Enlarged vessels (green) form in the presence of MMP-cleaved forms of VEGF.

VEGF stuck to the extracellular matrix induces branchy vessels that support tumorigenesis, but soluble VEGF prompts wider vessels that cannot provide for a tumor, as shown on page 681 by Lee et al.

VEGF directs vascular growth and patterning in developing, adult, and tumor tissues. This extracellular signaling is thought to require VEGF that is not bound to the matrix. Soluble VEGF is made by either mRNA splicing that removes the matrix attachment region, or by MMP-dependent matrix remodeling. Lee et al. now show that several MMPs also cleave VEGF directly, thus releasing it from the matrix.Contrary to expectations, however, soluble VEGF was less effective than its matrix-bound version at supporting the vessel growth needed for tumor survival. Uncleaved VEGF induced long, thin, branchy vessels that provided the circulation needed for tumorigeneis. But soluble VEGF resulted in wide but sparse vessels that did not allow tumor growth, probably because they support very low blood pressure.

The difference in vessel growth stems from the factor's localization. Both forms activated the same VEGF receptor, although preliminary evidence suggests they induce its phosphorylation on different residues. Soluble VEGF, which bathes the vessels, induced cell proliferation. The bound version, which is detected only at discrete sites, acted more like a chemokine—inducing filopodia and directed migration by generating a gradient. The different downstream signaling proteins that each version recruits to the receptors remain to be identified.

MMPs are provided in large quantities by inflammatory cells such as macrophages. Although inflammation is usually associated with the progression of cancer, in this case, the MMP reinforcements might combat tumor growth. In any case, the authors caution against the common practice of measuring circulating (soluble) VEGF levels to gauge tumor growth.

As most MMP knockout mice develop normally, VEGF cleavage might not be pertinent to developmental vascular patterning. There, alternative splicing seems to be the more important control mechanism. The authors plan to test this theory by creating knock-in mice that express an MMP-resistant VEGF that can still be spliced.