Tumor hypoxia profoundly influences tumor development and response to cancer treatment. Cells adapt to low oxygen conditions by decreasing the rate of energy-consuming processes, including protein synthesis. We previously showed that cells exposed to hypoxia activate the endoplasmic reticulum (ER) kinase PERK which phosphorylates eIF2α, resulting in a decrease in global rates of protein synthesis. Here we report that disruption of the PERK-eIF2α pathway in tumor cells dramatically affects their survival under conditions of prolonged hypoxia and affects tumor growth in vivo. PERK-/- MEFs transformed with large-T antigen and mutant Ki-ras exhibited reduced survival rates (approx. 3-fold) after exposure to hypoxia compared with the PERK+/+ transformed MEFs. This reduction in survival correlated with increased caspase-12, caspase-3 and PARP cleavage. However, Caspase-12 KO cells were as sensitive to hypoxia-induced apoptosis as WT-MEFs, indicating that caspase-12 cleavage is not the primary mechanism in ER stress-induced apoptosis under hypoxia. When injected into nude mice, transformed PERK-/- MEFs formed tumors that grew significantly slower (5-fold) compared to tumors from transformed PERK+/+ MEFs. These results suggest that the presence of wt-PERK contributes to optimal tumor growth conditions in vivo, perhaps by contributing to cellular adaptation to hypoxic or other stresses. To further investigate the role of eIF2α phosphorylation in tumor development, we overexpressed the C-terminal fragment of GADD34 which has been shown to induce active dephosphorylation of eIF2α by recruitment of the protein phosphatase PP1A. Cells stably overexpressing GADD34 (HT-29.A1) showed very little basal and hypoxia-induced eIF2α phosphorylation compared to HT-29.puro control cells. Furthermore, when exposed to prolonged hypoxia, HT-29.A1 cells exhibited lower overall survival rates and increased levels of cleaved PARP and caspase-3 compared to HT-29.puro cells. Transfection of a plasmid expressing ER-targeted bcl-2 (but not mitochondrial bcl-2) rescued the HT-29.A1 cells from hypoxia-induced apoptosis, indicating that an ER-initiated proapoptotic stimulus is responsible for the increased sensitivity of HT-29.GADD34 cells to hypoxia. We are currently investigating the growth characteristics of tumor xenografts from HT-29A1 and HT-29.puro cells. Our data support a model in which transformed cells activate PERK and induce eIF2α phosphorylation as adaptive responses to hypoxic stress and that disruption of this pathway pushes the cells into an ER-dependent apoptotic pathway. Pharmacologic or genetic approaches that target this response could have therapeutic implications for tumor hypoxia-specific therapies.

[Proc Amer Assoc Cancer Res, Volume 45, 2004]