Two functional properties shared by many tumor suppressor genes are the inhibition of cell-cycle progression and the induction of apoptosis. Recent work in the fruit fly has uncovered a new pathway for cellular homeostasis, involving the scaffold protein Salvador (Sav) and the two protein kinases Hippo (Hpo) and Warts (Wts)[1–3]. All components of this pathway have mammalian homologs, of which WW45 (the human Sav ortholog) and Lats-1/2 (the two human Wts orthologs) are considered tumor suppressors [1, 4]. According to the current model, the Sav protein acts as a scaffold that binds simultaneously to Hpo and Wts. This complex allows Hpo to efficiently phosphorylate Wts, which apparently triggers cell cycle arrest and apoptosis by downregulating cyclin E, Diap1, and other targets [2, 3, 5]. The interaction between Sav and Wts is mediated by the WW domains of Sav, which recognize a PPxY motif in Wts [1]. The recently discovered interaction between Hpo and Sav is, at least in part, mediated by their highly conserved carboxyl termini. We further analyzed this conserved region with the generalized profile method [6] in order to find more distantly related relatives of either family. After two rounds of iterative profile refinement, we found statistically significant evidence for a relationship between the two domains: members of the Sav family matched the Hpo profile with p values better than 0.01, and members of the Hpo family matched the Sav profile with similar significance (Figure 1). Besides Hpo and Sav, the profile searches also uncovered highly significant matches to members of a third protein family—the Rassf proteins. We refer to this new homology domain as SARAH (for Sav/Rassf/Hpo).

As shown in Figure 1, the SARAH domain spans approximately 50 residues. In all three protein families, it is located at the extreme carboxyl terminus. Secondary structure prediction programs suggest a long α-helix spanning most of the conserved region. The carboxy-terminal portion of the SARAH domain has a moderate to high coiled-coil propensity, as predicted by Coils-2 [7] and other programs. The predicted starting point of the coiled-coil region varies slightly between the member families. Interestingly, the amphipathicity pattern of the Sav subfamily predicts a heptad register differing from that of the other two families (Figure 1). While the multiple alignment of the SARAH domain does not reveal any invariant residues, several positions are highly conserved. While some of the conservation can be explained by generic coiled-coil properties, other residues shown in the consensus line of Figure 1 are specifically conserved in the SARAH family. Among them are polar residues found in ‘a’and ‘d’positions of the coiled-coil