TGFβ is a ubiquitously produced growth factor with important roles in pathologic processes such as cancer, fibrosis and autoimmunity, and in normal development, wound repair and homeostasis. There are three TGFβ isoforms, and all are secreted in a latent form. Latency is the result of a noncovalent association of the growth factor with its propeptide, which is called latency-associate peptide (LAP). TGFβ must be released from LAP before it can bind TGFβ receptors. The activation step is highly regulated but the specific activation mechanisms involved vivo have been poorly understood.

We discovered that αvβ6, a cell surface adhesion molecule in the integrin family, can activate latent TGFβ1 and TGFβ3 by interacting with an RGD sequence in the TGFβ1 and TGFβ3 LAPs. αvβ6 is expressed in epithelia, particularly after injury. Mice lacking αvβ6 have lung inflammation and are protected from lung fibrosis due to a relative lack of TGFβ signaling in the lung.

We also collaborated with S. Nishimura's lab at UCSF to show that a second RGD-binding integrin, αvβ8, activates TGFβ1 and TGFβ3. Mice lacking this integrin have abnormalities in vascular development. Comparison of knockouts of the two TGFβs and the two β integrin subunits reveals several partially or completely overlapping abnormalities in palate closure, immune regulation, and vascular development, suggesting that αvβ6 and αvβ8 are key TGFβ1/3 activators in vivo.

To determine the role of all RGD-binding integrins in the activation of latent TGFβ1, we made mice with a knock-in mutation of the TGFβ1 gene that changes the RGD site to RGE. These mice produce normal amounts of latent TGFβ1, but it cannot be activated by RGD-binding integrins. Strikingly, the mice have the same abnormalities seen in TGFβ1-null mice, indicating that RGD-binding integrins are indispensable for TGFβ1 activation.

We are currently further exploring the connection between these two integrins and the two RGD-containing TGFβ isoforms by generating mice with combined TGFβ1/3 mutations and mice with combined αvβ6/αvβ8 deficits. Our results support a tight functional system involving the two TGFβs and the two activating integrins. For example, mice with combined αvβ6/αvβ8 deficits have completely penetrant cleft palate, and mice with combined TGFβ1/3 mutations have the CNS vascular changes seen in αvβ8-null mice.

We are now focusing on the immune phenotype of mice lacking function of both αvβ6 and αvβ8. These mice develop severe autoimmune reactions that appear identical to (if not more severe than) those of TGFβ1-null mice. Also, we are testing the effectiveness of an inhibitory anti-αvβ6 mAb, developed by Biogen Idec, in mouse models of lung fibrosis. We find that anti-αvβ6 treatment effectively prevents radiation-induced lung fibrosis in mice, raising the possibility that such treatment might benefit patients with various forms of lung fibrosis.