Receptor-mediated cargo hitchhiking on bulk autophagy

Abstract

While the molecular mechanism of autophagy is well studied, the cargoes delivered by autophagy remain incompletely characterized. To examine the selectivity of autophagy cargo, we conducted proteomics on isolated yeast autophagic bodies, which are intermediate structures in the autophagy process. We identify a protein, Hab1, that is highly preferentially delivered to vacuoles. The N-terminal 42 amino acid region of Hab1 contains an amphipathic helix and an Atg8-family interacting motif, both of which are necessary and sufficient for the preferential delivery of Hab1 by autophagy. We find that fusion of this region with a cytosolic protein results in preferential delivery of this protein to the vacuole. Furthermore, attachment of this region to an organelle allows for autophagic delivery in a manner independent of canonical autophagy receptor or scaffold proteins. We propose a novel mode of selective autophagy in which a receptor, in this case Hab1, binds directly to forming isolation membranes during bulk autophagy.

Keywords: Saccharomyces cerevisiae; Atg8; Autophagy; Hab1; Selective Autophagy.

Figure 1. Proteomic analyses of autophagic bodies.

(A) Overview of the strategy used to isolate autophagic bodies and analyze their cargo. (B) Immunoblots of fractionated autophagic body samples. Logarithmically growing atg15Δ cells in SD medium were treated with rapamycin (200 ng/mL) for 2 h. Autophagic bodies were isolated from disrupted vacuoles by density gradient centrifugation and subjected to immunoblotting. Heavy fractions containing Pgk1 indicate autophagic body fractions. Control fractions were obtained from atg15Δ atg2Δ cells by the same procedure. Pho8 was used as a vacuolar marker. TL, total cell lysate; In, input. (C, D) The logarithm of protein enrichment in isolated autophagic bodies (C) 2 h rapamycin treatment; (D) 2 h nitrogen starvation) (y-axis) and the logarithm of abundance in total cell lysates (x-axis) are indicated. Red, known preferential cargo and autophagy machinery proteins; green, cytosolic proteins described in the main text; blue, glycogen-related enzymes; purple, histones. Source data are available online for this figure.

Figure 2. Hab1 is preferentially delivered to the vacuole via autophagy.

(A) GFP cleavage of Hab1-GFP, Pgk1-GFP, Fas1-GFP, and Ald6-GFP. Each sample was examined after appropriate dilution (Pgk1-GFP: 1/500, Fas1-GFP: 1/50, Ald6-GFP: 1/500). GFP′, free-GFP; +Rap, rapamycin treatment; –N, nitrogen starvation. Intensities of protein-GFP and GFP′ bands were measured to estimate the delivery of protein-GFP to vacuoles. The percentages of GFP′ to total GFP (GFP′ + full-length GFP) are shown as mean ± SD (n = 3). (B) Confocal fluorescence imaging of WT or atg2Δ cells expressing Hab1-GFP and Vph1-2xmCherry (a vacuolar membrane protein). Scale bar, 5 µm (C) GFP cleavage of Hab1-GFP was determined in WT, atg1Δ, atg2Δ, and atg15Δ cells at indicated time points of rapamycin treatment. Ape1 is shown as a marker of autophagic activity. Total protein was visualized by Ponceau S staining. (D) Dependence of Hab1 degradation on Atg proteins, as determined by Hab1-GFP cleavage following rapamycin treatment. * = intermediate degradation or mid-translation products of Hab1-GFP. Percentages of GFP′ to total GFP (GFP′ + full-length GFP) are shown as mean ± SD (n = 3). Blot data are representative of two (C) or three (A, D) independent experiments. Source data are available online for this figure.

Figure 3. The N-terminal amphipathic helix and the AIM of Hab1 are responsible for preferential delivery.

(A) Secondary structure of Hab1 as predicted by JPred 4. Truncation mutants of Hab1 are also depicted. (B) GFP cleavage of Hab1 C-terminal truncation variants. Percentages of GFP′ to total GFP (GFP′ + full-length GFP) are shown as mean ± SD (n = 3). (C) A helical wheel projection of residues 3–15 of Hab1. (D) Confocal fluorescence imaging of GFP-tagged Hab1 variant-expressing cells after 4 h rapamycin treatment. Scale bar, 5 µm. (E) GFP cleavage assay of Hab1 N-terminal helix mutants. (F) An illustration of the secondary structure of Hab1(1–42). (G) GFP cleavage of WT and putative AIM mutants (F35A, I38A, and F35A/I38A) of Hab1. Blot data are representative of two (E, G) or three (B) independent experiments. Source data are available online for this figure.

Figure 4. Hab1(1–42) specifically binds to lipidated Atg8.

(A) Binding of Atg8 to Hab1(1–42)-GFP as determined by immunoprecipitation. Samples were obtained from cells treated with rapamycin for 1 h before being subjected to immunoblotting. Quantifications of the immunoprecipitation band intensities are shown as mean ± SD (n = 3); data are normalized to the WT Atg8-PE band. (B) Binding of Atg8 to Hab1(1–42)-GFP in Atg8 lipidation-defective mutants assessed by immunoprecipitation. Quantifications of band intensities are shown as mean ± SD (n = 3); data are normalized to the WT Atg8-PE band. (C) Binding of Atg8 to Hab1(1–42)-GFP variants assessed by immunoprecipitation. Quantifications of band intensities are shown as mean ± SD (n = 3); data are normalized to the WT Atg8-PE band. Blot data are representative of three independent experiments (A–C). Source data are available online for this figure.

Figure 5. A possible role for Hab1 in ribosomal degradation.

(A) GFP cleavage of Hab1 truncation mutants in WT or atg24Δ cells. Percentages of GFP′ to total GFP (GFP′ + full-length GFP) are shown as mean ± SD (n = 3). ND, not determined. (B) Immunoblotting of subcellular fractionated lysates of Hab1-GFP strains. Total (T), supernatant (S), and pellet (P) fractions from cell lysates after rapamycin treatment of cells are shown. Rpl23 is a ribosomal protein control. Percentages of Hab1-GFP recovery of S and P (relative to T) are shown as mean ± SD (n = 3). (C) Immunoblots of sucrose density gradient centrifugated fractions after cross-linking using DSP. Lysates were obtained from atg2Δ cells expressing Hab1-GFP following 4 h of rapamycin treatment. Rpl23 and Rps25B-HA were used as markers of the 60S and 40S ribosomal subunits, respectively. Hsp90 is a negative control that does not bind to ribosomes. Absorbance at 254 nm reflects the amount of RNA. In, input. (D) Proteomic analyses of ribosomal proteins in autophagic bodies. Autophagic bodies isolated from atg15Δ, hab1Δ atg15Δ, and HAB1-overexpressing atg15Δ cells following 6 h rapamycin treatment were subjected to proteomic analyses. Fold changes are shown as normalized label-free quantification of each 40S or 60S ribosomal protein in autophagic bodies isolated from hab1Δ or HAB1-overexpressing cells relative to that of WT cells. “Total” indicates all detected proteins. ***p < 0.001. p values were calculated using an unpaired two-sided Welch’s t-test. Box plots are shown as median (middle bar) with 25th and 75th percentiles and 1.5 × interquartile ranges in whiskers. The numbers of annotated proteins are as follows: Total, 1888; 60S, 59; 40S, 47. (E) Representative TEM images of autophagic bodies in atg15Δ, atg15Δ hab1Δ, and HAB1-overexpressing atg15Δ cells. A TEM image of the rough endoplasmic reticulum (RER) is shown as a morphological control for ribosomes. PM, Plasma membrane. Scale bar, 200 nm. (F) Quantification of the frequency of ribosomes associated with autophagic body membranes, related to (E). The membrane localization frequencies of ribosomes in individual autophagic bodies are shown as dots in box plots. The n values for each sample are as follows: WT, 109; hab1Δ, 146; HAB1-overexpression, 151. ***p < 0.001. p values were calculated using an unpaired two-sided Welch’s t-test. Box plots are shown as median (middle bar) with 25th and 75th percentiles and 1.5 × interquartile ranges in whiskers. Blot data are representative of two (C, D) or three (A, B) independent experiments. Source data are available online for this figure.

Figure 6. Hab1(1–42) functions as an autophagy degron.

(A) An overview of the strategy used to determine Pho8Δ60-GFP delivery by binding to Hab1(1–42)-GBP. (B) The effect of Hab1(1–42) binding to Pho8Δ60 on its delivery as determined by ALP assay. Data were acquired from biological triplicates (mean ± SD). (C) Fluorescence micrographs of Pho8Δ60-GFP-expressing cells. Merged images shown. FM4-64, vacuolar marker. Scale bar, 5 µm. (D) An illustration of the organelle delivery assay employing synthetic tethering using the ALFA-tag system. (E) Rates of mitochondrial delivery to the vacuole were examined by GFP cleavage of Om45-GFP. Hab1(1–42) was tethered to mitochondria using the ALFA-tag system. –N, nitrogen starvation; mSc, mScarlet (Bindels et al, 2017); Percentages of GFP′ to total GFP (GFP′ + full-length GFP) are shown as mean ± SD (n = 3). ND, not determined. Blot data are representative of two independent experiments. (F) Rates of mitochondrial delivery to the vacuole were examined by GFP cleavage of Om45-GFP. Percentages of GFP′ to total GFP (GFP′ + full-length GFP) are shown as mean ± SD (n = 3). A subset of data is derived from samples used for quantification in (E). Blot data are representative of three independent experiments. Source data are available online for this figure.

Figure EV1. Fluorescence microscopy of Hab1-GFP (related Fig. 2B).

(A) Fluorescence micrographs of WT or atg1Δ cells expressing Hab1-GFP and Vph1-2xmCherry. Scale bar, 5 µm. (B) Fluorescence micrographs of Hab1-GFP and mCherry-Atg8-expressing cells after treatment with rapamycin for 1 or 4 h. Arrowheads show colocalization. Scale bar, 5 µm. (C) Time-lapse imaging of cells expressing Hab1-GFP and mCherry-Atg8. The time after rapamycin addition at which each snapshot was obtained is noted in the upper left corner of each. Arrowheads indicate mCherry-Atg8 or Hab1-GFP puncta. Scale bar, 5 µm.

Figure EV2. Hab1 N-terminal helix mutants (related to Fig. 3).

(A) GFP cleavage of WT and N-terminal truncation mutants of Hab1. (B) Fluorescence micrographs of Hab1-GFP and Sec63-mScarlet (ER-membrane marker) observed by spinning disc confocal fluorescence microscopy. Scale bar, 5 µm. (C) Fluorescence micrographs of Hab1(1–42)-GFP-expressing cells. Scale bar, 5 µm.

Figure EV3. Hab1 binds to ribosome (related to Fig. 5).

(A) A network map of Hab1(43–144)-interacting proteins. Proteins immunoprecipitated with Hab1(43–144)-GFP were detected by LC-MS/MS. Detected Hab1(43–144)-GFP interacting proteins were analyzed using STRING (Szklarczyk et al, 2019) after control sample (GFP-immunoprecipitated) proteins were excluded. (B) GO enrichment analysis of Hab1-interacting proteins. Analysis was performed on the cellular component for all proteins identified in (A) using the GO Consortium (accessed August, 2022) (Ashburner et al, 2000). GOs with FDR-adjusted p-values lower than 0.01 are indicated. FDR was calculated using the one-way Fisher’s exact test. (C) Confirmation of overexpression of Hab1 by placing HAB1 under the control of the TEF1 promoter. Cells expressing Hab1-GFP under the control of the HAB1 or TEF1 promoter were subjected to immunoblotting. Cells were cultured in SD media to the logarithmic growth phase and then treated with rapamycin to induce autophagy. (D) The effect of Hab1 on ribosome biosynthesis was evaluated by quantification of Rpl23. There was no detectable change in ribosomal protein levels in atg15Δ, atg15Δ hab1Δ, and HAB1-overexpressing atg15Δ cells in the logarithmic growth phase on YPD. Quantifications of band intensities are shown as mean ± SD. (n = 3); data are normalized to the WT band. (E) An analysis of non-ribosomal proteins in autophagic bodies detected in Fig. EV3A. Fold changes are shown as normalized label-free quantification of each 40 S or 60S ribosomal proteins in autophagic bodies isolated from hab1Δ (x-axis) or HAB1-overexpressing cells (y-axis) relative to that of WT cells. Green: non-ribosomal proteins detected in Fig. EV3A, red: 60S ribosomal proteins, blue: 40S ribosomal proteins. The same data set as shown in Fig. 5D was used. (F) Examination of abundance ratios of Hab1 to ribosomes by immunoblotting. Hab1-GFP- or Rpl10-GFP-expressing atg2Δ cells were cultured in YPD media to the logarithmic phase of growth and then treated with rapamycin or cultured in SD–N medium for 4 h. Cells were then harvested. For Rpl10-GFP-expressing cells, diluted samples were applied at the ratios described below each lane.

Figure EV4. The N- and C-terminal regions of Hab1 are important for the Hab1-dependent delivery of ribosomes to vacuoles.

(A) Hab1 orthologs of Zygosaccharomyces rouxii, Torulaspora delbrueckii, Lanchancea thermotolerans, and Kluyveromyces marxianus were identified based on sequence homology using PSI-BLAST searches. Amino acid sequences are shown. The alignment was performed using COBALT (https://www.ncbi.nlm.nih.gov/tools/cobalt/). Conserved residues are shown in red. (B) Immunoblotting of subcellular fractions. Lysates were obtained from cells expressing Hab1-GFP truncates. (C) Immunoblotting of subcellular fractions. Lysates were obtained from cells expressing Hab1-GFP point mutants. (D) GFP cleavage of Hab1 point mutants in WT and atg24Δ cells. The percentages of GFP′ to total GFP (GFP′ + full-length GFP) are shown as mean ± SD (n = 3). (E) Proteomic analyses of ribosomal proteins in autophagic bodies. Autophagic bodies isolated from hab1Δ atg15Δ cells harboring empty vector, HAB1(WT), HAB1(43–144) or HAB1(M58A) following 6 h rapamycin treatment were subjected to proteomic analyses. Fold changes are shown as normalized label-free quantifications of each 40S or 60S ribosomal protein in autophagic bodies. “Total” indicates all detected proteins. Box plots are shown as median (middle bar) with 25th and 75th percentiles and 1.5 × interquartile in whiskers.

Figure EV5. Binding to Hab1(1–42) allows receptor- and scaffold protein-independent delivery of organelles to vacuoles (related to Fig. 6).

(A) Fluorescence microscope images of Hab1(1–42)-bound mitochondria showing delivery to the vacuole. Om45-GFP-expressing cells were observed 24 h after shifting to SD–N medium. Scale bar, 5 µm. (B) Rates of peroxisomal delivery to the vacuole were examined by GFP cleavage of Pex14-GFP. Hab1(1–42) was tethered to mitochondria using the ALFA-tag system. –N, nitrogen starvation; Percentages of GFP′ to total GFP (GFP′ + full-length GFP) are showns as mean ± SD (n = 3). ND, not determined. Blot data are representative of two independent experiments. (C) Rates of peroxisomal delivery to the vacuole were examined by GFP cleavage of Pex14-GFP. Percentages of GFP′ to total GFP (GFP′ + full-length GFP) are shown as mean ± SD (n = 3). A subset of data is derived from samples used for quantification in (B). Blot data are representative of three independent experiments. (D) Fluorescence micrographs during nitrogen starvation (related to Figs. EV5B and 5C). Scale bar, 5 µm.