AMPK regulates phagophore-to-autophagosome maturation

Abstract

Autophagy is an important metabolic pathway that can non-selectively recycle cellular material or lead to targeted degradation of protein aggregates or damaged organelles. Autophagosome formation starts with autophagy factors accumulating on lipid vesicles containing ATG9. These phagophores attach to donor membranes, expand via ATG2-mediated lipid transfer, capture cargo, and mature into autophagosomes, ultimately fusing with lysosomes for their degradation. Autophagy can be activated by nutrient stress, for example, by a reduction in the cellular levels of amino acids. In contrast, how autophagy is regulated by low cellular ATP levels via the AMP-activated protein kinase (AMPK), an important therapeutic target, is less clear. Using live-cell imaging and an automated image analysis pipeline, we systematically dissect how nutrient starvation regulates autophagosome biogenesis. We demonstrate that glucose starvation downregulates autophagosome maturation by AMPK-mediated inhibition of phagophore tethering to donor membrane. Our results clarify AMPKs regulatory role in autophagy and highlight its potential as a therapeutic target to reduce autophagy.

Figure 1.

K-FOCUS: Live-cell high-throughput single-cell analysis of foci colocalization. (A) Illustration of the analysis pipeline for the Cellpose-TrackIt module, which incorporates CellPose segmentation into TrackIt (scale bar = 10 µm). (B) Workflow for manual ROI quality control and foci tracking using the TrackIt GUI. (C) A schematic outlining the colocalization criteria, including user-defined inputs, and key outputs. Additionally, it presents a visual representation of a colocalized and non-colocalized track, along with a kernel density plot of track length per cell for both colocalized and non-colocalized tracks.

Figure 2.

K-FOCUS is a robust object-based colocalization tool in live-cell imaging. (A) Example images demonstrating the accumulation of Halo-ATG9A in lysosomes, captured immediately after or 24 h following labeling with JFX650 in U2OS cells (scale bar = 10 µm). (B) A comparison between different colocalization analysis methods for the data shown in A, including threshold-based Pearson and Manders coefficients, as well as wavelet spot detection–based SODA and K-FOCUS. The box indicates the interquartile range, the whiskers the 10–90% confidence interval, the square indicates the average, and the horizontal line is the median; statistical significance was assessed by a two-tailed t-test. (C) Example images of U2OS cells expressing GFP-LC3B and Halo-ATG13 under both control and EBSS conditions (scale bar = 5 µm). (D) A comparison between different colocalization analysis methods for the data shown in C, including threshold-based Pearson and Manders coefficients, as well as wavelet spot detection–based SODA and K-FOCUS analyses. The box indicates the interquartile range, the whiskers the 10–90% confidence interval, the square indicates the average, and the horizontal line is the median; statistical significance was assessed by a two-tailed t-test.

Figure 3.

K-FOCUS analysis of the kinetics of autophagy factor foci formation under nutrient starvation. (A) Live-cell images of Halo-WIPI2 and GFP-LC3B in control and starvation medium (scale bar = 5 µm). (B) Quantification of colocalization kinetics of single cells in A and Fig. S1 using K-FOCUS, including total, colocalized, and non-colocalized phagophore formation rates. Data was downsampled to a pixel size of 370 nm. (C) Quantified fractions of LC3+ foci/cell from imaging in A and Fig. S1. For B and C, the box indicates the interquartile range, the whiskers the 10–90% confidence interval, the square indicates the average, and the horizontal line is the median. Statistical significance was assessed by one-way ANOVA. The letter assignments denote statistically significant differences among groups at P = 0.05, as determined by a Bonferroni post-hoc test.

Figure S1.

K-FOCUS analysis of the kinetics of autophagy factor foci formation under nutrient starvation. Live-cell images of ULK1-Halo, Halo-ATG13, or Halo-ATG2A and GFP-LC3B in control and starvation medium (scale bar = 5 µm).

Figure S2.

Autophagy flux analysis. (A) Western blot of control U2OS cells and Halo-ATG13, Halo-ATG2, or Halo-WIPI2 expressing U2OS cells stably coexpressing GFP-LC3B probed with an antibody against LC3B. Cell lysates were generated with CHAPS buffer split in half and treated with or without TEV protease. (B) Flow cytometry analysis of GFP-LC3B expression in control U2OS cells and Halo-ATG13, Halo-ATG2, or Halo-WIPI2 expressing U2OS cells stably coexpressing GFP-LC3B. (C) Quantification of median cellular GFP-LC3B signal intensity in Halo-ATG13, Halo-ATG2, Halo-ULK1, or Halo-WIPI2 expressing U2OS cells stably coexpressing GFP-LC3B cultured in complete media. (D) Western blot probed for LC3B of control U2OS cells, and U2OS cells expressing Halo-ATG13 or Halo-ATG2 grown in complete media, media lacking amino acids, or media lacking amino acids and FBS in the absence and presence of bafilomycin. (E) Western blot probed for LC3B of control U2OS cells, and U2OS cells expressing Halo-ULK1 or Halo-WIPI2 grown in complete media, media lacking amino acids, or media lacking amino acids and FBS in the absence and presence of bafilomycin. (F) Western blot probed for LC3B of control U2OS cells, and U2OS cells expressing Halo-ATG13, Halo-ATG2, Halo-ULK1, or Halo-WIPI2 grown in complete media, or media lacking glucose in the absence and presence of bafilomycin. (G) Quantification of the LC3B-II band intensity of the western blots shown in D and E (three biological replicates, mean and standard deviation). (H) Quantification of the LC3B-II band intensity of the western blot shown in panel F (two biological replicates, mean and standard deviation). (I) Western blot probed for GABARAP of control U2OS cells, and U2OS cells expressing Halo-ATG13, Halo-ATG2, Halo-ULK1, or Halo-WIPI2 grown in complete media, or media lacking glucose in the absence and presence of bafilomycin. Source data are available for this figure: SourceData FS2.

Figure S3.

K-FOCUS analysis of autophagy factor foci lifetimes. (A) Lifetimes of LC3B-positive and LC3B-negative trajectories of Halo-ATG13, ULK1-Halo, Halo-WIPI2, and Halo-ATG2A in different media conditions. (B) Lifetimes of LC3B-positive Halo-ATG13, ULK1-Halo, Halo-WIPI2, and Halo-ATG2A trajectories before and during colocalization with LC3B in different media conditions.

Figure S4.

Analysis of LC3B G120A aggregates, colocalization of Halo-ATG13 with GFP-P62, and assessment of alternate data processing. (A) K-FOCUS analysis of Halo-ATG13 colocalization with GFP-LC3B G120A expressed by viral transduction in cells grown in complete media, media lacking amino acids and FBS, or media lacking glucose. (B) K-FOCUS analysis of Halo-ATG13 colocalization with stably expressed GFP-P62 in cells grown in complete media, media lacking amino acids and FBS, or media lacking glucose. (C) Quantification of colocalization kinetics of single cells in Fig. 3 A and Fig. S1 using K-FOCUS including total, colocalized, and non-colocalized phagophore formation rates. Data was binned 2 × 2 to a pixel size of 217 nm. (D) Quantified fractions of LC3+ foci/cell from imaging in Fig. 3 A and Fig. S1. For B, C, and D, the box indicates the interquartile range, the whiskers the 10–90% confidence interval, the square indicates the average, and the horizontal line is the median. Statistical significance was assessed by one-way ANOVA. The letter assignments denote statistically significant differences among groups at P = 0.05, as determined by a Bonferroni post-hoc test.

Figure S5.

AMPK inhibits autophagosome maturation. (A) Western blots probed for ULK1 phosphorylated at S556 of control U2OS cells, and U2OS cells expressing Halo-ATG13, Halo-ATG2, Halo-ULK1, or Halo-WIPI2 grown in complete media, or media lacking glucose. (B) Western blots probed for ULK1 phosphorylated at S556 of control U2OS cells, and U2OS cells expressing Halo-ATG13, or Halo-WIPI2 grown in complete media, or media lacking glucose in the absence or presence of 1 µM or 20 µM of the AMPK inhibitor BAY3827. Total protein staining was carried out prior to protein transfer onto a nitrocellulose membrane and the membrane was cut into two pieces for western blot procedures. Both membrane pieces were stained in the same dish and exposed simultaneously. (C) Western blots probed for ULK1 phosphorylated at S556 of control U2OS cells, and U2OS cells expressing Halo-ATG13, or Halo-WIPI2 grown in complete media, in the absence and presence of 1 µM or 20 µM of the AMPK activator MK8722. (D) Live-cell images of ULK1-Halo, Halo-ATG13, or Halo-ATG2A and GFP-LC3B in the indicated media conditions in the presence and absence of the AMPK activator MK8722 or the AMPK inhibitor BAY3827 (scale bar = 5 µm). Source data are available for this figure: SourceData FS5.

Figure 4.

AMPK activation inhibits autophagosome maturation. (A) Formation rate and conversion ratio of LC3-positive and LC3-negative foci formed by Halo-ATG13, ULK1-Halo, Halo-WIPI2, and Halo-ATG2A in glucose-containing media in the absence and presence of the AMPK agonist MK8722. (B) Formation rate and conversion ratio of LC3-positive and LC3-negative foci formed by Halo-ATG13, ULK1-Halo, Halo-WIPI2, and Halo-ATG2A in media containing no glucose in the absence and presence of the AMPK inhibitor BAY3827. (C) Lifetimes of foci formed by Halo-ATG13, ULK1-Halo, Halo-WIPI2, and Halo-ATG2A before and during colocalization with GFP-LC3B in glucose-containing media in the presence and absence of the AMPK agonist MK8722 (left panels) or in media lacking glucose in in the absence and presence of the AMPK inhibitor BAY3827 (right panels). For all plots the box indicates the interquartile range, the whiskers the 10–90% confidence interval, the square indicates the average, and the horizontal line is the median; statistical significance was assessed by a two-tailed t-test.

Figure 5.

Contribution of the ULK1 complex to starvation-induced autophagy protein foci formation. (A) Formation rate of total, LC3B-positive, and LC3B-negative Halo-ATG13 foci in control, ULK1, FIP200, and ATG101 knock-out cells in various media conditions. (B) Lifetimes of LCB3-positive and LC3B-negative Halo-ATG13 foci in control, ULK1, FIP200, and ATG101 knock-out cells in various media conditions. Statistical significance was assessed by one-way ANOVA. The letter assignments denote statistically significant differences among groups at P = 0.05, as determined by a Bonferroni post-hoc test.

Figure 6.

AMPK regulates the tethering of WIPI2-positive phagophores. (A) Model depicting the diffusion properties of tethered phagophores and untethered ATG9 vesicles. (B and C) Step size distributions of Halo-ATG13, ULK1-Halo, Halo-WIPI2, and Halo-ATG2A trajectories for (B) LC3B-positive and (C) LC3B-negative autophagy factor foci. (D) Kernel density plots of the step size distributions of Halo-ATG13, ULK1-Halo, Halo-WIPI2, and Halo-ATG2A trajectories for LC3B-positive and -negative tracks in the indicated media conditions. The vertical white solid line represents the median of step size distribution. (E) Step size distributions of LC3B-positive and -negative Halo-WIPI2 trajectories fit with a two-state diffusion model (black line) encompassing tethered (red) and untethered (blue) populations in the different media conditions. (F) Diffusion coefficients of the tethered and untethered populations of LC3B-positive and -negative Halo-WIPI2 trajectories in the different media conditions, derived from the fits shown in E. (G) Distribution of LC3B-positive and -negative Halo-WIPI2 trajectories between tethered and untethered populations, derived from the fits shown in E. Statistical significance was assessed using a two-tailed t-test (*P < 0.05, **P < 0.01).

Figure S6.

Diffusion dynamics of autophagy factor foci. (A) Western blots probed for ATG9A of control U2OS cells and clonal U2OS cells expressing Halo-ATG13 or Halo-WIPI2 that were transfected with Cas9 and sgRNAs targeting the ATG9A gene. (B) TrackIt analysis of Halo-ATG13 foci in control cells and ATG9A knock-out cells grown in control media, media lacking amino acids and FBS, or media lacking glucose. The red solid line indicates the median. (C) TrackIt analysis of Halo-WIPI2 foci in control cells and ATG9A knock-out cells grown in control media, media lacking amino acids and FBS, or media lacking glucose. The red solid line indicates the median. (D) Step size distributions of LC3B-negative and LC3B-positive autophagy factor trajectories before and during colocalization with LC3B in indicated media conditions. (E) Step size distributions of LC3B-negative and LC3B-positive Halo-WIPI2 foci in the presence and absence of the AMPK agonist MK8722. (F) Kernel density plot of the step size distribution of LC3B-negative and LC3B-positive Halo-WIPI2 foci in the presence and absence of the AMPK agonist MK8722 (in media with glucose) or the presence and absence of the AMPK inhibitor BAY3827 (in media lacking glucose). Source data are available for this figure: SourceData FS6.

Figure S7.

Diffusion dynamics of autophagy factor foci. (A) Step size distributions of LC3B-positive and -negative ULK1-Halo, and Halo-ATG2A trajectories fit with a two-state diffusion model (black line) encompassing tethered (red) and untethered (blue) populations in the different media conditions. (B) Diffusion coefficients of the tethered and untethered populations of LC3B-positive and -negative ULK1-Halo, and Halo-ATG2A trajectories in the different media conditions, derived from the fits shown in Fig. S6 A. Statistical significance was assessed using a two-tailed t-test (*P < 0.05).

Figure 7.

Model of the regulation of autophagosome biogenesis by AMPK. Glucose withdrawal leads to AMPK activation, which promotes the recruitment of ATG13 to mobile ATG9 vesicles. Under these conditions, WIPI2 also accumulates on ATG9 vesicles, but their tethering to donor membranes to allow phagophore maturation is inhibited by AMPK activation. The exact molecular targets of AMPK that prevent phagophore tethering are unknown.