC CTC TGA CTT CAA CA-3’, GAPDH (R) 5’-GTT GCT GTA GCC AAA TTC GTT GT-3’.
Hoechst 33342 dye efflux assay for SP analysis and cell sorting
Adherent cells were harvested using accutase reagent (Sigma Inc). Human Tumor tissue grown in athymic nude mice was minced, enzymatically digested with 0.2% collagenase IV (Worthington Biochemical Corporation) prepared in 10% FBS containing medium for 60 min at 37°C. The digest was further disaggregated by passing through 10 ml pipette several times and filtered through 100/70-μm cell strainer to obtain a single cell suspension. Cells were washed and resuspended in HBSS at 1X10
6 cells/ml density and incubated with 4 μg/ml of Hoechst 33342 dye (Invitrogen) for 90 min at 37
0C in presence or absence of 1 μM FTC, as described by Goodell et al. [
21]. Cells were incubated with 2 μg/ml Propidium iodide (PI; Sigma Inc) before analysis to visualize and exclude the non-viable cells. The Hoechst 33342 dye was excited at 350 nm using UV laser and its fluorescence was analyzed using 400–500 nm BP filter for blue emission and 640–680 nm BP filter in combination with 655 nm LP-filter for red emission. Flow cytometers from BD Biosciences were used for data acquisition. Data were acquired using LSRII or FACS Vantage (DiVa), and sorted using FACS Vantage (DiVa) cell sorter. Data analyses were done using FlowJo software (Tree Star). Cell cycle analyses for fixed cells were performed for PI stained cells using Vindelov method with similar protocol as described earlier [
41].
Sorted SP or MP cells were plated in 96 well plates at the density of 10,000 cells/ml (1000 cells/well in 100 μl medium) in serum free stem cell selective media (DMEM/F12K (1:1) (Invitrogen), supplemented with 1X-N2 supplement (Invitrogen), 10 ng/ml EGF and 10 ng/ml bFGF (Sigma)) and allowed to grow as spheres for 10 days. Images of the spheres were taken using phase contrast microscope (Nikon) and total numbers were counted. To study the effect of drugs on the self-renewal of SP cells, drugs were added to the respective wells on day 1 and 5 and size and number of the spheres were analyzed on day 10.
Immunofluorescence
For immunostaining, spheres were transferred to poly D-lysine/Laminin coated glass surface for 18 h. For monolayer cultures, cells were directly plated over the poly D-lysin/Laminin coated glass surface and cultured or treated in stem cell selective media as indicated. Immunofluorescence staining was performed as described previously [
42]. Cells were observed using a Leica TCS SP5 confocal microscope (Leica Microsystems) at × 630 magnification.
Immunohistochemistry
Human lung cancer tissue microarray (TMA) slides with stage I/II or stage IV NSCLC patients were obtained through Lung Cancer Specialized Program of Research Excellence (SPORE). TMA slide with stage I/II tumor samples contained usable cores from 193 patients, and TMA slide with stage IV tumor samples contained usable cores from 103 patients including 17 adenocarcinoma samples from the metastatic sites. The Immunohistochemical staining was performed as described [
42]. The samples were scored by a pathologist (D. Coppola). The semiquantitative score was reached by taking into consideration both cellularity and intensity of expression (semiquantitative score = cellularity × intensity). Cellularity was scored as follows: a score of 3 equals to greater than 66% cellularity, a score of 2 equals to 34%–65% cellularity, and a score of 1 equals to less than 33% cellularity. Intensity was scored as follows: a score of 3 equals to strong intensity, a score of 2 equals to moderate intensity, and a score of 1 equals to weak intensity [
42]. The score of 1 or above was considered as positive expression of Sox2. The images were captured at × 200 magnification.
5-weeks-old female SCID-beige mice were used for these experiments under an IACUC approved protocol. For orthotopic implantation of tumor cells, sorted SP or MP cells from A549 cell line stably expressing luciferase gene (A549-Luc) were washed with serum-free DMEM-F12K medium and resuspended at indicated numbers in HBSS containing 500 μg/ml growth factor reduced Matrigel. Surgical procedure for orthotopic lung implantation was followed as suggested earlier for intrapulmonary implantation of tumor cells with some modifications [
43]. Specifically, cells were inoculated with 1 ml syringes with 30-gauge hypodermic needles in an open technique under direct visualization into the right lung tissue of SCID mice anesthetized by gas anesthesia (3% isoflurane). Tumor growth/metastases were imaged weekly using bioluminescence by IVIS-200 imaging system from Caliper Corporation. Mice were anesthetized and 30 mg/Kg of D-luciferin in PBS was administered by intraperitoneal (i.p.) injection. Ten minutes after injection, bioluminescence was imaged with a charge-coupled device camera (Caliper) with an imaging time of 2 min. At the end of the experiment, or when mice become moribund, all of the mice were euthanized and individual organs harvested for evaluation of tumor size; distant metastases was determined by bioluminescence of luciferase expressing cells.
Statistical methods
Data were presented as the mean ± standard deviation (SD). To assess the statistical significance of differences, student’s t test was performed. The data were considered statistically significant when the P value was less than 0.05.
Competing interest
We do not have any conflict of interest.
Authors’ contributions
SS conducted the experiments and wrote the initial version of the manuscript; JT and NBS conducted certain experiments; DC did pathological analysis of the samples; EH provided intellectual input; SA provided human tumor xenografts and input; SC directed the project and finalized the manuscript. All authors read and approved the final manuscript.