PURPOSE: Adenovirus-mediated arterial gene transfer is a promising tool in the study of vascular biology and the development of vascular gene therapy. However, intraluminal delivery of adenoviral vectors causes vascular inflammation and neointimal formation. Whether these complications could be avoided and gene transfer efficiency maintained by means of delivering adenoviral vectors via the adventitia was studied. METHODS: Replication-defective adenoviral vectors encoding a beta-galactosidase (beta-gal) gene (AdRSVnLacZ) or without a recombinant gene (AdNull) were infused into the lumen or the adventitia of rabbit carotid arteries. Two days after infusion of either AdRSVnLacZ (n = 8 adventitial, n = 8 luminal) or AdNull (n = 4 luminal), recombinant gene expression was quantitated by histochemistry (performed on tissue sections) and with a beta-gal activity assay (performed on vessel extracts). Inflammation caused by adenovirus infusion was assessed 14 days after infusion of either AdNull (n = 6) or vehicle (n = 6) into the carotid adventitia. Inflammation was assessed by means of examination of histologic sections for the presence of neointimal formation and infiltrating T cells and for the expression of markers of vascular cell activation (ICAM-1 and VCAM-1). To measure the systemic immune response to adventitial infusion of adenovirus, plasma samples (n = 3) were drawn 14 days after infusion of AdNull and assayed for neutralizing antibodies. RESULTS: Two days after luminal infusion of AdRSVnLacZ, approximately 30% of luminal endothelial cells expressed beta-gal. Similarly, 2 days after infusion of AdRSVnLacZ to the adventitia, approximately 30% of adventitial cells expressed beta-gal. beta-gal expression was present in the carotid adventitia, the internal jugular vein adventitia, and the vagus nerve perineurium. Elevated beta-gal activity (50- to 80-fold more than background; P <.05) was detected in extracts made from all AdRSVnLacZ-transduced arteries. The amount of recombinant protein expression per vessel did not differ significantly between vessels transduced via the adventitia (17.1 mU/mg total protein [range, 8.1 to 71.5]) and those transduced via a luminal approach (10.0 mU/mg total protein [range, 3.9 to 42.6]). Notably, adventitial delivery of AdNull did not cause neointimal formation. In addition, vascular inflammation in arteries transduced via the adventitia (ie, T-cell infiltrates and ICAM-1 expression) was confined to the adventitia, sparing both the intima and media. Antiadenoviral neutralizing antibodies were present in all rabbits after adventitial delivery of AdNull. CONCLUSION: Infusion of adenoviral vectors into the carotid artery adventitia achieves recombinant gene expression at a level equivalent to that achieved by means of intraluminal vector infusion. Because adventitial gene transfer can be performed by means of direct application during open surgical procedures, this technically simple procedure may be more clinically applicable than intraluminal delivery. Moreover, despite the generation of a systemic immune response, adventitial infusion had no detectable pathologic effects on the vascular intima or media. For these reasons, adventitial gene delivery may be a particularly useful experimental and clinical tool.