The utility of adenovirus (Ad) for the delivery and expression genes both in vitro and in vivo is based on the widespread host range and the remarkable efficiency of cellular entry processes allowing high-level gene expression, together with their relative ease of preparation and purification (28). To achieve effective infection, two virus-cellular receptor interactions are required. The adsorption of Ad to target cell receptors, which have been identified as coxsackievirus and Ad receptor (CAR), histocompatibility class I molecule, or sialoglycoproteins, is initiated via the knob portion of the fiber (13, 20, 28). Subsequently, the interaction of the penton base with αV integrins facilitates the entry of Ad via clathrin-mediated endocytosis (1, 11, 22, 28). Ad can infect a wide range of cell types because the primary cellular receptors for Ad are expressed on most cells (28). However, specific gene delivery to a target cell is precluded due to the ubiquitous expression of CAR. On the other hand, Ad gene transfer is poor to some cell types due to a low level of Ad receptors on these cells (23, 28). It is thus apparent that receptor recognition is one of the key factors involved in viral tropism. To develop effective and specific gene delivery to target cells (resulting in increased safety in gene therapy applications), a number of recombinant Ad vectors have been constructed. Major attention has been given to modification of virus-cellular receptor interactions, including bispecific conjugates (6, 10, 28, 36), genetic modification of the fiber knob (5, 17, 28, 35, 37), modification of the hexon (3) or penton (34), ablation of the binding site on the knob (25, 28), and construction of chimeric vectors (9, 18, 28). However, it has become clear that virus-cellular receptor interactions (knob-CAR and penton base-αV integrins) are not the sole determinants of viral tropism. In this regard, Roelvink et al. reported that Ad type 2 (Ad2) and Ad9 utilize the same cellular fiber receptor but that the shorter fiber of Ad9 permitted fiber-independent binding of Ad9 penton base to αV integrins (24). Moreover, they showed that Ads belonging to subgroups A, C, D, E, and F all bind to the same cellular fiber receptor CAR and suggested that differences in subgroup tropism is significantly influenced by the length of the fiber shaft (26). Additional data regarding the importance of shaft length were obtained by Ad5 capsid-based vectors with chimeric fibers (15, 30). These observations led to a more systematic analysis of the role of fiber shaft length in Ad infectivity. Shayakhmetov et al. generated viruses with chimeric fibers containing short shafts (Ad9 or Ad35) or long shafts (Ad5) in combination with CAR (Ad5 and Ad9)- or non-CAR (Ad35)-recognizing knob domains (29). For Ad5 or Ad9 knob-possessing vectors, long shafts were critical for efficient infection compared with the weak attachment of the engineered short-shafted vectors. In contrast, for the Ad35 knob-possessing vectors, which infected cells by a CAR-independent pathway, fiber shaft length had no significant influence on infection. This study clearly demonstrated that the length of the fiber shaft influenced CAR- and αV integrin-mediated infections. Taken together, these reports have established shaft length as a key parameter whose modulation might allow tropism alternations. In this regard, the long-shafted Ad uses its fiber interaction (higher affinity than penton base-αV integrin interaction) exclusively for attachment to cell surface proteins and for charge-dependent repulsion (24, 26, 29), followed by internalization by the αV integrins. In contrast, for short-shafted Ad, including artificial fiberless Ad (19, 33), direct binding to the αV integrins (lower affinity than knob-CAR interaction) becomes dominant. In this instance, infectivity is decreased. Indeed, when the short-shafted fiber of Ad9 was engineered into the Ad5 capsid, a significant decrease in infectivity was observed compared to wild-type Ad5 (29). However, there is no significant difference in infection between wild-type Ad2 and wild-type Ad9, which expresses its short-shafted fiber but also the natural Ad9 capsid proteins, including the penton base (24). It is expected that Ad9 capsid does not require long-shafted fibers for efficient infection. On this basis, we hypothesized that the shaft length provides optimal spatial proportion between knob-CAR interaction and penton base-αV integrin interaction, especially when Ad uses CAR-integrin pathway for infection. Therefore, the length of shaft is limited from 6 β-repeats to 23 β-repeats in the process of evolution. Paradoxically, these observations provided us with the basis to explore whether artificial extension of the shaft alters the infectivity profile of Ad. We focused on rescuing recombinant Ad vectors, which express artificial longer shaft over 23 β-repeats. The primary sequence of the fiber shaft consists of 15-residue pseudorepeats. The length of the shaft is determined by the number of β-repeats, which ranges from 6 (Ad3, Ad11, and Ad35) β-repeats to 23 (Ad12) β-repeats in human Ads (2). Green et al. predicted that these repeats contained two β-strands and two turns (the cross-β model) (12). Stouten et al. subsequently proposed a triple β-helical model, taking into account length measurements from electron microscopy and fiber diffraction patterns (31). Recently, van Raaij et al. revealed a novel structural motif for the shaft as a triple β-spiral model based on the crystallographic data (32). According to this model, we attempted to make Ad5 capsid-based longer-shafted Ad vectors by incorporating Ad2 shaft fragments of different lengths into Ad5 shaft and investigated whether extension of the shaft over 23 β-repeats altered the infectivity profiles of human Ads.