Abstract

The Tor kinases regulate responses to nutrients and control cell growth. Unlike most organisms that only contain one Tor protein, Saccharomyces cerevisiae expresses two, Tor1 and Tor2, which are thought to share all of the rapamycin-sensitive functions attributable to Tor signaling. Here we conducted a genetic screen that defined the global TOR1 synthetic fitness or lethal interaction gene network. This screen identified mutations in distinctive functional categories that impaired vacuolar function, including components of the EGO/Gse and PAS complexes that reduce fitness. In addition, tor1 is lethal in combination with mutations in class C Vps complex components. We find that Tor1 does not regulate the known function of the class C Vps complex in protein sorting. Instead class C vps mutants fail to recover from rapamycin-induced growth arrest or to survive nitrogen starvation and have low levels of amino acids. Remarkably, addition of glutamate or glutamine restores viability to a tor1 pep3 mutant strain. We conclude that Tor1 is more effective than Tor2 at providing rapamycin-sensitive Tor signaling under conditions of amino acid limitation, and that an intact class C Vps complex is required to mediate intracellular amino acid homeostasis for efficient Tor signaling.

THE Tor kinases are key components of an evolutionarily conserved nutrient-responsive pathway that regulates cell growth and proliferation in eukaryotic organisms. The Tor kinases were first identified in yeast cells as the targets of the antiproliferative drug rapamycin (Heitman et al. 1991). Thereafter, rapamycin has been instrumental in elucidating biological events governed by Tor signaling, including complex transcriptional and translational programs (reviewed in Rohde et al. 2001; Crespo and Hall 2002).

When yeast cells are grown in ample nutrient conditions, Tor activity promotes the expression of genes encoding tRNAs, ribosomal proteins, and rRNA, while inactivating genes required for utilization of poor nitrogen and carbon sources, and stress responses (Zaragoza et al. 1998; Beck and Hall 1999; Cardenas et al. 1999; Hardwick et al. 1999; Powers and Walter 1999; Komeili et al. 2000). Tor activity also supports translation, in large part by suppressing the general amino acid control response regulated by the Gcn2 kinase, and possibly by also affecting the stability of eIF4G (Berset et al. 1998; Valenzuela et al. 2001; Cherkasova and Hinnebusch 2003; Kubota et al. 2003; Rohde et al. 2004). Inhibition of Tor by rapamycin elicits many of the cellular responses that are triggered by starvation for nutrients, such as inhibition of protein synthesis, downregulation of amino acid permeases, protein degradation, autophagy, and cell cycle arrest (reviewed by Rohde et al. 2001; De Virgilio and Loewith 2006).

Unlike most organisms that express only one Tor protein, Saccharomyces cerevisiae has two highly homologous Tor proteins, Tor1 and Tor2, which are thought to share all of the rapamycin-sensitive functions attributable to Tor signaling, while only Tor2 serves a unique and essential rapamycin-insensitive role (reviewed in Crespo and Hall 2002). A recent study has suggested that Tor1 and Tor2 differ in providing a rapamycin-sensitive function in certain class C vps mutants (Xie et al. 2005); however, the exact nature of this function remains to be determined. The Tor proteins form two distinct multiprotein complexes: TORC1 and TORC2. Tor1 (and to a lesser extent Tor2) is a component of the TORC1 complex, which includes Lst8, Kog1, Tco89, and Bit61. TORC2 consists of Tor2, Lst8, and the Avo1, Avo2, and Avo3 proteins (Loewith et al. 2002; Wedaman et al. 2003; Reinke et al. 2004). FKBP12-rapamycin physically associates with TORC1 but not with TORC2, suggesting that this second complex mediates the rapamycin-insensitive Tor2 role in controlling polarization of the actin cytoskeleton (Loewith et al. 2002).

To understand why yeast cells express two functional Tor proteins, we sought to define novel Tor1- or Tor2-specific functions. Here we performed a genomewide screen searching for genes that when mutated in combination with tor1 reduce fitness or render cells inviable. These genes identified distinct functional networks including those involved in protein sorting, vacuolar inheritance, and microautophagy. In particular, we find that tor1 shows synthetic lethality or synthetic reduced fitness with mutations in different components of the class C Vps complex, which includes the Pep3, Pep5, Vps16, Vps33, Vps39, and Vps41 proteins (Banta et al. 1988; Raymond et al. 1992; Rieder and Emr 1997; Wurmser et al. 2000). The class C Vps complex plays a central role in protein sorting by regulating vesicle docking and fusion at the endosome and between the endosome and the vacuole (Srivastava et al. 2000; Peterson and Emr 2001). Viability of the tor1 pep3 double mutant was restored by expression of Tor1 but not Tor2, indicating that the function linking Tor1 and the class C Vps complex is unique to the rapamycin-sensitive TORC1 complex. Mutants lacking components of the class C Vps complex fail to recover from rapamycin-induced growth arrest and to survive nitrogen starvation, have low levels of amino acids, in particular glutamate, and show growth defects at 37° (this study; Kitamoto et al. 1988; Robinson et al. 1991). Remarkably, addition of glutamate or glutamine rescues the growth defect of class C single vps mutants and of a tor1 pep3ts conditional mutant at 37°. Our studies suggest that, in contrast to Tor2, Tor1 is specialized to support growth under conditions where intracellular amino acid concentrations are drastically reduced. We also conclude that an intact class C Vps complex is required to provide intracellular amino acid homeostasis for proper Tor1 signaling. These findings provide a physiological foundation to understand the duplication and divergence of Tor1 and Tor2 functions in S. cerevisiae.

MATERIALS AND METHODS

Yeast strains and media:

Strains used in this study are listed in Table 1. All the strains are isogenic derivatives of BY4741 or BY4742 and unless otherwise indicated were constructed by the Saccharomyces Genome Deletion Project (distributed by Invitrogen, Carlsbad, CA). Strains SZY25 and SZY21 were obtained by deletion of TOR1 with URA3 in strains BY4741 and 13652, respectively. Strains SZY26, 27, 28, 29, 30, 31, 32, and 33 as well as RPY50 and RPY51 were created by crossing strain SZY25 to the pep3, pep5, vps15, vps16, vps34, vac7, vac8, vac17, vps39, and vps41 MATα haploid strains, respectively. Strain SZY36 was obtained from strain BY4742 by deletion of the VPS33 gene with KanMX4. Strain SZY37 was constructed by replacing the URA3 gene in strain SZY25 with LEU2. Strain SZY40 was obtained by crossing strain SZY37 to the pep3 haploid mutant strain #14105 and the resulting diploid transformed with plasmid pBJ9113 expressing the pep3ts-108 (Srivastava et al. 2000) was sporulated and dissected to obtain strain SZY40-4. SZY43 is a meiotic segregant of strain SZY32. Strain SZY49 was constructed by crossing strains SZY25 and SZY36.

View this table:
TABLE 1

Yeast strains used in this study

Yeast synthetic medium (YNB) with ammonium was always supplemented with 2% glucose (SG). SG medium was supplemented with amino acids to satisfy any auxotrophic requirements (SC) and for the experiment presented in Figure 6C with 0.2% of the indicated amino acid. Sporulation medium was 1.5% potassium acetate (KAc) (pH 7.5), supplemented with any required amino acids. SLAD, YPD, and all other media were prepared as described previously (Gimeno et al. 1992; Sherman 2002). Rapamycin was added to the media from concentrated stock solutions in 90% ethanol, 10% Tween-20. Yeast transformations were performed by the lithium acetate method (Schiestl and Gietz 1989). Unless noted otherwise, mutant yeast strains were constructed by PCR-mediated gene disruption, replacing the entire open reading frame of the targeted gene with the indicated genes (Longtine et al. 1998; Goldstein and McCusker 1999). All gene deletions were confirmed by PCR.

Plasmids:

Low-copy centromeric plasmids, pRS315-TOR1 expressing TOR1, and pML40-3 expressing TOR2, as well as plasmids expressing Tor2-Tor1 hybrid proteins were described previously (Lorenz and Heitman 1995; Alarcon et al. 1996).

Plasmid pSZ12 (Hybrid 4) was created by gap repair; briefly, a 173-bp (from nucleotide 5316 to 5488 of TOR2) PCR product was generated with primers SZ167 5′–CATAATTGG GCCTTAGCTAATTTTGAAGTAATATCCATGCTAACATCTGTCTCTAAAAAGAAACAGGAAG–3′ and SZ168 5′–GAAAAAAGCCCTTGATCGCTGGAACAACATGTCTTTGAATAAGATTAGAAGAGTAATGAACTTC–3′ and plasmid pML40-3 as template. The PCR product was cotransformed with NcoI-digested pRS315-TOR1 into strain 16864. All hybrids were confirmed by sequencing. Plasmid pMEP2-GFP was kindly provided by J. Rutherford and will be published elsewhere. Plasmids pBJ9113 bearing the pep3ts-108 (Srivastava et al. 2000) and p188 containing the GCN4-lacZ reporter gene were provided by E. Jones and A. Hinnebusch, respectively.

Western blotting and α-factor processing:

Cell extracts from exponentially growing cultures in YEPD were prepared as previously described except that the breakage buffer consisted of 100 mm Tris-HCl, 50 mm KCl, 1 mm EDTA, 5% glycerol, and the protease inhibitors leupeptin, aprotinin, and pepstatin added at 1 μg/ml and 0.5 mm phenylmethyl sulfonyl fluoride (Rohde et al. 2004). Western blot analysis with 40 μg of protein for Ape1 and Alp1 and 10 μg for CpY was performed by standard techniques with antibodies specific for Ape1 and Alp1 (kindly provided by Y. Ohsumi) and CpY (Molecular Probes, Eugene, OR). The antisera recognizing amino acids 1–100 and 1–147 of Tor1 and Tor2, respectively, were previously described (Cardenas and Heitman 1995; Alarcon et al. 1996). For metabolic labeling cultures were grown to early exponential phase in SC medium. Cells were washed and resuspended in SC without methionine medium and treated with drug vehicle alone or 100 nm rapamycin and incubated for 20 min. Metabolic labeling of yeast cells with Trans 35S-LABEL (ICN), pulse chase, cell extract preparation, immunoprecipitation with specific antibodies for α-factor (a generous gift of T. Graham), and CpY were performed at 30° according to published protocols (Graham 1998).

Amino acid determination, Northern blot analysis, and β-galactosidase assays:

Amino acid extraction from exponentially growing cells in YPD medium was performed as described (Chen and Kaiser 2002). Amino acid analyses were carried out in duplicate by anion exchange chromatography employing a Beckman 6300 Li citrate-based analyzer followed by post-column ninhydrin reaction detection system at the Molecular Structure Facility, University of California at Davis. Northern blot analysis as well as β-galactosidase assays to determine GCN4-lacZ reporter gene activity were previously described (Cardenas et al. 1999; Rohde et al. 2004).

Fluorescent microscopy:

Cells were collected, spotted onto microscope slides, and imaged in a Nikon Eclipse E400 microscope equipped for epifluorescence and with a Nikon DXM1200F digital camera.

RESULTS

Mutation of TOR1 is synthetically lethal in combination with mutations in the class C VPS genes:

To understand the functional divergence between Tor1 and Tor2, we performed a genomewide scale screen to identify genes that when mutated exhibit a reduced fitness or synthetic lethal interaction with a tor1 mutation. This screen, performed by diploid-based synthetic lethality analysis on microarrays (dSLAM) (Pan et al. 2004), yielded 261 interactions that met the cut-off control/experimental hybridization ratio (C/E ratio) of ≥2 (Table 2). Remarkably, this set of genes comprises distinct clusters that share common functions, including transcriptional regulation, mRNA processing, ribosomal and mitochondrial functions, vesicle docking and fusion, protein transport, microautophagy, and vacuolar inheritance (Table 2, Figure 1A, and data not shown).

Figure 1.—

TOR1 exhibits synthetically lethal and reduced fitness interactions with genes involved in protein sorting and vacuolar functions. (A) Schematic of the distinctive functional categories, according to the published literature, identified by the tor1 dSLAM screen. (B) Heterozygous diploid strains tor1 pep3 (SZY26), tor1 pep5 (SZY27), tor1 vps16 (SZY29), and tor1 vps33 (SZY49) (see Table 1 for complete genotypes) were sporulated and dissected on YPD solid medium. After 3 days of incubation at 30°, colonies were replica plated onto YPD containing G418 (200 μg/ml) and SC-ura media. Plates were incubated for 2 days and photographed. (C) Heterozygous diploid strains tor1 vps15 (SZY28), tor1 vps34 (SZY30), tor1 vac7 (SZY31), tor1 vac17 (SZY33), tor1 vac8 (SZY32), tor1 vps39 (RPY50), and tor1 vps41 (RPY51) were sporulated and dissected as indicated above. Photographs show the colony size on the YPD plate of representative segregants from individual tetrads.

View this table:
TABLE 2

Synthetic lethal and reduced fitness interaction gene network of TOR1

In this study only the genetic interactions involved in vacuolar functions and protein trafficking were validated by classic tetrad analysis and the rest should be considered as potential synthetic interactions until subject to further analysis. Tetrad analysis was conducted in mating crosses between the pep3, pep5, vps16, vps33, vps15, vps34, vac7, vac8, vac17, gtr1, gtr2, and ego3 deletion mutants and strain SZY25, in which the entire TOR1 open reading frame was replaced with the URA3 selectable marker. As shown in Figure 1B the haploid meiotic progeny of the tor1 and pep3, pep5, vps16, and vps33 crosses were wild type (WT), ura+, or G418 resistant but no meiotic segregants with the ability to grow on both SD-ura and G418 selective media were recovered, confirming that these double mutants are synthetically lethal. In this analysis the rest of the genes examined exhibited a synthetic reduced fitness interaction when mutated in combination with tor1, as defined by the smaller size of the double mutant colony and a slow growth phenotype (illustrated in Figure 1C for the tor1 vps15, tor1 vps34, tor1 vac7, tor1 vac8, and tor1 vac17 and data not shown for tor1 gtr1, tor1 gtr2, and tor1 ego3 double mutants). The class C VPS genes also include VPS39 and VPS41; importantly, these genes were identified by the tor1 dSLAM screen but scored just below the C/E ratio of ≥2. On the basis of tetrad analysis, mutation of these vps genes in combination with tor1 also resulted in a synthetic reduced fitness phenotype (Figure 1C). These results validate the dSLAM screen and indicate that mutations in the class C VPS genes show a synthetic lethal interaction with tor1. For the remainder of this study, the focus is on understanding the basis for the synthetic lethality and reduced fitness defect exhibited by the tor1 class C vps double mutants.

Class C vps mutants fail to recover from rapamycin-induced growth arrest:

One hallmark of the genes linked to Tor signaling is that their mutation alters sensitivity to rapamycin. Mutation of 18 genes (including 14 involved in vacuolar functions and protein trafficking) out of the 62 shown in Table 2 resulted in rapamycin hypersensitivity as compared with the WT strain (Figure 2A) (Chan et al. 2000; Xie et al. 2005). In particular, mutants lacking components of the protein-sorting apparatus, including the class C Vps complex, the pre-autophagosomal (PAS) complex, as well as the vac8 mutant, are all extremely sensitive to rapamycin as compared to either the WT or the tor1 strains (Figure 2A). In addition, the class C vps mutants were tested for their ability to resume growth after rapamycin exposure. Actively growing cells were treated with rapamycin for 6 hr, washed, and spotted on YPD medium without drug. In contrast to the WT or tor1 strains, which readily recovered from rapamycin-induced growth arrest, mutations in class C VPS genes abolished recovery (Figure 2B). Interestingly, a similar phenotype was observed with strains harboring mutations in the EGO/Gse protein complex (Figure 1A), which was shown recently to play a role in combination with TORC1 in exit from rapamycin-induced quiescence (Dubouloz et al. 2005).

Figure 2.—

Class C vps mutants are hypersensitive to rapamycin and fail to recover from rapamycin-induced growth arrest. (A) The WT (BY4742), tor1 (#16864), and isogenic strains bearing mutations in different components of the class C Vps and PAS protein complexes as well as in genes involved in vacuolar inheritance, were grown overnight in liquid YPD medium. Equivalent numbers of cells were serially diluted and aliquots were spotted onto YPD and YPD containing 20 nm rapamycin. Plates were photographed following 3 days incubation at 30°. (B) Cultures of isogenic WT (BY4742), tor1 (#16864), pep5 (#10817), vps16 (#12783), and vps33 (SZY36) were grown to exponential phase. Cultures were divided in half and treated with drug vehicle alone or with 100 nm rapamycin for 6 hr. Cells were pelleted by centrifugation, washed twice, and equivalent numbers of cells were spotted on YPD medium. After incubation photographs were taken at 48 hr for both the untreated (YPD) and rapamycin-treated cells and 72 hr for the rapamycin-treated cells.

Expression of TOR1 but not TOR2 restores growth of a tor1 pep5 double mutant:

The synthetic lethality of tor1 class C vps double mutants is surprising as it has been proposed that either Tor1 or Tor2 can provide all of the rapamycin-sensitive essential functions attributable to the TORC1 signaling complex. The Tor proteins share a high degree of amino acid sequence identity with a few discrete stretches of dissimilarity, particularly at the extreme N-terminus and, to a lesser extent, toward the C-terminal region. To understand the structural basis for the differential functions of the two TOR genes, we created TOR2-TOR1 chimeric hybrid alleles cloned into low-copy, centromeric plasmids and examined their ability to provide TOR1 function and restore viability in tor1 pep5 meiotic segregants. First the TOR2-TOR1 hybrids were verified to be efficiently expressed, based on Western blot analysis with antibodies specific for Tor1 or Tor2 (Figure 3A). Second, we made use of the fact that a tor1 ssd1 strain is unable to grow at 39° (Alarcon et al. 1999), apparently due to inability of TOR2 to support cell integrity in the absence of the TOR1 and SSD1 genes (Reinke et al. 2004). Modest overexpression of TOR1 or TOR2 as well as of the TOR2-TOR1 hybrids efficiently rescued growth of the tor1 ssd1 mutant strain at 39°, indicating that the TOR2-TOR1 hybrids function to restore plasma membrane integrity (Figure 3B).

Figure 3.—

Expression of TOR1 but not TOR2 rescues viability of tor1 pep5 segregants. (A) The tor1 ssd1 mutant strain SZY21 was transformed with vector pRS315 and its derivatives encoding TOR1, TOR2, and the following TOR2–TOR1 hybrids (depicted in C): hybrid 1, Tor2 amino acids 1–1688 fused to Tor1 amino acids 1682–2470; hybrid 2, Tor2 amino acids 1–796 fused to Tor1 amino acids 788–2470; hybrid 3, Tor2 amino acids 1–483 fused to Tor1 amino acids 475–2470; hybrid 4, the Tor1 amino acid sequence from 1772–1815 was replaced by the homologous amino acid sequence of Tor2 from 1780–1818. Protein extracts of the different transformants were analyzed by Western blot with antibodies specific to an N-terminal region of Tor1 or Tor2. In these extracts, Cpr1 was also detected and serve as a loading control. (B) Equivalent cell numbers of the SZY21 transformants as indicated above were serially diluted and spotted in SD-leu medium. Plates were photographed after incubation at 30° or 39° and Tor1 function was scored by the ability to complement the conditional synthetic lethal phenotype of the tor1 ssd1 strain at 39°. (C) The tor1 pep5 diploid strain SZY27 transformed with the plasmids depicted at the left was sporulated and dissected on YPD medium. Following incubation at 30° for 3 days the plates were photographed and replica plated to YPD containing 200 μg/ml G418, SC-ura, and SC-leu media to score genotypes. The numbers on the right indicate the percent of tetrads (from a total of 10 scored) with a 4:0, 3:1, and 2:2 viable:inviable ratio.

While ectopic expression of plasmid-borne copies of TOR1 effectively rescued the growth of tor1 pep5 segregants, overexpression of TOR2 (also from a low-copy centromeric plasmid) failed to do so (Figure 3C). This result indicates that TOR1 has evolved to provide a function linked to the class C Vps complex which is not shared by TOR2. TOR2-TOR1 hybrids 1, 2, and 3 were able to suppress the synthetic lethality of tor1 pep5 segregants, albeit with an apparent differential efficiency. However, it should be noted that this assay examines both the increase in 4:0 segregation events and the relative growth of the rescued segregants. Based on an increase in 4:0 segregation events, hybrids 1, 2, and 3 all provide TOR1 function but it would be wrong to infer that hybrid 2 affords the best rescue because more 4:0 events were observed. Given the known segregation pattern of CEN-based plasmids (usually 2:2), the proportion of 4:0 segregation events seen with hybrids 1, 2, and 3 need not reflect a difference in rescue efficiency. In fact, hybrids 1, 3, and 4 rescue better based on colony growth. These results largely map this function to the C-terminal 788 amino acids of Tor1 (Figure 3C). Closer examination of this region in the Tor proteins revealed a discrete domain, from amino acid 1772 to 1815, which is highly dissimilar. However, replacement of this Tor1 region with the corresponding Tor2 sequence (hybrid 4) did not affect the ability to restore growth of tor1 pep5 segregants (Figure 3C). Thus, we conclude that the functional-structural differences between Tor1 and Tor2 map elsewhere in the C-terminal domain.

Tor1 does not regulate the functions of the class C Vps complex:

To gain insight into the molecular mechanisms that underlie the synthetic lethal interaction between TOR1 and class C VPS genes, we considered the hypothesis that Tor1 regulates the functions of this complex. This model is particularly attractive since Tor1 has been localized to internal membranes that resemble those associated with the endocytic pathway (Wedaman et al. 2003). To test this hypothesis, we examined if Tor1 mutation confers defects in class C Vps complex functions. The class C Vps protein complex is required for non-endosomal Golgi-to-vacuole transport, cytoplasm-to-vacuole targeting (Cvt), recycling from endosomes back to the late Golgi, endocytosis, and autophagy. The tor1 mutant did not show defects in the maturation of the vacuolar hydrolases CpY, Ape1, or Alp1 as compared to the WT strain or the pep3 mutant, which is defective in the proteolytic processing of these proteins (Figure 4A). Furthermore, CpY, Ape1, and Alp1 maturation was not altered by rapamycin treatment (Figure 4A). Moreover, mutation of TOR1 did not have any effect on endocytosis of the ammonium permease Mep2 elicited by ammonium addition to cells growing in ammonium-limiting medium (Figure 4B). These results indicate that the endosomal-to-vacuole, the Cvt, and the non-endosomal-to-vacuole protein-sorting routes are not regulated by Tor signaling and that Tor1 does not regulate endocytosis. This result is in accord with a previous study that identified a rapamycin-insensitive role for Tor2, but not for Tor1, in regulating endocytosis (deHart et al. 2003).

Figure 4.—

Tor1 signaling does not control class C Vps complex functions. (A) Tor1 does not control the maturation of vacuolar hydrolases. Exponentially growing cultures of the WT (BY4742), tor1 (#16864), and pep3 (#14105) strains in YPD medium were treated with either drug vehicle (−) or 100 nm rapamycin (+) for 1 hr. Protein extracts were prepared and analyzed by Western blot with antibodies specific for CpY, Ape1, and Alp1. (B) Tor 1 does not regulate endocytosis of the high-affinity ammonium permease Mep2. The WT, tor1, and pep3 strains indicated in A were transformed with plasmid pMep2-GFP. Transformants were grown to early exponential phase in SC-ura, washed twice, and resuspended in SLAD medium supplemented with the required amino acids to satisfy auxotrophic requirements. After 4 hr incubation in this medium, 50 mm ammonium sulfate was added to the cultures and incubation continued for 30 min. Cell samples were collected prior to and after addition of ammonium sulfate (Formula) and imaged for direct epifluorescence as indicated under experimental procedures. (C) Effects of TOR1 mutation in autophagy-mediated maturation of Ape1. The WT (BY4742), and the tor1 (#16864), pep3 (#14105), vac8 (#10253), and the tor1 vac8 (SZY43) mutant strains were grown and treated with rapamycin as indicated in A. Protein extracts were prepared and analyzed by Western blot with specific Ape1 antibodies.

Earlier studies have shown that treatment of yeast cells with rapamycin results in autophagy, indicating a role for Tor signaling in regulating this process (Noda and Ohsumi 1998). To test if mutation of TOR1 had any effect on autophagy, we made use of the fact that a vac8 mutant is defective in the Cvt pathway. Therefore, in a vac8 mutant, maturation of proApe1 in the vacuole is defective; however, this defect can be bypassed by induction of autophagy (Abeliovich et al. 2000). Maturation of Ape1 is enhanced and induced by rapamycin in the WT and vac8 strains, respectively (Figure 4C). Interestingly, mutation of TOR1 restores Ape1 processing in the vac8 cells, and rapamycin treatment of the tor1 vac8 double mutant dramatically enhances Ape1 processing (Figure 4C). As expected, mutation of PEP3 efficiently blocked the rapamycin-induced maturation of Ape1. These results show that mutation of TOR1 alone results in a low level of autophagy, whereas mutation of PEP3 blocks this process. We reasoned that if the basis for the synthetic lethality of the tor1 class C vps double mutants arose from a defect in effective execution of autophagy, tor1 mutation should also exhibit synthetic lethality in combination with mutations in other genes required for autophagy. Although our screen detected a synthetic reduced fitness interaction between tor1 and vps15 or vps34, two genes required for autophagy (Figure 1A) (Kihara et al. 2001), we found that a tor1 apg13 double-mutant strain that should be defective in autophagy was fully viable (data not shown).

Synthetic lethal interaction can arise between two genes whose products act in parallel or compensating pathways. Accordingly, we considered the hypothesis that TOR1 regulates a pathway that functions in parallel with the class C complex for protein sorting. A prediction of this model is that key genes working within this pathway should be synthetically lethal in combination with a pep3 mutation. A dSLAM screen with pep3 revealed 38 genes that are known when mutated to alter the sensitivity of cells to rapamycin (Figure 5A, data not shown) (Chan et al. 2000; Xie et al. 2005). Within this set, 13 genes share distinctive roles in protein trafficking between the ER to Golgi complex and protein cycling between the Golgi complex and endosomes (Figure 5A).

Figure 5.—

Tor1 does not regulate protein sorting from the ER to Golgi complex or protein cycling between the Golgi complex and endosomes. (A) Schematic of a fraction of the genes (grouped in functional categories according to the published literature) that when mutated alter rapamycin sensitivity and showed reduced fitness and synthetically lethal interactions in combination with mutation of PEP3. Interactions of the genes indicated in bold were confirmed by tetrad analysis. Genes enclosed by a box were synthetically lethal in combination with the pep3 mutation. (B) The WT (BY4742), tor1 (#16864), and pep3 (#14105) mutants were grown to early exponential phase and treated with drug vehicle (−) or 100 nm rapamycin for 20 min (+). Cell were pulse labeled with Trans 35S-LABEL for 8 min (min) and chased with unlabeled amino acids for 0, 5, 6, 14, 16, and 20 min as indicated in the figure (for further detail, see materials and methods). The mature (m) and processed (p) forms of α-factor (top) and CpY (bottom) were immunoprecipitated with α-factor and CpY-specific antibodies, respectively. Immunoprecipitated proteins were separated by SDS-PAGE and visualized by autoradiography. The migration of precursor forms characteristic for the Golgi complex and the ER compartment are indicated.

To test if Tor1 has a role in protein trafficking between these cell compartments, we tested the effects of TOR1 mutation and rapamycin treatment on α-factor maturation. The mating pheromone α factor is synthesized as a precursor and subjected to extensive glycosylation as it transverses the ER and Golgi complex, and then is cleaved at the late Golgi complex by the Kex2 protease to produce mature pheromone. Importantly, Kex2 normally cycles between the late Golgi complex and endosomal compartments. Thus, α-factor maturation serves as a reporter of protein trafficking between the ER to Golgi complex and protein cycling between the Golgi complex and endosomes. In this experiment we also monitored the processing of CpY in more detail. While mutation of PEP3 resulted in accumulation of glycosylated forms of α factor in the Golgi complex, and also blocked CpY processing, neither mutation of TOR1 nor rapamycin treatment of the WT strain perturbed α-factor or CpY processing to any significant extent (Figure 5B). Furthermore, rapamycin treatment in the pep3 strain did not alter the ratio of mature to immature α-factor forms observed in the pep3 untreated cells (Figure 5B). In addition, mutation of TOR1 or rapamycin treatment had no effect on the normal cycling of the SNARE protein Snc1 from early endosomes to the Golgi compartment and back to the plasma membrane (data not shown). Collectively, these results exclude models that evoke a role for Tor1 in regulating the class C Vps complex, or acting in a parallel compensating pathway to provide a known class C Vps complex function.

The synthetic lethality between TOR1 and PEP3 mutations is suppressed by amino acids:

Next, we entertained a model in which the class C Vps complex promotes Tor activity. Because Tor signaling is activated by nutrients, we tested the ability of the class C vps mutants to respond to nitrogen starvation. In contrast to the WT and tor1 strains, which effectively survived a 10-day period of incubation on ammonium-starvation medium, the class C vps mutants failed to resume growth (Figure 6A).

Figure 6.—

Class C vps mutants show severe defects in nitrogen metabolism. (A) The class C Vps complex is required for adaptation to nitrogen limitation. Isogenic WT (BY4742), and pep5 (#10817), vps16 (#12783), vps33 (SZY36), and tor1 (#16864) mutant strains were grown and spotted on YPD medium as indicated in the legend to Figure 2A. Following overnight incubation, the plate was photographed (see overnight image) and then starved for nitrogen by replica-plating to YNB without nitrogen medium. After incubating for 10 days, cells were replica plated back to YPD, incubated for 24 hr, and photographed. (B) Class C Vps complex mutations induce the general amino acid control response. Exponentially growing cultures of the WT (BY4742), and the pep5 (#10817), pep3 (#14105) mutant strains harboring the GCN4-lacZ reporter plasmid p180 were grown to exponential phase in SC-ura medium. Cultures were treated with 100 nm rapamycin for 0, 1, and 2 hr and analyzed for β-galactosidase activity. (C) Growth of the class C vps mutants at 37° is rescued by addition of glutamine, glutamate, and to a lesser extent by arginine. Isogenic WT (BY4742), and tor1 (#16864), pep3 (#14105), tor1 pep3 (strain SZY40-4 transformed with plasmid pBJ9113 expressing the pep3-108ts allele), pep5 (#10817), vps16 (#12783), vps33 (SZY36), vps39 (#13774), and vps41 (#14015) mutant strains were grown and spotted on SC medium (−) or SC medium supplemented with the indicated amino acid. Plates were incubated at 37° for 2 days and photographed. (D) Supplementation with glutamine or glutamate does not rescue the growth defect of the tor1 vac8, tor1 gtr1, and tor1 ego3 double mutants. Isogenic WT (BY4742), tor1 vac8 (SZY43), tor1 gtr1(RPY50), and tor1 ego3 (RPY51) double mutant strains were assayed as indicated in C except that plates were incubated at 30°.

Previous reports have shown that class C vps mutants contain severely fragmented vacuoles and are unable to store normal levels of basic amino acids in the vacuole (Kitamoto et al. 1988; Raymond et al. 1992). Low levels of intracellular amino acids, or rapamycin exposure, are known to trigger the general amino acid control response by inactivation of eIF2α, which results in a block to general protein synthesis and preferential translation of Gcn4. We examined if the low amino acid levels in class C vps mutants would suffice to activate Gcn4 translation. In accord with the above observation, pep3 or pep5 mutations resulted in robust Gcn4 translation, comparable to that observed in the WT strain treated with rapamycin (Figure 6B).

It has been proposed that glutamine and possibly glutamate, which are amino acids central to nitrogen metabolism, regulate nutrient signaling pathways, including the Tor pathway (Crespo et al. 2002; reviewed by Liu and Butow 2006). These observations prompted us to examine in more detail the intracellular amino acid content in the class C vps mutants. Under the growth conditions analyzed (active growth in YPD medium) the pep3, pep5, and vps33 mutants showed a lower level of the basic amino acids lysine, histidine, and arginine than that detected in the WT strain, in accord with an earlier report (Table 3) (Kitamoto et al. 1988). Remarkably, in contrast to the WT and tor1 strains, glutamate levels were significantly reduced (∼1.5-fold) in the class C vps mutants. The class C Vps complex is also required to support growth at 37° (Figure 6C) (Robinson et al. 1991, Koning et al. 2002). We sought to test if this growth defect is linked to amino acid levels. Exponentially growing cultures of the different strains were shifted from 30° to 37° and the amino acid levels were determined. Interestingly, shift to 37° for 3 hr caused a modest decline in glutamate and a near twofold decrease in glutamine in both the WT and the tor1 strains. In the pep3, pep5, and vps33 mutants the concentration of these amino acids declined more markedly, two- to threefold depending on the strain, and glutamate reached a threefold lower level than the one observed in the WT strain under similar conditions (Table 3). Moreover, supplementation of the growth media with glutamate, or with glutamine and to lesser extent with arginine, restored growth of the pep3, pep5, vps33, vps39, and vps41 mutant strains at 37°. This effect was specific and was not observed when the individual basic amino acids lysine or histidine were added as supplements (Figure 6C). The vps16 strain has a more severe growth defect at 37° than the rest of the class C vps mutants and this is only marginally alleviated by glutamine supplementation (see below for discussion). Strikingly, glutamate or glutamine and less efficiently arginine also supported the growth of a tor1 pep3 double-mutant strain carrying a partial loss of function pep3ts allele on a plasmid (Figure 6C). In contrast, supplementation with glutamine or glutamate did not rescue the growth defect of the tor1 vac8, tor1 gtr1, or tor1 ego3 double mutants (Figure 6D). Finally, supplementation with glutamine, glutamate, arginine, lysine, and histidine partially rescued the viability of tor1 pep5 segregants; however, these segregants grew poorly and rapidly accumulated suppressor mutations (data not shown). These results argue that the synthetic lethality of the tor1 pep3 double mutant derives from the metabolic derangement that underlies the reduced amino acid concentrations observed in the class C vps mutants.

View this table:
TABLE 3

The class C vps mutants show dramatically reduced levels of amino acids at 37°

DISCUSSION

Our genomewide search to define synthetic fitness or lethal interaction partners of TOR1 identified sets of genes that provide distinct functions. Among the synthetically lethal genes identified, four encode components of the class C Vps complex that functions in the recognition and fusion of vesicles with vacuolar and secretory membranes. This screen also identified a synthetic reduced fitness interaction between tor1 and mutations in genes encoding the components of the EGO/Gse complex, confirming and extending a previous study (Dubouloz et al. 2005). The EGO/Gse complex localizes to pre-vacuolar and vacuolar membranes and is required for sorting of the Gap1 amino acid permease to the plasma membrane in response to intracellular amino acids (Gao and Kaiser 2006).

Rapamycin exposure results in autophagy and thereby in a massive influx of membranes into the vacuolar membrane (Noda and Ohsumi 1998). In combination with TORC1, the EGO/Gse complex has been proposed to play a role in recovery from rapamycin-induced cell cycle arrest by enabling recycling of vacuolar membranes via microautophagy (Dubouloz et al. 2005). We have shown here that class C vps mutants are unable to recover from prolonged exposure to rapamycin. Although it is possible that the class C Vps complex is required for microautophagy, we were unable to address this question since class C vps mutants fail to complete autophagy and contain severely fragmented vacuoles.

We find that the structural difference between Tor1 and Tor2 to support viability of tor1 pep5 segregants maps to the C-terminal 788 amino acid region of Tor1. This result is in contrast with a previous study which concluded that the ability of a Tor1SR (rapamycin-resistant mutant) to confer rapamycin resistance in a vps16 mutant lies in the N-terminal 120 amino acid region of Tor1 (Xie et al. 2005). At present we do not have a good explanation for this discrepancy other than the biological assays to test for Tor1 function in the two studies were different.

Our screen also identified the Vps34 phosphatidylinositol 3-kinase and the Vps15 protein kinase, members of the PAS protein complex which functions in autophagy and protein sorting (Stack et al. 1995; Kihara et al. 2001), and genes involved in vacuolar inheritance (Figure 1A). Taken together these results reveal a prominent link between Tor signaling and vacuolar function. This view is further underscored by studies that have localized different components of the Tor pathway to the vacuolar membrane, including Tor2, Kog1, and Tco89 (Cardenas and Heitman 1995; Reinke et al. 2004; Araki et al. 2005). An important function of the vacuole is to preserve amino acid homeostasis by sequestering basic amino acids, which are toxic when present at high concentrations in the cytosol (reviewed by Klionsky et al. 1990).

Our results and those of others have shown that mutants lacking components of the class C Vps complex are unable to recover from ammonium starvation, they have low concentrations of intracellular amino acids, particularly glutamate and glutamine, and this concentration is further reduced upon shift of these mutants from 30° to 37°. Moreover, class C vps mutants have growth defects at 37° (Figure 6C) (Robinson et al. 1991, Koning et al. 2002). Remarkably, addition of glutamate or glutamine and to lesser extent arginine restored the growth at 37° of individual class C vps mutants and a tor1 pep3 mutant carrying a pep3ts allele, and partially rescued the viability of tor1 pep5 segregants. The effect of arginine could be explained by the ability of this amino acid to be converted into glutamate by the concerted actions of arginase and ornithine transaminase (reviewed by Davis 1986). We found that the vps16 strain was only marginally rescued by glutamine supplementation, suggesting that Vps16 has an additional function(s) different from that shared with the other members of the class C Vps complex. In this regard a role for Vps16 in regulating mRNA decapping and stability has been demonstrated (Zhang et al. 1999). Interestingly, genes encoding two subunits of the cap-binding complex, CBC1 and CBC2, were also identified by the tor1 synthetic screen and a rapamycin-sensitive role for Tor in controlling mRNA stability has been described (Albig and Decker 2001). Thus, it is possible that Tor1 function becomes compromised in the vps16 mutant via these activities independent of the class C Vps complex function.

In summary, our results indicate that the class C Vps complex is required to maintain amino acid homeostasis for effective Tor activity and lend further support to the proposal that glutamate/glutamine activates Tor signaling (Crespo et al. 2002). These studies also demonstrate that Tor1 is more efficient than Tor2 at providing rapamycin-sensitive Tor signaling under conditions where intracellular amino acids are drastically reduced. In support of this model, whereas Tor1 is a bona fide member of the rapamycin-sensitive TORC1 complex, Tor2 weakly interacts with this complex only in cells lacking Tor1 and no stable association of Tor2 with TORC1 has been detected in wild-type cells (Loewith et al. 2002; Reinke et al. 2004).

Strikingly, although both humans and many fungi express a single Tor kinase, the Tor genes have been independently duplicated in both the budding yeast S. cerevisiae and the fission yeast Schizosaccharomyces pombe (reviewed by Weisman 2004). Interestingly, in both cases one of the two paralogs is essential (TOR2) and the other is not under standard conditions (TOR1). Early studies in S. pombe revealed a role for Tor1 in amino acid uptake (Weisman et al. 2005). A more recent study has proposed a shared function for Tor1 and Tor2 in enabling cell proliferation and survival under both normal and adverse conditions and a positive and negative role for Tor1 and Tor2, respectively, in regulating G1 arrest and sexual differentiation (Uritani et al. 2006). Taken together, our findings provide a physiological basis for understanding the functional differences that distinguish Tor1 and Tor2 in S. cerevisiae and yield insight as to why the two genes are among the few (2–8%) retained following the genome duplication and massive gene loss event that marked the evolution of hemiascomycetous yeasts.

Acknowledgments

We thank Todd Graham for advice and suggestions with the α factor processing experiments. We thank Todd Graham, Elizabeth Jones, Yoshinori Ohsumi, Alan Hinnebusch, and Julian Rutherford for generous gifts of plasmids and antisera and Joseph Heitman and John McCusker for critical reading of the manuscript. This work was supported by R01CA114107 from the National Cancer Institute (to M.E.C); Xuewen Pan was supported by the Leukemia and Lymphoma Society and by National Institute of Health grant HG-002432 (to J.D.B).

Footnotes

  • 1 These authors contributed equally to this work.

  • Communicating editor: A. P. Mitchell

  • Received March 2, 2007.
  • Accepted May 25, 2007.

References

View Abstract