When Caroline Damsky moved her lab to the University of California, San Francisco, in 1985, she knew she would have an instant colleague in Zena Werb. There was a natural connection between their work. That connection would be reflected in their discoveries, which delineated an integrin-mediated pathway from the extracellular matrix (ECM) outside the cell to the internal, gene-expressing life of the cell. Figure 1 Cells with a ligated fibronectin receptor (top) express collagenase (bottom). Werb and Damsky's interests—metalloproteinase (MMP) gene expression and adhesion receptor antibodies—were both correlated with changes in cell shape and the actin cytoskeleton. Some dismissed the shape changes as simple, physical responses to a changing attachment environment, with no need to invoke signaling, but the two wondered if there was something more. “Were the cells getting signals from the ECM that could affect [protease] gene expression?” asks Damsky. “Zena was the doyenne of the protease side of things and I was the doyenne of the integrin side.” Until this point, integrins had been viewed as being important primarily in “sticking cells down,” says Damsky. Her hunch was that integrin's role went beyond glue to sending signals of some kind. At the same time, work paralleling Werb's suggested that proteases had some part in regulating cell connections to the ECM. Damsky says it was only natural to combine forces. When Damsky added her adhesion-blocking antibodies to fibroblasts in culture, thus specifically neutralizing a fibronectin receptor named α5β1 integrin, the cells started expressing collagenase and stromelysin, two ECM-degrading MMPs (Werb et al., 1989). Receptor clustering was needed: induction was increased when the antibodies were cross-linked and it did not occur with monovalent fragments. Although native fibronectin did not induce protease expression, the authors showed that fragments of fibronectin, ranging from 120 kD to six amino acids, did. The fragments shared the RGD cell binding sequence characterized by Pytela et al. (1985). This suggested that degradation products of fibronectin, bound to integrins, could turn on protease expression. “Cells were recognizing different states of the matrix molecule,” says Werb. “Embedded in a molecule was both the mechanism for homeostasis as well as for cell migration and changes that might occur, for example, in wound healing.” The idea fit with emerging studies showing distinct responses to different fibronectin fragments. For example, human monocytes had chemotactic activity to the cell-binding domain of fibronectin, but not to the intact molecule (Clark et al., 1988). Damsky notes wryly that although they were first to demonstrate integrin signaling, integrins became “respectable” signaling molecules after the labs of Joan Brugge, Thomas Parsons, and Lewis Romer showed that integrin ligation triggered the phosphorylation of focal adhesion kinase (Lipfert et al., 1992; Burridge et al., 1992). Nevertheless, the Werb/Damsky collaboration clearly set up a very large field, says Werb. She notes with pride that their work appears at the top of the list of 3,488 papers pulled up by a search for “integrin signaling.” Clark, R.A.F., et al. 1988. J. Biol. Chem. 263:12115–12123. [PubMed] Lipfert, L., et al. 1992. J. Cell Biol. 119:905–912. [PubMed] Pytela, R., et al. 1985. Cell. 40:191–198. [PubMed] Werb, Z., et al. 1986. J. Cell Biol. 102:697–702. [PubMed] Werb, Z., et al. 1989. J. Cell Biol. 109:877–889. [PubMed]