Identification of PPM1D as an essential Ulk1 phosphatase for genotoxic stress‐induced autophagy

Abstract

Autophagy is an evolutionary conserved process that degrades subcellular constituents. Unlike starvation‐induced autophagy, the molecular mechanism of genotoxic stress‐induced autophagy has not yet been fully elucidated. In this study, we analyze the molecular mechanism of genotoxic stress‐induced autophagy and identify an essential role of dephosphorylation of the Unc51‐like kinase 1 (Ulk1) at Ser⁶³⁷, which is catalyzed by the protein phosphatase 1D magnesium‐dependent delta isoform (PPM1D). We show that after exposure to genotoxic stress, PPM1D interacts with and dephosphorylates Ulk1 at Ser⁶³⁷ in a p53‐dependent manner. The PPM1D‐dependent Ulk1 dephosphorylation triggers Ulk1 puncta formation and induces autophagy. This happens not only in mouse embryonic fibroblasts but also in primary thymocytes, where the genetic ablation of PPM1D reduces the dephosphorylation of Ulk1 at Ser⁶³⁷, inhibits autophagy, and accelerates apoptosis induced by X‐ray irradiation. This acceleration of apoptosis is caused mainly by the inability of the autophagic machinery to degrade the proapoptotic molecule Noxa. These findings indicate that the PPM1D–Ulk1 axis plays a pivotal role in genotoxic stress‐induced autophagy.

Introduction

Autophagy is a catabolic process in which cellular contents are digested 1. The molecular basis of autophagy has been extensively studied mainly in the setting of the starvation‐induced type, in which functional complexes containing autophagy‐related (Atg) proteins drive for the formation of autophagosomes 2. Unc‐51‐like kinase 1 (Ulk1) is a serine/threonine kinase that forms the Ulk1 complex together with Fip200, Atg13, and Atg101 2, 3. The Ulk1 complex plays a role in the initial step of autophagy by recruiting downstream Atg proteins. In healthy conditions, Ulk1 is phosphorylated and inactivated by mammalian target of rapamycin complex 1 and AMP‐activated protein kinase at different serine/threonine residues 4, 5, 6. After nutrient starvation, Ulk1 translocates to pre‐autophagosomal membranes, where it activates the Vps34 complex to generate omegasomes, which are the initial platforms of the isolation membrane.

Autophagy can be induced not only by starvation but also by several stressors, including genotoxic stress 7, 8. Unlike starvation‐induced autophagy, the molecular mechanisms of genotoxic stress‐induced autophagy have not yet been fully elucidated. Genotoxic stress has been reported to induce the production of several autophagy‐inducing molecules, including DRAM, TSC2, and AMPK 7, 8. Others and we have also demonstrated the transcriptional induction of Ulk1 in response to DNA damage 9, 10. However, the mechanism by which the upregulation of these molecules induces autophagy has not been elucidated yet. To clarify the initial step of genotoxic stress‐induced autophagy, we approached this question from a different viewpoint and investigated the phosphorylation status of Ulk1 because this largely affects its activity.

In this study, we examined the molecular mechanisms of genotoxic stress‐induced autophagy and identified a key molecule, namely protein phosphatase 1D magnesium‐dependent delta isoform [PPM1D; also called Wip1 and PP2C 11, 12, 13]. PPM1D plays an essential role at the beginning of genotoxic stress‐induced autophagy by dephosphorylating Ulk1. We also demonstrated that PPM1D‐dependent Ulk1 dephosphorylation triggers Ulk1 puncta formation and induces autophagy after etoposide treatment. In thymocytes, PPM1D/Ulk1‐dependent autophagy suppressed genotoxic stress‐induced apoptosis by degrading the proapoptotic molecule Noxa. These findings indicate that the PPM1D‐Ulk1 axis plays a pivotal role in genotoxic stress‐induced cellular responses.

Results and Discussion

To clarify the initial step of genotoxic stress‐induced autophagy, we investigated the phosphorylation status of Ulk1, because Ulk1 is crucial at the beginning of starvation‐induced autophagy and is regulated by its phosphorylation. Upon genotoxic stress, the tumor suppressor p53 acts as a master regulator of various biological responses, including apoptosis and autophagy 7, 8, and hence, we speculated that p53 is involved in Ulk1 phosphorylation. We therefore compared Ulk1 phosphorylation between p53‐null mouse embryonic fibroblasts (p53^−/− MEFs) and its littermate wild‐type MEFs (p53^+/+ MEFs) after the addition of etoposide, a topoisomerase II inhibitor. As expected, we found a time‐dependent mobility shift in Ulk1 upon SDS–PAGE of cell lysates of etoposide‐treated p53^+/+ MEFs, but not of Ulk1 from p53^−/− MEFs (Fig 1A). This suggests that the dephosphorylation of Ulk1 proceeds in a p53‐dependent manner. Given that mouse Ulk1 is phosphorylated at Ser⁶³⁷ (corresponding to Ser⁶³⁸ of human Ulk1) and at Ser⁷⁵⁷ (corresponding to Ser⁷⁵⁸ of human Ulk1) by mTORC1 in healthy conditions, we next examined alterations of the phosphorylation status of these amino acid residues. We found that etoposide treatment induces a reduction in the level of Ulk1 phosphorylation at Ser⁶³⁷ (phospho‐Ulk1⁶³⁷) in p53^+/+ MEFs, whereas the phosphorylation level was not changed in p53^−/− MEFs (Figs 1A and EV1). This suggests that p53‐dependent phosphatases target Ser⁶³⁷ after etoposide treatment. Although the mobility of phospho‐Ulk1⁶³⁷ was still observed, this might be because of the presence of other unidentified Ulk1 modifications. The difference in the initial level of phospho‐Ulk1⁶³⁷ between p53^+/+ MEFs and p53^−/− MEFs might be caused by a decrease in Ulk1 kinase activity because of p53 deficiency. Unlike phospho‐Ulk1⁶³⁷, the extent of Ulk1 phosphorylation at Ser⁷⁵⁷ (phospho‐Ulk1⁷⁵⁷) was not altered in both types of cells (Figs 1A and EV1).

Figure 1. Involvement of PPM1D in Ulk1 dephosphorylation and autophagy induction.

• A
  p53‐dependent dephosphorylation of Ulk1 at Ser⁶³⁷. p53^+/+ and p53^−/− MEFs were treated with etoposide (10 μM), and the expression of each protein was examined by Western blotting. “Dead cells (%)” indicates the population of apoptotic cells assessed by propidium iodide (PI) staining. α‐Tubulin was used as a loading control. A semiquantitative analysis of protein expression is shown in Fig EV1.
• B, C
  PPM1D‐dependent dephosphorylation of Ulk1 at Ser⁶³⁷ and the induction of autophagy. PPM1D^+/+ and PPM1D^−/− MEFs were treated with etoposide (10 μM), and the expression of each protein was examined by Western blotting. “Dead cells (%)” indicate the population of apoptotic cells. (C) Semiquantitative analysis of protein expression is shown (n = 3; mean ± SD). *P < 0.05, **P < 0.01, NS: not significant. One‐way ANOVA followed by Tukey's post hoc test.
• D–I
  Suppression of etoposide‐induced autophagy in PPM1D^−/− MEFs. (D, E) The indicated MEFs were treated with etoposide (10 μM) for 6 h and then immunostained with an anti‐LC3 antibody (green). Representative images are shown in (D). LC3 puncta are observed in etoposide‐treated PPM1D^+/+ MEFs. (E) The proportion of cells with LC3 puncta was calculated (n > 100 cells in each experiment). Data are shown as the mean ± SD (n = 3 experiments). *P < 0.05, **P < 0.01. (F, G) The indicated MEFs were treated with etoposide (10 μM) for 6 h and then analyzed using electron microscopy. Representative images are shown in (F). Many autophagic vacuoles (arrows) can be seen in etoposide‐treated PPM1D^+/+ cells (upper panel). “N” indicates the nucleus. Bar = 2 μm. A representative autophagosome (AP) and autolysosome (AL) are shown in the insets. (G) The number of autophagosomes and autolysosomes in each cell were counted (n > 8 cells). Red lines indicate the mean value. *P < 0.05. (H, I) Analysis of autophagic flux. The indicated MEFs were treated with etoposide (10 μM) for 6 h in the presence or absence of bafilomycin A1 (10 nM), and the expression of each protein was examined by Western blotting. α‐Tubulin was included as a loading control. (I) Semiquantitative analyses of protein expression (n = 3; mean ± SD). *P < 0.05, **P < 0.01, NS: not significant. One‐way ANOVA followed by Tukey's post hoc test.

Source data are available online for this figure.

Figure EV1. p53‐dependent dephosphorylation of Ulk1 at Ser⁶³⁷ .

Semiquantitative analyses of protein expression shown in Fig 1A are shown (n = 3; mean ± SD). *P < 0.05, **P < 0.01, NS: not significant. One‐way ANOVA followed by Tukey's post hoc test.

To elucidate the molecule responsible for the etoposide‐induced reduction in phospho‐Ulk1⁶³⁷ levels, we focused on PPM1D, because it appeared to be the most likely candidate phosphatase targeting phospho‐Ulk1⁶³⁷, as it is activated in a p53‐dependent manner after genotoxic stress 11, 13. As expected, the level of PPM1D was increased in a time‐dependent manner in p53^+/+ MEFs, but not in p53^−/− MEFs, after etoposide treatment (Figs 1A and EV1). The expression levels of PPM1D changed in a reverse parallel manner to those of phospho‐Ulk1⁶³⁷, suggesting the involvement of PPM1D in Ulk1 dephosphorylation at Ser⁶³⁷.

To elucidate the involvement of PPM1D in the reduction in phospho‐Ulk1⁶³⁷ levels, we created PPM1D‐null (PPM1D^−/−) MEFs and control (PPM1D^+/+) MEFs from each embryo at day 14.5. In etoposide‐treated PPM1D^+/+ MEFs, we observed a time‐dependent mobility shift of Ulk1, a reduction in phospho‐Ulk1⁶³⁷ levels, and an increase in PPM1D (Fig 1B and C) to levels similar to those seen in p53^+/+ MEFs (Fig 1A). In contrast, none of these changes were observed in PPM1D^−/− MEFs (Fig 1B and C). The level of phospho‐Ulk1⁷⁵⁷ was not altered, irrespective of the presence of PPM1D (Fig 1B and C). In parallel with the increase in PPM1D and the reduction in phospho‐Ulk1⁶³⁷, autophagy was activated in etoposide‐treated PPM1D^+/+ MEFs, as assessed by the degradation of the p62 protein and the formation of LC3‐II (Fig 1B and C), both of which are well‐recognized autophagy markers. In contrast, autophagy was less activated in etoposide‐treated PPM1D^−/− MEFs (Fig 1B and C). Consistent results were obtained when autophagy was analyzed by LC3 puncta formation (Fig 1D and E) and electron microscopic (EM) analysis (Fig 1F and G). The effect of PPM1D on etoposide‐induced autophagy was further confirmed by examining autophagic flux, which is the dynamic process of autophagy. Autophagic flux can be measured by differences in the levels of p62 or LC3‐II between the presence and absence of lysosomal degradation inhibitors, such as bafilomycin A1. A significant difference, which indicates the activation of autophagic flux, was observed in etoposide‐treated PPM1D^+/+ MEFs (Fig 1H and I). In PPM1D^−/− MEFs, the difference was far smaller than that seen in PPM1D^+/+ MEFs, indicating a weaker induction of autophagic flux (Fig 1H and I). Although etoposide induces both autophagy and apoptosis, we finished the autophagy assay before the acceleration of apoptosis (see Fig 1A and B: Dead cells (%)), so that the effect of apoptosis could be disregarded. These results indicated that etoposide induced PPM1D accumulation, reduced phospho‐Ulk1⁶³⁷ levels, and activated autophagy in a p53‐dependent manner.

To confirm the involvement of PPM1D in the reduction of phospho‐Ulk1⁶³⁷ levels and the activation of autophagy by etoposide treatment, we stably expressed Flag‐tagged PPM1D in PPM1D^−/− MEFs. The exogenous expression of PPM1D in PPM1D^−/− MEFs restored the reduction in phospho‐Ulk1⁶³⁷ levels as well as the autophagic activation induced by etoposide (Fig 2A and B: p‐Ulk1⁶³⁷, p62, and LC3). Furthermore, when we applied GSK2830371, a recently developed specific inhibitor of PPM1D 14, to PPM1D^+/+ MEFs, we observed no reduction in phospho‐Ulk1⁶³⁷ levels (Fig 2C and D) and only a weak induction of autophagy (Fig 2C and D: p62 and LC3) that was similar to that seen in PPM1D^−/− MEFs (Fig 1B). GSK2830371 has been reported to suppress PPM1D expression 14, and this was confirmed in our Western blot analysis (Fig 2C and D). The involvement of PPM1D was further confirmed by the results that the transient overexpression of PPM1D reduced phospho‐Ulk1⁶³⁷ levels (Fig 2E and F) and induced autophagic flux (Fig 2E and F: p62 and LC3) without etoposide treatment, indicating that PPM1D is sufficient to induce autophagy. Although the expression level of PPM1D in Flag‐PPM1D‐overexpressing PPM1D^+/+ MEFs (Fig 2E) was much higher than that in Flag‐PPM1D stably transfected PPM1D^−/− MEFs (Fig 2A), this was due to the transient overexpression of PPM1D via electrophoretic transfection in the PPM1D^+/+ MEFs. Collectively, our results indicate that PPM1D is involved in the reduction of phospho‐Ulk1⁶³⁷ levels and the progression of autophagy induced by etoposide.

Figure 2. Confirmation of PPM1D involvement in Ulk1 dephosphorylation and the induction of autophagy.

• A, B
  Effect of stable expression of PPM1D in PPM1D^−/− MEFs. PPM1D^−/− MEFs and PPM1D‐transfected PPM1D^−/− MEFs were treated with etoposide (10 μM), and the expression of each protein was examined by Western blotting. (B) Semiquantitative analyses of protein expression in (A) (n = 3; mean ± SD). *P < 0.05, **P < 0.01, NS: not significant. One‐way ANOVA followed by Tukey's post hoc test.
• C, D
  Effect of a PPM1D inhibitor on etoposide‐treated PPM1D^+/+ MEFs. PPM1D^+/+ MEFs were treated with etoposide (10 μM) in the presence or absence of GSK2830371 (30 μM), and the expression of each protein was examined by Western blotting. (D) Semiquantitative analysis of protein expression in (C) (n = 3; mean ± SD). *P < 0.05, **P < 0.01, NS: not significant. One‐way ANOVA followed by Tukey's post hoc test.
• E, F
  Effect of the transient overexpression of PPM1D. PPM1D^+/+ MEFs were transfected with a PPM1D‐Flag plasmid in the presence or absence of bafilomycin A1 (10 nM), and the expression of each protein was examined by Western blotting. (F) Semiquantitative analyses of protein expression in (E) (n = 3; mean ± SD). *P < 0.05, **P < 0.01, NS: not significant. One‐way ANOVA followed by Tukey's post hoc test.

Source data are available online for this figure.

We next examined whether PPM1D binds and dephosphorylates Ulk1 at Ser⁶³⁷. To this end, we first performed an immunoprecipitation assay. As shown in Fig 3A, endogenous PPM1D was co‐immunoprecipitated with an anti‐Ulk1 antibody in lysates from etoposide‐treated PPM1D^+/+ MEFs (Fig 3A: lane 8). This interaction was not observed when lysates from healthy cells were used (Fig 3A: lane 7), but this is merely because of the low PPM1D expression level in these cells (Fig 3A: lane 3). When glutathione S‐transferase (GST)‐fused PPM1D was added to healthy cell lysates, we observed a physical interaction between endogenous Ulk1 and GST‐PPM1D (Fig 3B). Functionally, recombinant GST‐PPM1D efficiently reduced the level of phospho‐Ulk1⁶³⁷ in PPM1D^−/− MEF lysates (Fig EV2A). Consistently, when immunoprecipitated Ulk1 was incubated with GST‐PPM1D, we observed a reduction in phospho‐Ulk1⁶³⁷ (Figs 3C and EV2B: lanes 1 and 2). The level of phospho‐Ulk1⁶³⁷ was restored by the further addition of a phosphatase inhibitor cocktail (Figs 3C and EV2B: lanes 2 and 3), indicating that PPM1D directly binds and dephosphorylates Ulk1 at Ser⁶³⁷. Unlike phospho‐Ulk1⁶³⁷, the level of phospho‐Ulk1⁷⁵⁷ was not altered by the incubation with PPM1D (Figs 3C and EV2B). PPM1D primarily localizes to the nucleus after genotoxic stress and dephosphorylates several nuclear proteins 11, 13. We observed small cytoplasmic puncta mainly in the juxtanuclear region of etoposide‐treated PPM1D^+/+ MEFs (Fig 3D, arrowheads); importantly, some of them were co‐localized with Ulk1 puncta (Fig 3E: arrowheads). Consistent results were observed when Ulk1/Ulk2 double‐knockout (DKO) MEFs expressing HA‐Ulk1 were treated with etoposide (Fig EV2C). These co‐localized puncta are likely to be where PPM1D dephosphorylates Ulk1. The lack of PPM1D immunofluorescence signals in healthy MEFs (Fig 3D) and etoposide‐treated PPM1D^−/− MEFs (Fig EV2D) validates our experiment.

Figure 3. Interaction between Ulk1 and PPM1D and role of phospho‐Ulk1⁶³⁷ in autophagy.

• A
  Physical interaction between endogenous Ulk1 and endogenous PPM1D. PPM1D^+/+ MEFs were treated with etoposide (10 μM) or left untreated for 6 h. Cells were then lysed and immunoprecipitated with an anti‐Ulk1 antibody or a control IgG. Immune complexes and total lysates (1.8% input) were analyzed by Western blotting using anti‐PPM1D and anti‐Ulk1 antibodies.
• B
  Interaction between endogenous Ulk1 and recombinant PPM1D. Lysates from healthy PPM1D^+/+ MEFs were incubated with GST‐PPM1D or GST. Binding molecules were then analyzed by Western blotting using anti‐Ulk1 and anti‐GST antibodies. “Total lysate” indicates 7% of the lysates that were incubated with GST fusion protein.
• C
  In vitro Ulk1 dephosphorylation assay. Immunoprecipitated Ulk1 from PPM1D^+/+ MEF lysates was incubated with GST‐PPM1D (1 μg) with or without a phosphatase inhibitor cocktail for 1 h. Then, the extent of Ulk1 dephosphorylation was analyzed by Western blotting using anti‐phospho‐Ulk1⁶³⁷ and anti‐phospho‐Ulk1⁷⁵⁷ antibodies. Semiquantitative analyses are shown in Fig EV2B.
• D
  The presence of cytosolic PPM1D puncta in etoposide‐treated PPM1D^+/+ MEFs. PPM1D^+/+ MEFs were treated with etoposide (10 μM) for 6 h, immunostained with an anti‐PPM1D antibody, and their nuclei stained with Hoechst 33342 (50 ng/ml). Representative images of anti‐PPM1D (green; upper panels) and Hoechst 33342 (blue; lower panels) are shown. Arrowheads indicate cytoplasmic PPM1D puncta.
• E
  Colocalization of endogenous PPM1D and endogenous Ulk1 in etoposide‐treated PPM1D^+/+ MEFs. PPM1D^+/+ MEFs were treated with etoposide (10 μM) for 6 h, immunostained with anti‐PPM1D and anti‐Ulk1 antibodies, and their nuclei stained with Hoechst 33342 (50 ng/ml). Representative images of anti‐PPM1D (red; left), anti‐Ulk1 (green; middle), and a merged image (right) are shown. Magnified images of the areas within the dashed squares are shown at the bottom. Arrowheads indicate cytoplasmic PPM1D colocalized with Ulk1.
• F–J
  The crucial role of Ulk1 dephosphorylation at Ser⁶³⁷ in etoposide‐induced autophagy. (F–H) Ulk1/Ulk2 DKO MEFs that were stably transfected with HA‐Ulk1 or its mutants, S637D and S637A, were treated with etoposide (10 μM). Then, the cell lysates were collected time‐dependently, and the expression of each protein was examined by Western blotting. Asterisks in the Ulk1 blots are nonspecific bands. Semiquantitative analyses are shown in Fig EV2E. (I, J) The indicated MEFs were treated with or without etoposide (10 μM) for 6 h, followed by immunostaining with an anti‐LC3 antibody. Representative images are shown in (I). LC3 puncta are markedly observed in DKO MEFs transfected with wild‐type HA‐Ulk1. Puncta were absent and weakly observed in MEFs expressing the mutants S637D and S637A, respectively. (J) The population of cells with LC3 puncta was calculated (n > 100 cells in each experiment). Data are shown as the mean ± SD (n = 3 experiments). *P < 0.05, **P < 0.01. One‐way ANOVA followed by Tukey's post hoc test.

Source data are available online for this figure.

Figure EV2. Role of PPM1D and phospho‐Ulk1⁶³⁷ on autophagy.

• A
  Ulk1 dephosphorylation assay. PPM1D^−/− MEF lysates were incubated with GST‐PPM1D (1 μg) for 1 h. Then, the extent of Ulk1 dephosphorylation was analyzed by Western blotting using anti‐Ulk1, anti‐phospho‐Ulk1⁶³⁷, and anti‐phospho‐Ulk1⁷⁵⁷ antibodies.
• B
  Semiquantitative analyses of phosphorylated Ulk1 expression shown in Fig 3C are shown (n = 3; mean ± SD). *P < 0.05, **P < 0.01, NS: not significant. One‐way ANOVA followed by Tukey's post hoc test.
• C
  The colocalization of endogenous PPM1D and HA‐Ulk1 in etoposide‐treated Ulk1/Ulk2 double‐knockout (DKO) MEFs. Ulk1/Ulk2 DKO MEFs transfected with HA‐Ulk1 were treated with etoposide (10 μM) for 6 h, then immunostained with anti‐PPM1D and anti‐HA antibodies. Representative images of anti‐PPM1D (red; left), anti‐HA (green; middle), and a merged image (right) are shown. The nucleus was stained with Hoechst 33342 (50 ng/ml) in the merged image. Magnified images of the areas within the dashed squares are shown on the right. Arrowheads indicate cytoplasmic PPM1D colocalized with HA‐Ulk1.
• D
  Lack of PPM1D staining in PPM1D^−/− MEFs. PPM1D signals were not observed in the nuclei of etoposide‐treated PPM1D^−/− MEFs, despite such signals being seen in the nuclei of etoposide‐treated PPM1D^+/+ MEFs (see Fig 3D), validating the successful immunostaining of PPM1D.
• E
  Semiquantitative analyses of protein expression shown in Fig 3F–H are shown (n = 3; mean ± SD). *P < 0.05, **P < 0.01, NS: not significant. One‐way ANOVA followed by Tukey's post hoc test.
• F, G
  Analysis of etoposide‐induced autophagic flux in Ulk1/Ulk2 DKO MEFs. Ulk1/Ulk2 DKO MEFs stably transfected with HA‐Ulk1 (WT) or its mutants, S637D and S637A, were treated with etoposide (10 μM) for 6 h in the presence or absence of bafilomycin A1 (10 nM). The expression of each protein was examined by Western blotting. α‐Tubulin was included as a loading control. In (G), semiquantitative analyses of protein expression are shown (n = 3; mean ± SD). *P < 0.05, NS: not significant. One‐way ANOVA followed by Tukey's post hoc test.
• H
  Equivalent expression levels of the HA‐Ulk1 mutants. Ulk1/Ulk2 DKO MEFs were stably transfected with expression plasmids for HA‐Ulk1 or its mutants, S637D and S637A. Then, cells were lysed, and the expression levels of the Ulk1 mutants were examined by Western blotting.

Source data are available online for this figure.

Given that PPM1D dephosphorylates Ulk1 and activates autophagy after etoposide treatment, we examined the causal relationship between the level of phospho‐Ulk1⁶³⁷ and autophagy. To this end, we expressed HA‐Ulk1, a phosphomimetic mutant (S637D), and a phosphodeficient mutant (S637A) in Ulk1/Ulk2 DKO MEFs. We used DKO MEFs to avoid compensation of the Ulk1 deficit by Ulk2, as these cells were reported as autophagy‐resistant cells 15. We in fact did not observe etoposide‐induced autophagy in these DKO MEFs, as assessed by the degradation of p62 and the generation of LC3‐II (Figs 3F–H and EV2E), whereas it was restored by the expression of HA‐Ulk1 (Figs 3F and EV2E). Restoration of etoposide‐induced autophagy by HA‐Ulk1 was confirmed by the observation of LC3 puncta (Fig 3I and J) and the examination of autophagic flux (Fig EV2F and G). Importantly, the reduction in phospho‐Ulk1⁶³⁷ levels was also observed in exogenously expressed HA‐Ulk1 (Fig 3F). As expected, expression of the HA‐Ulk1 S637D mutant did not restore autophagy, as assessed by p62 and LC3 protein expression (Figs 3G and EV2E), LC3 puncta formation (Fig 3I and J), and autophagic flux (Fig EV2F and G), despite being expressed at levels similar to HA‐Ulk1 (Fig EV2H). These data indicated the crucial role of phospho‐Ulk1⁶³⁷ dephosphorylation in etoposide‐induced autophagy. Expression of the HA‐Ulk1 S637A mutant showed only a weak effect on etoposide‐induced autophagy (Figs 3H–J and EV2E–H), suggesting that a dynamic dephosphorylation process may be required for the full induction of autophagy. Note that PPM1D levels were increased to similar extents by etoposide, irrespective of the type of Ulk1 mutation at Ser⁶³⁷ (Fig 3F–H). All these data indicated that Ulk1 dephosphorylation at Ser⁶³⁷ is crucial for etoposide‐induced autophagy.

We further studied the detailed mechanism as to how PPM1D regulates autophagy through Ulk1 dephosphorylation. Because Ulk1 puncta formation is one of the initial steps of autophagy 16, we examined the subcellular localization of endogenous Ulk1. Ulk1 was localized diffusely throughout the cytosol in healthy PPM1D^+/+ MEFs, and cytosolic puncta formation was observed from 4.5 h after etoposide treatment (Fig 4A and B). This puncta formation was largely suppressed in PPM1D^−/− MEFs (Fig 4A and C) and GSK2830371‐treated PPM1D^+/+ MEFs (Fig 4A and D), indicating that PPM1D is required for etoposide‐induced Ulk1 puncta formation. Similar results were obtained when HA‐Ulk1‐expressing Ulk1/2 DKO MEFs were used (Fig EV3A–C). Importantly, Ulk1 puncta formation was largely dependent on Ulk1 phosphorylation at Ser⁶³⁷; the number of cells with HA‐Ulk1 puncta was increased time‐dependently in the etoposide‐treated DKO MEFs transfected with HA‐Ulk1, whereas no and weak Ulk1 puncta formation were observed in the MEFs transfected with the S637D and the S637A mutant, respectively (Fig 4E and F). The proportion of cells with Ulk1 puncta (Fig 4F) correlated with the extent of autophagy (Fig 3J). These results indicated that Ulk1 dephosphorylation at Ser⁶³⁷ is crucial for etoposide‐induced Ulk1 puncta formation.

Figure 4. Involvement of PPM1D in Ulk1 activity and Ulk1/DFCP1 puncta formation.

• A–D
  Ulk1 puncta formation was induced by etoposide in a PPM1D‐dependent manner. The indicated MEFs were treated with etoposide (10 μM) with or without GSK2830371 (20 μM) for the indicated times, followed by immunostaining with an anti‐Ulk1 antibody. Representative images are shown in (A). Ulk1 puncta are observed in etoposide‐treated PPM1D^+/+ MEFs. (B–D) The proportion of cells with Ulk1 puncta was calculated (n > 100 cells in each experiment). Data are shown as the mean ± SD (n = 3 experiments). *P < 0.05, **P < 0.01, NS: not significant. One‐way ANOVA followed by Tukey's post hoc test.
• E–I
  Role of the dephosphorylation of Ulk1 at Ser⁶³⁷ on etoposide‐induced Ulk1 puncta formation and Atg13 phosphorylation. Ulk1/Ulk2 DKO MEFs stably transfected with HA‐Ulk1 or its mutants, S637D and S637A, were treated with etoposide (10 μM) for the indicated times. (E, F) Cells were immunostained with an anti‐Ulk1 antibody (E), and the population of cells with Ulk1 puncta was calculated (n > 100 cells in each experiment) (F). Data are shown as the mean ± SD (n = 3 experiments). *P < 0.05, **P < 0.01. One‐way ANOVA followed by Tukey's post hoc test. (G–I) Cell lysates were collected in a time‐dependent manner, and Atg13 protein levels and its phosphorylation at Ser³¹⁷ were examined by Western blotting. Semiquantitative analyses are shown in Fig EV3D.
• J
  EGFP‐DFCP1‐expressing PPM1D^+/+ MEFs were treated with etoposide (10 μM) for the indicated times, followed by immunostaining with an anti‐Ulk1 antibody. Representative images of EGFP‐DFCP1 (green; left), anti‐Ulk1 (red; center), and a merged image (right) are shown. The nuclei were stained with Hoechst 33342 in the merged image. A magnified image of the area within the dashed square is also shown. Arrowheads indicate DFCP1 puncta localized close to Ulk1 puncta.
• K, L
  The role of Ulk1 dephosphorylation at Ser⁶³⁷ on etoposide‐induced DFCP1 puncta formation. Ulk1/Ulk2 DKO MEFs stably transfected with HA‐Ulk1 or its mutants, S637D and S637A, were transfected with EGFP‐DFCP1 and treated with etoposide (10 μM) for the indicated hours, followed by immunostaining with an anti‐HA antibody. Representative merged images of EGFP‐DFCP1 (green), anti‐HA (red), and Hoechst 33342 (blue) are shown in (K). Arrowheads indicate DFCP1 puncta localized close to Ulk1 puncta. The number of DFCP1 puncta in each cell was calculated and is shown in (L) (n > 8 cells in each experiment). Red lines indicate the mean values. **P < 0.01, NS: not significant. One‐way ANOVA followed by Tukey's post hoc test.

Source data are available online for this figure.

Figure EV3. Role of phospho‐Ulk1⁶³⁷ on Ulk1 puncta formation and Atg13 phosphorylation.

1. Ulk1/Ulk2 DKO MEFs stably transfected with HA‐Ulk1 were treated with etoposide (10 μM) for the indicated times and then immunostained with an anti‐HA antibody. Representative images are shown. HA‐Ulk1 puncta can be seen in etoposide‐treated MEFs.
2. The population of cells with HA‐Ulk1 puncta was calculated (n > 100 cells in each experiment). Data are shown as the mean ± SD (n = 3 experiments). **P < 0.01. One‐way ANOVA followed by Tukey's post hoc test.
3. The same experiment was performed with or without GSK2830371 (20 μM). Data are shown as the mean ± SD (n = 3 experiments). **P < 0.01. Unpaired two‐tailed Student's t‐test.
4. Semiquantitative analyses of protein expression in Fig 4G–I are shown (n = 3; mean ± SD). *P < 0.05, **P < 0.01. One‐way ANOVA followed by Tukey's post hoc test.

To investigate whether Ulk1 dephosphorylation at Ser⁶³⁷ also regulates its kinase function, we examined the phosphorylation status of Atg13 at Ser³¹⁷, which is the target phosphorylation site of Ulk1 17. As shown in Figs 4G and EV3D, phospho‐Atg13³¹⁷ was not observed in Ulk1/2 DKO MEFs, whereas it was increased time‐dependently after etoposide treatment by the transfection of HA‐Ulk1. This increase was not observed by transfection of the HA‐Ulk1 mutants S637D and S637A (Figs 4H and I, and EV3D), indicating the crucial role of Ulk1 dephosphorylation at Ser⁶³⁷ on downstream Atg13 phosphorylation. Ulk1 dephosphorylation at Ser⁶³⁷ also facilitates the puncta formation of double‐FYVE‐containing protein 1 (DFCP1), which occurs downstream of Ulk1 activation 17. DFCP1 puncta were localized relatively close to Ulk1 puncta (Fig 4J), which is consistent with a previous report 18. As shown in Fig 4K and L, etoposide treatment increased the number of cells with DFCP1 puncta in DKO MEFs transfected with HA‐Ulk1, but DFCP1 puncta were only weakly observed upon transfection of the HA‐Ulk1 mutants S637D and S637A. Taken together, these results indicated that the PPM1D‐dependent dephosphorylation of Ulk1 at Ser⁶³⁷ is required for Ulk1 puncta formation, Atg13 phosphorylation, DFCP1 puncta formation, and the induction of autophagy by etoposide.

We next examined whether PPM1D and Ulk1 dephosphorylation at Ser⁶³⁷ are crucial for autophagy in general. Unlike etoposide treatment, the induction of autophagy after starvation did not increase PPM1D expression levels (Fig EV4A and B). Furthermore, levels of starvation‐induced autophagy did not differ between PPM1D^+/+ MEFs and PPM1D^−/− MEFs, as assessed by the autophagic flux (Fig EV4C–F), indicating that PPM1D is not involved in starvation‐induced autophagy. Despite this, we still observed a reduction in phospho‐Ulk1⁶³⁷ levels upon starvation, which was not dependent on the presence of PPM1D (Fig EV4G and H), suggesting the possible involvement of any phosphatases other than PPM1D in starvation‐induced autophagy. PP2A, which belongs to the PP2C family, has actually been reported as such a molecule 19. The crucial role of Ulk1 dephosphorylation at Ser⁶³⁷ was evident in starvation‐induced autophagy because a high level of autophagic flux was observed in Ulk1/Ulk2 DKO MEFs expressing HA‐Ulk1, but was largely suppressed in Ulk1/Ulk2 DKO MEFs expressing the S637D and S637A mutants (Fig EV4I–K). Note that these phospho‐mutant proteins were not misfolded, but were catalytically inactivated, because they were able to bind to Fip200 and Atg13 to a similar extent to wild‐type Ulk1 (Fig EV4L).

Figure EV4. Role of PPM1D and phospho‐Ulk1⁶³⁷ on starvation‐induced autophagy.

• A, B
  No increase in PPM1D levels in starved MEFs. PPM1D^+/+ MEFs were treated with etoposide (10 μM) or starved for the indicated times. The expression of each protein was examined by Western blotting. (B) Semiquantitative analysis of the protein expression in (A) (n = 3; mean ± SD). **P < 0.01, NS: not significant. One‐way ANOVA followed by Tukey's post hoc test.
• C–F
  PPM1D‐independent induction of autophagy by starvation. PPM1D^+/+ MEFs and PPM1D^−/− MEFs were starved for the indicated times in the presence or absence of bafilomycin A1 (10 nM), and the expression of each indicated protein was examined by Western blotting. (E, F) Semiquantitative analyses of protein expression in (C) and (D) (n = 3; mean ± SD). *P < 0.05, NS: not significant. One‐way ANOVA followed by Tukey's post hoc test.
• G, H
  Reduction of phospho‐Ulk1⁶³⁷ by starvation irrespective of the presence of PPM1D. PPM1D^+/+ MEFs and PPM1D^−/− MEFs were starved for 1 h, and the expression of phospho‐Ulk1⁶³⁷ was examined by Western blotting. (H) Semiquantitative analysis of protein expression in (G) (n = 3; mean ± SD). *P < 0.05, NS: not significant. One‐way ANOVA followed by Tukey's post hoc test.
• I–K
  Involvement of Ulk1 dephosphorylation at Ser⁶³⁷ in starvation‐induced autophagy. (I, J) Ulk1/Ulk2 DKO MEFs stably transfected with HA‐Ulk1 or its mutants, S637D and S637A, were starved for the indicated times in the presence or absence of bafilomycin A1 (10 nM). The expression of each indicated protein was examined by Western blotting. In (K), semiquantitative analyses are shown (n = 3; mean ± SD). *P < 0.05, NS: not significant. One‐way ANOVA followed by Tukey's post hoc test.
• L
  Physical interaction between Ulk1 mutants and endogenous FIP200 and Atg13. Ulk1/Ulk2 DKO MEFs stably transfected with HA‐Ulk1 or its mutants, S637D and S637A, were lysed and immunoprecipitated with an anti‐Ulk1 antibody. Immune complexes and total lysates (6% input) were analyzed by Western blotting using anti‐FIP200, anti‐Atg13, and anti‐Ulk1 antibodies.

Source data are available online for this figure.

To generalize the involvement of PPM1D in the dephosphorylation of Ulk1 at Ser⁶³⁷ and the progression of autophagy during genotoxic stress, we irradiated primary thymocytes isolated from PPM1D‐deficient mice and control littermate mice. PPM1D^−/− thymocytes showed only a slight developmental abnormality (Fig EV5A), as described 20. In PPM1D^+/+ thymocytes, we observed the induction of PPM1D, a reduction in phospho‐Ulk1⁶³⁷ levels, and the progression of autophagy (assessed by p62 degradation), and these findings were not observed in PPM1D^−/− thymocytes (Figs 5A and EV5B), which is consistent with our observations in etoposide‐treated MEFs. Although the results of LC3 Western blotting were ambiguous (Fig 5A), a lower induction of autophagy in PPM1D^−/− thymocytes than control thymocytes was confirmed when autophagy was assessed using LC3 puncta formation (Fig 5B and C), EM analysis (Fig 5D), Cyto‐ID^™ fluorescence (an autophagy marker; Figs 5E and EV5C), and an autophagic flux assay using chloroquine (Figs 5F and EV5D: p62 and LC3). Thus, these results indicate that PPM1D is also crucial for irradiation‐induced autophagy in thymocytes.

Figure EV5. Impaired autophagy and decreased cell viability in PPM1D‐deficient thymocytes after X‐ray irradiation.

1. Slightly abnormal development of PPM1D‐knockout thymocytes. The percentages of CD4⁻CD8⁻ cells, CD4⁺CD8⁺ cells, CD4⁺CD8⁻ cells, and CD4⁻CD8⁺ cells were measured using a flow cytometer (mean ± SD; n = 7). Thymocytes were stained with a PE‐conjugated rat anti‐mouse CD4 antibody and an FITC‐conjugated rat anti‐mouse CD8a antibody (BD Biosciences) for 30 min at 4 ˚C and then analyzed by flow cytometry using a FACS Canto II (BD Biosciences). *P < 0.05, NS, not significant. Unpaired two‐tailed Student's t‐test.
2. Semiquantitative analyses of protein expression in Fig 5A are shown (n = 3; mean ± SD). *P < 0.05, **P < 0.01, NS: not significant. One‐way ANOVA followed by Tukey's post hoc test.
3. The indicated thymocytes were X‐ray irradiated (5 Gy), and 3 h later, they were stained with Cyto‐ID^™. Then, the rates of increase in mean Cyto‐ID fluorescence intensities were calculated using data from the flow cytometric analysis (n = 5); increased ratio of mean ± SD. *P < 0.05, NS, not significant. One‐way ANOVA followed by Tukey's post hoc test. Representative flow cytometric data are shown in Fig 5E.
4. Analysis of X‐ray‐induced autophagic flux in PPM1D^+/+ and PPM1D^−/− primary thymocytes. Semiquantitative analyses of protein expression in Fig 5F are shown (n = 3; mean ± SD). *P < 0.05, NS, not significant.

Figure 5. Impaired autophagy and decreased cell viability in PPM1D‐deficient thymocytes after X‐ray irradiation.

• A
  PPM1D‐dependent dephosphorylation of Ulk1 at Ser⁶³⁷, induction of autophagy, and degradation of Noxa. PPM1D^+/+ and PPM1D^−/− primary thymocytes were X‐ray irradiated (5 Gy), and 3 h later, thymocytes were lysed and the expression of each protein was examined by Western blotting. α‐Tubulin was used as a loading control. Semiquantitative analyses are shown in Fig EV5B.
• B–F
  Suppression of X‐ray‐induced autophagy in PPM1D^−/− thymocytes. (B, C) The indicated thymocytes were X‐ray irradiated (5 Gy), and 3 h later, thymocytes were immunostained with an anti‐LC3 antibody. Nuclei were stained with Hoechst 33342. Representative images are shown in (B). LC3 puncta can be seen in irradiated PPM1D^+/+ thymocytes (arrowheads). In (C), the proportion of cells with LC3 puncta was calculated (n > 100 cells in each experiment). Data are shown as the mean ± SD (n = 3 experiments). *P < 0.05. One‐way ANOVA followed by Tukey's post hoc test. (D) The indicated thymocytes were X‐ray irradiated (5 Gy), and 3 h later, thymocytes were subjected to EM analysis. An autophagic vacuole (arrowhead) can be seen in an irradiated PPM1D^+/+ thymocyte. (E) The indicated thymocytes were X‐ray irradiated (5 Gy), and 3 h later, thymocytes were stained with Cyto‐ID^™. Representative flow cytometric data are shown. Quantitative data are shown in Fig EV5C. Cyto‐ID fluorescence in irradiated PPM1D^+/+ thymocytes was higher than that in the other thymocytes. (F) Analysis of autophagic flux. The indicated thymocytes were X‐ray irradiated (5 Gy) in the presence or absence of chloroquine (120 μM), and the expression of each protein was examined by Western blotting. α‐Tubulin was included as a loading control. Semiquantitative analyses are shown in Figure EV5D.
• G–J
  Contribution of PPM1D/Ulk1‐dependent autophagy to irradiation‐induced apoptosis. (G) Thymocytes were X‐ray irradiated (5 Gy), and 6 h later, cell death was determined using the PI assay. Data represent the mean ± SD (n = 3 independent thymi). (H, I) Thymocytes were X‐ray irradiated (5 Gy), and 3 h later, the expression of each protein was examined by Western blotting. α‐Tubulin was included as a loading control. (I) Semiquantitative analyses are shown (n = 3; mean ± SD). (J) Indicated thymocytes were X‐ray irradiated (5 Gy), and 3 h later, caspase‐3/7 activity was examined using the Caspase‐Glo 3/7 assay (Promega) according to the manufacturer's protocol. Data are shown as the mean ± SD (n = 3 experiments). (G–J) *P < 0.05, **P < 0.01, NS: not significant. One‐way ANOVA followed by Tukey's post hoc test.
• K
  The effect of Noxa siRNA on thymocytes. The indicated thymocytes were transfected with the pmaxGFP vector and siRNAs. After 12 h, cells were X‐ray irradiated (5 Gy), and 6 or 10 h later, the population of dead cells among the GFP‐positive cells was determined using the PI assay. Data are shown as the mean ± SD (n = 3 experiments). *P < 0.05, NS: not significant. Unpaired two‐tailed Student's t‐test.

Source data are available online for this figure.

In some cases, autophagy affects the extent of apoptosis 7, 8. Therefore, we examined whether and how PPM1D/Ulk1‐mediated autophagy affects apoptosis. When thymocytes from wild‐type, Ulk1^−/−, PPM1D^−/−, and Ulk1^−/−/PPM1D^−/− DKO mice were irradiated, we observed increased apoptosis in Ulk1^−/− thymocytes compared with wild‐type thymocytes (Fig 5G). Consistent results were observed when apoptosis was assessed by Western blotting for caspase‐3 (Fig 5H and I) and by a caspase‐3/7 activity assay (Fig 5J), suggesting that Ulk1‐dependent autophagy plays a protective role against irradiation‐induced apoptosis. PPM1D^−/− thymocytes were even more susceptible to irradiation‐induced apoptosis than Ulk1^−/− thymocytes (Fig 5G–J), but this susceptibility was not further enhanced in Ulk1^−/−/PPM1D^−/− thymocytes (Fig 5G). These results may be owing to PPM1D regulating not only Ulk1 but also Ulk2 or other unidentified molecules.

We further investigated the mechanism by which autophagy suppresses irradiation‐induced apoptosis. Because irradiation‐induced apoptosis occurs via a p53‐dependent increase in proapoptotic BH3‐only molecules, including Puma, Bim, and Noxa 21, 22, 23, and because autophagy may degrade such proteins to reduce the amount of apoptosis, we aimed to identify molecules that are the target substrates of autophagy. The expression levels of Puma and Bim were not altered between irradiated PPM1D^−/− thymocytes and irradiated wild‐type thymocytes at 3 h (Figs 5A and EV5B). At a later time point, the expression levels of Puma were increased, but we still did not observe higher expression levels in PPM1D^−/− thymocytes than in wild‐type thymocytes (Fig EV6A). Unlike Puma and Bim, several of our experimental results indicated that Noxa is the target of autophagy in irradiated thymocytes. Firstly, the level of Noxa in Ulk1^−/− thymocytes was higher than that in wild‐type thymocytes, and the level of Noxa in PPM1D^−/− thymocytes was much higher than that in Ulk1^−/− thymocytes after irradiation (Fig 5H and I), which was parallel to the extent of apoptosis, suggesting that autophagy inhibits apoptosis via the degradation of Noxa. Second, chloroquine, an autophagy inhibitor, suppressed the autophagic degradation of Noxa in wild‐type thymocytes, but not in PPM1D^−/− thymocytes (Figs 5F and EV5D). Third, immunostaining analysis showed the colocalization of LC3 and Noxa, and more importantly, that the addition of chloroquine increased the number of LC3 puncta colocalized with Noxa (Fig EV6B). Fourth, Noxa contains an LC3‐interacting region (LIR) motif, which ensures the targeting of autophagy substrates to LC3, at the site ⁹⁹FNLV¹⁰². In fact, wild‐type Noxa, but not its LIR mutant Noxa (F99A/V102A), was degraded by etoposide‐induced autophagy (Fig EV6C and D). Fifth, irradiation‐induced Noxa degradation was not observed in thymocytes lacking Atg7 (a core autophagy molecule; Fig EV6E and F). Sixth, the enhanced susceptibility of Ulk1^−/− and PPM1D^−/− thymocytes to irradiation was significantly suppressed by the introduction of an siRNA for Noxa (Fig 5K). Note that the silencing of Noxa did not reduce irradiation‐induced apoptosis in wild‐type thymocytes (Fig 5K), which is consistent with previous reports 21, 22. This is thought to be owing to the autophagic degradation of Noxa in wild‐type thymocytes. Collectively, our results demonstrate that PPM1D‐dependent, Ulk1‐mediated autophagy suppresses irradiation‐induced thymocyte apoptosis via degradation of the proapoptotic Noxa protein.

Figure EV6. Involvement of Noxa in the irradiation‐induced apoptosis.

• A
  Effect of PPM1D on the irradiation‐induced increase in levels of proapoptotic molecules at a later time point. The indicated thymocytes were irradiated with X‐ray (5 Gy) for 6 h. The expression of each apoptosis‐associated protein was examined by Western blotting.
• B
  The colocalization of endogenous LC3 and endogenous Noxa in X‐ray‐treated PPM1D^+/+ thymocytes. PPM1D^+/+ thymocytes were X‐ray irradiated for 3 h in the presence or absence of chloroquine (120 μM), then immunostained with anti‐LC3 and anti‐Noxa antibodies. Representative images of staining with anti‐LC3 (green; left), anti‐Noxa (red; middle), and a merged image (right) are shown. The nucleus was stained with Hoechst 33342 in the merged image.
• C, D
  Crucial role of the LIR domain in the autophagic degradation of Noxa. PPM1D^+/+ MEFs expressing wild‐type Noxa (WT) or LIR mutant Noxa (F99A/V102A) were treated with etoposide (10 μM), q‐VD‐OPh (10 μM), and MG132 (10 μM) in the presence or absence of bafilomycin A1 (10 nM). The expression of each protein was examined by Western blotting. (D) Semiquantitative analysis of Noxa is shown (n = 3; mean ± SD). *P < 0.05, NS: not significant. One‐way ANOVA followed by Tukey's post hoc test.
• E, F
  Suppression of the irradiation‐induced degradation of Noxa by the lack of Atg7. The Atg7‐deficient thymocytes isolated from PIPC‐injected Atg7^flox/flox:Mx1‐cre mice and control thymocytes were X‐ray irradiated (5 Gy) in the presence or absence of chloroquine (120 μM). After 3 h, thymocytes were lysed, and the expression of Noxa was examined by Western blotting. (F) Semiquantitative analysis of Noxa in (E) (n = 3; mean ± SD). *P < 0.05, NS: not significant. One‐way ANOVA followed by Tukey's post hoc test.
• G
  A schematic model of the PPM1D‐dependent dephosphorylation of Ulk1 and its effect on autophagy and apoptosis. Genotoxic stress induces PPM1D accumulation in a p53‐dependent manner. Then, PPM1D dephosphorylates Ulk1 at serine⁶³⁷, stimulates Ulk1 puncta formation, and induces autophagy. This autophagy contributes to the suppression of genotoxic stress‐induced apoptosis through the degradation of Noxa in primary thymocytes.

Source data are available online for this figure.

This study demonstrated the essential role of PPM1D in genotoxic stress‐induced autophagy via the dephosphorylation of Ulk1 at Ser⁶³⁷. PPM1D is expressed at low levels in healthy cells and its expression is increased in the nucleus after genotoxic stress. PPM1D mainly functions in the nucleus by the dephosphorylation of various nuclear proteins. In contrast, we showed here that PPM1D also plays a role in the cytoplasm. After exposure to genotoxic stress, a small amount of PPM1D is expressed in the cytoplasm and induces autophagy. Because PPM1D puncta appeared from an early period after etoposide treatment, it is likely not a result of the cell destruction process, that is, an increase in nuclear membrane permeability. Rather, it is likely to be a molecular signal tightly regulated by p53.

This study also demonstrated that cytosolic PPM1D interacts with and dephosphorylates Ulk1 at Ser⁶³⁷. The dephosphorylation sequence preferred by PPM1D is known as the pSQ site 13, which is present in nuclear substrate proteins. However, the PPM1D target sequence preference is slightly flexible, since it also dephosphorylates the pTGY site in p38, a cytosolic protein 13. In the case of Ulk1, the target sequence at Ser⁶³⁷ corresponds to the pSSQ site, not the pSQ site. Therefore, PPM1D may dephosphorylate sites other than the pSQ site with regard to cytoplasmic substrate proteins.

In this study, we demonstrated the essential role of PPM1D in genotoxic stress‐induced autophagy, namely its dephosphorylation of Ulk1. PPM1D is an established factor in the negative feedback against the p53‐dependent DNA damage response 13. This function is mediated by the dephosphorylation of and the inactivation of DNA damage‐signaling molecules, such as Chk2, ATM, and even p53 itself 13. Although Ulk1 is not categorized as a DNA damage‐signaling molecule, PPM1D‐dependent Ulk1 dephosphorylation is expected to function as a negative regulator of the p53‐dependent DNA damage response. We first examined whether this type of autophagy might degrade and hence decrease the level of DNA damage‐signaling molecules. However, the expression of these molecules was regulated only transcriptionally and not by autophagy. We next considered whether p53‐dependent apoptosis might be reduced by PPM1D‐dependent autophagy. This hypothesis appeared to be correct; the loss of PPM1D enhanced irradiation‐induced thymocyte apoptosis via both p53 signaling and Ulk1‐mediated autophagy. Thus, one physiological role of PPM1D‐mediated Ulk1 dephosphorylation is to reduce the amount of DNA damage‐induced apoptosis.

Taken together, the results presented here demonstrate that PPM1D is required for the dephosphorylation of Ulk1 at Ser⁶³⁷ and the subsequent induction of autophagy. PPM1D‐dependent autophagy contributes to the suppression of irradiation‐induced apoptosis through the degradation of Noxa in primary thymocytes.

Materials and Methods

Mice

The generation of Ulk1^−/− mice has been described previously 24. PPM1D^−/− mice were kindly provided from professor L.A. Donehower 25. Ulk1/PPM1D DKO mice were generated by crossbreeding Ulk1‐deficient mice with PPM1D‐deficient mice. The Atg7‐flox mice were described in other studies 26. Mice were bred at the Laboratory for Recombinant Animals, Medical Research Institute, Tokyo Medical and Dental University, Tokyo, Japan.

Antibodies and chemicals

The antibodies are listed in Table EV1. Etoposide, bafilomycin A1, and GSK2830371 were obtained from Sigma‐Aldrich, Adipogen, and ChemieTek, respectively. All other chemicals were purchased from Nacalai Tesque.

Plasmid construction

The HA‐tagged mouse Ulk1 plasmid was kindly provided by Professor Muramatsu (Tokyo Medical and Dental University). The introduction of point mutations into mouse Ulk1 was performed using PCR with Pfu Turbo (Agilent Technologies). The pCMV‐PPM1D expression plasmid which contained DNA encoding mouse PPM1D was generated by amplifying the appropriate cDNA from a mouse cDNA library and subcloning it into plasmid p3xFlag CMV14 (Sigma‐Aldrich). All constructs were confirmed by sequence analysis.

Cell culture and DNA transfection

MEFs were generated from p53^+/+, p53^−/−, PPM1D^+/+, and PPM1D^−/− embryos on embryonic day 14.5 by immortalization with the SV40 T antigen. Ulk1/Ulk2 DKO MEFs were kindly provided by Professor Tooze (London Research Institute). To generate MEFs stably expressing HA‐Ulk1 and its mutants, each plasmid was transfected into MEFs (1 × 10⁶) using the Amaxa electroporation system (Lonza) and selected using hygromycin B (Invivogen). The pEGFP‐C1B‐mmDFCP1 plasmid was transiently transfected into MEFs (1 × 10⁶) using the Neon transfection system (Thermo Fisher Scientific). The transfection efficiency was more than 75%, as assessed by the number of GFP‐fluorescent cells. Cells were treated with etoposide (10 μM), or treated with starvation in Hanks' balanced salt solution supplemented with 1 mM sodium pyruvate, 10 mM HEPES/Na⁺ (pH 7.4), and 0.05 mM 2‐mercaptoethanol.

PPM1D^+/+, PPM1D^−/−, Ulk1^−/−, and Ulk1^−/−/PPM1D^−/− primary thymocytes were harvested from the respective mice at 5–6 weeks of age. Primary thymocytes were also harvested from Atg7^flox/flox:Mx1‐Cre mice, in which poly (I:C) was intraperitoneally injected (300 μg per day for five consecutive days). Nucleofection of thymocytes was performed with the Mouse T cell Nucleofector kit according to the manufacturer's protocols. Thymocytes were transfected with the pmaxGFP vector together with Noxa siRNA or control siRNA and incubated for 12 h. A nontargeting siRNA pool (D‐001206‐13‐05) and the Noxa siRNA SMARTpool (M‐048269‐01‐0005) were purchased from Dharmacon. The transfection efficiency was more than 25%, as assessed by the number of GFP‐fluorescent cells. MEFs and thymocytes were cultured in the medium described previously 24.

Immunoblot analysis

Cells were washed with ice‐cold PBS, frozen, and then lysed in cell lysis buffer containing 20 mM HEPES (pH 7.5), 100 mM NaCl, 1.5 mM MgCl₂, 1 mM EGTA, 10 mM Na₂P₂O₇, 10% glycerol, 1% NP‐40, 1 mM dithiothreitol, 1 mM Na₃VO₄, and 1% protease inhibitor cocktail (Nacalai). After vortexing for 15 s, insoluble material was removed by centrifugation. Supernatants were loaded onto 5–20, 15, or 8% SDS–polyacrylamide gels. After electrophoresis, the proteins were blotted onto a PVDF membrane (Millipore). The membranes were blocked with 5% skim milk in TBS containing 0.05% Tween‐20 (TBS‐T) and incubated with the primary antibody overnight at 4°C. After washing with TBS‐T, the membranes were incubated with a horseradish peroxidase‐labeled secondary antibody and visualized with Chemi‐Lumi One Super reagent (Nacalai). All experiments were conducted at least in duplicate. Protein band densities were semiquantified using ImageJ software.

GST pull–down assays

GST‐PPM1D was purchased from Abnova (#H00008493‐P01). Cell lysates from healthy MEFs were preincubated with 20 μl of 50% glutathione–Sepharose 4B (GE Healthcare) for 30 min at 4°C. After centrifugation at 4,400 g for 3 min, supernatants were incubated with GST and GST‐PPM1D immobilized on new glutathione–Sepharose 4B for 2 h at 4°C. After washing with 1 ml PBS three times, bound proteins were released from the beads by heating at 100°C for 3 min in 2× Laemmli sample buffer.

Immunoprecipitation

Control and etoposide‐treated MEFs were harvested and lysed with RIPA buffer. Immunoprecipitation was performed using anti‐Ulk1 and control antibodies in the presence of protein G–Sepharose (GE Healthcare) for 2 h at 4°C. The beads were then washed three times with PBS. Proteins were released from the beads by heating at 100°C for 3 min in 2× Laemmli sample buffer. Immunoblotting was performed as described above, except that the EasyBlot anti Rabbit IgG kit (GeneTex) was used.

Phosphatase assay

Ulk1 was immunoprecipitated from healthy MEFs with an anti‐Ulk1 antibody and suspended in phosphatase buffer (50 mM Tris–HCl (pH 7.5), 30 mM MgCl₂, 1 mg/ml bovine serum albumin, 0.05% 2‐mercaptoethanol). Samples were treated with or without 1 μg GST‐PPM1D and a phosphatase inhibitor cocktail (Calbiochem, #524627) for 1 h at 30°C, and then, the levels of phosph‐Ulk1⁶³⁷ were examined using the EasyBlot anti Rabbit IgG kit (GeneTex).

Immunofluorescence analysis

Cells were fixed in 4% paraformaldehyde containing 8 mM EGTA for 10 min and then permeabilized using 50 μg/ml digitonin for 5 min. Then, cells were stained with the indicated primary antibodies for 1 h at room temperature. After washing, the cells were stained with secondary antibodies and Hoechst 33342, mounted in Prolong Gold antifade reagent (Life Technologies) and observed using a fluorescence microscope (IX71, Olympus) and a laser scanning confocal microscope (LSM510 and LSM710, Zeiss). Thymocytes were fixed in 4% paraformaldehyde containing 8 mM EGTA for 10 min, pelleted onto slides using a Cytospin3 (Shandon) centrifuge and permeabilized using 50 μg/ml digitonin for 5 min. The cells were then stained as described above.

Cell viability and Cyto‐ID assay

Thymi were prepared from young mice (about 5–10 weeks old). Primary thymocytes were harvested and seeded onto dishes and maintained in modified DMEM. The cells were X‐ray irradiated (5 Gy), and their viability was determined using PI staining and caspase‐3 Western blotting. To determine autophagic activity, cells were stained with both PI and the Cyto‐ID Autophagy detection kit (ENZO Life Sciences) according to the manufacturer's protocol. The autophagic cells among the PI‐negative thymocytes were detected by flow cytometry (BD; FACS Canto II). Data analysis was performed using BD FACSDiva and FlowJo software.

Electron microscopy

Cells were fixed with 1.5% paraformaldehyde/3% glutaraldehyde in 0.1 M phosphate buffer (pH 7.2) and then treated with 1% OsO₄. After dehydration, the fixed cells were embedded in Epon 812. Thin sections were cut and stained with uranyl acetate lead citrate for observation under a JEM‐1010 electron microscope (JEOL Co. Ltd.) at 80 K.

Statistical analysis

Results are expressed as the mean ± standard deviation (SD). Statistical evaluation was performed using Prism (GraphPad) software. Comparisons of two datasets were performed using unpaired two‐tailed Student's t‐tests. All other comparisons of multiple datasets were performed using one‐way ANOVA followed by Tukey's post hoc test. A P‐value < 0.05 was considered to indicate a statistically significant difference between two groups.

Author contributions

ST designed the research, performed the biological analyses, and wrote the manuscript. TY designed the research and performed the biological analyses. SA performed the EM analyses. SH and AN supported the biological analyses. SS designed the research and wrote the manuscript. All authors edited the manuscript.

Conflict of interest

The authors declare that they have no conflict of interest.

Supporting information

Expanded View Figures PDF

Table EV1

Source Data for Expanded View

Review Process File

Source Data for Figure 1

Source Data for Figure 2

Source Data for Figure 3

Source Data for Figure 4

Source Data for Figure 5

EMBO Reports (2016) 17: 1441–1451