A Rag GTPase dimer code defines the regulation of mTORC1 by amino acids

Abstract

Amino acid availability controls mTORC1 activity via a heterodimeric Rag GTPase complex that functions as a scaffold at the lysosomal surface, bringing together mTORC1 with its activators and effectors. Mammalian cells express four Rag proteins (RagA-D) that form dimers composed of RagA/B bound to RagC/D. Traditionally, the Rag paralogue pairs (RagA/B and RagC/D) are referred to as functionally redundant, with the four dimer combinations used interchangeably in most studies. Here, by using genetically modified cell lines that express single Rag heterodimers, we uncover a Rag dimer code that determines how amino acids regulate mTORC1. First, RagC/D differentially define the substrate specificity downstream of mTORC1, with RagD promoting phosphorylation of its lysosomal substrates TFEB/TFE3, while both Rags are involved in the phosphorylation of non-lysosomal substrates such as S6K. Mechanistically, RagD recruits mTORC1 more potently to lysosomes through increased affinity to the anchoring LAMTOR complex. Furthermore, RagA/B specify the signalling response to amino acid removal, with RagB-expressing cells maintaining lysosomal and active mTORC1 even upon starvation. Overall, our findings reveal key qualitative differences between Rag paralogues in the regulation of mTORC1, and underscore Rag gene duplication and diversification as a potentially impactful event in mammalian evolution.

Fig. 1. The RagC and RagD paralogues differentially regulate mTORC1.

a, Schematic representation of the presence (black squares) or absence of the RagA–D genes in the genome of the indicated species. b, Immunoblots with lysates from HEK293FT WT, or qKO cells stably expressing HA-tagged RagA/C or RagA/D, treated with medium containing (+) or lacking (–) AA, in basal (+), starvation (–) or add-back (–/+) conditions, probed with the indicated antibodies. Arrowheads indicate bands corresponding to different protein forms, when multiple bands are present. P, phosphorylated form; S, SUMOylated form. c–f, Quantification of TFEB (c and d), TFE3 (e) and S6K (f) phosphorylation from b. n = 5 independent experiments. g, Co-localization analysis of mTOR with LAMP2 (lysosomal marker) in HEK293FT WT, or qKO cells stably expressing RagA/C or RagA/D, using confocal microscopy. Magnified insets shown to the right. Scale bars, 10 μm. h, Quantification of mTOR/LAMP2 co-localization from n = 40 individual cells per condition from a representative experiment out of two independent replicates. Data in graphs shown as mean ± s.e.m. *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.001. Source numerical data and unprocessed blots are available in source data. Source data

Fig. 2. RagC and RagD differentially regulate TFE3 localization, target gene expression and lysosomal biogenesis.

a, TFE3 localization analysis in HEK293FT WT or qKO cells, stably expressing HA-tagged RagA/C, RagA/D or Luc as a negative control, using confocal microscopy. Nuclei stained with DAPI. Magnified insets shown to the right. Scale bars, 10 μm. n = 3 independent experiments. b, Scoring of TFE3 localization from a. Individual cells were scored for nuclear, intermediate or cytoplasmic TFE3 localization, as indicated in the example images. Scale bars, 10 μm. n = 3 independent experiments. c, Expression analysis of the TFE3 target genes GPNMB and UAP1L1 in HEK293FT WT, qKO cells or qKOs stably expressing HA-tagged RagA/C or RagA/D. n_GPNMB = 7 independent experiments; n_UAP1L1 = 6 independent experiments. d, LysoTracker staining in HEK293FT WT, qKO cells or qKOs stably expressing HA-tagged RagA/C or RagA/D. e, Quantification of LysoTracker signal intensity from n = 50 individual cells per condition from a representative experiment out of three independent replicates. Data in graphs shown as mean ± s.e.m. *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.001. Source numerical data are available in source data. Source data

Fig. 3. RagD shows higher affinity to p18 and associates with lysosomes more strongly than RagC.

a, Co-localization analysis of stably expressed HA-tagged RagA/C or RagA/D dimers with LAMP2 (lysosomal marker) in HEK293FT qKO cells, using confocal microscopy. Magnified insets shown to the right. Scale bars, 10 μm. b, Quantification of HA/LAMP2 co-localization from n = 40 individual cells per condition from a representative experiment out of three independent replicates. c, LysoRag IP experiments in HEK293FT qKO cells stably expressing HA-tagged RagA/C, RagA/D or Luc as a negative control. Intact lysosomes were immunopurified by HA–Rag IPs under native conditions and the presence of LAMP2, CTSD, mTOR and Raptor proteins in the lysosomal fractions was analysed by immunoblotting. d, Quantification of relative Rag–lysosome affinity. n = 3 independent experiments. e, p18/LAMTOR1 binds more strongly to RagD, compared with RagC. Co-IP experiments in HEK293FT qKO cells, transiently expressing FLAG-tagged p18 or Luc as negative control, and HA-tagged RagA with RagC or RagD. Binding of the Rags to p18 was analysed by immunoblotting. f, Quantification of relative Rag-p18 binding. n = 3 independent experiments. g, Working model for the differential regulation of mTORC1 by RagC- or RagD-containing dimers. RagD-containing dimers show stronger binding to p18/LAMTOR1, lysosomal localization, lysosomal recruitment of mTORC1, and phosphorylation of the TFE3/TFEB mTORC1 substrates. In contrast, RagC-containing dimers bind much less to p18, localize less to lysosomes and are less potent in recruiting mTORC1 to lysosomes to phosphorylate TFE3/TFEB. Both complexes are similarly capable of driving S6K phosphorylation. See main text for details. Created with BioRender.com. Data in graphs shown as mean ± s.e.m. **P < 0.01. Source numerical data and unprocessed blots are available in source data. Source data

Fig. 4. Differences in the N- and C-terminal RagD regions are responsible for its differential behaviour, compared with RagC.

a, Superposition of the structure of RagC (from PDBID: 6S6D; shown in cyan) with RagD (modelled; shown in yellow) shows high structural similarity between the two structures. Side chains of variable positions shown as dark grey (RagC) or light grey (RagD) sticks. b, Schematic representation of HA-tagged RagC, RagD and the RagDCD chimaera, in which the N- and C-terminal tails of RagC were replaced with those of RagD. The AA sequences around the fusion points are shown as insets. c, Immunoblots with lysates from HEK293FT qKO cells stably expressing HA-tagged RagA with RagC, RagD or the RagDCD chimaera, probed with the indicated antibodies. Arrowheads indicate bands corresponding to different protein forms, when multiple bands are present. P, phosphorylated form; S, SUMOylated form. d,e, Quantification of TFEB (d) and TFE3 (e) phosphorylation. n = 4 independent experiments. f, Co-IP experiments in HEK293FT qKO cells transiently expressing FLAG-tagged p18 or Luc as control, and HA-tagged RagA with RagC, RagD or the RagDCD chimaera. Binding of p18 to the Rags was analysed by immunoblotting. g, Quantification of relative Rag-p18 binding. n = 3 independent experiments. Data in graphs shown as mean ± s.e.m. **P < 0.01, ****P < 0.001. Source numerical data and unprocessed blots are available in source data. Source data

Fig. 5. Cancer-associated RagC mutations enhance TFE3/TFEB phosphorylation and mTOR lysosomal recruitment.

a, Schematic representation of HA-tagged WT RagC and the cancer-related T90N, W115R RagC point mutants. b, Immunoblots with lysates from HEK293FT qKO cells stably expressing HA-tagged RagA with WT RagC or RagD, or the T90N, W115R RagC mutants, probed with the indicated antibodies. Arrowheads indicate bands corresponding to different protein forms, when multiple bands are present. P, phosphorylated form; S, SUMOylated form. c–f, Quantification of TFEB (c and d), TFE3 (e) and S6K (f) phosphorylation from b. n = 3 independent experiments. g, Co-localization analysis of mTOR with LAMP2 (lysosomal marker) in HEK293FT qKO cells stably expressing HA-tagged RagA with the RagC proteins shown in a or RagD, using confocal microscopy. Magnified insets shown to the right. Scale bars, 10 μm. h, Quantification of mTOR/LAMP2 co-localization from n = 50 individual cells per condition from a representative experiment out of three independent replicates. Data in graphs shown as mean ± s.e.m. *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.001. Source numerical data and unprocessed blots are available in source data. Source data

Fig. 6. The RagA and RagB paralogues differentially control mTORC1 activity upon AA starvation.

a, Immunoblots with lysates from HEK293FT WT, or qKO cells stably expressing RagA/D or RagB/D, treated with medium containing (+) or lacking (–) AA, in basal (+), starvation (–) or add-back (–/+) conditions, probed with the indicated antibodies. Arrowheads indicate bands corresponding to different protein forms, when multiple bands are present. P, phosphorylated form; S, SUMOylated form. b–e, Quantification of TFEB (b and c), TFE3 (d) and S6K (e) phosphorylation from the blots shown in a. n = 4 independent experiments. f, Co-localization analysis of mTOR with LAMP2 (lysosomal marker) in HEK293FT WT or qKO cells stably expressing RagA/D or RagB/D, using confocal microscopy. Magnified insets shown to the right. Scale bars, 10 μm. g, Quantification of mTOR/LAMP2 co-localization from n = 40 individual cells per condition from a representative experiment out of two independent replicates. Data in graphs shown as mean ± s.e.m. *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.001. Source numerical data and unprocessed blots are available in source data. Source data

Fig. 7. The RagB-specific N-terminal tail is responsible for its differential effect towards mTORC1 upon AA starvation, compared with RagA.

a, Schematic representation of HA-tagged WT RagA, WT RagB and the RagBΔN, RagB^AQVHS chimaeras. b, Immunoblots with lysates from HEK293FT qKO cells stably expressing the proteins shown in a as dimers with HA-tagged RagD, probed with the indicated antibodies. Arrowheads indicate bands corresponding to different protein forms, when multiple bands are present. P, phosphorylated form. c, Quantification of TFEB phosphorylation. n = 3 independent experiments. d, Co-localization analysis of mTOR with LAMP2 (lysosomal marker) in HEK293FT qKO cells stably expressing HA-tagged RagBΔN or RagB^AQVHS as dimers with RagD, treated with medium containing (+) or lacking (–) AA, using confocal microscopy. Magnified insets shown to the right. Scale bars, 10 μm. e, Quantification of mTOR/LAMP2 co-localization from n = 50 individual cells per condition from a representative experiment out of three independent replicates. f, Working model for the differential regulation of mTORC1 by RagA- or RagB-containing dimers, in basal or AA starvation conditions. Whereas RagA-containing dimers allow for mTORC1 de-localization away from lysosomes and for its inactivation upon AA starvation, RagB-containing dimers retain lysosomal and active mTORC1 even in AA starvation conditions. See main text for details. Created with BioRender.com. Data in graphs shown as mean ± s.e.m. *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.001. Source numerical data and unprocessed blots are available in source data. Source data

Extended Data Fig. 1. Characterization of the Rag genomic alterations in qKO HEK293FT cells.

(a) CRISPR/Cas9-mediated knockout of RRAGA. The associated genomic changes in the two RRAGA alleles and the resulting changes in the RagA protein are shown. (b) CRISPR/Cas9-mediated knockout of RRAGB. The associated genomic changes in RRAGB are shown. (c) CRISPR/Cas9-mediated knockout of RRAGC. The associated genomic changes in RRAGC and the resulting changes in the RagC protein are shown. (d) CRISPR/Cas9-mediated knockout of RRAGD. The associated genomic changes in RRAGD are shown. Source data

Extended Data Fig. 2. Characterization of the HEK293FT qKO and Rag dimer reconstituted cell lines.

(a-b) Two independent quadruple RagA-D knockout (qKO) HEK293FT clones show no lysosomal accumulations of mTOR. Colocalization analysis of mTOR with LAMP2 (lysosomal marker) in WT and qKO cells (clones qKO1, qKO2), using confocal microscopy. Magnified insets shown to the right. Scale bars = 10 μm (a). Quantification of mTOR/LAMP2 colocalization from n = 50 individual cells per condition from a representative experiment out of 2 independent replicates is shown in (b). Data shown as mean ± SEM. **** p<0.001. (c) Genetic ablation of all four Rags blunts mTORC1 reactivation by amino acids. Immunoblots with lysates from HEK293FT WT and qKO cells (clones qKO1, qKO2), treated with media containing or lacking AA, in starvation (–) or add-back (–/+) conditions, probed with the indicated antibodies. Arrowheads indicate bands corresponding to different protein forms, when multiple bands are present. P: phosphorylated form; S: SUMOylated form. n = 2 independent experiments. (d) Reconstitution of qKO cells with different Rag dimers reveals qualitative differences in the regulation of mTORC1. Immunoblots with the indicated antibodies using lysates from HEK293FT WT and qKO cells stably expressing comparable amounts of the four different Rag dimer combinations, or Luciferase (Luc) as a negative control, as HA-tagged proteins. Arrowheads indicate bands corresponding to different protein forms, when multiple bands are present. P: phosphorylated form; S: SUMOylated form. n = 3 independent experiments. Source numerical data and unprocessed blots are available in Source Data. Source data

Extended Data Fig. 3. Analysis of independent RagA/C- and RagA/D-expressing clones.

(a-e) Immunoblots with lysates from qKO HEK293FT cells stably expressing HA-tagged RagA/C or RagA/D, grown under basal, AA-replete culture conditions, probed with the indicated antibodies. Arrowheads indicate bands corresponding to different protein forms, when multiple bands are present. P: phosphorylated form; S: SUMOylated form (a). Quantification of TFEB, TFE3 and S6K phosphorylation from (a), shown in (b-c), (d) and (e), respectively. Data in graphs shown as mean ± SEM. ** p<0.01, *** p<0.005. n = 3 independent experiments. (f-g) Colocalization analysis of mTOR with LAMP2 (lysosomal marker) in qKO HEK293FT cells stably expressing RagA/C or RagA/D, grown under basal, AA-replete culture conditions, using confocal microscopy. Magnified insets shown to the right. Scale bars = 10 μm (f). Quantification of mTOR/LAMP2 colocalization from n = 50 individual cells per condition from a representative experiment out of 3 independent replicates is shown in (g). Source numerical data and unprocessed blots are available in Source Data. Source data

Extended Data Fig. 4. Interaction properties of RagC- and RagD-containing dimers.

(a) RagC and RagD bind similarly to RagA. Co-immunoprecipitation experiments in HEK293FT qKO cells, transiently expressing HA-tagged RagA with FLAG-tagged RagC, RagD, or Luciferase (Luc) as negative control. Binding of RagA to RagC or RagD was analyzed by immunoblotting as indicated. n = 2 independent experiments. (b-f) Co-immunoprecipitation experiments in HEK293FT qKO cells transiently expressing FLAG-tagged RagA/C, RagA/D, or Luciferase (Luc) as a negative control. Binding of the Rags to the indicated proteins was analyzed by immunoblotting. P: phosphorylated protein form; S: SUMOylated protein form(s) (b). Quantification of Rag binding to mTOR, Raptor, TFEB, and TFE3, shown in (c), (d), (e), (f), respectively. n = 3 independent experiments. Data in graphs shown as mean ± SEM. * p<0.05. Source numerical data and unprocessed blots are available in Source Data. Source data

Extended Data Fig. 5. Structural comparison between RagC and RagD suggests main differences localize to the unstructured N- and C-terminal tails.

(a) Structure-based sequence alignment of RagC and RagD prepared with ESPript. Similar and identical residues are marked by yellow and red boxes, respectively. Secondary structure assignment is based on PDBID: 6S6D. (b) Minimal surface residue differences between RagC and RagD. Surface representation of the RagA (blue) / RagC (cyan) heterodimer (PDBID: 6S6D). Variable positions between RagC and RagD are coloured yellow. (c) No residue differences between the RagC and RagD structures localize at the Rag dimer / LAMTOR complex interface. Model of the RagA (blue) / RagC (cyan) heterodimer in the active conformation (PDBID: 6S6D) bound to the pentameric LAMTOR1-5 complex (PDBID: 6EHP). Variable positions between RagC and RagD are shown as yellow sticks. The ultimate residues that could be modelled at the N- and C-termini of the RagA, RagC and LAMTOR1 in the published structures are also labelled (N, C). Source data

Extended Data Fig. 6. Cancer-associated RagC mutations show increased lysosomal localization and p18 binding.

(a-b) Colocalization analysis between the Rags and LAMP2 (lysosomal marker) in HEK293FT qKO cells stably expressing HA-tagged RagA with WT RagC, RagD, or the T90N and W115R RagC mutants, using confocal microscopy. Magnified insets shown to the right. Scale bars = 10 μm (a). Quantification of HA/LAMP2 colocalization from n = 50 individual cells per condition from a representative experiment out of 2 independent replicates is shown in (b). Data shown as mean ± SEM. *** p<0.005, **** p<0.001. (c) The cancer-associated RagC mutants bind more strongly to p18/LAMTOR1, compared to wild-type RagC. Co-immunoprecipitation experiments in HEK293FT qKO cells, transiently expressing FLAG-tagged p18 or Luciferase (Luc) as negative control, and HA-tagged RagA with WT RagC, RagD, or the T90N and W115R RagC mutants. Binding of the Rags to p18 was analyzed by immunoblotting as indicated. n = 2 independent experiments. Source numerical data and unprocessed blots are available in Source Data. Source data

Extended Data Fig. 7. Analysis of independent RagA/D- and RagB/D-expressing clones.

(a-e) Immunoblots with lysates from HEK293FT WT or qKO cells stably expressing RagA/D or RagB/D, treated with media containing (+) or lacking (–) AA, probed with the indicated antibodies. Arrowheads indicate bands corresponding to different protein forms, when multiple bands are present. P: phosphorylated form; S: SUMOylated form (a). Quantification of TFEB, TFE3 and S6K phosphorylation from the blots in (a), shown in (b-c), (d), and (e), respectively. n = 3 independent experiments. (f-g) Colocalization analysis of mTOR with LAMP2 (lysosomal marker) in HEK293FT WT or qKO cells stably expressing RagA/D or RagB/D, using confocal microscopy. Magnified insets shown to the right. Scale bars = 10 μm (f). Quantification of mTOR/LAMP2 colocalization from n = 50 individual cells per condition from a representative experiment out of 3 independent replicates is shown in (g). Data in all graphs shown as mean ± SEM. * p<0.05, ** p<0.01, *** p<0.005, **** p<0.001. Source numerical data and unprocessed blots are available in Source Data. Source data

Extended Data Fig. 8. Structural comparison between RagA and RagB suggests main differences localize to the unstructured N-terminal tail.

(a) Structure-based sequence alignment of RagA and RagB prepared with ESPript. Similar and identical residues are marked by yellow and red boxes, respectively. Secondary structure assignment is based on PDBID: 6S6D. (b) Residue differences in the structures of RagA and RagB are not predicted to cause overall structural changes. Superposition of the structure of RagA (from PDBID: 6S6D; shown in blue) with RagB (modelled; shown in red) shows high structural similarity between the two structures. Side chains of variable positions shown as dark grey (RagA) or light grey (RagB) sticks. (c) Minimal surface residue differences between RagA and RagB. Surface representation of the model of RagB (red) / RagD (yellow) heterodimer in the active conformation (modelled based on PDBID: 6S6D). Variable positions between RagA and RagB are coloured blue. (d) Ribbon representation of the model of the RagB (red) / RagD (yellow) heterodimer in the active conformation (modelled based on PDBID: 6S6D). Variable positions between RagA and RagB are coloured blue and side chains are shown as sticks. (e) Same as in (d), but for the inactive RagB/D dimer (modelled based on PDBID: 6ULG).