TOR complex 1 (TORC1) is an evolutionarily conserved protein kinase complex that promotes cellular macromolecular synthesis and suppresses autophagy. Amino-acid-induced activation of mammalian TORC1 is initiated by its recruitment to the RagA/B-RagC/D GTPase heterodimer, which is anchored to lysosomal membranes through the Ragulator complex. We have identified in the model organism Schizosaccharomyces pombe a Ragulator-like complex that tethers the Gtr1-Gtr2 Rag heterodimer to the membranes of vacuoles, the lysosome equivalent in yeasts. Unexpectedly, the Ragulator-Rag complex is not required for the vacuolar targeting of TORC1, but the complex plays a crucial role in attenuating TORC1 activity independently of the Tsc1-Tsc2 complex, a known negative regulator of TORC1 signaling. The GATOR1 complex, which functions as Gtr1 GAP, is essential for the TORC1 attenuation by the Ragulator-Rag complex, suggesting that Gtr1GDP-Gtr2 on vacuolar membranes moderates TORC1 signaling for optimal cellular response to nutrients.https://doi.org/10.7554/eLife.30880.001
Target of rapamycin (TOR) is serine/threonine-specific protein kinase highly conserved among eukaryotes from yeast to humans. By forming two distinct protein complexes known as TOR complex 1 (TORC1) and TOR complex 2 (TORC2), mammalian TOR (mTOR) and its counterparts in diverse eukaryotes control cell growth, proliferation and metabolisms in response to various intra- and extracellular signals (reviewed in [Wullschleger et al., 2006; Laplante and Sabatini, 2012]). Mammalian TORC1 (mTORC1) signaling has been attracting much attention, partly because its inhibition by rapamycin is effective in suppressing growth of some human cancers (Guertin and Sabatini, 2007; Shimobayashi and Hall, 2016). mTORC1 promotes cellular growth through the regulation of both anabolic and catabolic processes; amino acids serve as stimulatory signals that activate mTORC1, which phosphorylates S6K1 and 4E-BP1 to increase protein synthesis (Hara et al., 1998). Activated mTORC1 also phosphorylates ULK1, resulting in suppression of autophagy (Ganley et al., 2009; Jung et al., 2009).
The molecular mechanism that mediates mTORC1 activation in response to amino acid stimuli has recently been a focus of very active research. In the currently prevailing model, a critical step in amino-acid-induced activation of mTORC1 is its translocation to the surface of lysosomes, where the small GTPase Rheb (Inoki et al., 2003; Tee et al., 2003) triggers mTORC1 activation. Recruitment of mTORC1 to lysosomes is mediated by the Rag GTPase heterodimer that is anchored to lysosomal membranes through a protein complex called Ragulator (Kim et al., 2008; Sancak et al., 2008; Sancak et al., 2010). The Rag heterodimer consists of RagA or RagB bound to either RagC or RagD, and it has been proposed that guanine nucleotide loading to RagA/B is regulated in response to amino acid signals; the Ragulator complex, which is composed of LAMTOR1/p18, LAMTOR2/p14, LAMTOR3/MP1, LAMTOR4/C7orf59 and LAMTOR5/HBXIP, functions as a guanine nucleotide exchange factor (GEF) for RagA/B (Bar-Peled et al., 2012). In the presence of amino acid stimuli, RagA/B are in the GTP-bound form and the Rag heterodimer can interact with the raptor subunit of mTORC1 for lysosomal recruitment of mTORC1 (Sancak et al., 2008). On the other hand, the GATOR1 complex composed of DEPDC5, Nprl2, and Nprl3 has GTPase-activating protein (GAP) activity toward RagA/B, whose GDP-bound form induces the release of mTORC1 from lysosomes in the absence of amino acids (Bar-Peled et al., 2013). Without functional GATOR1, mTORC1 signaling is resistant to amino acid starvation and, interestingly, inactivating mutations to the GATOR1 components have been found in human cancers. GATOR1 forms a larger GATOR complex together with GATOR2, which comprises WDR24, WDR59, MIOS, SEH1L and SEC13, and gene knockdown experiments suggest that the GATOR2 subcomplex negatively regulates the GAP activity of GATOR1 (Bar-Peled et al., 2013). It should be noted, however, that further intricacy may remain to be added to the model described above. There is a discrepant report that guanine nucleotide loading to the Rag GTPases is not responsive to amino acids (Oshiro et al., 2014). RagA/B-independent activation of mTORC1 has also been observed (Kim et al., 2014; Jewell et al., 2015).
The mTORC1 regulators enumerated above are conserved also in unicellular eukaryotes. The Gtr1-Gtr2 GTPase heterodimer (Nakashima et al., 1999) is the budding yeast counterpart of RagA/B-RagC/D, interacting with the Ego1-Ego2-Ego3 complex that tethers Gtr1-Gtr2 to the membrane of the vacuole, a lysosome-like organelle (Dubouloz et al., 2005; Gao and Kaiser, 2006; Powis et al., 2015; Kira et al., 2016). The Ego ternary complex is likely to be the yeast equivalent of mammalian Ragulator, although their constituents share little sequence homology (Kogan et al., 2010). The yeast SEACIT and SEACAT complexes apparently correspond to mammalian GATOR1 and GATOR2, respectively (Neklesa and Davis, 2009; Dokudovskaya et al., 2011; Wu et al., 2011; Panchaud et al., 2013a; Panchaud et al., 2013b; Kira et al., 2014). However, these TORC1 regulators in budding yeast might function differently from those in mammals, because the Rheb GTPase, the primary mTORC1 activator, is not part of the TORC1 signaling pathway in budding yeast (Urano et al., 2000).
On the other hand, a Rheb ortholog called Rhb1 is an essential activator of TORC1 in the fission yeast Schizosaccharomyces pombe, another model eukaryote distantly related to budding yeast (Mach et al., 2000; Urano et al., 2005; Uritani et al., 2006). Like mammalian Rheb, Rhb1 is under the regulation of the Tsc1-Tsc2 complex that functions as GAP for Rhb1 (Matsumoto et al., 2002; van Slegtenhorst et al., 2004; Murai et al., 2009). Thus, S. pombe is expected to serve as an excellent experimental system to explore the TORC1 regulatory mechanisms that are conserved also in mammals. Fission yeast TORC1 is composed of the Tor2 kinase associated with the regulatory subunits Mip1 and Wat1, which are orthologous to mammalian raptor and mLST8, respectively (Álvarez and Moreno, 2006; Matsuo et al., 2007; Hayashi et al., 2007). The heterodimeric Rag GTPases Gtr1-Gtr2 are also implicated in S. pombe TORC1 regulation, although their exact role is ambiguous because of contradictory reports of their mutant phenotypes (Valbuena et al., 2012; Ma et al., 2013; Laor et al., 2014; Ma et al., 2016). In this study, we have identified Ragulator- and GATOR1-like complexes in fission yeast, which regulate the cellular localization and nucleotide-binding state of Gtr1-Gtr2, respectively, as has been found with their mammalian counterparts. Unexpectedly, however, these conserved regulatory machineries are all required to attenuate TORC1 activity, and mutants lacking any of them show severe growth defects due to deregulated TORC1 activation. Our data collectively suggest that TORC1 activation in S. pombe does not require the Rag-like GTPases and that they rather play an important role in moderating TORC1 activity on vacuolar membranes for optimal cellular response to nutrients.
Aiming to identify proteins that physically interact with the Rag-family GTPases in fission yeast, affinity purification of the Gtr1-Gtr2 heterodimer was performed using strains that express Gtr1 and Gtr2 from their chromosomal loci as fusions with different epitope tags, FLAG and myc. Two successive immunoprecipitation procedures using anti-FLAG and anti-myc antibodies to collect the Gtr1-Gtr2 heterodimer complex were followed by mass spectrometry, which identified four co-purified proteins encoded by open-reading frames SPBC29A10.17, SPBC1778.05c, SPAC222.19, and SPAC23D3.16 in the S. pombe genome database (Figure 1A and B). The protein encoded by SPBC1778.05c has been named Lam2 because of its sequence similarity to the mammalian Ragulator subunit LAMTOR2 (Ma et al., 2016). On the other hand, the protein products of SPBC29A10.17, SPAC222.19 and SPAC23D3.16 show no apparent sequence homology to any known proteins. However, they form a complex with Lam2, and the predicted secondary structure of the SPAC222.19 protein resembles the αββαβββα structure of the roadblock domain (Koonin and Aravind, 2000), which has been identified among the human Ragulator components and the Ego proteins in budding yeast (Bar-Peled et al., 2012) (Figure 1—figure supplement 1A); therefore, we refer to those proteins as Lam1 (SPBC29A10.17), Lam3 (SPAC222.19), and Lam4 (SPAC23D3.16). Their physical interaction with the Gtr1-Gtr2 GTPases was further confirmed by co-immunoprecipitation experiments (Figure 1C). In addition, immunoprecipitating any one of the Lam proteins resulted in co-purification of the other three Lam proteins (Figure 1—figure supplement 1B–E), confirming complex formation among Lam1, Lam2, Lam3 and Lam4.
Gene disruption analyses found that the null mutation of the lam1, lam2, lam3 and lam4 exhibit very similar growth defects on rich yeast extract (YES) medium (Figure 1D). We also noticed that their growth phenotypes were comparable to those of the cells lacking Gtr1 or Gtr2. Moreover, the defective phenotypes of the lam∆ gtr1∆ and lam∆ gtr2∆ double mutants are indistinguishable from those of the respective single mutants, indicating that the lam∆ and gtr∆ phenotypes are not additive (Figure 1D and E). Together, these biochemical and genetic data suggest that the Gtr1-Gtr2 GTPases and the Lam1~4 proteins form a complex and function together to promote cell growth.
Similar to the lysosomal localization of the mammalian Rag GTPases, the Gtr1-Gtr2 heterodimer localizes to membranes of the vacuole, the lysosome equivalent in yeasts (Valbuena et al., 2012). To examine whether the Lam proteins are also localized to vacuolar membranes, we constructed strains in which the GFP-encoding sequence was inserted at the 3’ end of the chromosomal lam1+, lam2+, lam3+ and lam4+ open-reading frames. By fluorescence microscopy, all four Lam-GFP fusion proteins were detected on vacuolar membranes labeled by the FM4-64 dye (Figure 2A). The Lam proteins were also detectable on vacuoles even in strains lacking Gtr1 and Gtr2 (Figure 2—figure supplement 1), indicating that the vacuolar targeting of the Lam proteins does not require the Gtr1-Gtr2 GTPases. We next assessed the interdependency among the Lam proteins for their vacuolar localization. In the lam1∆ mutant, Lam2, Lam3 and Lam4 diffused throughout the cytoplasm, while Lam1 was observed on vacuoles even in lam2∆, lam3∆ and lam4∆ cells (Figure 2B, Figure 2—figure supplement 2); thus, Lam1 appears to be indispensable for anchoring the Lam complex to vacuoles. On the other hand, the vacuolar localization of Lam2 and Lam3 is also dependent on Lam4 but not vice versa (Figure 2B, Figure 2—figure supplement 2), implying that Lam4 mediates association of Lam2 and Lam3 with Lam1 anchored to vacuoles. We also observed that Lam2 and Lam3 are mutually dependent for their vacuolar localization.
Consistent with the inferred role of Lam1 as a vacuolar membrane anchor of the Lam complex, its N-terminal region has potential myristoylation and palmitoylation sites (Figure 2C) similar to those in the Ragulator component LAMTOR1 and the budding yeast Ego1 protein (Kogan et al., 2010). Moreover, the vacuole/lysosome localization signal sequence composed of acidic residues followed by di-leucine (Darsow et al., 1998) is also conserved within the N-terminal regions of Lam1, LAMTOR1, and Ego1 (Figure 2C, underlines). We found that simultaneous alanine substitutions of both myristorylation and palmitoylation sites in Lam1 resulted in its dispersal throughout the cytoplasm (Figure 2D 'lam1GCC-AAA'). On the other hand, Lam1 was distributed to both vacuolar and plasma membranes when the conserved di-leucine motif of Leu-21 and Leu-22 was replaced by two alanine residues (lam1LL-AA). These observations suggest that specific targeting of Lam1 to vacuolar membranes requires its N-terminal lipid modifications as well as the di-leucine motif.
In addition to the detectable sequence similarity of Lam2 to mammalian LAMTOR2, the aforementioned characteristics conserved among Lam1, LAMTOR1 and Ego1 suggest that the Lam protein complex is a Ragulator equivalent in fission yeast. The vacuolar localization of the Lam proteins (Figure 2A) and their physical association with the Gtr1-Gtr2 GTPases (Figure 1) are also consistent with such a possibility. Therefore, we investigated whether the Lam protein complex is responsible for the localization of the Gtr1-Gtr2 GTPases to vacuolar membranes, as is the case with mammalian Ragulator for the lysosomal localization of the Rag GTPases (Sancak et al., 2010). Indeed, the vacuolar localization of Gtr1-Gtr2 is impaired in the absence of any one of the Lam proteins (Figure 3A, Figure 3—figure supplement 1A), indicating that the intact Lam complex is required for the recruitment of Gtr1-Gtr2 to vacuoles. Heterodimer formation between Gtr1 and Gtr2 is also essential for their localization to vacuoles, because these proteins diffused throughout the cytoplasm in the absence of the other (Figure 3A ‘gtr2∆’ and Figure 3—figure supplement 1A ‘gtr1∆’). Intriguingly, artificial targeting of the Lam1 protein to the nucleus by fusing the nuclear localization signal sequence of the SV40 large T-antigen resulted in the translocation of Gtr1-Gtr2 to the nucleus (Figure 3B). Thus, the Lam protein complex appears to be a key determinant of the cellular localization of Gtr1-Gtr2, serving as a Ragulator-like complex that recruits the fission yeast Rag GTPases to vacuolar membranes.
In a currently prevailing model, recruitment of mTORC1 to lysosomal membranes by the Ragulator-Rag complex is a key step in mTORC1 activation by the Rheb GTPase (Sancak et al., 2010). Vacuolar localization of fission yeast TORC1 was previously reported (Valbuena et al., 2012), which we confirmed by constructing a strain that expresses GFP-tagged Mip1, a raptor ortholog, from the chromosomal mip1+ locus (Figure 3C ‘wild-type’). To our surprise, Mip1-GFP was observed mostly on vacuolar membranes even in lam1∆, gtr1∆ and gtr2∆ mutant cells. In addition, Mip1 persisted on vacuoles even when the Gtr1-Gtr2 GTPases were translocated to the nucleus by NLS-fused Lam1 (Figure 3B, bottom panels). These observations indicate that fission yeast TORC1 can localize to vacuolar membranes in a manner independent of the Ragulator-Rag complex.
The GATOR1 complex, which consists of DEPDC5, Nprl2 and Nprl3, has GAP activity for RagA/B, and its inactivating mutations in cancers result in mTORC1 hyperactivation (Bar-Peled et al., 2013). Nprl2 has a fission yeast ortholog, Npr2, and genetic analysis suggested that Npr2 negatively regulates TORC1 also in fission yeast (Ma et al., 2013). To determine whether Npr2 is part of a GATOR1-like complex, Npr2 was affinity-purified from the lysate of a strain expressing FLAG-tagged Npr2 (Figure 4A), and the co-purified proteins were analyzed by mass spectrometry (Figure 4B). The identified Npr2-interacting proteins include Iml1 (SPBC26H8.04c) and Npr3 (SPBC543.04), which show detectable sequence similarity to the mammalian GATOR1 components DEPDC5 and Nprl3, respectively (Figure 4—figure supplement 1). Physical association of Npr2, Iml1 and Npr3 was also confirmed by pair-wise immunoprecipitation experiments (Figure 4—figure supplement 2A–C), suggesting that the three proteins form a protein complex equivalent to mammalian GATOR1. Strains lacking the Iml1, Npr2 or Npr3-coding genes exhibited similar growth defects on YES agar medium (Figure 4C). In addition, the double- and triple-deletion mutants of these genes showed defective phenotypes comparable to the respective single mutants, consistent with the notion that Iml1, Npr2 and Npr3 function as a complex. We also noticed that the amino acid sequences of Sea4, Sec13, Sea3, and Seh1 detected in the FLAG-Npr2 immunoprecipitates (Figure 4B) show similarity to those of the mammalian GATOR2 components MIOS, Sec13, WDR59, and SEH1, respectively (http://www.pombase.org; our unpublished results), implying conservation of the entire GATOR complex in S. pombe.
Fluorescence microscopy of fission yeast strains whose chromosomal iml1+ gene was tagged with the GFP sequence revealed Iml1 localization to vacuolar membranes in wild-type cells (Figure 4D) as well as in strains lacking Npr2 or Npr3 (Figure 4—figure supplement 2D). In addition, the vacuolar localization of Iml1 did not appear to require the Gtr1-Gtr2 GTPases and the Ragulator-like complex on vacuolar membranes (Figure 4D, Figure 4—figure supplement 2E). Furthermore, the Iml1-Npr2-Npr3 complex was found to be dispensable for targeting Gtr1-Gtr2 to vacuoles (Figure 4E, Figure 4—figure supplement 2F–H). Thus, the Ragulator-Rag complex and the GATOR1 complex in fission yeast are both localized to vacuolar membranes, but in a manner independent of each other.
Loss of functional GATOR1 has been identified in cancer cell lines, implying that hyperactive mTORC1 in the absence of GATOR1 contributes to cell proliferation (Bar-Peled et al., 2013). Therefore, the growth defect observed in the absence of the GATOR1-like complex in fission yeast (Figure 4C) was an unexpected result, and further genetic analysis was conducted to determine the relationship between the Iml1-Npr2-Npr3 complex and the Gtr1-Gtr2 Rag GTPases. The compromised growth phenotypes of the iml1∆, npr2∆ and npr3∆ mutants on YES medium were found to be comparable to those of the gtr1∆ and gtr2∆ strains, and the double mutant analyses indicated that the GATOR1 and Rag GTPase mutant phenotypes are not additive (Figure 5A and B). Thus, S. pombe GATOR1 appears to contribute to cell growth in the same pathway as the Gtr1-Gtr2 GTPases. To further examine the possibility that the GATOR1 complex regulates the nucleotide-binding state of Gtr1, we constructed strains whose gtr1+ locus was replaced with gtr1Q61L or gtr1S16N that are expected to express constitutively GTP- and GDP-bound Gtr1, respectively (Nakashima et al., 1999; Valbuena et al., 2012). Cells expressing GTP-locked Gtr1Q61L showed a severe growth defect similar to the gtr1∆ mutant, while the growth of the gtr1S16N mutant was comparable to that of wild-type cells, implying an important function of the GDP-bound form of Gtr1 (Figure 5C). On the other hand, the equivalent mutations to Gtr2 did not significantly affect the phenotype. The gtr2∆ mutation, however, caused a growth defect indistinguishable from that of the gtr1∆ mutant and was found to be epistatic to any of the gtr1 mutations including gtr1S16N (Figure 5D), indicating the essential role of Gtr2 for the Gtr1 function.
If, like mammalian GATOR1 (Bar-Peled et al., 2013), the Iml1-Npr2-Npr3 complex functions as GAP for Gtr1, GTP-bound Gtr1 should accumulate in the absence of the Iml1-Npr2-Npr3 complex. Indeed, the GTP-locked gtr1Q61L mutation and the GATOR1 mutations caused very similar growth phenotype, and their defects were not additive in the double mutants (Figure 5E). On the contrary, the GDP-locked gtr1S16N mutation suppressed the phenotype of the GATOR1-defective mutants (Figure 5F). These results support the notion that GATOR1 functions as GAP for Gtr1 and that loss of functional GATOR1 is complemented by expression of the GDP-bound form of Gtr1. Consistently, iml1R854A, a mutation to the ‘arginine finger’ conserved within the GAP domain of Iml1 orthologs (Figure 4—figure supplement 1A) (Panchaud et al., 2013a), caused a growth defect similar to those of the lam1∆ and gtr1Q61L strains (Figure 4—figure supplement 2I).
The observations above strongly suggested an important function of GDP-bound Gtr1 (Gtr1GDP) in growth of S. pombe, and we hypothesized that optimal growth of fission yeast under nutrient-rich conditions, such as yeast extract medium, might require attenuation of TORC1 by the heterodimer of Gtr1GDP and Gtr2. As expected, the TORC1 inhibitors rapamycin and caffeine (Takahara and Maeda, 2012; Rallis et al., 2013) suppressed the defective growth phenotypes of the strains lacking Gtr1-Gtr2 or the GATOR1 components as well as that of the GTP-locked gtr1Q61L mutant (Figure 5G, Figure 5—figure supplement 1A). Similarly, temperature-sensitive hypomorphic tor2 alleles, tor2-13 and tor2-287 (Uritani et al., 2006; Hayashi et al., 2007; Ikai et al., 2011), also complemented the loss of Gtr1-Gtr2, GATOR1 or Ragulator (Figure 5H–K, Figure 5—figure supplement 1B and C), corroborating that cells with reduced TORC1 activity can grow in the absence of Gtr1-Gtr2 and GATOR1. Indeed, absence of Gtr1-Gtr2, GATOR1 or Ragulator allowed tor2-287 mutant cells to grow even at the restrictive temperature (Figure 5I and K and Figure 5—figure supplement 1C), suggesting that loss of Gtr1GDP-Gtr2 leads to augmented TORC1 activity in this tor2 mutant.
Collectively, the above data are consistent with a model where the GATOR1 GAP gives rise to Gtr1GDP, which forms heterodimer with Gtr2 to moderate TORC1 activity to a level optimal for cell growth. Because the Tsc1-Tsc2 complex also negatively regulates TORC1 signaling as GAP for the Rhb1 GTPase (Urano et al., 2005; Uritani et al., 2006; van Slegtenhorst et al., 2004; Nakase et al., 2006), we examined the relationship of Tsc1-Tsc2 with GATOR1-Gtr1. Without the functional Tsc complex in the tsc1∆ or tsc2∆ background, both gtr1∆ and iml1∆ mutations brought about growth defects much severer than those of the single mutants (Figure 5L and M), and these synthetic phenotypes were suppressed by rapamycin (Figure 5—figure supplement 1D). Thus, it appears that the GATOR1-Gtr signaling axis and the Tsc-Rhb1 pathway independently impinge on TORC1 (Figure 5N) and that loss of both negative regulatory mechanisms results in a severe growth defect.
The genetic analyses presented above led to a working model that GATOR1 and Gtr1GDP-Gtr2 temper TORC1 activation induced by the Rhb1 GTPase (Figure 5N); loss of Gtr1GDP-Gtr2 from vacuolar membranes results in deregulated TORC1 signaling and growth defects that can be suppressed by rapamycin or Tor2 hypomorphs. In order to further substantiate this model, we next examined how GATOR1 and Gtr1-Gtr2 affect TORC1 activity by monitoring phosphorylation of Psk1, a S. pombe S6K1 ortholog. The hydrophobic motif phosphorylation of Psk1 by TORC1 results in reduced electrophoretic mobility of Psk1 and is also detectable using antibodies against phosphorylated S6K1 (Nakashima et al., 2012). As previously reported (Hatano et al., 2015), phosphorylated Psk1 became undetectable in wild-type cells within 30 min of nitrogen starvation that inactivates TORC1 (Figure 6A). In contrast, Psk1 remained phosphorylated until later time points in strains lacking the Tsc1-Tsc2 complex, a known negative regulator of TORC1 signaling, as well as in those carrying the activating mutations of Tor2 and Rhb1, tor2E2221K (Urano et al., 2007) and rhb1-DA4 (Murai et al., 2009), respectively (Figure 6B). Similar delays in TORC1 inactivation were observed in the gtr1∆ and gtr2∆ mutants (Figure 6C), indicating that absence of Gtr1-Gtr2 results in deregulated TORC1 signaling that cannot promptly respond to starvation. The GTP-locked gtr1Q61L mutant and iml1∆ mutant also exhibited a delay in the nitrogen-starvation response, while no apparent difference was observed between the GDP-locked gtr1S16N mutant and the wild type (Figure 6D and E). These data are consistent with our model that GATOR1 functions as GAP for Gtr1 and that the Gtr1GDP-Gtr2 heterodimer plays a role in restraining TORC1 activation. As expected, the GDP-locked gtr1S16N mutation suppressed the delayed starvation response of the iml1∆ mutant that lacks the functional GATOR1 complex (Figure 6F), consistent with the observed phenotypic complementation (Figure 5F). In addition, the parallel TORC1 regulation by the Tsc1-Tsc2 and GATOR1-Gtr pathways (Figure 5N) was also observed by monitoring the TORC1 activity; the TORC1 response to nitrogen starvation was further delayed in the gtr1∆ tsc2∆ (Figure 6G) and iml1∆ tsc2∆ double mutants (Figure 6H) in comparison to the respective single mutants. We thus examined further the possibility that the GATOR1-Gtr pathway can regulate TORC1 signaling independently of the Rhb1 GTPase. The Rhb1 is an essential activator of TORC1 and the rhb1∆ mutant is viable only in the presence of an activated allele of Tor2, such as tor2E2221K (Mach et al., 2000; Urano et al., 2007). Also in the tor2E2221K background, the gtr1∆ and iml1∆ mutations caused significant delays in the TORC1 response to starvation, indicating that TORC1 carrying the tor2E2221K mutation is still subject to the regulation by the GATOR1-Gtr pathway (Figure 6I). Interestingly, even in the absence of Rhb1, the gtr1∆ and iml1∆ mutations caused apparent delays in the TORC1 inactivation upon starvation (Figure 6J), suggesting that Gtr1GDP-Gtr2 can control TORC1 activity in a Rhb1-independent manner, at least in the tor2E2221K background.
In the presence of plentiful nutrients, Psk1 appeared to be mostly in the phosphorylated, slow-migrating form and very little unphosphorylated Psk1 was detectable (e.g. Figure 6A, middle panel at time 0). On the other hand, the slow-migrating, phosphorylated form of Sck1, which is another known substrate of TORC1 (Nakashima et al., 2012), significantly increased in growing gtr1∆, gtr2∆ and iml1∆ mutant cells, comparing to that in wild-type cells (Figure 6K). This observation suggests that absence of GATOR1 or Gtr1-Gtr2 results in elevated TORC1 activity also under nitrogen-replete conditions. In contrast, loss of the Tsc1-Tsc2 complex seems to have little impact on TORC1 activity under the same conditions.
Together, the above results demonstrate hyperactive TORC1 signaling in the absence of GATOR1 or Gtr1GDP-Gtr2, consistent with the phenotypic analyses shown in Figure 5.
Our studies described above showed that loss of Gtr1GDP-Gtr2 results in deregulated TORC1 activation that brings about a growth defect in the yeast extract medium. We noticed that the defective phenotype was alleviated by supplementing the medium with ammonium (Figure 7A, Figure 7—figure supplement 1A), implying that loss of the functional GATOR1-Gtr pathway starves S. pombe cells of nitrogen in the standard yeast extract medium. Because TORC1 is known to control cellular amino-acid uptake (Matsumoto et al., 2002; van Slegtenhorst et al., 2004; Aspuria and Tamanoi, 2008; Liu et al., 2015), we examined the possibility that the observed growth defect is due to compromised amino-acid uptake. The gtr, GATOR1 and Ragulator mutants could grow on EMM minimal medium that contains ammonium as nitrogen source, but little growth of the mutant cells was observed when ammonium was replaced with amino acid (Figure 7B, Figure 7—figure supplement 1B). Addition of ammonium chloride significantly mitigated their growth defect on the amino-acid media, implying that mutant cells lacking Gtr1GDP or Ragulator can utilize ammonium but not amino acids as nitrogen source. The defective amino-acid uptake of the gtr, GATOR1 and Ragulator mutants was also evidenced by their resistance to canavanine and ethionine, toxic analogs of arginine and methionine, respectively (Figure 7C, Figure 7—figure supplement 1C). As expected, those mutants with amino-acid auxotrophy failed to grow by utilizing exogenous amino acids in the growth medium (Figure 7—figure supplement 1D and E).
It has been reported that the Pub1 ubiquitin ligase and its arrestin-related adaptor, Any1, negatively regulate the plasma-membrane localization of amino acid transporters through endocytosis and their retention to Golgi/endosomes (Nakase et al., 2013; Nakashima et al., 2014). The pub1∆ and any1∆ mutations partially suppressed the growth defect of the gtr1∆ and iml1∆ mutants on YES medium and abrogated their resistance to the toxic amino acid analogs, implying ameliorated amino-acid uptake in the double mutants (Figure 7D). Indeed, the pub1∆ and any1∆ mutations failed to improve the growth of gtr1∆ and iml1∆ cells on EMM, whose nitrogen source is ammonium rather than amino acid. We therefore examined whether the gtr1∆ and iml1∆ mutations affect the plasma-membrane localization of Cat1, an amino acid transporter known to be under the regulation of Pub1 and Any1 (Nakashima et al., 2014). The majority of Cat1 with a fluorescent tag was detected at the plasma membrane in wild-type cells, but the protein was mostly sequestered in the cytoplasm of the gtr1∆ and iml1∆ mutants (Figure 7E). Furthermore, the pub1∆ (Figure 7F) and any1∆ (Figure 7G) mutations restored the plasma membrane localization of Cat1 in the gtr1∆ and iml1∆ mutants, consistent with the observed genetic complementation (Figure 7D).
Taken together, these results strongly suggest that elevated TORC1 activity in cells lacking GATOR1, Gtr1GDP-Gtr2 or Ragulator leads to reduced amino-acid uptake, which partly contributes to their growth defects on amino acids as nitrogen source. Although inactivation of TORC1, such as that by tor2 mutations, also results in growth defects, the tor2 phenotypes are not complemented by ammonium in the growth medium (Figure 7—figure supplement 1F). In addition, the tor2 mutant cells are smaller than wild-type cells (Matsuo et al., 2007), while the mutants lacking Gtr1GDP-Gtr2, GATOR1 or Ragulator do not show such a phenotype (Figure 7—figure supplement 1G). Thus, elevated TORC1 activity brings about growth defects that are apparently different from those from TORC1 inactivation.
Like in mammalian cells, the Rheb GTPase Rhb1 is an essential activator of TORC1 in fission yeast (Mach et al., 2000; Urano et al., 2005; Uritani et al., 2006). In this study, we have identified and characterized the S. pombe counterparts of Ragulator and GATOR1, the mammalian protein complexes that control the Rheb-dependent activation of mTORC1 on the lysosomal surface (Sancak et al., 2010; Bar-Peled et al., 2013). Because the Rag-like heterodimeric Gtr1-Gtr2 GTPases are also implicated in the regulation of S. pombe TORC1 (Valbuena et al., 2012), we delved into the roles and relationships of these conserved regulatory factors in the control of TORC1 signaling.
Affinity-purification of the Gtr1-Gtr2 heterodimer followed by mass-spectrometry led to identification of four co-purified proteins, Lam1, Lam2, Lam3 and Lam4 (Figure 1). The following observations strongly suggest that the four Lam proteins form a complex equivalent to mammalian Ragulator that recruits the Rag GTPase heterodimer to lysosomes. First, among the components of the Lam protein complex, Lam2 shows detectable sequence similarity to the Ragulator subunit LAMTOR2/p14 (Ma et al., 2016). Second, the predicted secondary structure of Lam3 suggests it has a roadblock domain, which has been found among the Ragulator components. Third, the lipid modification sites and the lysosome/vacuole-targeting di-leucine motif at the N-termini of mammalian LAMTOR1 and Ego1 in budding yeast are also conserved in the S. pombe Lam1 protein (Figure 2). Lastly, the Lam protein complex is localized to the membranes of vacuoles, lysosome-like organelles in yeast and is essential for the vacuolar localization of the Gtr1-Gtr2 GTPase heterodimer (Figures 2 and 3, Figure 3—figure supplement 1).
In contrast to the limited sequence similarity of the Ragulator subunits between mammals and fission yeast, the primary structures of the GATOR1 subunits are relatively well conserved through evolution; Iml1, Npr2 and Npr3 in fission yeast are orthologous to human DEPDC5, Nprl2 and Nprl3, respectively (Figure 4—figure supplements 1 ; Ma et al., 2013) and the three proteins indeed form a complex (Figure 4—figure supplement 2A–C). Furthermore, the iml1∆, npr2∆ and npr3∆ defects are complemented by expression of the GDP-locked Gtr1 (Figures 5F and 6F), consistent with the idea that the GATOR1 function as GAP for the RagA/B GTPases is conserved in S. pombe, as has also been found in the budding yeast Saccharomyces cerevisiae (Panchaud et al., 2013a). On the contrary, Ma et al. (2016) recently reported that expression of Gtr1S20L, their assumed GDP-locked mutant of Gtr1, failed to complement the npr2∆ and npr3∆ defects, concluding that those mutations do not alter the nucleotide-bound state of Gtr1 (Ma et al., 2016); however, the Ser-20 substitution has never been characterized in any Rag family GTPases (Figure 5—source data 1) and may not behave as intended. Their study also proposed the function of Npr2 and Npr3 in tethering Gtr1-Gtr2 to vacuoles, based on the result that the expression and vacuolar localization of Gtr1-GFP and Gtr2-GFP were compromised when either of them was ectopically overexpressed in the npr2∆ and npr3∆ mutants (Ma et al., 2016). Unbalanced expression of the heterodimer constituents may explain the observations, as the GATOR1 mutations did not affect the protein levels and vacuolar localization of GFP-fused Gtr1 and Gtr2 that were expressed from their chromosomal loci (Figure 4E, Figure 4—figure supplement 2F–H). Thus, the vacuolar localization of GATOR1 appears to be independent of the Ragulator-Gtr complex, and may be mediated by an additional mechanism (Peng et al., 2017; Wolfson et al., 2017).
Our gene deletion experiments found that the null mutants of the Gtr1-Gtr2 GTPases, Ragulator (Lam1, Lam2, Lam3 and Lam4) and GATOR1 (Iml1, Npr2 and Npr3) exhibit very similar growth defects, which are rescued by the TORC1 inhibitors (Figure 5G, Figure 5—figure supplement 1A). The double mutant analyses confirmed that they all belong to the same epistasis group, consistent with the idea that both Ragulator and GATOR1 complexes function through the Gtr1-Gtr2 heterodimer. Together with other data presented in this paper, we propose that the Gtr1GDP-Gtr2 heterodimer tethered to the vacuolar surface by the Ragulator complex plays a role in moderating TORC1 activation induced by the Rhb1 GTPase. Although TORC1 activity is essential for cell viability in fission yeast (Álvarez and Moreno, 2006; Matsuo et al., 2007; Weisman and Choder, 2001), its attenuation by Gtr1GDP-Gtr2 enables appropriate response to limitation in nitrogen source, allowing localization of the amino acid transporters to the plasma membrane (Figures 6 and 7). It is currently unknown how Gtr1GDP-Gtr2 counteracts the Rhb1-dependent activation of TORC1. In amino-acid-starved mammalian cells, Tsc2 is recruited to lysosomal membranes by the Rag GTPases and inhibits Rheb, resulting in inactivation of mTORC1 (Demetriades et al., 2014). However, such a Tsc2-dependent mechanism cannot explain the observations that the effects of the gtr1∆ and tsc2∆ mutations are additive (Figures 5L and 6G) and that the Gtr1-dependent attenuation of TORC1 appears to be in effect even in the absence of Rhb1 (Figure 6J).
It remains possible that GTP-bound Gtr1 is capable of promoting TORC1 activation, as has been found for GTP-bound RagA/B in higher eukaryotes (Kim et al., 2008; Sancak et al., 2008; Sancak et al., 2010). However, such functionality of Gtr1GTP would be cryptic in S. pombe, where TORC1 can be strongly activated in the gtr1 null mutant as well as in strains expressing mostly Gtr1GTP, such as gtr1Q61L and GATOR1 mutants (Figure 6). In this organism, TORC1 localization to vacuolar membranes is not dependent on the Ragulator-Gtr complex (Figure 3C), suggesting Gtr-independent targeting of TORC1 to the vacuolar surface where the Rhb1 GTPase activates TORC1. Rag-independent recruitment of mTORC1 to lysosomal membranes has also been reported (Jewell et al., 2015), and there might be a conserved mechanism for such TORC1 localization. Moreover, it is conceivable that GDP-bound RagA/B also plays a role in attenuation of mTORC1 signaling, a possibility compatible with the observed hyperactivation of mTORC1 in the absence of functional GATOR1 on lysosomes (Bar-Peled et al., 2013; Peng et al., 2017; Wolfson et al., 2017). Because deregulated mTORC1 activation is implicated in cancerous cell proliferation and neurological disorders including epilepsy (Huang and Manning, 2008; Baldassari et al., 2016), the molecular mechanisms that negatively regulate mTORC1 warrant further investigation.
S. pombe strains used in this study are listed in the Supplementary File 1. Growth media and genetic manipulations for S. pombe have been described previously (Moreno et al., 1991; Shiozaki and Russell, 1997; Bähler et al., 1998). S. pombe cells were grown in yeast extract medium YES and synthetic minimal medium EMM. For nitrogen starvation experiments, cells exponentially growing in EMM liquid medium were filtered onto 0.45 μm mixed cellulose ester membrane (Advantec, Japan) and resuspended in liquid EMM lacking ammonium chloride (EMM–N). More than two biological replicates were tested for each experiment. For cell size determination, EMM liquid cultures from three biological replicates were diluted 500 times with the Cellpack Reagent (Sysmex, Japan), and cell volumes were measured using the Particle Analyzer CDA-1000 (Sysmex).
For spot test assay, EMM liquid cultures were adjusted to cell density equivalent to OD600 1.0. Serial dilutions of the cells were spotted onto solid media. Images were captured by the LAS-4000 system (Fujifilm, Japan).
Cells were harvested in 10% trichloroacetic acid (TCA) and crude cell lysates were prepared as described previously (Tatebe and Shiozaki, 2003). Protein concentration was determined by the Bio-Rad protein assay kit (Bio-Rad, Hercules, California). Proteins were resolved by SDS-PAGE, transferred to nitrocellulose, and probed with antibodies as follows: anti-phospho-p70 S6K (Cell Signaling Technology, Danvers, Massachusetts), anti-Psk1 (custom antibody raised in rabbits against the peptide ‘SDDEIAEEGYDFEELEAS’) (Sigma, St. Louis, Missouri), anti-Spc1 (Tatebe and Shiozaki, 2003), anti-FLAG (M2, Sigma), anti-myc (9E10, Covance, Princeton, New Jersey; A-14, Santa Cruz Biotechnolgy, Dallas, Texas), and anti-GFP (Roche, Switzerland).
For immunoprecipitation, cells were grown to exponential phase in YES and filtered onto 0.45 μm mixed cellulose ester membrane (Advantec). The cells were disrupted in lysis buffer composed of 20 mM HEPES-KOH (pH 7.5), 150 mM potassium glutamate, 10% glycerol, 0.25% Tween-20, 10 mM sodium fluoride, 10 mM p-nitrophenylphosphate, 10 mM sodium pyrophosphate, 10 mM β-glycerophosphate, and 0.1 mM sodium orthovanadate containing PMSF, leupeptin, and protease inhibitor cocktail (Sigma) with glass beads using Multi-beads Shocker (Yasui Kikai, Japan). Anti-FLAG magnetic beads (Sigma) and anti-myc affinity gel (Sigma) were used to immunoprecipitate FLAG- and myc-tagged proteins, respectively. For serial immunoprecipitation, cell lysates were first subjected to anti-FLAG immunoprecipitation, followed by elution using 3X FLAG peptides (Sigma) and anti-myc immunoprecipitation.
Samples to be analyzed by mass spectrometry were first resolved in a 12% Mini-PROTEAN TGX precast gel (Bio-Rad). Each lane was sliced into four pieces and digested with trypsin. Mass spectrometric analysis was performed using the LTQ-Orbitrap XL-HTC-PAL system. MS/MS spectra were analyzed by Mascot server (Matrix Science) and compared against NCBInr protein database (Taxonomy: S. pombe).
Fluorescence microscopic analysis was carried out using DeltaVision Elite Microscopy System (GE Healthcare, United Kingdom). Cells grown exponentially in EMM liquid were stained with FM4-64 dye (Invitrogen, Carlsbad, California; Biotium, Fremont, California) for vacuole visualization. Z-axial images were taken at 0.4 µm with a 100X objective lens. Deconvolution of images was performed using DeltaVision SoftWoRx software. More than 200 cells were analyzed in each microscopy experiment.
Fission yeast Tor2 promotes cell growth and represses cell differentiationJournal of Cell Science 119:4475–4485.https://doi.org/10.1242/jcs.03241
The Tsc/Rheb signaling pathway controls basic amino acid uptake via the Cat1 permease in fission yeastMolecular Genetics and Genomics 279:441–450.https://doi.org/10.1007/s00438-008-0320-y
GATOR1 complex: the common genetic actor in focal epilepsiesJournal of Medical Genetics 53:503–510.https://doi.org/10.1136/jmedgenet-2016-103883
ULK1.ATG13.FIP200 complex mediates mTOR signaling and is essential for autophagyJournal of Biological Chemistry 284:12297–12305.https://doi.org/10.1074/jbc.M900573200
Amino acid sufficiency and mTOR regulate p70 S6 kinase and eIF-4E BP1 through a common effector mechanismJournal of Biological Chemistry 273:14484–14494.https://doi.org/10.1074/jbc.273.23.14484
The TSC1-TSC2 complex: a molecular switchboard controlling cell growthBiochemical Journal 412:179–190.https://doi.org/10.1042/BJ20080281
Rheb GTPase is a direct target of TSC2 GAP activity and regulates mTOR signalingGenes & Development 17:1829–1834.https://doi.org/10.1101/gad.1110003
ULK-Atg13-FIP200 complexes mediate mTOR signaling to the autophagy machineryMolecular Biology of the Cell 20:1992–2003.https://doi.org/10.1091/mbc.E08-12-1249
Regulation of TORC1 by Rag GTPases in nutrient responseNature Cell Biology 10:935–945.https://doi.org/10.1038/ncb1753
Rag GTPases are cardioprotective by regulating lysosomal functionNature Communications 5:4241.https://doi.org/10.1038/ncomms5241
Dynamic relocation of the TORC1-Gtr1/2-Ego1/2/3 complex is regulated by Gtr1 and Gtr2Molecular Biology of the Cell 27:382–396.https://doi.org/10.1091/mbc.E15-07-0470
Structural conservation of components in the amino acid sensing branch of the TOR pathway in yeast and mammalsJournal of Molecular Biology 402:388–398.https://doi.org/10.1016/j.jmb.2010.07.034
Isp7 is a novel regulator of amino acid uptake in the TOR signaling pathwayMolecular and Cellular Biology 34:794–806.https://doi.org/10.1128/MCB.01473-13
Loss of the TOR kinase Tor2 mimics nitrogen starvation and activates the sexual development pathway in fission yeastMolecular and Cellular Biology 27:3154–3164.https://doi.org/10.1128/MCB.01039-06
Amino acids activate mammalian target of rapamycin (mTOR) complex 1 without changing Rag GTPase guanyl nucleotide chargingJournal of Biological Chemistry 289:2658–2674.https://doi.org/10.1074/jbc.M113.528505
Identification of Cdc37 as a novel regulator of the stress-responsive mitogen-activated protein kinaseMolecular and Cellular Biology 23:5132–5142.https://doi.org/10.1128/MCB.23.15.5132-5142.2003
The Saccharomyces cerevisiae Rheb G-protein is involved in regulating canavanine resistance and arginine uptakeJournal of Biological Chemistry 275:11198–11206.https://doi.org/10.1074/jbc.275.15.11198
The Vam6 and Gtr1-Gtr2 pathway activates TORC1 in response to amino acids in fission yeastJournal of Cell Science 125:1920–1928.https://doi.org/10.1242/jcs.094219
Tsc1+ and tsc2+ regulate arginine uptake and metabolism in Schizosaccharomyces pombeThe Journal of Biological Chemistry 279:12706–12713.https://doi.org/10.1074/jbc.M313874200
The fission yeast TOR homolog, tor1+, is required for the response to starvation and other stresses via a conserved serineThe Journal of Biological Chemistry 276:7027–7032.https://doi.org/10.1074/jbc.M010446200
Selective regulation of autophagy by the Iml1-Npr2-Npr3 complex in the absence of nitrogen starvationMolecular Biology of the Cell 22:4124–4133.https://doi.org/10.1091/mbc.E11-06-0525
Roger J DavisReviewing Editor; University of Massachusetts Medical School, United States
In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.
Thank you for submitting your article "GDP-bound form of the Rag family GTPase Gtr1 counteracts the Rheb-dependent activation of TOR complex 1" for consideration by eLife. Your article has been favorably evaluated by Philip Cole (Senior Editor) and three reviewers, one of whom is a member of our Board of Reviewing Editors.
The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.
Mammalian cell studies have led to a model in which TORC1 activation by amino acids is triggered by its recruitment to the RagA/B GTPase heterodimer that is anchored to lysosomal membranes by the multi-subunit Ragulator complex, which functions as a guanine nucleotide exchange factor (GEF). These activities are counteracted by GATOR, which has GAP activity toward RagA/B to cause the release of TORC1 from lysosomal membranes. Chia et al. identify the Ragulator-like complex in fission yeast and show that it tethers the Gtr1-Gtr2 Rag heterodimer to vacuolar membranes. However, in contrast to mammals, TORC1 localizes to vacuolar membranes independently of the Ragulator-Rag complex. Moreover, Chia et al. report that Gtr1-GDP acts to suppress TORC1 activity.
This is an interesting and well-executed study of the Rag pathway in fission yeast that provides important new insights relevant to the evolution and function of this system of regulation and amino acid sensing at the vacuole/lysosome. The difference in Rag function between mammals and fission yeast is intriguing and suggests that adaptation has occurred in each organism to meet the regulatory needs for the control of TORC1 activity.
1) The title of the paper should be changed to better reflect the study. First, the title should note that the analysis was performed with fission yeast (or S. pombe). Second, while Gtr1-GDP is identified as a suppressor of TORC1 activity, the link to Rheb is not fully established and should not be stated in the title. The title should reflect the more balanced approach to stating conclusions employed by the authors in the Abstract.
2) "Growth" in all of the genetic experiments in the study is classified as a single cellular process/phenotype. Are growth defects from too much or too little TORC1 activity really the same (proliferation effects, cell cycle, cell size)? Given that TORC1 is established to promote cell growth, rather than impair it, these cellular phenotypes are worth a more careful analysis.
3) The presumptive effects on amino acid uptake in the final section of the study are interesting and might provide insight into the growth defect of the mutants analyzed. The authors should directly test whether elevated TORC1 signaling causes a defect in the uptake or utilization of exogenous amino acids in the context of these mutants.
4) Figure 7A. Is the rescue by ammonium due to restoration of nitrogen or attenuation of TORC1?https://doi.org/10.7554/eLife.30880.023
- Tomoyuki Fukuda
- Tomoyuki Fukuda
- Tomoyuki Fukuda
- Kazuhiro Shiozaki
- Kim Hou Chia
- Fajar Sofyantoro
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
We thank M Uritani, T Matsumoto, and the NBRP Japan for reagents. We also thank T Toda and H Tatebe for insightful comments and L Takeuchi, A Hishinuma, and R Kurata for technical assistance.
- Roger J Davis, Reviewing Editor, University of Massachusetts Medical School, United States
© 2017, Chia et al.
This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.