page 609, Sawada and Sheetz provide significant new support for the second model.
The authors grew cells on collagen-coated silicon, and then used detergent to strip the cells down to their cytoskeletons. When the cytoskeletons were stretched 10% and incubated with cytoplasmic proteins, the proteins bound at distinct spots. Relaxed cytoskeletons produced a different protein binding pattern. Biochemical analysis identified a distinct subset of proteins, including paxillin, focal adhesion kinase, and p130Cas, that bind in a stretch-dependent manner. Confirming the relevance of the system, the stretch-dependent binding pattern of GFP–paxillin appears identical in vitro and in vivo.The absence of a cell membrane in these experiments rules out the involvement of changes in ion concentration, and suggests that matrix forces directly cause conformational changes in the cytoskeleton that can alter protein binding patterns. Although cytoskeletal binding alone could drive matrix-induced responses, the new data do not exclude the possibility that ionic changes could also influence force-dependent signaling. ▪