Ribosome profiling reveals a functional role for autophagy in mRNA translational control

Abstract

Autophagy promotes protein degradation, and therefore has been proposed to maintain amino acid pools to sustain protein synthesis during metabolic stress. To date, how autophagy influences the protein synthesis landscape in mammalian cells remains unclear. Here, we utilize ribosome profiling to delineate the effects of genetic ablation of the autophagy regulator, ATG12, on translational control. In mammalian cells, genetic loss of autophagy does not impact global rates of cap dependent translation, even under starvation conditions. Instead, autophagy supports the translation of a subset of mRNAs enriched for cell cycle control and DNA damage repair. In particular, we demonstrate that autophagy enables the translation of the DNA damage repair protein BRCA2, which is functionally required to attenuate DNA damage and promote cell survival in response to PARP inhibition. Overall, our findings illuminate that autophagy impacts protein translation and shapes the protein landscape.

Fig. 1. Minimal effects of ATG12 genetic deletion on global translation.

a Representative ³⁵S methionine incorporation autoradiogram from Atg12^f/f and Atg12^KO MEFs in control media or following 2 h HBSS starvation. p62/SQSTM1 and LC3 immunoblotting on the same blot is shown below. bAtg12^f/f or Atg12^KO MEFs were switched to either control, low serum, glucose free, or glutamine-free media, or HBSS for 2 h. Cells were labeled with ³⁵S methionine for 30 min prior to lysis. Relative ³⁵S methionine incorporation rate is quantified, shown as a boxplot with dotplot overlay for each biological replicate, normalized to loading control. Statistical significance was assessed by ANOVA with Tukey’s HSD post hoc test, p = 1.0 for Atg12^f/f cells compared with Atg12^KO cells, in both control and HBSS. c, dAtg12^f/f and Atg12^KO MEFs were cultured in control, glucose free, or glutamine-free media or HBSS for 2 h. Addition of ³⁵S methionine and puromycin was included for 30 min prior to lysis. c Representative autoradiogram and immunoblot and (d) relative puromycin incorporation rate is quantified, displayed as a boxplot including each biological replicate, normalized to loading control; p > 0.95 for Atg12^f/f cells compared with Atg12^KO cells, in both control media and HBSS. e, fAtg12^f/f and Atg12^KO MEFs in control media or following 2 h HBSS starvation were lysed and immunoblotted for markers of cap-dependent protein translation inhibition (p-eIF2α). e Representative immunoblots are shown. f Relative protein levels of p-eIF2α were quantified, normalized to loading control, and shown as a boxplot including each biological replicate. Statistical analysis by ANOVA with Tukey’s HSD post hoc test. g, h Protein lysate from Atg12^f/f and Atg12^KO MEFs in control media or following 2 h HBSS starvation was subject to pulldown with γ-amino-phenyl-m7-GTP cap analog conjugated to agarose beads. g A representative immunoblot is shown and (h) 4EBP1 relative to eIF4G1 levels are quantified (mean + SD, n = 3 independent biological replicates), p value by t test. i Quantification (mean + SEM, n = 5 independent biological replicates, p value by t test) of Cricket paralysis virus IRES translation, normalized to cap translation rates. Cells were treated with PP242 (2 μM for 1 h) to inhibit mTORC1, and Thapsigargin (Tg, 1 μM for 1 h) to induce IRES-mediated translation (additional data in Supplementary Fig. 1f).

Fig. 2. Amino acid levels and mTORC1 activation not impaired by loss of ATG12.

a Changes in intracellular metabolite levels detected by GC-TOF with MTBSTFA (n = 4 biologically independent replicates), in Atg12^f/f and Atg12^KO MEFs in control media or following HBSS starvation for the indicated times. Fold change is relative to Atg12^f/f in control media. b–dAtg12^f/f and Atg12^KO MEFs in control media or following 2 h HBSS starvation were lysed and immunoblotted for markers of mTORC1 signaling (p-S6, p-4EBP1). b Representative immunoblot is shown. Relative protein levels of (c) p-S6 and (d) p-4EBP1 were quantified, normalized to loading control, and shown as a boxplot with dotplot overlay for each biological replicate. Statistical analysis by ANOVA with Tukey’s HSD post hoc test was performed and p values for Atg12^f/f samples compared with Atg12^KO samples in control media or HBSS are shown. e Representative immunoblot for markers of mTORC1 signaling (p-S6, p-4EBP1) in Atg12^f/f and Atg12^KO MEF protein lysate following a timecourse of HBSS starvation.

Fig. 3. Ribosome profiling reveals that autophagy supports the translation of proteins involved in DNA damage response and cell-cycle control.

a–c Violin plots of number of read counts of ribosome protected footprints (RPFs) per gene per biological replicate (above) and histogram of the mean of the number of read counts of ribosome protected footprints per gene (below) in (a) Atg12^f/f and Atg12^KO MEFs in control media (b) Atg12^f/f and Atg12^KO MEFs following 2 h HBSS starvation or (c) Atg12^f/f MEFs in control media or following 2 h HBSS starvation, p = 0.005 by t test. d–f Fold change of RPF counts versus fold change of mRNA counts. Labeled points in orange are mRNAs whose change in ribosome occupancy was significant, and protein level changes confirmed by immunoblotting (see Supplementary Fig. 3c). g, h Molecular functions of mRNAs whose ribosome occupancy is g increased (p < 0.01, n = 36 significant genes) or h decreased (p < 0.005, n = 60 significant genes) in Atg12^KO cells versus Atg12^f/f cells in either fed or starved conditions.

Fig. 4. Reduced BRCA2 protein translation in autophagy-deficient cells.

a, b Newly synthesized BRCA2 levels were assessed following 8 h AHA incorporation and BRCA2 immunoprecipitation from control or Atg12^KO HEK293T cells. a Representative immunoblot is shown. b Quantification (boxplot with dotplot overlay for each independent biological replicate) of newly synthesized BRCA2 (StrepHRP levels), p = 0.05 by t test. c, d Protein lysate was collected from Atg12^f/f and Atg12^KO MEFs in control media or following 2-h HBSS starvation. BRCA2 levels were measured by immunoblotting: c representative immunoblot is shown; d relative BRCA2 protein levels were normalized to loading control, and quantified shown as a boxplot with dotplot overlay for each biological replicate. Statistical analysis was performed by ANOVA with Tukey’s HSD post hoc test (p = 0.05 for control conditions). Comparing Atg12^f/f to Atg12^KO in HBSS, one outlier (Dixon test, p = 0.001) was excluded from statistical analysis, p = 0.05 by t test. e, f Protein lysate was collected from HEK293T cells with CRISPR deleted ATG12 and (e) immunoblotted for the indicated proteins; f relative BRCA2 protein levels normalized to loading control was quantified and shown as boxplot with dotplot overlay for each independent replicate, p = 0.003 by t test. gBrca2 levels (mean ± SD, n = 7 independent biological replicates) in Atg12^f/f and Atg12^KO MEFs were measured by qPCR, with Gapdh as the endogenous control. h, iAtg12^f/f and Atg12^KO MEFs following cycloheximide (100 μg/ml) treatment for time indicated to inhibit protein translation. h Representative immunoblots for BRCA2 and Mcl-1. i Quantification (mean ± SD, n = 3 independent biological replicates) of relative BRCA2 and Mcl-1 levels from immunoblotting, normalized to loading control. p values for cyclohexamide treatment between t = 0 and 6 h were calculated using ANOVA with Tukey’s post hoc test and are shown for Atg12^f/f and Atg12^KO, respectively.

Fig. 5. The 5′UTR of Brca2 determines translational sensitivity to autophagy due to structure complexity, requiring the helicase eIF4A1.

aAtg12^f/f and Atg12^KO MEFs were transfected with pcDNA3 expressing Gfp, Gfp preceded by the 5′UTR of Brca2, Gfp followed by the 3′UTR of Brca2, or Gfp flanked by both 5′ and 3′ UTRs of Brca2. Representative immunoblot for levels of GFP is shown, three independent biological replicates performed. bAtg12^f/f and Atg12^KO MEFs were transfected with pNL1.1 driving the expression of nano-luciferase, nano-luciferase preceded by the 5′UTR of Brca2, nano-luciferase followed by the 3′UTR of Brca2, or nano-luciferase flanked by both 5′ and 3′ UTRs of Brca2. Luciferase activity was measured by Nano-glo (Promega). Quantification shown (mean ± SEM, n = 4 independent biological replicates), statistics calculated by t test for biological replicates only. c Local minimum free energy (MFE) was predicted by RNALfold in the 5′UTRs from mRNAs with significantly lower than expected ribosome occupancy in Atg12^KO MEFs compared with a randomly generated gene set. A violin plot with boxplot overlay of the MFEs is shown, p = 0.05 by Kolmogorov–Smirnov test. d Quantification (mean ± SD, n = 6 independent biological replicates) of the fold enrichment of Brca2 interaction with eIF4A1 over IgG control in Atg12^f/f or Atg12^KO MEFs by RNA immunoprecipitation. p = 0.05 by t test. e Protein lysate from Atg12^f/f and Atg12^KO MEFs was captured by cap pulldown, and total protein lysate and pulldown was immunoblotted as indicated. f Quantification (mean ± SD, n = 4 independent biological replicates) of eIF4A1 capture by cap pulldown relative to eIF4G1 capture. p = 7.4E−08 by t test.

Fig. 6. eIF4A1 is sequestered by accumulated p62 in autophagy-deficient cells.

a Representative immunofluorescence images for eIF4A1 (red in merged imaged) and p62/SQSTM1 (green in merged imaged) in Atg12^f/f and Atg12^KO MEFs, nuclei were counterstained with Hoechst (blue). Yellow box indicates region of inset panel in the top left corner. Far right panels show the points of colocalization (white) of eIF4A1 in p62/SQSTM1. Bars = 50 μm. b Manders’ coefficient (percent of colocalization) of eIF4A1 in either NBR1 or p62/SQSTM1 in Atg12^f/f and Atg12^KO MEFs was calculated, and boxplot with dotplot overlay representing one field is shown (n = 3 biologically independent replicates). An outlier (Dixon test p = 0.03) was excluded from statistical analysis and the p values between Atg12^f/f and Atg12^KO were calculated by t test. c Representative immunoprecipitation of eIF4A1 and immunoblot for the autophagy cargo receptors p62/SQSTM1 and NBR1. Arrow indicates p62/SQSTM1, asterisk indicates immunoglobulin heavy chain. Immunoprecipitation was performed with three biologically independent replicates. d Protein lysate from Atg12^f/f and Atg12^KO MEFs that were knocked down for p62/SQSTM1 or treated with nontargeting shRNA was captured by cap pulldown, and the ratio of eIF4A1 to eIF4G1 was quantified (mean ± SD, n = 4 biologically independent replicates). e Protein lysate from Atg12^f/f and Atg12^KO MEFs that were knocked down for p62/SQSTM1 or treated with nontargeting shRNA or HEK293Ts that overexpress p62ΔLIR or empty vector control was captured by cap pulldown, and immunoblotted as indicated. f–h Protein lysate from Atg12^f/f or Atg12^KO MEFs stably infected with shRNA to p62/SQSTM1 or nontargeting control was collected. Relative quantification from immunoblots for f p62/SQSTM and g BRCA2, normalized to loading control are shown in the boxplot with dotplot overlay per biological replicate. p values were calculated using ANOVA with Tukey’s post hoc test and significant values are reported. h The change in BRCA2 levels between Atg12^f/f and Atg12^KO in shNT-treated cells or shp62-treated cells is plotted, p = 0.76 between shNT and shp62 for ΔBRCA2, calculated by t test.

Fig. 7. Increased MSI1 in autophagy-deficient cells impairs Brca2 translation efficiency.

a Quantification (mean ± SD, n = 4 biologically independent replicates) of the fold enrichment of Brca2 interaction with MSI1 over IgG control in Atg12^f/f or Atg12^KO MEFs by RNA immunoprecipitation. An outlier (Dixon test p = 0.007) was excluded from statistical analysis, and the p value by t test is shown. b Boxplot, with dotplot overlay for each biological replicate, of relative MSI1 protein levels normalized to loading control and representative immunoblots from autophagy-inhibited MEFs, assayed by immunoblotting. Statistical analysis was performed by t test. c Representative immunoblot of immunoprecipitation of myc-tagged overexpressed LC3 family members interaction with MSI1 in HEK293T cells. Immunoprecipitation was performed with three biologically independent replicates. d Protein lysate was collected from Atg12^f/f MEFs treated with shRNA to Msi1, and immunoblotted as indicated. Knockdown of ~50% was consistent among three independent biological replicates. e–g Protein lysate was collected from Atg12^f/f MEFs that were stably knocked down for MSI1 and subsequently treated with 4OHT or control, and assayed by immunoblotting. e Representative immunoblot is shown. f Quantification of relative BRCA2 protein levels normalized to loading control, are plotted as a boxplot with dotplot overlay for each independent biological replicate, p values by ANOVA with Tukey’s post hoc test. g Difference in BRCA2 steady state protein levels (ΔBRCA) between Atg12^f/f and Atg12^KO MEFs following MS1 depletion, p value by t test.

Fig. 8. Decreased BRCA2 levels in Atg12-deleted cells result in increased DNA damage and centrosome abnormalities.

a Immunofluorescence for γH2AX (red) in Atg12^f/f and Atg12^KO MEFs; nuclei are counterstained with Hoechst (blue). Bar = 50 μm. Percent of γH2AX-positive cells was quantified in Atg12^f/f and Atg12^KO MEFs; three random fields were counted over three independent biological replicates, p = 0.002 by t test. b Immunofluorescence for γH2AX (red) and 53BP1 (green) in Atg12^f/f and Atg12^KO MEF; nuclei are counterstained with Hoechst (blue). Bar = 50 μm. Three biologically independent replicates were performed. c ROS-glo assay (Promega) in Atg12^f/f and Atg12^KO MEFs treated with vehicle control or menadione (50 μM for 2 h, positive control) (mean ± SEM, n = 2 biologically independent replicates). d Protein lysate was collected from Atg12^f/f and Atg12^KO MEFs treated with vehicle control or NAC (5 mM for 8 h), or ectopically overexpressing either GFP (pGFP) or human BRCA2 (huBRCA2). A representative immunoblot for γH2AX is shown, as well as boxplots with independent biological replicates, for γH2AX levels normalized to loading control. Statistical analysis was performed by t test. e Protein lysate was collected from Atg12^f/f and Atg12^KO MEFs treated for 16 h with vehicle control, rucaparib (100 nM), olaparib (100 nM), or BMN (2 nM), and immunoblotted as shown. Three independent biological replicates were performed. f A clonogenic replating assay was performed on Atg12^f/f or Atg12^KO MEFs treated for 16 h with vehicle control, rucaparib (100 nM), olaparib (100 nM), or BMN (2 nM). Colony number is shown as a boxplot including biological replicates, p value by t test, g Immunofluorescence of centrosomes (γ-tubulin, red) and mitotic cells (pH3, green), nuclei counterstained by Hoechst (blue), in Atg12^f/f and Atg12^KO MEFs. Yellow box indicates magnified region below. White arrows indicate non-mitotic cells with multiple centrosomes (3+) or non-clustered centrosomes. Bar = 100 μm. h Quantification (mean ± SEM, n = 3 biologically independent replicates) of distance between mother and daughter centrosomes measured on immunofluorescence of γ-tubulin in non-pH3 positive cells, p value by t test. i Quantification of Atg12^f/f and Atg12^KO MEFs with abnormal numbers (3+) of centrosomes from immunofluorescence images (n = 4 biologically independent replicates). p values by t test on logit transformed percent per replicate. Fraction above the bar plots indicates number of cells with 3+ centrosomes out of number of cells enumerated.

Fig. 9. Atg12 deletion in vivo leads to reduced BRCA2 and increased DNA damage.

a Diagram of Atg12^f/f;Cag-Cre^ER+ mouse treatment and tissue collection. b Protein lysate was collected from tissues two weeks following vehicle or 0.2 mg/g tamoxifen treatment and immunoblotted for markers of autophagic flux (p62/SQSTM1, LC3). c Representative images of male and female Atg12^f/f and Atg12^KO littermates. d Body weights of male mice following Atg12 deletion (Atg12^f/fn = 17, Atg12^KOn = 14). e Protein lysate was collected from mouse tissues 10 weeks following vehicle or tamoxifen treatment, and immunoblotted for BRCA2. f Boxplot with dotplot overlay for biological replicates of BRCA2 protein levels, normalized to total protein levels, assayed by immunoblotting. p = 0.02 by t test. g Boxplot with dotplot overlay for biological replicates of percent of γH2AX-positive nuclei by immunofluorescence, counted over four randomly selected fields of stained tissue per a minimum of four mice. p value calculated by t test. h Representative immunofluorescence for γH2AX (red) in mouse tissues from the cerebral cortex and small intestine, with nuclei counterstained by Hoechst (blue). Bar = 50 μm.