Epithelial-to-mesenchymal transition (EMT) is strongly associated with cancer progression, but its potential role during premalignant development has not been studied. Here, we show that a 4-week exposure of immortalized human bronchial epithelial cells (HBEC) to tobacco carcinogens can induce a persistent, irreversible, and multifaceted dedifferentiation program marked by EMT and the emergence of stem cell–like properties. EMT induction was epigenetically driven, initially by chromatin remodeling through H3K27me3 enrichment and later by ensuing DNA methylation to sustain silencing of tumor-suppressive microRNAs (miRNA), miR-200b, miR-200c, and miR-205, which were implicated in the dedifferentiation program in HBECs and also in primary lung tumors. Carcinogen-treated HBECs acquired stem cell–like features characterized by their ability to form spheroids with branching tubules and enrichment of the CD44high/CD24low, CD133, and ALDH1 stem cell–like markers. miRNA overexpression studies indicated that regulation of the EMT, stem-like, and transformed phenotypes in HBECs were distinct events. Our findings extend present concepts of how EMT participates in cancer pathophysiology by showing that EMT induction can participate in cancer initiation to promote the clonal expansion of premalignant lung epithelial cells. Cancer Res; 71(8); 3087–97. ©2011 AACR.
Lung cancer associated with tobacco use may soon become the leading cause of cancer-related death worldwide due to the advanced stage at detection and the ineffectiveness of chemotherapy to achieve long-term remission (1). Investigations using malignant tumors and precursor lesions to adenocarcinoma and squamous cell carcinoma have provided insight into some of the genes and pathways that contribute to malignancy (2, 3). Studies using animal models and human tumors have established a link between exposure to chemical carcinogens in tobacco smoke with genomic instability (3), gene mutation (4), LOH (5), and epigenetic silencing of genes through DNA methylation of cytosines and histone modifications (2). Moreover, DNA methylation is now recognized as a causal epigenetic event that occurs during lung cancer progression to affect gene expression (6).
Recently, the activation of the epithelial-to-mesenchymal transition (EMT) program has been implicated as an important step in the metastasis of lung and other tumors (7). This process in cancer pathogenesis involves disruption of normal epithelial integrity, with loss of morphologic features of polarized epithelia, and gain of mesenchymal markers accompanied by the progressive acquisition of a motile and invasive phenotype (8). EMT is characterized by changes in several molecular pathways and networks, with the loss of E-cadherin expression emerging as a critical step driving this developmental program in human cancers, including lung (9–11). Mechanisms for loss of E-cadherin function include promoter CpG hypermethylation, histone modifications, and direct inhibition by zinc finger transcriptional repressors ZEB1, ZEB2, Snail1, and Twist1 (12–17). Recently, specific microRNAs (miRNA) have been described as crucial regulators of the EMT process. miRNAs are evolutionary conserved small RNAs that can modulate gene expression by inhibiting the protein translation process and/or degrading the respective target mRNA (18). The miR-200 family and miR-205 are key determinants of the epithelial phenotype by directly targeting ZEB1 and ZEB2, showing that miRNAs can indirectly regulate E-cadherin expression (19, 20). Adding to the complexity of the EMT regulatory program is the recent discovery of a reciprocal feedback loop in which transcriptional repressors such as ZEB1 and ZEB2 bind to E-boxes within the promoter regions of the miR-200 family to repress their transcription (21, 22).
The morphologic heterogeneity seen in lung tumors suggests that EMT could be an active process driving differentiation and dedifferentiation in early tumor development (23). The E-cadherin gene is somatically inactivated in many cancers types, such as lobular carcinoma of the breast and diffuse gastric carcinoma, in which neoplastic cells throughout the entire tumor mass have lost epithelial characteristics and display a highly invasive phenotype with features of EMT (7). Reduced expression of E-cadherin is also seen in solid cancers at the tumor–stroma boundary where single invading cells exhibit growth patterns that resemble EMT (7). Local invasion of carcinoma in situ spreading through the basement membrane may also involve increased expression of zinc finger transcriptional repressors, such as ZEB1, which, in turn, promotes EMT (24). We have developed an in vitro premalignancy lung model that uses human bronchial epithelial cells (HBEC) to identify genes and pathways critical for neoplastic transformation associated with exposure to tobacco carcinogens (25). Our initial study showed differences in transformation efficiency associated with DNA repair capacity for two HBECs after low-dose treatment with the carcinogens methylnitrosourea (MNU) or benzo(a)pyrene-diolepoxide (BPDE), or both, for 12 weeks (25). Protein levels of cytosine DNA methyltransferase 1 (DNMT1) increased significantly during carcinogen exposure and were associated with promoter hypermethylation of 5 to 10 genes in each transformed cell line (25). The HBECs displayed a change in morphology to a mesenchymal-like appearance, suggestive of EMT, after 4 weeks of carcinogen treatments that persisted throughout the remaining treatments and in transformed cells (colonies recovered from soft agar). The purpose of this study was to characterize the cellular and molecular changes associated with morphology and the impact on transformation.
Materials and Methods
Cell lines and samples
HBECs (from Drs. Shay and Minna, Southwestern Medical Center, Dallas, TX) were established from two different people (HBEC1; smoker without lung cancer; HBEC2; smoker with cancer; ref. 26). Carcinogen exposures were previously described (25). Thirteen lung cancer-derived cell lines (Calu6, Calu3, A549, H358, H522, H23, H1435, H1975, H1993, H2023, H2085, H2228, and HCC827) were obtained from and authenticated by the American Type Culture Collection. Experiments were conducted in cell lines passed for a maximum of 6 months postresuscitation.
Twenty-four frozen carcinomas (stage I, TNM staging system) with distant normal lung tissue were obtained from our New Mexico Lung Cancer Cohort. White bloods cells (WBC) and normal HBECs from cancer-free smokers were used as controls. All persons providing tissue specimens signed informed consent, and the Institutional Review Board of Lovelace Respiratory Research Institute approved this study.
Soft agar assays
Sphere formation assay
A total of 6 × 103 cells were plated in 24-well plates and analyzed in a nonadherent culture conditions, using 10% Matrigel matrix (BD Biosciences). Plates were inspected for colony growth and branching morphology on day 10.
Cells were labeled with antibodies conjugated with fluorescent dyes, anti–CD44-PE (clone G44-26; BD Bioscience), anti–CD24-FITC (clone ML5; BD Bioscience), and anti–CD133/2-APC (clone 293C3; Miltenyi Biotec). The antibodies were diluted in fluorescence-activated cell-sorting (FACS) buffer (1× PBS, 5% FBS) containing 15% blocking reagent (Miltenyi Biotec) and sorted with a flow cytometer. The ALDEFLUOR kit (STEMCELL Technologies Inc.) was used to isolate a cell population with ALDH1 enzymatic activity (29).
Gene expression analysis
RNA was isolated with TRI reagent (Sigma) and reverse transcribed using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems). Quantitative real-time PCR (qRT-PCR) was carried out with the ABI PRISM 7900HT, using inventoried TaqMan assays (Applied Biosystems). Experiments were normalized to GAPDH (glyceraldehyde 3-phosphate dehydrogenase). Data were analyzed with respect to a calibrator sample using the 2−ΔΔCt method (30).
The anti–E-cadherin, anti-vimentin, anti-ZEB1, anti–β-actin were from Santa Cruz Biotechnology and used at 1:1,000 dilution.
miRNA expression analysis
Total RNA (1 μg) was reverse transcribed using the miScript Reverse Transcription Kit (Qiagen). qRT-PCR was carried out using the miScript PCR Kit (Qiagen). Experiments were normalized to RNU5A. Results were reported as RQ with respect to a calibrator sample by using the 2−ΔΔCt method.
Chromatin immunoprecipitation assay
Antibodies specific to histone H3, dimethyl H3K4, dimethyl H3K9 (Abcam), DNMT1 (Imgenex), and trimethyl H3K27 (from Dr. Thomas Jenuwein, Institute of Molecular Pathology, Austria; ref. 31) were used to capture protein–DNA complexes. Rabbit and mouse immunoglobins (IgG) were used for isotype controls. Results were generated by qRT-PCR carried out in triplicate, using Power SYBR Green PCR Master Mix (Applied Biosystems). Primer sequences and PCR conditions are available on request. Results were quantified using a 2−ΔΔCt method.
CpG cluster prediction
The enrichment for CpG dinucleotides were predicted using the online sliding window algorithm CpGPlot in the EMBOSS package from EMBL-EBI (32).
DNA methylation analysis
DNA was isolated from cell lines and tumor samples by a standard phenol–chloroform extraction. DNA was modified using the EZ DNA Methylation-Gold Kit (Zymo Research), and 50 ng of modified DNA was used for PCR in COBRA or bisulfite sequencing. PCR products were cloned into the pCR 4-TOPO Vector (Invitrogen), and 10 clones per promoter were sequenced (Sequetech). The promoter regions of the miRNAs studied are upstream of the pre-miRNA transcript (+1); miR-200b −3,318 to −2,978 (20 CpGs), miR-200c −1,301 to −920 (16 CpGs), and miR-205 −467 to −175 (3 CpGs). Primer sequences and PCR conditions are available on request.
Decitabine and trichostatin A treatment
Cells were treated in duplicate as follows: vehicle (0.6 μL ethanol in 10 mL medium), trichostatin A (TSA; Sigma), 300 nmol/L for 18 hours, or decitabine (DAC; Sigma), 500 nmol/L for 96 hours, with fresh medium containing the drug changed every 24 hours.
Migration and invasion assays
Cells were serum starved for 48 hours, and cell migration and invasion were measured using the CytoSelect 24-Well Cell Migration Assay (8 μm; Colorimetric) and Invasion Assay (Basement Membrane; Colorimetric) kits (Cell Biolabs). Both assays were done according to the protocol, and serum containing growth media of the respective cell lines were used as a chemoattractant for 24 hours.
Cells were seeded at 70% confluency. After 24 hours, cells were transfected using Lipofectamine 2000 (Invitrogen) with miRNA or control mimics (Applied Biosystems). Plasmids for stable expression of miR-200c and miR-205 were purchased from Cell Biolabs and cotransfected with linear hygromycin selection marker (Clontech). The pSilencer2.1 (Ambion) siRNA expression vector with hygromycin resistance was used for stable gene knockdown.
All data are presented as mean ± SEM. Correlation of the quantitative data was determined using the Pearson correlation coefficient among the different normalized expression values. Statistical significance was P < 0.05.
Tobacco carcinogens induce EMT
Our previous study showed that treatment of HBEC1 and HBEC2 for 12 weeks (once a week for 1 hour) with genotoxic but not cytotoxic doses of MNU or BPDE, or both, induced transformation assessed by growth in soft agar (25). Moreover, the cells recovered from soft agar (transformed cells) had acquired a fibroblast-like appearance (Fig. 1A and Supplementary Fig. S1). HBEC2 cells treated with this combination of tobacco carcinogens for 4, 8, or 12 weeks were compared with the transformed cell line. After 4 weeks of exposure to MNU/BPDE, HBEC2 had acquired a fibroblast-like mesenchymal appearance consistent with EMT. The mesenchymal-like morphology remained throughout the remaining 8 weeks of treatment and persisted after removal of the carcinogens at all time points and in transformed cells (Fig. 1A). Transformation efficiency was greatest in HBEC2 treated with MNU/BPDE; therefore, this cell line was selected for detailed temporal characterization of the mechanisms underlying the changes in their growth pattern. HBEC1 MNU/BPDE-transformed cells were used to confirm studies related to EMT.
In vitro motility and invasion assays were used to determine whether the induction of EMT affected the motility and invasive potential of the transformed cells. Although EMT was evident after 4 weeks of carcinogen exposure, only the transformed cells were migratory and invasive compared with untreated HBEC2 (Fig. 1B). The molecular hallmark of EMT is decreased expression of E-cadherin and KRT19, and expression of these genes was reduced 99- and 8-fold after 4 weeks and 659- and 474-fold in HBEC2-transformed cells (Fig. 2, Supplementary Fig. S2, and Table 1). An increase in expression was seen for the mesenchymal markers N-cadherin (150- to 237-fold), vimentin (6- to 7-fold), and fibronectin (1- to 21-fold) over 4 to 12 weeks (Fig. 2, Supplementary Fig. S2, and Table 1). Expression of the EMT-inducing transcription factors ZEB1, ZEB2, Snail1, and Twist1 were significantly increased 2- to 1,418-fold at week 4 and 5- to 2,219-fold in transformed HBEC2 cells (Fig. 2, Supplementary Fig. S2, and Table 1). HBEC1-transformed cells exhibited reduced expression of epithelial markers E-cadherin and KRT19 (2- to 5-fold), increased expression of mesenchymal markers N-cadherin, vimentin, and fibronectin (2- to 19-fold), and increased expression of EMT-inducing transcription factors ZEB1, ZEB2, Snail1, and Twist1 (3- to 109-fold; Table 1). Expression of the miRNAs regulating EMT (miR-200b, miR-200c, and miR-205) was reduced 4- to 12-fold at 4 weeks and 10- to 55-fold in the HBEC2-transformed cell line (Fig. 3A). A 2-fold reduction in expression of these miRNAs was also seen in HBEC1-transformed cells (Table 1). Transient overexpression of miR-200b, miR-200c, or miR-205 in HBEC2-transformed cells did not induce a morphologic mesenchymal-to-epithelial transition (MET). However, reexpression of these miRNAs with mature mimics in HBEC2-transformed cells prevented growth in soft agar (Fig. 4A). Expression of E-cadherin increased 85- and 68-fold with transient expression of miR-200b and miR-200c, respectively, while a modest increase in expression (1.2-fold) was seen following transfection of miR-205 (Fig. 4B). These findings were replicated in HBEC2-transformed cells with stable integration and expression of miR-200c and miR-205 (not shown; not tested for miR-200b because the construct was not available). Although knockdown of ZEB1, ZEB2, Snail1, and Twist1 or combinations (ZEB1/ZEB2, ZEB1/Snail1, ZEB1/Twist1, ZEB2/Snail1, ZEB2/Twist1, and Snail1/Twist1) in HBEC2-transformed cells did not induce morphologic MET, transformation was reversed as reflected by a dramatic reduction in the number and size of the colonies in soft agar (Supplementary Table S1).
Transformed HBECs acquire stem cell–like features
The induction of EMT has been associated with the acquisition of stem cell–like features that include nonadherent growth and change in expression of cell-surface glycoproteins (33, 34). Therefore, we analyzed the ability of the transformed versus HBEC2 treated for 4, 8, or 12 weeks with carcinogens to grow in a laminin-rich growth factor–reduced Matrigel assay indicative of a nonadherent growth property of cancer stem-like/tumor progenitor cells. Although EMT was present after 4 weeks of carcinogen treatment, only the 12-week treated cells grew from single cells into small aggregate colonies and formed spherical structures 50 μm in diameter by the 10th day. Moreover, the transformed cells formed large spheroids 100 μm or greater with branching tubules, an in vitro characteristic of invasive cells (Fig. 5A). Reexpression of miR-200b or miR-200c with mature mimics in the transformed cells prevented growth in Matrigel (Fig. 5B). In contrast, transformed cells with transient expression of miR-205 grew as small aggregate colonies forming spheroids (50 μm in diameter) but did not form branching tubules. A similar growth pattern in the transformed cells with stable expression of miR-200c and miR-205 was observed (not shown). The stem cell–like phenotype of HBEC2 and HBEC1 was further characterized by assessing the expression of recently reported stem cell surface markers CD44, CD24, CD133, and ALDH1 (29, 33, 35). The CD44high/CD24low cell population was absent in HBEC2 and HBEC1. In marked contrast, after only 4 weeks of carcinogen exposure, the CD44high/CD24low cell population had increased by approximately 14%, with a striking increase in approximately 73% of the HBEC2-transformed cells displaying this stem cell marker (Fig. 5C and D and Table 1). Interestingly, the transformed cells with stable reexpression of miR-200c and miR-205 retained the stem cell CD44high/CD24low cell population (65% and 82%, respectively). The CD133 marker was expressed in approximately 3% and less than 1%, whereas ALDH1 was detected in less than 1% and 9% of HBEC1- and HBEC2-transformed cells, respectively. CD44high/CD24low, CD133, and ALDH1 were also present in 1% to 82% of lung cancer cell lines; however, stem cell surface markers were not associated with expression changes of the EMT markers (Fig. 5C and D and Table 1).
Epigenetic silencing of miR-205 and miR-200 family
The regulatory regions of miR-200b and miR-200c contain CpG-rich sequences. The recently characterized promoter of miR-200b has a CpG island (length 2,541, % GC 65.0, O/E ratio 0.69, 183 CpGs). The miR-200c promoter resides within a CpG-enriched region (length 1,796, % GC 50, O/E ratio 0.60, 66 CpGs). In contrast, the regulatory region of miR-205 is nearly devoid of CpGs. HBEC2-transformed cells were treated with the demethylating agent DAC or histone deacetylase inhibitor TSA to assess the impact of cytosine DNA methylation and chromatin remodeling on the expression of these miRNAs. DAC induced robust reexpression (4- to 7-fold) of each miRNA, whereas expression increased 2- to 3-fold with TSA (Fig. 3A). Bisulfite sequencing revealed a small increase in CpG methylation in these promoters at week 4 (3%–16%) that increased to 27% to 42% in the transformed cell line (Fig. 3B). The steady increase in hypermethylation was associated with a more profound reduction in expression of the miRNAs. Following treatment with DAC, the transformed cells showed near-complete demethylation (∼10% methylation remaining) of CpGs within these miRNA promoters that was associated with up to a 7-fold increase in expression. The 4- to 12-fold reduction in expression of these miRNAs in conjunction with 16% or less CpG promoter methylation after 4 weeks of carcinogen exposure suggests that histone modifications may contribute to their transcriptional silencing. Chromatin immunoprecipitation coupled to qRT-PCR was used to investigate changes in active and repressive histone marks. The H3K4me2 (indicative of active/open chromatin) mark was absent at the promoters of these miRNAs during carcinogen exposure relative to HBEC2 (Fig. 6). In striking contrast, H3K27me3 (indicative of inactive/closed chromatin) was enriched 14- to 30-fold at week 4 but declined dramatically to 0.7- to 7-fold by week 8 (Fig. 6). The occupancy of H3K9me2 (inactive/closed chromatin mark) was 0.3- to 1-fold at week 4 and increased up to 8-fold in transformed cells. Enrichment of DNMT1 at these miRNA promoters was observed at week 4, with the highest level of occupancy observed at week 12 (Fig. 6).
Markers of EMT in lung cancer cells
Established lung cancer cell lines Calu6 and A549 display morphologic features of EMT reflected by reduced expression of E-cadherin and KRT19 that is associated with methylation of miR-200b, miR-200c, and miR-205 (Table 1 and Supplementary Fig. S1). The lung cancer cell lines H1435 and HCC827 have an epithelial morphology associated with normal levels of E-cadherin, KRT19, miR-200b, miR-200c, and miR-205 expression (Table 1 and Supplementary Fig. S1). COBRA was used to evaluate the methylation status of these miRNAs in HBEC1-transformed cells, lung cancer cell lines, and bronchial epithelial cells. Reduced expression of miR-200b, miR-200c, and miR-205 was strongly correlated with hypermethylation of these promoters in HBEC1-transformed cells and was associated with an EMT phenotype (Table 1). Cancer-specific methylation within the promoter of miR-200b, miR-200c, and miR-205 was found in 69%, 54%, and 92% of lung cancer cell lines, respectively (Table 1 and Supplementary Table S2), whereas no methylation was detected in bronchial epithelial cells (Supplementary Table S3). Bisulfite sequencing detected dense methylation in Calu6 cells (64%–98%) that correlated with reduced miRNA expression and modest enrichment of H3K27me3, H3K9me2, and DNMT1 (Figs. 3 and 6). In contrast, miR-200b and miR-200c were sparsely methylated (∼4%) and highly expressed in H1435 cells (Fig. 3). Moderate methylation (21% of CpGs) and lower expression of miR-205 were seen in H1435, along with enrichment of H3K27me3, H3K9me2, and DNMT1 (Figs. 3 and 6). The expression of E-cadherin and ZEB1 reflected the methylation and expression status of these miRNAs in lung cancer cell lines (Table 1). A recent study (36) reported cell-type–specific methylation of miR-200c in fibroblasts, suggesting differential regulation in this cell type. Primary tumors are heterogeneous and contain stromal cells consisting of fibroblasts, endothelial, and inflammatory cells. Assessing the prevalence of DNA methylation in these miRNAs from primary blood lymphocytes (WBCs) revealed extensive and frequent methylation of these promoters (Supplementary Table S3), an effect that may support the nonadherent properties of this circulating cell population. The high prevalence for methylation in stromal cells confounds the evaluation of promoter methylation of these miRNAs in primary lung tumors. As an alternative approach, qRT-PCR was used to investigate miRNA expression patterns in 24 primary lung carcinomas relative to distant normal lung tissue (Supplementary Table S4). Reduced expression of at least one miRNA was observed in 9 of 24 (38%) primary lung tumors. The expression of miR-200b and miR-200c was significantly correlated (r = 0.595, P = 0.002). The EMT markers ZEB1 (r = −0.365, P = 0.07) and KRT19 (r = 0.396, P = 0.05) were associated with reduced expression of miR-200b, suggesting that EMT is a feature of some nonmetastatic lung tumors.
These studies show that the exposure of immortalized HBECs to tobacco carcinogens rapidly induces a multifaceted dedifferentiation program characterized by EMT and stem-like properties. The induction of EMT is epigenetically driven, initially by chromatin remodeling with ensuing promoter DNA methylation sustaining stable silencing of the miRNAs implicated in this developmental program. Finally and of great importance is the finding that the regulation of EMT morphology, induction of stem cell–like phenotype, and transformation are distinct events in response to carcinogen exposure.
EMT occurs during embryogenesis to support tissue remodeling and has been proposed as a key step in the metastasis of epithelial tumors (9, 37). Experimental models to characterize the genotypic and phenotypic changes of EMT during tumor invasion have used ectopic expression of transcription factors such as Twist or Snail or, alternatively, stimulation of these transcription factors by the growth factors TGF-β and TNF-α, which are present in the tumor microenvironment (14, 22). Abrogating the increased expression of these transcription factors induced MET and reversed phenotypic (growth in soft agar) and gene expression changes (e.g., E-cadherin and miR-200 family) linked to this developmental program. Our studies offer new insight into the role and regulation of EMT in lung carcinogenesis. We show that EMT can be induced by DNA damage generated from carcinogens present within tobacco smoke. An attribute of this premalignancy model is the ability to define the timing of events occurring during carcinogen exposure. Although EMT was induced after 4 weeks, these cells failed to exhibit enhanced proliferative and invasive phenotypes (growth in soft agar and Matrigel). Rather, after 12 weeks of carcinogen exposure, HBEC2 became highly enriched for the CD44high/CD24low stem cell marker and transformed cells exhibited nonadherent growth associated with invasive properties of branching morphogenesis and increased migration and invasion. Interestingly, the acquisition of this stem cell–like feature is not required for the enhanced proliferative and invasive properties displayed by the transformed HBECs. This conclusion is based on the fact that stable overexpression of miR-200c or miR-205 reverses branching morphogenesis and nonadherent growth of the transformed cells without reducing the population of cells enriched for the CD44high/CD24low cell marker or changing the morphology (EMT to MET). These results may reflect the multitude of signaling networks regulating EMT (e.g., Wnt and receptor tyrosine kinase signaling) that are likely impacted by carcinogen exposure (24, 38). Thus, the deregulation of pathways controlling cell morphology, miRNAs, transcriptional regulators, and phenotype associated with EMT are not always coordinately regulated. This conclusion is further substantiated by the extension of our findings in this model of premalignancy to malignant lung tumor–derived cell lines and primary tumors. The plasticity of EMT is clearly evident on the basis of the fact that some cancer cell lines display many of the hallmarks of EMT: reduced expression of E-cadherin and requisite miRNAs and increased expression of ZEB1 while exhibiting an epithelial rather than a mesenchymal morphology. Nonmetastatic primary lung tumors also displayed features of EMT as evident by the reduced expression of miRNAs in our study. Interestingly, Yanagawa and colleagues (39) showed that increased expression of Snail in adenocarcinoma was associated with poor prognosis and heterogeneous E-cadherin expression, with a greater reduction seen in the poorly differentiated component of the tumors. Together, these findings extend the role of EMT from cancer metastasis to initiation and progression of lung cancer.
The histone methyltransferase EZH2 along with EED and SUZ12 comprises the polycomb repressive complex 2 to catalyze the repressive trimethylation mark at lysine 27 of histone H3 (H3K27me3). Stem cells rely on polycomb group proteins to reversibly repress genes encoding transcription factors required for differentiation (40). Several recent studies support the hypothesis of a stem cell signature in cancer in which differentiated cells undergoing transformation acquire stem cell characteristics through a process of dedifferentiation by enrichment of the H3K27me3 and de novo cytosine DNA methylation (41). This scenario was observed during transformation of HBECs in which carcinogen exposure first led to a reduction in expression of these EMT-regulating miRNAs by recruitment of H3K27me3, followed by epigenetic switching in which this histone modification was replaced by cytosine DNA methylation within the miRNA promoters. Epigenetic switching has recently been described for genes in a prostate cancer cell line (42).
Substantial evidence linking DNA damage to the acquisition of epigenetic silencing via chromatin remodeling and cytosine DNA methylation is now emerging. One factor contributing to aberrant de novo cytosine DNA methylation during DNA damage is the rapid recruitment of DNMT1 to sites of DNA damage (43). Le Gac and colleagues (44) found that in cells treated with doxorubicin that induces double-strand breaks, DNMT1 is recruited by activated p53 and binds to functional Sp1 sites within promoters of the survivin, CDC2, and CDC25 genes. Moreover, the transcriptional repressor HDAC1 and the repressive chromatin marker H3K9me2 were also found at these promoters following DNA damage (44, 45). Cuozzo and colleagues (46) provide stronger support for a mechanistic link between DNA damage and DNA methylation. In that study, a recombinant plasmid containing a 1-Sce1 restriction site within one copy of two inactivated tandem repeated green fluorescent protein (GFP) genes was introduced into HeLa or mouse embryonic stem cells. The restriction endonuclease 1-Sce1 was added to the cells to induce a double-strand break in the GFP gene at this site. Rapid gene silencing associated with homologous recombination and DNA methylation of GFP was seen and could be blocked by treatment with DAC. The initiation of epigenetic silencing in our study of miR-200b, miR-200c, and miR-205 by trimethylation of H3K27 provides another link between chromatin remodeling and transcriptional repression in response to DNA damage. Finally, our previous study revealed that reduced DNA repair capacity is associated with increased gene promoter methylation in the aerodigestive tract of smokers (47). Collectively, these findings provide both in vitro and in vivo support for a major role of tobacco carcinogens in affecting epigenetic regulation of miRNAs and genes involved in lung carcinogenesis.
Our model of tobacco carcinogen–induced premalignancy shows temporally that initiation of EMT occurs after just 4 weeks of carcinogen exposure, followed by the acquisition of other stem cell–like traits: spheroids with branching tubules and a marked enrichment for stem cell–like markers CD44high/CD24low and/or CD133/ALDH1 in HBEC-transformed cells. However, in spite of displaying in vitro characteristics of highly invasive cells, the HBEC-transformed cell lines did not grow orthotopically in nude mice (25). The ability to confer a complete malignant and metastatic phenotype has been associated with the presence of EMT, K-ras, and p53 mutations in murine cell lines (9). Mutation of the K-ras and p53 genes, generally late events in tumor development (48, 49), was not observed in carcinogen-treated HBECs (25). However, our in vitro model establishes a link between tobacco carcinogen exposure, epigenetic induction of EMT, and the development of stem cell–like phenotype that together may represent some of the early events driving the clonal expansion of premalignant lung cells.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
This work was supported by a research grant from NIH (R01 ES008801).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
- Received August 17, 2010.
- Revision received February 2, 2011.
- Accepted February 11, 2011.
- ©2011 American Association for Cancer Research.