Galectins Control mTOR in Response to Endomembrane Damage

Abstract

The Ser/Thr protein kinase mTOR controls metabolic pathways, including the catabolic process of autophagy. Autophagy plays additional, catabolism-independent roles in homeostasis of cytoplasmic endomembranes and whole organelles. How signals from endomembrane damage are transmitted to mTOR to orchestrate autophagic responses is not known. Here we show that mTOR is inhibited by lysosomal damage. Lysosomal damage, recognized by galectins, leads to association of galectin-8 (Gal8) with the mTOR apparatus on the lysosome. Gal8 inhibits mTOR activity through its Ragulator-Rag signaling machinery, whereas galectin-9 activates AMPK in response to lysosomal injury. Both systems converge upon downstream effectors including autophagy and defense against Mycobacterium tuberculosis. Thus, a novel galectin-based signal-transduction system, termed here GALTOR, intersects with the known regulators of mTOR on the lysosome and controls them in response to lysosomal damage. VIDEO ABSTRACT.

Keywords: AMPK; APEX2; LC3; TAK1; TFEB; autophagy; catabolism; galectins; lysosome; mTOR.

Figure 1. Lysosomal damage inhibits mTOR signaling

(A) Dose-response mTOR activity (pS6K1 (T389) and pULK1 (S757)), to lysosomal damage by GPN (full medium, 1 h). (B) As in A after 1h washout in full medium. (C) As in A, dose-response to silica in full medium, 1 h. (D) mTOR activity in primary human macrophages treated with 100 μM GPN in full medium or starved in EBSS for 1 h (Ctrl, full medium only). (E) Quantification by automated high-content imaging and analysis (HC) of overlaps between mTOR and LAMP2 (images, Figure S1D); cells treated with 100 μM GPN, 2mM LLOMe, or 400 μg/mL Silica for 1 h in full medium. (F) Immunofluorescence confocal microscopy of mTOR localization relative to LAMP2-positive lysosomes. 100 μM GPN in full medium, 1 h; endogenous LAMP2 (green florescence, Alexa-488) and mTOR (red florescence, Alexa-568). Scale bar, 5 μm. (G) TFEB nuclear translocation; (treatments: 100 μM GPN, 2mM LLOMe, or 400 μg/mL Silica in full medium 1 h. HC (blue: nuclei, Hoechst 33342; red: anti-TFEB antibody, Alexa-568). White masks, computer algorithm-defined cell boundaries (primary objects); pink masks, computer-identified nuclear TFEB based on the average intensity of Alexa-568 fluorescence. (H) Endogenous LC3 puncta quantified by HC. Tretments as in G. Green masks, computer-identified LC3 puncta (target objects). (I) Endogenous ATG13 puncta quantification by HC; treatments as in G. Red masks, computer-identified ATG13 puncta (target objects). Data, means □ SEM; immunoblots: n ≥ 3; HC: n ≥ 3 (each experiment: 500 valid primary objects/cells per well, ≥ 5 wells/sample). *p < 0.05, **p < 0.01, ANOVA. See also Figure S1.

Figure 2. Ragulator-Rag complex responds to lysosomal damage in control of mTOR

(A) mTOR activity (immunoblot analysis of S6K1 (T389) phosphorylation) in TSC2-deleted (TSC2^-/-) and wild type (TSC2^WT) cells treated with 100 μM GPN in full medium (Full) or starved in EBSS for 1 h. Ctrl (control): untreated cells. (B) Co-immunoprecipitation analysis of changes in interactions between Ragulator and Rag GTPases following treatment with GPN. HEK293T cells stably expressing FLAG-metap2 (control) or FLAG-p14 were treated with 100 μM GPN in full medium for 1 h. Cell lysates were immunoprecipitated (IP) with anti-FLAG antibody and immunoblotted for endogenous RagA or RagC. (C) Immunoprecipitation (IP) analysis of interactions between RagB and mTOR/Raptor in cells treated with GPN. HEK293T cells overexpressing FLAG-metap2 (control) or FLAG-RagB were treated with 100 μM GPN in full medium for 1 h. Cell lysates were IP-ed with anti-FLAG antibody and immunoblotted for endogenous mTOR or Raptor. (D) mTOR activity in HEK293T cells or HEK293T cells stably expressing constitutively active RagB GTPase (RagB^Q99L) treated and analyzed as in A. (E) Immunofluorescence confocal microscopy visualization of mTOR localization relative to LAMP2-positive lysosomes. Cells as in D were treated as in A, and immunostained for endogenous LAMP2 (green florescence, Alexa-488) and mTOR (red florescence, Alexa-568). Scale bar, 1 μm. (F) Quantification by HC of overlaps between mTOR and LAMP2 (images, Figure S2C) in cells as in D, treated as in A. Data, means □ SEM; immunoblots: n ≥ 3; HC: n ≥ 3 (each experiment: 500 valid primary objects/cells per well, ≥ 5 wells/sample). † p ≥ 0.05 (not significant), *p < 0.05, **p < 0.01, ANOVA. See also Figure S2.

Figure 3. Gal8 is in dynamic complexes with mTOR and its regulators and adaptors

(A) Galectin puncta formation in response to GPN. Cells expressing YFP-galectin fusions were treated with 100 μM GPN or without (Ctrl) in full medium for 1 h and galectin puncta quantified by HC. Left, images of galectins 1, 3, 8, and 9. White masks, algorithm-defined cell boundaries (primary objects); green masks, computer-identified galectin puncta (target objects). (B) Co-immunoprecipitation (Co-IP) analysis of galectins and mTOR or RagA. Cells expressing FLAG-tagged galectins were subjected to anti-FLAG immunoprecipitation followed by immunoblotting for endogenous mTOR or RagA. (C) Co-IP analysis of endogenous proteins in macrophage-like cells treated with 100 μM GPN in full medium1 h. IP: anti-Gal8; immunoblotting: endogenous RagA, p14, mTOR and Raptor. (D) APEX2 proximity biotinylation analysis. Cells were transfected with APEX2 fusions with Gal3, 8 and 9, incubated or not with biotin-phenol, pulsed with H₂O₂, and biotinylated proteins affinity-isolated on streptavidin-beads analyzed by immunoblotting. (E) Proximity biotinylation as in D in response to GPN. BP, biotin-phenol. (F)(i-ii) GST pulldown assay of in vitro translated and radiolabeled Myc-tagged p18 with GST, or GST-tagged Gal8 and Gal9. Data (% binding). (G)(i-ii) GST pulldown assay of in vitro translated Myc-tagged Gal8 or Gal9 with GST or GST-tagged RagB/D. Data as in F. (H) Cells transfected with GFP-Gal8 and FLAG-tagged metap2 (negative control) or RagB variants (RagB^WT, RagB^T54L or RagB^Q99L) were subjected to anti-GFP IP, followed by immunoblotting for FLAG-tagged proteins or GFP. (I) Cells transfected with GFP-Gal8 and FLAG-tagged metap2 or RagC variants (RagC^WT, RagC^S75L or RagC^Q120L) processed as in H; immunoblotting: FLAG or GFP. Data, means □ SEM; blots: n ≥ 3, HC: n ≥ 3 (each experiment: 500 valid primary objects/cells per well, ≥ 5 wells/sample). † p ≥ 0.05 (not significant), *p < 0.05, **p < 0.01, ANOVA. See also Figure S3.

Figure 4. Gal8 is required for mTOR inactivation in response to lysosomal damage

(A) mTOR activity, monitored by phosphorylation of S6K1 (p-T389) and ULK1 (S757), in parental (Gal8WT^HeLa) and Gal8-knockout (Gal8KO^HeLa) HeLa cells treated with 100 μM GPN in full medium for 1 h. (B) HC analysis of overlaps between mTOR and LAMP2 in Gal8WT^HeLa and Gal8KO^HeLa cells treated as in A. Ctrl (control): untreated cells. Red and green masks, computer-identified mTOR and LAMP2, respectively (target objects). (C) Immunoblotting analysis of LC3 lipidation (LC3-II) in Gal8WT ^eLa and Gal8KO^HeLa treated as in A. (D) mTOR activity (monitored as in A) in bone marrow-derived macrophages (BMMs). BMMs of wild type C57BL (Gal8WT^BMM) and their littermate Gal8-knockout mice (Gal8KO^BMM) were treated with 400 μM GPN in full medium for 1 h. (E) Analysis of autophagy induction (monitored as in C) in Gal8WT^BMM and Gal8KO^BMM treated with 400 μM GPN in full medium for 1 h. (F) HC analysis of TFEB nuclear translocation in Gal8WT^BMM and Gal8KO^BMM treated with 400 μM GPN in full medium for 1 h. Nuclei: blue pseudocolor, Hoechst 33342; TFEB: red fluorescence, Alexa-568. Ctrl (control): untreated cells. White masks, algorithm-defined cell boundaries (primary objects); pink masks, computer-identified nuclear TFEB based on average intensity. Data, means □ SEM; blots: n ≥ 3, HC: n ≥ 3 (each experiment: 500 valid primary objects/cells per well, ≥ 5 wells/sample). † p ≥ 0.05 (not significant), *p < 0.05, **p < 0.01, ANOVA. See also Figure S4.

Figure 5. Lysosomal damage promotes interactions between Gal8 and the amino acid and cholesterol sensor SLC38A9

(A) Cells expressing FLAG-SLC38A9 were treated with 100 μM GPN in full medium or starved in EBSS for 1 h. Cell lysates were subjected to anti-FLAG immunoprecipitation and immunoblotted for endogenous Gal8. Control (Ctrl), untreated cells. Note: SLC38A9 is heavily glycosylated protein with smear pattern in immunoblots. (B) Co-IP analysis of interactions between SLC38A9 and Gal8. Cells expressing GFP-tagged Gal8 or glycan recognition-mutant forms of Gal8 (individual R69H, R232H or double/combined R69H & R232H; see panel C) were treated with 100 μM GPN for 1 h in full medium. IP, anti-SLC38A9 (SLC) antibody; immunoblotting with anti-GFP. *, non-specific bands. Note: input SLC3A89 was deglycosylated with PNGase F. (C) Schematic diagram of Gal8 domains (CRD and CRD2, carbohydrate recognition domains 1 and 2) and summary of interactions between SLC38A9 and Gal8. +++, strong; +, weak; -, not detectable. (D) mTOR activity in wild type (WT) and Gal8-knockout (Gal8^KO) HeLa cells. Gal8^KO cells transfected wtih FLAG-tagged Gal8 or double glycan (R69H & R232H) recognition-mutant form of FLAG-Gal8^R69H&R232H were treated with 100 μM GPN for 1 h in full medium. mTOR activity was monitored by S6K1 (T389) phosphorylation. (E) Co-IP analysis of Gal8 with RagB or p18 in response to GPN treatment. Wild type (WT) or SLC38A9-knockout (SLC38A9^KO) cells transfected with FLAG vector or FLAG-Gal8 were treated with 100 μM GPN in full medium for 1 h. IP, anti-FLAG; immunoblot, antibodies against RagB and p18. (F) HC quantification of overlaps between galectins and LAMP2 in response to GPN treatment. Wild type (WT) and SLC38A9 knockout (SLC38A9^KO) HEK293T cells expressing GFP-galectin fusions were treated with 100 μM GPN in full medium for 1 h. White masks, automatically defined cell boundaries (primary objects); red and green masks, computer-identified LAMP2 and GFP-galectin profiles (target objects). HC data, means □ SEM; n ≥ 3 (each experiment: 500 valid primary objects/cells per well, ≥ 5 wells/sample). † p ≥ 0.05 (not significant), *p < 0.05, ANOVA. See also Figure S5.

Figure 6. SLC38A9 is required for mTOR reactivation during recovery from lysosomal damage

(A) mTOR activity (immunoblots; S6K1 phosphorylation at T389) and autophagy induction (LC3 immunoblots) in WT (wild type) and SLC38A9 knockout (SLC38A9^KO) HEK293T cells treated with 100 μM GPN in full medium (time course). (B) HC analysis of autophagy induction (endogenous LC3 puncta) in WT and SLC38A9^KO cells treated with 100 μM GPN in full medium for 30 min.. Control (Ctrl), untreated cells. White masks, algorithm-defined cell boundaries (primary objects); green masks, computer-identified LC3 puncta (target objects). (C) mTOR activity recovery and autophagy inhibition in SLC38A9^KO cells after GPN washout. WT and SLC38A9^KO HEK293T cells were treated with 100 μM GPN for 1 h followed by 1 h washout in full medium. mTOR activity and autophagy were monitored as in A. (D) mTOR activity and autophagy induction (monitored as in A) cells overexpressing FLAG-SLC38A9 or FLAG (vector control) treated with 100 μM GPN in full medium (time course). (E) HC analysis of autophagy induction in SLC38A9-overexpressing cells treated with GPN. FLAG and FLAG-SLC38A9 (F-SLC) expressing HEK293T cells were treated with 100 μM GPN in full medium for 30 min, and LC3 puncta were quantified by HC. Ctrl, no GPN. Masks, as in B. Data (B and E), means □ SEM, n ≥ 3 independent experiments (500 primary objects counted per well; ≥ 5 wells/sample per each experiment), **p < 0.01, ANOVA. (F) Schematic, strategy for APEX2-Gal8 LC-MS/MS proteomic analysis (see STAR methods). (G) Cytoscape depiction of dynamic changes in protein interactions/proximity relative to Gal8 in response to lysosomal damage caused by exposure to GPN (GPN+), based on protemic data in Table S1. Red lines, key changes in interactions/proximities (using > 100-fold change in precursor peak intensities as a cutoff) observed in each of the three complete biological replicates of HEK293T cells transfected with pJJiaDEST-APEX2-Gal8 subjected to separate LC-MS/MS analyses. Dotted black lines, examples of proteins identified in the LC/MS/MS analysis that did not display changes when comparing control (full medium without GPN; GPN-) vs. lysosomal damage (full medium with 100 μM GPN for 1 h; GPN+) (see Table S1). The MS/MS proteomic data have been deposited at MassIVE, ID MSV000081788 and linked to ProteomeXchange accession ID PXD008390). See also Figure S5 and Table S1.

Figure 7. Galectin 9 interacts with AMPK and activates it during lysosomal damage

(A) Co-IP analysis of interactions between galectins and AMPKα. Cells transfected with FLAG-tagged galectins were subjected to anti-FLAG IP followed by immunoblotting for endogenous AMPKα. (B) AMPK activity in parental (Gal9WT^293A) and Gal9-knockout (Gal9KO^293A) HEK293A cells treated with 100 μM GPN in full medium for 1 h. AMPK activation was monitored by immunoblotting analysis of phosphorylated AMPKα (p-T172) and its targets acetyl-CoA carboxylase (ACC, p-S79) and ULK1 (p-S317; vs. p-S757 (phosphorylated by mTOR)) relative to total AMPKα, ACC and ULK1. (C) mTOR activity in HeLa cells transfected with scrambled siRNA (Scr) or AMPKα siRNA (AMPKα^KD) treated with 100 μM GPN in full medium for 1 h. mTOR activity was monitored by immunoblotting for p-S6K1 (T389) and p-ULK1 (S757). (D) Co-IP analysis of interactions using Gal9 antibody or control IgG for IP and TAK1, LKB1 or CaMKK2 in immunoblots. (E) APEX2 biotinylation proximity analysis of AMPKα and its upstream regulators. Cells transfected with APEX2-Vector or APEX2-Gal9 were subjected to in vivo biotinylation (BP, biotin phenol), biotinylation proteins enriched on streptavidin beads, and the samples immunoblioted for TAK1, LKB1 and CaMKK2. (F) HC analysis of autophagy induction (LC3 puncta) in parental (Gal9WT^293A) and Gal9-knockout (Gal9KO^293A) HEK293A cells treated with 100 μM GPN in full medium for 1 h. White masks, algorithm-defined cell boundaries (primary objects); green masks, computer-identified LC3 puncta (target objects). (G) Analysis of autophagy induction in Gal8WT^BMM and Gal8KO^{B M} primary macrophages treated with 400 μM GPN in full medium for 1 h. LC3 puncta were quantified by HC as in F. HC data, means □ SEM, n ≥ 3 independent experiments (500 primary objects counted per well; ≥ 5 wells/sample per each experiment), **p < 0.01, ANOVA. (H) Survival curves of Gal8 wild type mice and their Gal8-knockout littermates in a model of respiratory infection with M. tuberculosis. Initial lung deposition, 700 CFU of M. tuberculosis Erdman. See also Figures S5 and S6.