The response of breast epithelial cells and breast cancer cell to cigarette smoke has been previously examined [
15,
16,
30,
31], but these studies focused on short-term treatment (up to one week) while we have analyzed the effect of continuous long-term exposure. We demonstrated that chronic exposure to tobacco smoke in the form of CSE or CSC can alter the phenotype of mammary epithelial cells, promoting the acquisition of mesenchymal traits such as increased anchorage-independent growth, motility, invasion, and the expression of markers associated with self-renewal and tumor initiation. Numerous groups have demonstrated the emergence of a CD44
+/CD24
-/low stem-like signature from CD44
+/low/CD24
+ cells upon the induction of an EMT phenotype characterized by loss of E-cadherin and gain of vimentin [
11,
32]. The CD44
+/CD24
-/low phenotype has been consistently associated with self-renewing mammary epithelial cells, which are also more tumorigenic and basal-like than CD44
+/CD24
+ cells [
33]. Similarly, we showed that treatment of MCF 10A cells with CSE leads to the emergence of a CD44
hi/CD24
low population, and our
in vivo experiments demonstrated that CSE-treated MCF 10A cells have increased survival and colonization ability. Although MCF 10A cells did not become malignant, treatment of the MCF7 cancer cell line led to increased metastatic potential, consistent with published evidence that the differentiation state of the cell of origin is a strong determinant of the cellular phenotype of the final transformed state [
34]. Other studies in animal models have previously shown that tobacco smoke can increase the risk of metastasis from breast cancer, but this has been attributed mainly to smoking-induced inhibition of host antitumor immune defenses, or to damage of the host tissue [
35,
36]. In contrast, our data from
ex vivo exposure followed by orthotopic or subcutaneous transplantation into mice indicate that tobacco smoke can directly affect the ability of breast epithelial cells to invade or metastasize, independent of other cigarette smoke effects on the host and stromal environment.
The phenotypical alterations induced by cigarette smoke were accompanied by multiple gene expression changes. We concentrated our analysis on genes associated with EMT, loss of tumor suppression and the acquisition of malignancy traits. Our data indicates that ERβ is epigenetically repressed by tobacco smoke, which is consistent with a recent study showing that methylation of ERβ is a frequent event in breast cancer [
37]. Contrary to the better known and structurally similar ERα, ERβ does not induce mitogenic response and can reduce basal, hormone-independent cell proliferation [
28]. ERβ is widely expressed in normal mammary epithelium, but frequently lost in breast cancer, where its presence generally correlates with better prognosis [
28,
38]. Knock down of ERβ in MCF 10A or MCF7 cells was shown to cause a significant growth increase of both cell types in a ligand-independent manner [
38], while expression of exogenous ERβ in the receptor negative breast cancer cell line MDA-MB-231 inhibited proliferation [
39]. Cigarette smoke also caused downregulation of claudin 1, 3, 4, 7, and 8. The claudins are integral components of tight junctions, and their expression in cancer appears to be tissue specific, with some claudins downregulated in certain tumors and upregulated in others [
40]. A small subgroup of breast cancer has been identified as expressing low levels of claudins, and is referred to as the “claudin low” group [
41‐
43]. Claudin low tumors represent 12-13% of breast cancers, are generally basal like, and overexpress EMT markers [
41‐
43]. Mouse claudin-low tumors generated in a p53-null animal model were found to be markedly enriched in tumor-initiating cells [
44]. Consistently, our claudin-low CSE-treated breast cells are more tumorigenic than untreated cells, and exhibit gene expression changes indicative of EMT, such as downregulation of E-cadherin and occludin, and upregulation of N-cadherin, fibronectin and vimentin. Downregulation of occludin can reduce cancer sensitivity to apoptogenic factors by modulating apoptosis-associated genes. In addition, occludin decreases cellular invasiveness and motility, thus its downregulation can potentially favor cancer metastasis [
45]. The downregulation of occludin and claudin 1 [
46] may also be the result of epigenetic regulation, since we have observed increased methylation at the promoter of these genes and in the case of claudin 1, the gene can be re-expressed with demethylating agents such as 5-azacytidine and decitabine [
46]. CSE treatment upregulated
TGFBR3 and
TGFB2 in MCF 10A cells, which is consistent with the reported observation that endothelial cells undergoing EMT express
TGFBR3, and TGFBR3-specific antisera can inhibit mesenchyme formation and migration [
24]. Moreover, ectopic overexpression of
TGFBR3 in non-transforming ventricular endothelial cells conferred transformation in response to TGFB2 [
24]. Since we observed upregulation of
TGFBR3 and
TGFB2 in MCF 10A cells that are undergoing EMT-like changes, but are not completely transformed by cigarette smoke, our results suggest that overexpression of these two genes by cigarette smoke may be a component of EMT that is not associated with transformation. Alternatively, this could be a very early event in transformation and cancer development. We also observed that the EMT-promoting transcription factors
TWIST1,
TWIST2,
ZEB1,
ZEB2 and
FOXC2 were upregulated, while
FOXC1 and
SNAI1 (Snail) were downregulated by CSE. Except for the decreased
SNAI1, these data are consistent with recent reports that the MDA-MB-231 and MDA-MB-435 basal B cell lines express higher levels of fibronectin, N-cadherin,
SNAI1 and
ZEB2, and lower E-cadherin and
FOXC1 than the luminal epithelial cell line, MCF7 [
25]. The same study showed that overexpression of
TWIST1, as well as the EMT-promoting factor TGF-β1, consistently upregulates
ZEB1 and
ZEB2 and
FOXC2 in human mammary epithelial cells. Interestingly, TGF-β1 is up-regulated by TWIST1, but is not required for TWIST1-induced up-regulation of
FOXC2, which occurs in mammary epithelial cells overexpressing
TWIST1 even in the presence of a TGF-β signaling inhibitor [
25]. Taken together our observations in the MCF 10A breast epithelial cell line exposed to CSE are consistent with a model of EMT where TWIST drives the transition and upregulates
FOXC2,
ZEB1 and
ZEB2, with potential involvement of TGFβ signaling.