ced for 2 days prior to intracranial injections. Tumor volumes, plotted in a log scale, decreased significantly from HIF1 in reference to -gal. After infection with lentiviruses expressing gene products as indicated, GSC20 cells were grown in neurobasal-A medium and images of tumor spheres were captured at 25 and 50 magnifications, with scale bars of 1 mm and 400 m, respectively. Representative tumor lesions from 2 individual mice of each group are presented at 25 and 200 magnifications, with scale bars of 1 mm and 100 m, respectively. Tumor lesions are demarcated in dash lines. Mitoses are indicated by yellow arrowheads and multi-nucleation by blue arrows. doi:10.1371/journal.pone.0125125.g006 expression after continuous culture, yet retained the acquired malignant traits in the in vivo setting. These results indicate that repeated activation of HIF-1 can program cancer cells to acquire perpetual signaling possibly through feed-forward biochemical/metabolic loops or genetic/epigenetic changes, even though the underlying mechanism requires further investigation. The study also suggests that HIF-1 can promote malignant progression at its own expense, which might account for HIF-1 inactivation in human cancer. Although repeated cycles of hypoxia and reoxygenation have long been known to promote tumor progression, our experimental system allows us to directly interrogate HIF-1 and HIF-2, excluding other possible aspects such as changes in reactive oxygen species, which may also contribute to malignant progression. It should be noted that the cancer cells treated for 8 weeks in our study, although they no longer expressed HIF1, did maintain endogenous HIF-1 expression under hypoxia, a potential contributing factor to the process and a disadvantage of our experimental system. Interestingly, knockdown of endogenous HIF-1 by shorthairpin RNA in the cancer cells treated for PubMed ID:http://www.ncbi.nlm.nih.gov/pubmed/19777101 8 weeks impeded tumor growth but failed to prevent invasion, suggesting that intermittent induction of HIF1 in culture is key to programming MedChemExpress MRT-67307 glioma cells for invasion whereas endogenous HIF-1 potentially facilitates tumor growth. This interpretation is consistent with our results that tumors from the control groups, -gal and HIF2, exhibited much less invasion despite endogenous HIF-1 expression; however, we cannot exclude the possibility that endogenous HIF-1 was required for programming cancer 10 / 15 Lasting Effect of HIF-1 on Malignant Progression cells during intermittent induction of HIF1. This possibility will be better addressed in HIF-1-deficient cancer cells or conditional genetic models. It is interesting to note that GSC20-derived tumor cells in the control group featured rampant mitoses and numerous giant-sized, hyperchromatic nuclei, indicative of rapid cell proliferation and aberrant DNA replication; however, HIF1 expression not only retarded tumor growth but also markedly reduced mitosis of tumor cells and diminished multinucleation. These effects are 
in agreement with the inhibitory roles of HIF-1, but not HIF-2, in cell-cycle progression and DNA replication upon transient induction. Although these findings PubMed ID:http://www.ncbi.nlm.nih.gov/pubmed/19776277 are reminiscent of previous reports that HIF-1 retards tumor growth, our data collectively indicate that HIF-1 has pleotropic effects that are context dependent. Our study also indicates that HIF-1 is a potent inducer of glioma invasion. This is particularly interesting because investigations of glioma invasion have been hampered by the scarcity of represe