BACKGROUND: Anti-angiogenic therapies like bevacizumab are promising for cancer, but acquired resistance, which often includes increased invasiveness, can limit their utility. We investigated mediators of this invasive resistance using a large archive of bevacizumab-resistant glioblastoma (BRG) tissue and by developing novel xenograft models of anti-angiogenic therapy resistance. METHODS: Microarray analysis, immunostaining, immunoprecipitation, and proximity ligation assays (PLAs) were used to define mediators of invasive anti-angiogenic therapy resistance. RESULTS: Microarray analysis and immunostaining revealed increased hypoxia and reactive-oxygen species (ROS), along with upregulated ß1 integrin and receptor tyrosine kinase c-Met, in BRGs compared to pre-treatment (P < 0.005). An siRNA screen and western blot revealed upregulation of AP-1 (P < 0.01), an ROS-activated transcription factor, in BRGs and BRG-derived xenografts with siRNA targeting AP-1 components reducing ß1 integrin and c-Met expression in BRG-derived cells. Immunoprecipitation and PLA revealed formation of a c-Met-ß1 integrin complex involving ß1 heterodimer partners alpha5 and alphaV integrins, with the formation of this complex increased in culture by hypoxia, a feature of anti-angiogenic therapy resistance, and prolonged bevacizumab treatment in vivo. Cross-activation occurred in this complex, with c-Met activation by HGF increasing cell adhesion to fibronectin and fibronectin leading to c-Met phosphorylation. We also found that increasing VEGF concentration suppressed formation of the c-Met-ß1 integrin complex (P < 0.05). We developed U87-BevR, a novel xenograft model of acquired bevacizumab resistance through serial multi-generational treatment of U87MG xenografts. U87-BevR exhibited upregulated c-Met and ß1 integrin upon development of resistance. Treating U87-BevR with ß1 or c-Met inhibitors transiently suppressed invasive growth and reduced c-Met-ß1 complex formation. Use of 3- and 10-fold less bevacizumab to treat xenografts led to less initial efficacy but in a more durable fashion associated with less c-Met-ß1 integrin complex formation (P < 0.05). CONCLUSIONS: Invasive resistance limits the efficacy of anti-angiogenic therapy. We demonstrated increased activity of ß1 integrin and c-Met in BRGs and the formation of a c-Met-ß1 complex. The combination of chemotactic c-Met and haptotactic ß1 integrin endows BRG cells with their tremendous invasive capacity. The formation of this complex is promoted by two parallel mechanisms–increased transcription of each individual factor driven by hypoxia-induced ROS leading to AP-1 activation and relief of VEGF suppression of the formation of the complex. One option to prevent the formation of a c-Met-ß1 complex and thereby counteract the evolution of therapeutic resistance is dose reduction, which may be associated with lesser, but more durable responses, suggesting that, with anti-angiogenic therapy, “less is more.” SECONDARY CATEGORY: Preclinical Experimental Therapeutics.