To the editor—We were surprised to read two papers in the January issue of Nature Medicine claiming both human fibroblast growth factor receptor1 (FGFR) and αVβ5 integrin2 as co–receptors for adeno-associated virus 2 (AAV2) infection.

FGFR is the high-affinity receptor for basic fibroblast growth factor3 (bFGF), although it has been shown that low-affinity binding to heparan sulfate molecules is prerequisite for the binding of bFGF with FGFR (ref. 4). This interaction between FGFR and heparan sulfate led to the proposal that herpes simplex virus, which uses heparan sulfate as an attachment molecule5, may use FGFR as its cellular receptor6. Qing et al. made an analogous suggestion1 about the binding of AAV2 to FGFR. The data on the relationship between HSV/FGFR and AAV2/FGFR were almost identical: bFGF was able to compete the viral binding and infection, and transfection of FGFR cDNA was capable of increasing viral binding and infection in 'nonpermissive' cells. However, subsequent research has shown that FGFR is not required (e.g., ref. 7) for HSV infection. Part of the viral-inhibiting properties of the bFGF is due to competition with binding to cell surface heparan sulfate, and cell lines such as Hep-2 and A431 that do not have FGFR on their cell surface are fully susceptible to HSV. We have done similar experiments for AAV2 (Table). Flow cytometry confirmed the absence of FGFR on Hep-2 or A431 cells7, yet they were readily transduced by rAAV2, showing that FGFR is also not required for AAV2 infection.

Table 1 FGFR and AAV2
Full size table

In the paper by Summerford et al.2, αVβ5 integrin was identified as a co-receptor for AAV, mainly on the basis of a viral overlay reported to show binding to denatured β5. Using the same viral overlay technique, we found a strongly binding, 150-kDa AAV2 binding protein8. We have also identified a weaker band at about 110 kDa, and have identified the protein as human nucleolin, based on peptide sequencing and antibody detection9. However, using viral overlay, we were unable to detect any binding of AAV2 to purified or recombinant β5 integrin. In addition, despite the title of the Summerford paper ("αVβ5 integrin..."), the gels seem to have been truncated so that binding to αV (150 kDa) could not be seen. This is surprising, because for adenovirus, most of the viral interaction is through the αV motif, and any β subunit can substitute (although typically αVβ3 is used for attachment and internalization)(ref. 10). In addition, adenovirus binding seems to depend on conformation and RGD motif and seems to be not detected by viral overlay. Finally, the transfection of β5 cDNA only increased by 260% the transduction of AAV2 infection, with significant transduction in CS cells (which lack β5). As in the Qing paper1, the ability to significantly infect or transduce cells that do not have the putative receptor suggests that these molecules are not essential in initiation of AAV2 infection.