Tumor Angiogenesis: Putting a value on plastic GEMMs

It has long been appreciated that angiogenesis can contribute to the induction and/or progression of several diseases, including retinopathies, arthritis, psoriasis, and cancer^1,2. Since endothelial cells typically exist in a quiescent state in normal vessels, the molecular changes required for these cells to become proliferative and invasive have long been considered potential targets for the control of both neovascularization, and for control of the accompanying pathology. To this end, over the last 3 decades, a number of novel effectors of angiogenesis have been characterized, ranging from growth factors such as VEGF ³ to signaling effectors such as RasGAPs ⁴ to cell adhesion molecules ^5,6 , including the alpha-V integrins featured in the article of Steri et al. on page xxx. Nonetheless, as an aggregate, the efficacy of anti-angiogenic agents have only produced a modest impact on disease ⁷.

We and others have demonstrated that antagonists of alpha V integrins target and suppress pathological angiogenesis resulting in decreased tumor growth ^8,9, inflammatory disease ^8,10 and retinal disease ⁵. In glioblastoma patients treated with these agents some clinical efficacy was noted ¹¹ even though overall survival was not significantly impacted in recently announced phase III trials. Studies by Hynes and colleagues showed that mice deficient in either αv or β3 showed robust blood vessel development ^12,13. In fact, these animals displayed enhanced endothelial cell VEGFR expression associated with increased tumor angiogenesis. This led the authors to conclude that αvβ3 plays a negative role in angiogenesis. How, then, can one reconcile the fact that αvβ3 antagonists suppress angiogenesis in mice and man? In an elegant series of studies, Steri et al. examined the effects of acute depletion of endothelial cell integrin αvβ3 on neovascularization, via Cre-mediated inactivation of a floxed β3 gene. Complementary approaches were used in this multicenter tour-de-force of mouse genetics, including Tie2 driven expression of Cre, and hydroxytamoxifen-inducible expression of Cre driven by the PDGFb promoters.

The findings by Steri and colleagues that αvβ3 is required for angiogenesis are consistent with previous studies by Byzoya and coworkers ¹⁴, who showed that mice expressing a signaling mutant of the β3 integrin subunit in which both cytosolic tyrosine residues were mutated to phenylalanine were deficient in angiogenesis. This was associated with decreased interactions between VEGFR2 and integrin αvβ3, which was in turn associated with decreased VEGFR2 activation. The deficiency may have resulted from the absence of an active integrin signaling complex, since this mutant eliminated the binding sites for kindlins, which stabilize integrin-ligand interaction ^15-17.

Steri et al. found that while long-term knockdown of integrin αvβ3 resulted in no impact on the growth of tumors in either knockdown system, the acute down-regulation resulted in an immediate, decreased angiogenesis and growth of the tumor. The results corresponded to a decreased invasiveness of the temporally induced ECs in aortic ring assays, which suggested that overall competence of the ECs to invade the tissue was compromised by acute deletion of integrin β3. However, over the longer term, the plasticity of the endothelial cells, and/or the tumor cells eliciting the response, permitted a ‘cellular rewiring’ that promotes alternative means by which endothelial cells execute the key angiogenic functions of invasiveness, survival and proliferation. Notably, the authors were able to identify one potentially important mechanism; the decreased expression of a crucial integrin downstream signaling protein, focal adhesion kinase (FAK), appeared to permit an increased invasiveness among the integrin-deficient endothelial cells. This corresponded with recent studies implicating haplo-sufficient endothelial FAK expression in increased tumor growth. Together, the results paint a compelling picture of the adaptations endothelial cells can make to execute a critical physiological program.

Still, all is not as clear as it might be, and certain aspects of these genetic studies defy easy explanation.

The depletion of β3 integrin was shown to have an inhibitory effect on the expression of integrin αvβ5 in these genetic models, and this feature has not been observed before in β3-integrin targeted models. It is possible that with the loss of β3, some stabilizing influence on β5 is lost. Given that β3 and β5 appear to have unique and somewhat discrete roles in angiogenesis ⁵, the concomitant loss of both integrin heterodimers is surprising, and may be significant. A dominant role for a cytokine-activated integrin β5 may cause a de facto requirement for its own down-regulation; alternatively, the enhanced expression of VEGF in the β3-depleted cells could result in increased internalization and turnover of β5. Conversely, it is not known if integrin expression changes in response to therapies that target growth factors.

How the findings by Steri insights could, or should, be applied to our current knowledge of tumor therapy remains an open question. For example, if depletion of β3 integrin results in increased VEGF dependence, is this an indication that these tumors will exhibit increased sensitivity to bevacizumab¹⁸, or to other VEGF pathway targeting molecules?

In a similar vein, if decreased FAK activity is part of a long-term adaptation of integrin loss, will the use of FAK inhibitors, over time, promote tumor angiogenesis and progression? Interestingly, FAK expression is required for angiogenesis in vivo ¹⁹, and FAK inhibitors have already been linked to decreased tumor vasculature, tumor growth and metastasis ^20-22. Indeed, VEGF-producing tumor cells promote vascular permeability (a critical initial common event in both angiogenesis and tumor extravasation), an effect that is abrogated in the presence of FAK inhibition or genetic expression of kinase dead FAK ²³.

The question remains as to whether genetic deletion of an integrin is truly a good representation of what occur following treatment with an integrin antagonist. Given that integrin antagonism can actively promote cell death ^9,10,24, while deletion of the integrin prevents this ²⁵, it would seem at face value that this is another example of genetics and pharmacology conflicting. Indeed, the current study by Seri and colleagues goes a long way to resolving this. Precisely how these two systems compare, however, may not be truly known until directed tests are done using integrin antagonists and genetic models in parallel. Preferably, these would be performed in an orthotopic tumor setting, with inducible loss of the integrin, alone and in combination. Such a study would also more permit a clearer elucidation of specific non-endothelial effects of integrin antagonists.

Nonetheless, the acute deletion model described by Steri et al., clearly identifies at least one aspect of β3 function that may also be disrupted pharmacologically; as a regulator of an endothelial ‘subroutine’ governing early/initial tumor vascularization. Although the molecular aspects of this subroutine are not identified, it may involve feedback with FAK. By contrast, endothelial cells lacking β3 for an extended period of time do not appear to use, or require, the same subroutine. So what might this mean in the clinic? Both integrin antagonists and FAK inhibitors are well tolerated, with minimal toxicity ^25,26. At a minimum, the antagonism of integrin αvβ3 would be predicted to slow the maturation of micro-metastases, while FAK inhibition could provide a second barrier to tumor spread. The final impact of these on patient progression, however, would depend intimately upon the disease status itself, and the genetics of the particular cancer. Once an agent is proven in mice, the only way to determine efficacy in human patients is in the clinic. The insights provided by the studies of Steri et al. provide useful clues into optimizing approaches for anti-angiogenic therapy, and ultimately may help to guide current clinical approaches.