Cell Biology

Noncanonical roles of ATG5 and membrane atg8ylation in retromer assembly and function

Masroor Ahmad Paddar
Fulong Wang
Einar S Trosdal
Emily Hendrix
Yi He
Michelle Salemi
Michal Mudd
Jingyue Jia
Thabata L A Duque
Ruheena Javed
Brett Phinney
Vojo Deretic author has email address

Autophagy, Inflammation and Metabolism Center of Biochemical Research Excellence
Department of Molecular Genetics and Microbiology, University of New Mexico School of Medicine, 915 Camino de Salud, NE, Albuquerque, NM 87131, USA
Department of Chemistry & Chemical Biology, The University of New Mexico, Albuquerque, NM, USA
Proteomics Core Facility, UC Davis Genome Center, University of California, Davis, CA 95616, USA

https://doi.org/10.7554/eLife.100928.1

Open access
Copyright information

Figures and data

ATG5 interacts with retromer.
A. ATG5 functions. X, a postulated additional function. B. 2D scatter plot (log2 fold changes; color coded p value cutoff matrix for comparisons between samples as per the lookup table) of proximity biotinylation LC/MS/MS datasets: FlpIn-HeLa^{APEX2-ATG5-WT} cells (X-axis) and FlpIn-HeLa^{APEX2-ATG5-K130R} (Y-axis) treated with 2 mM LLOMe for 30 min ratioed vs. HeLa^{APEX2-ATG5-WT} without LLOMe treatment (Ctrl). C-E. Co-IP analyses and quantification of VPS26A (C), VPS29 (D) VPS35, (E), interaction with ATG5 in HeLa cells treated with or without 2 mM LLOMe for 30 min. Data, means ± SE (n=3); unpaired t-test; p values indicated above the bars.

Retromer affects lysosomal sensitivity to damage.
A,B. High content microscopy (HCM) imaging and quantification of Gal3 response (puncta/cell of endogenous Gal3 profiles stained for immunofluorescence) in HeLa^WT, HeLa^ATG5-KO, and HeLa^VPS35-KO cells subjected to lysosomal damage by Leu-Leu-O-Me ester hydrobromide (LLOMe; 2mM, 30 min). HCM, an unbiased machine-driven image acquisition and data analysis based on presets of >500 valid primary objects/cells per well (representative images shown; white mask, cell; red masks, Gal3 puncta), with a minimum of 5 wells per sample (sampling error), and n≥3, independent biological replicates (experimental error) in separate 96-well plates. Data, means ± SE (n=3), two-way ANOVA with Tukey’s multiple comparisons. C,D. Complementation analysis of VPS35^KO LyHYP (lysosome hypersensitivity) phenotype monitored by HCM quantification of Gal3 puncta in HeLa^VPS35-KO cells transfected with GFP (control) or GFP-VPS35 expressing plasmids. Data, means ± SE (n=3); one-way ANOVA with Tukey’s multiple comparisons. E,F. Comparative HCM analysis of ubiquitin (immunofluorescence; FK2 antibody) response to lysosomal damage (LLOMe; 2mM, 30 min) in HeLa^ATG5-KO and HeLa^VPS35-KO cells. Yellow profiles, colocalization of ubiquitin and LAMP1. Top panels, ubiquitin immunostaining alone. Data, means ± SE (n=4), two-way ANOVA with Tukey’s multiple comparisons. G,H. HCM quantification of ALIX localization to endolysosomal compartments (% of LAMP1 profiles positive for ALIX immunostaining) in Huh7^WT, Huh7^ATG5-KO, and Huh7^VPS35-KO cells following lysosomal damage. Yellow profiles, colocalization of ALIX and LAMP1. Data, means ± SE (n=3); two-way ANOVA with Tukey’s multiple comparisons. HCM images in all relevant panels, examples from a bank of unbiased operator-independent machine-collected and algorithm-processed fields containing a minimum of 500 primary objects/cells per well (5 wells minimum per 96-well plate; 3 plates minimum), per cell line/condition.

ATG5 affects sorting of the retromer cargo GLUT1.
A. Co-IP analysis of endogenous ATG5 with GFP-VPS35 or YFP-VPS29 (transient transfection). B. Reverse Co-IP analysis GFP-VPS35 or YFP-VPS29 (transient transfection) and endogenous ATG5. C. Confocal images illustrating localization of GLUT1 and LAMP2 in Huh7^WT, Huh7^VPS35-KO, and Huh7^ATG5-KO cells. Scale bar, 10 μm. D,E. HCM quantification of GLUT1 (endogenous protein immunostaining) puncta/cell (D) and GLUT1 colocalization with LAMP2 (% of LAMP2 profiles positive for GLUT1; overlap area) (E) in Huh7^WT, Huh7^ATG5-KO, and Huh7^VPS35-KO cells. Scale bar, 20 μm. Data, means ± SE (n =6); one-way ANOVA with Tukey’s multiple comparisons. F,G. HCM quantification of GLUT1-SNX27 overlap (% of SNX7 area positive for GLUT1) in Huh7^WT, Huh7^ATG5-KO, and Huh7^VPS35-KO cells. Scale bar, 20 μm. Data, means ± SE (n=5); one-way ANOVA with Tukey’s multiple comparisons. H,I. Immunoblot analysis and quantification of proteins in lysosomes purified/enriched by LysoIP (immunoisolation with TMEM192-3xHA) from Huh7^WT, Huh7^ATG5-KO, and Huh7^VPS35-KO cells. TMEM192-2xFLAG, negative control. Data, means ± SE (n=3), one-way ANOVA with Tukey’s multiple comparisons. HCM images in panel F, examples from a bank of unbiased operator-independent machine-collected and algorithm-processed fields containing a minimum of 500 primary objects/cells per well (5 wells minimum per 96-well plate; 3 plates minimum), per cell line.

Membrane atg8ylation apparatus affects sorting of the retromer cargo GLUT1.
A,B. HCM quantification of GLUT1 (endogenous protein immunostaining) puncta/cell in Huh7^WT, Huh7^ATG3-KO, Huh7^ATG5-KO, Huh7^ATG7-KO, Huh7^ATG16-KO, and Huh7^VPS35-KO cells. Scale bar, 20 μm. Data, means ± SE (n=5); one-way ANOVA with Tukey’s multiple comparisons. C,D. HCM quantification of GLUT1 (endogenous protein immunostaining) colocalization with LAMP2 (% of LAMP2 profiles positive for GLUT1; overlap area) in Huh7^WT, Huh7^ATG3-KO, Huh7^ATG5-KO, Huh7^ATG7-KO, Huh7^ATG16-KO, and Huh7^VPS35-KO cells. Scale bar, 20 μm. Data, means ± SE (n=5); one-way ANOVA with Tukey’s multiple comparisons. E,F. HCM quantification of GLUT1 (endogenous protein immunostaining) colocalization with LAMP2 (% of LAMP2 profiles positive for GLUT1; overlap area) in HeLa^WT, and HeLa^Hexa-KO cells. Scale bar, 20 μm. Data, means ± SE (n=5); one-way ANOVA with Tukey’s multiple comparisons. HCM images in all relevant panels, examples from a bank of unbiased operator-independent machine-collected and algorithm-processed fields containing a minimum of 500 primary objects/cells per well (5 wells minimum per 96-well plate; 3 plates minimum), per cell line.

Canonical autophagy does not affect sorting of the retromer cargo GLUT1.
A,B. HCM quantification of GLUT1 (immunostaining of endogenous protein) puncta/cell in in Huh7^WT, Huh7^ATG13-KO, Huh7^FIP200-KO, Huh7^ATG5-KO, and Huh7^VPS35-KO cells. Scale bar, 20 μm. Data, means ± SE (n=5); one-way ANOVA with Tukey’s multiple comparisons. C,D. HCM quantification of GLUT1 (immunostaining of endogenous protein) colocalization with LAMP2 (% of LAMP2 profiles positive for GLUT1; overlap area) in in Huh7^WT, Huh7^ATG13-KO, Huh7^FIP200-KO, Huh7^ATG5-KO, and Huh7^VPS35-KO cells. Scale bar, 20 μm. Data, means ± SE (n=5); one-way ANOVA with Tukey’s multiple comparisons. E-G. HCM quantification of GLUT1 (immunostaining of endogenous protein) puncta/cell (E) and GLUT1 colocalization with LAMP2 (% of LAMP2 profiles positive for GLUT1; overlap area) (F) in Huh7^WT, Huh7^FIP200-KO, Huh7^FIP200-KO ⁺ ^siATG5, and Huh7^VPS35-KO cells. Scale bar, 20 μm. Data, means ± SE (n =6); one-way ANOVA with Tukey’s multiple comparisons. HCM images in all relevant panels, examples from a bank of unbiased operator-independent machine-collected and algorithm-processed fields containing a minimum of 500 primary objects/cells per well (5 wells minimum per 96-well plate; 3 plates minimum), per cell line.

Membrane atg8ylation machinery is required for proper RAB7 localization.
A-C HCM quantification of Rab7 (endogenous protein immunostaining) puncta/cell (B) and Rab7 colocalization with LAMP2 (% of LAMP2 profiles positive for GLUT1; overlap area) (C) in Huh7^WT, Huh7^ATG3-KO, Huh7^ATG5-KO, Huh7^ATG7-KO, Huh7^ATG16-KO, and Huh7^VPS35-KO cells. Scale bar, 20 μm. Data, means ± SE (n =6); one-way ANOVA with Tukey’s multiple comparisons. D-F. HCM quantification of Rab7 (endogenous protein immunostaining) puncta/cell (E) and Rab7 colocalization with LAMP2 (% of LAMP2 profiles positive for GLUT1; overlap area) (F) in Huh7^WT, Huh7^ATG13-KO, Huh7^FIP200-KO, Huh7^ATG5-KO, and Huh7^VPS35-KO cells. Scale bar, 20 μm. Data, means ± SE (n =6); one-way ANOVA with Tukey’s multiple comparisons. G,H. Immunoblot analysis and quantification of proteins in lysosomes purified/enriched by LysoIP (immunoisolation with TMEM192-3xHA) from Huh7^WT, Huh7^ATG5-KO, and Huh7^VPS35-KO cells. TMEM192-2xFLAG, negative control. Data, means ± SE (n=3), one-way ANOVA with Tukey’s multiple comparisons. HCM images in all relevant panels, examples from a bank of unbiased operator-independent machine-collected and algorithm-processed fields containing a minimum of 500 primary objects/cells per well (5 wells minimum per 96-well plate; 3 plates minimum), per cell line.

CASM induction affects sorting of the retromer cargo GLUT1.
A,B. HCM imaging and quantification of LC3 response (puncta/cell of endogenous LC3 immunofluorescent profiles) in HeLa^WT, and HeLa^VPS35-KO cells in response to lysosomal damage by LLOMe (1mM, 30 and 60 min). Scale bar, 20 μm. Data, means ± SE (n=5), one-way ANOVA with Tukey’s multiple comparisons. C-E. HCM quantification of GLUT1 response (puncta/cell of endogenous GLUT1, D) and its localization to endolysosomal compartments (% of LAMP1 profiles positive for GLUT1 immunostaining, E) in Huh7^WT treated with or without Monensin (100µM), LLOMe (100µM), and Bafilomycin A1 (100nM) for 45 minutes. Scale bar, 20 μm. Data, means ± SE (n=5), one-way ANOVA with Tukey’s multiple comparisons. F-H. Analysis of GLUT1 puncta/cell (F) and its localization to endolysosomal compartments (% of LAMP1 profiles positive for GLUT1 immunostaining, G) phenotype monitored by HCM quantification in Huh7^WT cells transfected with GFP (control) or GFP-Rab7^WT, GFP-Rab7^Q67L, and GFP-Rab7^T22N, expressing plasmids. Scale bar, 20 μm. Data, means ± SE (n=3); one-way ANOVA with Tukey’s multiple comparisons. HCM images in all relevant panels, examples from a bank of unbiased operator-independent machine-collected and algorithm-processed fields containing a minimum of 500 primary objects/cells per well (5 wells minimum per 96-well plate; 3 plates minimum), per cell line/condition.

Membrane atg8ylation and endolysosomal homeostasis affect retromer-dependent sorting.
A-C. HCM quantification of GLUT1 (immunostaining of endogenous protein) puncta/cell (B) and GLUT1 colocalization with LAMP2 (% of LAMP2 profiles positive for GLUT1; overlap area, C) in U2OS^WT and U2OS^ATG2A/B-DKO cells treated with or without LLOMe (100µM) for 45 minutes. Scale bar, 20 μm. Data, means ± SE (n =6); one-way ANOVA with Tukey’s multiple comparisons. D-F. HCM quantification of GLUT1 (immunostaining of endogenous protein) puncta/cell (E) and GLUT1 colocalization with LAMP2 (% of LAMP2 profiles positive for GLUT1; overlap area, F) in Huh7^WT cand Huh7^VPS37A-KO cells treated with or without LLOMe (100µM) for 45 minutes. Scale bar, 20 μm. Data, means ± SE (n =6); one-way ANOVA with Tukey’s multiple comparisons. HCM images in all relevant panels, examples from a bank of unbiased operator-independent machine-collected and algorithm-processed fields containing a minimum of 500 primary objects/cells per well (5 wells minimum per 96-well plate; 3 plates minimum), per cell line/condition. G. Schematic: Membrane atg8ylation maintains membrane homeostasis under basal or stress conditions and, independently of canonical autophagy, affects retromer function under. A functional atg8ylation apparatus is required for proper sorting of the retromer cargo GLUT1.

Cornell model of M. tuberculosis latent infection in mice and effects of Atg5 loss in myeloid lineage on disease reactivation.
A. Summary of mice mortality data in acute infection model of Mtb infection (aerosol) based on survival curves in ref. 35. B. Details and timeline of the Cornell latency model experiments (see narrative in Methods). C. Effects of Atg5 loss on spontaneous reactivation of M. tuberculosis infection in Atg5^fl/fl LysM-Cre⁺ mice (loss of Atg5 in myeloid lineage) vs. Atg5^fl/fl LysM-Cre^- (control) mice. Mice were infected with an aerosol of M. tuberculosis (initial deposition, 100-120 CFUs per lung). After a period of 2.5 weeks, initial bacterial growth was assessed by determining lung CFUs (triangles), mice were treated PO with antibiotics (0.1g/L INH and 0.15g/L RIF in drinking water) for 8 weeks, bacterial clearance after chemotherapy assessed by determining lung CFUs (open circles), and remaining mice subjected to antibiotic washout for 7 weeks plus spontaneous reactivation period with no treatment of 3 weeks at which time the mice were sacrificed and lung CFUs determined by plating (filled circles). Data and statistics for spontaneous reactivation: means, †p≥0.05, t test; n=8 mice per group. D. Cornell murine model of M. tuberculosis latent infection and dexamethasone (DXM) induced reactivation and effects of Atg5 loss in myeloid lineage of Atg5^fl/fl LysM-Cre⁺ mice (vs. Atg5^fl/fl LysM-Cre^- control mice). Mice were infected with M. tuberculosis aerosols (initial lung deposition 100-120 CFUs), bacteria allowed to replicate in vivo, mice subjected to antibiotic regimen, followed by antibiotic washout period, after which immunosuppression with DXM was carried out to reactivate infection/bacterial replication (details in Methods). Data, CFU’s per mouse lungs (means ± SE, t test, n=10 mice per group).

ATG5 interactome analysis.
A. Volcano plot of proximity biotinylation LC-MS/MS interactome comparing FlpIn-HeLa^{APEX2-ATG5-WT} treated with or without 2 mM LLOMe for 30 min. Diameter of symbols reflects relative number of unique peptides identified. B. Volcano plot of proximity biotinylation LC-MS/MS interactome comparing FlpIn-HeLa^{APEX2-ATG5-K130R} treated with 2 mM LLOMe for 30 min and FlpIn-HeLa^{APEX2-ATG5-WT} without LLOMe treatment. C. VPS proteins in the MS DIA data. D. VPS proteins identified here compared to VPS proteins in Ref. 86. E. Table: Control MS data (APEX2-SGALS1) for comparison with data in panel C; note no increase in retromer subunits with LLOMe treatment in the APEX1-SGALS1 dataset.

Retromer affects a subset of responses to lysosomal damage.
A,B. HCM imaging and quantification of LC3 (puncta/cell of immunofluorescently stained endogenous LC3 profiles) in HeLa^WT, and HeLa^VPS35-KO cells in response to lysosomal damage by LLOMe. Data, means ± SE (n=3), one-way ANOVA with Tukey’s multiple comparisons. C,D. HCM analysis of ubiquitin (immunofluorescence; FK2 antibody) response to lysosomal damage (LLOMe; 1mM, 2 h) in Huh7^WT, Huh7^ATG5-KO, and Huh7^VPS35-KO cells. Yellow profiles, colocalization of ubiquitin and LAMP1. Data, means ± SE (n=3), one-way ANOVA with Tukey’s multiple comparisons. E,F. HCM quantification of ALIX localization to endolysosomal compartments (% of LAMP1 profiles positive for ALIX immunostaining) in HeLa^WT and HeLa^VPS35-KO cells following lysosomal damage. Yellow profiles, colocalization of ALIX and LAMP1. Data, means ± SE (n=3); two-way ANOVA with Tukey’s multiple comparisons. HCM images in all relevant panels, examples from a bank of unbiased operator-independent machine-collected and algorithm-processed fields containing a minimum of 500 primary objects/cells per well (5 wells minimum per 96-well plate; 3 plates minimum), per cell line/condition.

Membrane atg8ylation regulates retromer function.
A. Immunoblots of CRISPR KOs: (i) ATG3, ATG5, and ATG7 in Huh7 cells; (ii) VPS35 in Huh7 cells; (iii) ATG5 in HeLa cells; and (iv) VPS35 in Hela cells. B,C. HCM images (example form a bank of unbiased operator-independent machine collected and processed images containing a minimum of 500 primary objects/cells) of GLUT1 (immunostaining of endogenous protein) puncta/cell (B) and GLUT1 colocalization with LAMP2 in Huh7^WT, Huh7^ATG5-KO, and Huh7^VPS35-KO cells (C). Scale bar, 20 μm. D,E. HCM quantification of GLUT1 (immunostaining of endogenous protein) puncta/cell in HeLa^WT, HeLa^ATG5-KO, and HeLa^VPS35-KO cells. Scale bar, 20 μm. Data, means ± SE (n =5); one-way ANOVA with Tukey’s multiple comparisons. F,G. HCM quantification of GLUT1 colocalization with LAMP2 (% of LAMP2 profiles positive for GLUT1; overlap area) in HeLa^WT, HeLa^ATG5-KO, and HeLa^VPS35-KO cells. Scale bar, 20 μm. Data, means ± SE (n =5); one-way ANOVA with Tukey’s multiple comparisons. H. Confocal images illustrating localization of GLUT1 and LAMP2 in HeLa^WT, HeLa^ATG5-KO, and HeLa^VPS35-KO cells. Scale bar, 10 μm. I,J. HCM quantification of GLUT1-SNX27 overlap (% of SNX7 area positive for GLUT1) in HeLa^WT, HeLa^ATG5-KO, and HeLa^VPS35-KO. Scale bar, 20 μm. Data, means ± SE (n=4); one-way ANOVA with Tukey’s multiple comparisons. HCM images in all relevant panels, examples from a bank of unbiased operator-independent machine-collected and algorithm-processed fields containing a minimum of 500 primary objects/cells per well (5 wells minimum per 96-well plate; 3 plates minimum), per cell line.

ATG5 knockout has no effect on protein levels of retromer subunits.
A,B. Immunoblot analysis (A) and quantification (B) of retromer complex proteins VPS35 (i), VPS26 (ii), and VPS29 (iii) from Huh7^WT, Huh7^ATG3-KO, Huh7^ATG5-KO, and Huh7^ATG7-KO cells (total cell extract). C,D. Immunoblot analysis (C) and quantification (D) of retromer complex proteins VPS35 (i), VPS26 (ii), and VPS29 (iii) from HeLa^WT, and HeLa^ATG5-KO cells (total cell extract). E. HCM quantification of GLUT1-LAMP2 colocalization in HeLa^LC3-TKO, HeLa^GABA-TKO, and HeLa^HEXA-KO cells. Data, means ± SE (n =4), one-way ANOVA with Tukey’s multiple comparisons. F. HCM images of GLUT1 puncta/cell in Huh7^WT, Huh7^FIP200-KO, Huh7^FIP200-KO+ ^siATG5, and Huh7^VPS35-KO cells. G. Immunoblot analysis showing the siRNA mediated knockdown of ATG5 in Huh7^FIP200-KO cells. HCM images in panel F, examples from a bank of unbiased operator-independent machine-collected and algorithm-processed fields containing a minimum of 500 primary objects/cells per well (5 wells minimum per 96-well plate; 3 plates minimum), per cell line/condition.

Loss of membrane atg8ylation but not of canonical autophagy diverts of RAB7 to lysosomal compartments.
A. HCM images of Rab7 (immunostaining of endogenous protein) showing Rab7 colocalization with LAMP2 (% of LAMP2 profiles positive for GLUT1; overlap area) in Huh7^WT, Huh7^ATG3-KO, Huh7^ATG5-KO, Huh7^ATG7-KO, Huh7^ATG16-KO, and Huh7^VPS35-KO cells. Scale bar, 20 μm. B. HCM images of Rab7 (puncta/cell of endogenous Rab7 profiles stained for immunofluorescence) in Huh7^WT, Huh7^ATG13-KO, Huh7^FIP200-KO, Huh7^ATG5-KO, and Huh7^VPS35-KO cells. Scale bar, 20 μm. C,D. HCM images (C) and quantification of RAB7 puncta (D) in Huh7^WT, Huh7^FIP200-KO, Huh7^FIP200-KO ⁺ ^siATG5, and Huh7^VPS35-KO cells. Scale bar, 20 μm. Data, means ± SE (n =8), one-way ANOVA with Tukey’s multiple comparisons. E,F. HCM images (E) and quantification of RAB7-LAMP2 colocalization (F) in Huh7^WT, Huh7^FIP200-KO, Huh7^FIP200-KO ⁺ ^siATG5, and Huh7^VPS35-KO cells. Scale bar, 20 μm. Data, means ± SE (n =8), one-way ANOVA with Tukey’s multiple comparisons. HCM images in all relevant panels, examples from a bank of unbiased operator-independent machine-collected and algorithm-processed fields containing a minimum of 500 primary objects/cells per well (5 wells minimum per 96-well plate; 3 plates minimum), per cell line.

CASM agonists effects on retromer cargo GLUT1 in the absence of changes in TBC1D5-LC3A association.
A. HCM images of GLUT1 puncta in Huh7^WT cells upon treatment with Monensin (100µM), LLOMe (100µM), and Bafilomycin A1 (100nM) for 45 minutes. B,C. Co-IP analysis (B) and quantification (C) of TBC1D5 and GFP-LC3 interaction in Huh7 cells treated with or without Monensin (100µM), LLOMe (100µM), and Bafilomycin A1 (100nM) for 45 minutes. Data, means ± SE (n=3), one-way ANOVA with Tukey’s multiple comparisons. HCM images in A, examples from a bank of unbiased operator-independent machine-collected and algorithm-processed fields containing a minimum of 500 primary objects/cells per well (5 wells minimum per 96-well plate; 3 plates minimum), per condition.

A. HCM representative images of GLUT1 puncta in U2OS^WT and U2OS^ATG2A/B-DKO cells upon treatment with LLOMe (100µM) for 45 minutes. B. HCM representative images of GLUT1 puncta in Huh7^WT and Huh7^VPS37A-KO cells upon treatment with LLOMe (100µM) for 45 minutes. C,D. HCM quantification of LC3 puncta in Huh7^WT, Huh7^ATG3-KO, Huh7^ATG5-KO, Huh7^ATG7-KO, and Huh7^VPS35-KO cells induced for autophagy in EBSS for 90 minutes. Plot in C, an example of a whole 96-well plate HCM readout; X axis, well positions; Y axis, HCM parameter quantified. Plot in D, nested data from three different plates each one as in C. FM, full medium; EVSS, starvation medium. Data, means ± SE (n=3); one-way ANOVA with Tukey’s multiple comparisons. E,F. Co-IP analysis (E) and quantification (F) of VPS35 and YFP-VPS29 interaction in HeLa^WT, and HeLa^ATG5-KO cells. Data, means ± SE (n=3), one-way ANOVA with Tukey’s multiple comparisons. G,H. Co-IP analysis (G) and quantification (H) of VPS26 and YFP-VPS29 interaction in HeLa^WT, and HeLa^ATG5-KO cells. Data, means ± SE (n=3); unpaired t-test.

Sign up for email alerts