Spatially distinct pools of TORC1 balance protein homeostasis

ABSTRACT

In eukaryotes, TORC1/MTORC1 is a critical regulator of growth and proliferation. In response to nutrient abundance TORC1/MTORC1 favors anabolic processes and retards degradative ones. In S. cerevisiae, TORC1 is conventionally known to localize on the vacuolar membrane. In the course of their recent investigations, Hatakeyama et al. discovered a novel second site of TORC1 localization— the prevacuolar endosome. Their article, highlighted here, discusses the mechanism of TORC1 localization to the prevacuolar endosome and highlights a hitherto unappreciated mechanism by which 2 spatially separated pools of TORC1 execute the distinct functions of promoting anabolism and inhibiting degradation.

KEYWORDS:

• Autophagy
• lysosome
• MTOR
• stress
• vacuole
• yeast

Living organisms accumulate mass by the assimilation of nutrients. An abundance of nutrients promotes growth and cellular proliferation, whereas a dearth of nutrients delays them. In complex multicellular eukaryotes, growth and proliferation are also regulated by neuroendocrine circuits, which convey the organismal nutritional status to individual tissues [1]. In eukaryotes, the integration of nutritional cues with cellular growth and proliferation is mediated by a Ser/Thr kinase of the PtdIns3K-related kinase family known as TOR (target of rapamycin) in Saccharomyces cerevisiae or MTOR (mechanistic target of rapamycin kinase) in mammals [2].

Figure 1. Two distinct pools of TORC1 regulate anabolic and catabolic processes. Amino acids simulate both the vacuolar pool and a novel endosomal pool of TORC1 via the small GTPases Gtr1 and Gtr2 (indicated by solid black arrows). Upon activation TORC1 (designated as ‘ON’) promotes general anabolism and protein synthesis while concomitantly downregulating degradative processes. The vacuolar pool of TORC1 is primarily involved in upregulating translation by phosphorylating the AGC-family kinase Sch9. The inhibition of degradative processes is largely mediated by the endosomal pool of TORC1 that phosphorylates Atg13 and Vps27 to inhibit macroautophagy and microautophagy, respectively (Adapted from [16]).

In S. cerevisiae, there are 2 TOR genes, and the gene products assemble into either of 2 distinct subcellular complexes — Tor complex 1 (TORC1) and TORC2 [3]. These 2 complexes share common subunits but also contain complex-specific components; along these lines, TORC2 only harbors Tor2, whereas TORC1 preferentially contains Tor1. Similarly, in mammals, MTOR is a component of both MTORC1 and MTORC2 that contain shared and unique subunits [4]. TORC1/MTORC1 is the better characterized of the 2 complexes, is more sensitive to the macrocyclic lactone rapamycin, and has been implicated in regulating growth by promoting anabolism and preventing catabolic processes such as macroautophagy [5].

How TORC1 integrates nutritional cues and growth signals at the cellular level has been the focus of intense research, and several proteins and protein complexes involved in the regulation of TORC1/MTORC1 activity have been identified. In mammalian cells, inactive MTORC1 is cytosolic but is recruited to the lysosomal surface upon nutrient stimulation. This recruitment is dependent on the Ragulator-RRAG complex, which contains the membrane binding component LAMTOR1 and a heterodimer of small GTPases RRAGA or RRAGB with RRAGC or RRAGD. For recruiting MTORC1 to the lysosome the Ragulator complex needs to be in an active conformation where RRAGA/B is GTP-loaded, and RRAGC/D is GDP-bound [6]. Activation of the RRAG complex is promoted by the stimulation of molecular sensors by certain amino acids such as leucine, arginine and glutamine. Most of these sensors regulate the RRAG complex through upstream GTPase activating proteins (GAPs) of the GATOR complexes [7–9]. The intricacy in MTORC1 activation is highlighted by the recent finding that lysosomal lumenal arginine levels also influence MTORC1 [10]. Recruitment of MTORC1 to the lysosomal membrane is not sufficient for activation, however, and depends on a second stimulation by the small GTPase RHEB which also localizes on the lysosomal membrane. In the absence of growth factors such as INS (insulin) or IGF1, RHEB is inactivated by the TSC complex which acts as a GAP. Upon growth factor stimulation, TSC activity is inhibited allowing RHEB to be loaded with GTP and activated, whereupon it facilitates the complete activation of MTORC1 [11,12].

The mechanism of activation of TORC1 in S. cerevisiae is similar. In yeast, the EGO complex (EGOC) functions analogous to the Ragulator-RRAG complex and contains the membrane binding component Meh1/Ego1, Ego2 and Slm4/Ego3, which recruit a heterodimer of the small GTPases Gtr1 (analogous to RRAGA/B) and Gtr2 (analogous to RRAGC/D) to the vacuole membrane. Yeast TORC1 is, however, primarily activated by amino acid stimulation. Also, unlike MTORC1, yeast TORC1 is constitutively localized on the vacuolar membrane. From its vacuolar location, activated TORC1 is proposed to promote protein translation by phosphorylating the AGC-kinase family protein Sch9 (a putative homolog of the mammalian RPS6KB kinase) [13]. TORC1 also inhibits macroautophagy by phosphorylating Atg13, although the precise subcellular location of this interaction is unclear [14]. However, whether yeast TORC1 may exist in other cellular locations is an intriguing question, especially because several subcellular locations for MTORC1 have been sporadically suggested in the literature [15]. In the research highlighted here, Hatakeyama and colleagues report the existence of a second pool of TORC1 and propose that the 2 pools of TORC1 that localize in distinct cellular compartments execute different functions [16] (Figure 1).

Previously, the De Virgilio group had reported that yeast expressing a nucleotide-binding mutant of Gtr1 (Gtr1^S20L) exhibit a significant growth defect when cultured in complete medium. This phenotype was dependent on the presence of the TORC1 subunit Tco89, indicating that the nucleotide-binding status of Gtr1 can either promote or inhibit growth [17]. Their present analysis was initiated by a screen to identify mutations that suppress this growth defect. Among the suppressor mutations, a mutation in Tco89 was identified, as expected. A point mutation in Tor1 was also found that promises to be an exciting avenue for future study. A constellation of suppressor mutations were identified in the EGOC components, including Meh1/Ego1, as well as the Golgi-resident Akr1 that palmitoylates Meh1/Ego1 allowing it to bind the vacuolar membrane. These findings indicate that the EGOC needs to be appropriately assembled for Gtr1 to function and hence for the Gtr1^S20L mutant to exhibit a growth defect. However, how the EGOC components are delivered to the vacuolar membrane for assembly was yet undefined.

The AP3-HOPS pathway transports proteins selectively from the trans-Golgi network (TGN) to the vacuole [18]. Because Meh1/Ego1 is palmitoylated at the TGN, and suppressor mutations were identified in multiple AP3-HOPS pathway components, the authors test whether this pathway is involved in targeting EGOC components to the vacuolar membrane. In wild-type cells EGOC members such as Gtr1 and Meh1/Ego1 localize to the vacuolar membrane, as expected. Using fluorescence microscopy, the authors show that in cells lacking either AP3 pathway proteins (apl5∆ or apl6∆ cells) or HOPS complex components (vam6∆ or vps41∆ cells) Meh1/Ego1 and Gtr1 exhibit a partial redistribution to the plasma membrane, whereas Tor1 still localizes to the vacuolar membrane. The partial redistribution highlights the importance of the AP3-HOPS pathway in Meh1/Ego1 and Gtr1 transport, but also indicates that a second protein-targeting pathway may be compensating for the loss, with the Prc1/carboxypeptidase Y (CPY) pathway [19] being a prime candidate. Indeed, the vacuolar localization of Meh1/Ego1 is completely lost in cells lacking both the AP3 component Apl5 and the CPY pathway t-SNARE Pep12. Furthermore, mutating the di-leucine-containing signal sequence on Meh1/Ego1 critical for recognition by the AP3 adaptor to a di-alanine (Ego1[LLAA]) also leads to a redistribution of Meh1/Ego1 almost exclusively to the plasma membrane.

Imaging analysis revealed that in addition to their localization on the vacuolar membrane, Meh1/Ego1-GFP and GFP-Gtr1 also exhibit a punctate perivacuolar localization reminiscent of endosomes. GFP-Tor1 exhibits a similar distribution pattern. This localization is retained when the AP3-HOPS and CPY pathways are manipulated. The authors, therefore, test whether the formation of these perivacuolar foci is dependent on AP1 or the monomeric GGA adaptors, which transport cargo from the TGN to endosomes [20]. Loss of AP1 components Apl2 and Apl4 does not abolish the foci, but the combined depletion of Gga1 and Gga2 results in the loss of the perivacuolar Meh1/Ego1, Gtr1 and Tor1 puncta. In addition, the authors find that these puncta almost exclusively colo-calize with mCherry-tagged Vps21 and Vps27, 2 well-defined endosomal markers, as well as a third endosomal protein, Pib2 — a FYVE-domain containing protein previously known to promote the tethering of TORC1 to the vacuolar membrane. These findings led the authors to propose that in addition to the well-established localization of TORC1 on the vacuole, a pool of EGOC, and hence TORC1, also localizes to what are defined as prevacuolar endosomes.

The identification of 2 distinct pools of TORC1 begs the question of whether both TORC1 assemblies are sensitive to nutritional cues. To test this, Hatakeyama et al. generated protein chimeras that target the C terminus of the TORC1 kinase substrate Sch9, containing the target residue T737, to either the endosomal or the vacuolar membrane. The endosome-targeting chimera (ET) was designed to contain a FYVE domain for binding PtdIns3P, a lipid enriched in endosomal membranes, followed by GFP and the Sch9 fragment. The vacuole-targeting chimera (VT) is composed of the N-terminal transmembrane domain of the vacuolar Pho8 phosphatase, which follows the Sch9 fragment and GFP. The authors find that upon glutamine-mediated activation, both the ET and VT chimeras, along with endogenous Sch9, are phosphorylated indicating that both endosomal and vacuolar pools of TORC1 are functional and responsive to nutrient stimulation. Phosphorylation of both ET and VT is dependent on the presence of Gtr1 and Pib2, and phosphorylation intensities correlate with qualitative and quantitative changes in the amino acid stimulus.

The authors next wanted to discern whether these 2 spatially separated pools of TORC1 are functionally autonomous. To make this assessment requires the selective inactivation of one pool of TORC1 and the subsequent monitoring of the activity of the other pool. In these assays, the authors utilized glutamine stimulation to test TORC1 activity. The deletion of the gene encoding the vacuolar ATPase component Vph1 specifically inhibits the activity of the vacuolar pool of TORC1, identified by the absence of the phosphorylation of the VT reporter, but does not affect the activity of the endosomal pool because the ET reporter remains phosphorylated. The phosphorylation of endogenous Sch9 also decreases substantially in vph1∆ cells, indicating that the vacuolar pool of TORC1 is critical for Sch9 phosphorylation. The combined deletion of the VPH1 and the endosomal v-ATPase subunit-encoding STV1 genes leads to the complete loss of endogenous Sch9 as well as both ET and VT reporter phosphorylation, indicating that these 2 pools represent the functionally active TORC1 assemblies. In a complementary approach, the expression of an endosome-localizing FYVE-domain tagged hyperactive Tor1^I1954V allele (FYVE-Tor1^I1954V) in a tor1∆ background results in the phosphorylation of ET but not VT, indicating that the endosomal pool of TORC1 can function autonomously. To test whether the vacuolar pool of TORC1 is also autonomously functional the authors utilized the internally tagged Tor1[D330]-GFP allele, which they discovered localizes exclusively on the vacuolar membrane. Yeast expressing Tor1[D330]-GFP [21] exhibit abundant phosphorylation of VT but strongly reduced ET phosphorylation, highlighting the concept that the vacuolar pool of TORC1 can function independently of the endosomal pool. Finally, the loss of the vacuolar TORC1 subunit Tco89 completely abolishes VT phosphorylation in the presence or absence of glutamine, consistent with previous literature. However, in tco89∆ cells, ET phosphorylation is detected during growth on proline (in the absence of glutamine), but phosphorylation levels do not increase upon the administration of glutamine. This indicates that while Tco89 completely controls the activity of vacuolar TORC1, it may only partially regulate endosomal TORC1 activity.

The endosomal localization of TORC1 lends credence to a hypothesis that TORC1 may regulate biological processes that occur at/originate from this compartment. One likely candidate is the temporal regulation of the ESCRT complex for multivesicular body (MVB) formation at the endosome or the utilization of ESCRT machinery at the vacuolar membrane for microautophagy [22]. Because the authors find that the ESCRT-0 component Vps27 colocalizes with EGOC on prevacuolar endosomes, they investigated whether Vps27 is a direct TORC1 substrate. They find that purified TORC1 can phosphorylate Vps27 in vitro. Mass spectrometry of in vitro phosphorylated Vps27 complemented with phosphoproteomic SILAC analysis using differential TORC1 activation allowed the authors to identify Ser/Thr residues on Vps27 that are directly phosphorylated by TORC1. Consistent with these findings, phosphorylated forms of Vps27 are detected by SDS-PAGE analysis using Phos-Tag gels that specifically retard migration of phosphorylated moieties. Under normal conditions, Vps27 migrates as 4 discrete bands in a Phos-tag matrix with greater TORC1 activation increasing the population of the slower migrating (hyperphosphorylated) bands, and TORC1 inactivation enhancing the intensity of the faster migrating (hypophosphorylated/unphosphorylated) species. The authors then proceed to show, using this technique, that the selective endosomal targeting of the FYVE-Tor1^I1954V allele in tor1Δ cells promotes Vps27 phosphorylation. Conversely, Vps27 remains predominantly hypophosphorylated when the vacuolar membrane-localizing Tor1[D330]-3GFP allele is expressed, indicating that the endosomal pool of TORC1 is responsible for Vps27 phosphorylation. Finally, the authors show that microautophagy may be regulated by the TORC1-dependent phosphorylation status of Vps27. Their assay for microautophagy follows the degradation of the GFP-tagged vacuolar membrane protein Pho8 (GFP-Pho8), determined by immunoblotting as the concomitant appearance of the free GFP moiety. Upon rapamycin treatment, yeast expressing a mutant of Vps27 (Vps27[7D]) that mimics the TORC1-catalyzed hyperphosphorylated state show reduced free GFP generation compared to yeast expressing wild-type Vps27. The authors also confirm the direct role of Vps27 phosphorylation status in this process by monitoring the translocation of Vps27 to the vacuolar membrane to initiate microautophagy and its subsequent internalization within the lumen by the same process. Indeed, the phosphomimetic Vps27[7D] mutant shows reduced internalization compared to wild-type Vps27 as measured by the generation of free pHlourin moieties from pHlourin-Vps27 in rapamycin-treated cells. These findings are notable, because they establish Vps27 as a novel TORC1 substrate and a proximal effector of TORC1 signaling.

Finally, Hatakeyama and colleagues show that endosome-localized TORC1 also phosphorylates Atg13, a protein critical for the initiation of macroautophagy. Atg13 is a well-characterized TORC1 substrate, but the subcellular location of the TORC1-Atg13 interaction was previously unclear. Atg13 phosphorylation by TORC1 inhibits macroautophagy. The authors suggest that the phosphorylation likely occurs at or near the prevacuolar endosomes where this pool of TORC1 is ideally poised to respond to nutrient availability and regulate macroautophagy. Consistent with this idea, the expression of endosome-localizing FYVE-Tor1^I1954V in tor1∆ cells promotes Atg13 phosphorylation more efficiently than the expression of the vacuole-localizing Tor1[D330]-3GFP allele.

This work from Hatakeyama et al. adds another layer to the conventional knowledge regarding yeast TORC1 because it uncovers a hitherto unidentified prevacuolar endosomal localization of TORC1. They also provide a molecular basis by which the selective sorting of the EGOC components, which recruit Tor1/Tor2 to these spatially distinct locations, occurs. The authors propose a division of labor between the 2 pools with the vacuolar TORC1 more intimately connected with the regulation of protein translation via Sch9 phopshorylation, and the endosomal pool primarily concerned with the inhibition of microautophagy and macroautophagy. As a part of their investigations they also identify a novel TORC1 target, Vps27, that acts as a proximal TORC1 effector. Further analysis regarding the functional relevance of these pools of TORC1 will shed more light on the mechanisms by which TORC1 integrates nutritional information with growth responses in eukaryotic cells.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by the National Institute of General Medical Sciences [GM053396].