mTOR inhibits autophagy by controlling ULK1 ubiquitylation, self-association and function through AMBRA1 and TRAF6

Francesca Nazio ,2, Flavie Strappazzon1,2, Manuela Antonioli3, Pamela Bielli2, Valentina Cianfanelli1,2, Matteo Bordi1,2, Christine Gretzmeier4,5, Joern Dengjel4,5, Mauro Piacentini3,6, Gian Maria Fimia3 and Francesco Cecconi1,2,7

Summary

Autophagy is important in the basal or stress-induced clearance of bulk cytosol, damaged organelles, pathogens and selected proteins by specific vesicles, the autophagosomes. Following mTOR (mammalian target of rapamycin) inhibition, autophagosome formation is primed by the ULK1 and the beclin-1–Vps34–AMBRA1 complexes, which are linked together by a scaffold platform, the exocyst. Although several regulative steps have been described along this pathway, few targets of mTOR are known, and the cross-talk between ULK1 and beclin 1 complexes is still not fully understood. We show that under non-autophagic conditions, mTOR inhibits AMBRA1 by phosphorylation, whereas on autophagy induction, AMBRA1 is dephosphorylated. In this condition, AMBRA1, interacting with the E3-ligase TRAF6, supports ULK1 ubiquitylation by LYS-63-linked chains, and its subsequent stabilization, self-association and function. As ULK1 has been shown to activate AMBRA1 by phosphorylation, the proposed pathway may act as a positive regulation loop, which may be targeted in human disorders linked to impaired autophagy.
Autophagy is a highly conserved catabolic process, by which cytoplasmic material (for example, proteins, lipids and organelles) is transported to the lysosome for degradation by means of doublemembraned vesicles, the autophagosomes. The process of de novo autophagosome formation has been shown to be regulated by at least four molecular complexes including the ULK1 complex, the beclin-1–VPS34–AMBRA1 complex, two transmembrane proteins, ATG9 and WIPI complexes, and two ubiquitin-like protein conjugation systems (ATG12 and LC3; ref. 1). The mammalian orthologue of yeast Atg1, the serine/threonine kinase ULK1, plays a key role in autophagy induction2. ULK1 forms a stable complex with ATG13 (refs 3,4) and with FIP200 (ref. 5), the mammalian counterpart of ATG17. Under normal conditions, the ULK1 complex is de-activated by mTORC1 kinase-dependent phosphorylation of ULK1 and ATG13. Inhibition of mTORC1 enhances ULK1 kinase activity, which, in turn, triggers ATG13 and FIP200 phosphorylation4,6,7. ULK1 also interacts with ATG101, another key player in autophagosome formation, in an ATG13-dependent manner6,8. On autophagy, the ULK1 complex is localized in the isolation membranes5, where it promotes the formation of PI(3)P-enriched membrane compartments9 through the beclin-1–VPS34–AMBRA1 complex, which recruits downstream effectors to the site where nucleation occurs. AMBRA1, on its ULK1-dependent release from a cytoskeletal docking site, induces autophagosome nucleation by promoting beclin 1 interaction with its target lipid kinase VPS34 (refs 10,11). In this work, we found that AMBRA1isaULK1-bindingpartner,requiredforULK1stabilityandkinase activity. Interestingly, during autophagy, ULK1 undergoes Lys 63ubiquitylation and this modification is mediated by AMBRA1 and a specificRINGfingerE3ubiquitinligase,TRAF6(tumournecrosisreceptorassociated factor 6). Moreover, we determined that this Lys-63-linked ubiquitylationfacilitatesULK1self-associationduringautophagy.
Taken together, our results lead us to propose a role for AMBRA1 as a substrate receptor for the TRAF6 ligase in mediating ULK1 Lys63-linked ubiquitylation that proves to be relevant for ULK1 protein stability and activation. Finally, we identified a specific inhibitory phosphorylation by mTOR on AMBRA1, which is removed during autophagy induction. We propose in this paper that mTOR, besides its negative regulation of ULK1 through its direct phosphorylation, may also indirectly inhibit ULK1 stability and activity by modifying AMBRA1 and impairing its E3-ligase-related functions.

RESULTS

AMBRA1 interacts with the serine/threonine kinase ULK1 in mammalian cells

On autophagy induction, ULK1 phosphorylates AMBRA1, a modification essential for AMBRA1 translocation to the endoplasmic reticulum, where it can prime autophagosome formation11. To examine the correlation between AMBRA1 and ULK1, we investigated whether AMBRA1 interacted with ULK1. Indeed, AMBRA1 interacts with both wild-type (WT) and K46I (dominant negative) ULK1 (ref. 3), showing that their interaction may not depend on ULK1 kinase activity (Fig. 1a). Furthermore, AMBRA1 and ULK1 form a stable protein complex in normal conditions and on autophagy induction by starvation (Fig. 1b,c), as revealedbyco-immunoprecipitationandco-localizationexperiments.
Next, we attempted to identify the ULK1 domain responsible for this binding to AMBRA1. To this end, we used the full-length form of ULK1 (FL) and two different ULK1 deletion mutants (ULK1 1829–1,051 and 11–828; ref. 4). As shown in Fig. 1d, the carboxy-terminal domain (CTD) of ULK1, which binds ATG13 (ref. 3), is sufficient to bind AMBRA1. Vice versa, when mapping the AMBRA1 region responsible for ULK1 binding, we found that the AMBRA1 amino-terminal (F1) and C-terminal (F3) regions are sufficient to interact with ULK1, whereas the central region (F2), which binds beclin 1 (ref. 10), shows no interaction (Fig. 1e).
On the basis of these data, we examined the capacity of AMBRA1 to associate with ATG13, a ULK1-binding protein. As shown in Fig. 1f, AMBRA1 is able to interact with ATG13, suggesting that ULK1, ATG13 and AMBRA1 coexist in a complex. However, AMBRA1–ULK1 interaction does not significantly depend on ULK1–ATG13 binding (Fig. 1g). In conclusion, AMBRA1 forms a stable ternary complex with ULK1 and ATG13, which does not depend on ULK1 kinase activity or on ULK1 binding to ATG13.

AMBRA1 is important for ULK1 stability and kinase activity

By analysing ULK1 expression, we noticed that the ULK1 protein was expressed at higher levels on AMBRA1 transfection (Fig. 2a). Thus, we reasoned that AMBRA1 binding could stabilize ULK1 and analysed the expression of Ulk1/2 in Ambra1-deficient (Ambra1gt/gt) mouse tissues. As illustrated in Fig. 2b and Supplementary Fig. S1a, the expression levels of Ulk1/2 in Ambra1gt/gt embryos were much lower than in the wild type (Ambra1+/+). Similar results were obtained in Ambra1+/gt mouse embryonic fibroblasts (MEFs) (Supplementary Fig. S1b), suggesting an AMBRA1-dose-dependent effect in this stabilization. As shown in SupplementaryFig.S1c,therearenodifferencesintheexpressionofbeclin1 and ATG12, two other autophagy-related proteins, in Ambra1+/+ versus Ambra11gt/gt MEFs, indicating a ULK1-specific effect for AMBRA1. Furthermore, Ulk1 was detected as a smeared band with faster mobility in Ambra1gt/gt tissues. As the faster gel migration of ULK1 indicates a reduced phosphorylation level of the protein2,5, AMBRA1 could also be important for ULK1 basal phosphorylation. Indeed, the same results were observed by downregulation of AMBRA1 in HeLa cells, an effect thatcanbecompensatedbyAMBRA1reconstitutionassays(Fig. 2c,d).
By analysing Ulk1 messenger RNA expression, we found that there was no difference between Ambra1+/+ and Ambra1gt/gt MEFs, and that Ulk1 mRNA, as expected5, was enhanced after starvation in both the MEF genotypes analysed (Supplementary Fig. S1d). In contrast, an analysis of ULK1 decay in the presence of the translational inhibitor cycloheximide showed that, in the presence of AMBRA1, ULK1 expression remained essentially unchanged during the treatment (Fig. 2e). Also, ULK1 loss on AMBRA1 deficiency is proteasome dependent (Supplementary Fig. S1e). These results confirm that AMBRA1 is involved in ULK1 protein stability.
Next, to verify whether the ULK1 gel mobility shift, observed on AMBRA1 downregulation, was due to an impairment of ULK1 kinase activity, we investigated ATG13 phosphorylation (pATG13), a ULK1mediated event. In AMBRA1-shRNA-treated cells, the pATG13 level decreases with respect to the control (Fig. 2d). In contrast, on AMBRA1 overexpression, the ATG13 phosphorylation level increases and an in vitro kinase assay shows that ULK1 kinase activity is, indeed, higher (Fig. 2f,g). However, by analysing the interaction between ULK1 and two known ULK1-binding partners, ATG13 and mTOR, we observed that the ULK1 complex is normally formed in AMBRA1-deficient cells (Supplementary Fig. S1f). Further, as previously reported, ULK1 forms, on autophagy induction, specific puncta within the cytosol12; although fewer ULK1 puncta could be seen in AMBRA1-deficient cells, highlighting a general decrease in ULK1 levels, no consequences could be detected on its mobilization and co-localization with LC3 puncta during autophagy (Supplementary Fig. S1g).

AMBRA1 is important for ULK1 Lys-63-linked ubiquitylation

Promptedbythesefindings,weanalysedthemolecularstepsinvolvedin this regulation. Post-translational modifications such as Lys-63-linked ubiquitylation often regulate kinase activity and stability13–15. In particular, ULK1 is a Lys-63-linked ubiquitylated protein during axon growth and endocytosis in the sensory growth cones of the nervous system16. How exactly polyubiquitylation affects the autophagy regulatory functions of ULK1 and whether this modification is sensitive to autophagy induction remain unknown.
Therefore, we investigated whether ULK1 underwent Lys-63-linked ubiquitylation in our cellular system and found that this was the case (Fig. 3a and Supplementary Fig. S2a). Furthermore, overexpression of a K63R ubiquitin mutant (unable to form Lys-63-linked chains) blocked the polyubiquitylation of ULK1, whereas a K48R mutant (unable to form K48-linked chains), as well as wild-type ubiquitin, did not (Fig. 3b), thus confirming that the polyubiquitylation we observed on ULK1 depends on Lys-63-linked ubiquitin. Furthermore, AMBRA1-siRNA-treated cells exhibit a decrease of Lys-63-linked ubiquitin in the ULK1 immunoprecipitates (Fig. 3c), while the amount of ULK1-linked ubiquitin is larger in cells transfected with AMBRA1 than in untransfected cells (Fig. 3d), indicating that AMBRA1 is required for ULK1 Lys-63-linked ubiquitylation.
Next, we analysed whether this modification on ULK1 was sensitive to autophagy induction. We found a rapid and transient increase in the Lys-63-linked ubiquitylation level of ULK1 using two different stimuli (rapamycin treatment and nutrient deprivation; Fig. 3e,f). Indeed, after autophagy induction by starvation, we observed a precise co-localization between ULK1 and Lys-63-linked ubiquitins (Supplementary Fig. S2b). Finally, we found that in the presence of a construct encoding ubiquitin Lys-63-linked chains, there was an appreciable increase in ULK1 levels and in the ATG13 phosphorylation level (Supplementary Fig. S2c). This supports our idea that this modification enhances ULK1 stability and activity. Of note, it is reasonable to believe that the population of Lys-63-ubiquitylated ULK1 isthatwhichisreleasedfromthemTORcomplexduringautophagy2.

AMBRA1 regulates ULK1 ubiquitylation by forming a complex with the TRAF6-ubiquitin ligase

We next attempted to identify the E3 ligase that could mediate this modification on ULK1. The E3 ligase TRAF6 can, indeed, ubiquitylate ULK1 in vitro16, and human TRAF6 is linked to the induction of autophagy in macrophages (through a physical interaction with beclin 1), resulting in its Lys-63-linked ubiquitylation17. In light of this evidence, we measured the expression levels of p62 and LC3 in TRAF6-overexpressing cells, observing a strong induction of autophagy on-rate, both in basal and starvation conditions (Fig. 4a and Supplementary Fig. S3a). However, when TRAF6 overexpression is coupled to simultaneous depletion of AMBRA1, the capability of TRAF6 to induce autophagy is reduced in all conditions analysed (Fig. 4b,c and Supplementary Fig. S3b), demonstrating that the TRAF6 pro-autophagic effect may depend, at least in part, on the presence of AMBRA1. Such a finding led us to examine whether TRAF6 was able to bind ULK1 and AMBRA1. As shown in Fig. 4d, FLAG–TRAF6 can bind AMBRA1 and ULK1 both in basal and starvation conditions and the three factors show vast subcellular co-localization (Supplementary Fig. S3c). Interestingly, when AMBRA1 is downregulated, the capability of TRAF6 to interact with ULK1 is strongly affected (Fig. 4e), proving that AMBRA1 is essential for the TRAF6–ULK1 interaction.
Data are presented as means ± s.d. and significance is P = 0.006 for ULK1 and P = 0.044 and P = 0.011 for pATG13. (e) HeLa cells were transfected or not with vector encoding AMBRA1 and treated with 35µM cycloheximide (CHX) for 6 and 8h before extraction. ULK1, actin and AMBRA1 were analysed by western blotting. Quantification is shown below. n=3 extracts prepared from independent experiments. Data are presented as means±s.d. and significance is P =0.049 and P =0.02. (f) HEK293 cells were transfected as in a; ATG13, pATG13, ULK1, AMBRA1 and actin were analysed by western blotting. (g) Endogenous ULK1 is immunoprecipitated from HeLa cells transfected with β-galactosidase or AMBRA1 plasmid respectively using an anti-ULK1 antibody. As a control, rabbit IgG was used. The resulting precipitates were subjected to ULK1 kinase assay using MBP as the substrate. Purified complexes were analysed by autoradiography and by western blotting using an anti-ULK1 antibody. Total extracts were analysed using an anti-AMBRA1 antibody. Uncropped images of blots/gels are shown in Supplementary Fig. S6. Source data of statistical analysis are shown in Supplementary Table S1.
Next, we investigated whether TRAF6-mediated modification of ULK1 was as important as AMBRA1 deficiency for ULK1 protein stability. The depletion of TRAF6 promotes a massive proteasomedependent reduction of ULK1, while TRAF6 overexpression enhances ULK1 levels (Fig. 4f and Supplementary Fig. S3a,d,f). These results demonstrate that both TRAF6 and AMBRA1 are required for ULK1 stability. In contrast, TRAF6 modulation has no such effect on the levels of other autophagy-related proteins, such as ATG13 or AMBRA1 (Fig. 4f).
Furthermore, the amount of ULK1-linked ubiquitin is larger in cells transfected with FLAG–TRAF6 than in control cells (Fig. 4g), the implication being that TRAF6 can ubiquitylate ULK1 in vivo. Interestingly, AMBRA1 downregulation led to a significant decrease in the capability of TRAF6 to ubiquitylate ULK1 (Fig. 4h), proving that AMBRA1isnecessaryto mediateULK1ubiquitylationbyTRAF6.
Last, we analysed ULK1 activation after TRAF6 overexpression by a kinase assay in vitro and found that, after TRAF6 overexpression, ULK1 proved more active (Fig. 4i). However, the capability of TRAF6 to induce ULK1 activation is impaired after AMBRA1 downregulation (Fig. 4i), as also confirmed by monitoring ULK1 gel mobility and ATG13 phosphorylation levels (Supplementary Fig. S3e–g). Taken together, these findings demonstrate that, by promoting its ubiquitylation, both TRAF6 and AMBRA1 are required to enhance ULK1 phosphorylation.

AMBRA1–TRAF6 interaction is necessary for ULK1 Lys-63-linked ubiquitylation

By analysing the amino-acid sequence of human AMBRA1 we found five putative TRAF6-binding sites (P-X-E-X-X-aromatic/acidic)18 located at residues 618–623, 640–645, 681–686, 916–921 and 1,132–1,137, respectively. We mutated the glutamate residue to alanine in each potential binding site, and observed that the two sites identical to the canonical TRAF6-binding site (residues 618–623 and 681–686) are essential for AMBRA1–TRAF6 interaction (Fig. 5a,b and Supplementary Fig. S4a,b). By generating a double-mutant construct (AMBRA1AA), we further confirmed the importance of these two sites in the AMBRA1–TRAF6 interaction, even though their mutation still allowsAMBRA1tobindULK1(Fig. 5candSupplementaryFig.S4c). We then examined the functional significance of AMBRA1–TRAF6 interaction in ULK1 ubiquitylation and activation. In the presence of the AMBRA1AA construct, we found that ULK1 ubiquitylation and activation are almost abolished (Fig. 5d,e and Supplementary Fig. S4d,e), supporting the hypothesis that AMBRA1–TRAF6 binding is essential for ULK1-linked ubiquitylation and for regulating ULK1 stability and activity. TRAF6 overexpression only slightly rescued this phenotype (Supplementary Fig. S4d).

ULK1 Lys-63-linked ubiquitylation is required for ULK1 self-association

As Lys-63-linked ubiquitylation is often linked to protein selfassociation19,20 and a recent report shows that Atg1, the yeast ULK1 orthologue, is able to self-associate under autophagy induction21, we checked for this process in our system. We found that ULK1 is able to self-associate in mammalian cells and that autophagy induction by starvation enhances this capability (Fig. 6a). In contrast, downregulation of TRAF6 (Fig. 6b) or AMBRA1 (Fig. 6c) reduces the amount of ULK1–Myc interacting with ULK1–HA: this strongly suggests that both proteins are necessary for ULK1 self-association. Consistently, in the presence of the AMBRA1AA mutant construct, ULK1self-associationisstronglyreduced(Fig. 6d),implyingthatULK1 ubiquitylation promotes ULK1 self-association.

mTOR phosphorylates AMBRA1 at Ser 52, inhibiting its role in ULK1 modification

Taken together, all of the results above indicate that AMBRA1 plays an upstream role in ULK1 regulation. On the basis of the observation that mTORC1 is a negative regulator of the ULK1 complex, and one of the moreupstreamfactorsinautophagysignalling,wereasonedthatmTOR could also modulate AMBRA1 activity. To identify potential AMBRA1 inhibitory modifications on autophagy induction, we performed a differential high-throughput mass spectrometry study (Supplementary Fig. S5a), by using cell extracts from starved and untreated cells. AMBRA1 was found to be phosphorylated in normal conditions at its Ser 52 residue, and the same site proved to be dephosphorylated on autophagy induction. Considering that the region surrounding this site is similar to the consensus phospho-acceptor motif for mTORC1 (ref. 22), we therefore postulated that mTOR could be responsible for this AMBRA1 phosphorylation in basal conditions, an event that would have an inhibitory role on autophagy (Supplementary Fig. S5b). Indeed, mutation of this site (AMBRA1S52A) decreased the level of phosphorylation of AMBRA1 by mTOR in vitro (Fig. 7a), whereas AMBRA1 is able to physically interact with the mTORC1 complex (Fig. 7b), suggesting that AMBRA1 could be an mTOR substrate and that Ser 52 could be an mTOR target site. Moreover, we generated an antibody against phospho-Ser-52 AMBRA1 (Supplementary Fig. S5c) and, by means of this antibody, we confirmed a decrease in the phosphorylation status of AMBRA1 at Ser 52 in cells cultured in the presence of Torin1 (Fig. 7c), an mTOR inhibitor23,24, or in nutrient-depleted media.
The functional role of this site in the capability of AMBRA1 to induce autophagy was then analysed by using LC3 and p62 antibodies, in the presence or the absence of chloroquine, an inhibitor of the autophagy flux (Fig. 7d). After AMBRA1WT reintroduction in AMBRA1-deficient Supplementary Fig. S3b. (d) HEK293 cells were transfected with AMBRA1 and FLAG–TRAF6 as indicated. Protein extracts were immunoprecipitated using an anti-FLAG antibody. (e) HeLa cells were transfected as in b and FLAG–TRAF6 was purified as in d. Quantification of immunoprecipitated TRAF6 is shown on the right. n =3 extracts prepared from independent experiments. Data are presented as means±s.d. and significance is P =0.0003. (f) In HeLa cells, TRAF6 was downregulated by RNAi. AMBRA1, ULK1, TRAF6, ATG13 and actin were analysed by western blotting. (g) HeLa cells were co-transfected with vectors encoding ULK1–Myc, FLAG–TRAF6 and Ub-Lys-63–HA, as indicated. Protein extracts were immunoprecipitated using an anti-Myc antibody. (h) HeLa cells stably interfered for AMBRA1 were transfected with FLAG–TRAF6 and Ub-Lys-63–HA as indicated. Protein extracts were immunoprecipitated using anti-ULK1 antibody. (i) Endogenous ULK1 is immunoprecipitated from HeLa cells transfected as in b. The resulting precipitates were subjected to a ULK1 kinase assay using MBP as the substrate. Purified complexes were analysed by autoradiography and by western blotting using an anti-ULK1 antibody. Uncropped images of blots/gels are shown in Supplementary Fig. S6. Source data of statistical analysis are shown in Supplementary Table S1. HeLa cells, a strong induction of autophagy can be observed, as indicated by LC3 lipidation and p62 decrease. However, the addition of AMBRA1S52A in the same system has a stronger effect, suggesting that the dephosphorylation of this site contributes to the induction of autophagy. We also transfected HeLa cells with an AMBRA1S52E phospho-mimicking mutant construct, again in an AMBRA1-deficient context. In this case, the capability of AMBRA1 to induce autophagy was significantly reduced (Supplementary Fig. S5d).
To highlight the effect of this construct in ex vivo primary cells, we reconstituted Ambra1-deficient MEFs with AMBRA1S52A and measured the occurrence of GFP–LC3-positive puncta, in comparison with AMBRA1WT,obtaininganevenstrongerautophagyinduction(Fig. 7e). Then, the functional effect of Ser 52 phosphorylation in ULK1 modifications was analysed in the same reconstitution systems. As shown in Fig. 7f, in the presence of AMBRA1S52A, ULK1 ubiquitylation is more abundant, compared with the expression of AMBRA1WT. Furthermore, on transfection of the same construct, the level of ATG13 phosphorylation is increased, together with ULK1 protein levels and the capability of ULK1 to self-associate (see total extracts in Fig. 7f–h and Supplementary Fig. S5e). Our conclusion, therefore, is that AMBRA1 phosphorylation by mTOR at Ser 52 impairs ULK1 ubiquitylation, stability, activity and self-association, besides inhibiting autophagy.

DISCUSSION

We have shown here that AMBRA1, a component of the beclin 1 complex, regulates the activation and stability of the kinase ULK1, on the right. n =3 extracts prepared from independent experiments. Data are presented as means±s.d. and significance is P =0.037. (c) HEK293 cells were transfected with AMBRA1 RNAi oligonucleotide (AMBRA1 siRNA) or unrelated oligonucleotides as a control (Ctrl siRNA). Then, cells were co-transfected and analysed as in a. Quantification is shown on the right. n =3 extracts prepared from independent experiments. Data are presented as means±s.d. and significance is P =0.044. (d) HEK293 cells were co-transfected with vectors encoding for ULK1–HA, ULK1–Myc and AMBRA1WT or AMBRA1AA respectively. Protein complexes were immunoprecipitated and analysed as in a. Quantification is shown on the right. n =3 extracts prepared from independent experiments. Data are presented as means±s.d. and significance is P =0.014. Uncropped images of blots/gels are shown in Supplementary Fig. S6. Source data of statistical analysis are shown in Supplementary Table S1.
by promoting its Lys-63-linked ubiquitylation. Our data indicate that, on autophagy induction, ULK1 is rapidly Lys-63-ubiquitylated by an AMBRA1–TRAF6 complex, a key step in ULK1 self-association and in the enhancement of its kinase activity. In the absence of autophagy signalling, mTOR downregulates these series of events through a specific inhibitory phosphorylation on AMBRA1 (see Fig. 8).
Overall, these results imply that a much closer cross-talk between ULK1 and beclin 1 complexes exists long before their hierarchical localization at the omegasome on autophagy induction25. Indeed, the exocyst, a protein scaffold complex involved in tethering intracellular vesicles to the plasma membrane, has been previously shown to provide an assembly and activation platform for members of the autophagy machinery. In this higher complex, all subcomplexes, including ULK1, beclin 1 and mTOR coexist, their interaction and functional reciprocal control being regulated by ad hoc protein modifications26. mTOR, a master regulator of a cell life27, is one of the more upstream negative regulators of autophagy: this protein, in response to the nutrient levels availabletothecell,caninhibitULK1andATG13byphosphorylation2,4. We here demonstrate that mTOR exerts a further brake on autophagy, by directly phosphorylating AMBRA1, thereby inhibiting its action on ULK1. AMBRA1 is thus one of the few known functional targets of mTOR and one of the more upstream autophagy signallers. innate immune response by regulating the Toll-like pathway. In Interestingly, and according to its secondary and tertiary putative particular, in this signalling pathway, TRAF6 ubiquitylates some sequence, AMBRA1 can be considered as an intrinsically disordered protein kinases such as TAK1 and IKKγ, leading to their subsequent protein28 (data not shown). These proteins are excellent candidates activation29. In our case, we found that TRAF6 overexpression for the role of scaffold factors, linking different molecules and driving enhances ULK1 activity and that the capability of TRAF6 to modify different cell processes, a capability that may explain the multi-task role ULK1 is affected when AMBRA1 is downregulated. Interestingly, of AMBRA1 at different steps during autophagy induction. AMBRA1 has been identified as DDB1- and cullin4-associated factor
As for the role of Lys-63-linked ubiquitylation in autophagy, we show 3 (DCAF3), and proposed to bridge this ligase complex to its target here that this modification is an early and crucial step of autophagy substrates30,31, supporting the idea that AMBRA1 could act as a TRAF6 induction and is mediated by TRAF6. TRAF6-dependent Lys-63-linked adaptor during autophagy. In fact, TRAF6 needs adaptor proteins ubiquitylation of beclin 1 was shown to be essential for beclin 1 to ubiquitylate its substrates, such as in the regulation of TAK1 pro-autophagic function17. TRAF6 is also known to be involved in kinase through the TAB1–TAB2 adaptor complex in the IL-1 signal embryos were transduced with indicated lentiviral vectors and quantification of LC3 puncta per cell is shown on the right. Bars represent mean±s.e.m. of triplicate samples and 50 cells analysed per sample. n =3 independent experiments. Significance is P =0.001 and P =0.003. Scale bar, 10µm. (f) HeLa cells stably interfered for AMBRA1 were co-transfected with vectors encoding Ub-Lys-63–HA and AMBRA1WT or AMBRA1S52A, respectively. Protein extracts were immunoprecipitated using an anti-ULK1 antibody and ubiquitin Lys-63–HA, ULK1, AMBRA1 and actin were analysed by western blotting. (g) HeLa cells stably interfered for AMBRA1 were transduced as in d; protein extracts were analysed by western blotting using anti-AMBRA1, anti-ATG13, anti-pATG13 and anti-actin antibodies. Quantification of the p-ATG13 band is shown below. (h) HeLa cells transfected with ULK1–HA, ULK1–Myc and different AMBRA1 constructs, respectively. Protein complexes were immunoprecipitated using anti-HA antibody and analysed by western blotting using anti-HA, anti-Myc and anti-AMBRA1 antibodies. Uncropped images of blots/gels are shown in Supplementary Fig. S6. Source data of statistical analysis are shown in Supplementary Table S1.
transduction pathway32. Furthermore, it remains unknown how the failure in Lys-63-linked ubiquitylation of ULK1 affects its protein stability. It has been demonstrated that an enzyme called A20, a deubiquitylase for TRAF6, could deubiquitylate Lys-63-linked chains of TRAF6-target proteins to trigger K48-linked ubiquitylation, leading to these proteins degradation. A20 will therefore be a good candidate to exert this function on ULK1 (ref. 33). Of note, it has been recently published that, in yeast, the ULK1 orthologue Atg1 forms an Atg1–Atg1 complex, stimulating the autophosphorylation of Atg1 and autophagy induction21. According to our data, a self-association that enhances ULK1 activity occurs also in higher eukaryotes, and depends on the AMBRA1–TRAF6 complex.
Of note, we have previously shown that AMBRA1 itself is one of the targets of the ULK1 kinase activity11. Indeed, this ULK1-mediated AMBRA1 phosphorylation has the potential to prime AMBRA1 detachment from cytoskeletal docking sites. Given the additional and more upstreamroleofAMBRA1thatwedescribehere,theULK1-dependentregulation of AMBRA1 localization may represent a positive feedback loop regulatingautophagyenhancement.Asfortheroleofthesetwoproteins in vivo, it should be noted that both AMBRA1 and ULK1 play crucial roles in the developing and adult brain compartments10,16,34. Although these roles have not yet been clearly related to the ULK1 pro-autophagic functions, our findings reveal that AMBRA1 may participate in a common pathway in neurogenesis, linked to autophagy regulation.

References

1. Mizushima, N. & Komatsu, M. Autophagy: renovation of cells and tissues. Cell 147, 728–741 (2012).
2. Hosokawa, N. et al. Nutrient-dependent mTORC1 association with the ULK1-Atg13FIP200 complex required for autophagy. Mol. Biol. Cell. 20, 1981–1991 (2009).
3. Chan, E. Y., Longatti, A., McKnight, N. C. & Tooze, S. A. Kinase-inactivated ULK proteins inhibit autophagy via their conserved C-terminal domains using an Atg13-independent mechanism. Mol. Biol. Cell. 29, 157–171 (2009).
4. Jung, C. H. et al. ULK-Atg13-FIP200 complexes mediate mTOR signaling to the autophagy machinery. Mol. Biol. Cell 20, 1992–2003 (2009).
5. Hara, T. et al. FIP200, an ULK-interacting protein, is required for autophagosome formation in mammalian cells. J. Cell Biol. 181, 497–510 (2008).
6. Hokosawa, N. et al. Atg101, a novel mammalian autophagy protein interacting with Atg13. Autophagy 5, 973–979 (2009).
7. Ganley, I. G. et al. ULK1–ATG13-FIP200 complex mediates mTOR signaling and is essential for autophagy. J. Biol. Chem. 284, 12297–12305 (2009).
8. Mercer, C. A., Kaliappan, A. & Dennis, P. B. A novel, human Atg13 binding protein, Atg101, interacts with Ulk1 and is essential for macroautophagy. Autophagy 5, 649–662 (2009).
9. Hayashi-Nishino, M. et al. A subdomain of the endoplasmic reticulum forms a cradle for autophagosome formation. Nat. Cell Biol. 11, 1433–1437 (2009).
10. Fimia, G. M. et al. Ambra1 regulates autophagy and development of the nervous system. Nature 447, 1121–1125 (2007).
11. Di Bartolomeo, S. et al. The dynamic interaction of AMBRA1 with the dynein motor complex regulates mammalian autophagy. J. Cell Biol. 191, 155–168 (2010).
12. Chan, E. Y., Kir, S. & Tooze, S. A. siRNA screening of the kinome identifies ULK1 asa multidomain modulator of autophagy. J. Biol. Chem. 282, 25464–25474 (2007). 13. Chen, Z. J. & Sun, L. J. Non proteolytic functions of ubiquitin in cell signaling. Mol. Cell 33, 275–286 (2009).
14. Hoeller, D. & Dikic, I. Targeting the ubiquitin system in cancer therapy. Nature 458, 438–444 (2009).
15. Raiborg, C. & Stenmark, H. The ESCRT machinery in endosomal sorting ofubiquitylated membrane proteins. Nature 458, 445–445 (2009).
16. Zhou, X. et al. Unc-51-like kinase 1/2-mediated endocytic processes regulate filopodia extension and branching of sensory axons. Proc. Natl Acad. Sci. USA 104, 5842–5847 (2007).
17. Shi, C. S. & Kehrl, J. H. Traf6 and A20 differentially regulate TLR4-inducedautophagy by affecting the ubiquitination of Beclin 1. Sci. Signal 3, 123 (2010).
18. Ye, H. et al. Distinct molecular mechanism for initiating TRAF6 signalling. Nature 418, 443–447 (2002).
19. Liu, Z., Zhang, W. P., Xing, Q., Ren, X., Liu, M. & Tang, C. Noncovalent dimerizationof ubiquitin. Angew. Chem. Int. Ed. Engl. 51, 469–472 (2012).
20. Humphrey, R. K., Yu, S. M., Bellary, A., Gonuguntla, S., Yebra, M. & Jhala, U. S.Lysine 63-linked ubiquitination modulates mixed lineage kinase-3 interaction with JIP1 scaffold protein in cytokine-induced pancreatic β cell death. J. Biol. Chem. 288, 2428–2440 (2012).
21. Yeh, Y. Y., Shah, K. H. & Herman, P. K. An Atg13 protein-mediated self-associationof the Atg1 protein kinase is important for the induction of autophagy. J. Biol. Chem. 286, 28931–28939 (2011).
22. Hsu, P. P. et al. The mTOR-regulated phosphoproteome reveals a mechanism of mTORC1-mediated inhibition of growth factor signaling. Science 332, 1317–1322 (2011).
23. Feldman, M. E. et al. Active-site inhibitors of mTOR target rapamycin-resistant outputs of mTORC1 and mTORC2. PLoS Biol. 7, e38 (2009).
24. Thoreen, C. C. et al. An ATP-competitive mammalian target of rapamycin inhibitor reveals rapamycin-resistant functions of mTORC1. Biol. Chem. 284, 8023–8032 (2009).
25. Itakura, E. & Mizushima, N. Characterization of autophagosome formation siteby a hierarchical analysis of mammalian Atg proteins. Autophagy 6, 764–776 (2010).
26. Bodemann, B. O. et al. RalB and the exocyst mediate the cellular starvation VPS34 inhibitor 1 response by direct activation of autophagosome assembly. Cell 144, 253–267 (2011).
27. Hosking, R. mTOR: the master regulator. Cell 149, 955–957 (2012).
28. Dyson, H. J. & Wright, P. E. Intrinsically unstructured proteins and their functions.Nat. Rev. Mol. Cell Biol. 6, 253–267 (2005).
29. Sorrentino, A. et al. The type I TGF-β receptor engages TRAF6 to activate TAK1 in a receptor kinase-independent manner. Nat. Cell Biol. 10, 1199–1207 (2008).
30. Jin, J., Arias, E. E., Chen, J., Harper, J. W. & Walter, J. C. A family of diverseCul4-Ddb1-interacting proteins includes Cdt2, which is required for S phase destruction of the replication factor Cdt1. Mol. Cell 23, 709–721 (2006).
31. Behrends, C., Sowa, M. E., Gygi, S. P. & Harper, J. W. Network organization of thehuman autophagy system. Nature 466, 68–76 (2010).
32. Takaesu, G. et al. TAB2, a novel adaptor protein, mediates activation of TAK1 MAPKKK by linking TAK1 to TRAF6 in the IL-1 signal transduction pathway. Mol. Cell 5, 649–658 (2000).
33. Wertz, I. E. et al. De-ubiquitination and ubiquitin ligase domains of A20 downregulate NF-κB signalling. Nature 430, 694–699 (2004).
34. Tomoda, T., Bhatt, R. S., Kuroyanagi, H., Shirasawa, T. & Hatten, M. E. A mouseserine/threonine kinase homologous to C. elegans UNC51 functions in parallel fiber formation of cerebellar granule neurons. Neuron 24, 833–846 (1999).