The above data prove beyond reasonable doubt that human CD46 is a receptor for several adenoviruses. Again the issue of the relevance of CD46 for viral tropism has to be raised, and again the existence of alternative receptors, coreceptors, and postentry virulence determinants has to be invoked to explain the distinctive tropism and pathology of the different adenovirus-based serotypes. An interesting observation in this respect is that the CD46 C isoform, which is highly expressed in the eye (cornea/conjunctiva) and brain, is preferentially used by Ad37.
In the perspective of adenovirus vector applications, the fact that certain strains contact a ubiquitous protein rather than the much less readily available CAR is good news. Present gene therapy protocols rely on vectors derived from Ad2 and Ad5, two species C adenoviruses that use CAR as a receptor. However, CAR is expressed only in a few tissues, and efficient adenovirus-mediated transgene expression is associated with CAR availability (11
). The question now becomes how successful gene transfer protocols can be, based on CD46 usage and, more in general, whether and how a ubiquitous molecule can be exploited for vector targeting.
The fact that certain viruses become attenuated when they bind tightly to ubiquitous receptors expressed at high levels (reference 20
and references therein) is a warning of the difficulties that have to be overcome. For example, tick-borne encephalitis virus adapts to negatively charged heparan sulfate: local patches of predominantly positively surface charges evolve in its attachment protein (20
). The positively charged amino acids are selected in tissue culture, but when a tissue culture-adapted virus is passed into mice neuroinvasiveness is attenuated.
This does not mean that any level of heparan sulfate binding may preclude vector targeting. In contrast, low-affinity binding may concentrate viral particles in a given organ expressing high heparan sulfate levels and facilitate subsequent interactions with a high-affinity receptor or a coreceptor. Detailed knowledge of the organ distribution and concentration of receptor and coreceptor molecules is necessary to predict the efficiency of targeting approaches. Characterization of the interactions of the viral attachment proteins with their receptors is also required to plan the production of recombinant viruses with carefully balanced affinities to different receptors. Towards this it was shown that selectively receptor-blind viruses can be created through the introduction of point mutations in their attachment proteins (5
The art of vector targeting is only beginning to develop into a science (6
). Ligand-directed targeting of viral vectors to disease sites is based on the present, incomplete framework of knowledge about receptor hierarchy and tissue distribution and proceeds slowly. Alternative approaches to targeting that rely on the characterization of the tropism of viruses collected from primates or other mammals are being pursued (10
). Moreover, chimeras of different viral species are being produced in the perspective of generating vector libraries with novel receptor specificities from which individual vectors can be selected. These approaches will synergize with studies of viral structure and assembly to guide the development of vectors with incrementally higher tissue and cell specificity.