This study was undertaken to analyse the expression and role of transcription factors zeb1, twist and snai1 in human breast carcinoma. The study shows that nuclear expression of these transcription factors in breast carcinoma epithelial tumor cell compartment is uncommon with only 3.6% and 3.1% of cases being positive for twist and snai1 and with no positivity for zeb1. The cells in the stromal compartment, however, showed abundant nuclear positivity for zeb1 and twist, though no positive cases with snai1 were found where nuclear expression of fusiform stromal cells would have exceeded the 5% expression level considered as the limit for positive expression.
The low expression of zeb1, twist and snai1 in epithelial tumor cell compartment of breast carcinoma is surprising since these transcriptional factors are believed to be important in the spread of carcinomas through induction of EMT. Similar findings have been found in pancreatic adenocarcinomas where twist expression was lower in tumor cells than in non-neoplastic epithelial cells [25
]. However, EMT is defined as a phenomenon where tumor cells lose their epithelial features, and thus a proportion of the cells in stromal tissue might represent transformed malignant cells. In fact, during EMT, epithelial cancer cells have been postulated to be transformed to mesenchymal type cells then invade the blood vessels or lymphatics after which they regain epithelial features at a metastatic site through mesenchymal-epithelial transformation [26
]. Evidence for this kind of concept comes from the detection of similar genetic changes in both tumor and stromal cells in primary tumor sites while in metastases, stromal cells do not display these kinds of genetic changes [26
]. Furthermore, epithelial cells have been shown to transdifferentiate to myofibroblastic cells in tissue fibrosis and cancer [27
]. Our results could thus merely indicate that epithelial tumor cells initiate zeb1 and twist synthesis after being transformed to myoepithelial type cells. This is in line with the proposed hierarchy of the expression of such transcriptional factors in EMT [3
] leading to that zeb1 and twist, which apparently maintain the migratory phenotype of tumor cells, are mostly detected in the stromal compartment in breast tumor tissue. On the other hand, evidently stromal fibroblast-derived cells are also activated to express zeb1 and twist which makes it difficult to differentiate these cell types from each other by morphologic means, a phenomenon which has also been described by others [28
Even though the low frequency of cases with nuclear expression in epithelial type tumor cells, twist expression in these cells has clinical importance in breast carcinoma. Nuclear expression of twist was related to a poorer outcome of the patients. It has been shown previously that downregulation of twist in aggressive breast cancer cell lines leads to an abrogation of the invasive and metastatic phenotype of the cells and furthermore twist expression is associated with high grade breast tumors [3
]. In addition to EMT twist has also other features which are linked to the growth or metastasis of tumor cells, i.e. it downregulates the expression of tissue inhibitors of metalloproteinase 1 (TIMP1) mRNA in Saos cells [30
] promotes angiogenesis by inducing vascular endothelial growth factor (VEGF) and stimulates cancer cell migration [31
]. Prevention of twist transcription in breast cancer blocked EMT, invasion and the development of multidrug resistance induced by adriamycin [33
]. In line with this, expression of twist has been shown to be prognostically important also in other types of carcinomas such as cervix [34
], hepatocellular [31
], esophageal [35
] and gastric carcinoma [36
Nuclear expression of snai1 in the epithelial compartment was seen in 3.1% of cases which is considerably lower than has been reported in some other types of tumors. In endometrioid carcinomas, nuclear snail expression was found in 29% [14
] and in ovarian tumors in 23-38% of cases [14
]. Snai1 expression has variably been linked to prognosis in hormone-sensitive ovarian carcinomas [14
]. Curiously, there are no extensive studies on clinical materials in breast cancer. Our present experiments reveal that in addition to being rare, nuclear snai1 expression in breast carcinoma does not seem to influence patient prognosis. Furthermore, it was not related to the size of the tumors or to the presence of metastases. The fact that snai1 was not observed in stromal cells might suggests that in contrast to some other tumors it does not play an important role in EMT of breast carcinoma. Clearer and stronger expression of snai1 has been detected in squamous cell carcinoma of the pharynx where stromal expression was associated with tumor size and prognosis (Jouppila-Mättö A, Tuhkanen H, Soini Y, Pukkila M, Närkiö-Mäkelä M, Sironen R, Virtanen V, Mannermaa A, Kosma V-M: Transcription factor Snail 1 expression and poor survival in pharyngeal squamous cell carcinoma, submitted). Evidently tumors of different sites and histology vary in their expression of snai1.
In our large set of breast carcinomas, no tumors were found with a positive nuclear expression of zeb1 in epithelial tumor cells. In contrast, zeb1 was found in fusiform stromal cells with a 75% proportion. In uterine tumors, ZEB 1 was expressed in stromal cells of low grade endometrial adenocarcinomas, but not in tumor cells [38
]. In contrast, aggressive types of endometrial cancers also showed zeb1 expression in tumor cells [4
]. Our results showed a lower expression of zeb1 in stromal cells in mucinous and medullary carcinomas than was the case in ductal carcinomas. These tumors have a slightly better prognosis than breast ductal or lobular carcinomas [21
]. A similar association was observed for twist in mucinous carcinomas. The results could suggest that EMT activity would be lower in these histological types of tumors. Overall, stromal expression of zeb1 did not associate with the prognosis of the patients.
Stromal zeb1 expression was significantly lower in in situ
type of breast carcinoma compared to invasive cases indicating that induction of zeb1 in stromal cells associates with the invasive phenotype. Some of these cells could represent transformed tumor cells undergoing EMT. Another proportion of the cell population could represent stromal fibroblastic cells undergoing activation of these transcription factors in response to the growth factors produced by the tumor tissue during the development of an invasive tumor. This kind of activation might stimulate stromal cells to transform to more motile alpha-smooth muscle actin-producing myofibroblasts. In fact, zeb1 serves as a transcriptional activator inducing genes such as alpha-smooth muscle actin, vimentin and collagens and heterozygous mutation of zeb1 gene leads to impaired smooth muscle actin and myosin expression [39
In our set of breast carcinomas stromal expression of zeb 1 was associated with the estrogen and progesterone receptor status of the tumors. Previous studies have revealed that estrogen and progesterone may stimulate zeb1 expression. In ovarectomised mice, zeb1 was upregulated in uterine stroma and myometrium after progesterone or estrogen treatment [38
]. Our results indicate that positive selection of the tumor cells for estrogen and progesterone receptors is reflected in the upregulation of zeb1 in stromal cells induced by the positive trophic stimuli of these hormones. Previously it has been shown that higher amounts of estradiol and estrone are present in estrogen positive tumors [40
]. It remains to be clarified whether breast epithelial tumor cells per se do not express zeb1 even in hormone positive carcinomas. However, in experiments on ten breast and ovarian carcinoma cell lines, upregulation of the zeb1 gene was seen in only one cell line, suggesting that the zeb1 mRNA response in neoplastic cells has become deranged [4
]. This is also reflected in findings on ovarian and uterine non-neoplastic and neoplastic tissues where zeb1 mRNA levels were associated with the estrogen level in non-neoplastic tissue but this association was lost in neoplastic tissues [4
Curiously, however, stromal twist expression was not associated with a poorer survival of the patients nor was there any relation to tumor size or to the presence of metastases. Similarly no association was found for zeb1. The findings may be due to the fact that twist and zeb1 positive cells in the stroma represent a mixture of cell types, one proportion representing non-neoplastic activated fusiform fibroblastic cells, the other being EMT transformed tumor cells. Since these cells cannot be distinguished properly by morphology their relationships or even their relative numbers are difficult to measure. Thus the real quantity of EMT transformed cells in the stroma cannot be assessed by zeb1 or twist immunohistochemistry.