Jean-François Dedieu

Genetic manipulations of adenovirus type 5 fiber resulting in liver tropism attenuation

The development of genetically modified adenoviral vectors capable of specifically transducing a given cell population requires the addition and functional presentation of particular tropism determinants within the virus capsid, together with the abrogation of the molecular determinants that dictate their natural tropism in vivo. The human adenovirus serotype 5 (Ad5) first attaches to the cell surface following high-affinity binding of the C-terminal knob of the fiber capsid protein to the coxsackie and adenovirus receptor (CAR). Here we have assessed whether genetic shortening of the fiber shaft (virus BS1), or replacing the Ad5 fiber shaft and knob with their Ad3 counterparts (virus DB6), could cripple this interaction in vitro and in vivo. A 10-fold decrease in the binding of the modified capsids to soluble CAR was evidenced, which correlated with a similar reduction of their ability to transduce CAR-positive cells in vitro. The ability of BS1 to interact with cellular integrins was also impaired, suggesting that the penton base and the short-shafted fiber when embedded in the capsid preclude each other from efficiently interacting with their cognate cell surface receptors (CAR and integrins respectively). BS1 and DB6 intravenous injections in mice further supported a profound impairment of the ability of the capsid-modified viruses to transduce the liver as demonstrated by a 10-fold reduction of intracellular viral DNA and transgene expression. Interestingly enough, the host humoral response was also specifically weakened in BS1- and DB6-inoculated animals. Taken together, these observations indicate that (i) fiber shortening and (ii) pseudo-typing of Ad5-based vectors with the shaft and knob from non-CAR-binding serotypes constitute two promising strategies to successfully attenuate their native tropism in vitro and most importantly in vivo.

Targeted adenoviral vectors

Subgroup C human adenoviruses (Ad2 or Ad5) can transduce a wide array of dividing and quiescent adherent cell types in vitro and in vivo, essentially because their primary receptor (CAR) is widely expressed at the cellular surface. On the other hand, specific target cell types (eg some tumor cells, smooth muscle cells, fibroblasts) display limiting amounts of the CAR receptor so that large viral doses are needed for efficient infection, a constraint which favors vector dissemination in vivo, even after loco-regional injection. Abrogating the native virus-CAR interaction while allowing the virus to use a CAR-independent pathway for entry would thus represent a major milestone in order to better control the vector tropism in vivo. Towards this goal, we first engineered a series of capsid-modified adenoviruses by genetically inserting targeting peptides within protruding loops of the fiber and hexon capsid monomers (Vigne E, et al: J Virol 1999, 73:5156). The most interesting virus – AE43 – displays within the fiber HI loop a vitronectin-derived high affinity peptide for the urokinase-type plasminogen activator (uPAR or CD87), a cell surface receptor upregulated during cellular activation, and which controls cellular migration and invasion. Most interestingly, AE43 could increase transduction of cells normally refractory to adenovirus infection more than 100-fold in vitro. We could also demonstrate that AE43 retained its ability to enter the cell via a CAR-dependent pathway in vitro, and displayed a normal tropism following systemic injection in mice. To cripple the virus-CAR interaction, we then evaluated various strategies, including the introduction of CAR-ablating mutations within the fiber, and the shortening of the fiber shaft. Interestingly, some of these constructs could decrease transduction of CAR-positive cells at least 10-fold in vitro, with a similar reduction in liver transduction following systemic injection in naive mice. We are currently assessing the effect of combining within a single vector the uPAR-binding peptide of AE43 with our best CAR-ablating mutations.