Abstract 2982: Epithelial-to-mesenchymal transition in vitro: clinical implications for breast cancer biology

Abstract

Abstract Background: Epithelial-to-mesenchymal transition (EMT) is a crucial step in not only embryonic development but also cancer progression. From the perspective of cancer biology, EMT has been assumed to relate to the aggressiveness of cancer, which is defined by metastatic potential, and the properties of cancer stem cells. There are very few models that recapitulate EMT in vivo or in vitro. Takahashi et al. previously reported that ARPE-19 retinal pigmental epithelial cells transformed to mesenchymal cells after extrinsic stimulation with TNF-α and TGF-β in vitro. In this model, ARPE-19 cells developed cellular foci with fibronectin deposits and lost E-cadherin expression. Thus, this in vitro model was a very good representation of EMT. In the current study, using this model, we clarified the biological processes and clinical implications of EMT in breast cancer. Our objectives were to 1) understand the molecular dynamics of EMT using time-course whole-genome transcriptional profiles of this EMT model and 2) define the clinical implications on the basis of these transcriptional profiles. Methods: mRNAs were extracted from ARPE-19 cells treated with TNF-α and TGF-β and subjected to Affymetrix U133 plus 2.0 arrays (GSE12548). We compared the gene expressions at four time points: 0, 6, 24, and 60 hr after treatment. For analyzing microarray data, the BRB-array tool was used. All statistical tests were two-sided. Results: Samples were paired by time point as follows: 0 vs 6, Phase I; 6 vs 24, Phase II; and 24 vs 60, Phase III. In each comparison, the numbers of significantly differentially expressed probes (false discovery rate<0.05) were 1652 (Phase I), 805 (Phase II), and 110 (Phase III), suggesting that gene expression changes were most remarkable early in EMT. Among the three phases (I, II, or III), no probes overlapped each other, indicating that different genes were associated with different dynamic phases of EMT. Next, over- and underexpressed biological pathways associated with EMT dynamics in each phase were analyzed using Ingenuity pathway analysis. The overexpressed pathway in phase I was correlated with the “hepatic fibrosis activation bioprocess”. This pathway was also identified in mouse breast cancer stem cells of our model (data on submission). Also, in this pathway, IL-8 and TGF-β receptor 2 were upregulated, and both of them are overexpressed in triple-negative breast cancer, one of the most aggressive subtypes of breast cancer. Interestingly, tight junction signaling was upregulated at Phase III, and fibronectin and twist1 were still overexpressed at this very late phase of EMT. These findings imply that tight junction proteins could be expressed very late in the EMT process. Conclusion: Different biological processes are associated with different dynamic phases of EMT; this finding implies that analyzing EMT dynamics can be a promising approach to identifying molecular targets for breast cancer. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research; 2012 Mar 31-Apr 4; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2012;72(8 Suppl):Abstract nr 2982. doi:1538-7445.AM2012-2982