Cyclooxygenase (COX)-2-derived prostanoids can influence several processes that are linked to carcinogenesis. We aimed to address the hypothesis that platelets contribute to aberrant COX-2 expression in HT29 colon carcinoma cells and to reveal the role of platelet-induced COX-2 on the expression of proteins involved in malignancy and marker genes of epithelialmesenchymal transition (EMT). Human platelets cocultured with HT29 cells rapidly adhered to cancer cells and induced COX-2 mRNA expression, but not protein synthesis, which required the late release of platelet-derived growth factor and COX-2 mRNA stabilization. Platelet-induced COX-2-dependent prostaglandin E 2 (PGE 2 ) synthesis in HT29 cells was involved in the downregulation of p21 WAF1/CIP1 and the upregulation of cyclinB1 since these effects were prevented by rofecoxib (a selective COX-2 inhibitor) and rescued by exogenous PGE 2 . Galectin-3, which is highly expressed in HT29 cells, is unique among galectins because it contains a collagen-like domain. Thus, we studied the role of galectin-3 and platelet collagen receptors in plateletinduced COX-2 overexpression. Inhibitors of galectin-3 function (b-lactose, a dominant-negative form of galectin-3, Gal-3C, and anti-galectin-3 antibody M3/38) or collagen receptor-mediated platelet adhesion (revacept, a dimeric platelet collagen receptor GPVI-Fc) prevented aberrant COX-2 expression. Inhibition of platelet-cancer cell interaction by revacept was more effective than rofecoxib in preventing platelet-induced mRNA changes of EMT markers, suggesting that direct cell-cell contact and aberrant COX-2 expression synergistically induced gene expression modifications associated with EMT. In conclusion, our findings provide the rationale for testing blockers of collagen binding sites, such as revacept, and galectin-3 inhibitors in the prevention of colon cancer metastasis in animal models, followed by studies in patients.