Transcriptomic analysis of an in vitro murine model of ovarian carcinoma: functional similarity to the human disease and identification of prospective tumoral markers and targets.

Ovarian cancer is an aggressive disease of poor prognostic when detected at advanced stage. It is widely accepted that the ovarian surface epithelium plays a central role in disease etiology, but little is known about disease progression at the molecular level. To identify genes involved in ovarian tumorigenesis, we carried out a genome-wide transcriptomic analysis of six spontaneously transformed mouse ovarian surface epithelial (MOSE) cell lines, an in vitro model for human ovarian carcinoma. Loess normalization followed by statistical analysis with control of multiple testing resulted in 509 differentially expressed genes using an adjusted P-value < or = 0.05 as cut-off. The top 20 differentially expressed genes included 10 genes (Spp1, Cyp1b1, Btg1, Cfh, Mt1, Mt2, Igfbp5, Gstm1, Gstm2, and Esr1) implicated in various aspects of ovarian carcinomas, and other 3 genes (Gsto1, Lcn7, and Alcam) associated to breast cancer. Upon functional analysis, the majority of alterations affected genes involved in glutathione metabolism and MAPK signaling pathways. Interestingly, over 20% of the aberrantly expressed genes were related to extracellular components, suggestive of potential markers of disease progression. In addition, we identified the genes Pura, Cnn3, Arpc1b, Map4k4, Tgfb1i4, and Crsp2 correlated to in vivo tumorigenic parameters previously reported for these cells. Taken together, our findings support the utility of MOSE cells in studying ovarian cancer biology and as a source of novel diagnostic and therapeutic targets. Copyright 2005 Wiley-Liss, Inc.