B-80

J. S. Smith¹ , S. C. Stevenson² , J. Tian¹ , A. P. Byrnes¹ , ¹CBER, FDA, Bethesda, MD, ²Department of Diabetes and Metabolism, Novartis Institutes of Biomedical Research Inc., Cambridge, MA

When adenoviral vectors are administered systemically, they are quickly taken up by phagocytes of the reticuloendothelial system, including Kupffer cells (KCs) in the liver. We have recently shown in mice that E1/E3-deleted Ad5 vectors cause rapid KC death, followed by a slower depletion of KCs from the liver (Mol. Ther. 13:108-117, 2006). KCs lost membrane integrity within 10 min after i.v. injection of Ad5 vectors, and this eventually resulted in significant depletion of KCs from the liver. In the current study, we investigated which features of the adenoviral fiber contribute to KC toxicity by administering Ad5 vectors with the fiber partially or completely replaced by the type 35 fiber. When compared to the type 5 fiber, the type 35 fiber does not bind to CAR, has a much shorter shaft, and lacks a putative heparin-binding domain (KKTK) in the shaft. We also directly examined the role of the KKTK domain in the Ad5 shaft. All vectors were provided by Cell Genesys, Inc. and were described in Hum. Gene Ther. 14:777-787, 2003.

C57Bl/6 mice were injected i.v. with vectors and F4/80⁺ KCs in the liver were counted at 18 h. An Ad5 vector dose of 10¹¹ vp/kg depleted KCs by approximately 80%. KCs were depleted to a similar extent when the type 5 fiber head was replaced with the type 35 fiber head (5S35H), indicating that the type 5 and type 35 fiber heads were interchangeable and that CAR binding was not essential. However, when the type 5 shaft was replaced with the type 35 shaft, no KC depletion was seen, regardless of whether the fiber was completely replaced (35F) or only the shaft was replaced (35S5H). In addition, KC depletion was also absent with an Ad5 vector that had the shaft KKTK sequence replaced with GAGA (S*). This indicated that the shaft length did not determine toxicity, but rather that the putative heparin-binding domain was involved. Of note, all of the vectors were capable of producing KC depletion to some degree when injected at higher doses (10¹² vp/kg), indicating that none of the alterations provided absolute protection against KC toxicity.

In summary, we found that the type 5 fiber shaft made a major contribution to KC toxicity. Vectors that lacked the KKTK motif or contained the type 35 fiber shaft were impaired in their ability to deplete KCs. There was no obvious contribution from the type 5 fiber head, and CAR binding played no apparent role in KC toxicity. This information is likely to prove valuable in developing less toxic adenoviral vectors for systemic administration.