When is a lipid kinase not a lipid kinase? When it is a protein kinase

About 10 years ago there was a heated debate over whether v-Src and several other retrovirally encoded protein-tyrosine kinases had lipid kinase activity in addition to protein kinase activity. Ultimately, the analysis of the unusual lipid kinase activity associated with v-Src led to the discovery of a novel type of phosphatidylinositol (PI) kinase, the PI 3-kinases. This in turn led to the characterization of a type of PI 3-kinase that associates with phosphotyrosine-containing proteins via a regulatory subunit, ~85, that has two SH2 domains. The catalytic domain of this PI 3-kinase, ~110, proved to contain a C-terminal region distantly related to the catalytic domain of the protein kinase superfamily and containing motifs conserved in subdomains VI6 (motif I) and VII (motif II) of the protein kinase catalytic domain (Hiles et al., 1992) (Figure 1). ~110 has PI 3-kinase activity in the absence of ~85, and this activity is abolished when Arg-916 in one of the conserved protein kinase motifs is mutated to Pro. This shows that the PI 3-kinase activity is intrinsic and that the protein kinase-like region is required for this activity, although the boundaries of the minimal catalytic domain have not been precisely defined (Dhand et al., 1994). Since that time, a large family of proteins related to the catalytic domain of the ~110 PI 3-kinase subunit has been uncovered (Table 1). Based on the demonstrable PI kinase activities of pl 1 Oa, ~1106, Vps34p, and the PI 4-kinases, one might have predicted that all of these proteins would have PI kinase activity. Ironically, however, we have now come full circle, and it appears that a subset of the PI 3-kinase family members have protein kinase activity. Some PI 3.Kinase Family Members Are Bona Fide Lipid Kinases pl 1 Oa, ~1108, and pl 1 Oy, Vps34p and its human homolog, and the PI 4-kinases have all been shown to have intrinsic PI kinase activity, although they have distinct substrate specificities; pl lOa, ~1108, and plOOy can phosphorylate the 3’position on the inositol ring of PI and Pl(4)P, but preferentially phosphorylate Pl(4,5)P2, whereas the yeast and human VPS34 PI 3-kinases and the PI 4-kinases can only phosphorylate PI. (Note that the Pl(4)P 5kinases form a separate, unrelated family [Boronenkov and Anderson, 19951). In contrast, there is little evidence that other members of the family have PI kinase activity. FRAP (Sabatini et al., 1995) and ToRp (Cardenas and Heitman, 1995) have been shown to possess associated PI 4-kinase activity when isolated by immunoprecipitation. In neither case, however, is the PI 4-kinase activity inhibited by rapamycin, even though FRAP and ToRp bind rapamycinl FKBP and have been identified as targets for inhibition by rapamycin, a macrolide that inhibits the growth of many cells. When a temperature-sensitive mutant of ToRp was examined, TorPp-associated PI 4-kinase activity was 1 Ofold lower in cells grown at the restrictive temperature, but the high level of activity associated with the mutant ToRp at the permissive temperature was not temperature sensitive in vitro (Cardenas and Heitman, 1995). Unfortunately, neither study included the critical control of testing a protein mutated in one of the conserved kinase motifs. This is important, because type II PI 4-kinase is activated by nonionic detergents, such as those used in cell lysis, and has a tendency to bind nonspecifically to immunoprecipitated proteins (Whitman et al., 1987). Indeed, in a second study of FRAP, the associated PI 4-kinase activity was found to be unaffected by mutation of either Asp-2338 or Asp-2357, which are, respectively, in motifs I and II in the PI 3-kinase domain, indicating that the PI 4-kinase activity is associated rather than intrinsic (Brown et al., 1995). It is possible that the conditions needed to reveal an intrinsic PI kinase activity have not been found, but this suggests that the PI 3-kinase domain might have another activity. Some PI 3-Kinase Family Members Have Protein Kinase Activity The first hints that a PI 3-kinase-related protein might have protein kinase activity came from a study of the pl 1 O/p85 complex itself, in which it was shown that ~110 has an associated Mn2+-dependent protein kinase activity that can phosphorylate a specific serine, Ser-608, in the ~85 subunit, resulting in inhibition of PI 3-kinase activity (Carpenter et al., 1993; Dhand et al., 1994). Mutation of Arg916 in motif I in the PI 3-kinase domain not only eliminates PI 3-kinase activity but also abolishes the protein-serine kinase activity, establishing that this activity is intrinsic (Dhand et al., 1994). This phosphotransfer usually occurs within the ~1101~85 heterodimer (Dhand et al., 1994) but under some conditions, phosphorylation of exogenous