Report

GSK3-TIP60-ULK1 Signaling Pathway Links Growth Factor Deprivation to Autophagy

See allHide authors and affiliations

Science  27 Apr 2012:
Vol. 336, Issue 6080, pp. 477-481
DOI: 10.1126/science.1217032

This article has a correction. Please see:

Abstract

In metazoans, cells depend on extracellular growth factors for energy homeostasis. We found that glycogen synthase kinase-3 (GSK3), when deinhibited by default in cells deprived of growth factors, activates acetyltransferase TIP60 through phosphorylating TIP60-Ser86, which directly acetylates and stimulates the protein kinase ULK1, which is required for autophagy. Cells engineered to express TIP60S86A that cannot be phosphorylated by GSK3 could not undergo serum deprivation–induced autophagy. An acetylation-defective mutant of ULK1 failed to rescue autophagy in ULK1−/− mouse embryonic fibroblasts. Cells used signaling from GSK3 to TIP60 and ULK1 to regulate autophagy when deprived of serum but not glucose. These findings uncover an activating pathway that integrates protein phosphorylation and acetylation to connect growth factor deprivation to autophagy.

Autophagy is an evolutionarily conserved cellular process by which a cell degrades its own organelles and large protein aggregates. Autophagy functions to maintain energy homeostasis under nutrient-poor conditions to allow cell survival (14). Unlike unicellular organisms, metazoans depend on extracellular growth factors to regulate uptake and digestion of environmental nutrients. Withdrawal of growth factors or culture of cells in serum-free medium initiates autophagy (2, 5). Dysregulation of autophagy is also associated with diverse diseases, including cancers and neurodegeneration (6). Products of a series of autophagy genes (ATGs) mediate and regulate various aspects of autophagy (710). ATG1 (ULK1 and ULK2 in mammals) encodes a protein kinase that is inhibited by mTOR (mammalian target of rapamycin) complex 1 (mTORC1) (1113). Inhibition of mTOR leads to activation of ULK1 and initiation of autophagy (14). ULK1 is also activated by the energy sensor kinase adenosine monophosphate-dependent protein kinase (AMPK) upon nutrient deprivation (15, 16).

We observed that the acetyltranferase TIP60 (HIV-1 Tat interactive protein, 60 kD) displayed two bands with different electrophoretic mobility (fig. S1). Mass spectrometry revealed that the upper-band TIP60 was phosphorylated on Ser86 and Ser90, and the lower one singly phosphorylated on Ser90 (fig. S1). We produced a monoclonal antibody that specifically recognizes Ser86-phosphorylated TIP60. When Ser90 was mutated, Ser86 was no longer phosphorylated, indicating that Ser90 phosphorylation may be a prerequisite for Ser86 phosphorylation (Fig. 1, A and B). The Ser86 and Ser90 residues are located in a conserved S/TXXXS/T (where S/T represents serine/threonine, and X can be any amino acid) phosphorylation motif recognized by glycogen synthase kinase-3 (GSK3) (fig. S2) (17, 18). Co-expression of GSK3 enhanced phosphorylation of TIP60 on Ser86, which was inhibited by treatment with GSK3 inhibitors SB216763, SB415286, or LiCl (fig. S3). In vitro phosphorylation assays demonstrated that TIP60 appears to be phosphorylated by GSK3 and cyclin-dependent kinase 1 (CDK1) through a dual-kinase mechanism (Fig. 1C and fig. S4) (19).

Fig. 1

Phosphorylation of TIP60 by two kinases in cells deprived of serum. (A) Differential electrophoresis mobilities among WT-TIP60 and TIP60 mutants. (B) Dephosphorylation of immunoprecipitated Myc-tagged WT and S86A but not S90A-TIP60 by calf-intestinal alkaline phosphatase (CIP). (C) In vitro phosphorylation of TIP60. In vitro kinase assays using His-tagged WT-TIP60 or S90A-TIP60 as substrates for immunoprecipitated Myc-cyclinB1/CDK1 complexes from human embryonic kidney (HEK) 293 T cells and purified bacterially expressed GST-GSK3β. (D) Time course of serum-deprived HCT116 cells. Endogenous TIP60 was immunoprecipitated with antibody against TIP60. Total cell lysates (TCL) and immunoprecipitates (IP) were immunoblotted as indicated. (E) Modification of endogenous TIP60. Lentivirus-mediated GSK3α and GSK3β double-knockdown HCT116 cells (GSK3-DKD) as well as LacZ siRNA-expressing control cells (ctrl) were placed in serum-free medium or standard medium for 12 hours. (F) HCT116 cells deprived of serum were treated with dimethyl sulfoxide (DMSO) (mock) or GSK3 inhibitors SB216763 or SB415286. (G) GSK3 activation increased phosphorylation of TIP60 on Ser86. HCT116 cells were placed in medium containing 20 μM inhibitor IV of AKT1 and AKT2 or 40 nM rapamycin for 16 hours. TCLs and TIP60 IPs were immunoblotted as indicated. Statistical analyses on the impacts of AKT inhibitor or rapamycin on GSK3β Ser9 phosphorylation were performed as described in the supplementary materials and are shown in fig. S5A.

GSK3β is inhibited through phosphorylation at Ser9 in cells stimulated by growth factors. Several signaling pathways induce the inhibitory phosphorylation of GSK3, including the phosphoinositide-3-kinase (PI3K) and AKT signaling pathway, the mitogen-activated protein kinase (MAPK) cascade, and the mTOR pathway (2024). We therefore tested whether serum deprivation would increase phosphorylation of TIP60 on Ser86. In human colorectal cancer HCT116 cells, Ser86 phosphorylation was increased in cells deprived of serum, concurring with a decreased phosphorylation of Ser9 of GSK3β and a stronger interaction between GSK3β and TIP60 (Fig. 1, D and E). Phosphorylation of Ser86 of TIP60 was decreased in cells lacking GSK3 or in cells treated with inhibitors of GSK3, AKT, or mTOR (Fig. 1, E and G, and figs. S3 and S5A). Mouse embryonic fibroblasts (MEFs) lacking the gene for tuberous sclerosis protein 2 (TSC2) that have attenuated GSK3 activity (24) showed reduced phosphorylation of Ser86 (fig. S5B).

To test whether TIP60 functions in autophagy induced by serum deprivation, we generated an HCT116 cell line (TIP60 KD) expressing small interfering RNA (siRNA) to TIP60 (fig. S6). Depletion of TIP60 impaired the lipidation of microtubule-associated protein 1 light chain 3 (LC3)—a marker of autophagosome formation—in cells deprived of serum (Fig. 2A). The numbers of autophagic vacuoles (AVs) and vesicles containing GFP-LC3 were also reduced (figs. S7 and S8). We next tested whether Ser86 phosphorylation contributes to the essential role of TIP60 in autophagy induction. A MEF line in which both TIP60 alleles are replaced by TIP60S86A was generated (fig. S9). Similar deficiencies in autophagy induction as accessed by LC3 lipidation, p62 degradation, and autophagosome formation were observed in the TIP60S86A cells, which could be rescued by introduction of wild-type (WT) TIP60 (Fig. 2B and figs. S10 to S12). The heart of postnatal mice shows a sharp increase in autophagy activity during the first 24 hours after birth (25). We therefore measured phosphorylation of TIP60 on Ser86 in the heart of rat embryos and neonates. Ser86 phosphorylation was significantly increased in 3 hours after birth and subsequently reduced in 36 hours. The temporal pattern of TIP60 Ser86 phosphorylation correlates well with that of lipidation of LC3 and degradation of p62 (Fig. 2C and fig. S13).

Fig. 2

TIP60 along with its Ser86 phosphorylation is essential in serum starvation–induced autophagy. (A) LC3 lipidation in TIP60 siRNA-expressing HCT116 cells (TIP60-KD) or the control cells (ctrl). Cells were deprived of serum for indicated times. Total proteins were extracted and immunoblotted as indicated. CQ, chloroquine. The relative amounts of LC3II were calculated from densitometry performed on immunoblots and normalized to the amount of tubulin. Data represent mean ± SEM of three independent experiments. Statistical significance was determined by analysis of variance (ANOVA); †P < 0.01, *P < 0.05, **P < 0.01 (ANOVA followed by Tukey). (B) LC3 lipidation in WT and TIP60S86A MEFs. Cytosolic proteins were then extracted and analyzed by means of immunoblotting. The graphs indicate p62 (degradation of which is a marker for autophagy activation)–to–glyceraldehyde-3-phosphate dehydrogenase (GAPDH) ratio and fold induction of LC3II. Data represent mean ± SEM of three independent experiments. †P < 0.01 (ANOVA), *P < 0.05, **P < 0.01 (ANOVA followed by Tukey); N.S., not significant. (C) TIP60 from heart homogenates isolated from 3 hours after birth or prenatal rat embryos was immunoprecipitated by using antibody to TIP60, followed by immunoblotting. Each lane represents a different individual. The relative Ser86 phosphorylation levels were calculated and normalized to total TIP60. Data are shown as mean ± SEM (n = 4 rats, each group). **P = 0.0034 (Student’s t test).

Because the highly conserved ATG genes are core components of the autophagy pathway, we tested for interaction between TIP60 and ATGs. Strong association of exogenous Myc-tagged TIP60 with hemagglutinin (HA)–tagged ULK1 (ATG1) was detected (Fig. 3A). In HCT116 cells deprived of serum, the interaction between endogenous TIP60 and ULK1 was increased (Fig. 3B). We also observed decreased interaction between exogenous TIP60 and ULK1 in cells treated with the GSK3 inhibitor SB216763 because Ser86-phosphorylated TIP60 may preferentially bind ULK1 (Fig. 3C).

Fig. 3

Acetylation of ULK1 by TIP60 upon serum deprivation. (A) Association of Myc-tagged TIP60 with cotransfected HA-tagged ATGs. Protein extracts were immunoprecipitated with antibody to HA 20 hours after transfection. (B) Increased association between TIP60 and ULK1 in cells deprived of serum. TIP60 immunoprecipitates from HCT116 cells serum starved for 16 hours were immunoblotted as indicated. Ratio of coimmunoprecipitated ULK1 to total ULK1 was calculated. Data represent mean ± SEM of three independent experiments. **P < 0.01 (Student’s t test). (C) Association of HA-ULK1 with Myc-TIP60. HEK293 T cells transfected with HA-ULK1 and Myc-TIP60 were treated with DMSO or 10 μM GSK3 inhibitor SB216763 for 12 hours. (D) Acetylation of endogenous ULK1. TIP60-KD HCT116 cells and ctrl cells were serum-starved for 12 hours. Acetylated proteins were immunoprecipitated with antibody to acetylated lysine. Ratio of acetylated ULK1 to total ULK1 was calculated. Data represent mean ± SEM of three independent experiments. N.S., not significant; **P < 0.01 (ANOVA followed by Tukey). (E) Inhibition of the acetylation of ULK1 by GSK3 inhibitor. HCT116 cells were pretreated with DMSO or 10 μM SB216763 for 4 hours before 12-hour serum starvation. Acetylated ULK1 was analyzed as in (D). (F) Phosphorylation-dependent activity of TIP60 toward ULK1. FLAG-TIP60 was immunoprecipitated from HEK293 T cells treated with DMSO or 10 μM SB216763 for 12 hours; TIP60 cotransfected with HA-GSK3β was used as a comparison. In vitro acetylation assays with purified His-ULK1 as a substrate were performed. Relative acetyltransferase activities of TIP60 toward ULK1 were calculated as the ratio of acetylated ULK1 to total ULK1. Data represent mean ± SEM of three independent experiments. **P < 0.01 (ANOVA followed by Tukey).

We then examined the biochemical consequences of the interaction between TIP60 and ULK1 in an in vitro acetylation assay. TIP60 acetylated ULK1 (fig. S14). In HCT116 cells deprived of serum, the amount of acetylated ULK1 was significantly increased, which was almost undetectable in cells resupplemented with growth factors such as insulin or epidermal growth factor as well as in cells expressing siRNA to TIP60 (Fig. 3D and fig. S15). GSK3 inhibitor SB216763 also led to decreased acetylation of ULK1 (Fig. 3E), indicating a requirement of TIP60 Ser86 phosphorylation in promoting its acetyltransferase activity. TIP60 precipitated from GSK3 inhibitor–treated cells or S86A-TIP60 mutant showed compromised acetyltransferase activities toward ULK1; on the other hand, coexpression of GSK3 (to enhance Ser86 phosphorylation) increased the activity (Fig. 3F and fig. S16).

We subjected acetylated ULK1 protein to mass spectrometry. Seven lysine residues were found to be candidate acetylation sites (fig. S17). We then created mutants changing these lysine residues to arginine singly or in combination. Simultaneous mutation of Lys162 and Lys606 (2KR) on ULK1 almost entirely abolished its acetylation by TIP60 in vitro (Fig. 4A). Acetylation of ULK1 by TIP60 in vitro increased its kinase activity by using myelin basic protein (MBP) as a substrate (Fig. 4B). LC3II production and the number of LC3 puncta were reduced in 2KR-expressing ULK1−/− cells (Fig. 4, C and D).

Fig. 4

Requirement of TIP60-mediated acetylation of ULK1 in serum deprivation–induced autophagy. (A) In vitro acetylation assays were performed to determine TIP60 acetylation of immunoprecipitated HA-tagged WT-ULK1 or its lysine-to-arginine mutants. (B) In vitro coupled acetylation-phosphorylation assays were performed as described in the supplementary materials, materials and methods. WT-TIP60 or acetyltransferase-dead TIP60-DN was incubated with WT-ULK1 or 2KR-ULK1 in different combinations indicated. MBP was used as the substrate for ULK1 kinase in the presence of 32P-ATP. Data represent mean ± SEM of three independent experiments. N.S., not significant; **P < 0.01 (ANOVA followed by Tukey). (C) LC3 lipidation in ULK1−/− MEFs stably expressing WT-ULK1 or 2KR-ULK1. Cells were serum starved for 9 hours, and total cell lysates were analyzed for LC3 lipidation by means of immunoblotting. (D) Autophagosome formation in ULK1−/− MEFs. The cells were serum-starved for 9 hours after 24 hours of infection and were then fixed. LC3 positive puncta were shown as mean ± SEM of 5 random areas. N.S., not significant; **P < 0.01 (ANOVA followed by Tukey). (E) Comparison of LC3 lipidation in cells deprived of serum or glucose. HCT116 cells were pretreated with DMSO or GSK3 inhibitor SB216763 for 4 hours and then deprived of serum or glucose. Total proteins were analyzed for LC3 lipidation.

In the absence of glucose, ULK1 is phosphorylated and activated by AMPK (15, 16). However, during 16 hours of glucose starvation neither the inhibitory Ser9 phosphorylation of GSK3β nor the phospho-Ser86 signal was affected (fig. S18). Treatment of cells with GSK3 inhibitors decreased serum starvation–induced autophagy but not that caused by glucose deprivation (Fig. 4E). Moreover, autophagy was as effectively induced by serum starvation in AMPKα1/2-DKO (double knockout) MEFs as in control MEFs (fig. S19). These results indicate that the autophagy signaling axis mediated by the GSK3 sensor kinase is distinct from that mediated by AMPK. To investigate possible interplay between mTOR-AMPK phosphorylation and TIP60 acetylation of ULK1, we examined the phosphorylation of WT-ULK1 or acetylation-deficient 2KR-ULK1 and the acetylation of phosphorylation-deficient ULK1 proteins reintroduced into the ULK1−/− MEFs. These phosphorylation events appeared not to alter ULK1 acetylation by TIP60 or vice versa (figs. S20 and S21).

GSK3 functions in several pathways that regulate cellular homeostasis of energy and growth, including those mediated by PI3K and AKT, mTOR, and MAPK (17, 18). Under situations such as insulin or insulin-like factor withdrawal, activity of AKT is attenuated, leading to the activation of GSK3. Inhibition of mTOR by rapamycin leads to inactivation of its target kinases, including AKT, p70S6K, and p85S6K, which in turn promotes GSK3 activation (2024). Mitogen-activated kinase signaling also inhibits GSK3 activity through the downstream kinase p90RSK (22). Our present study demonstrates an autophagy-activating pathway that comprises GSK3, TIP60, and ULK1. In this pathway, GSK3 is activated by the removal of its inhibitory Ser9 phosphorylation upon withdrawal of growth factors. Activated GSK3 catalyzes phosphorylation of TIP60 at Ser86, which depends on a prior phosphorylation at Ser90 and results in higher affinity of TIP60 for ULK1 and increased acetylation and kinase activity of ULK1. Inhibition of GSK3 can block serum but not glucose deprivation–induced autophagy, all pointing to the conjecture that glucose and growth factor deprivation use distinct machineries in autophagy induction (fig. S22). In line with an essential role of TIP60 for autophagy induction is the previous observation that TIP60−/− mouse blastocysts fail to undergo implantation and die around embryonic day 3.5, when autophagy activity is strongly elicited for normal implantation (26, 27). The pathway we discovered may have a general role in sustaining cellular energy and metabolic substrates, considering the common sensor role of GSK3 for growth-related signaling.

Supplementary Materials

www.sciencemag.org/cgi/content/full/336/6080/477/DC1

Materials and Methods

Figs. S1 to S22

References (2830)

References and Notes

  1. Acknowledgments: We thank D. Luo for help with statistical analyses, M. Kundu and H. Zhang for providing ULK1−/− and TSC2−/− MEFs, and B. Viollet and K. R. Laderoute for providing AMPKα1/2-DKO MEFs. This work was supported by grants from the 973 Program (2011CB910800), National Natural Science Foundation of China (31130016, 30921005, and 31000621), Fundamental Research Funds for the Central Universities (2010121094), and the 111 Project of Ministry of Education of China (B06016). S.-C.L. is a Cheung Kong Scholar.
View Abstract

Navigate This Article