UFM1 is a member of the ubiquitin like protein family. While the enzymatic cascade of UFM1 conjugation has been elucidated in recent years, the biological function remains largely unknown. In this report we demonstrate that the recently identified C20orf116, which we name UFM1-binding protein 1 containing a PCI domain (UFBP1), and CDK5RAP3 interact with UFM1. Components of the UFM1 conjugation pathway (UFM1, UFBP1, UFL1 and CDK5RAP3) are highly expressed in pancreatic islets of Langerhans and some other secretory tissues. Co-localization of UFM1 with UFBP1 in the endoplasmic reticulum (ER) depends on UFBP1. We demonstrate that ER stress, which is common in secretory cells, induces expression of Ufm1, Ufbp1 and Ufl1 in the beta-cell line INS-1E. siRNA-mediated Ufm1 or Ufbp1 knockdown enhances apoptosis upon ER stress. Silencing the E3 enzyme UFL1, results in similar outcomes, suggesting that UFM1-UFBP1 conjugation is required to prevent ER stress-induced apoptosis. Together, our data suggest that UFM1-UFBP1 participate in preventing ER stress-induced apoptosis in protein secretory cells.

Using tumor cell-restricted overexpression of glutathione peroxidase 4 (GP x 4), we investigated the contribution of tumor cell eicosanoids to solid tumor growth and malignant progression in two tumor models differing in tumorigenic potential. By lowering cellular lipid hydroperoxide levels, GP x 4 inhibits cyclooxygenase (COX) and lipoxygenase (LOX) activities. GP x 4 overexpression drastically impeded solid tumor growth of weakly tumorigenic L929 fibrosarcoma cells, whereas B16BL6 melanoma solid tumor growth was unaffected. Yet, GP x 4 overexpression did markedly increase the sensitivity of B16BL6 tumors to angio-destructive TNF-alpha therapy and abolished the metastatic lung colonizing capacity of B16BL6 cells. Furthermore, the GP x 4-mediated suppression of tumor cell prostaglandin E(2) (PGE(2)) production impeded the induction of COX-2 expression by the tumor stress conditions hypoxia and inflammation. Thus, our results reflect a PGE(2)-driven positive feedback loop for COX-2 expression in tumor cells. This was further supported by the restoration of COX-2 induction capacity of GP x 4-overexpressing L929 tumor cells when cultured in the presence of exogenous PGE(2). Thus, although COX-2 expression and eicosanoid production may be enabled by PGE(2) from the tumor microenvironment, our results demonstrate the predominant tumor cell origin of protumoral eicosanoids, promoting solid tumor growth of weakly tumorigenic tumors and malignant progression of strongly tumorigenic tumors.