Epigenetic Alterations Induced by Tobacco Smoke

点击翻译按钮获取中文摘要

A concerted effort has been underway to establish an in-vitro system to examine sequential epigenetic effects of cigarette smoke in respiratory epithelia. Briefly, normal human small airway epithelial cells (SAEC) and cdk4/hTERT-immortalized human bronchial epithelial cells (HBEC) have been cultured in normal media (NM) with or without cigarette smoke condensate (CSC) for up to 24 months under potentially relevant exposure conditions. Western blot analysis demonstrated that CSC mediated dose- and time-dependent diminution of H4K16Ac and H4K20Me3, while increasing relative levels of H3K27Me3; these histone alterations coincided with decreased DNMT1 and increased DNMT3b expression. Pyrosequencing and quantitative RT-PCR experiments revealed time-dependent hypomethylation of D4Z4, NBL2, and LINE-1 repetitive DNA sequences; up-regulation of H19, IGF2, MAGE-A1, and MAGE-A3 as well as activation of Wnt signaling; and, hypermethylation of tumor suppressor genes such as RASSF1A, RUNX3, and RAR-beta which are frequently silenced in human lung cancers. Array-based DNA methylation profiling identified additional novel DNA methylation targets in soft agar clones derived from CSC-exposed HBEC; a CSC gene expression signature was also identified in these cells. Progressive genomic hypomethylation and locoregional DNA hypermethylation induced by CSC coincided with a dramatic increase in soft agar clonogenicity but not tumorigenicity of HBEC. A manuscript pertaining to results of the initial analysis of the model (9 months of CSC exposure) were published recently in Oncogene. Comprehensive analyses of global DNA methylation and gene expression are underway using samples pertaining to baseline, as well as 6, 12, 18 and 24 months of continuous exposure to normal media or CSC. We expect to identify novel targets that have been silenced or activated by epigenetic mechanisms coinciding with duration of CSC exposure. Additional experiments utilizing this unique model system have been performed to ascertain if random stochastic events rather than instructive mechanisms establish epigenetic signatures in lung cancer cells. Briefly, Illumina methylation array analysis revealed 48 gene loci to be hypomethylated whereas 56 were hypermethylated in HBEC exposed to CSC for nine months relative to control HBEC. Hypermethylated- but not hypomethylated genes in CSC-treated HBEC were highly overlapped with stem cell Polycomb target genes (PCTGs). Further analysis revealed that stem cell PCTGs were seven-fold more likely to be hypermethylated than non-PCTGs in HBEC following CSC exposure. Pyrosequencing analysis of multiple genes silenced by CSC treatment revealed progressive de novo DNA hypermethylation. ChIP analysis revealed that genes, which were hypermethylated and completely silenced by CSC treatment exhibited relatively high levels of H3K27Me3 prior to CSC exposure. In contrast, genes that were resistant to de-novo DNA methylation had very low or undetectable H3K27Me3 and high level H3K4Me3 prior to treatment. Interestingly, progressive increases in site-specific DNA methylation coincided with gradual reductions of H3K27Me3 of several epigenetically-silenced genes. These novel findings suggest that de-novo DNA methylation and gene silencing are not the result of random events during malignant transformation. Our data provide the first direct evidence for an instructive mechanism of de-novo DNA methylation during pulmonary carcinogenesis, and highlight the potential for our model to yield new insights regarding epigenetic mechanisms contributing to the DNA methylation paradox in lung cancer cells. Because it recapitulates critical epigenetic alterations observed in established lung cancers, this unique model may prove quite useful for evaluating novel chromatin remodeling agents for lung cancer prevention using surrogate epigenetic markers of malignant transformation as endpoints. Several studies suggest that smoking status at diagnosis or during treatment effects response to therapy and overall survival of lung cancer patients. As such, additional experiments have been performed to examine the effects of CSC on cultured lung cancer cells. Briefly, several lung cancer lines were treated with CSC using a variety of concentrations and exposure durations. Under conditions mimicking 1 pack per day (ppd) exposures for five days, CSC had no effects on in-vitro proliferation rates, yet dramatically enhanced tumorigenicity of A549 and Calu-6 cells in nude mice. Affymetrix arrays and quantitative RT-PCR experiments revealed that CSC markedly inhibited expression of Dkk-1, a secreted Wnt signaling antagonist and putative tumor suppressor. Subsequent ChIP, pyrosequencing, and methylation specific PCR experiments demonstrated that inhibition of Dkk-1 expression by CSC coincided with a dose-dependent increase in H3K27 trimethylation, and recruitment of polycomb repressor complexes without an appreciable increase in DNA methylation within the Dkk-1 promoter despite continuous CSC exposures ranging from 5-60 days. Cessation of CSC exposure resulted in diminution of Dkk-1 promoter-associated polycomb proteins, and restoration of Dkk-1 expression in these cancer cells. Western blot and focused qRT-PCR array experiments indicated that CSC mediated dose-dependent increases in Wnt signaling in A549 and Calu-6 cells. Similar findings were noted following siRNA mediated knock-down of Dkk-1 in these cells. CSC exposure also inhibited expression of Dkk-1 in normal respiratory epithelia. Interestingly, CSC exposure or knock-down of Dkk-1 in lung cancer cells dramatically up-regulated expression of Wnt5a, a non-canonical Wnt ligand implicated in cancer stem cell signaling. Knock-down of Dkk-1 recapitulated the pro-tumorigenic effects of CSC in lung cancer cells. Results of these studies were published in Cancer Research. Additional experiments have been conducted to specifically examine the effects of Wnt5a in lung cancer cells. Briefly, Calu-6 and H841 cells, which exhibit low endogenous levels of Wnt5a, and A549 cells, which express relatively high endogenous levels of this ligand, were transduced with lentiviral vectors expressing Wnt5a, shRNA targeting Wnt5a, or control sequences. Calu-6, and H841- but not A549 cells constitutively expressing Wnt5a exhibited significantly increased proliferation, migration, and invasion relative to vector controls. This phenomenon, which was also observed following exposure of parental Calu-6 cells to Wnt5a, was partially abrogated by Wnt5a-derived hexapeptide, Box-5. Knock-down of Wnt5a diminished proliferation, migration and invasion of A549 cells. Relative to vector controls, Calu-6 cells constitutively expressing Wnt5a exhibited significantly increased tumorigenicity in nude mice. Western blot, promoter-reporter, and focused quantitative RT-PCR array experiments demonstrated that Wnt5a increased expression of the nuclear orphan receptor, Ror2, and activated non-canonical Wnt signaling. Microarray array analysis revealed surprisingly little over-lap regarding either gene or micro-RNA expression between lung cancer lines following constitutive expression of Wnt5a, suggesting that the effects of Wnt5a in lung cancer cells are highly pleiotropic. Collectively, these data demonstrate that Wnt5a enhances the malignant phenotype of lung cancer cells, and support the development of therapies targeting Wnt5a-mediated signaling for lung cancer therapy. A manuscript pertaining to these experiments has been submitted for publication.