It is established that tumor cell-derived VEGF acts on endothelial cells to promote angiogenesis and tumor growth. Here, we demonstrate that in K5-SOS-dependent mouse skin tumors, autocrine VEGF is required for tumor cell proliferation in a cell-autonomous and angiogenesis-independent manner. VEGF is upregulated in SOS-expressing tumors, and its deletion in epidermal cells delays tumorigenesis by suppressing angiogenesis and tumor cell proliferation. Epidermis-specific Flt1 deletion also impairs tumorigenesis and proliferation. Surprisingly, complete tumor inhibition occurs in the absence of VEGF in EGFR mutant mice, demonstrating that VEGFR and EGFR synergize in neoplastic cells to promote tumor growth. Mechanistically, K5-SOS upregulates VEGF, Flt1, and Neuropilin-1 in an Erk-dependent manner, thereby activating an autocrine proliferation loop, whereas EGFR prevents tumor cells from apoptosis. Moreover, Flt1 is upregulated in human SCC, and its inhibition in SCC cells impairs proliferation. Thus, in addition to regulating angiogenesis, VEGF has to be considered as a potent growth factor for epidermal tumors.

The coxsackievirus and adenovirus receptor (CAR) is a high affinity receptor used by adenoviruses, including adenovirus type 5 (Ad5). The adenovirus fibre molecule bears the high affinity cell binding domain of Ad5, allowing virions to attach to CAR. The avian adenovirus CELO displays two fibre molecules on its capsid and it was logical to expect that the cell binding functions of CELO might also reside in one or both of these fibres. We had previously shown that the cell binding properties of CELO resemble Ad5, suggesting that the two viruses use similar receptors. Experiments with CAR-deficient CHO cells and CHO cells modified to express CAR demonstrated that CELO has CAR-dependent transduction behaviour like Ad5. Mutations were introduced into the CELO genome to disrupt either the long fibre 1 or the short fibre 2. A CELO genome with fibre 2 disrupted did not generate virus, demonstrating that fibre 2 is essential for some stage in virus growth, assembly or spread. However, a CELO genome with disrupted fibre 1 gene produced virus (CELOdF1) that was capable of entering chicken cells, but had lost both the ability to efficiently transduce human cells and the CAR-specific transduction displayed by wild-type CELO. The ability of CELOdF1 to transduce chicken cells suggests that CELOdF1 may still bind, probably via fibre 2, to a receptor expressed on avian but not mammalian cells. CELOdF1 replication was dramatically impaired in chicken embryos, demonstrating that fibre 1 is important for the in vivo biology of CELO.

No abstract