Seeding prosthetic arterial grafts with genetically modified endothelial cells (ECs) has the potential to substantially improve graft function. However, preliminary applications suggest that grafts seeded with retrovirally transduced ECs yield a significantly lower percent surface coverage than those seeded with nontransduced ECs. The objective of this study was to test the hypothesis that canine ECs transduced with the human tissue plasminogen activator (tPA) gene would have a lower rate of adherence to pretreated expanded polytetrafluoroethylene (ePTFE) both in vitro and in vivo and that they would proliferate at a slower rate on pretreated ePTFE in vitro. Early passage ECs derived from canine external jugular vein were transduced with the retroviral MFG vector containing the gene for human tPA. ECs exposed to media alone served as controls. Iodine 125-labeled ECs were seeded in vitro onto ePTFE graft segments pretreated with canine whole blood, fibronectin (50 micrograms/ml), or media alone, and the percent of ECs adherent at 1 hour were determined (n = 3). Additional tPA-transduced and -nontransduced ECs were grown for 10 days on either fibronectin (50 micrograms/ml)-pretreated ePTFE wafers or tissue culture plastic pretreated with gelatin (1%) or fibronectin (50 micrograms/ml), and the EC proliferation rates were determined (n = 3). Furthermore, 125I-labeled ECs were seeded onto fibronectin (50 micrograms/ml)-pretreated ePTFE graft segments implanted as carotid and femoral artery interposition grafts (n = 3). The grafts were harvested after 1 hour, and the percent of ECs adherent was determined. Human tPA was detected by immunohistochemical staining in 61% +/- 5% of the transduced ECs and was expressed at 35.4 +/- 12.9 ng/hr/10(6) cells. Fibronectin and whole blood pretreatment of the ePTFE grafts led to greater EC adherence in vitro than did media alone (90.9% +/- 5.3% vs 77.8% +/- 5.8% vs 4.7% +/- 1.1%, p < or = 0.05). No significant difference in the rates of adherence or proliferation was seen in vitro between the transduced and nontransduced ECs. No significant difference in proliferation was found for the transduced ECs on the three matrices tested in vitro. In contrast, adherence of the transduced ECs in vivo was significantly lower than that of nontransduced ECs (64.7% +/- 2.1% vs 73.7% +/- 4.1%, p < or = 0.05) 1 hour after implantation. Lower rates of surface endothelialization by genetically modified ECs in vivo do not appear to be due to an impaired capacity to initially adhere or proliferate on the synthetic graft but may result from decreased adherence after exposure to in vivo hemodynamic forces.