The yeast TOR1 and TOR2 proteins were first identified as the targets of the immunosuppressive natural product rapamycin bound to the cellular protein FKBP12. Subsequently, a mammalian FKBP12-rapamycin–binding protein was identified that shares ~50% sequence identity with the yeast TOR proteins. Two C-terminal domains that are highly conserved between yeast and mammalian TOR have been identified. The first, the FKBP12-rapamycin–binding or FRB domain, was identified as the target for rapamycin binding to both yeast and mammalian TOR, based on both biochemical and genetic approaches (Chen et al., 1995
; Lorenz and Heitman, 1995
), and the x-ray crystal structure of the FKBP12–rapamycin–FRB domain ternary complex has been solved (Choi et al., 1996
). The second domain is the C-terminal kinase domain that also shares identity with other members of the PI kinase-related kinase superfamily, including PI-3 and PI-4 kinases from yeast and man, DNA-dependent protein kinase, and the yeast and human checkpoint control proteins MEC1, TEL1, ATM, and ATR.
We have taken biochemical and genetic approaches to determine the catalytic activity and to define functional domains in the yeast TOR proteins. Our studies show that TOR1 possess a robust protein kinase activity capable of phosphorylating PHAS-I, the only known substrate of the mammalian TOR kinase homologue. Although we do not detect autophosphorylation activity, the TOR1 kinase activity shares several other hallmarks exhibited by the mammalian TOR kinase in that the TOR1 kinase activity is enhanced by the presence of MnCl2
ions and is inhibited by FKBP12–rapamycin and wortmannin (Brunn et al., 1996
). TOR1 kinase activity was also abolished by the D2275A-active site mutation and rendered resistant to rapamycin by a mutation in the FRB domain (S1972I) that interferes with FKBP12–rapamycin binding. In contrast to the yeast TOR2 and mammalian TOR homologues, FRAP and RAFT1, we found no PI-4 kinase activity associated with the yeast TOR1 enzyme. Taken together, our findings reveal that yeast TOR1, like the mammalian mTOR homologue, has protein kinase activity.
We find that the integrity of the TOR1 kinase domain is required for TOR1 functions in yeast, and for a dominant rapamycin-resistant TOR1 mutant to confer rapamycin resistance. In addition, we find that overexpression of a kinase-inactive TOR1 mutant is toxic to yeast cells. These findings are in accord with observations previously reported by Zheng et al. (1995)
. The toxicity of the overexpressed kinase-inactive TOR1 mutant could be overcome by overexpressing wild-type TOR1, illustrating that this is a specific toxic effect resulting from interference with TOR function in vivo. Toxicity of the TOR1 kinase-inactive mutant was not inhibited by FKBP12–rapamycin, indicating that the FRB domain is not responsible for this toxic effect. Finally, in contrast to the previous report of Zheng et al. (1995)
, we did not observe any toxic effect upon overexpression of two different TOR1 rapamycin-resistant mutants bearing single amino substitutions in the FRB domain. Our sequence analysis of the region rescued by gap repair in the S1972I or S1972R mutants revealed no extraneous mutations. Thus, either subtle differences in experimental conditions or additional mutations in the studies of Zheng et al.
must explain this discrepancy.
We proceeded to map regions of the TOR1 protein that are required for the toxic effect of the TOR1 kinase-inactive mutant. To our surprise, simply deleting the entire kinase domain, the FRB domain, or both the kinase domain and the FRB domain, resulted in C-terminally or internally truncated forms of the TOR1 protein that were also toxic when overexpressed. Thus, inactivation of the kinase activity of TOR1 by either a point mutation or a large deletion resulted in a toxic TOR1 protein, suggesting that regions of the protein other than the kinase domain itself are responsible for the toxicity of the TOR1 kinase-inactive mutant.
To map regions of the TOR1 protein responsible for this toxic effect, we took two approaches. First, a large central region of TOR1 was deleted, after which progressively larger and larger portions of the missing section were replaced. This analysis revealed that readdition of residues 290-1207 was not sufficient to render that attached kinase-inactive domain toxic, but that readdition of residues 290-1682 was sufficient to restore toxicity, implicating residues 1207–1682 in the toxic effect. In the second approach, short defined central regions of the TOR1 protein were overexpressed and tested for toxic effects. By this approach, a 567-amino acid segment of TOR1 from residue 1207 to 1774 was found to be partially toxic, and addition of another 200 C-terminal residues rendered this domain fully toxic (754 amino acids, 1207–1961). These studies therefore defined a central domain of TOR1 that, when overexpressed on its own, was toxic to the cell. Importantly, this domain is amino-terminal to, and completely distinct from, the FRB and kinase domains that have been previously defined. The same domain derived from TOR2 was also toxic when overexpressed in yeast. We term this novel TOR domain, common to the yeast TOR1 and TOR2 proteins, the toxic central effector domain.
What might be the function of this TOR effector domain? One plausible explanation for the dominant negative toxic effect observed upon overexpression of the TOR effector domain is that this domain is involved in mediating dimerization between TOR1 and itself, or between TOR1 and TOR2. However, we have been unable to find any evidence of coprecipitation between the overexpressed toxic TOR1 domain (detected with the fused HA epitope) and endogenous yeast TOR1 or TOR2 (detected with polyclonal sera against TOR1 or TOR2) (our unpublished results). We therefore propose that the TOR effector domain interacts with substrates or regulators of the TOR-dependent, rapamycin-sensitive signaling cascade. For example, the effector domain might dock substrates onto TOR for subsequent phosphorylation by the adjacent kinase domain. In this model, the TOR proteins could serve a function as a scaffold upon which to assemble other interacting proteins for appropriate interaction with the TOR kinase domain. An alternative possibility is that this domain mediates localization to membranes and competes with the endogenous TOR1 for localization.
What role does the toxic effector domain play in known TOR functions in yeast? The TOR proteins have two defined functions in yeast. One TOR function is essential and unique to TOR2 and involves polarization of the yeast actin cytoskeleton though an effect of TOR2 on the Rho-like GTPases, RHO1 and RHO2, via the exchange factor ROM2 (Schmidt et al., 1996
). The second TOR function is also essential but shared by both TOR1 and TOR2 and involves regulation of translational initiation (Barbet et al., 1996
). Thus far, no direct target of the TOR1 or TOR2 protein involved in either function in yeast has been identified, and the only protein implicated as a direct target of the mammalian TOR protein is PHAS-I, but the domains with which PHAS-I interacts have not yet been identified. Because we find that the toxic effector domain is common to both TOR1 and TOR2, and toxicity of this domain is overcome by TOR1 overexpression (which does not provide the TOR2 unique function), these findings implicate the toxic effector domain in the translational regulatory function common to both TOR1 and TOR2. Consistent with this interpretation, overexpression of the most potent TOR1-toxic domain yields a G1
cell cycle arrest that is indistinguishable from rapamycin inhibition of TOR function or genetic depletion of yeast TOR1 and TOR2 (Table ). Shorter, less potent forms of the TOR-toxic domain yield a G1
arrest, but the cells are smaller than those observed with rapamycin, TOR depletion, or the larger TOR-toxic domain (Table ). One plausible hypothesis is that these forms of the toxic domain yield a more transient cell cycle arrest from which cells escape, reenter the cell cycle, and hence do not form the large G1
arrested cells observed with rapamycin exposure. Taken together, our observations support the hypothesis that overexpression of the TOR-toxic domain interferes with the shared TOR function in the regulation of translation.
The direct physical targets of the TOR-toxic domain remain to be identified. We have thus far been unable to identify overexpression suppressors of the TOR1-toxic domain from screens of either 2μ genomic libraries or a GAL-regulated cDNA library. In addition, we have not isolated any proteins interacting with the yeast TOR1-toxic domain using the two-hybrid system. Thus, the target of the toxic domain may have multiple subunits or may not be a protein and could be an RNA or small molecule. The yeast phospholipase C gene PLC1
was identified as a multicopy suppressor of growth inhibition by the TOR-toxic domain. Interestingly, the yeast PLC1
gene was also recently identified as a multicopy suppressor of conditional lethal TOR2
mutations in yeast (Helliwell et al., 1998
). The effects of PLC1 may be indirect, as we have failed to observe a direct physical interaction between the TOR-toxic domain and PLC1 in both the two-hybrid system or by coimmunoprecipation (our unpublished results). The yeast PLC1 enzyme, like its mammalian counterpart, is involved in phosphatidylinositol metabolic cascades and cleaves PI-4P and PI-4,5P2
to yield the soluble second messengers inositol diphosphate and inositol triphosphate. Although the functions and targets of these cascades remain to be fully elucidated, yeast plc1
mutants are viable but exhibit a number of phenotypes, including poor growth, temperature-sensitive growth in some strain backgrounds, sensitivity to osmotic stress, and an inability to utilize many carbon sources (Flick and Thorner, 1993
). Given the homology of the TOR kinase domains with lipid kinases, previous studies that the TOR proteins are associated with a lipid kinase in yeast and mammalian cells (Brown et al., 1995
; Cardenas and Heitman, 1995
; Sabatini et al., 1995
), the fact that one downstream element of the TOR-regulated cascade contains an essential pleckstrin homology domain (the Rho1 GTP exchange factor Rom1, which could be a target of PI-4,5 P2
) (Schmidt et al., 1997
), and recent studies that reveal the growth factor-activated PI-3 lipid kinase may function upstream of mTOR in mammalian cells (Gingras et al., 1998
), our findings provide evidence for another interesting link between TOR function and PI metabolic cascades in the regulation of cell function. Further studies will be required to determine whether the yeast TOR proteins and PLC1 function in the same or related signaling cascades.
Similar structure–function approaches have been recently applied to the mammalian ATM and ATR proteins, which share identity with the C-terminal TOR kinase domain. Overexpression of fragments of the ATM protein containing a leucine zipper motif have a dominant negative activity in cultured mammalian cells (Morgan et al., 1997
). This region of ATM may either inhibit endogenous ATM function or bind to and inhibit other members of the ATM signaling pathway. Interestingly, overexpression of the ATM kinase domain alone was sufficient to complement many of the defects in atm mutant fibroblasts, including radiosensitivity and S phase checkpoint function, indicating that the kinase domain is responsible for much of the activity of ATM (Morgan et al., 1997
). In related studies, Cliby et al. (1998)
demonstrated that overexpression of a kinase-inactive mutant of the mammalian ATR protein caused sensitivity to ionizing radiation, methyl methanesulfonate, and cis
-platinum and also abolished the G2/M checkpoint after DNA damage with ionizing radiation. Taken together, these studies implicate the kinase domains of ATM and ATR in mediating cellular responses to DNA damage and cell cycle progression.
The TOR-toxic domain is conserved between the yeast and mammalian TOR proteins and, in BLAST searches, shares more limited identity over an ~240- amino acid region with the PIK family members ATR, RAD3, mei-41, and ATM (Figure ). Yeast TOR1 and mammalian ATR share 26% overall identity, and 48% similarity, in this region. This finding suggests that the TOR-toxic domain, like the more C-terminal kinase domain, has been conserved between different PIK family members and might play a role in the signaling functions of these different proteins. In general, members of the PIK-related family have been implicated in signaling cascades that regulate responses of the cell to exogenous signals (TOR, mTOR, PI-3 kinase) or endogenous signals (ATR, ATM, Rad3, Mec1, Tel1). This conserved region may interact with effectors or regulators of the TOR- and ATR-signaling pathways. Further studies will be required to address the functions of this domain in other PIK family members.
Figure 9 The TOR toxic domain is conserved and shares identity with regions of the PIK family members, ATR, Rad3, mei-41, and ATM. The TOR1 (amino acids 1376–1614), TOR2 (amino acids 1383–1620), and mTOR/RAFT/FRAP-toxic domains (amino acids 1427–1663) (more ...)