Cell culture and treatments

All cell lines were grown at 37 °C, 5% CO2. Human female embryonic kidney HEK293FT cells (#R70007, Invitrogen; RRID: CVCL_6911) and the resulting genetically modified cell lines were cultured in high-glucose DMEM (#41965039, Thermo Fisher Scientific), containing 10% foetal bovine serum (FBS) and 1% penicillin–streptomycin. The parental HEK293FT cells were purchased from Invitrogen before the initiation of the project. Their identity was validated by the Multiplex human Cell Line Authentication test (Multiplexion GmbH), which uses a single-nucleotide polymorphism typing approach, and was performed as described at www.multiplexion.de. All cell lines were regularly tested for Mycoplasma contamination using a PCR-based approach and were confirmed to be Mycoplasma free.

AA starvation experiments were performed as described previously13. In brief, custom-made starvation media were formulated according to the Gibco recipe for high-glucose DMEM, specifically omitting the AAs. The media were filtered through a 0.22-μm filter device and tested for proper pH and osmolality before use. For the respective AA-replete (+AA) treatment media, commercially available high-glucose DMEM was used (#41965039, Thermo Fisher Scientific). All treatment media were supplemented with 10% dialysed FBS. For this purpose, FBS was dialysed against 1× PBS through 3,500 MWCO dialysis tubing. For basal (+AA) conditions, the culture media were replaced with +AA treatment media 60–90 min before lysis or fixation. For AA starvation, culture media were replaced with starvation media for 1 h. For AA add-back experiments, cells were first starved as described above and then starvation media were replaced with +AA treatment media for 30 min.

Antibodies

Antibodies against phospho-TFEB (Ser211) (#37681), TFEB (#4240), TFE3 (#14779), phospho-S6K (Thr389) (#9205), S6K (#9202), 4E-BP1 (#9452), phospho-4E-BP1 (Thr37/46) (#9459), phospho-4E-BP1 (Ser65) (9451), ULK1 (#8054), phospho-ULK1 (Ser757) (#14202), DYKDDDDK (FLAG) tag (#2368), mTOR (#2983), RagA (#4357), RagB (#8150), RagC (#9480), RagD (#4470), FLCN (#3697) and CTSD (#2284) proteins were purchased from Cell Signaling Technology. Anti-Raptor (#20984-1-AP) and anti-LARS (#21146-1-AP) antibodies were purchased from Proteintech. A monoclonal antibody recognizing human and mouse α-tubulin (#T9026) was purchased from Sigma, and the anti-HA (3F10; #11867423001) antibody was purchased from Roche. The anti-LAMP2 monoclonal antibody (DSHB Hybridoma Product H4B4) was purchased from Developmental Studies Hybridoma Bank (DSHB) and was deposited to the DSHB by August, J.T./Hildreth, J.E.K. For immunoprecipitation (IP) experiments, FLAG-tagged proteins were pulled down using anti-FLAG M2 affinity gel (#A2220, Sigma). For immunoblotting, all primary antibodies were used 1:1,000 in PBS-T, 5% BSA, except for anti-FLAG, for which 1:3,000 was used. Peroxidase-conjugated AffiniPure anti-rabbit, anti-mouse and anti-rat secondary antibodies (#711-035-152, #715-035-151 and #712-035-153, respectively; all from Jackson ImmunoResearch) were used 1:10,000 in PBS-T (1× PBS and 0.1% Tween-20), 5% powdered milk. For immunofluorescence (IF), all primary antibodies were used 1:200 in BBT solution (1× PBS, 0.1% Tween-20 and 0.1% BSA). Anti-mouse rhodamine (TRITC)-conjugated (#715-025-150, Jackson ImmunoResearch) and anti-rabbit fluorescein (FITC)-conjugated AffiniPure secondary antibodies (#711-095-152, Jackson ImmunoResearch) were used 1:100 in BBT, whereas anti-rabbit Alexa Fluor 488-conjugated (#711-545-152, Jackson ImmunoResearch) and anti-rat Alexa Fluor 647-conjugated AffiniPure secondary antibodies (#712-605-153, Jackson ImmunoResearch) were used 1:500 in BBT.

Plasmid constructs

Expression plasmids for FLAG- and HA-tagged RagA and RagC, as well as for FLAG–Luc, were described previously2. The respective expression vectors for RagB and RagD were PCR amplified from pRK5–HA–GST plasmids (described in ref. 2) and cloned in pcDNA3-HA and pcDNA3-FLAG vectors as EcoRI/NotI inserts. For the pcDNA3-FLAG-p18 expression construct, p18 was PCR-amplified from cDNA using appropriate primers, and cloned into the EcoRI/NotI sites of pcDNA3-FLAG. The pcDNA3–HA–RagC T90N and W115R point mutants were generated by site-directed mutagenesis using appropriate primers. The RagA/B chimaeric constructs were described previously2. The RagDCD chimaera was generated by first constructing a RagDC plasmid (containing the RagD N-terminal tail) using a two-step overlap PCR and appropriate primers to amplify parts of RagD and RagC. The end product was cloned into the pcDNA3-HA vector as NdeI/NotI fragment. Then a GeneArt string (Thermo Fisher Scientific) was used to introduce the C-terminal RagD part as a HpaI/NotI fragment in the RagDC plasmid, generating the RagDCD expression vector.

For the cell lines stably expressing HA-tagged Rag GTPase dimers (WT, mutants and chimaeras), the respective pcDNA3-puro vectors were generated by replacing the EcoRI/ClaI pcDNA3 fragment, containing the neomycin cassette, with the StuI/BstBI fragment of the MCSV-puro plasmid (Addgene #68469, RRID:Addgene_68469; described in ref. 61), containing the PGK-puro cassette. The backbone and insert fragment ends were blunt before ligation.

All restriction enzymes were purchased from Fermentas/Thermo Scientific. The integrity of all constructs was verified by sequencing. Sequences of all cloning primers are provided in Supplementary Table 1.

Generation of KO cell lines

HEK293FT knock-out (KO) cell lines were generated using the CRISPR/Cas9 system developed by the Zhang lab62. Double-stranded DNA oligos that encode single guide RNAs (sgRNAs) against target genes were designed using online tools. For RagB, RagC and RagD, two sgRNAs were designed per gene targeting the 5′ coding sequence or untranslated region and the 3′ coding sequence or untranslated region, respectively (Supplementary Fig. 1). RagA was efficiently knocked out using a single sgRNA. Each sgRNA was cloned into the BbsI restriction sites of the PX459 vector. The oligo sequences for all sgRNAs are provided in Supplementary Table 1.

In brief, cells were seeded in six-well plates and transfected on the following day with the respective sgRNA-expressing vectors using Effectene reagent (QIAGEN), according to the manufacturer’s instructions. Forty-eight hours post-transfection, cells were selected with 3 μg ml−1 puromycin (#A1113803, Thermo Fisher Scientific) for 3 days. Single-cell clones were picked by single-cell dilution, and KO clones were validated by genomic DNA PCR/sequencing (Extended Data Fig. 1) and immunoblotting using specific antibodies.

Transient DNA transfection

Plasmid DNA transfections were performed using Effectene (QIAGEN), according to the manufacturer’s instructions.

Stable cell line generation

For the generation of monoclonal stable lines expressing HA-tagged Rag GTPases (WT, mutants and chimaeras), HEK293FT qKO cells were transfected using the indicated Rag dimer expression vectors. Forty-eight hours post-transfection, cells were selected with 2 μg ml−1 puromycin for 2 days and then propagated in maintenance selection media containing the same puromycin concentration. To specifically assess the qualitative differences between the various Rag GTPase paralogues, single-cell clones that express comparable Rag levels were selected for functional characterization experiments. For the RagA versus RagB comparison, RagB expression is lower than RagA, resembling the endogenous RagA/B expression differences. Rag expression levels were validated by immunoblotting.

Despite the complications that generating monoclonal stable cell lines may introduce to a study (for example, due to clonal propagation and interclonal variability), and the fact that this is a tedious and lengthy process, this proved to be the best and only way that allows for a direct comparison between different Rag dimers and the functional characterization of their qualitative properties in the regulation of mTORC1 by AAs: while transiently overexpressing Rags could show the differences in interactions between RagC/D and other proteins in co-IP experiments, it largely masked the qualitative effects towards mTORC1 activity. This was probably due to overexpression artefacts, as Rag levels were massively higher in transiently transfected cells, compared with stable cell lines. Moreover, cells expressing such high Rag levels showed non-physiological localization patterns, with the majority of cells showing non-lysosomal Rag localization, regardless of the dimer expressed. Although polyclonal stable cell lines performed much better in maintaining the physiological regulation of mTORC1 by the Rags, they were still not appropriate for this study: because individual cells in the polyclonal population express uneven/variable Rag levels, some cells demonstrated almost undetectable Rag expression, while others had massive Rag overexpression. This led to large cell-to-cell variability, especially in microscopy studies, where we assessed mTOR or Rag localization at the single-cell level. In sum, monoclonal cell lines that express comparable Rag levels for the different Rag dimers and show low cell-to-cell variability were the only way to reliably investigate the Rag dimer code that defines the mTORC1 response to AAs.

Immunoblotting

For immunoblotting analyses, cells were washed once in-well with serum-free DMEM, to remove FBS, and lysed in 250 µl Triton lysis buffer (50 mM Tris pH 7.5, 1% Triton X-100, 150 mM NaCl, 50 mM NaF, 2 mM Na-vanadate, 0.011 g ml−1 β-glycerophosphate, 1× PhosSTOP phosphatase inhibitors and 1× Complete protease inhibitors) for 10 min on ice. Samples were clarified by centrifugation (14,000g, 15 min, 4 °C), and supernatants were transferred to new tubes. Protein concentration was measured using the Protein Assay Dye Reagent (#5000006, Bio-Rad).

Protein samples were subjected to electrophoretic separation on SDS–PAGE and analysed by standard western blotting techniques. In brief, proteins were transferred to nitrocellulose membranes (#10600002, Amersham) and stained with 0.2% Ponceau solution (Serva) to confirm equal loading. Membranes were blocked with 5% powdered milk in PBS-T (1× PBS and 0.1% Tween-20) for 1 h at room temperature, washed three times for 10 min with PBS-T and incubated with primary antibodies (1:1,000 in PBS-T, 5% BSA) rotating overnight at 4 °C. The next day, membranes were washed three times for 10 min with PBS-T and incubated with appropriate HRP-conjugated secondary antibodies (1:10,000 in PBS-T, 5% milk) for 1 h at room temperature. Signals were detected by enhanced chemiluminescence, using the ECL Western Blotting Substrate (#W1015, Promega), or SuperSignal West Pico PLUS (#34577, Thermo Scientific) and SuperSignal West Femto Substrate (#34095, Thermo Scientific) for weaker signals. Immunoblot images were captured on film (#28906835, GE Healthcare) and quantified using the GelAnalyzer software (v19.1; www.gelanalyzer.com).

Co-IP

For co-IP experiments, 1 × 106 cells were transiently transfected with the indicated plasmids and lysed 40–48 h post-transfection in IP lysis buffer (50 mM Tris pH 7.5, 0.3% CHAPS, 150 mM NaCl, 50 mM NaF, 2 mM Na-vanadate, 0.011 g ml−1 β-glycerophosphate, 1× PhosSTOP phosphatase inhibitors and 1× Complete protease inhibitors). FLAG-tagged proteins were incubated with 30 μl pre-washed anti-FLAG M2 affinity gel (Sigma, #A2220) for 3 h at 4 °C and washed four times with IP wash buffer (50 mM Tris pH 7.5, 0.3% CHAPS, 150 mM NaCl and 50 mM NaF). Samples were then boiled for 6 min in 2× Laemmli sample buffer and analysed by immunoblotting using appropriate antibodies.

LysoRag IP

To purify Rag-bound lysosomes, we developed the LysoRag IP method, a modified version of the Lyso-IP method that was previously described by the Sabatini lab44. This method allows for the purification of intact lysosomes, using HA-tagged Rags as bait. As a result, Rag dimers that bind to lysosomes more strongly pull down larger amounts of lysosomal material. HEK293FT qKO monoclonal cell lines, stably expressing HA-tagged RagC or RagD as dimers with RagA, were used to compare the relative affinities of RagC and RagD to lysosomes. In brief, 2 × 107 cells were seeded in a 15 cm dish and allowed to settle for 24 h. On the next day, cells were washed once with ice-cold PBS and scraped in 1 ml of ice-cold PBS containing 1× PhosSTOP phosphatase inhibitors (#04906837001, Roche) and 1× Complete protease inhibitors (#11697498001, Roche). Cells were then pelleted by centrifugation (1,000g, 2 min, 4 °C) and resuspended in 1 ml of 1× ice-cold PBS with inhibitors. For input samples, 25 µl of the suspension was transferred in a new tube and lysed by the addition of 125 µl CHAPS lysis buffer (50 mM Tris pH 7.5, 0.3% CHAPS, 150 mM NaCl, 50 mM NaF, 2 mM Na-vanadate, 0.011 g ml−1 β-glycerophosphate, 1× PhosSTOP phosphatase inhibitors and 1× Complete protease inhibitors) on ice for 10 min. Lysed input samples were then cleared by centrifugation (14,000g, 15 min, 4 °C), and the supernatant was transferred to new tubes containing 37.5 μl of 6× Laemmli and boiled for 6 min.

For the lysosomal fractions, the remaining cell suspension was homogenized with 20 strokes in pre-chilled 2 ml hand Dounce homogenizers kept on ice. The homogenate was cleared by centrifugation (1,000g, 2 min, 4 °C) and incubated with 100 µl pre-washed Pierce anti-HA magnetic beads (#88837, Thermo Fisher Scientific) on a nutating mixer for 3 min at room temperature, followed by three washes with ice-cold PBS, containing phosphatase and protease inhibitors, on a DynaMag spin magnet (#12320D, Invitrogen). After the last wash, lysosomes were eluted from the beads by addition of 60 μl 2× Laemmli sample buffer and boiling for 6 min.

IF and confocal microscopy

ΙF/confocal microscopy experiments and quantification of co-localization were performed as previously described13. In brief, cells were seeded on fibronectin-coated coverslips and treated as indicated in each experiment. After treatments, cells were fixed for 10 min at room temperature with 4% PFA in PBS. Samples were washed/permeabilized with PBT solution (1× PBS and 0.1% Tween-20), and blocked with BBT solution (1× PBS, 0.1% Tween-20 and 0.1% BSA). Staining was performed with the indicated primary antibodies in BBT (1:200 dilution) and then with appropriate highly cross-adsorbed secondary fluorescent antibodies (1:100 in BBT for FITC- or TRITC-conjugated antibodies; 1:500 in BBT for Alexa Fluor-conjugated antibodies). Finally, nuclei were stained with DAPI and cells mounted on slides using Fluoromount-G (#00-4958-02, Invitrogen). Images from single-channel captures are shown in greyscale. For the merged images, FITC, Alexa 647 (for anti-HA IFs), and Alexa 488 (for anti-TFE3 IFs) are shown in green, TRITC in red and DAPI in blue. Images were captured using a 40× objective lens on an SP8 Leica confocal microscope.

To quantify co-localization of mTOR or HA signal with the lysosomal marker LAMP2, the Fiji software (version 2.1.0/1.53c)63 was used to define regions of interest corresponding to individual cells, excluding the nucleus. Forty to 50 individual cells from approximately ten independent fields were selected per experiment for the analysis. The Coloc2 plugin was used to calculate the Manders’ co-localization coefficient, using automatic Costes thresholding64,65. The Manders’ co-localization coefficient yields the fraction of the signal of interest (mTOR or HA-Rag in this study) that overlaps with a second signal (in our case, lysosomes).

Subcellular localization of TFE3 was analysed by scoring cells on the basis of the signal distribution of TFE3, as shown in the example images in Fig. 2b. Signal was scored as nuclear (more TFE3 signal in the nucleus), cytoplasmic (more TFE3 signal in the cytoplasm) or intermediate (similar TFE3 signal between nucleus and cytoplasm). Approximately 50 individual cells were scored per genotype for each experiment.

Gene expression analysis (RT–qPCR)

For gene expression analysis, RNA was isolated with TRIzol (#15596018, Thermo Fisher Scientific) and reverse transcription was performed using the RevertAid H Minus Reverse Transcriptase kit (#EP0451, Thermo Fisher Scientific). The cDNAs were diluted 1:10 in nuclease-free H2O and 4 µl of diluted cDNA was used per reaction, along with 5 µl of 2× Maxima SYBR Green/ROX qPCR master mix (#K0223, Thermo Fisher Scientific) and 1 µl of primer mix (2.5 µM of forward and reverse primers). For each replicate experiment, reactions were set in technical triplicates in a StepOnePlus Real-Time PCR system (Applied Biosystems) and analysed with the StepOne software (v2.2.2; Applied Biosystems). Relative gene expression levels were calculated with the 2−ΔΔCt method. RPL13a expression was used for normalization as internal control.

LysoTracker staining

For LysoTracker staining experiments, cells were seeded in fibronectin-coated coverslips and grown until they reached 80–90% confluency. Lysosomes were stained by the addition of 100 nM LysoTracker Red DND-99 (#L7528, Invitrogen) in complete medium for 1.5 h in standard culturing conditions. Cells were then fixed with 4% PFA in PBS for 10 min at room temperature, washed and permeabilized with PBT solution (1× PBS and 0.1% Tween-20), and nuclei stained with DAPI (1:2,000 in PBT) for 10 min. Coverslips were mounted on slides using Fluoromount-G (#00-4958-02, Invitrogen). Images were captured using a 40× objective lens on an SP8 Leica confocal microscope, using the Leica Application Suite X software (v3.5.7.23225). LysoTracker signal intensity was measured from 50 individual cells per genotype for each experiment using Fiji (version 2.1.0/1.53c)63.

Phylogenetic analysis

Rag orthologues were identified by performing a blastp search (blastp suite) against the NCBI Reference proteins (refseq_protein; version 2021-07) database, using the AA sequences of human RRAGA (Uniprot ID: Q7L523), RRAGB (Uniprot ID: Q5VZM2-2), RRAGC (Uniprot ID: Q9HB90) and RRAGD (Uniprot ID: Q9NQL2-1) as query proteins, filtering for each of the organisms shown in Fig. 1a (H. sapiens, taxid: 9606; M. mulatta, taxid: 9544; C. lupus, taxid: 9615; M. musculus, taxid: 10090; X. laevis, taxid: 8355; D. rerio, taxid: 7955; D. melanogaster, taxid: 7227; C. elegans, taxid: 6239; S. cerevisiae, taxid: 4932). The expect threshold for identified proteins was set at 1 × 10−30; with a maximum of 100 target sequences; disabled low-complexity region filtering; using the BLOSUM62 matrix; a word size of 6; and gap existence and extension costs of 11 and 1, respectively.

Sequence alignment and structure modelling

Structure-based sequence alignments of RagA and RagB or RagC and RagD were prepared with Clustal Omega66 and ESPript67. To generate a model of the RagA/RagC/LAMTOR complex, we superposed the crystal structure of the active RagA–Q66L–GTP/RagC–S75N–GDP heterodimer (PDBID: 6S6D)45 with the complex structure of LAMTOR with the dimerization domains of RagA and RagC (PDBID: 6EHR)46. To model the active RagB or RagD GTPases, we introduced AA substitution in the RagA or RagC GTPases (PDBID: 6S6D), respectively, in Coot68, followed by structure idealization using refmac569. The inactive RagB/RagD dimer conformation was modelled accordingly on the basis of the cryo-EM structure of the inactive RagA/RagC dimer bound to the FLCN–FNIP2 complex (PDBID: 6ULG)70.

Statistics and reproducibility

Statistical analysis and data presentation in graphs was performed using the GraphPad Prism software (v9.1.0). For all quantifications, data in the graphs are shown as mean ± standard error of the mean (s.e.m.). Normal distribution was tested using the Shapiro–Wilk or the Kolmogorov–Smirnov tests, and correction for multiple comparisons was performed using the Tukey test. Significance was calculated using unpaired two-tailed t-test (for pairwise comparisons, see Fig. 3b,d,f and Extended Data Figs. 3b–e,g and 4c–f) or one-way analyis of variance (for multiple comparisons, see Figs. 1c–f,h, 2b,c, 4d,e,g, 5c–f,h, 6b–e,g and 7c,e and Extended Data Figs. 2b, 6b and 7b–e,g) for normally distributed data, or the Kruskal–Wallis test for non-normally distributed data (Fig. 2e). P values are described in the figures and figure legends (*P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.001; NS, non-significant). Statistics source data are provided in the numerical source data table.

All findings were reproducible over multiple independent experiments, within a reasonable degree of variability between replicates. The number of replicate experiments for each assay is provided in the respective figure legends. No statistical method was used to pre-determine sample size, which was determined in accordance with standard practices in the field. No data were excluded from the analyses. The experiments were not randomized, and the investigators were not blinded to allocation during experiments and outcome assessment.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.