Lung cancer metastases are common in people living with lung cancer and significantly contribute to lung cancer death . EMT is believed to play crucial roles in cancer metastasis . In the present study, we analyzed a genome-wide DNA methylation profile and studied the relationship between TGF-β-induced EMT and the whole DNA methylome in lung cancer cells. We report the global methylation signature of A549 cells and its relationship with gene expression, histone modifications and short RNA coding regions. Furthermore, our data provide evidence that TGF-β-induced EMT had almost no effect on the global DNA methylome. Notably, we found that expression of DNMTs and TETs were altered after EMT. Additionally, histone methylation modifications including H3K4me3, H3K9me2 and H3K36me3, which were demonstrated to have a role in the EMT of AML12 cells , were also changed during cancer cell EMT in our study.
TGF-β-induced rapid senescence of A549 cells has long been noticed [23–25]. Detection of cell markers including E-cadherin, N-cadherin, Snail1 and Vimentin has been used to identify typical EMT processes . TGF-β induces changes in these cell markers and cell shape by activating diverse downstream pathways [27–29]. According to our data, after TGF-β treatment, A549 cells showed characteristics of EMT.
From the methylation signature of the A549 cell line, we observed low methylation levels at the TSS region of the genes. This may be due to the lack of nucleosomes, which are required substrates of de novo methylation, at active TSSs [11, 30]. A low overall methylation level of CGIs in A549 cells was also found. CGIs are protected from methylation through two possible mechanisms: recognition of common cis-acting sequences [31, 32] and active demethylation . The relationship between DNA methylation and gene expression has long been considered. Methylation in the immediate vicinity of the TSS blocks the initiation of gene expression, and high gene expression levels may also in turn block de novo methylation of the promoter . The genes with the lowest average methylation levels around the TSS included FANCD2, FAM58A, HDAC1, and C14orf166 according to our data, and FCGBP, CAPN12, FES, and CBLC had the highest average TSS methylation levels in A549 cells. Higher methylation within the gene body of genes with hyperexpression was also found in the present study. It has been proposed that transcriptional elongation may stimulate DNA methylation and that H3K36me3, which is also associated with elongation, might be involved in the recruitment of DNMTs .
The methylation signature of the genome is also closely correlated with many histone modifications, and we observed an effect of histone acetylation and methylation on CCGG methylation. It has been found that acetylated histones are associated with open chromatin, and unmethylated DNA tends to get repackaged into an open configuration. Conversely, non-acetylated histones are often associated with compact chromatin and methylated DNA [12, 36, 37]. Different types of histone methylation also serve as marks for the genome-wide DNA methylation pattern. An inverse relationship between H3K4me and H2az histone marks and DNA methylation has long been considered [30, 38], and here, we report that H3K79me2 is anti-correlated with CCGG methylation. Additionally, H3K9me3 and H3K27me3 have been found to be repressive marks of gene transcription and positively correlated with DNA methylation by interacting with methylated DNA-binding proteins [39–41]. Furthermore, our data indicate that DNA methylation might be associated with short RNA expression.
In A549 cells undergoing EMT, we found that there was no genome-wide DNA methylation reprogramming, and we observed changes in H3K4me3, H3K9me2 and H3K26me3. The regulation of gene expression by histone modifications and DNA methylation has long been considered [42–45]. Histone modifications provide unstable transcriptional repression, and DNA methylation serves as a highly stable silencing mark that is not easily reversed . During just 4 days of A549 cell treatment with TGF-β, the status of histone modifications was easily changed. However, DNA methylation remains relatively stable and cannot be altered in a short time, while the expression of DNMTs and TETs was regulated according to our data. The altered expression levels of the genes during and after EMT that were found by Sartor et al.  may not have been induced by altered DNA methylation but by altered histone modifications.
In the present study, we quantified the methylation levels of approximately 300,000 CCGG sites, which is approximately 13% of all CCGG sites and 1.1% of all CG sites in humans; thus, the methylation status of other CG sites and non-CG sites were not included. Although the methylation levels of the CCGG sites sequenced in our study could well reflect the methylation status of all CG sites, the employment of some higher throughput DNA methylation methods, for example, reduced representation bisulfite sequencing (RRBS) and whole-genome bisulfite sequencing (WGBS), may further confirm our findings. We tested for changes in whole histone methylation levels during EMT, and ChIP-Seq technology will help generate high-resolution profiles of histone modifications and allow more detailed study of the role of histone modifications in EMT of cancer cells. Additionally, more detailed molecular biology research may provide information regarding the direct role of DNMTs and TETs in EMT. In summary, our work explored the genome-wide DNA methylation signature of A549 lung cancer cells and the influence of EMT on the epigenome. Although the exact role of DNMTs, TETs and histone modifications in TGF-β-induced EMT and whether the levels of these influence genome-wide DNA methylation over longer periods of time were not determined, our findings may contribute to the uncovering of the epigenetic mechanisms of EMT in cancer cells and benefit the development of cancer metastasis treatments.