If interaction of signaling proteins is determined through binding of interaction domains that potentially recognize domains in many other proteins, how is inappropriate cross talk of signaling pathways avoided? For example, the yeast osmosensor Sho has an SH3 domain that recognizes Pro-X-X-Pro motifs, and Sho's specific interaction partner Pbs2 has an SH3 binding motif. But yeast have 26 other known SH3 domains that occur in many proteins. One solution to the problem, now proposed by Zarrinpar et al., is that negative selection actually removes potentially competing domains. The evidence consistent with such a strategy came from experiments in which the authors replaced the SH3 domain of Sho with each of the other 26 possible domains and looked for interaction of the modified protein with Sho's biological partner Pbs2. None of the other domains from yeast allowed detectable interaction in vitro or would support the osmolarity-sensing signaling pathway in vivo in which Sho and Pbs2 normally function. However, 6 out of 12 SH3 domains swiped from proteins belonging to other organisms did allow protein interaction in vitro and restoration of at least partial biological function in vivo. Expression of a Sho mutant that bound with high affinity to Pbs2, but also showed interaction with other SH3 domains, reduced fitness in yeast. Together the results suggest that, in this case, specificity has been generated by negative selection, which may have simply removed potential cross-reacting SH3-binding partners. Endy and Yaffe provide commentary.
A. Zarrinpar, S.-H. Park, W. A. Lim, Optimization of specificity in a cellular protein interaction network by negative selection. Nature 426, 676-680 (2003). [Online Journal]
D. Endy, M. B. Yaffe, Molecular monogamy. Nature 426, 614-615 (2003). [Online Journal]