Genetic engineering using recombinant adenoviruses offers an opportunity to modify islet grafts in order to prevent allograft rejection. We have used an adenovirus coding for CTLA4Ig to compare its efficacy in preventing islet rejection depending on local or systemic production after gene transfer either into the islets or intramuscularly, respectively. Islet allograft survival was also evaluated using recombinant CTLA4Ig administered intraperitoneally or incubated ex vivo with islets prior to transplantation. Transduction of islets with 10(3) or 10(4) plaque-forming units (pfu) per islets of AdCTLA4Ig prolonged islet survival (mean +/- standard deviation [SD] days = 19.5 +/- 5.8 and 19.5 +/- 5.6, respectively, vs. 10.6 +/- 2.4 in control islets, p < 0.001), with low levels of circulating CTLA4Ig. In contrast, long-term survival (>60 days) was obtained after intramuscular injection of AdCTLA4Ig that resulted in sustained high levels of circulating CTLA4Ig. Islets incubated in vitro with CTLA4Ig did not show prolonged survival (10.3 +/- 2.5 days). Graft rejection was delayed after one injection of CTLA4Ig (23 +/- 7.6 days, p < 0.003 vs. control). Recipients of long-term surviving grafts after intramuscular AdCTLA4Ig gene transfer were not tolerant because second islet grafts of donor origin were rejected. These recipients also had a strong inhibition of humoral responses against nominal antigens, whereas animals receiving transduced islets showed normal responses. These data demonstrate that local production of CTLA4Ig after gene transfer was as efficient as a single injection of CTLA4Ig in preventing graft rejection. Furthermore, intramuscular gene transfer of CTLA4Ig was the most efficient way to induce long-term islet graft survival but no donor-specific tolerance was induced.