During cancer progression, cells encounter many stress signals and all along they have built-in mechanisms to eliminate themselves. The successful cancer cells managed to foil this hardwired stress response. Emerging evidence indicates that some of the genes that normally function to eliminate the cells are co-opted to become oncogenes. How the cellular (normal vs. cancerous) context determines that some genes undergo this “Jekyll-and-Hyde” conversion is an intriguing but largely unresolved issue in cancer biology.
In my work over the past 5 years, I found ATF3, an ATF/CREB family transcription factor encoded by an adaptive-response gene, is a new regulatory molecule with a dichotomous role. It enhances apoptosis in untransformed cells, but protects the cells from stress-induced death and promotes cell motility in malignant cancer cells. In an in vivo xenograft mouse model, in addition to promoting primary tumor growth, ATF3 also increases lung metastasis.
To explore the potential mechanisms by which ATF3 promotes metastasis, I set out to address three major questions. First, as a transcription factor, will ATF3 regulate the expression of some target genes involved in cell motility? Second, as an adaptive response gene, will ATF3 be induced in cancer cells in response to stromal signals and whether the induction of ATF3 mediates cancer cells’ response to stromal signals? Third, will the induction of ATF3 in cancer cells feed back on stromal cells to affect stroma-cancer interaction?
By examining potential ATF3 target genes, I found ATF3 regulates a set of genes involved in cell motility, some of which were proved to be direct targets of ATF3. As an adaptive-response gene, ATF3 was induced by multiple stromal signals, such as TGFβ, TNFα and IL-1β. When focusing on one multifunctional cytokine, TGFβ, I found that ATF3 can mediate TGFβ effects on target gene expression and cell motility regulation. The interaction between ATF3 and Smad2/3 offers a partial mechanistic understanding. Besides mediating the intracellular signaling of stromal factors, ATF3 is also involved in stroma-cancer interaction, as demonstrated by the effect of ATF3 expression in cancer cells on macrophage recruitment and angiogenesis.
Finally, by examining ATF3 status in human breast cancer, I found that wild-type ATF3 gene is frequently amplified and its expression elevated in human breast tumors, suggesting a pathophysiological relevance of ATF3 to human cancer. Through the database analysis of microarray data generated by two other groups, I found that the higher expression of ATF3 correlates with worse outcome of breast cancer patients. This, in combination with the functional consequences of ATF3 in both cell system and mouse model, suggests ATF3 is a new oncogene in malignant breast cancer cells that is likely to have pathophysiological relevance to human cancer.