Amino acid homeostatic control by TORC1 in Saccharomyces cerevisiae under high hydrostatic pressure

Microorganisms have evolved intricate sensing systems to detect and rapidly respond to changes in a variety of environmental factors such as nutrients, pH, temperature, or osmolarity. Organisms also encounter a variety of mechanical stresses such as hydrostatic pressure, tension, hydrodynamic shear, or compression. In mammals, a small increase in the tension on skeletal muscles at millinewton levels can activate protein synthesis with enhanced translational efficiency through the activation of mammalian target of rapamycin (TOR) complex 1 (TORC1) (Hornberger et al., 2004; You et al., 2012). Unlike stretching or tension, hydrostatic pressure is the application of uniform stress without cellular deformation. In the joint, cartilage is typically exposed to hydrostatic pressures between 3 and 10 MPa (Afoke et al., 1987), with stress as high as 18 MPa in the hip joint (Hodge et al., 1989) (atmospheric pressure is nearly equal to 0.1 MPa = 1 bar = 0.9869 atm = 1.0197 kg/cm²; to avoid confusion, MPa is used throughout). Hydrostatic pressure at physiological levels, up to 50 MPa, has been widely used as a tool for mechanical stimulation in tissue engineering and as a method of differentiating cells towards a chondrogenic phenotype (Elder and Athanasiou, 2009; Natenstedt et al., 2015). Microorganisms also have the ability to sense and respond to hydrostatic pressure for survival (Abe, 2007; Aertsen and Michiels, 2005; Marietou et al., 2015). However, little is known regarding how living cells sense such stimuli and convert them into intracellular biochemical responses.

To elucidate the underlying molecular mechanisms for high-pressure sensing and adaptation, we have previously explored genes required for the growth of Saccharomyces cerevisiae under high pressure, applying functional genomic strategies (Abe and Minegishi, 2008; Kurosaka et al., 2019). Our findings emphasized that amino acids and/or nitrogen availability are a key factor in providing fundamental insights into the cellular events occurring under non-lethal levels of high pressure (<50 MPa). Cell growth under high pressure primarily depends on the availability of tryptophan, and thus tryptophan auxotrophic strains are highly sensitive to high pressure (Abe and Horikoshi, 2000; Abe and Iida, 2003). Using the yeast deletion library, we identified 84 mutants defective in growth under high pressure (25 MPa at 25°C) and/or low temperature (0.1 MPa at 15°C) (Abe and Minegishi, 2008). Among them, mutants lacking any one of the following proteins: Ego1 (also known as Meh1), Ego3 (also known as Slm4), Gtr1, or Gtr2, which play pivotal roles in the TORC1 signaling pathway, have severe growth defects under pressure in the BY4742 strain background (leu2 his3 lys2 ura3) (Abe and Minegishi, 2008). Defects in growth under high pressure (25 MPa) are rescued for 24 of the identified deletion mutants by the introduction of a combination of four plasmids (LEU2, HIS3, LYS2 and URA3), suggesting close links between the genes and permease-dependent nutrient uptake (Kurosaka et al., 2019). However, the growth defects of the ego1Δ, ego3Δ, gtr1Δ and gtr2Δ mutants are not rescued by introduction of the four plasmids (Kurosaka et al., 2019). Taken together, our previous findings suggest that high-pressure signaling and nutrient sensing converge on common cellular pathways that give rise to the regulation of TORC1 and amino acid homeostasis.

TOR is a serine/threonine kinase that is evolutionarily conserved across eukaryotes. It regulates cellular homeostasis by coordinating anabolic and catabolic processes in the presence of nutrients (Dokudovskaya and Rout, 2015; González and Hall, 2017; Heitman et al., 1991; Lim and Zoncu, 2016; Powis and De Virgilio, 2016). TOR forms two structurally and functionally different conserved complexes termed TOR complex 1 and 2 (TORC1 and TORC2, respectively) in S. cerevisiae (Eltschinger and Loewith, 2016; González and Hall, 2017). TORC1 is composed of one of two alternative catalytic subunits, Tor1 and Tor2, and the regulatory subunits Lst8, Kog1 and Tco89 (Loewith et al., 2002; Reinke et al., 2004). TORC1 partially localizes to the vacuolar membrane where it directly phosphorylates Sch9, Gln3, and Atg13 (Bertram et al., 2000; Jiang and Broach, 1999; Kamada et al., 2000; Urban et al., 2007). In the presence of favorable nutrients, TORC1 promotes growth by activating multiple steps in protein and ribosome synthesis, in part through direct phosphorylation of the protein kinase Sch9 (Urban et al., 2007). In contrast, when cells are starved for nutrients, or exposed to toxic stresses, TORC1 signaling is repressed, leading to a downregulation of protein synthesis, effectively arresting growth (Brauer et al., 2008; Gasch et al., 2000; Powers and Walter, 1999; Urban et al., 2007). The mechanisms underlying TORC1 inhibition under nitrogen starvation are governed by members of the conserved Rag family of small GTPases, Gtr1 and Gtr2. Nitrogen starvation triggers the GTP-binding state of Gtr1 and Gtr2 (Powis and De Virgilio, 2016; Shimobayashi and Hall, 2016). The EGO complex (EGOC) is composed of Ego1, Ego2 and Ego3, and tethers Gtr1 and Gtr2 to the vacuolar membrane in preparation for TORC1 stimulation (Dubouloz et al., 2005; Kira et al., 2016; Nicastro et al., 2017; Zhang et al., 2019). Pib2 is a recently identified regulator of TORC1 and localizes to the vacuolar membrane via binding of its FYVE domain to phosphatidylinositol 3-phosphate (Kim and Cunningham, 2015).

Despite being involved in responses to nitrogen and amino acid starvation, Npr2 and Npr3 [components of the GTPase-activating protein (GAP) complex, SEACIT], and Gtr1 and Gtr2, have little role in transmitting other environmental stress signals, including glucose starvation, osmotic stress, heat stress and oxidative stress, to TORC1 (Binda et al., 2009; Hughes Hallett et al., 2014). Glucose starvation inhibits TORC1 in cells constitutively expressing the active form of Gtr1^GTP–Gtr2^GDP, suggesting the dispensability of Gtr1 and Gtr2 in this TORC1 inhibition (Hughes Hallett et al., 2015). Instead, the AMP-activated protein kinase, Snf1 is required for TORC1 inhibition in response to glucose starvation. The MAP kinase Hog1 inhibits signaling through the TORC1 pathway under osmotic stress. Overexpression of Pbp1, a poly(A)-binding protein (Pab1)-binding protein and a component of stress granules (SGs) (Swisher and Parker, 2010), represses TORC1 by inducing SG formation, and this process is independent of Npr2 and Npr3 (Takahara and Maeda, 2012). Thus, some stress signaling pathways drive TORC1 into each of its signaling states in a way different from nitrogen starvation signaling. In contrast, the effect of high hydrostatic pressure at non-lethal levels (for example, 25 MPa, as commonly used in our experiments) in S. cerevisiae has similarities to nitrogen starvation in many aspects (Abe and Horikoshi, 2000; Abe and Iida, 2003; Abe, 2007; Abe and Minegishi, 2008; Suzuki et al., 2013). This prompted us to further investigate the role of TORC1 in amino acid homeostatic control under high pressure.

In the present study, we demonstrated that high pressure causes aberrant accumulation of intracellular glutamine, and that Pib2 loaded with glutamine may stimulate TORC1 to downregulate glutamine synthesis. Under these specific stress conditions, TORC1 is thought to localize to the vacuolar membrane through EGOC-dependent recruitment.

We first examined the role of upstream regulators of TORC1 in growth under high pressure. In addition to Ego1, Ego3, Gtr1 and Gtr2 (Abe and Minegishi, 2008), Ego2, a newly identified subunit of EGOC (Kira et al., 2016; Powis et al., 2015; Zheng and Jiang, 2005), was also required for growth at 25 MPa (Fig. 1A). Vam6 is a guanine-nucleotide-exchange factor for Gtr1, and the Lst4–Lst7 complex functions as a GAP for Gtr2 (Binda et al., 2009; Peli-Gulli et al., 2015). The vam6Δ, lst4Δ and lst7Δ mutants displayed marked high-pressure sensitivity at 25 MPa (Fig. 1A). Gtr1 and Gtr2 form a heterodimer (Gong et al., 2011; Jeong et al., 2012), and two guanine-nucleotide-bound forms, Gtr1^GTP and Gtr2^GDP, activate TORC1 in response to nutrient status (Binda et al., 2009; Panchaud et al., 2013). Guanine-nucleotide-binding forms can be reverted by critical amino acid substitutions in Gtr1 and Gtr2 (Boguski and McCormick, 1993; Nakashima et al., 1999). We confirmed that the expression of the constitutively active forms Gtr1^GTP (Q65L) or Gtr2^GDP (S23L) restored the ability to grow under high pressure in the gtr1Δ or gtr2Δ mutants, respectively, whereas the constitutively inactive forms Gtr1^GDP (S20L) or Gtr2^GTP (Q66L) did not (Fig. 1B). These results suggest that the activation of Gtr1 and/or Gtr2 via the functionality of these protein modifiers is necessary for growth under high pressure.

Deletions of TOR1, TCO89 and PIB2 also appeared to result in high-pressure sensitivity at 25 MPa (Fig. 2A). It has been demonstrated that the pib2Δ mutant exhibits either synthetic lethality or strong fitness defects when combined with any of the ego1Δ, ego2Δ, ego3Δ, gtr1Δ, or gtr2Δ mutations (Kim and Cunningham, 2015). Therefore, high pressure might attenuate both EGOC- and Pib2-dependent stimulatory pathways of TORC1, and hence, any defect in one of the two pathways results in high-pressure sensitivity.

N-terminally GFP-tagged Tor1 was visualized under a confocal microscope after incubation of the cells at 0.1 or 25 MPa for 3 h (Fig. 2B). Consistent with previous reports, GFP–Tor1 localized to the vacuolar membranes and was also present on foci (puncta) (Hatakeyama et al., 2019; Kira et al., 2014, 2016; Varlakhanova et al., 2017). Therefore, we classified the localization of GFP–Tor1 into four patterns and quantified them by counting on average ∼750 cells per culture from three independent experiments. The patterns of localization used for classification were: vacuolar membrane and the cytoplasm (V/C); the cytoplasm (C); foci (F); or the vacuolar membrane, the cytoplasm and foci (V/C/F). In the wild-type strain and the pib2Δ mutant, GFP–Tor1 mainly localized to V/C and C, and occasionally to F at 0.1 MPa (Fig. 2B; statistical data are presented in Table S1). In contrast, the proportion of the cell population with GFP–Tor1 at the vacuolar membrane dramatically reduced in the ego1Δ, ego3Δ, gtr1Δ, gtr2Δ and tco89Δ mutants, and there was a concurrent remarkable increase in the incidence of F localization at 0.1 MPa (Fig. 2B; Table S1). This is in good agreement with previous reports (Hatakeyama et al., 2019; Kira et al., 2014, 2016; Varlakhanova et al., 2017). We found that GFP–Tor1 remained at the vacuolar membrane in a substantial proportion of the wild-type strain population after high-pressure incubation. However, high pressure decreased the proportion of cells with vacuolar and/or foci localization of GFP–Tor1 in the mutants to varied degrees, with concurrent dislocation to the cytoplasm observed (Fig. 2B; Table S1). These results suggest that vacuolar membrane (V/C) localization of TORC1 is required for growth under high pressure. It should be noted that GFP–Tor1 remained at the vacuolar membrane in the pib2Δ mutant to a greater extent than observed in the other mutants (see below). High hydrostatic pressure generally increases protein hydration, because water confined to the surface of proteins is denser than bulk water (the system volume becomes smaller upon hydration of proteins) (Hata et al., 2020; Ohmae et al., 2013; Winter, 2015). Therefore, the assembly of TORC1 to form foci could be attenuated, or the disassembly of pre-existing TORC1 foci could be facilitated by high pressure if one of Ego1, Ego3, Gtr1 or Gtr2 is lost. Further experiments are needed to assess the dynamic changes in TORC1 localization upon high-pressure incubation.

To determine the significance of TORC1 kinase activity, we analyzed the effects of the constitutively active Tor1 mutations A1954V and I1957V (which have three- and two-fold greater activity than wild-type Tor1, respectively), and the kinase-null mutation W2176R, on growth under high pressure (Reinke et al., 2006). We found that the A1954V and I1957V mutations moderately enhanced growth at 25 MPa, whereas the W2176R mutation did not (Fig. 3A). Interestingly, the constitutively active mutations also restored the ability of the pib2Δ mutant to grow at 25 MPa, whereas these mutations had marginal effect in the ego1Δ and tco89Δ mutants (Fig. 3B–D). The restoration of high-pressure growth in the pib2Δ mutant can be rationalized by moderate vacuolar localization of GFP–Tor1 observed at 25 MPa (Fig. 2B). These results are consistent with our finding that TORC1 must be localized to the vacuolar membrane through an interaction with EGOC and Tco89 for growth under high pressure. In this regard, high pressure causes a more deleterious effect on Pib2-dependent TORC1 activation than on EGOC-dependent activation.

We next analyzed the effect of high pressure on the phosphorylation level of Sch9, a well-known substrate of TORC1 (Binda et al., 2009), using an anti-p-Sch9 (T737) antibody. Remarkably, high pressure considerably promoted Sch9 phosphorylation by twofold in the wild-type strain, while the total Sch9 level detected using an anti-HA antibody was unaffected (Fig. 4A). The ego3Δ and tco89Δ mutations significantly reduced the basal level of Sch9 phosphorylation at 0.1 MPa, whereas the ego1Δ and pib2Δ mutations did not. The ego1Δ, ego3Δ and pib2Δ mutations abolished the pressure-induced phosphorylation of Sch9. High pressure promoted Sch9 phosphorylation in the tco89Δ mutant, but the quantity of phosphorylated Sch9 was lower than that in the wild-type strain (Fig. 4A). These results suggest that Pib2 mediates a high-pressure stimulus to activate TORC1, which must be localized in the vacuolar membrane in an EGOC-dependent manner. It has been demonstrated that TORC1 is reactivated by the addition of various amino acids to nitrogen-starved cells in a Gtr1- and Pib2-dependent manner. For example, leucine elicits a transient phosphorylation of Sch9, whereas addition of glutamine leads to prolonged Sch9 activation (Stracka et al., 2014; Varlakhanova et al., 2017; Ukai et al., 2018). It should be noted that high-pressure-induced TORC1 activation is distinct from nitrogen-induced reactivation because the cells are continuously cultured in nitrogen-rich synthetic complete (SC) medium at both 0.1 and 25 MPa. In this regard, high pressure elicits hyperactivation of TORC1 in growing cells in a manner dependent on Ego1, Ego3 and Pib2.

The significance of the phosphorylation of Sch9 was verified by analyzing the growth of the rrd1Δsch9Δ strain expressing mutant forms of Sch9. Single deletion of SCH9 only caused a moderate sensitivity to high pressure, but a combined mutation with RRD1, which encodes protein that is associated with Tap42 (an essential protein in TORC1 signaling) (Jiang and Broach, 1999; Luke et al., 1996; Zheng and Jiang, 2005), resulted in a severe growth defect at 25 MPa (Fig. 4C). Sch9 has five amino acid residues (S711, S726, T737, S758 and S765) that can be phosphorylated by TORC1. We confirmed that substitution of alanine for each of the five residues individually did not significantly alter total Sch9 protein levels (Fig. 4B). We found that the T737A mutation markedly impaired the ability of cells to grow at 25 MPa, whereas the others did not (Fig. 4C). The T737E mutation, which mimics phosphorylation, allowed the rrd1Δsch9Δ mutant to grow at 25 MPa (Fig. 4C). These results suggest that Sch9 could be phosphorylated by TORC1 at T737, and this phosphorylation is required for growth under high pressure.

High hydrostatic pressure potentially promotes dissociation of multimeric proteins (Cordeiro et al., 2013; Fourme et al., 2012; Gross and Jaenicke, 1994; Winter, 2015). For example, the 30S and 50S ribosomal subunits from Escherichia coli dissociate under varying magnitudes of hydrostatic pressure (Infante et al., 1982). Therefore, we assessed whether high pressure at 25 MPa physically caused dissociation of ribosomes in TORC1 mutants, thereby diminishing translation and cell growth, or exerted attenuation of cell growth resulting from TORC1 deficiency. To verify this, we performed polysome profile analysis according to the method described by Inada and Aiba (2005). Exponentially growing cells (OD₆₀₀<1.0) were incubated in SC medium at 0.1 or 25 MPa for 5 or 15 h. Immediately after decompression, the cells were treated with 100 µg/ml cycloheximide to arrest translation (Inada and Aiba, 2005) and collected by centrifugation for the polysome profile analysis. The polysome ratio was 47.8% for wild-type cells cultured at 0.1 MPa for 5 h (Fig. 5). There were slight decreases in the polysome ratio in the ego1Δ (37.8%) and tor1Δ (32.3%) mutant strains. Interestingly, high-pressure incubation at 25 MPa for 5 h did not substantially alter the polysome ratio in the wild-type (43.7%), ego1Δ (41.7%) or tor1Δ (39.6%) strains (Fig. 5). This result suggested that the translational apparatus was maintained even in the ego1Δ and tor1Δ mutants at 25 MPa, despite cell growth arrest. Prolonged incubation of the cells for 15 h at 25 MPa moderately decreased the polysome ratio in the wild-type and two mutant strains; however, there was no marked difference among them (Fig. 5). In contrast, exposure of the wild-type cells to severe ethanol stress (10% for 30 min) dramatically decreased the polysome ratio to 10.6%, which was consistent with that reported previously (Yamauchi and Izawa, 2016). Therefore, we speculate that high pressure causes metabolic perturbations when the TORC1 signaling pathway is deficient, rather than disruptive effects on translation machinery.

Next, we focused on the levels of intracellular amino acids in TORC1 mutants. Mülleder et al. (2016) performed a functional metabolomic analysis to probe the extent of genome–metabolism interactions in 4913 gene-deletion nutrient-prototrophic mutants cultured in amino-acid-free SD medium. They found that rapamycin treatment causes an increase in levels of alanine, aspartate, asparagine, glutamate, glutamine, branched-chain amino acids and aromatic amino acids. These metabolic characteristics closely resemble those observed for the tco89Δ, ego1Δ, ego2Δ, ego3Δ and gtr1Δ mutants. However, cycloheximide treatment does not replicate the metabolic profile of TORC1 inhibition (Mülleder et al., 2016). Despite the differences in experimental conditions (i.e. nutrient auxotrophy, culture medium and quantification procedures), we confirmed that the ego1Δ, gtr1Δ, tor1Δ and pib2Δ mutations also caused increases in the levels of similar amino acids in our analysis (BY4742 strain background; SC medium; Fig. 6, left panels). Interestingly, high pressure (25 MPa) dramatically increased the levels of glutamine and alanine in the ego1Δ, gtr1Δ, tor1Δ and pib2Δ mutants, whereas it had no measurable effect in the wild-type strain (Fig. 6). These results suggest that high pressure has a tendency towards increasing levels of glutamine and alanine, and that EGOC–TORC1 and Pib2–TORC1 coordinately act to maintain these amino acid concentrations at certain levels under high pressure. The increases in amino acid levels observed upon cycloheximide treatment (Mülleder et al., 2016) did not match the amino acid profiles observed under high-pressure incubation (Fig. 6).

Tanigawa and Maeda (2017) demonstrated that glutamine and cysteine effectively stimulate in vitro TORC1 kinase activity in permeabilized semi-intact cells, and this stimulation depends on Pib2 but not Gtr1 and Ego3. Ukai et al. (2018) showed that radio-labeled glutamine directly binds to purified Pib2 proteins in the presence of cell lysate. Hence, we speculate that high pressure induces the intracellular accumulation of amino acids, particularly glutamine (Fig. 6), thereby stimulating TORC1 to phosphorylate Sch9 (Fig. 4A). TORC1 then inhibits the transcriptional activators Rtg1, Rtg3 and Gln3, all of which mediate glutamine synthesis (Crespo et al., 2002). Defects in TORC1 signaling resulted in uncontrolled amino acid biosynthesis and aberrant amino acid accumulation in the ego1Δ, gtr1Δ, tor1Δ and pib2Δ mutants under high pressure (Fig. 6). To test this hypothesis, we deleted GLN3 in the ego1Δ, gtr1Δ, tor1Δ and pib2Δ mutants, aiming to decrease the level of intracellular glutamine. According to the previous report, gln3Δ mutation produces a twofold decrease in glutamine levels (28.5 mM in the wild-type strain versus 15.7 mM in the gln3Δ mutant; Mülleder et al., 2016). In our system, glutamine levels were also markedly decreased in the four gln3Δ double mutants, at both 0.1 and 25 MPa, although the degree of reduction varied (Fig. 6, right panels). Whereas glutamine levels in the tor1Δgln3Δ and pib2Δgln3Δ mutants were comparable to those in the wild-type strain, they were still elevated in the ego1Δgln3Δ and gtr1Δgln3Δ mutants (Fig. 6). If the increased glutamine levels were the sole reason for high-pressure sensitivity, the tor1Δgln3Δ and pib2Δgln3Δ mutants should have recovered the ability to grow under high-pressure. Indeed, the tor1Δgln3Δ mutant grew comparably to the wild-type strain at 25 MPa, but the pib2Δgln3Δ mutant displayed only a partial restoration in growth (Fig. 7A). In our preliminary experiments, we examined whether other mutants with enhanced glutamine accumulation displayed sensitivity to high pressure. Deletions for HAP2, HAP5 (which both encode subunits of the Hap2/3/4/5 CCAAT-binding complex, a global regulator of respiratory gene expression; McNabb et al., 1995), or YOL057C cause marked intracellular accumulation of glutamine (to levels approximately three times higher than that of the wild-type strain); the three deletion mutants were classified into a cluster distinct from the TORC1 deficiencies (Mülleder et al., 2016). We did not find any deficiency in high-pressure growth in the hap2Δ, hap5Δ and yol057cΔ mutants (data not shown). Although we have not quantified intracellular glutamine levels of these mutants, it might be an oversimplification to conclude that increased glutamine levels are the sole reason for high-pressure sensitivity.

TORC1 inhibition has similar effects on intracellular amino acid homeostasis in mutants with defects in retrograde transport from endosomes to the Golgi apparatus, from the Golgi apparatus to the endoplasmic reticulum, in the multivesicular body (MVB) or HOPS/CORVET pathways, and in vacuolar H⁺-ATPase subunits (Mülleder et al., 2016). Upon defects in retrograde transport, substantial amounts of proteins might be destabilized and subject to degradation. Consequently, intracellular amino acid levels could be increased. To elucidate whether perturbations to intracellular membrane traffic abolish high-pressure growth, we examined deletions for genes relevant to the GARP complex (VPS51, VPS52 and VPS54), CORVET complex (VPS11, VPS18 and VPS16), SNARE complex (VPS45 and VPS6) and vacuolar protein sorting (VPS19, VPS15 and VPS34). Although there were some variations in growth ability among these mutants, the 11 deletion mutants displayed marked sensitivity to high-pressure conditions (Fig. 7B). Therefore, high pressure may exaggerate the aberrant accumulation of intracellular amino acids, leading to diminished cell growth.

In the present study, we demonstrated that high pressure stimulates the activation of TORC1, which must be localized in the vacuolar membrane in an EGOC-dependent manner for this activation to occur. This activation could be mediated by Pib2 that resides at the vacuolar membrane. A previous functional metabolomics study has revealed a similarity between the changes to intracellular amino acid levels in rapamycin-treated cells and those in mutants lacking components of EGOC–TORC1 and Pib2–TORC1 (Mülleder et al., 2016). These changes in amino acid levels are clearly different from those caused by treatment with cycloheximide. Therefore, amino acid homeostasis governed by TORC1 is not principally controlled via protein biosynthesis. One of the remarkable findings of the present study is that the levels of intracellular glutamine and alanine dramatically increased when the ego1Δ, gtr1Δ, tor1Δ and pib2Δ mutants were cultured at 25 MPa. Therefore, high pressure perturbs intracellular amino acid homeostasis, leading to increases in glutamine and alanine. As a result, TORC1 can be activated to exert downregulation of the biosynthesis of glutamine and, probably, other amino acids. Our present findings provide a unique framework for understanding nitrogen sensing and metabolic adaptation to mechanical stress.

Ukai et al. (2018) demonstrated that radio-labeled glutamine binds to Pib2 only in the presence of cell lysates, suggesting that the existence of unidentified components is essential for glutamine–Pib2 binding. Tanigawa and Maeda (2017) demonstrated that the addition of glutamine in a semi-intact cell system enhances Pib2–TORC1 kinase activity. Thus, glutamine is likely to strengthen the binding affinity of an as-yet-uncharacterized Pib2 complex to TORC1, thereby stimulating kinase activity. The pib2Δgtr1Δ double mutant exhibits synthetic lethality (Kim and Cunningham, 2015; Ukai et al., 2018), and thus, EGOC–TORC1 and Pib2–TORC1 complexes are functionally redundant. EGOC–TORC1 and Pib2–TORC1 are mutually distinct regarding their functionality, because GFP–Tor1 normally localized to the vacuolar membrane in the pib2Δ mutant but was aberrantly localized in EGOC-defective mutants (Fig. 2B). Kira et al. (2016) described colocalization of GFP–Tor1 puncta with a vacuolar–nuclear junction marker, Vac8–mCherry. Prouteau et al. (2017) demonstrated that TORC1 oligomerizes into a higher-level hollow helical assembly, a TOROID (TORC1 organized in inhibited domain), upon glucose starvation, and that this assembly emerges in the gtr1Δgtr2Δ mutant in the presence of glucose. Recently, Hatakeyama et al. (2019) demonstrated that GFP–Tor1 colocalizes in the foci with Vps21 and Vps27, which are marker proteins for prevacuolar endosomes (PVEs), and showed that Ego1 is separately sorted to the vacuoles and PVEs to form EGOC at both surfaces. Therefore, the dynamic relocations of GFP–Tor1 responsive to high pressure in the TORC1 mutants (Fig. 2B) should be investigated with respect to colocalization with some marker proteins, and protein–protein associations. The expression of the active mutant forms of Tor1 restored growth under high pressure in the pib2Δ mutant but was less effective in the ego1Δ and tco89Δ mutants (Fig. 3). The primary role of EGOC in growth under high pressure might be to recruit TORC1 to the vacuolar membrane to induce its full activity, whereas the role of Pib2 under high pressure could be to mediate the direct activation of TORC1 through increased glutamine concentration. TORC1 activation could lead to inhibition of the transcriptional factor Gln3 and a reduction in glutamine biosynthesis. This is seemingly contradictory to our finding that the glutamine level in the wild-type strain was unchanged by high pressure (Fig. 6). We assume that increasing the glutamine level surrounding the vacuole immediately activates TORC1 to downregulate glutamine synthesis under high pressure. This is in agreement with our observation that TORC1 must be strictly confined to the vacuolar membrane to enable growth under high pressure.

Hatakeyama et al. demonstrated that TORC1 in PVEs mediates phosphorylation of Vps27 to prevent microautophagy under nutrient-rich conditions, and this phosphorylation does not affect degradation of plasma membrane permeases via the MVB pathway (Hatakeyama and De Virgilio, 2019; Hatakeyama et al., 2019). Thus, TORC1 function might connect to amino acid homeostasis via its role in membrane transport systems and microautophagy. Nevertheless, aberrant amino acid homeostasis arising from various genetic defects has no marked effect on normal cell growth in nutrient-rich conditions under atmospheric pressure. Therefore, high pressure might somehow exaggerate the aberrant accumulation of intracellular amino acids, leading to diminished cell growth.

Taking these findings together, we propose the following model to explain the regulation of EGOC–TORC1 in response to high pressure (Fig. 8). Upon amino acid perturbation by high pressure, TORC1 is first activated in a Pib2-dependent manner within the vacuolar membrane. The activated TORC1 then phosphorylates Sch9 and other substrates, such as Gln3, which serve to recover from the increases in glutamine and other amino acids. Tap42 is an essential protein in TORC1 signaling, and is associated with Sit4 and Rrd1 upon phosphorylation by TORC1 (Jiang and Broach, 1999; Wang et al., 2003; Yan et al., 2006; Zheng and Jiang, 2005). The Tap42–Sit4–Rrd1 complex is a serine/threonine phosphatase, which mediates dephosphorylation of Sap155 and Par32 (Luke et al., 1996). We have previously demonstrated that the sap155Δ and par32Δ mutants display marked sensitivity to high pressure (Abe and Minegishi, 2008), and that the high-pressure sensitivity of these mutants can be recovered by prototrophy rescue via the introduction of three plasmids coding for HIS3, LEU2 and URA3 (Kurosaka et al., 2019). In this regard, the Tap42–Sit4–Rrd1 pathway might play a role in protecting nutrient permeases against high-pressure perturbation by as-yet-unknown mechanisms (Fig. 8).

Unlike other perturbations used in biological systems, such as starvation, osmolarity, pH or toxic chemicals, hydrostatic pressure does not introduce or remove any components of the system. It merely changes the equilibria among pre-existing components. As changes in equilibria can be many orders of magnitude, pressure can highlight minor or masked species for direct observation. Because the TORC1 machinery is highly conserved across eukaryotes, our findings provide a unique insight into the hidden dynamics of the complex metabolic network in the TORC1 signaling pathways.

The parental wild-type strain BY4742 (MATα his3Δ1 leu2Δ0 lys2Δ0 ura3Δ0) and its deletion mutants were used in this study (Table S2). The cells were grown at 25°C at 0.1 MPa or 25 MPa in synthetic complete (SC) medium, with slight modifications. The amino acid concentrations were as follows: 100 μg/ml leucine, 30 µg/ml isoleucine, 40 µg/ml tryptophan, 20 µg/ml histidine, 30 µg/ml lysine, 20 µg/ml methionine, 50 µg/ml phenylalanine, 30 µg/ml tyrosine, 20 µg/ml arginine, 100 µg/ml aspartic acid, 100 µg/ml glutamic acid, 400 µg/ml serine and 200 µg/ml threonine. In our analysis, 1.0 OD₆₀₀ is comparable to 1.65×10⁷ cells/ml.

Exponentially growing cells were diluted in SC medium to give an OD₆₀₀ value of 0.1. The cells in sterilized tubes were placed in a high-pressure chamber and subjected to high pressures and 25°C for 24 h. The ability of cell growth was evaluated by measuring OD₆₀₀.

Plasmids used in this study are listed in Table S3. pPL132 (3HA–Tor1), pPL155 (3HA–Tor1 A1957V), pPL156 (3HA–Tor1 I1954V) and pPL157 (3HA–Tor1 W2176R) were kind gifts from Ted Powers (University of California, Davis, CA) (Reinke et al., 2006). pSK108 was a kind gift from Shintaro Kira and Takeshi Noda (Osaka University, Japan), and was used to generate genomic N-terminal tagged GFP–TOR1 strains as described previously (Kira et al., 2014, 2016). pUA127, produced by modifying pUA35 (Uemura et al., 2017), was used to construct plasmids expressing fusion proteins containing a C-terminal triple hemagglutinin (3HA) tag in pRS316 (URA3, CEN).

To construct pUA181 (GTR1-3HA, URA3, CEN), the GTR1 open-reading frame (ORF) containing its own promoter (pGTR1) was amplified using genomic DNA prepared from BY4741 and primers 5′-CCACGTGCTGGAAATAACACGTCCAC-3′ and 5′-GGTGGTGGCTTAATGACCTTCTACTTATG-3′. The resulting fragments were cloned into pGEM-T Easy (Promega) to generate pUA27. A DNA fragment containing the GTR1 ORF with pGTR1 was amplified using pUA27 as a template and primers 5′-CGGGCCCCCCCTCGAGCCACGTGCTGGAAATAACACGTCCA-3′ and 5′-ACTCATGGTTCCCCCGGGTTGGAAAAACTCTTTGGCTTTTTTGATGTTTTCC-3′. The resulting fragment was cloned into the XhoI-SmaI site of pUA127 using an In-Fusion HD cloning kit (Takara Bio Inc., Shiga, Japan) to generate pUA181. The pUA182 (GTR2-3HA, URA3, CEN) plasmid was constructed by similar procedures.

To construct pUA224 (SCH9-3HA, URA3, 2µ), the SCH9 ORF with its own promoter (pSCH9) was amplified using BY4741 genomic DNA and primers 5′-CGGGCCCCCCCTCGAGCCCAACTCATGTTAGAATGCACGC-3′ and 5′-ACTCATGGTTCCCCCGGGTATTTCGAATCTTCCACTGACAAATTCGTC-3′. The resulting fragment was cloned into the XhoI-SmaI site of pUA127 using an In-Fusion HD cloning kit to generate pUA194. The 3.7-kb XhoI-NotI fragment in pUA194 was inserted into the XhoI-NotI site of pRS426 to generate pUA224.

Point mutations were created in the Gtr1-3HA, Grt2-3HA and Sch9-3HA plasmids (Table S3) by site-directed mutagenesis using PrimeSTAR^® Mutagenesis Basal Kit (TaKaRa, Bio Inc.) according to the manufacturer's instructions.

Whole cell extract preparation was performed as described previously, with a slight modification (Usami et al., 2014). In brief, cells (1.65×10⁸) were collected by centrifugation, washed serially with 10 mM NaN₃-10 mM NaF and lysis buffer A [50 mM Tris-HCl, 5 mM EDTA, 10 mM NaN₃ and Complete™ EDTA-free protease inhibitors (Roche, Mannhein, Germany)], and disrupted with glass beads at 4°C. Unbroken cells and debris were removed by centrifugation at 900 g for 5 min, and cleared lysates were diluted with 5× sample buffer (312.5 mM Tris-HCl pH 6.8, 50% glycerol, 10% SDS, 0.005% Bromophenol Blue and 25% 2-mercaptoethanol) then boiled for 5 min.

For the analysis of phosphorylation of Sch9, whole cell lysates were prepared as described previously with a slight modification (Urban et al., 2007). The cell culture was mixed with trichloroacetic acid (final concentration 6%) and kept on ice for 5 min, then washed with ice-cold 70% ethanol. The cells were suspended with urea buffer (50 mM Tris-HCl pH 7.5, 5 mM EDTA, 6 M urea, 1% SDS, 50 mM NaF, 10 mM β-glycerophosphate and Complete™ EDTA-free protease inhibitors), disrupted with glass beads at 4°C and heated at 65°C for 10 min. The cell lysates were centrifuged at 15,000 g for 10 min. The supernatants were diluted with 5× sample buffer and incubated at 65°C for 15 min.

Proteins were separated by SDS–PAGE and transferred to an Immobilon PVDF membrane (Millipore, Billerica, MA). The membrane was then incubated for 1 h with anti-HA (1:1000; cat. no. ab-hatag, ver. no. 05E31-SV; InvivoGen, San Diego, CA), anti-Adh1 (1:1000; cat. no. 100-4143; Rockland antibodies and assays, Gilbersville, PA), or anti-phospho-Sch9 (Thr737) [1:4000; a kind gift from Dr Tatsuya Maeda of Hamamatsu University School of Medicine, Japan (Takahara and Maeda, 2012)] antibodies. After washing, the membrane was incubated for 30 min with horseradish peroxidase-conjugated donkey anti-mouse IgG (GE Healthcare Bio-Science, Piscataway, NJ). Labeling was detected using an ECL select kit (GE Healthcare Bio-Science).

Cells expressing GFP-tagged proteins were imaged on a confocal laser microscope model FV-3000 equipped with a 60× objective lens (NA 1.5) (UPLAPO60XOHR; Olympus, Co. Ltd, Tokyo, Japan).

Extraction of intracellular amino acids was performed essentially as described previously (Ohsumi et al., 1988; Tone et al., 2015). Exponentially growing cells in SC medium were incubated at 0.1 MPa or 25 MPa at 25°C for 5 h, and 10 OD₆₀₀ cells were washed twice with a buffer (2.5 mM potassium phosphate pH 6.0, 0.6 M sorbitol and 10 mM glucose). The cells were resuspended in 500 µl of distilled water and boiled for 15 min. After centrifugation, the supernatants were collected and dried by an evaporator. The dried fractions were dissolved in 450 µl of 20 mM HCl. Labeling of amino acids with phenyl-isothiocyanate (PTIC, FUJIFILM Wako Pure Chemical Corporation, Tokyo, Japan) was performed according to the manufacturer’s instructions. Briefly, 50 µl of the amino acid mixture was dried and dissolved in 20 µl of a solution (ethanol:water:triethylamine=2:2:1). After evaporation, the dried sample was dissolved in a reaction solution (ethanol:triethylamine:PTIC=7:7:1) to label amino acids for 20 min. The labeled amino acids were analyzed using an LC2010A HT high-performance liquid chromatograph (Shimadzu, Kyoto, Japan).

Cells cultured at 0.1 or 25 MPa were immediately treated with 100 µg/ml cycloheximide to arrest translation. The cells were collected by centrifugation (900 g) and washed with a buffer (20 mM HEPES pH 7.4, 2 mM magnesium acetate, 2 mM potassium acetate and 100 µg/ml cycloheximide). Because culture volume of the high-pressure chamber was limited (∼100 ml), we subjected cell mixtures from 4–8 independently prepared cultures to two independent polysome profile analyses. Polysome profile was analyzed according to the method described by Inada and Aiba (2005) using a gradient master and fractionator (cat. no. 107-201M and 152-002, BioComp Instruments Ltd., New Brunswick, Canada). The polysome ratio was determined as the percentage of area under polysomal ribosome peaks relative to that under total ribosome peaks (Hofmann et al., 2012).

We thank Ted Powers, Takeshi Noda and Shintaro Kira for providing plasmids; Michael N. Hall and Takashi Ushimaru for providing 3GFP–Tor1 and Tco89–GFP strains, respectively (not used in this version of the manuscript); Tatsuya Maeda for providing anti-phospho-Sch9 (Thr737); Nobuyoshi Baba and Kiyoshi Takahashi for supporting the construction of high-pressure chambers; Miyuki Kawada for valuable discussions.