Endostatin has been considered a highly specific inhibitor of endothelial cell proliferation and/or migration. To explore the use of endostatin in antiangiogenic gene therapy, we generated a recombinant adenovirus, AdEndo, carrying the gene for mouse endostatin. Injection of 10(9) PFU of AdEndo resulted in a low but significant suppression (25%) of preestablished tumor growth in murine models involving murine Lewis lung carcinoma (LLC) and human breast cancer MDA-MB-231 tumors. Greater anticancer activity was observed when the same dose of AdEndo was injected into two other preestablished murine models involving C51 murine colon cancer and HT29 human colon cancer (55 and 47% tumor growth reduction, respectively). In vitro, endostatin derived from AdEndo-infected MRC-5 fibroblasts inhibited the growth of C51 and HT29 cell lines (72 and 61%, respectively). The extent of this inhibition was comparable to that observed in endothelial cells: 75% for microcapillary endothelial cell line HMEC-1, 52% for human dermal microvascular endothelial cells, 46% for human umbilical vein endothelial cells, and 67% for calf pulmonary arterial endothelial cells. Both endothelial and colon cancer cells showed a clear increase in cell apoptosis (4- to 5-fold for endothelial cells and 5- to 10-fold for colon cancer cells) and an accumulation in the G(1) phase of the cell cycle. This antiproliferative activity was not observed in other tumor cell lines: LLC, MDA-MB-231, murine colon adenocarcinoma MC38, human prostate cancer cell line DU145, and human breast cancer cell line CAL51. Taken together, these results provide evidence that, in addition to its antiangiogenic activity, endostatin exerts a direct anticancer action that appears to be restricted to some tumor cell lines. Thus, endostatin could be used in some colon cancer treatments and its clinical efficacy would depend on the response of tumor cells themselves.