Present addresses:* University of Sheffield, Sheffield, S57AU, United Kingdom;† Cancer Research UK, London, SE1 7EH, United Kingdom. embryonic lethality in mouse embryos. The phenotypes differ slightly, but a major feature in each case is that the yolk sac and embryonic vascular networks fail to form properly; ie, vasculogenesis and angiogenesis are defective (George et al. 1993; Yang et al. 1993). In both cases, endothelial cells (defined by expression of PECAM-1/CD31) do develop from their angioblast precursors. In α5-null embryos, endothelial tubes do form, but they are dilated and disrupted and the pattern of vessels is abnormal (Yang et al. 1993; Goh et al. 1997; Francis et al. 2002). The defects seen in fibronectin-null embryos are more severe, ranging from severely disrupted extraembryonic and intraembryonic vessels and heart to a complete absence of a heart, depending on the genetic background (George et al. 1993, 1997; Georges-Labousse et al. 1996). In the most severe cases, the two heart primordia fail to migrate to the midline to form the heart (George et al. 1997). Major deficits in vascular formation are also seen in α5-null embryoid bodies (Taverna and Hynes 2001) and in α5-null teratocarcinomas (Taverna and Hynes 2001; Francis et al. 2002). Fibronectin-null embryoid bodies, like fibronectin-null embryos, show more severe deficits than do their α5-null counterparts (see Table 1)(Francis et al. 2002), but some rescue of vasculogenesis can be achieved by adding back fibronectin to the developing embryoid bodies (Francis et al. 2002). In complete accord with these genetic analyses, antibodies against α5β1 or against fibronectin, or peptides that selectively block their interaction, have been shown to inhibit angiogenesis in several systems, including the chicken chorioallantoic membrane and implanted Matrigel plugs (Kim et al. 2000), and both α5β1 and fibronectin are up-regulated by the angiogenic growth factor FGF-2 (Kim et al. 2000). It seems clear from these results that α5β1 binding to fibronectin plays an essential role in vasculogenesis and angiogenesis. Given what is known about this receptor–ligand pair, those roles very likely include promotion of cell migration, appropriate ECM (eg, BM) assembly, and signal transduction into the cell, affecting cell proliferation and survival and other processes. Accordingly, this receptor–ligand pair seems a possible candidate for antiangiogenic drugs. Peptides can be chosen that are highly selective for this pair, even distinguishing them from other integrins that also recognize RGD, as does α5β1 (Koivunen et al. 1993, 1994). The major challenge in targeting α5β1 is likely to be toxicity, since this integrin is relatively widely distributed, as is fibronectin, and side effects are likely. α1β1 AND α2β1

These two integrins are also expressed by endothelial cells and are up-regulated by vascular endothelial growth factor (VEGF)(Senger et al. 1997). Both these integrins are receptors for a variety of collagens and, in some cells and situations, also for laminins. Antibodies to α1β1 plus α2β1 inhibit angiogenesis in response to VEGF in the chicken chorioallantoic membrane (Senger et al. 1997), and in mouse skin in response to human tumor xenografts (Senger et al. 2002), and block endothelial migration in collagen gels (Senger et al. 2002). Genetic data are in overall agreement; the genes for both α1 and α2 have been knocked out in mice (Gardner et al. 1996; Pozzi et al. 2000; Chen et al. 2002). Neither produces embryonic lethality; the mice are viable and fertile. The double knockout has not yet been studied, which will be necessary because of possible overlapping functions of these two integrins …