KICSTOR recruits GATOR1 to the lysosome and is necessary for nutrients to regulate mTORC1

Abstract

The mechanistic target of rapamycin complex 1 (mTORC1) is a central regulator of cell growth that responds to diverse environmental signals and is deregulated in many human diseases, including cancer and epilepsy. Amino acids are a key input to this system, and act through the Rag GTPases to promote the translocation of mTORC1 to the lysosomal surface, its site of activation. Multiple protein complexes regulate the Rag GTPases in response to amino acids, including GATOR1, a GTPase activating protein for RAGA, and GATOR2, a positive regulator of unknown molecular function. Here we identify a protein complex (KICSTOR) that is composed of four proteins, KPTN, ITFG2, C12orf66 and SZT2, and that is required for amino acid or glucose deprivation to inhibit mTORC1 in cultured human cells. In mice that lack SZT2, mTORC1 signalling is increased in several tissues, including in neurons in the brain. KICSTOR localizes to lysosomes; binds and recruits GATOR1, but not GATOR2, to the lysosomal surface; and is necessary for the interaction of GATOR1 with its substrates, the Rag GTPases, and with GATOR2. Notably, several KICSTOR components are mutated in neurological diseases associated with mutations that lead to hyperactive mTORC1 signalling. Thus, KICSTOR is a lysosome-associated negative regulator of mTORC1 signalling, which, like GATOR1, is mutated in human disease.

Extended Data Figure 1. GATOR1 and GATOR2 associate with endogenous KICSTOR components in an amino acid insensitive fashion

a) An endogenously tagged GATOR1 component co-immunoprecipitates endogenous KICSTOR. Anti-FLAG immunoprecipitates were prepared from HEK-293T cells expressing endogenously FLAG-tagged DEPDC5, a GATOR1 component, that had been starved of amino acids for 50 min or starved and restimulated with amino acids for 10 min. Immunoprecipitates and cell lysates were analyzed by immunoblotting for the indicated proteins. b) An endogenously tagged GATOR2 component co-immunoprecipitates endogenous GATOR1 and KICSTOR. Anti-FLAG immunoprecipitates were prepared from HEK-293T cells expressing endogenously FLAG-tagged WDR59, a GATOR2 component, and treated as in (a). Immunoprecipitates and cell lysates were analyzed by immunoblotting for the indicated proteins. c) An anti-KPTN antibody co-immunoprecipitates endogenous components of KICSTOR, GATOR1, and GATOR2. Anti-KPTN immunoprecipitates were prepared from wild-type HEK-293T treated as in (a) and immunoprecipitates and cell lysates analyzed by immunoblotting for the indicated proteins. Anti-GSK3β immunoprecipitates were used to monitor the non-specific binding of proteins to the beads. d) KPTN and ITFG2 form a heterodimer that requires SZT2 to associate with C12orf66. Anti-FLAG immunoprecipitates and lysates prepared from HEK-293T cells expressing the indicated cDNAs were analyzed by immunoblotting for the relevant epitope tags. e) Loss of KICSTOR components did not have a significant effect on the expression levels of GATOR1 or GATOR2 components. HEK-293T cell clones deficient for indicated KICSTOR components or Nprl3 or WDR24 were generated via the CRISPR/Cas9 system and single cell cloning. Cell lysates were analyzed by immunoblotting for the indicated proteins. DNA sequencing of the C12orf66 gene was used to verify out of frame mutations in the genomic locus of the sgC12orf66 cells because an antibody that detects the C12orf66 protein in cell lysates is not available. The HEK-293T cell clones analyzed here were used in subsequent figures, where indicated. f) KPTN interacts with ITFG2 even in cells lacking other KICSTOR components. Immunoprecipitates and cell lysates prepared from wild-type, SZT2-deficient, or C12orf66-deficient HEK-293T cell clones expressing the indicated proteins were analyzed by immunoblotting. g) Expression levels of KPTN and ITFG2 in HEK-293T cells stably expressing the indicated sgRNAs under amino acid starved or replete conditions. Cell lysates were analyzed by immunoblotting for the levels of the indicated proteins. Raptor serves as a loading control. These same cell lines were also analyzed for phospho-S6K1 and S6K1 levels in Extended Data Fig. 5a–c. h) KPTN-ITFG2 does not compete with C12orf66 for associating with SZT2. HEK-293T cells expressing the indicated cDNAs were treated and analyzed as in (d).

Extended Data Figure 2. Size exclusion chromatography analysis of KICSTOR components

Lysates of wild-type or SZT2-deficient (sgSZT2) HEK-293T cells were fractionated with tandem Superose 6 size exclusion chromatography columns and the collected fractions analyzed by immunoblotting for the indicated proteins. Colored bars indicate fractions that contain the protein denoted by that color in the key. Fractions containing the molecular weight standards are indicated. Note that the C12orf66 antibody exhibits significant background when used to probe total cell lysates by immunoblotting so we are only confident that the bands in the high molecular weight fractions that disappear in the SZT2-deficient cells actually represent C12orf66.

Extended Data Figure 3. SZT2 interacts with KPTN-ITFG2, C12orf66, and GATOR1

a) SZT2 interacts with GATOR1 in the absence of other KICSTOR components. HEK-293T cells stably expressing a control sgRNA (sgAAVS1) or sgRNAs targeting the indicated KICSTOR components were transfected with the indicated cDNAs. Anti-FLAG immunoprecipitates were prepared and analyzed, along with cell lysates, by immunoblotting for the relevant epitope tags. The ratios of the intensities of the HA-SZT2 to FLAG-DEPDC5 bands are indicated below the FLAG-DEPDC5 blot. See Extended Data Fig. 1g for the expression levels of the KICSTOR components in the cell lines used here. b) SZT2 links the other KICSTOR components to the GATOR complexes. Anti-FLAG immunoprecipitates prepared from wild-type or SZT2-deficient HEK-293T cells expressing the indicated cDNAs were analyzed by immunoblotting for the indicated proteins. * marks non-specific bands. See Extended Data Fig. 1e for the expression level of SZT2 in the SZT2-deficient HEK-293T cells. c) C12orf66 interacts with SZT2 at a distinct site than KPTN-ITFG2. HEK-293T cells expressing the indicated cDNAs were analyzed as in (b). d) GATOR1 interacts with the first and second regions of SZT2. HEK-293T cells expressing the indicated cDNAs were analyzed as in (b). e) The association of SZT2 with KPTN-ITFG2 persists in the absence of GATOR1 or (f) GATOR2. Anti-FLAG immunoprecipitates were prepared from wild-type, Nprl3-deficient, or WDR24-deficient HEK-293T cells expressing the indicated cDNAs and analyzed as in (b).

Extended Data Figure 4. Subcellular localization of the GATOR and KICSTOR complexes

a) Expression levels of GFP-tagged GATOR1, GATOR2, or KICSTOR components used in the localization experiments. Nprl2-deficient or wild-type HeLa cells stably expressing the indicating GFP-tagged proteins were single cell sorted for the low GFP population and single cell clones were analyzed by immunoblotting for levels of the indicated proteins. b) Quantitation of the imaging data in Figures 1g, h, i. Values are mean ± standard error. c) Amino acids do not control the localization of GATOR2 to the lysosomal surface. Wild-type HeLa cells stably expressing GFP-tagged WDR24, a component of GATOR2, were starved or starved and restimulated with amino acids for the indicated times prior to processing for immunofluorescence detection of GFP and LAMP2. Scale bars represent 10 μm. Quantitation of the imaging data is shown in the bar graph on the right. d) Amino acids do not regulate the amounts of GATOR1, GATOR2, or KICSTOR components on purified lysosomes. Lysosomes immunopurified with anti-HA beads from wild-type HEK-293T cells expressing HA-tagged LAMP1 and treated as in (c) were analyzed by immunoblotting for the levels of the indicated proteins.

Extended Data Figure 5. KICSTOR loss affects the sensitivity of the mTORC1 pathway to nutrients but not growth factors

a) CRISPR/Cas9-mediated depletion of KPTN, (b) ITFG2, or (c) C12orf66 renders mTORC1 signaling insensitive to amino acid deprivation. HEK-293T cells stably expressing the indicated sgRNAs were starved of amino acids for 50 min or starved and restimulated with amino acids for 10 min. Cell lysates were analyzed by immunoblotting for the levels and phosphorylation states of the indicated proteins. See Extended Data Fig. 1g for the expression levels of the KICSTOR components in the cell lines used here. d) Testing of indicated SZT2 sgRNA-treated HeLa cell clones for levels of SZT2 and raptor by immunoblotting. Note that not all clones have complete loss of the SZT2 protein. e) mTORC1 signaling in SZT2-deficient cells is insensitive to amino acid deprivation. Indicated HeLa cell SZT2-deficient clones from (d) were treated and analyzed as in (a). f) CRISPR/Cas9-mediated depletion of ITFG2 renders mTORC1 signaling partially insensitive to amino acid deprivation. HeLa cells stably expressing the indicated ITFG2 sgRNAs were treated and analyzed as in (a). g) Glucose levels do not affect GATOR1 localization, as monitored by GFP-Nprl2. HeLa cells expressing GFP-Nprl2 from Figure 1h were starved of glucose for 50 min or starved and restimulated with glucose for 10 min prior to processing for immunofluorescence for GFP and LAMP2 (left). Scale bars represent 10 μm. Quantitation of the imaging data is shown in bar graph on the right. h) In SZT2-deficient cells mTORC1 signaling is still sensitive to serum starvation and insulin stimulation. Indicated HeLa cell SZT2-deficient clones from (d) were starved of serum for 50 min or starved of serum and restimulated with insulin for 10 min. Cell lysates were analyzed by immunoblotting for the levels and phosphorylation states of the indicated proteins. i) In cells with CRISPR/Cas9-mediated depletion of ITFG2, mTORC1 signaling is still sensitive to serum starvation and insulin stimulation. HeLa cells stably expressing the indicated ITFG2 sgRNAs were treated and analyzed as in (h). j) Quantitation of phospho-S6K1 blots in (h) and (i). Values are mean ± standard deviation.

Extended Data Figure 6. Activation of mTORC1 signaling in tissues from Szt2^GT/GT mice

a) SZT2 inhibits mTORC1 signaling in mouse liver and muscle. Mice with the indicated genotypes were treated and analyzed by immunoblotting for the levels and phosphorylation state of S6 as in Figure 2d. The animals examined here represent an additional two animals of each genotype beyond those analyzed in Figures 2d, e. b) Quantitation of the ratio of phospho-S6 to S6 bands in (a) and in Figures 2d, e. Values are mean ± standard deviation for n=4. c) SZT2 inhibits mTORC1 signaling in hepatocytes and cardiomyocytes in vivo. Liver and heart sections prepared from mice treated as in Figure 2d were analyzed by immunohistochemistry for phospho-S235/236 S6 levels and serial sections were stained with hematoxylin and eosin (H&E). Liver image is centered over a central vein. Scale bar represents 40 μm.

Extended Data Figure 7. GATOR1, like KICSTOR, functions downstream of or parallel to GATOR2 in the mTORC1 pathway

a) GATOR1, like KICSTOR, is epistatic to GATOR2. Wild-type, WDR24-deficient, or Nprl3- and WDR24-deficient HEK-293Ts were starved of amino acids for 50 min or starved and restimulated with amino acids for 10 min. Cell lysates were analyzed by immunoblotting for the indicated proteins and phosphorylation states. b) Quantitation of the imaging data in Figure 3d. Values are mean ± standard error.

Extended Data Figure 8. KICSTOR regulates the lysosomal localization of GATOR1 but not of GATOR2 or the Rag GTPases

a) Quantitation of the imaging data in Figures 4a, b. Values are mean ± standard error. b) Loss of KICSTOR components does not affect the lysosomal localization of GATOR2. GFP-WDR24 expressing HeLa cells prepared as in Extended Data Fig. 4c were subsequently modified with the CRISPR/Cas9 system to create KPTN-deficient cells. These cells as well as wild-type and sgAAVS1-treated control cells were starved or starved and restimulated with amino acids for the noted times prior to processing for the detection of GFP and LAMP2 by immunofluorescence. Insets depict selected fields magnified 3.24× and their overlays. Scale bars represent 10 μm. Quantitation of the imaging data is shown in the bar graph below images (mean ± standard error). c) The Rag GTPases localize to lysosomes regardless of SZT2 expression. Wild-type and SZT2-deficient HEK-293T cells were treated as in (b) prior to processing for the detection of RagC and LAMP2 by immunofluorescence. Insets depict selected fields magnified 3.24× and their overlays. Scale bars represent 10 μm. Quantitation of the imaging data is shown in the bar graph on the right (mean ± standard error).

Extended Data Figure 9. Loss of KICSTOR disrupts the GATOR1-GATOR2 interaction

a) Loss of SZT2 disrupts the GATOR1-GATOR2 interaction. Anti-FLAG immunoprecipitates were prepared from wild-type or SZT2-deficient HeLa cells stably expressing the indicated cDNAs and starved of amino acids for 50 min or starved and restimulated with amino acids for 10 min. Immunoprecipitates and cell lysates were analyzed by immunoblotting for the indicated proteins. b) KPTN is necessary for the interaction of GATOR2 with GATOR1. HEK-293T cells stably expressing the indicated sgRNAs were transfected with the indicated cDNAs and subsequently treated and analyzed as in (a). c) ITFG2 is also necessary for the GATOR1-GATOR2 interaction. Cells were prepared, treated, and analyzed as in (b). d) C12orf66 is also necessary for the GATOR1-GATOR2 interaction. Cells were prepared, treated, and analyzed as in (b).

Figure 1. The KICSTOR complex interacts with GATOR1 and localizes to lysosomes

a) Mass spectrometric analyses identified KICSTOR-derived peptides in immunoprecipitates prepared from HEK-293T cells expressing endogenously FLAG-tagged DEPDC5. b) Presence or absence of KICSTOR component gene orthologs in model organisms. c) GATOR1 co-immunoprecipitates more KICSTOR than does GATOR2. Immunoprecipitates were prepared from HEK-293E cells stably expressing the indicated FLAG-tagged proteins and analyzed by immunoblotting for the indicated proteins. d) GATOR2 requires GATOR1 to associate with KICSTOR. Anti-FLAG immunoprecipitates were prepared from wild-type and Nprl3-deficient HEK-293T cells transiently expressing the indicated cDNAs. Anti-FLAG immunoprecipitates and lysates were analyzed as in (c). e) GATOR1 does not require intact GATOR2 to interact with KICSTOR. Wild-type and WDR24-deficient HEK-293T cells transiently expressing the indicated cDNAs were treated and analyzed as in (c). f) Model proposing that the KICSTOR complex interacts with GATOR1, which in turn interacts with GATOR2. KPTN, ITFG2, and C12orf66 are indicated by the first letter of each. g) KICSTOR localizes to lysosomes in an amino acid independent fashion. Wild-type HeLa cells stably expressing GFP-ITFG2 were starved or starved and restimulated with amino acids for the indicated times prior to processing for immunofluorescence detection of GFP and LAMP2, a lysosomal marker. In all images, insets depict selected fields magnified 3.24× and their overlays. Scale bar represents 10 μm. h) GATOR1 localizes to lysosomes regardless of amino acid levels. Nprl2-deficient HeLa cells were reconstituted with GFP-Nprl2 and treated and analyzed as in (g). i) GATOR2 localizes to lysosomes regardless of amino acid levels. Wild-type HeLa cells stably expressing Mios-GFP were treated and analyzed as in (g).

Figure 2. Regulation of mTORC1 signaling by nutrients requires KICSTOR

a) mTORC1 signaling is insensitive to amino acid starvation in cells lacking any KICSTOR component. HEK-293T cell clones deficient for each KICSTOR component were starved of amino acids for 50 min or starved and restimulated with amino acids for 10 min. Cell lysates were analyzed by immunoblotting for the levels and phosphorylation states of the indicated proteins. See Extended Data Fig. 1e for validation of the KICSTOR deficient cells. b) Activation of autophagy induced by amino acid starvation requires KICSTOR. The HEK-293T cell clones as in (a) were monitored for LC3B processing and accumulation after one hour of amino acid starvation in the absence or presence of chloroquine. c) mTORC1 signaling is insensitive to glucose starvation in cells lacking KICSTOR. Experiment was performed as in (a) except that cells were starved of and stimulated with glucose. d) SZT2 inhibits mTORC1 signaling in mouse liver. Mice with the indicated genotypes were fasted for 8 hours and liver lysates analyzed by immunoblotting for the levels and phosphorylation states of the indicated proteins. Two wild-type and two Szt2^GT/GT mice were analyzed in this experiment and in (e). e) SZT2 inhibits mTORC1 signaling in the mouse gastrocnemius muscle. Mice were treated and muscle lysates analyzed as in (d). f) SZT2 inhibits mTORC1 signaling in mouse neurons in vivo. Brain sections prepared from mice treated as in (d) were analyzed by immunohistochemistry for phospho-S235/236 S6 levels and stained with hematoxylin and eosin (H&E). Equivalent regions of the cortex and cerebellum are shown for the two genotypes. Scale bar represents 40 μm.

Figure 3. KICSTOR acts upstream of the Rag GTPases

a) KICSTOR functions upstream of the Rag GTPases. Wild-type or SZT2-deficient HEK-293Ts expressing the indicated cDNAs were starved of amino acids for 50 min or starved and restimulated with amino acids for 10 min. FLAG immunoprecipitates and cell lysates were analyzed by immunoblotting for the levels and phosphorylation states of the indicated proteins. b) Modest GATOR1 overexpression inhibits mTORC1 signaling to a lesser extent in SZT2-deficient than in wild-type cells. FLAG immunoprecipitates and cell lysates prepared from cells expressing the indicated cDNAs and in amino acid replete conditions were analyzed as in (a). In this experiment 0.5 and 2.0 ng of the cDNA for each GATOR1 component was transfected while 100 ng of each was used in (a). c) KICSTOR functions downstream of or in parallel to GATOR2. Wild-type, SZT2-deficient, or double SZT2- and WDR24-deficient HEK-293Ts were treated and analyzed as in (a). d) Amino acid insensitive localization of mTOR to lysosomes in cells lacking KICSTOR components. Wild-type and KPTN-, ITFG2-, or SZT2-deficient HEK-293T cells were starved or starved and restimulated with amino acids for the indicated times prior to processing for immunofluorescence detection of mTOR and LAMP2. In all images, insets depict selected fields magnified 3.24× and their overlays. Scale bars represent 10 μm.

Figure 4. The lysosomal localization of GATOR1 requires KICSTOR

a) SZT2 loss renders GATOR1 dispersed throughout the cytoplasm. Nprl2-deficient HeLa cells were reconstituted with GFP-Nprl2 and subsequently modified with the CRISPR/Cas9 system to create SZT2-deficient cells expressing GFP-Nprl2. These cells were starved or starved and restimulated with amino acids for the noted times prior to the detection of GFP and LAMP2 by immunofluorescence. Insets depict selected fields magnified 3.24× and their overlays. Scale bars represent 10 μm. b) Loss of C12orf66 or KPTN also disrupts the localization of GATOR1 to the lysosome. Cell lines were generated, treated, and analyzed as in (a). c) In SZT2-deficient cells, GATOR1 is not present on immunopurified lysosomes. Lysosomes were immunopurified from wild-type or SZT2-deficient HEK-293T cells expressing HA-tagged LAMP1 and starved of and stimulated with amino acids as in (a). Lysosomes and whole cell lysates were analyzed by immunoblotting for the levels of the indicated proteins. d) Loss of SZT2 disrupts the GATOR1-Rag GTPase interaction. Control or SZT2-deficient HEK-293T cells with or without the stable expression of HA-Nprl3 were starved of amino acids for 50 min or starved and restimulated with amino acids for 10 min. Lysates and anti-HA immunoprecipitates were prepared and analyzed by immunoblotting for the levels of the indicated proteins. e) Proposed role for KICSTOR in the nutrient sensing pathway upstream of mTORC1.