Glioblastoma multiforme is distinguished from its less malignant astrocytoma precursors by intense angiogenesis and frequent loss of tumor suppressor genes on chromosome 10. Here we link these traits by showing that when a wild-type chromosome 10 was returned to any of three human glioblastoma cell lines U251, U87, or LG11, they lost their ability to form tumors in nude mice and switched to an antiangiogenic phenotype, as measured by the inhibition of capillary endothelial cell migration and of corneal neovascularization. This change in angiogenesis was directly due to the increased secretion of a potent inhibitor of angiogenesis, throm-bospondin-1, because: (a) neutralizing thrombospondin completely relieved the inhibition; (b) the inhibitory activity of thrombospondin was not dependent on transforming growth factor β; and (c) chromosome 10 introduction did not alter secreted inducing activity. The inducing activity was dependent on vascular endothelial cell growth factor and had an ED50 of 10 µg/ml in media conditioned by parental cells and 9–13 µg/ml in media conditioned by chromosome 10 revertants. Normal human astrocytes were also antiangiogenic due to secreted thrombospondin. The effect of chromosome 10 on thrombospondin production in vitro was reflected in patient material. Normal brain and lower grade astrocytomas known to retain chromosome 10 stained strongly for thrombospondin, but 12 of 13 glioblastomas, the majority of which lose chromosome 10, did not. These data indicate that the loss of tumor suppressors on chromosome 10 contributes to the aggressive malignancy of glioblastomas in part by releasing constraints on angiogenesis that are maintained by thrombospondin in lower grade tumors.
This work was supported by NIH Grants CA52750 (to N. P. B.), CA56041 (to P. A. S.), HL39926 (to P. J. P.), a Searle Graduate Fellowship, and Department of the Army Grant DAMD17-94-J-4466 (to S. C. H.).