Tor, a Phosphatidylinositol Kinase Homologue, Controls Autophagy in Yeast*

Autophagy is a bulk protein degradation process that is induced by starvation. The control mechanism for induction of autophagy is not well understood. We found that Tor, a phosphatidylinositol kinase homologue, is involved in the control of autophagy in the yeast, Saccharomyces cerevisiae. When rapamycin, an inhibitor of Tor function, is added, autophagy is induced even in cells growing in nutrient-rich medium. A temperature-sensitive tor mutant also leads to induction of autophagy at a nonpermissive temperature. These results indicate that Tor negatively regulates the induction of autophagy. Tor is the first molecule that is identified as a pivotal player in the starvation-signaling pathway of autophagy. Furthermore, we found that a high concentration of cAMP is inhibitory for induction of autophagy. APG gene products are involved in autophagy induced by starvation. Autophagy was not induced in apg mutants in the presence of rapamycin, indicating that the site of action of Tor is upstream of those of Apg proteins. In nutrient-rich medium, Apg proteins are involved also in the transport of aminopeptidase I from the cytosol to the vacuole. Tor may act to switch Apg function between autophagy and transport of aminopeptidase I.

Autophagy is a bulk protein degradation process that is induced by starvation. The control mechanism for induction of autophagy is not well understood. We found that Tor, a phosphatidylinositol kinase homologue, is involved in the control of autophagy in the yeast, Saccharomyces cerevisiae. When rapamycin, an inhibitor of Tor function, is added, autophagy is induced even in cells growing in nutrient-rich medium. A temperaturesensitive tor mutant also leads to induction of autophagy at a nonpermissive temperature. These results indicate that Tor negatively regulates the induction of autophagy. Tor is the first molecule that is identified as a pivotal player in the starvation-signaling pathway of autophagy. Furthermore, we found that a high concentration of cAMP is inhibitory for induction of autophagy. APG gene products are involved in autophagy induced by starvation. Autophagy was not induced in apg mutants in the presence of rapamycin, indicating that the site of action of Tor is upstream of those of Apg proteins. In nutrient-rich medium, Apg proteins are involved also in the transport of aminopeptidase I from the cytosol to the vacuole. Tor may act to switch Apg function between autophagy and transport of aminopeptidase I.
When nutrients are depleted from the environment, mammalian as well as yeast cells begin to degrade their own cytosol and organelles. This bulk protein degradation is mediated mainly by autophagy, the process in which cytoplasmic components are delivered to the lysosome (1). In the process of autophagy, a portion of cytoplasm is sequestered by a double or multi-layered membrane structure, referred to as the autophagosome (1). The autophagosome fuses with endosomal or lysosomal membranes, resulting eventually in degradation of cytoplasmic materials (1,2).
To initiate this dynamic reorganization of intracellular membranes, as occurs during formation of autophagosome, cells are expected to have a regulatory mechanism of autophagy. Although the overall mechanism remains to be elucidated, some factors have been reported to be involved in this control in mammalian cells. For example, formation of autophagosomes is inhibited by amino acids (3,4), by insulin (5), by cell swelling (6), and by cAMP (7). Blommaart et al. reported that the phosphorylation of ribosomal protein S6 and autophagy are controlled by the same signal transduction pathway (8). S6 phosphorylation is mediated by the serine/threonine kinase p70S6k (9).
In the yeast Saccharomyces cerevisiae, autophagy is induced by starvation (10). The sequence of events in yeast is essentially the same as that of mammalian cells: formation of an autophagosome is followed by fusion of the autophagosome to the vacuole, which is equivalent to a lysosome in mammalian cells (11). The inner membrane structures of the autophagosome are delivered to the vacuole as the autophagic bodies (11). The autophagic bodies are destroyed depending on proteinase B activity in the vacuole (10). However, the induction mechanism of autophagy in yeast is not well understood.
Nutrient starvation not only induces autophagy but also arrests cells at early G 1 phase. Tor, a phosphatidylinositol kinase-related kinase, is known to be involved in the signaling pathway from nutrient starvation to G 1 arrest in yeast (12)(13)(14)(15). The immunosuppressive drug rapamycin is effective in arresting yeast at early G 1 phase (12). The effect of rapamycin is mediated by FKBP, and it is the FKBP-rapamycin complex that binds to Tor (16,17). There are two Tors in yeast, and inactivation of both Tors leads to G 1 arrest (15). In addition to inhibiting G 1 progression, Tor2p has another essential function that appears to be involved in the organization of the actin cytoskelton (13,14,18). Tor1p does not have this function, which is not inhibited by rapamycin (19). To investigate the control mechanism of autophagy in yeast, we determined the effects of inactivating Tor and of a high concentration of cAMP.
Yeast Culture-Yeast cells were grown in YPD (1% yeast extract, 2% polypeptone, 2% glucose). For the nitrogen starvation, the cells in log phase were washed with water and suspended in SD(ϪN) medium (0.17% yeast nitrogen base without amino acids and ammonium sulfate (Difco) and 2% glucose). Rapamycin (Sigma) was added at a final concentration of 0.2 g/ml using a stock solution of 20 g/ml in 90% ethanol and 10% Triton X-100. Hydroxyurea (Sigma) was added to 0.2 M, and nocodazole (Sigma) was added to 10 g/ml using stock solution of 10 mg/ml in dimethyl sulfoxide. cAMP (Sigma) was added at a final concentration of 1 mM.
Flow Cytometry-Cultures (0.3 ml) of yeast in YPD medium (A 600 ϭ 0.5) were added to 0.7 ml of ethanol. The samples were incubated overnight at Ϫ20°C, washed, then resuspended, and sonicated for 30 s in 200 mM Tris-Cl (pH 7.5), 50 mM EDTA. The samples were then * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. incubated with RNase (0.2 mg/ml) for 4 h at 37°C. DNA was stained by 0.2 mg/ml propiodium iodide in 200 mM Tris-Cl (pH 7.5), 50 mM EDTA for 1 h at 4°C. 10,000 events were analyzed for DNA content using Coulter Epics XL.
Alkaline Phosphatase Assay-The assay method was basically that of Nothwehr et al. (23). Cells (1-2 ϫ 10 7) were harvested and washed with water. The cells were then suspended in 0.2 ml of assay buffer (250 mM Tris-HCl (pH 9.0), 10 mM MgSO 4 , 10 M ZnSO 4 ) and disrupted by vortexing with glass beads. After centrifugation, 50 l of the supernatant was added to 0.5 ml of assay buffer and 50 l of 55 mM ␣-naphtylphosphate to start the reaction. After incubation at 30 min for 20°C, 0.5 ml of 2 M glycine-NaOH (pH 11.0) was added to stop the reaction. Fluorescence intensity of emission at 472 nm after excitation at 345 nm was measured. One unit is defined as the activity to release 1 mol ␣-naphtol/min/mg protein. Protein concentration was determined by the BCA method (Pierce) or the Bradford method (24).

RESULTS
Autophagy in yeast is induced under starvation conditions (10) and can be followed as an accumulation of autophagic bodies in the vacuole when vacuolar degradation is inhibited by inactivation of proteinase B (Fig. 1A). In nutrient-rich YPD medium, autophagy usually does not occur (Fig. 1B). However, when rapamycin was added to cells growing in YPD, it was found that autophagic bodies accumulate in the vacuole, indicating that autophagy was induced (Fig. 1C). This morphological observation was confirmed by use of the biochemical assay for autophagy developed previously (21). The assay is based on the fact that the vacuolar membrane protein alkaline phosphatase, Pho8p, becomes localized in the cytoplasm if its transmembrane region is truncated (21). When autophagy occurs, this truncated Pho8p, Pho8⌬60p, is delivered to the vacuole where it becomes active as a result of processing by vacuolar proteases (21). Therefore, the measurement of phosphatase activity provides a simple assay for autophagic activity (21). When rapamycin was added to YPD culture of cells harboring pho8⌬60, alkaline phosphatase activity increased, implying that autophagy had been induced (Fig. 2, lane 3).
Rapamycin is known to bind to the FK506 binding protein, FKBP, to effect cell cycle arrest in yeast (12). We investigated whether FKBP is involved in this induction of autophagy by rapamycin. Disruption of FKB1, the gene for FKBP, resulted in the failure of rapamycin to induce autophagy (Fig. 2, compare  lane 3 and 5). This result indicates that the action of rapamycin in inducing autophagy is mediated by FKBP.
The FKBP-rapamycin complex inhibits Tor function (16,17). Inactivation of the two TOR genes leads to G 1 arrest (15). The Pho8⌬60 gene was introduced into cells in which both TOR1 and TOR2 were deleted and harbored temperature-sensitive allele of TOR2 in a plasmid. This strain grows at 30°C but not at 37°C (15). 1 Alkaline phosphatase activity was increased when tor2 ts cells harboring Pho8⌬60 were incubated in YPD medium for 3 h at 37°C, whereas it was not increased at 30°C in the same cells or in the wild-type TOR cells at 37°C (Fig. 2,  lanes 6 -8). Based on these results, we concluded that inactivation of Tor function is sufficient for induction of autophagy.
It has been reported that treatment with rapamycin or inactivation of Tors leads to cell arrest at early G 1 phase (15). This raises the possibility that autophagy is induced when the cell cycle is arrested at any phase. To investigate this possibility, we treated cells with hydroxyurea to arrest the cell cycle at S phase or with nocodazole to arrest the cell cycle at M phase. After 3 h of treatment with these reagents, flow cytometry analysis confirmed that most cells were arrested at S or M phase. After treatment with hydroxyurea, 83% of the are with 1 N DNA as compared with 56% in control cells. After treatment with nocodazole, 76% of the cells are with 2 N DNA as compared 1 M. Hall, personal communication.  with 44% in control cells. However, in those cell cycle arrested cells, autophagy was not induced (Fig. 3). Moreover, when rapamycin was added to the cells arrested at S or M phase, autophagy was still induced, indicating that cell cycle arrest at G 1 is not prerequisite for induction of autophagy (Fig. 3).
In our laboratory, 14 autophagy defective mutants, called apg mutants, have been isolated (22). Most of the APG genes have been cloned and sequenced; however, their sites of action in the autophagic process are still to be determined (25)(26)(27). We have investigated the effect of rapamycin on apg mutants by observing the accumulation of autophagic bodies in the presence of rapamycin. It was found that none of the apg mutants accumulated autophagic bodies, implying that autophagy was not induced (panels D-F in Fig. 1 show the representative three apg mutants, but the other apg mutants were not shown). Therefore, the sites of action of all APG gene products are presumed to be downstream of the site of action of Tor.
Next we investigated the effect of cAMP, which is thought to play some role in cell cycle control by nutrient availability (28). It has been reported that cells harboring a triple mutation in cAMP pathway (pde1, pde2, and cyr1) are able to respond to exogenously added 1 mM cAMP, because phosphodiesterases (Pde1p, Pde2p), which are thought to degrade exogenously added cAMP, are defective (20). This strain grown in the presence of cAMP showed the same phenotypes as mutants with an elevated activity of the cAMP pathway, such as low glycogen accumulation (20). In pho8⌬60 cells having these mutations as well as in wild-type cells, nitrogen starvation induced autophagy (Fig. 4). When cAMP was added to the medium, the induction of autophagy in these cells was inhibited (Fig. 4). Furthermore, this inhibition was not overcome by rapamycin (Fig. 4). These results indicated that 1 mM cAMP is inhibitory for the induction of autophagy. DISCUSSION In this study, we showed that Tor function is involved in the induction of autophagy in yeast. Inactivation of Tor by rapamycin treatment or by using a temperature-sensitive allele of TOR2 promoted autophagy even under nutrient-rich condi-tions. Although the kinase domain of Tor was shown to be necessary for G 1 cell cycle progression (19), it remains to be determined whether Tor possesses actual kinase activity. Therefore, the mechanism through which Tor transmits signals is still unknown. However, from our results, it would be reasonable to speculate that under nutrient-rich conditions, Tor functions by blocking a factor needed for initiation of autophagy. This model would be consistent with what is known in mammalian cells, where the condition for phosphorylation of the ribosomal protein S6 is inhibitory for autophagy (8). S6 phosphorylation is sensitive to rapamycin, suggesting that a rapamycin-sensitive factor is involved upstream of S6 phosphorylation and autophagy (8,29). We assume that this factor is RAFT/FRAP, the mammalian counterpart of Tor, which mediates S6 kinase activity (29 -32).
Baedet et al. reported that inactivation of Tor leads to several G 0 /stationary phase responses, which are also produced by nitrogen starvation (15). We showed that cell cycle arrest at S or M phase does not induce autophagy, implying that cell cycle arrest by itself is not the cause of induction of autophagy. Morover, even after the cell cycle is arrested at S or M phase, autophagy was induced by rapamycin treatment. This means that G 1 arrest itself is not the cause of autophagy, and autophagy and G 1 arrest are induced in a parallel manner under starvation conditions by the inactivation of Tor. Baedet et al. reported that the loss of Tor function also causes an inhibition of translation and proposed that the inhibition of protein synthesis causes G 1 arrest (15). We observed that the cells treated with cycloheximide in YPD did not induce autophagy (data not shown). Moreover, we have found that treatment with cycloheximide is inhibitory for induction of autophagy under starvation conditions (10). Therefore, additional experiments are needed to investigate the role of mRNA translation in autophagy.
There is a report that the two Tor proteins are involved in the initiation of sporulation (33). Sporulation is usually induced only in starvation medium, but inhibition of Tor function by rapamycin or by expressing a kinase-inactive mutant leads to initiation of sporulation even in nutrient-rich medium (33). Autophagy-defective mutants do not sporulate (22) that autophagy is prerequisite for sporulation. So our finding that Tors are involved in control of autophagy is consistent with the report (33) on the involvement of Tor in initiation of sporulation. Tor2p has been shown to signal to the actin cytoskelton via the small GTPase Rho (18,34). Because this function is specific to Tor2p and insensitive to rapamycin (19), we assume that it is not involved in autophagy.
The Ras-cAMP pathway is involved both in G 1 progression and initiation of sporulation, although the cAMP level does not change with nitrogen source availability (28,35). Our finding indicates that increasing the concentration of cAMP inhibits autophagy even in the presence of rapamycin. The inhibitory effect of cAMP is observed also in mammalian cells (7). Baedet et al. reported that G 1 arrest by rapamycin is not affected by hyperactivation of the Ras-cAMP pathway using bcy1 or ras2 mutants, suggesting that the Ras-cAMP pathway is independent of the Tor pathway to G 1 arrest (15). Thus, unlike the case of G 1 arrest, cAMP may have to be kept below a certain critical level to initiate autophagy. Too high a concentration of cAMP may perturb some factors involved in the induction of autophagy.
APG and other autophagy-relative genes were reported to function not only in autophagy but also in the transport of aminopeptidase I (API) 2 to the vacuole (36,37). Unlike most of the vacuolar enzymes, API is delivered from cytosol to the vacuole via nonsecretory and vesicle-mediated pathway (38,39). Although both API transport and autophagy involve cytoplasm to vacuole delivery, API transport has different characteristics from autophagy. For example, API is delivered even under nutrient-rich condition, and its transport is highly selective. In contrast, autophagy is induced only under starvation, and autophagic delivery of the cytosol is nonselective (36). Therefore a factor that switches between autophagy and API transport might exist. Our analysis with apg mutants indicated that all APG gene products function downstream of Tors. Tor might be a molecule that switches Apg function directly or indirectly between autophagy and API transport.
Tor is the first molecule identified in the signal pathway in autophagy. We have found significant similarities in the control mechanism for autophagy of mammalian cells and yeast. Further analysis of autophagy in yeast would help our understanding of the process in mammalian cells.