EGOC inhibits TOROID polymerization by structurally activating TORC1

doi:10.21203/rs.3.rs-1238390/v1

PDF

Article

EGOC inhibits TOROID polymerization by structurally activating TORC1

https://doi.org/10.21203/rs.3.rs-1238390/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted 02 Feb, 2022

You are reading this latest preprint version

Target Of Rapamycin Complex 1 (TORC1) is a protein kinase controlling cell homeostasis and growth in response to nutrients and stresses. Glucose depletion triggers a pronounced redistribution of TORC1 from a dispersed localisation over the vacuole surface into a large, inactive condensate called a TOROID (TORC1 Organised in Inhibited Domain). However, the molecular mechanisms governing this transition were unclear. Here, we show that acute depletion and repletion of EGO Complex (EGOC) activity is sufficient to control TOROID distribution, independently of other nutrient-signalling pathways. A 3.8 Å-resolution 3D reconstruction of TORC1 from TOROID cryo-EM data together with interrogation of key interactions in vivo provided structural insights into TORC1-TORC1’ and TORC1-EGOC interaction interfaces. These data support a model in which glucose-dependent activation of EGOC triggers binding to TORC1 at an interface required for TOROID assembly, resulting in inhibition of TORC1 polymerisation and release of active TORC1 dimers.

The Target Of Rapamycin, TOR, is a Phosphatidylinositol Kinase related protein Kinase (PIKK) family member ^1,2 as it resembles lipid kinases but is rather a Ser/Thr protein kinase. Similarly to other PIKKs, TOR functions as the catalytic subunit of distinct multiprotein complexes, TORC1 and TORC2. These complexes are broadly conserved across eukaryote evolution, however, the rapamycin-sensitivity of TORC1 has facilitated studies of this complex and consequently TORC1 signalling is relatively better characterized compared to rapamycin-insensitive TORC2 ^3,4. Both complexes are essential regulators of cell homeostasis: in yeast, TORC1 controls cell mass and volume according to nutrient availabilities and abiotic stresses while TORC2 maintains biophysical properties of the plasma membrane ⁵.

Yeast TORC1 is composed of four proteins: either of the two TOR paralogs Tor1 or Tor2, Kog1, Lst8, and Tco89 ^6,7. In humans, mTORC1 (mammalian/mechanistic Target Of Rapamycin Complex 1) respectively contains mTOR, Raptor and mLst8; a Tco89 orthologue is not apparent. TORC1 phosphorylates the AGC-family kinase Sch9 which served as a convenient reporter to demonstrate that TORC1 is active in the presence of good carbon (glucose), nitrogen (glutamine), and phosphate sources and inactivated upon nutrient deprivation as well as exposure to osmotic, redox or thermal stresses ⁸. Sch9 and other TORC1 effectors subsequently regulate a plethora of molecular processes ranging from ribosome biogenesis to autophagy; from lipid biosynthesis to membrane trafficking ^9–12. mTORC1, which also phosphorylates an AGC-family kinase, S6K1, is additionally regulated downstream of growth-factor signalling and loss of mTORC1 regulation has been implicated in a large range of illnesses including cancers, neurodegenerative diseases and metabolic disorders ¹³. Given its central role in physiological and pathophysiological growth control, molecular dissection of (m)TORC1 signalling is of considerable interest.

Structures of mTORC1 have been recently solved by combinations of X-ray crystallography and electron cryo-microscopy (cryo-EM) ^14–16. These studies demonstrated that mTORC1 is dimeric and, when bound, the macrolide rapamycin recruits the 12 kDa proline isomerase FKBP12 such that substrate access to the kinase active site is sterically occluded. Under physiological conditions, mTORC1 activity is regulated through the actions of multiple GTPases. The Rheb GTPase couples growth factor and energy cues to mTORC1 with the binding of GTP-loaded Rheb leading to allosteric rearrangements that increase the kinase activity of mTORC1 ¹⁶. In contrast, the RAG GTPases have a more nuanced role in mTORC1 regulation, primarily downstream of nutrient cues.

There are two sets of RAG paralogues in mammalian cells: RAGA and RAGB, and RAGC and RAGD. These proteins form obligate heterodimers with GTP-loading status determining the distance and relative orientation between the two G-domains ^17,18. Unlike other GTPases, the RAGs are not modified with lipid but are instead anchored to the lysosome membrane via the heteropentameric Ragulator complex^19–21. Recent structures of the Raptor•RAG•Ragulator complex suggest that active RagA^GTP/RagC^GDP is able to bind the Armadillo (ARM) domain of Raptor and grasp onto a specific short α−helical fragment termed the ‘Claw’ ^22,23. This binding recruits mTORC1 from a mysterious localization in the cytoplasm to the lysosome where it can be subsequently activated by Rheb^GTP.

The regulation of TORC1 in Saccharomyces cerevisiae (budding yeast) does not completely parallel the situation in mammalian cells. In contrast to mammalian cells and Schizosaccharomyces pombe ²⁴, there are no reported roles for the budding yeast Rheb orthologue, Rhb1, in TORC1 regulation. Furthermore, yeast TORC1 appears to be constitutively localized to vacuole and/or endosome membranes, regardless of its activity status ^25,26. We recently reported that acute glucose starvation triggers the condensation of TORC1 into a large, vacuole-associated helical polymer we named a TOROID (TORC1 organised in Inhibited Domains; ²⁷. TOROID formation tracked with, and was necessary for, Sch9 dephosphorylation, implying that this polymerization was mechanistically responsible for TORC1 inactivation. Consistently, a low-resolution TOROID structure suggested that stacking of TORC1 protomers (i.e. dimers) into the TOROID polymer would sterically occlude access to the kinase active site analogously to rapamycin-FKBP12 binding ²⁷. TOROID formation was found to be dependent upon the two yeast RAG orthologues GTR1 (RAGA/B) and GTR2 (RAGC/D): Δgtr1 Δgtr2 cells present constitutive TOROIDs even in the presence of glucose and are impaired in their ability to downregulate Sch9 phosphorylation upon glucose starvation ²⁷.

Gtr1 and Gtr2, like the RAGs, are obligate heterodimers anchored to the vacuolar membrane via the EGO ternary complex (EGO-TC, the functional equivalent of Ragulator) composed of Ego1, Ego2 and Ego3 ²⁸. Ego1 is myristoylated and palmitoylated on its N-terminal extremity, which serves to tether the complex to the vacuolar membrane ²⁹. The remainder of Ego1, together with Ego2 and Ego3, cradle the C-terminal dimerization domains of Gtr1 and Gtr2 ²⁸. This orientation exposes the N-terminal G-domains of Gtr1 and Gtr2 to upstream regulators as well as their effector, TORC1. The GTPase activating Proteins (GAPs) that act on Gtr1 (RagA/B) and Gtr2 (RagC/D) are SEACIT (GATOR1) and Lst4/7 (FNIP–FLCN) respectively ^30,31. In contrast to the relatively well-understood and accepted GAPs, numerous Guanine nucleotide Exchange Factors (GEFs) that act on Gtr1 (RagA/B) have been reported ^18,25,31. A GEF for Gtr2 (RagC/D) may not be necessary as this GTPase appears to possess an intrinsically high dissociation constant for GDP ³². Similar, but not identical to the situation in mammalian cells, guanine nucleotide loading status of the Gtrs appears to have a direct role on TORC1 localization in yeast. Indeed, expressing mutants that lock the Gtrs in an inactive, Gtr1^GDP/Gtr2^GTP conformation, TORC1 is concentrated in a TOROID punctum on the vacuolar membrane whereas in cells expressing active, Gtr1^GTP/Gtr2^GDP, TORC1 is diffusely localized over the surface of the vacuole ²⁷. Molecular details explaining how the EGO-TC / Gtrs (collectively known as the EGO Complex, EGOC) govern TOROID assembly, alone or in conjunction with other nutrient-regulated signalling pathways, remain unknown.

To specifically assess the contribution of the EGOC to TORC1 localization dynamics, independently of the multitude of other glucose-responsive signalling pathways ³³, we developed orthogonal approaches by which we could acutely abrogate or reintroduce EGOC activity. We observed that proteolytic shaving of the EGOC off of the vacuole membrane triggered TORC1 puncta (TOROID) formation while reintroduction of the EGOC to the vacuole membrane triggered puncta disassembly. We additionally observed that the EGOC co-localizes with TORC1 puncta in a Gtr-dependent fashion. To better characterize this interaction, we used cryo-EM to obtain a 3.8 Å-resolution structure of TORC1 from TOROID cryo-EM maps. We found that a central hub of interactions locks the Kog1 Armadillo (ARM) domain on top of the kinase domain of Tor2’ and sterically occludes the kinase active site. CrispR/Cas9 mutagenesis and pull-down assays were used to verify these core features and additionally revealed that this same hub is engaged by the EGOC to mediate glucose-dependent regulation of TORC1 signalling. Based on these data, we propose that the EGOC is necessary and sufficient for TOROID regulation through competition for a common binding hub in Kog1; trans binding of this hub to Tor1’/2’ sequesters TORC1 into an inactive TOROID, while binding to active EGOC liberates TORC1 dimers able to signal to downstream targets.

Acute abrogation or restoration of EGOC activity on the vacuole membrane impacts TOROID dynamics independently of glucose changes

We previously proposed that TOROID regulation downstream of glucose-derived signals was mediated via the EGOC ²⁷. This was based on constitutive presence of TOROIDs in Δgtr1 Δgtr2 cells and an inability of cells expressing GTP/GDP-locked Gtr1/2 mutants to regulate TOROID assembly/disassembly according to glucose cues. However, knowing that glucose signals regulate TORC1 activity via additional, parallel pathways including Snf1/AMPK and pH ^34,35, we wished to determine if orthogonal, glucose-independent manipulation of EGOC function alone would be sufficient to alter TOROID formation.

To rapidly abrogate EGOC signalling, we introduced a Tobacco Etch Virus (TEV) Protease cleavage site into the amino-terminal region of Ego1, downstream of the lipidation sites (Ego1^TEV; Figure 1A and S1A) ^28,36. TEV protease was expressed downstream of the CTH2 promoter, whose tight repression by Fe²⁺ is quickly reversed by addition of the iron chelator Bathophenanthrolinedisulfonic acid (BPS; ³⁷. Neither acute BPS treatment nor TEV protease induction affected TORC1 activity in otherwise WT cells (Figure S1B).

While the EGO1^TEV allele per se does not display a phenotype, expression of the TEV protease renders these, but not WT cells, hypersensitive to rapamycin demonstrating a loss of EGOC function similar to cells lacking functional EGOC (Figure S1C; ^25,38. In EGO1^TEV cells, prior to TEV induction, the EGOC was diffusely localized around the vacuole membrane; this signal was diminished by 50% after 30 minutes of TEV induction (Figure 1B and S1D). After 75 minutes, the EGOC was almost completely shaved off of the vacuole similarly to Δego1 strain (Figure 1C). TORC1 puncta formation tracked with EGOC removal from the vacuolar membrane displayed twice more cells presenting TOROIDs after ~60 minutes of BPS treatment (Figure 1C, 1D and Figure S1D). Measurements from single-cell estimate the half-time of TOROID formation upon EGOC depletion to be about 7 minutes (Figure 1E and 1F). Taken together, these data demonstrate that acute depletion of EGOC from the vacuole is sufficient to induce TOROID formation.

To rapidly re-establish EGOC signalling, we exploited the vacuole fusion that occurs during zygote formation ³⁹. Specifically, we crossed a MATa Δgtr1 Δgtr2 strain, which constitutively displays a TOROID marked with GFP-Tor1, with a MATα partner containing Vph1-mCherry to delineate the vacuolar membrane (Figure 1G). As a control, we also used a MATα Δgtr1 Δgtr2 VPH1-mCherry partner. After mixing of the mating partners, budding zygotes were staged as either early, intermediate, or late (Figure 1G). Early zygotes possessed a small bud and fragmented vacuoles that remained in the MATα parent. Intermediate zygotes possessed larger buds in which some Vph1-mCherry signal was observed. Late zygotes possessed a bud roughly the size of the parent cells and vacuole exchange between the parents in addition to the bud was observed. In crosses between Δgtr1 Δgtr2 parents, a TOROID, initially present in the MATa parent, started to be also detected in about 50% of buds and some MATα parents, already at the intermediate zygote stage (Figure 1G, and 1H). This pattern was not observed in crosses with MATα GTR1 GTR2 VPH1-mCherry. In this case, TORC1 puncta were observed in only very few buds, and, at late stages even disappear from the MATa parent (Figure 1H). Live, single-cell imaging also demonstrated that TOROIDs from this mating dissolve promptly upon entry into the bud, upon mixing with vacuolar membrane possessing functional EGOC (Figure 1I). The half-time to complete punctum dissolution was approximately 2.5 minutes (Figure 1J). These data indicate that acute loss or re-establishment of EGOC activity, in the absence of potentially confounding inputs from other nutrient-regulated signalling pathways, is sufficient to trigger or dissolve TOROIDs, respectively. This supports the notion that the EGOC regulates TOROID formation, potentially through direct physical interaction with TORC1 and/or TOROIDs.

TOROIDs partially co-localize with EGOC puncta

Previous work demonstrated that Gtr1/2 can physically interact with TORC1 ^25,40 and our results suggested that this interaction is necessary for regulating TOROID formation. To probe this idea, we assessed how the EGOC and TORC1 localize with respect to one another. At the outset, we ensured that introduction of fluorescent tags did not compromise protein function. This was critical as we frequently encountered rapamycin hyper-sensitivity in singly, and especially doubly tagged strains that correlated with aberrant protein localizations. For example, using a GFP-Tor1, Ego3-3xmCherry strain, it has recently been shown that TORC1 and EGOC co-localize in discrete puncta ^41,42. However, in this background there appears to be a synthetic interaction between these TOR1 and EGO3 alleles as TORC1 puncta are present even in glucose-replete conditions (Figure S2A). We noted similar, synthetic loss-of-function phenotypes between GFP-TOR1 and EGO3-mScarlet, EGO2-mScarlet, and EGO1-mCherry (data not shown). Importantly, when assessed separately, both TORC1 (marked with GFP-Kog1) and the EGOC (marked with EGO3-GFP) localized diffusely around the vacuole membrane in glucose-replete cells but collapsed into a punctum within 10 minutes of acute glucose depletion. Both puncta dissolved again within minutes of glucose replenishment (Figure 2A-B). We note that failure to appreciate the hypomorphic nature of tagged alleles may lead to incorrect conclusions and likely underlies discrepancies found in the literature.

A strain expressing GFP-Kog1 and Ego3-3xmCherry appeared to be relatively uncompromised with nearly WT levels of TOROIDs in glucose-replete and post-diauxic shift (PDS) cultures (Figure 2C and S2A). When these cells were cultured PDS, co-localization between TORC1 and EGOC was moderate, as reflected by a Pearson’s coefficient of 0.44 (Figure S2B). This co-localization also appeared to by asymmetric: while the EGOC signal overlapped strongly with the TORC1 signal (Mander’s coefficient (MC): 0.71; Figure S2B), the TORC1 signal overlap with the EGOC signal was less robust (MC: 0.45; Figure S2B). Further characterization showed that small EGOC puncta did not co-localize with the TORC1 signal (Pearson’s coefficient (PC_<500nm): -0.004 ± 0.095; Figure S2C) while large EGOC puncta did (PC_>500nm: 0.525 ± 0.175). In contrast, large and small TORC1 puncta displayed similar overlaps with the EGOC signal (PC_<500nm: 0.315 ± 0.023; PC_>500nm: 0.361 ± 0.095; Figure S2C). Together, these data show that both EGOC and TORC1 form condensates upon acute glucose depletion or in PDS cells, and that large EGOC puncta seem to require a TOROID but not vice versa.

We next wished to characterize how guanine nucleotide loading status of the Gtrs affects EGOC puncta formation. Although deletion of GTR1 and GTR2 promotes TORC1 puncta formation (Prouteau et al., 2017, Figure 1), EGO-TC puncta were absent in these strains (Figure 2D). Reintroduction of “active” GTR alleles (Gtr1^GTP/Gtr2^GDP) antagonized both TORC1 and EGOC puncta formation regardless of the growth state while reintroduction of “inactive” GTR alleles (Gtr1^GDP/Gtr2^GTP) enhanced formation of TOROIDs but, curiously, not EGOC puncta (Figure 2E-F). In Δgtr1 Δgtr2 and Gtr1^GDP/Gtr2^GTP strains which contain TOROIDs, TORC1 activity is reduced as evidenced by their rapamycin hypersensitivity (Figure 2G). As a complementary means to assess the effects of nucleotide loading status of the Gtrs on EGOC and TORC1 puncta formation we also used mutant strains in which Gtr GAP activities had been altered. Deletion of IML1/SEA1, which compromises SEACIT activity leading to Gtr1^GTP accumulation, antagonized EGOC and TORC1 puncta formation (Figure S2D-F). In contrast, deletion of SEA3, a proposed SEACAT component, augmented EGOC and TORC1 puncta formation (Figure S2D-F). Finally, deletion of LST4 or LST7, which leads to Gtr2^GTP accumulation, promoted EGOC and TORC1 puncta formation (Figure S2D-F). Collectively, these results confirm and extend our previous observations ²⁷ that the nucleotide loading status of the Gtrs dictate TOROID formation: whether generated by nutrient levels, mutations in the GAPs or mutations in the GTRs themselves, Gtr1^GTP/Gtr2^GDP antagonize, while Gtr1^GDP/Gtr2^GTP promote, TOROID formation. The EGOC also forms a punctum, which in most, but not all cases, correlates with the presence of a TOROID, the notable exceptions being the lack of EGOC puncta in Δgtr1 Δgtr2 cells and in exponentially growing cells expressing “inactive” GTRs. These last results demonstrate that EGOC puncta are not necessary for TOROID formation but can assemble with TOROIDs in a Gtr-dependent manner. In contrast, active EGOC is needed to prevent TOROID formation.

Wishing to further characterize the molecular mechanism by which the EGOC interacts with TORC1 to regulate its supramolecular organization we first performed in vitro binding assays between EGOC variants and free TORC1 dimers purified from exponentially growing cells or TOROIDs from PDS cells. These assays demonstrated that purified “active” EGOC interacts preferentially, albeit weakly, with TORC1 dimers (Figure 2H-I). In contrast, purified “inactive” EGOC interacts robustly with TORC1 dimers and even more so with TOROIDs (Figure 2H-I). Thus, in contrast to the situation reported for mammalian cells ²¹, in yeast, the EGOC binds TORC1 regardless of which guanine nucleotides are bound to the Gtrs. This is consistent with prior genetic observations that the EGOC can both positively and negatively regulate TORC1 activity ²⁵, see discussion).

High resolution Cryo-EM TOROID Structure

To better understand the molecular regulation of TOROID (dis)assembly, we purified TORC1 from KOG1-TAP ∆tor1 cells grown PDS and acquired a high-resolution TOROID structure by cryo EM. We hypothesized that TOROIDs containing only Tor2, rather than a mix of Tor1 and Tor2, could yield a higher resolution structure. Initial attempts to determine the TOROID structure using helical reconstruction resulted in 3D reconstructions at rather low resolution (FSC_0.143: 9.1 Å), despite the apparent high quality of the obtained data and ensuing 2D class averages (Figure 3A, Figure S3A-B). The relatively poor precision of helical symmetry determination on the un-symmetrized map (Figure S3A) suggested that instead of being a rigid helix, TOROID filaments resemble a flexible, slinky-like spring with varying pitch. Therefore, helical reconstruction most probably results in averaging out of TORC1 units along the helical assembly. To overcome this issue, we employed signal subtraction protocols ^43,44, using the known positions of TORC1 units in the helical reconstruction, to obtain a set of signal subtracted particles that could be further analyzed using a regular single particle analysis (SPA) pipeline (Figure S3A). We used a broad 3D mask including parts of neighboring TORC1 units surrounding a central TORC1 in the TOROID helix to ensure that information on the TORC1-TORC1’ interfaces remained intact.

3D reconstruction using signal-subtracted TORC1 particles and employing D1 symmetry and non-uniform refinement in CryoSPARC ⁴⁵ resulted in a 3.85 Å map according to the gold standard FSC=0.143 criterion, which enabled us to build a nearly complete structural model of yeast TORC1 in the TOROID (Figure 3B, Figure S3A-C). While the final map displays heterogeneity in terms of resolution (Figure S3B-C), the symmetric nature of the TOROID assembly could be exploited to unambiguously build the majority of the TORC1 structure, with many regions showing density for amino acid side chains (Figure 3D). Moreover, it is apparent from the 3D reconstruction and local resolution estimation that the interactions between adjacent TORC1 units along the TOROID helix (intra-coil) are far more prominent than interactions along the helix z-direction (inter-coil; Figure 3A, S3B-C), which is also reflected by the resolution difference at the intra-coil and inter-coil interfaces.

The overall yeast TORC1 structure extracted from TOROIDs is very similar to its mammalian counterpart (mTORC1, Figure 4A), as shown by a structural alignment of individual Tor2/mTOR (Figure 4B), Kog1/RAPTOR (Figure 4C) and Lst8 subunits (Figure S4A, B). While yeast and human Lst8 are virtually identical (RMSD= 0.937 Å over 252 aligned main-chain atoms), the main differences between Tor2/mTOR (RMSD= 3.311 over 1608 aligned main-chain atoms) and Kog1/RAPTOR (RMSD= 2.371 over 866 aligned main-chain atoms), can be traced to a conformational difference of Tor2, and the presence of previously uncharacterized additional α-helical regions for Kog1 (Figure 4B-C). We compared our structure of yeast TORC1 present in TOROIDs with structures of inactive mTORC1 and active mTORC1 bound to Rheb ¹⁶. Tor2 in TORC1 in TOROIDs is structurally more similar to mTOR in inactive mTORC1 (RMSD_inactive= 3.311 Å) than in Rheb bound mTORC1 (RMSD_active= 10.388 Å; Figure 4B). Indeed, TORC1 in TOROIDs has an even more “open” conformation than inactive human mTORC1 (Figure 4A): the HORN region of Tor2 makes a 7° or 23° outward shift when compared to inactive or Rheb-bound mTORC1 respectively (Figure 4B).

Comparison of yeast Kog1 and human RAPTOR reveals a ‘Twix’ region contributed by two α-helices connected by a poly-glutamine linker that is present in Kog1 but completely absent from human RAPTOR. Alignment of S. cerevisiae Kog1 with orthologous sequences from H. sapiens, S. Pombe and C. thermophilum indicated that the Twix is uniquely present in budding yeast (Figure S4C-D). The Twix region is situated between α-helices 26 and 29 of Kog1 (Figure S4C-D), and interacts with the FRB’ domain of Tor2’ in an adjacent TORC1’ subunit with residues Gln2039, Asp2042 and Tyr2045 of Tor2’ respectively interacting with Glu865, Tyr829/Asn872 and Asn 876 of Kog1 (Figure 4D). Adjacent to the Twix region, we observed density for an α-helix not directly connected to surrounding parts of the structure, yet in close proximity to Lys1016 of Kog1 which terminates α-helix 34 (residues 1005 – 1016, Figure S4D). Careful inspection of the sequence succeeding Lys1016 and ensuing model building in Coot led to the identification of the region as being a helix with sequence ‘P₁₀₆₉MRTSLAKLFQSLGFSES₁₀₈₆’ (Figures 3D, 4D, S4E), which we termed the ‘Tack’ region. Sequence alignments demonstrated that this region precedes the sequence that aligns with the RAPTOR ‘Claw’ which was shown to interact with RagC in RAPTOR-Rag-Ragulator reconstructions ^22,23. In this vicinity, the structure displays a number of prominent molecular interactions leading to a robust TORC1-TORC1’ binding interface that we term the hub of interactions. Phe1083 and Ser1086, at the very tip of the Tack α-helix, make Van der Waals interactions with Pro2201 of the KD’ region of Tor2’ and Arg884’ of Kog1’ respectively (Figure 4D). Adjacent to Pro2201, Trp2207 sits in between Arg884, Lys885 and Arg888 of Kog1, forming a prominent cation-pi interaction with Arg884 situated near the tip of the Tack α-helix and the base of the Twix region (Figure 4D). The Twix and the Tack shield the kinase active site and thereby block substrate access, providing, together with the large allosteric change described above, a structural explanation for why TORC1 in TOROIDs is inactive (Figure 4D, Figure S4C).

Mutations confirm observed TORC1/TORC1’ interfaces

To interrogate the structural insights obtained from the observed interfaces, we selected sets of residues in the inter- and intra-coil interfaces for mutation using CrispR/Cas9 (Figure 5A-B and S5). The impact of these mutations was assessed by determining the relative size of TORC1 puncta in cells grown PDS (Figure S5E). In parallel, all mutants were tested in spot assays on rapamycin plates to identify mutations that compromise TORC1 activity (Figure S5F). In these analyses, we first assessed what we considered to be control mutations – mutation of exposed residues in the vicinity of the intra- and inter-coil interfaces, but not obviously engaged in TORC1-TORC1’ interactions. As expected, these mutations (Kog1^K1383A and Kog1^Δ1544−end) did not significantly affect TOROID size or rapamycin sensitivity (Figure S5A, C and F). In contrast, mutation of W1279 of Tor1, which is not predicted to mediate contacts within the helix but is predicted to be important to stabilize the FAT domain, yielded a reduction in TOROID size as well as rapamycin hypersensitivity, consistent with a loss of TORC1 particle integrity (Figure S5A, C and F).

Inter-coil interactions, although clearly visible in the helical reconstruction, appear almost completely absent in the signal subtracted map, and only become apparent when using a higher contour threshold level during visualization. Based on our helical reconstruction, the inter-coil interface is predominantly formed between the Tor2 HORN and the Kog1’ WD domain (Figure 3). To validate this, two exposed loops in the HORN, Tor2^337–345 (Tor1^326–334) and Tor2^379–381 (Tor1^368–370), were substituted by poly alanine-glycine stretches. These strains displayed a significant reduction in TOROID size as well as an increase in rapamycin resistance (Figure S5B and S5D-F). An additional protruding region of the Tor2 kinase domain also appeared to interact with Kog1’. Although attempts to generate mutations in this region of Tor2 were unsuccessful, mutation of the equivalent region of Tor1 (Tor1^1449/54/56A) similarly resulted in a reduced TOROID size and increased rapamycin resistance (Figure S5B and S5D-F).

Given the higher resolution of this region in the signal subtracted map, amino acids involved in intra-coil interactions were easier to trace (Figure 5 and S5A, S5C and S5E-F). Intra-coil interactions are plentiful, and mainly contributed by Kog1 interacting with Kog1’, Lst8’, and the KD’ and FAT’ regions of Tor2’ from an adjacent TORC1’ unit. Glu784 and Ser896 of Kog1 interacts with His292 of Lst8’ and alanine substitution of this His reduces TOROID formation (Lst8^H292A; Figure 5A-B). Similar results were obtained with reciprocal mutations in Kog1 (Kog1^E784A, Kog1^S895–896A Figure 5A-B and S5E-F). The backbone carbonyl oxygen of Gly781 in Kog1 interacts with Gln29 of Lst8’ and alanine substation of this Gln residue strongly reduces TOROID formation and leads to a significant increase in Rapamycin resistance (Figure S5A, S5C and S5E-F).

Striking amongst the intra-coil interactions, and as alluded to above, is the TORC1-TORC1’ hub of interactions, formed through extensive contacts between Kog1 and Tor2’. These contacts include interactions between the Twix helices of Kog1 and the FRB domain of Tor2’ (Figure S4C-D), the tip of the Tack helix of Kog1 and the KD of Tor2’ (Figure 4D, S4C and 5), and a basic pocket of Kog1 that hosts Trp2207 from the Lst8 Binding Element (LBE) of Tor2’. The apparent cation-pi interaction between Tor2^W2207 and Kog1^R884 where particularly intriguing as mutations in the corresponding residues of mTOR and RAPTOR (mTOR^R2266P and Raptor^D635N) have been identified in human carcinomas (https://cancer.sanger.ac.uk/cosmic). To test if this cation-pi interaction is important for TORC1 regulation, we assessed the consequences of combined mutations of Trp2207 of Tor2 and the analogous site in Tor1 (Tor1^W2203R, Tor2^W2207R), anticipating to create a charge repulsion with Kog1’. Consistently, in this strain, TOROID formation was absent (Figure 5C and S5E), and TORC1 was hyperactive as shown by increased rapamycin resistance (Figure S5F). We attempted to rescue these phenotypes by reverting Arg884 of Kog1 to Asp, the corresponding amino acid found in RAPTOR. However, KOG1^R884D neither blocked TOROID formation on its own, nor did it rescue the increased rapamycin resistance and lack of TOROIDs of TOR1^W2203R TOR2^W2207R cells (data not shown), perhaps owing to the remaining positively charged residues surrounding Arg884 (see below). The potential significance of these cancer associated mutations is discussed below.

TOROID-like structures have not been reported in other systems, including mammals and we wondered if the yeast-specific Twix helices of Kog1 might be a reason why. To test if the Twix region is necessary for TOROID formation, and considering the poor local sequence conservation, we initially, and rather conservatively, replaced a region from α26 to α29 (including the Twix helices: α27 α28), with the corresponding, smaller region of human Raptor to generate an allele that we named kog1 Humanized Chimera (kog1^HC; Figure 5C). Cells expressing this allele of KOG1 cultured PDS formed TOROIDs indistinguishably from WT cells (Figure 5B, C and S5E) suggesting that the Twix region is not required for TOROID formation (see also below).

kog1 ^HC allele unveils the TORC1/EGOC interaction domain

The kog1^HC allele also replaces helix α29 of yeast Kog1 which contains the three basic residues (Kog1^R884, Kog1^K885 and Kog1^R888) thought to interact with Trp2207 of Tor2 with the equivalent helix α29 of Raptor which contains acidic amino acids instead of these arginines (Figure S4D). Although KOG1^R884D did not suppress the inability of TOR1^W2203R TOR2^W2207R cells to form TOROIDs we wondered if kog1^HC might. Indeed, we found this to be the case (Figure S5E) but, as presented below, not necessarily through charge complementation of Tor1^W2203R Tor2^W2207R.

As expected, the TOR1^W2203R TOR2^W2207R kog1^HC strain presented a strong reduction in rapamycin resistance compared to the TOR1^W2203R TOR2^W2207R strain (Figure S5F). However, we unexpectedly noticed that kog1^HC cells displayed hypersensitivity to rapamycin, to a level comparable to ∆gtr1 ∆gtr2 cells (Figure S5F). One explanation for the rapamycin hypersensitivity of kog1^HC cells would be if they, like ∆gtr1 ∆gtr2 cells, constitutively from TOROIDs even in the presence of glucose. Thus, we decided to re-assess TORC1 localization in kog1^HC cells, as well as in all of the other mutant strains that we had created, in glucose-replete conditions. Amongst all of these strains, the kog1^HC strain alone displayed TOROIDs in glucose-replete conditions similarly to ∆gtr1 ∆gtr2 cells (Figure 6A-D and data not shown).

Guessing that these phenotypes might be due to the human sequences (helices α28 and/or α29 of Raptor) that were introduced during the construction of kog1^HC, we generated a new kog1 allele in which the Twix helices α27 and α28 of Kog1 were “cleanly” replaced with a short (AG)₄ linker (Figure 6A). We named this allele, kog1^ΔTwix. kog1^ΔTwix, like kog1^HC, formed TOROIDs normally when cells were cultured PDS (Figure 6B), but, unlike kog1^HC, do not form TOROIDs in exponential phase and displayed nearly normal sensitivity to rapamycin (Figure 6C). These observations confirm that the Twix helices are not a prerequisite for TOROID formation or normal TORC1 activity but that Kog1 helix α26 and/or α29 is important for basal TORC1 activity in a manner that cannot be complemented by the corresponding Raptor sequence.

We hypothesized that the phenotypes observed in the kog1^HC strain could be explained if TORC1 in these cells was unable to engage the EGOC. Several additional observations support this model. Similarly to ∆gtr1 ∆gtr2 cells, in kog1^HC cells, EGOC puncta were not observed (Figure 6D-E), and TORC1 activity, as assessed by Sch9 phosphorylation, was reduced and largely insensitive to acute changes in glucose levels (Figure 6G). Furthermore, in pulldown experiments, unlike WT TORC1, TORC1 containing Kog1^HC failed to robustly interact with purified EGOC (Figure 6H-I, S6A). Collectively, these results suggest that a key binding site for EGOC lies within helix α26 and/or α29 of Kog1, i.e. within the TOROID hub of interactions. They also demonstrate that TOROID assembly is necessary, but not sufficient for EGOC puncta formation (see Figure 2E, F; cells expressing inactive Gtr1^GDP/Gtr2^GTP, do not efficiently form EGOC puncta in glucose-replete conditions despite the presence of TOROIDs).

Based on these observations we wondered if kog1^HC suppressed the inability of TOR1^W2203R TOR2^W2207R cells to form TOROIDs because of its own inability to bind the EGOC rather than, or in addition to, charge complementation. Consistent with this idea we observed that the ∆gtr1 ∆gtr2 double deletion was epistatic to TOR1^W2203R TOR2^W2207R: ∆gtr1 ∆gtr2 TOR1^W2203R TOR2^W2207R cells presented constitutive TOROIDs (Figure S6B). The implications of this observation are discussed below.

CrispR-based random mutagenesis confirms that the EGOC-TORC1 interaction interface overlaps with the TORC1-TORC1’ hub of interactions

Using the strains generated thus far, we observed a strong anti-correlation between TOROID size in PDS and rapamycin sensitivity (Pearson=-0.75; Figure S6C). Mutations compromising TORC1-TORC1’ interactions have small TOROIDs and are resistant to rapamycin while mutations abrogating TORC1-EGOC binding have large TOROIDs and are sensitive to rapamycin. Thus it appears that sensitivity to rapamycin depends on the proportion of free and active TORC1 dimers (Figure S6C). Exploited this idea we set up a CrispR-based PCR random mutagenesis screen to further define the TORC1-EGOC interface. A portion of the ARM domain of Kog1 surrounding α26 - α29 (residues 696-922) was chosen for mutagenesis (Figure 7A). The first selection step was rapamycin-hypersensitivity, expecting that mutants of interest would present a hypersensitive phenotype similar to Δgtr1/2 and kog1^HC strains. The second step was to select for TOROID formation in glucose-replete cultures. A total of 3500 mutants were screened on 2.5 and 4 nM rapamycin plates. While the latter concentration appeared to be too stringent, selection with 2.5 nM rapamycin yielded 42, rapamycin-sensitive clones. Among these, 13 presented TOROIDs in glucose-replete media and these were retained and sequenced (Figure 7A and S7A-D). 17-point mutations were identified, of which 7 (marked in bold below) were the only mutation present in the clone and could thus be deemed causative (Figure 7B, S7B-D). Remarkably, while the Twix-encoding region was included in this screen, no mutation within this sequence was recovered, supporting the idea that it does not participate in EGOC-dependent regulation of TOROIDs.

Several mutations, alone (bold) or in combination, altered leucine residues involved in hydrophobic interactions between α-helices (L762P, L765P, L766P, L900P, L912Q; Figure 7B and S7C-D) suggesting that local perturbation of domain packing may affect EGOC binding. L737Q (found in combination with L762P), at the junction of α21-22 helices, C777R and M778T in helix α24, I802N and A804E in helix α26, and I891N in helix α29 were recovered (Figure 7B and S7B-D) and we note that many of these map to a region which in Raptor is important for interaction with RagA (Raptor residues N557, C594 and 634-36; Figure 7B and alignment Figure S4D); ²³). Building on this observation we found that Kog1 helix α21 (which corresponds to α23 in Raptor, known to be important for RagA binding) is essential for proper TOROID dynamics since its substitution to an Alanine-Glycine stretch triggers constitutive TOROID formation (Figure 7C). Finally, we found that purified TORC1 harboring Kog1^C777R or AG-substituted Kog1 helix α21 respectively presented much reduced or no EGOC binding activity (Figure 7D). Thus, Kog1 appears to interact with the GTRs similarly to how Raptor interacts with the RAGs.

Remarkably, the Tack helix appears to correspond to the Claw helical fragment of Raptor described to interact with RagC. To probe the importance of the Tack helix in TOROID regulation we made three poly AG substitutions. Replacement of the proximal α-helix 34 (Kog1^1004–1022) had no effect on TOROID regulation, while replacement of the entire Tack helix (Kog1^1121–1131) leads to a mild, but significant, impairment of TOROID formation (Figure S7E, F). Replacement of the putative yeast Claw sequence (Kog1^1121−1131; ²³ presented no significant effect on either TOROID formation or regulation (Figure S7E-F), whereas equivalent mutations in Raptor prevent Rag-binding ²³ which we predict would lead to constitutive TOROID formation. From these observations we conclude that, although similarly localized, the Tack helix does not function equivalently to the Raptor Claw. Instead, it provides structural elements (Kog1^F1083 binding to Tor2’^P2201) needed to stabilize the TOROID.

Based on these collective observations, we propose that, analogous to the binding observed in mammalian cells, EGOC binds to helices α21, α24, α26 and α29 of Kog1. This binding (Figure 7G) competes with a critical interface needed for TOROID formation, and would thus release/sequester free TORC1 dimers that are competent to signal to downstream effectors.

We previously reported ²⁷ that glucose withdrawal triggers a rapid, Gtr1/2-dependent reorganization of TORC1 from a disperse localization over the surface of the vacuole to a hollow helical assembly that we named a TOROID. Our, at this time, low-resolution, 3D reconstruction revealed that TOROID-sequestered TORC1 is inactive due, at least in part, to rapamycin-FKBP12-like occlusion of the kinase active site. This unexpected mechanism underlying TORC1 regulation differs from the widely accepted model of mammalian TORC1 regulation and raised some important questions, some of which we have addressed in the present study.

We show here that the EGOC alone is sufficient to regulate TORC1/TOROID dynamics. Although we conclude that EGOC-mediated TOROID formation represents an important mechanism for the cell to regulate TORC1 signaling output, we do not wish to imply that this is the only mechanism through which glucose or other stimuli regulate TORC1. Indeed, in the absence of TORC1-EGOC interactions (e.g. ∆gtr1 ∆gtr2 or kog1^HC cells) TORC1 is still partially regulated by glucose withdrawal/repletion (Figure 6G,²⁷) and by osmotic shock (data not shown). This clearly indicates the existence of EGOC-independent TORC1 regulators. How such regulators (e.g. Snf1 and Pib2 ^46,47 interface with TORC1/TOROID-based regulation remains an important unknown. The observation that some TORC1 activity remains in ∆gtr1 ∆gtr2 cells – and mTORC1 activity in ragA ragB ragC ragD cells ⁴⁸ – implies that not all TORC1 is sequestered into TOROIDs in this genetic background. Indeed, we favor a model in which there is a dynamic equilibrium between active, free TORC2 dimers and inactive TOROID polymers in a cell. Mutations or stimuli that shift the equilibrium towards free dimers would stimulate TORC1 signaling output, and vice versa. This hypothesis is supported by the strong anti-correlation that we see between TORC1 puncta size and rapamycin resistance (Supplementary Figure 6B). In such a model, one could envision that non-EGOC TORC1 regulators independently act on the pool of free TORC1 dimers. Additionally, distinct pools of TORC1 (e.g. endosomal TORC1; ²⁶ could be subjected to distinct upstream regulatory mechanisms.

The major advance of our present work, however, is the molecular insights we provide to explain how binding of EGOC liberates free TORC1 from TOROID polymers. Our new, high-resolution TORC1 map reveals that TOROIDs are rather spring-like with loose inter-coil interactions and robust intra-coil interactions. The weak inter-coil interactions bestow flexibility upon the TOROID structure for an as of yet unknown purpose. In contrast, the strong intra-coil interactions serve to occlude access to the active site, which, together with allosteric changes to the catalytic spine, ensures that TOROID-associated TORC1 is inactive. Closer inspection of the TORC1-TORC1’ interface proximal to the kinase active site revealed an intriguing hub of interactions primarily involving Kog1 and Tor2’. Within this hub lies the budding-yeast-specific Twix helices (Kog1 α27 and α 28), that we posited might underlie TOROID formation which, to date, has also only been observed in budding yeast. While performing experiments from which we ultimately concluded that the Twix per se does not play an essential role in TOROID formation, we discovered that the region flanking the Twix does. Specifically, using a CrispR/cas9-supported random mutagenesis screen, we found that α-helices 21-26 and 29 of Kog1 serve as a binding platform for Gtr1 and Gtr2 of the EGOC. This observation was satisfying but not obvious: satisfying because the corresponding region of Raptor (α23- α29) was previously found to bind to RagA and RagC (Rogala et al., 2019); not obvious because the positioning of the Twix helices should occlude GTPase binding. We hypothesize that the Twix helices are flexible - potentially more so at the TOROID termini - and that they are displaced upon EGOC binding. The molecular details of this displacement await a TORC1-EGOC structure. Nevertheless, these observations clearly demonstrate that EGOC binding to this region of Kog1 disrupts the hub of interactions needed to sustain TOROID assembly, providing a mechanism to explain how the EGOC can liberate and/or maintain free, active TORC1 dimers – the major conclusion of this paper.

Although this model of TORC1 regulation is appealing, it is oversimplified, as it does not take into account the nucleotide loading status of Gtr1 and Gtr2. A priori, one could posit that only “active” EGOC binds TORC1, preventing its association into a TOROID. However, compelling evidence suggests that inactive EGOC also binds TORC1:

i/ In in vitro pulldowns we observe that TORC1 interacts with “inactive” EGOC even more robustly than “active” EGOC (Figure 2H).

ii/ Expression of “inactive” gtr alleles produce phenotypes that are more severe than simple gtr1/2 deletion ²⁵. Consistently, deletion of LST4 or LST7 (Subunits of the Gtr2 GAP; LST = Lethal with Sec Thirteen) shows synthetic lethality with hypomorphic alleles of sec13 ⁴⁹ which encodes a SEACAT component.

iii/ “Inactive” EGOC partially colocalizes with TOROIDs (Figure S2).

The RAG GTPases do not solely interact with TORC1; they also interact with their respective GAPs. Interestingly, recent evidence suggests that with both GATOR1 and Folliculin-Fnip, the RAGs have multiple binding modes ^18,50,51. Based on this previous literature and the arguments presented above, we speculate that the EGOC may also have multiple TORC1-interaction modes.

Is mammalian TOR similarly regulated via a TOROID like structure? We do not yet have an answer for this important question. As far as we are aware, comparison of yeast and mammalian TORC1 structures does not present any obvious structural argument to preclude the existence of a mammalian TOROID. mTORC1 resides on lysosomes, which range in diameter from 0.1 to 1 µm ⁵². In contrast, a yeast vacuole is around 2 – 3 µm in diameter. Thus, there are likely far fewer mTORC1s per lysosome compared to the 500-1000 yeast TORC1s per vacuole. This implies that if mTOROIDs do exist their appearance by light microscopy would be far less striking than what we observe in yeast. Furthermore, the mTOR field almost exclusively uses a single antibody - Cell Signaling 7C10 – for IHC studies of mTOR. This antibody recognizes a phospho-epitope that may not be accessible or even present in a putative mTOROID helix. Thus, it is very possible that mTORC1 in mTOROIDs would not be visible using this reagent. We note that the RagA^GDP RagC^GTP conformation of the RAGs, often considered to be inactive, in some studies has been shown to potently inhibit mTORC1 function ⁵³. This observation is not consistent with the current model that active RAGs simply recruit mTORC1 to the lysosome for subsequent activation by Rheb. In conclusion, we believe that it is not yet appropriate to rule out the existence of mTOROIDs.

The presence of mTOROIDs could explain why mTOR^R2266P, and possibly Raptor^D635N, are associated with carcinoma based on our finding that substitution of the analogous residue in Tor1^W2203R Tor2^W2207R generated cells that did not form TOROIDs and appeared to possess hyperactive TORC1 signaling (i.e. were resistant to rapamycin). This double mutant is intriguing as it still requires the Gtrs to prevent TOROID formation. Perhaps this mutant can use the binding of “inactive” EGOC to prevent TOROID formation. Alternatively, the mutant may possess a higher affinity for low levels of “active” EGOC in PDS cells. Further characterization of this mutant is clearly warranted.

ACKNOWLEDGMENTS

R.L. acknowledges support from the Canton of Geneva, the Swiss National Science Foundation, SNF Sinergia METEORIC, the Swiss National Centre for Competence in Research Chemical Biology and the European Research Council AdG TENDO. I.G. is supported by the European Union’s Horizon 2020 research and innovation programme under grant agreement No 647784 to IG. S.N.S. is supported by the Flanders Institute for Biotechnology (VIB). J.F. was supported by a long-term European Molecular Biology Organization (EMBO) fellowship (ALTF441-2017) and a Marie Skłodowska-Curie actions individual fellowship (789385, RespViRALI). Support and use of iNEXT resources, the cryo-electron microscopy (cryo-EM) service platform at EMBL Heidelberg and CryoGEnic facility (DCI Geneva) is acknowledged. J.F. and A.D. acknowledge the use of the Grenoble IBS-EMBL computer cluster and the technical assistance of Aymeric Peuch.

AUTHOR CONTRIBUTIONS

Genetic engineering and cellular biology experiments were carried out by M.P., C.B., L.B. and R.L. EGOC expression and purification were performed by C.G. and TOROID purifications, pull downs and cryo-EM grid preparations were performed by M.P. and C.B. EM data collection were performed by M.P., C.B., J.F. and A.D. Cryo-EM data processing, model building and refinement were carried out by J.F. and A.D. with additional input from Y.S. Experimental design was conducted by M.P., C.B., J.F., I.G., A.D. and R.L. The first draft of the manuscript was written by M.P., C.B., J.F. and R.L. with inputs from all authors.

DECLARATION OF INTERESTS

The authors declare no conflicts of interest.

Keith, C.T. & Schreiber, S.L. PIK-related kinases: DNA repair, recombination, and cell cycle checkpoints. Science 270, 50–1 (1995).
Tatebe, H. & Shiozaki, K. Evolutionary Conservation of the Components in the TOR Signaling Pathways. Biomolecules 7(2017).
Gaubitz, C., Prouteau, M., Kusmider, B. & Loewith, R. TORC2 Structure and Function. Trends Biochem Sci 41, 532–545 (2016).
Riggi, M. et al. Decrease in plasma membrane tension triggers PtdIns(4,5)P-2 phase separation to inactivate TORC2. Nature Cell Biology 20, 1043-+ (2018).
Eltschinger, S. & Loewith, R. TOR Complexes and the Maintenance of Cellular Homeostasis. Trends in Cell Biology 26, 148–159 (2016).
Loewith, R. et al. Two TOR complexes, only one of which is rapamycin sensitive, have distinct roles in cell growth control. Mol Cell 10, 457–68 (2002).
Reinke, A. et al. TOR complex 1 includes a novel component, Tco89p (YPL180w), and cooperates with Ssd1p to maintain cellular integrity in Saccharomyces cerevisiae. J Biol Chem 279, 14752–62 (2004).
Urban, J. et al. Sch9 is a major target of TORC1 in Saccharomyces cerevisiae. Mol Cell 26, 663–74 (2007).
Huber, A. et al. Characterization of the rapamycin-sensitive phosphoproteome reveals that Sch9 is a central coordinator of protein synthesis. Genes Dev 23, 1929–43 (2009).
Kamada, Y. et al. Tor-mediated induction of autophagy via an Apg1 protein kinase complex. J Cell Biol 150, 1507–13 (2000).
MacGurn, J.A., Hsu, P.C., Smolka, M.B. & Emr, S.D. TORC1 regulates endocytosis via Npr1-mediated phosphoinhibition of a ubiquitin ligase adaptor. Cell 147, 1104–17 (2011).
Zimmermann, C. et al. TORC1 Inhibits GSK3-Mediated Elo2 Phosphorylation to Regulate Very Long Chain Fatty Acid Synthesis and Autophagy. Cell Rep 18, 2073–2074 (2017).
Liu, G.Y. & Sabatini, D.M. mTOR at the nexus of nutrition, growth, ageing and disease. Nat Rev Mol Cell Biol 21, 183–203 (2020).
Aylett, C.H. et al. Architecture of human mTOR complex 1. Science 351, 48–52 (2016).
Baretic, D., Berndt, A., Ohashi, Y., Johnson, C.M. & Williams, R.L. Tor forms a dimer through an N-terminal helical solenoid with a complex topology. Nat Commun 7, 11016 (2016).
Yang, H. et al. Mechanisms of mTORC1 activation by RHEB and inhibition by PRAS40. Nature 552, 368–373 (2017).
Shen, K., Choe, A. & Sabatini, D.M. Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability. Molecular Cell 68, 552-+ (2017).
Shen, K. & Sabatini, D.M. Ragulator and SLC38A9 activate the Rag GTPases through noncanonical GEF mechanisms. Proc Natl Acad Sci U S A 115, 9545–9550 (2018).
Kim, J. & Kim, E. Rag GTPase in amino acid signaling. Amino Acids 48, 915–928 (2016).
Nada, S. et al. The novel lipid raft adaptor p18 controls endosome dynamics by anchoring the MEK-ERK pathway to late endosomes. EMBO J 28, 477–89 (2009).
Sancak, Y. et al. Ragulator-Rag complex targets mTORC1 to the lysosomal surface and is necessary for its activation by amino acids. Cell 141, 290–303 (2010).
Anandapadamanaban, M. et al. Architecture of human Rag GTPase heterodimers and their complex with mTORC1. Science 366, 203–210 (2019).
Rogala, K.B. et al. Structural basis for the docking of mTORC1 on the lysosomal surface. Science 366, 468–475 (2019).
Morozumi, Y. & Shiozaki, K. Conserved and Divergent Mechanisms That Control TORC1 in Yeasts and Mammals. Genes (Basel) 12(2021).
Binda, M. et al. The Vam6 GEF controls TORC1 by activating the EGO complex. Mol Cell 35, 563–73 (2009).
Hatakeyama, R. et al. Spatially Distinct Pools of TORC1 Balance Protein Homeostasis. Mol Cell 73, 325-338 e8 (2019).
Prouteau, M. et al. TORC1 organized in inhibited domains (TOROIDs) regulate TORC1 activity. Nature 550, 265–269 (2017).
Zhang, T. et al. Structural insights into the EGO-TC-mediated membrane tethering of the TORC1-regulatory Rag GTPases. Sci Adv 5, eaax8164 (2019).
Gao, M. & Kaiser, C.A. A conserved GTPase-containing complex is required for intracellular sorting of the general amino-acid permease in yeast. Nat Cell Biol 8, 657–67 (2006).
Peli-Gulli, M.P., Sardu, A., Panchaud, N., Raucci, S. & De Virgilio, C. Amino Acids Stimulate TORC1 through Lst4-Lst7, a GTPase-Activating Protein Complex for the Rag Family GTPase Gtr2. Cell Rep 13, 1–7 (2015).
Powis, K. & De Virgilio, C. Conserved regulators of Rag GTPases orchestrate amino acid-dependent TORC1 signaling. Cell Discov 2, 15049 (2016).
Sekiguchi, T., Hirose, E., Nakashima, N., Ii, M. & Nishimoto, T. Novel G proteins, Rag C and Rag D, interact with GTP-binding proteins, Rag A and Rag B. J Biol Chem 276, 7246–57 (2001).
Broach, J.R. Nutritional Control of Growth and Development in Yeast. Genetics 192, 73–105 (2012).
Dechant, R., Saad, S., Ibanez, A.J. & Peter, M. Cytosolic pH regulates cell growth through distinct GTPases, Arf1 and Gtr1, to promote Ras/PKA and TORC1 activity. Mol Cell 55, 409–21 (2014).
Sullivan, A., Wallace, R.L., Wellington, R., Luo, X. & Capaldi, A.P. Multilayered regulation of TORC1-body formation in budding yeast. Mol Biol Cell 30, 400–410 (2019).
Gao, X.D., Wang, J., Keppler-Ross, S. & Dean, N. ERS1 encodes a functional homologue of the human lysosomal cystine transporter. FEBS J 272, 2497–511 (2005).
Prouteau, M., Daugeron, M.C. & Seraphin, B. Regulation of ARE transcript 3' end processing by the yeast Cth2 mRNA decay factor. EMBO J 27, 2966–76 (2008).
Dubouloz, F., Deloche, O., Wanke, V., Cameroni, E. & De Virgilio, C. The TOR and EGO protein complexes orchestrate microautophagy in yeast. Mol Cell 19, 15–26 (2005).
Weisman, L.S. Organelles on the move: insights from yeast vacuole inheritance. Nat Rev Mol Cell Biol 7, 243–52 (2006).
Kira, S. et al. Reciprocal conversion of Gtr1 and Gtr2 nucleotide-binding states by Npr2-Npr3 inactivates TORC1 and induces autophagy. Autophagy 10, 1565–78 (2014).
Kira, S. et al. Dynamic relocation of the TORC1-Gtr1/2-Ego1/2/3 complex is regulated by Gtr1 and Gtr2. Mol Biol Cell 27, 382–96 (2016).
Ukai, H. et al. Gtr/Ego-independent TORC1 activation is achieved through a glutamine-sensitive interaction with Pib2 on the vacuolar membrane. PLoS Genet 14, e1007334 (2018).
de la Rosa-Trevin, J.M. et al. Scipion: A software framework toward integration, reproducibility and validation in 3D electron microscopy. Journal of Structural Biology 195, 93–99 (2016).
Ilca, S.L. et al. Localized reconstruction of subunits from electron cryomicroscopy images of macromolecular complexes. Nature Communications 6(2015).
Punjani, A., Zhang, H.W. & Fleet, D.J. Non-uniform refinement: adaptive regularization improves single-particle cryo-EM reconstruction. Nature Methods 17, 1214-+ (2020).
Hughes Hallett, J.E., Luo, X. & Capaldi, A.P. Snf1/AMPK promotes the formation of Kog1/Raptor-bodies to increase the activation threshold of TORC1 in budding yeast. Elife 4(2015).
Tanigawa, M. et al. A glutamine sensor that directly activates TORC1. Communications Biology 4(2021).
Lawrence, R.E. et al. A nutrient-induced affinity switch controls mTORC1 activation by its Rag GTPase-Ragulator lysosomal scaffold. Nat Cell Biol 20, 1052–1063 (2018).
Roberg, K.J., Bickel, S., Rowley, N. & Kaiser, C.A. Control of amino acid permease sorting in the late secretory pathway of Saccharomyces cerevisiae by SEC13, LST4, LST7 and LST8. Genetics 147, 1569–1584 (1997).
Shen, K. et al. Cryo-EM Structure of the Human FLCN-FNIP2-Rag-Ragulator Complex. Cell 179, 1319-1329 e8 (2019).
Lawrence, R.E. et al. Structural mechanism of a Rag GTPase activation checkpoint by the lysosomal folliculin complex. Science 366, 971–977 (2019).
Novikoff, A.B., Beaufay, H. & De Duve, C. Electron microscopy of lysosomerich fractions from rat liver. J Biophys Biochem Cytol 2, 179–84 (1956).
Yang, S. et al. The Rag GTPase Regulates the Dynamic Behavior of TSC Downstream of Both Amino Acid and Growth Factor Restriction. Developmental Cell 55, 272-+ (2020).
Gibson, D.G. Synthesis of DNA fragments in yeast by one-step assembly of overlapping oligonucleotides. Nucleic Acids Res 37, 6984–90 (2009).
Laughery, M.F. et al. New vectors for simple and streamlined CRISPR-Cas9 genome editing in Saccharomyces cerevisiae. Yeast 32, 711–20 (2015).
Guichard, P., Hamel, V., Neves, A. & Gonczy, P. Isolation, cryotomography, and three-dimensional reconstruction of centrioles. Centrosome & Centriole 129, 191–209 (2015).
Zheng, S.Q. et al. MotionCor2: anisotropic correction of beam-induced motion for improved cryo-electron microscopy. Nature Methods 14, 331–332 (2017).
Rohou, A. & Grigorieff, N. CTFFIND4: Fast and accurate defocus estimation from electron micrographs. Journal of Structural Biology 192, 216–221 (2015).
Tang, G. et al. EMAN2: An extensible image processing suite for electron microscopy. Journal of Structural Biology 157, 38–46 (2007).
Scheres, S.H. RELION: implementation of a Bayesian approach to cryo-EM structure determination. J Struct Biol 180, 519–30 (2012).
Zhang, K. Gctf: Real-time CTF determination and correction. Journal of Structural Biology 193, 1–12 (2016).
Rosenthal, P.B. & Henderson, R. Optimal determination of particle orientation, absolute hand, and contrast loss in single-particle electron cryomicroscopy. Journal of Molecular Biology 333, 721–745 (2003).
Punjani, A., Rubinstein, J.L., Fleet, D.J. & Brubaker, M.A. cryoSPARC: algorithms for rapid unsupervised cryo-EM structure determination. Nature Methods 14, 290-+ (2017).
Sanchez-Garcia, R. et al. DeepEMhancer: a deep learning solution for cryo-EM volume post-processing. Commun Biol 4, 874 (2021).
Kelley, L.A., Mezulis, S., Yates, C.M., Wass, M.N. & Sternberg, M.J.E. The Phyre2 web portal for protein modeling, prediction and analysis. Nature Protocols 10, 845–858 (2015).
Roy, A., Kucukural, A. & Zhang, Y. I-TASSER: a unified platform for automated protein structure and function prediction. Nature Protocols 5, 725–738 (2010).
Pettersen, E.F. et al. UCSF chimera - A visualization system for exploratory research and analysis. Journal of Computational Chemistry 25, 1605–1612 (2004).
Lopez-Blanco, J.R. & Chacon, P. iMODFIT: Efficient and robust flexible fitting based on vibrational analysis in internal coordinates. Journal of Structural Biology 184, 261–270 (2013).
Kidmose, R.T. et al. Namdinator - automatic molecular dynamics flexible fitting of structural models into cryo-EM and crystallography experimental maps. Iucrj 6, 526–531 (2019).
Emsley, P., Lohkamp, B., Scott, W.G. & Cowtan, K. Features and development of Coot. Acta Crystallographica Section D-Biological Crystallography 66, 486–501 (2010).
Jumper, J. et al. Highly accurate protein structure prediction with AlphaFold. Nature 596, 583–589 (2021).
Tunyasuvunakool, K. et al. Highly accurate protein structure prediction for the human proteome. Nature 596, 590–596 (2021).

Strains, plasmids and media

Yeast strains, plasmids, media and reagents are listed in Supplementary Table3.

Yeast growth conditions

Yeasts were grown in complete synthetic media (CSM) buffered at pH 6.25 (Sorensen Buffer), lacking the appropriate(s) amino acid(s) to maintain plasmid(s) selection when necessary. Saturated yeast liquid pre-cultures grown in complete synthetic medium (CSM) were diluted into fresh CSM at OD₆₀₀ 0.3 and cultured 4 hours at 30°C to reach exponential growth (OD₆₀₀ ̴ 0.6) or 24 hours to reach Post Diauxic Shift (PDS). Confocal Z-stack images were acquired on LSM800 confocal microscope (Plan-Apochromat 63x/1.4 Oil Objective, Zeiss) at the relevant growth states.

For spot assays, exponentially growing cells were normalized to OD₆₀₀ 0.1, 5 μL from 4, serial 10-fold dilutions, were spotted on selective SC plates containing either rapamycin (at the indicated dilutions), Mohr salt (100 μM) or BPS (25 μM) and grown at 30 °C for 2-3 days. To measure the half-maximal effective concentration (EC₅₀) of Rapamycin, plates were imaged and the intensity of each spot was measured using ImageJ. At each Rapamycin concentration, a relative growth coefficient was measured as the slope of the linear regression based on the intensities of the 4 dilution spots. The relative growth coefficients were then plotted as a function of the Rapamycin concentration and EC₅₀s were extracted with Prism Software (version 8.3.0).

Iron deprivation treatment

EGO1^TEV cells transformed with a plasmid p(CTH2p)-3HA-TEV-(Cth2t) were grown overnight in buffered CSM-URA supplemented with 0.3 μM Mohr salt ((NH₄)₂Fe(SO₄)₂). Cultures were diluted to OD₆₀₀ 0.2-0.3 and grown for 5 hours before starting the iron deprivation treatment. This ensures the repression of the CTH2 promoter. Cells was filtered with hydrophilic mixed cellulose esters membrane, 65 μm pore size and rinsed with buffered CSM-URA to wash out potential aggregates of Mohr salt. Cells were resuspended in buffered CSM-URA supplemented with 0.1 μM BPS and either imaged by confocal microscopy or harvested for immunoblotting assay at different time points during 90 minutes.

Glucose deprivation and repletion

At OD₆₀₀ 0.6 cells growing exponentially at 30 °C in liquid CSM were filtered, washed once with starvation medium (CSM-glucose) and then transferred to CSM-glucose. 2% Glucose was subsequently added after 30 min of starvation. Confocal Z-stack images were acquired on LSM800 confocal microscope and/or aliquots for western blot analyses were taken at the relevant time points.

Yeast Mating experiments

Experiments were performed by co-spotting equivalent amounts of exponentially growing MATa and MATalpha cells, onto a YPD plate and leaving it at 30°C for 2.5 hours. The yeast mixture was then subsequently resuspended in liquid CSM at 30 °C and deposited on μ-slides VI^0.4 channels (ibidi) coated with Concanavalin A (0.5mg/mL) to be subsequently imaged by Z-stacks or time-series (1 image/ 30 seconds) Z-stacks using TCS SP5 confocal microscope (Plan-Apochromat 63x/1.4 Oil Objective, Leica).

Light Microscopy and analyses

All confocal images presented in the main and extended data are maximum projections of 3-5 images within a Z-stack (ImageJ). Z slices were initially separated by 500 nm. Image deconvolution was performed with Huygens Software. Analysis of TORC1 foci into size/volume classes followed 3D reconstructions of the stack image and intensity measurements using Imaris software ²⁷. The vacuolar enrichment of EGO3-GFP, EGO1^TEVsubjected to BPS treatment was measured with ImageJ. The mean intensity value in the cytosol was subtracted from the vacuolar membrane signal. The time of TOROID formation or dissolution upon EGOC depletion or replenishment was analysed with imageJ by measuring the intensity of TORC1 focus across the Z-stacks and over the duration of the acquisition. To quantify the degree of colocalization between TORC1 and EGOC, the intensity based colocalization function of Imaris Software was used to extract the Mander’s and Pearson coefficients. Alternatively, measure of reciprocal puncta co-localisation was performed using cell segmentation (FIJI) followed by puncta segmentation according to their size in each channel (inferior or superior to 500 nm; Imaris). Signals from both channels were extracted and subjected to Pearson analysis to measure their correlation.

CrispR/Cas9 mutageneses

pML104 plasmid expressing Cas9 enzyme and Guide RNA (from addgene) was linearized by PCR (Supplementary Table 3) and the 20-mer guide sequence targeting a specific yeast genome site was inserted by Gibson cloning ⁵⁴. Mutagenesis was achieved by genomic cut and recombination performed by co-transforming yeast strain with the specific plasmid and the codon optimized PCR product ⁵⁵. Short mutagenic PCR products (< 100 nucleotides) were obtained using overlapping forward and reverse primers (see Supplementary Table 2) while long mutagenic PCR products (> 100 nucleotides) were obtained using codon optimized gene sequences (GenScript; See Supplementary Table 3). Yeast were plated onto CSM-URA plates and then replicated on the same plates to decrease the number of false positive clones. CrispR efficiency was estimated by comparing the number of clones obtained with and without co-transformation with PCR product. 10 clones were then subjected to colony PCR (and subsequently sequenced to verify the mutagenesis. CrispR/Cas9 based Random PCR mutagenesis was performed using GeneMorphII Random Mutagenesis Kit (Agilent Technologies) aiming for 1-2 mutation(s) per 500 bp of PCR product. Transformants were plated onto 10-cm CSM-URA plates at a density of 100-200 clones per plate, and then replicated onto CSM-URA containing or not 2.5 nM Rapamycin. Rapamycin-sensitive clones were then sequenced using the same approach as previously described

Western Blot analysis

Yeast cultures were treated following standard TCA-Urea extraction procedures. Protein lysates were loaded on a 4-20 gradient SDS-PAGE gel and transferred to a nitrocellulose membrane using the iBlot^® system (ThermoFisher). The membrane was probed with primary antibodies overnight at 4 °C, washed and incubated with secondary antibodies for 45 min at RT. The membrane was developed using the Odyssey^® imaging system (LI-COR) and the results were quantified using ImageStudio™ Lite (LI-COR).

Primary antibodies used were rabbit polyclonal anti-Sch9, mouse monoclonal anti-P-Sch9^S758 (In house at a dilution of 1:5’000 and 1:2’500, respectively) and mouse monoclonal anti-HA (1:15’000, Sigma n°H9658). Secondary antibodies used were donkey anti-rabbit, IRDye 800 (Li-Cor Biosciencesand donkey anti-mouse, IRDye 680 (Li-Cor Biosciences). All secondary antibodies were used at a dilution of 1:10’000.

Plasmid design and protein purification

For the expression of all components of the EGOC, two vectors were designed: one for the expression of EGO-TC and the second for the dimer of GTPases. Coding sequences of EGO2(∆1-21)-EGO3-EGO1(∆1-111: His₆) were codon optimized for bacterial expression and synthetized by GenScript in pUC57. Each ORF was separated with an operon sequence (ttaactttaaaaaaaaaaaaacaggaggcaatatacat). The synthetized fragment was digested with NcoI and XhoI. The tricistronic ORF was cloned by T4 DNA ligation into a vector derived from pACYC (digested with NcoI and XhoI). The DNA sequences encoding full length GTR1 and GTR2 were amplified from S. cerevisiae gDNA (TB50 background). The bicistronic ORFs, separated by an operon sequence (aggaggaaaaaaaa), was cloned by T4 DNA ligation into a pET-derived vector. Alternatively, plasmids carrying mutations that lock Gtr1 and Gtr2 in their active or inactive conformations (Q65L/S23L or S20L/Q66L, respectively) were used as template for the PCR ²⁵.

GTPases or all EGOC components were (co)expressed in E. coli (BL21 DE3) and transformed cells were grown in 2xYT media at 37°C for 4 hours followed by overnight induction at 18°C with 0.1mM isopropyl-β-D-thiogalactopyranoside (IPTG). Cells were harvested by centrifugation and resuspended in lysis buffer for Ni-NTA purification [50mM Tris-Cl pH 7.4, 300mM NaCl, 5% Glycerol, 20mM Imidazole, 0.15% CHAPS, 1mM MgCl₂, 1 μg/ml DNase, 1 μg/ml Lysozyme, supplemented with protease inhibitors 1mM PMSF and cOmplete™ Protease Inhibitor Cocktail] and lysed using an Emulsiflex system (AVESTIN). Total lysate was cleared by centrifugation at 15,000 rpm for 45 minutes at 4°C. The soluble fraction was subjected to an affinity purification using a chelating HiTrap FF crude column (GE Healthcare) charged with Ni²⁺ ions on an AKTA-HPLC explorer. Proteins were washed and eluted in lysis buffer containing 250mM Imidazole. The purest fractions were concentrated to about 10mg/ml (Amicon 50kDA) and loaded on a Superdex GF200 Increase, equilibrated with the storage buffer [50mM MES-NaOH pH:6.0, 300mM NaCl, 0,15% CHAPS, 1mM MgCl₂, 0,5mM DTT]. Purified proteins were concentrated to 1-1.8 mg/ml.

Protein sequence Alignments

Protein sequence alignments were done using sequences from Saccharomyces cerevisiae, Schizosaccharomyces pombe, Chaetomium thermophilum and Homo sapiens sapiens and generated with ClustalX2. Kog1 orthologue sequences: S.c. QHB09177, S.p. CAB08769, C.t. XP_006691974 and H.s. Q8N122. Lst8 orthologue sequences: S.c. QHB11374, S.p. BAA32427, C.t. XP_006696584 and H.s. Q9BVC4. TOR orthologue sequences: S.c. Tor1: CAA52849 and S.c. Tor2: CAA50548, S.p. Tor2: NP_595359 and S.p. Tor1: NP_596275, C.t. XP_006695016 and H.s. mTOR P42345.

6x HIS Pull-down assay

Cells expressing KOG1 -TAP or kog1^HC-TAP were cultured in CSM and harvested either in exponential phase or PDS (OD₆₀₀ 2.5 or OD₆₀₀ 16, respectively). Cell pellets were frozen and further lysed by grinding in a cold mortar and pestle under liquid nitrogen. Cell extracts were resuspended in lysis buffer (50 mM Tris pH 7.4, 150 mM NaCl, 10 mM MgCl2, 0.1% NP-40, 1 mM PMSF, 3 tabs of cOmplete™ Protease Inhibitor Cocktail /100 mL of lysis buffer). Cell lysates were cleared by centrifugation at 4000 g at 4 °C for 10 min. Total protein concentrations were measured using Bradford protein assay and normalized to 5 mg/mL. Dynabeads® His-Tag Isolation & Pulldown (Novex) were washed once with binding buffer (50 mM Na-MES pH6, 600 mM NaCl, 1.5g/L CHAPS) and incubated with 20 µg of purified proteins (HIS₆-GST or Ego2-Ego3-His₆-Ego1, Gtr1&Gtr2) for 30 min at 4°C. Beads were washed once with lysis buffer and resuspended in the initial bead volume. For pull-down assays, 15 mg of total proteins and 25 µl microliters of coated Dynabeads were used per condition. The mixes were incubated for 60 min at 4 °C. The beads were washed 10 x with 2 ml ice-cold lysis buffer, boiled with 2 × SDS-PAGE sample buffer to elute the proteins and loaded onto 4–20% Mini-PROTEAN® TGX™ Precast Protein Gels. For immunoblot analysis, rabbit polyclonal anti-TAP (Open Biosystems CAB1001) and mouse monoclonal anti-polyHistidine antibody (clone HIS-1, Sigma H1029) were used at 1:2’000 and 1:4’000, respectively.

CryoEM sample preparation

TORC1 filament samples were adsorbed for approximately 1 min onto Lacey carbon film grids (300 microMesh), blotted with Whatman filter paper according to the method described in ⁵⁶, and subsequently plunged in liquid ethane using a homemade plunging apparatus.

CryoEM data collection

Cryo-EM images of TORC1 filaments were collected on a Titan Krios microscope (Thermo Scientific) at EMBL Heidelberg, Germany, operated at 300 kV and equipped with a K2 summit direct electron detector (Gatan) camera operated in counting mode. A total of 4901 movies of 16 frames were collected with a dose rate of 2.5 e-/Å2/s and a total exposure time of 8 s, corresponding to a total dose of 20 e-/Å2/. Movies were collected at a magnification of 37,000x, corresponding to a calibrated pixel size of 1.35 Å/pixel at the specimen level.

CryoEM data processing

Motion correction and dose weighting of the recorded movies were performed using MotionCor2 ⁵⁷, discarding the first frame. Initial CTF estimation was performed on the aligned and dose-weighted summed frames using CTFFIND4 ⁵⁸. TORC1 filaments were picked manually using the e2helixboxer module in EMAN2 ⁵⁹ from a total of 3864 micrographs, resulting in 8625 picked filaments. A total of 219,455 particles were extracted in RELION 2.0 ⁶⁰ using the --helix option, with an extract size of 600 pixels (810 Å), an outer diameter of 620 Å, and a helical rise of 26.75 Å. Following particle extraction, per-particle CTF correction was performed using Gctf⁶¹.

Initially, helical reconstruction was performed in RELION2.0 using binned particles (2.7 Å/pixel), a previously acquired low-resolution map of TOROID ²⁷ low-pass filtered to 30 Å, and corresponding previously determined helical parameters as input, while imposing D1 symmetry and allowing for a helical symmetry search to optimize the respective helical parameters. The result of this helical reconstruction converged to a map with a resolution of around 9.1 Å based on the 0.143 gold-standard Fourier shell correlation (FSC) criterion ⁶². 3D classification and subsequent helical refinement in RELION2.0 did not result in an increase in resolution. Next, we imported non-binned particles and the reconstructed map resampled to a voxel size of 1.35 Å in cryoSPARC 3.01 ⁶³, and performed helical reconstruction imposing D1 symmetry, while allowing a helical symmetry search starting from the helical parameters obtained from the RELION helical reconstruction. The resulting 3D reconstruction had a resolution of 9.3 Å, similar to helical reconstruction in RELION. The broad peak of the mean squared error in a plot showing the estimated twist and rise (Figure S3A) indicates a relatively poor accuracy at determining the helical symmetry parameters on the un-symmetrized map of the last 3D refinement round. Indeed, similar scores are obtained for a twist ranging from approximately 46.9° to 47.7°, and rise from approximately 25.2 Å to 28.2 Å, which would translate into a pitch change from 201 to 205 Å and 192 to 215 Å, respectively (given an average twist for variable rise and vice versa).

We thus argued that the rather low resolution obtained after helical reconstruction in RELION and cryoSPARC could be the consequence of a varying pitch along the helices, resulting in an averaging of the structure while implying one set of helical parameters. To test this, we used the 9.1 Å helical reconstruction map (resampled to a voxel size of 1.35 Å) to create a mask containing the whole TOROID filament segment except for one central TORC1 assembly and its interacting regions with adjacent TORC1 complexes. We then performed signal subtraction in RELION2.0, using this mask and the structure and segment coordinates of the helical reconstruction, to generate a set of 219,455 subtracted particles containing one isolated TORC1 assembly per segment. Next, we employed the ‘localrec’ module in Scipion ^43,44 to crop and re-center the signal subtracted particles in a smaller box of 300 pixels (405 Å), as well as assigning to each particle a refined defocus based on the helical geometry. Single-particle analysis (SPA) of the subtracted particles using 3D refinement in RELION2.0, while imposing D1 symmetry, resulted in a map with a resolution of 4.5 Å (FSC = 0.143). We then imported the particles from the RELION2.0 refinement in cryoSPARC 3.01. After 2D classification and class selection, a set of 213808 selected particles was used for Non-Uniform refinement ⁴⁵ in cryoSPARC 3.01, employing a dynamic mask, imposing D1 symmetry, using the option to keep particles from the same helix in the same half-set, and allowing high-order aberration estimation and correction, which resulted in a final map with a resolution of 3.85 Å (FSC = 0.143). A final post-processed and locally sharpened map was obtained using DeepEMhancer ⁶⁴. Local resolution estimation of the final 3D reconstruction was calculated in cryoSPARC 3.01. A summary of cryo-EM data collection parameters and image processing procedures can be found in Supplementary Information (Supplementary Figure 3, Supplementary Table 1).

Model building and refinement

Initial homology models of Tor2 and Lst8 were generated using Phyre2 ⁶⁵, while a model of Kog1 was generated using ITasser ⁶⁶. Homology models of Tor2, Lst8 and Kog1 were first manually placed in the final 3D reconstruction followed by rigid-body fitting in Chimera ⁶⁷. The rigid-body fitted models were subsequently subjected to a round of flexible fitting using Imodfit ⁶⁸ followed by automatic molecular dynamics flexible fitting using NAMDINATOR ⁶⁹. The flexibly fitted structure was then refined using the Phenix software package (Adams et al., 2010) employing global minimization, local grid search, ADP refinement and secondary structure, Ramachandran and non-crystallographic symmetry (NCS) restraints. Initial refinement was followed by several cycles of extensive manual building in Coot ⁷⁰ followed by additional rounds of refinement in Phenix using a nonbonded weight parameter of 200. A final refinement round was performed in Phenix using global minimization, local grid search, ADP refinement, secondary structure, Ramachandran and non-crystallographic symmetry (NCS) restraints, and a nonbonded weight parameter of 300.

Alphafold ^71,72 structural predictions for Kog1, Lst8, and Tor2 were released during preparation of this manuscript. Individual subunits within the built structural model of the TOROID assembly as well as the corresponding EM map density were thoroughly compared with the Kog1, Lst8 and Tor2 Alphafold predictions. The Alphafold structural prediction for Lst8 is virtually identical to Lst8 extracted from our TOROID structural model (RMSD: 0.882 Å over 259 aligned main-chain atoms). The Alphafold Tor2 model is equally predicted well, but displays a closed conformation (similar to active mTOR, Figure 4B) starkly contrasting the open conformation of Tor2 extracted from our TOROID structural model, thus resulting in a RMSD value of 8.484 Å (1894 aligned main-chain atoms). While the Kog1 Alphafold model shows a high overall agreement with Kog1 extracted from the TOROID structure (RMSD: 0.829 Å over 1016 aligned main-chain atoms), the ‘Twix’ region displays a tilt which markedly differs from the TOROID structure, and the ‘Tack’ region, although correctly predicted as a helix surrounded by large loop regions, is predicted to be in a completely incorrect location. Accordingly, the model confidence score of this particular region in the Kog1 Alphafold structural prediction is very low. Nonetheless, we were able to use the Kog1 Alphafold model to improve the TOROID Kog1 structure in regions were the EM density was very weak, notably the Kog1 N-terminus and the C-terminal WD domain, resulting in significantly improved protein geometry validation statistics. A summary of refinement and validation statistics can be found in Supplementary Table 1.

Data Availability

Cryo-EM TOROID maps obtained using either helical reconstruction or signal subtraction and ensuing single particle analysis have been submitted to the EM Data Bank (EMDB) with accession codes EMD 13595 and 13594 respectively. The atomic model built in the signal subtracted TOROID cryo-EM map is submitted to the Protein Data Bank (PDB) with accession code 7PQH.

There is NO Competing Interest.

SupplementaryTable1.docx
Supplementary Table 1: Cryo-EM data collection, refinement, and validation statistics.
SupplementaryTable2.docx
Supplementary Table 2: List of oligonucleotides used for plasmid design and CrispR/Cas9 mutageneses.
SupplementaryTable3.docx
Supplementary Table 3
floatimage8.jpeg
Figure S1: Inducible in vivo proteolysis of EGOC - technical set up. a) The HA tagged TEV protease coding sequence has been cloned under the control of the CTH2 promoter and terminator. The terminator contains AU Rich Elements (AREs) recognized by the Cth2 protein that, when bound, destabilizes the transcript triggering its fast turnover. The promoter is additionally activated by Aft1 and Aft2 transcription factors which are repressed by the presence of Iron. Bathophenantrolinedissulfonic acid is a specific Iron chelator used to quickly induce the TEV protease expression. The TEV site has been inserted into the N-terminal non-conserved sequence of Ego1 downstream of the lipidated residues. This expression system, unlike other, commonly used inducible promoters (such as the GAL1 promoter), was chosen as it does not per se effect TORC1 signalling. b) Western Blot analyses of TEV expression (a-HA tag) and TORC1 activity monitored by anti-Sch9 Ser758 phosphorylation (a-P-Sch9) and anti-total Sch9 (a-tot-Sch9) antibodies. WT cells transformed by empty or TEV plasmid (pTEV) were subjected to BPS and rapamycin treatments as indicated. Cells were harvested at the indicated time points. Averages of the TEV expression normalized to the maximum are indicated with the standard deviations for N=3 replicates. The induction half-time is 23 minutes (EC50 Pearson’s coefficient = 0.989). c) Serial 10-fold dilutions of strains of the indicated genotype were spotted onto plates containing vehicle or 2.5nM rapamycin and 100µM Mohr salt or 25 µM BPS. d) Bar graph representing the quantity of Ego3 present around the vacuole membrane. EGOC depletion from the vacuolar membrane was determined by measuring the normalized Vacuolar Enrichment (see graphic): vacuole associated Ego3-GFP signal minus cytosolic background signal from Figure 1b. Vacuolar depletion half-time is 25.5 minutes after TEV protease induction in a Ego1^TEV strain (EC50 Pearson’s coefficient: 0.969). The measures are based on N=3 replicates and error bars represent standard deviation.
floatimage9.jpeg
Figure S2: EGOC puncta formation requires TOROIDs and is regulated by Gtr-GAPs. a) Cells expressing GFP-Kog1 or GFP-Tor1, and Ego3-3xmCherry were imaged in exponential phase, after 30 minutes of acute glucose depletion or in PDS by confocal microscopy. b) Confocal images of PDS GFP-Kog1- Ego3-3xmCherry-expressing cells were used to assess EGOC puncta and TOROID colocalization. Global 3D Pearson coefficient is indicated in black and reflects the global colocalisation. Manders coefficients using mCherry (1) or GFP (2) channels are indicated in gray and reflects the overlaps of GFP in mCherry and mCherry in GFP signals, respectively. Imaris quantifications on N=3 replicates. Error bars correspond to standard deviation and indicates P-value lower than 0.01. c) Workflow of the image analysis used to calculate the extent of TOROID colocalization with EGOC puncta and vice versa, A and B respectively. Cells were segmented using an in house designed FIJI macro and Imaris 3D analyses were performed to segment puncta in each channel according to their size (inferior or superior to 500 nm). The bar graph shows the subsequent Pearson’s coefficients from each channel (N=3, n>50). Error bars correspond to standard deviation; * indicates P-value lower than 0.1 indicates P-value lower than 0.01. d) WT, Δgtr1 Δgtr2, ∆lst4, ∆lst7, ∆sea1 and ∆sea3 strains expressing GFP-Tor1 or Ego3-GFP were imaged in exponential phase or in PDS by confocal microscopy. e) Bar graph representing the quantification of TOROID from d) with N=3 and n>100. Error bars correspond to standard deviation. f) Bar graph representing the quantification of EGOC puncta from d with N=3 and n>100. Error bars correspond to standard deviation.
floatimage10.jpeg
Figure S3: TOROID structure determination by helical reconstruction and signal subtraction of Cryo-EM data. a) Processing workflow for TOROID structure determination by helical reconstruction and signal subtraction followed by single particle analysis (SPA). Software packages used during the different steps are indicated. The plot on the right shows the variation in estimated helical twist (in degrees) and rise (in Å) of individual TOROID segments during helical reconstruction. A summary of data collection/processing parameters, and refinement/validation statistics can be found in Supplementary Table 1. b) and c) Local resolution colouring of 3D reconstructions of TOROID by signal subtraction and SPA (b) and helical reconstruction (c). d) Gold-standard FSC curves for the TOROID map obtained after signal subtraction and SPA (left) or helical reconstruction (right). FSC curves are shown after applying either no mask (blue), a soft spherical mask (orange), a loose mask (green) or a tight mask (red) to both half maps prior to calculating the FSC. The corrected FSC curve (purple) is calculated by using the tight mask with correction by noise substitution. The estimated resolution at FSC = 0.143 is shown for the corrected FSC curves (purple). For the TOROID map obtained after signal subtraction and SPA, the model vs map FCS curve is shown in dark gray, as well as the resolution at FSC = 0.5.
floatimage11.jpeg
Figure S4: Evolutionary conservation of TORC1 components. a) Lst8 secondary structures from the TORC1 molecular model (Arrows: Beta-strands) aligned with the similarity sequence diagram based on the Lst8 orthologues aligned with ClustalX2. b) Side and top cartoon views of a structural alignment of Lst8 in TOROID (orange) with Lst8 from inactive mTORC1 (red, PDB ID: 6BCX) and Lst8 from active TORC1 bound to Rheb (green, PDB ID: 6BCU). Subunits are coloured according to the scheme shown below the panel. c) Zoom of the interaction interface between Kog1 (light pink) and the kinase domain (KD) of Tor2 (green). The Twix and Tack regions of Kog1, and the FRB domain and active site of Tor2 are annotated. d) Kog1 secondary structures (Arrows: Beta-strands, Rectangles: alpha helices, dash lines: unstructured loops) and domains (RNC, ARM WD40) from the TORC1 molecular model aligned with the similarity sequence diagram based on the Kog1/Raptor orthologues aligned with ClustalX2. Zoom on the a26 to a29 helix alignments show that Kog1 R884 is substituted by acidic residues in other organisms. e) Fit of the ‘Tack’ α-helix of Kog1 in the corresponding carved-out region of the TOROID Cryo-EM map. The helix is shown as a cartoon with labelled side chains shown as sticks. f) Tor2 domains (HORN, FAT and Kinase Domain; KD) aligned to the similarity sequence diagram based on the paralogues and orthologues aligned with ClustalX2. A closer look at the alignment within the Kinase domain reveals that the W2207 of the first alpha helix of the Lst8 binding Element of Tor2 (LBE1) is substituted by basic residues in other organisms.
floatimage12.jpeg
Figure S5: Additional mutations to test intra- and inter-coil TORC1-TORC1 interactions. a) TORC1 molecular model extracted from the TOROID Cryo-EM map. The different TORC1 subunits are coloured as indicated. Adjacent protomers are shown in lighter shades. Locations of amino acid substitutions chosen to probe intra-coil interactions are indicated with black boxes. b) Same TORC1 molecular model as a), rotated 180 degrees. Locations of amino acid substitutions chosen to probe inter-coil interactions are indicated with black boxes. c) Left panel: Distribution of TORC1 focus sizes (monitored by the number of GFP–Tor1 molecules) in the WT strain and in lst8^Q29A, tor2^{K2050F, K2053F}, tor1^W1279A, kog1^R1383A mutants presumed to compromise intra-coil interactions. GFP-Tor1 strains were grown PDS and imaged by confocal microscopy. The same control data are used in all graphs. Right insets: Zooms showing the different intra-coil interaction interfaces. d) Same as c), but for tor1^326-334AGn tor2^337-345AGn, tor1^368-370AGn tor2^379-381AGn, tor1^1449/54/56Aand kog1^∆1544-end mutants presumed to compromise inter-coil interactions. e) Statistical analyses of the TORC1 focus size distributions described in c) and d). GFP-Tor1 WT and mutated strains were imaged in exponential phase (E) or in PDS (P). Median and standard deviations are calculated over N replicates and n counted cells. T-test analyses were performed between WT and each mutated strain. n.s. indicates not significant, indicates P-value lower than 0.01, * indicates P-value lower than 0.001. f) Serial 10-fold dilutions of cells of the indicated genotypes were spotted onto plates containing the indicated concentrations of Rapamycin. The relative growth was measured and plotted for each strain; the corresponding EC₅₀ were extracted and quantified at the right of the panel. The EC₅₀ values vary from 1 (dark blue) to 9 nM (dark red).
floatimage13.jpeg
Figure S6. TOROID size is anti-correlated with rapamycin sensitivity. a) Gel filtration profile and corresponding SDS–polyacrylamide gel electrophoresis of the five subunits of EGOC^WT (Ego1, Ego2, Ego3, Gtr1^WT and Gtr2^WT) as visualized with Coomassie Blue staining. Fractions 3-5 contain the fully assembled complex and were pooled together. The coomassie blue-stained polyacrylamide gel (bottom right) shows the different forms of EGOC: WT (Gtr1^WT/Gtr2^WT), “active” (Gtr1^GTP/Gtr2^GDP) and “inactive” (Gtr1^GDP/Gtr2^GTP). b) WT, Δgtr1 Δgtr2, tor1^W2203R tor2^W2207R and Δgtr1 Δgtr2 tor1^W2203R tor2^W2207R cells expressing GFP-Tor1 were imaged in exponential phase or in PDS by confocal microscopy. Numbers indicate the quantification with standard deviation of TOROID formation based on biological replicates (N=3) and a minimum of 500 cells counted per replicate. c) Dot plot representing TOROID average size as a function of Rapamycin sensitivity of all the strains in S5f. The slope represents the linear regression with the Pearson coefficient (R).
FigureS7.pdf
Figure S7: CrispR mutagenesis screen reveals mutations in the Armadillo domain of Kog1 involved in the TORC1-EGOC interaction. Workflow of the CrispR mutagenesis screen. a) The screen is based on generation of mutagenized linear DNA fragments through repetitive rounds of error-prone PCR, introduced into GFP-TOR1 yeast cells with CrispR/Cas9. ~3500 primary transformants were replicated from SC-URA plates to SC-URA containing 2.5 or 4nM rapamycin. 99% of the clones grew in the presence of 2.5 nM Rapamycin, whereas only 39% grew in the presence of 4 nM rapamycin. b) Cells from 42 clones presenting hypersensitivity to 2.5 nM rapamycin were imaged in exponential phase by confocal microscopy. Clones presenting hypersensitivity to Rapamycin, constitutive TOROIDs and harboring punctual mutation(s) are labelled (from 1 to 12). c) Table listing mutations identified in the labelled clones from b). d) Kog1 structural model (Upper panel). The region colored in grey (aa 717-895) was mutagenized by error-prone PCR. Lower panel: Zoom showing the locations of the altered amino acids that confer rapamycin-sensitivity and constitutive TOROIDs. e) Zoom of the Kog1 secondary structure and alignment from S4d. Two additional features are added: a helix (a1004-1022) upstream the Tack and the putative claw (a1121-1132) proposed by Rogala et al., 2019. f) Bar graph representing the percentage of KOG1^WT, kog1^Da1004-1022, kog1^Dtack and kog1^Da1121-1132 cells displaying a TORC1 focus, marked with GFP-TOR1, in exponential phase or PDS. The analysis was performed on biological replicates (N=3) with a minimum of 500 cells counted per replicate. Error bars correspond to standard deviations, and *** represents P-values lower than 0.001.

PDF

Version 1

posted 02 Feb, 2022

You are reading this latest preprint version

EGOC inhibits TOROID polymerization by structurally activating TORC1

Status:

Version 1

Abstract

Figures

Introduction

Results

TOROIDs partially co-localize with EGOC puncta

High resolution Cryo-EM TOROID Structure

Mutations confirm observed TORC1/TORC1’ interfaces

Discussion

Declarations

References

Methods

Competing Interests

Supplementary Files

Status:

Version 1