A Double Negative Feedback Loop Between Mtorc1 and AMPK Kinases Guarantees Precise Autophagy Induction Upon Cellular Stress

International Journal of Molecular Sciences

Article A Double Negative Feedback Loop between mTORC1 and AMPK Kinases Guarantees Precise Autophagy Induction upon Cellular Stress

Marianna Holczer 1, Bence Hajdú 1, Tamás L˝orincz 2 , András Szarka 2,Gábor Bánhegyi 1,3 and Orsolya Kapuy 1,* 1 Department of Medical Chemistry, Molecular Biology and Pathobiochemistry, Semmelweis University, 1085 Budapest, Hungary; [email protected] (M.H.); [email protected] (B.H.); [email protected] (G.B.) 2 Laboratory of Biochemistry and Molecular Biology, Department of Applied Biotechnology and Food Science, Budapest University of Technology and Economics, 1111 Budapest, Hungary; [email protected] (T.L.); [email protected] (A.S.) 3 Pathobiochemistry Research Group of the Hungarian Academy of Sciences and Semmelweis University, 1085 Budapest, Hungary * Correspondence: [email protected]; Tel.: +36-1266-2615

Received: 11 October 2019; Accepted: 6 November 2019; Published: 7 November 2019

Abstract: Cellular homeostasis is controlled by an evolutionary conserved cellular digestive process called autophagy. This mechanism is tightly regulated by the two sensor elements called mTORC1 and AMPK. mTORC1 is one of the master regulators of proteostasis, while AMPK maintains cellular energy homeostasis. AMPK is able to promote autophagy by phosphorylating ULK1, the key inducer of autophagosome formation, while mTORC1 downregulates the self-eating process via ULK1 under nutrient rich conditions. We claim that the feedback loops of the AMPK–mTORC1–ULK1 regulatory triangle guarantee the appropriate response mechanism when nutrient and/or energy supply changes. In our opinion, there is an essential double negative feedback loop between mTORC1 and AMPK. Namely, not only does AMPK downregulate mTORC1, but mTORC1 also inhibits AMPK and this inhibition is required to keep AMPK inactive at physiological conditions. The aim of the present study was to explore the dynamical characteristic of AMPK regulation upon various cellular stress events. We approached our scientific analysis from a systems biology perspective by incorporating both theoretical and molecular biological techniques. In this study, we confirmed that AMPK is essential to promote autophagy, but is not sufficient to maintain it. AMPK activation is followed by ULK1 induction, where protein has a key role in keeping autophagy active. ULK1-controlled autophagy is always preceded by AMPK activation. With both ULK1 depletion and mTORC1 hyper-activation (i.e., TSC1/2 downregulation), we demonstrate that a double negative feedback loop between AMPK and mTORC1 is crucial for the proper dynamic features of the control network. Our computer simulations have further proved the dynamical characteristic of AMPK–mTORC1–ULK1 controlled cellular nutrient sensing.

Keywords: mTOR; AMPK; autophagy; systems biology; double negative feedback

1. Introduction The maintenance of cellular homeostasis against external and internal stimuli (such as starvation, inﬂammatory mediators) is controlled by complex protein–protein regulatory pathways. The energy and food supply of a cellular system is directly measured by the AMP/ATP ratio [1,2]. When the ATP level is high, the conditions are optimal for cellular growth, while increasing the level of AMP/ATP

Int. J. Mol. Sci. 2019, 20, 5543; doi:10.3390/ijms20225543 www.mdpi.com/journal/ijms Int. J. Mol. Sci. 2019, 20, x FOR PEER REVIEW 2 of 17

When the ATP level is high, the conditions are optimal for cellular growth, while increasing the level of AMP/ATP quickly blocks the anabolic processes of the cell [2,3]. This mechanism is tightly Int. J. Mol. Sci. 2019, 20, 5543 2 of 17 regulated by the two sensor elements of nutrient conditions and energy supply: one of the mTOR (mammalian target of rapamycin) complexes and AMPK (AMP activated protein kinase) kinase [4– 6]. quickly blocks the anabolic processes of the cell [2,3]. This mechanism is tightly regulated by the two mTOR issensor a serine/threonine elements of nutrient protein conditions kinase and and energy is the supply: main one component of the mTOR of (mammalian mTOR target of pathway-regulatedrapamycin) cellular complexes proteostasis and AMPK by integrating (AMP activated different protein signals kinase) such kinase as growth [4–6]. factors, amino acids, glucosemTOR, and is a serineenergy/threonine status to protein control kinase growth and is and the main metabolism component [7,8 of]. mTORFor proper pathway-regulated function, mTORcellular kinas proteostasise has to form by a integrating complex, mTORC1, different signalswith other such regulatory as growth subunits factors, aminosuch as acids, glucose, regulatory-associatedand energy protein status of to mTOR control (Raptor), growth mammalian and metabolism lethal [7 with,8]. For SEC13 proper protein function, 8 (MLST8), mTOR kinase has to PRAS40, andform DEPTOR a complex, [9]. mTORC1,Under physiological with other regulatory conditions subunits, mTORC1 such becomes as regulatory-associated active by a protein of phosphatidylinositolmTOR (Raptor), 3 kinase mammalian (PI3K)–Akt lethal kinase with-depend SEC13 proteinent manner, 8 (MLST8), while PRAS40, the TSC1 and DEPTOR-TSC2 [9]. Under (hamartin-tuberin)physiological complexconditions, is a critical mTORC1upstream becomesnegative activeregulator bya of phosphatidylinositol mTORC1 [8,10]. Inhibition 3 kinase (PI3K)–Akt of mTORC1 resultskinase-dependent in a drastic block manner, of protein while the synthesis TSC1-TSC2 by de (hamartin-tuberin)-phosphorylating the complex crucial istargets a critical upstream of mTOR suchnegative as ribosomal regulator protein of mTORC1 S6 kinase [8(p70S6K),10]. Inhibition and translation of mTORC1 initiation results factor in a 4E drastic binding block of protein protein-1 (4E-synthesisBP1) [8,10 by]. de-phosphorylating the crucial targets of mTOR such as ribosomal protein S6 kinase Besides (p70S6K) mTORC1, and AMPK translation tightly initiation controls factor ATP- 4Econsuming binding protein-1 processes (4E-BP1) such as [8 ,10 gly].cogen or protein syntheses,Besides fatty acids mTORC1, and cholesterol AMPK tightly syntheses controls, and ATP-consuming upregulates processes such that as increase glycogen or protein ATP (i.e., glycolysis,syntheses, β-oxidation) fatty acids [11,12 and] cholesterol by sensing syntheses, energy levels and to upregulates maintain cellular processes homeostasis that increase ATP (i.e., [13]. AMPK isglycolysis, a heterotrimericβ-oxidation) protein [11 complex,12] by sensingand it is energy fairly sensitive levels to maintainto the change cellular in the homeostasis cellular [13]. AMPK AMP/ATP ratio.is a When heterotrimeric the AMP proteinlevel is complexhigh in the and cell, it isthe fairly free sensitiveAMP directly to the binds change to inAMPK the cellular and AMP/ATP turns it on [2].ratio. AMPK When becomes the AMP active level when is high it is inphosphorylated the cell, the free at AMP the Thr directly-172 residue binds to [14 AMPK]. and turns it on [2]. Both AMPKAMPK and becomes mTORC1 active control when it cellular is phosphorylated homeostasis at the via Thr-172 an evolutionary residue [14 ]. conserved digestive process Both called AMPK autophagy and mTORC1 [7,15,16 control]. During cellular autophagy homeostasis, cellular via an components evolutionary become conserved digestive sequestered intoprocess a double called membrane autophagy vesicle,[7,15,16 ]. called During an autophagy,autophagosome, cellular whose components contents become are then sequestered into delivered to anda double degraded membrane by the vesicle,lysosome called [17,18 an].autophagosome, Cells always have whose some contents residual are autophagic then delivered to and activity even degradedunder physiological by the lysosome conditions; [17,18 ].however, Cells always the process have some becomes residual more autophagic efficient during activity even under various stressphysiological events (i.e., immune conditions; signals, however, starvation, the process and growth becomes factor more depr efficientivation) during to promote various stress events the degradation(i.e., immuneof several signals, cellular starvation, components and growth [19,20 factor]. Although deprivation) autophagy to promote seems the degradationto be a of several cyto-protectivecellular process components by recycling [19 ,20 the]. harmful Although and/or autophagy unnecessary seems to components be a cyto-protective of the cell, process an by recycling excessive levelthe of harmfulautophagy and might/or unnecessary induce cell components death [21,22 of]. theOne cell, of the an excessivekey inducers level of of the autophagy complex might induce required for autophagosomecell death [21,22 ].formation One of theis Unc key- inducers51 like autophagy of the complex activating required kinase for (ULK1/2) autophagosome [23]. formation Knock-down isof Unc-51ULK1 is like able autophagy to inhibit activating autophagy kinase activation (ULK1 in/2) several [23]. Knock-downcell lines [13,23,24 of ULK1]. After is able to inhibit starvation, ULK1/2autophagy, together activation with Fip200 in several and cellAtg13 lines, forms [13,23 a, 24complex]. After that starvation, is able to ULK1 become/2, together active with Fip200 through auto-andphosphorylatio Atg13, formsn [ a13 complex]. This proc thatess is in able turn to leads become to translocation active through of auto-phosphorylationthe entire complex [13]. This to the pre-autophagosomalprocess in turn membrane leads to translocation and to autophagy of the entire induction complex [13]. to the pre-autophagosomal membrane and AMPK andto autophagy mTORC1 induction are known [13 ]. to oppositely control the autophagy inducing complex ULK1/2-Fip200-Atg13AMPK [13] and. Under mTORC1 nutrient are rich known condition to oppositelys, mTORC1 control directly the dow autophagynregulate inducings the complex self-eating processULK1 / by2-Fip200-Atg13 phosphorylating [13]. various Under Ser nutrient-residues rich on conditions, ULK1 (such mTORC1 as Ser757) directly [13,16 downregulates]. In the contrast, AMPKself-eating has both processdirect and by indirect phosphorylating positive effects various on autophagy Ser-residues induction. on ULK1 On (such one hand, as Ser757) [13,16]. AMPK is ableIn to contrast, promote AMPK the self has-eating both mechanism direct and indirect by phosphorylating positive effects different on autophagy phosphor induction.ylation On one hand, sites on ULK1AMPK (e.g., isSer able555) to in promote response the self-eatingto nutrient mechanism limitation by[13,25 phosphorylating]. On the other di ffhanderent, phosphorylationAMPK sites inhibits the mTORon ULK1 complex (e.g., Ser555)either vi ina responsethe TSC1/2 to nutrientpathway limitation or by direct [13,25 phosphorylation]. On the other hand, of Raptor AMPK inhibits the [13,26–28]. mTOR complex either via the TSC1/2 pathway or by direct phosphorylation of Raptor [13,26–28]. In addition, In ULK1 addition, is able ULK1 to is inhibit able to the inhibit mTOR the mTOR complex complex throughout throughout its Raptor its Raptor subunit, subunit, generating generating a soa so called called double double negative negative feedback feedback loop loop (i.e (i.e.,., ULK1 ┤mTORC1 mTORC1 ┤ULK1) ULK1) [13 [13,29,30,29,30]. Recently,]. it has Recently, it hasalso also been been shown shown that ULK1 that ULK1 antagonizes antagonizes autophagy autophagy via downregulating via downregulating AMPK AMPKactivity [13,31]. This activity [13,31regulatory]. This regulatory connection connection betweenAMPK between and AMPK ULK1 results and ULK1 in a negative results feedback in a negative loop in the control feedback loopnetwork. in the Through control ULK1-dependent network. Through negative ULK1 regulation-dependent of its negative up-stream regulation regulators, of the its control network up-stream regulatorshelps to, fine-tunethe control the network cellular autophagichelps to fine response-tune the with cellular respond autophagic to cellular response stress eventswith [13,32]. respond to cellularTo stres understands events [13,32 the dynamic]. characteristics of the AMPK–mTORC1–ULK1 regulatory triangle, a computational analysis was performed by Szymanska et al. [33]. Their mathematical model was formulated to study the systems-level consequences of interactions among AMPK, mTORC1, and

ULK1 by mainly focusing on autophagy induction upon various stress events [33]. By changing Int. J. Mol. Sci. 2019, 20, x FOR PEER REVIEW 2 of 17

2.1. Downregulation of ULK1 Blocks AMPK Induction Recently, Szymanska et al. built a mathematical model based on the mTOR–AMPK–ULK1 (from this point mTOR refers to mTORC1) regulatory triangle to explain cellular autophagy induction upon rapamycin treatment or glucose starvation [33]. However, other well-documented mutant behaviors have not been studied with this model yet. To further investigate the systems-level properties of the control network, we created an ordinary differential equation-based mathematical model using their network as a template (see the colored boxes and black lines of the wiring diagram in Figure1). A double negative feedback has already been described between AMPK and ULK1 (see regulatory connections “a” and “b” in Figure1); furthermore, mTOR–ULK1 mutual antagonism is also well-known (see regulatory connections “c” and “d” ion Figure1). AMPK directly inhibits mTOR (see regulatory connection “e” in Figure1), which blocks autophagy (aka ATG) (see regulatory connection “f” in Figure1); meanwhile, ULK1 directly promotes ATG activation (see regulatory connection “g” in Figure1). We assume that when the ATG level is high, that the cellular self-cannibalism is active. By first running computer simulations, we reproduced glucose starvation and rapamycin treatment (the details of the mathematical code can be found in the Supplementary Materials and also in Figure S1A,B). Next, we tested whether this simple model was also able to describe many other well-known mutant behaviors (these results are summarized in Tables S1 and S2). However, we observed that neither modeling cellular ULK1 depletion (ULK1T set to almost zero), nor generating mTOR hyper-activation (2-fold increase of kamtor) could mimic the relevant biological experiment (Figure S1C,D and Table S2). Namely, both ULK1 depletion or mTOR hyper-activation resulted in a significant increase in the AMPK-P level in the model already at physiological conditions, but no AMPK phosphorylation was actually detected when siULK1 or siTSC1/siTSC2 (aka mTOR hyper-activation) was expressed in human cells [27,34–36]. Our preliminary data assumed that one or more important feedback(s) was still missing from the mTOR–AMPK–ULK1 regulatory triangle. To reveal the missing regulatory connection(s) of this simple control network, we first studied the ULK1 silencing experimentally (Figure2). ULK1 was directly silenced with siRNA in human embryonic kidney cells (HEK293T), and the efficiency of siULK1 was tested both on the mRNA (data not shown) and protein (Figure2) levels. To further confirm the e ffect of ULK1 silencing on autophagy induction, the expression of ULK1 siRNA was also followed by rapamycin (i.e., mTOR inhibition) treatment (100 nM, 2h) or starvation (carbohydrate-free medium, 6 h and 24 h) (i.e., AMPK hyper-activation). At the end of the treatments, the key indicators of autophagy (LC3 and p62) were detected by immunoblotting and the efficiency of the self-eating process was also checked by fluorescence microscope (Figure2A–C). Int. J. Mol. Sci. 2019, 20, x FOR PEER REVIEW 3 of 17

To understand the dynamic characteristics of the AMPK–mTORC1–ULK1 regulatory triangle, a computational analysis was performed by Szymanska et al. [33]. Their mathematical model was formulated to study the systems-level consequences of interactions among AMPK, mTORC1, and ULK1 by mainly focusing on autophagy induction upon various stress events [33]. By changing the intracellular level of either AMPK (simulating starvation) or mTORC1 (simulating rapamycin treatment) kinases, the regulatory network generated a graded response to stress inputs achieved by a bistable switch between autophagy and protein synthesis [33]. Although Szymanska et al. have already presented a testable mechanistic model of the AMPK– mTORC1–ULK1 regulatory network, they did not take into account many cellular stress events already proven by experimental methods. Here, we directly show that some mutant phenotypes cannot be explained by this simple model (such as ULK1 depletion by using siULK1, or mTORC1 hyper-activation by using siTSC1/2). We claim that mTORC1 is able to downregulate AMPK, generating an extra double negative feedback loop in the control network (i.e., AMPK ┤ mTORC1 ┤ AMPK). By using systems biology techniques, we confirmed that the feedback loops of the AMPK– mTORC1–ULK1 regulatory triangle guarantee the appropriate response mechanism of the cell upon a broad spectrum of external and internal cellular stress effects.

2. Results

Figure 1. The wiring diagram of the AMPK–mTOR–ULK1 regulatory triangle controlled stress response

mechanism. AMPK, mTOR, and ULK1 are denoted by isolated red, green and blue boxes, respectively. ATG (see purple box) refers to the autophagy activator complex. Dashed line shows how the components can inﬂuence each other, while blocked end lines denote inhibition. The feedback loop proven here both experimentally and theoretically is orange-colored.

After the addition of rapamycin or inducing starvation, a high LC3II/GAPDH ratio and significantly decreased p62 level were observed, suggesting that autophagy was activated and worked properly. When these treatments were preceded by ULK1 silencing, neither the LC3II/GAPDH ratio nor p62 level changed significantly, suggesting that functional autophagy was not detected (Figure2A,B). HEK293T cells were fixed and immunolabeled for endogenous LC3II/LC3I detection, where LC3 (referring to autophagy) could be observed in discrete foci (Figure2C). In the control cells treated with rapamycin, a more than 5-fold increase of autophagosomes was detected. In contrast, the relative amount of autophagosomes remained significantly low in both the absence of ULK1 and siULK1 combined with rapamycin treatment. These results fit our western blot analysis and further suggest that ULK1 is essential for autophagy induction. To explore the activation profile of both AMPK and mTOR in the absence of ULK1, AMPK phosphorylation and the key marker of active mTOR (such as p70S6K-P) was also followed by immunoblotting when ULK1 siRNA was expressed in the cells (Figure2D). Interestingly, AMPK phosphorylation did not show a significant reduction, while mTOR activity significantly increased when the treatments were carried out in human HEK293T cells expressing siULK1 (see the amount of p70S6K-P in Figure2D), suggesting that although ULK1-dependent inhibition on AMPK was missing, something else could keep it in its inactive de-phosphorylated state. Since autophagy was not detected when ULK1 silencing was combined with mTOR inactivation/AMPK hyper-activation, our results further confirm that ULK1 is essential for proper autophagy activation. We also claim that the AMPK-P level cannot increase in the absence of ULK1. Int. J. Mol. Sci. 2019, 20, x FOR PEER REVIEW 4 of 17

Figure 1. The wiring diagram of the AMPK–mTOR–ULK1 regulatory triangle controlled stress response mechanism. AMPK, mTOR, and ULK1 are denoted by isolated red, green and blue boxes, respectively. ATG (see purple box) refers to the autophagy activator complex. Dashed line shows how the components can influence each other, while blocked end lines denote inhibition. The feedback loop proven here both experimentally and theoretically is orange-colored.

By first running computer simulations, we reproduced glucose starvation and rapamycin treatment (the details of the mathematical code can be found in the Supplementary Materials and also in Figure S1A,B). Next, we tested whether this simple model was also able to describe many other well-known mutant behaviors (these results are summarized in Tables S1 and S2). However, we observed that neither modeling cellular ULK1 depletion (ULK1T set to almost zero), nor generating mTOR hyper-activation (2-fold increase of kamtor) could mimic the relevant biological experiment (Figure S1C,D and Table S2). Namely, both ULK1 depletion or mTOR hyper-activation resulted in a significant increase in the AMPK-P level in the model already at physiological conditions, but no AMPK phosphorylation was actually detected when siULK1 or siTSC1/siTSC2 (aka mTOR hyper-activation) was expressed in human cells [27,34–36]. Our preliminary data assumed that one or more important feedback(s) was still missing from the mTOR–AMPK–ULK1 regulatory triangle. To reveal the missing regulatory connection(s) of this simple control network, we first studied the ULK1 silencing experimentally (Figure 2). ULK1 was directly silenced with siRNA in human embryonic kidney cells (HEK293T), and the efficiency of siULK1 was tested both on the mRNA (data not shown) and protein (Figure 2) levels. To further confirm the effect of ULK1 silencing on autophagy induction, the expression of ULK1 siRNA was also followed by rapamycin (i.e., mTOR inhibition) treatment (100 nM, 2h) or starvation (carbohydrate-free medium, 6 h and 24 h) (i.e., AMPK hyper-activation). At the end of the treatments, the key indicators of autophagy (LC3 and p62) were detected by immunoblotting and the efficiency of the self-eating process was also Int.checked J. Mol. Sci.by2019 fluorescence, 20, 5543 microscope (Figures 2A–C). 5 of 17

Figure 2. The effect of ULK1 silencing on the AMPK level under various stress events. HEK293T cells were starved (for 6h or 24h) or treated with rapamycin (rap – rapamycin —100 nM, 2 h) without/with silencing of ULK1 by siRNA. (A) The markers of autophagy (LC3, p62) and ULK1 were followed by immunoblotting. GAPDH was used as the loading control. (B) Densitometry data represent the intensity of p62, ULK1, and LC3 II normalized for GAPDH. (C) The connection of ULK1 silencing and autophagy induction was checked by immunofluorescence microscopy. LC3 was stained by green fluorescence dye. Rapamycin (rap–rapamycin—100 nM, 2h) was used as the positive control. Quantification and statistical analysis of the immunofluorescence microscopy data. Error bars represent standard deviation, asterisks indicate statistically significant difference from the control: ns—nonsignificant; * p < 0.05; ** p < 0.01, (D). The markers of AMPK and mTOR (p70S6K-P) were followed in HEK293T cells without/with silencing of ULK1 by siRNA. Densitometry data represent the intensity p70S6K-P normalized for the total level of p70S6K and AMPK-Thr172-P normalized for the total level of AMPK. For each of the experiments, three independent measurements were carried out. Error bars represent standard deviation, asterisks indicate statistically significant difference from the control: ns—nonsignificant; * p < 0.05; ** p < 0.01.

2.2. Hyper-Activation of mTOR Blocks AMPK Induction To explore the role of mTOR kinase in autophagy regulation, mTOR was directly hyper-activated by depleting its stoichiometric inhibitors, called TSC1 and TSC2, respectively (Figure3). First, the mixture of siTSC1 and siTSC2 was expressed in HEK293T cells. Then, combined treatment was also carried out, when silencing of RNAs was followed by starvation (carbohydrate-free medium, 6 h and 24 h) or rapamycin addition (100 nM, 2 h). The efficiency of siTSC1 and siTSC2 was tested on both the mRNA (data not shown) and protein (Figure3A,B) levels. At the end of the treatments, the key markers of autophagy (i.e., p62, LC3II/GAPDH) were detected by immunoblotting and the efficiency of the self-eating process was also checked by fluorescence microscope (Figure3A–C). Int. J. Mol. Sci. 2019, 20, 5543 6 of 17 Int. J. Mol. Sci. 2019, 20, x FOR PEER REVIEW 6 of 17

FigureFigure 3. The 3. The effect effect of mTORof mTOR hyper-activation hyper-activation (by(by TSC1TSC1/2/2 silencing) silencing) on on AMPK AMPK level level under under various various stressstress events. events. HEK293T HEK293T cells cells were were starved starved (for (for 6h6h or 24h) or or treated treated with with rapamycin rapamycin (rap (rap – rapamycin – rapamycin —100—100 nM, 2nM, h) without2 h) without/with/with silencing silencing of TSC1of TSC1/2/2 by siRNA.by siRNA. (A )(A The) The markers markers of of autophagy autophagy (LC3,(LC3, p62), mTORp62), (p70S6K-P), mTOR (p70S6K-P), TSC1, and TSC1, TSC2 and wereTSC2 followedwere followed by immunoblotting. by immunoblotting. GAPDH GAPDH waswas used as as the loadingthe control. loading (control.B) Densitometry (B) Densitometry data represent data represent the intensity the inte ofnsity p62 LC3of p62 II, TSC1,LC3 II, and TSC1, TSC2 and normalized TSC2 for GAPDH,normalized and for p70S6K-P GAPDH, and normalized p70S6K-P fornormalized the total for level the total of p70S6K. level of p70S6K. (C) The (C connection) The connection of ULK1 silencingof ULK1 and silencing autophagy and induction autophagy was induction checked was by checked immunofluorescence by immunofluorescence microscopy. microscopy. LC3 was LC3 stained was stained by green fluorescence dye. Rapamycin (rap – rapamycin —100 nM, 2h) was used as the by green fluorescence dye. Rapamycin (rap – rapamycin —100 nM, 2h) was used as the positive positive control. Quantification and statistical analysis of immunofluorescence microscopy data. control. Quantification and statistical analysis of immunofluorescence microscopy data. Error bars Error bars represent standard deviation, asterisks indicate statistically significant difference from the representcontrol: standard ns—nonsignificant; deviation, asterisks* p < 0.05; indicate ** p statistically< 0.01, D) The significant markers di ffoference AMPK from and theULK1 control: ns—nonsignificant;(ULK1-Ser555-P) * pwere< 0.05; followed ** p < 0.01,in HEK293T (D) The cells markers without/with of AMPK silencing and ULK1 of TSC1/2 (ULK1-Ser555-P) by siRNA. were followedDensitometry in HEK293T data cells represent without the/with intensity silencing and ofAMPK-Thr172-P TSC1/2 by siRNA. normalized Densitometry for the datatotal representlevel of the intensityAMPK and and AMPK-Thr172-P ULK1-Ser555-P normalized normalized for for the the total total leve levell of ULK1. of AMPK For each and of ULK1-Ser555-P the experiments, normalized three for theindependent total level ofmeasurements ULK1. For eachwere of ca therried experiments, out. Error bars three represent independent standard measurements deviation, asterisks were carried out. Errorindicate bars statistically represent significant standard difference deviation, from asterisks the control: indicate ns—nonsignificant; statistically significant * p < 0.05; di** ffp erence< 0.01. from the control: ns—nonsignificant; * p < 0.05; ** p < 0.01. Silencing of mTOR inhibitors resulted in a successful hyper-activation of mTOR (see the Silencingenhanced ofphosphorylation mTOR inhibitors of p70S6K resulted in inFigures a successful 3A,B). hyper-activationThe moderate ratio of of mTOR LC3II/GAPDH (see the enhanced and phosphorylationthe high level of of p70S6K p62 suggested in Figure that3A,B). autophagy The moderate could not ratiobe active of LC3II (Figures/GAPDH 3A,B). andThis theobservation high level of coincided with the assumption that mTOR has a drastic negative effect on autophagy induction. p62 suggested that autophagy could not be active (Figure3A,B). This observation coincided with the Autophagy induction was also analyzed by using immunofluorescence microscopy with and assumption that mTOR has a drastic negative effect on autophagy induction. Autophagy induction was without TSC1/2 silencing (Figure 3C). Rapamycin-treatment showed an almost 5-fold increase of also analyzed by using immunofluorescence microscopy with and without TSC1/2 silencing (Figure3C). Rapamycin-treatment showed an almost 5-fold increase of autophagosomes, while in the absence of TSC1/2, no significant amount of discrete foci was detected. These results suppose that hyper-active mTOR blocks autophagy induction. Interestingly, siTSC1/2 combined with rapamycin treatment Int. J. Mol. Sci. 2019, 20, x FOR PEER REVIEW 2 of 17

When the ATP level is high, the conditions are optimal for cellular growth, while increasing the level of AMP/ATP quickly blocks the anabolic processes of the cell [2,3]. This mechanism is tightly regulated by the two sensor elements of nutrient conditions and energy supply: one of the mTOR (mammalian target of rapamycin) complexes and AMPK (AMP activated protein kinase) kinase [4– 6]. mTOR is a serine/threonine protein kinase and is the main component of mTOR pathway-regulated cellular proteostasis by integrating different signals such as growth factors, amino acids, glucose, and energy status to control growth and metabolism [7,8]. For proper function, mTOR kinase has to form a complex, mTORC1, with other regulatory subunits such as regulatory-associated protein of mTOR (Raptor), mammalian lethal with SEC13 protein 8 (MLST8), PRAS40, and DEPTOR [9]. Under physiological conditions, mTORC1 becomes active by a phosphatidylinositol 3 kinase (PI3K)–Akt kinase-dependent manner, while the TSC1-TSC2 (hamartin-tuberin) complex is a critical upstream negative regulator of mTORC1 [8,10]. Inhibition of mTORC1 results in a drastic block of protein synthesis by de-phosphorylating the crucial targets of mTOR such as ribosomal protein S6 kinase (p70S6K) and translation initiation factor 4E binding protein-1 (4E-BP1) [8,10]. Besides mTORC1, AMPK tightly controls ATP-consuming processes such as glycogen or protein syntheses, fatty acids and cholesterol syntheses, and upregulates processes that increase ATP (i.e., glycolysis, β-oxidation) [11,12] by sensing energy levels to maintain cellular homeostasis Int. J. Mol. Sci. 2019, 20, 5543 7 of 17 [13]. AMPK is a heterotrimeric protein complex and it is fairly sensitive to the change in the cellular AMP/ATP ratio. When the AMP level is high in the cell, the free AMP directly binds to AMPK and turns it on [2]. AMPK becomesresulted active when in a modestit is phosphorylated increase of autophagosomesat the Thr-172 residue (Figure [143]C).. These data fit to the immunoblot Both AMPK and mTORC1analysis, control where cellular significant homeostasis increase was via observed an evolutionary in LC3II/GAPDH conserved ratio upon treatment (Figure3A,B), digestive process called autophagysupposing [7,15,16 that artificial]. During downregulation autophagy, cellular of mTOR components by rapamycin become addition might win against silencing sequestered into a double membraneTSC1/2. vesicle, called an autophagosome, whose contents are then delivered to and degraded by theIn lysosome the next [17,18 step,] we. Cells also always tested the have possible some residual influence autophagic of mTOR kinase on AMPK activity by activity even under physiologicalfollowing conditions; the phosphorylation however, the process status ofbecomes AMPK more in cells efficient expressing during siTSC1 and siTSC2 (Figure3D). various stress events (i.e., immuneAlthough signals, ULK1-controlled starvation, and inhibition growth onfactor AMPK depr wasivation) completely to promote missing due to mTOR-dependent the degradation of several downregulation cellular components of ULK1 [19,20 (see]. Although regulatory autophagy connections seems “c” and to “d” be ain Figure1), AMPK remained cyto-protective process by recyclingde-phosphorylated, the harmful supposing and/or unnecessary that AMPK components was also inhibited of the cell, (Figure an 3D). In addition, the AMPK- excessive level of autophagy mightdependent induce phosphorylation cell death [21,22 of]. ULK1One of at the Ser555 key inducers was also of not the observed complex (Figure 3D), further supposing required for autophagosome thatformation ULK1 wasis Unc not-51 active. like autophagy activating kinase (ULK1/2) [23]. Knock-down of ULK1 is able to inhibitTaken together,autophagy these activation results confirm in several that cell mTOR lines upregulation [13,23,24]. keepsAfter autophagy inactive. Although starvation, ULK1/2, together withULK1 Fip200 activity and was Atg13 not, observed forms a complex due to the that hyper-activation is able to become of mTOR active in the presence of siTSC1 and through auto-phosphorylationsiTSC2, [13]. This AMPK proc remainsess in turn completely leads to translocation inactive. Since of only the entire mTOR complex is active in this mTOR–AMPK–ULK1 to the pre-autophagosomal membraneregulatory and triangle, to autophagy we assume induction that mTOR [13]. has some direct or indirect negative effect on AMPK in the AMPK and mTORC1 areabsence known of ULK1. to oppositely control the autophagy inducing complex ULK1/2-Fip200-Atg13 [13]. Under nutrient rich conditions, mTORC1 directly downregulates the self-eating process by phosphorylating2.3. The Proper various Dynamical Ser-residues Characteristic on ULK1 of ULK1 (such Depletion as Ser757) Could [13,16 Be Managed]. In by a Double Negative Feedback Loop between mTOR and AMPK contrast, AMPK has both direct and indirect positive effects on autophagy induction. On one hand, AMPK is able to promote the self-eatingOur results mechanism have shown by phosphorylating that phosphorylation-dependent different phosphor activationylation of AMPK was not observed sites on ULK1 (e.g., Ser555) inwhen response either siULK1 to nutrient or siTSC1 limitation/siTSC2 [13,25 were]. expressedOn the other in the hand cells., AMPK Since ULK1-dependent inhibition was inhibits the mTOR complex eithernot present, via the but TSC1/2 mTOR pathway remained or active by direct in both phosphorylation treatments (see Figuresof Raptor2 and 3), we supposed that mTOR [13,26–28]. might act negatively on AMPK induction. To investigate the importance of this mTOR-dependent In addition, ULK1 is ableinhibition to inhibit of AMPK the mTOR on the dynamic complex properties throughout of our its Raptor control network, subunit, we extended our model with generating a so called doublethis negative negative feedback connection loop (i.e.,(i.e., mTORULK1 ┤AMPK, mTORC1 see ┤ the ULK1) orange [13,29,30 colored]. line of the wiring diagram in Recently, it has also been shownFigure that1). Thereby, ULK1 antagonizes an extra double autophagy negative via feedback downregulating loop was generated AMPK between mTOR and AMPK activity [13,31]. This regulatory connection between AMPK and ULK1 results in a negative (AMPK mTOR AMPK). feedback loop in the control network. Through ULK1-dependent negative regulation of its Mimicking starvation (i.e., induction of AMPK phosphorylation) through computer simulations, up-stream regulators, the control network helps to fine-tune the cellular autophagic response with AMPK quickly became active and phosphorylated ULK1 (Figure S2A). The active AMPK and ULK1 respond to cellular stress events [13,32]. together induced autophagy, however, AMPK-P was later downregulated due to ULK1-dependent de-phosphorylation (Figure S2A). In the absence of ULK1 (i.e., silencing of ULK1 by siRNA was mimicked by setting ULK1T = 0.001), the active mTOR level remained high, and due to the

mTOR-dependent AMPK inhibition, it was now able to keep AMPK in its inactive dephosphorylated form in this model (Figure S2B). Our computer simulations also nicely ﬁt with our experimental data when mTOR hyper-activation was generated by increasing mTOR activity (kamtor = 0.05). Although hyper-activation of mTOR rapidly downregulated ULK1, AMPK did not become active because mTOR also had a negative eﬀect on AMPK (Figure S2C). However, in the absence of the mTOR AMPK connection, AMPK became active when ULK1 was depleted or mTOR was hyper-activated (see Figure S1A,B), further suggesting the importance of mTOR-dependent downregulation of AMPK. Our theoretical analysis suggests that mTOR can inhibit AMPK. We suppose that the double negative feedback loop generated by mTOR and AMPK in the control network has a key role in keeping AMPK inactive at physiological conditions. We claim that the level of AMPK-P remains low until mTOR is present.

2.4. AMPK Activation Always Precedes ULK1 Induction during Autophagic Stress Response Since both ULK silencing and mTOR hyper-activation (via TSC1/TSC2 silencing) were carried out under normal growth conditions, we cannot rule out that AMPK did not become active due to unchanged energy status of the cell, rather than via the mTOR-dependent control mechanism. Int. J. Mol.Int. Sci. J. Mol.2019 Sci., 20 ,2019 5543, 20, x FOR PEER REVIEW 8 of 17 8 of 17

Therefore, to further confirm the importance of the mTOR-dependent downregulation of AMPK-P, Therefore,the AMPK to further phosphorylation conﬁrm the importance profile was of followed the mTOR-dependent in time when mTOR downregulation was inhibited. of AMPK-P, HEK293T cells the AMPKwere phosphorylation treated with 100 proﬁle nM rapamycin was followed and the in time time when-dependency mTOR was of the inhibited. key markers HEK293T (i.e., cellsautophagy, were treatedAMPK, with mTOR, 100 nM and rapamycin ULK1) andwere the followed time-dependency both experimentally of the key markers (via (i.e.,immunoblotting) autophagy, and AMPK,theoretically mTOR, and ULK1)(via computer were followed simulations) both experimentally (Figure 4). (via immunoblotting) and theoretically (via computer simulations) (Figure4).

Figure 4. AMPK induction precedes ULK1 activation upon rapamycin treatment. HEK293T cells were Figure 4. AMPK induction precedes ULK1 activation upon rapamycin treatment. HEK293T cells denoted in time after 100 nM rapamycin treatment. (A) The markers of autophagy (LC3, p62), AMPK, were denoted in time after 100 nM rapamycin treatment. (A) The markers of autophagy (LC3, p62), ULK1 (ULK1-Ser555-P), and mTOR (p70S6K-P, 4E-BP1-P) were followed by immunoblotting. GAPDH AMPK, ULK1 (ULK1-Ser555-P), and mTOR (p70S6K-P, 4E-BP1-P) were followed by was used as the loading control. (B) Densitometry data represent the intensity of p62 and LC3 II immunoblotting. GAPDH was used as the loading control. (B) Densitometry data represent the normalized for GAPDH, ULK1-Ser555-P normalized for the total level of ULK1, p70S6K-P normalized intensity of p62 and LC3 II normalized for GAPDH, ULK1-Ser555-P normalized for the total level of for the total level of p70S6K, 4E-BP1-P normalized for the total level of 4E-BP1, and AMPK-Thr172-P ULK1, p70S6K-P normalized for the total level of p70S6K, 4E-BP1-P normalized for the total level of normalized for the total level of AMPK. For each of the experiments, three independent measurements 4E-BP1, and AMPK-Thr172-P normalized for the total level of AMPK. For each of the experiments, were carried out. Error bars represent standard deviation, asterisks indicate statistically significant three independent measurements were carried out. Error bars represent standard deviation, asterisks difference from the control: ns—nonsignificant; * p < 0.05; ** p < 0.01. (C) Computer simulation of indicate statistically significant difference from the control: ns—nonsignificant; * p < 0.05; ** p < 0.01. rapamycin treatment. The relative activity of AMPK-Thr172-P, mTOR, ULK1-Ser555-P and autophagy (C) Computer simulation of rapamycin treatment. The relative activity of AMPK-Thr172-P, mTOR, (ATG) are plotted in time. ULK1-Ser555-P and autophagy (ATG) are plotted in time. Upon rapamycin treatment, the mTOR targets (i.e., p70S6K, 4E-BP1) were quickly downregulated. The active phosphorylatedUpon rapamycin form treatment, of p70S6K the disappeared, mTOR targets while (i.e., the thirdp70S6K, phosphorylation 4E-BP1) were band quickly of 4E-BP1-Pdownregulated. appeared, suggestingThe active the phosphorylated complete inactivation form of mTORp70S6K (Figure disappeared,4). AMPK-P while could the third be observedphosphorylation when active band mTOR of 4E-BP1-P disappeared appeared, in the cell,suggesting further the confirming complete that inactivation mTOR might of mTOR have some(Figure negative 4). AMPK-P effects oncould AMPK-P. be observed Parallel when with theactive inactivation mTOR disappeared of mTOR, AMPK in the became cell, further hyper-phosphorylated.confirming that mTOR AMPK-P might was ablehave to some promote negative the phosphorylation effects on AMPK-P. and activationParallel with of ULK1 the inactivation (see ULK1-Ser555-Pof mTOR, in Figure AMPK4) within became a certain hyper-phosphoryl delay. However,ated. after two-hourAMPK-P long was rapamycin able to treatment, promote the AMPK-Pphosphorylation disappeared due and to ULK1-dependentactivation of ULK1 downregulation. (see ULK1-Ser555-P Autophagy in Figure became 4) activewithin when a certain both delay. AMPK-PHowever, and active after ULK1 two-hour were long present rapamycin in the cell treatm (Figureent,4 AMPK-P). Although disappeared the ratio of due LC3II to ULK1-dependent/GAPDH increaseddownregulation. after 30 min-long Autophagy rapamycin became treatment, active p62 when significantly both AMPK-P decreased and only active when ULK1 ULK1 were became present in the cell (Figure 4). Although the ratio of LC3II/GAPDH increased after 30 min-long rapamycin

Int. J. Mol. Sci. 2019, 20, 5543 9 of 17 phosphorylated (Figure4). These results suggest that a two-hour long rapamycin treatment is required for dramatic activation of autophagy in HEK293T cells. Interestingly, autophagy remained active at the end of the rapamycin treatment, despite the ULK1-dependent downregulation of AMPK-P (Figure4), supposing that AMPK-P is not essential to maintain autophagy. Our results demonstrate that AMPK can be active if mTOR is downregulated, and its activity always precedes ULK1 induction. Although autophagy becomes signiﬁcantly active only when both AMPK-P and ULK1-P are present in the cell, AMPK-P is not essential to keep the cellular self-cannibalism active.

2.5. AMPK Is Not Sufficient to Induce Autophagy during mTOR-Downregulation Our results clearly suggest that AMPK remains de-phosphorylated until mTOR is active, even in the absence of ULK1 upon various cellular stress events. In order to further confirm this connection, the time-dependency of rapamycin treatment was detected in the absence of ULK1. First, ULK1 gene expression was silenced by siRNA, and then rapamycin treatment (100 nM, for 2 h) was carried out. The key markers (i.e., autophagy, AMPK, mTOR, and ULK1) were followed both experimentally (via immunoblotting) and theoretically (via computer simulations) (Figure5). Rapamycin drastically decreased the activity of mTOR after 30 min of treatment. AMPK quickly became significantly phosphorylated one and a half hours later, and remained active in the absence of ULK1 when mTOR was inhibited. These results further confirm that mTOR is crucial to keep AMPK in its inactive de-phosphorylated form when ULK1 is not present in the cell. However, no autophagy induction was observed at all (see the low LC3II/GAPDH ratio and the high p62 level in Figure5B,C). Our computer simulations showed that the level of AMPK-P became permanently high while mTOR completely disappeared and ULK1 was silenced (via decreasing the total level of ULK1) (Figure5D). AMPK became hyper-phosphorylated, suggesting that its negative regulators (i.e., mTOR and ULK1) were missing from the control network. Although AMPK seemed to be active and mTOR was inactive, in the absence of ULK1, no autophagy induction was detected. Taken together, we can conclude that the AMPK-P level increases when both ULK1 and mTOR are missing from the cell. Although AMPK becomes active, alone is not sufficient to induce autophagy when both ULK1 and mTOR are inhibited. The control network definitively requires ULK1 for autophagy induction. Int. J. Mol. Sci. 2019, 20, x FOR PEER REVIEW 9 of 17 treatment, p62 significantly decreased only when ULK1 became phosphorylated (Figure 4). These results suggest that a two-hour long rapamycin treatment is required for dramatic activation of autophagy in HEK293T cells. Interestingly, autophagy remained active at the end of the rapamycin treatment, despite the ULK1-dependent downregulation of AMPK-P (Figure 4), supposing that AMPK-P is not essential to maintain autophagy. Our results demonstrate that AMPK can be active if mTOR is downregulated, and its activity always precedes ULK1 induction. Although autophagy becomes significantly active only when both AMPK-P and ULK1-P are present in the cell, AMPK-P is not essential to keep the cellular self-cannibalism active.

2.5. AMPK Is Not Sufficient to Induce Autophagy during mTOR-Downregulation Our results clearly suggest that AMPK remains de-phosphorylated until mTOR is active, even in the absence of ULK1 upon various cellular stress events. In order to further confirm this connection, the time-dependency of rapamycin treatment was detected in the absence of ULK1. First, ULK1 gene expression was silenced by siRNA, and then rapamycin treatment (100 nM, for 2 h) was carried out. The key markers (i.e., autophagy, AMPK, mTOR, and ULK1) were followed both experimentally (via immunoblotting) and theoretically (via computer simulations) (Figure 5). Int. J. Mol. Sci. 2019, 20, 5543 10 of 17

Figure 5. Silencing of ULK1 arrests autophagy induction upon rapamycin treatment. ULK1 was Figuresilenced 5. by Silencing siRNA in of HEK293T ULK1 arrests cells, followedautophagy by 100induction nM rapamycin upon rapamycin treatment treatment. in time. (A )ULK1 Test of was the silencedULK1 silencing by siRNA by immunoblotting,in HEK293T cells, total followed level ofby ULK1 100 nM and rapamycin ULK1-Ser555-P treatment were in followed. time. (A GAPDH) Test of wasthe ULK1 used as silencing the loading by control.immunoblotting, Densitometry total data level represent of ULK1 the intensityand ULK1-Ser555-P of ULK1 and were ULK1-Ser555-P followed. GAPDHnormalized was for used GAPDH as the and loading ULK1-Ser555-P control. Densitom normalizedetry fordata total represent level of the ULK1. intensity (B) The of markersULK1 and of ULK1-Ser555-Pautophagy (LC3, normalized p62), AMPK, for andGAPDH mTOR and (p70S6K-P, ULK1-Ser 4E-BP1-P)555-P normalized were followed for total by level immunoblotting. of ULK1. (B) TheGAPDH markers was usedof autophagy as the loading (LC3, control. p62), AMPK, (C) Densitometry and mTOR data (p70S6K-P, represent 4E-BP1-P) the intensity were of p62followed and LC3 by immunoblotting.II normalized for GAPDH,GAPDH p70S6K-Pwas used normalized as the loading for the control. total level (C) of Densitometry p70S6K, 4E-BP1-P data normalizedrepresent the for intensitythe total levelof p62 of 4E-BP1,and LC3 and II normalized AMPK-Thr172-P for GAPD normalizedH, p70S6K-P for the normalized total level offor AMPK. the total For level each of p70S6K,the experiments, 4E-BP1-P three normalized independent for the measurements total level of 4E-BP1, were carried and AMPK-Thr172-P out. Error bars representnormalized standard for the deviation, asterisks indicate statistically significant difference from the control: ns—nonsignificant; * p < 0.05; ** p < 0.01. (D) Computer simulation of rapamycin treatment. The relative activity of AMPK-Thr172-P, mTOR, ULK1-Ser555-P and autophagy (ATG) are plotted in time.

2.6. AMPK Is Essential to Induce Autophagy during mTOR-Downregulation To further explore the role of AMPK in autophagy induction, we checked the time-dependency of rapamycin treatment, where human cells were pre-treated by Compound C (also called dorsomorphin). Compound C is a widely used cell permeable pyrrazolopyrimidine compound that inhibits a number of kinases including AMPK in a dose-dependent manner [37]. HEK293T cells were pre-treated with 2 µM Compound C for 0.5 h followed by the addition of 100 nM rapamycin, while AMPK, mTOR, ULK1, and autophagy markers were detected by immunoblotting (Figure6A,B). Compound C completely blocked AMPK activation during the whole treatment, therefore no AMPK phosphorylation was observed. Due to the addition of rapamycin, mTOR was also inhibited (see the intensive dephosphorylation of p70S6K and the appearance of the third phosphorylation band of 4E-BP1-P in Figure6A,B). In the absence of AMPK-P, no ULK1-Ser555-P phosphorylation was detected, suggesting that ULK1 also remained inactive. The low LC3II/GAPDH ratio and the high p62 level suggest that autophagy is inactive during the treatment. Int. J. Mol.Int. Sci. J. Mol.2019 ,Sci.20, 2019 5543, 20, x FOR PEER REVIEW 11 of 17 11 of 17

Figure 6. Inhibition of AMPK arrests ULK1 induction upon rapamycin treatment. HEK293T cells were Figure 6. Inhibition of AMPK arrests ULK1 induction upon rapamycin treatment. HEK293T cells pre-treated with the AMPK inhibitor, Compound C (2 µM, 0.5 h), followed by 100 nM rapamycin were pre-treated with the AMPK inhibitor, Compound C (2 µM, 0.5 h), followed by 100 nM treatment in time. (A) The markers of autophagy (LC3, p62), AMPK, ULK1 (ULK1-Ser555-P), and mTOR rapamycin treatment in time. (A) The markers of autophagy (LC3, p62), AMPK, ULK1 (p70S6K-P, 4E-BP1-P) were followed by immunoblotting. GAPDH was used as the loading control. (ULK1-Ser555-P), and mTOR (p70S6K-P, 4E-BP1-P) were followed by immunoblotting. GAPDH was (B) Densitometry data represent the intensity of p62 and LC3 II normalized for GAPDH, ULK1-Ser555-P used as the loading control. (B) Densitometry data represent the intensity of p62 and LC3 II normalized for the total level of ULK1, p70S6K-P normalized for the total level of p70S6K, 4E-BP1-P normalized for GAPDH, ULK1-Ser555-P normalized for the total level of ULK1, p70S6K-P normalized for the total level of 4E-BP1, and AMPK-Thr172-P normalized for the total level of normalized for the total level of p70S6K, 4E-BP1-P normalized for the total level of 4E-BP1, and AMPK. For each of the experiments, three independent measurements were carried out. Error bars AMPK-Thr172-P normalized for the total level of AMPK. For each of the experiments, three represent standard deviation, asterisks indicate statistically significant difference from the control: independent measurements were carried out. Error bars represent standard deviation, asterisks ns—nonsignificant; * p < 0.05; ** p < 0.01. (C) Computer simulation of rapamycin treatment. The relative activityindicate of AMPK-Thr172-P, statistically mTOR,significant ULK1-Ser555-P, difference from and the autophagy control: (ATG)ns—nonsignificant; are plotted in * time. p < 0.05; ** p < 0.01. (C) Computer simulation of rapamycin treatment. The relative activity of AMPK-Thr172-P, mTOR, When theULK1-Ser555-P, combined addition and autophagy of both (ATG) Compound are plotted C (viain time. decreasing the rate constant of AMPK activation) and mTOR inhibitor (via decreasing the total level of mTOR) was theoretically tested, ULK1 remainedWhen inactive the combined (Figure 6additionC). Although of both mTOR-dependent Compound C (via inhibition decreasing on ULK1the rate was constant missing, of AMPK ULK1 couldactivation) not be and active mTOR because inhibitor its activation (via decreasing term requires the total AMPK. level Since of mTOR) neither was active theoretically AMPK nor tested, active ULK1ULK1 wasremained detected, inactive autophagy-induced (Figure 6C). Although cellular survivalmTOR-dependent could not becomeinhibition active on ULK1 either. was missing, BothULK1 experimental could not andbe active theoretical because results its activation indicate thatterm AMPK requires activation AMPK. isSinc essentiale neither to induceactive AMPK autophagynor byactive promoting ULK1 was ULK1 detected, phosphorylation autophagy-induced with respect cellular to various survival cellular could stress not events,become even active in either. the absenceBoth of mTOR. experimental and theoretical results indicate that AMPK activation is essential to induce autophagy by promoting ULK1 phosphorylation with respect to various cellular stress events, even 3. Discussionin the absence of mTOR. The maintenance of cellular homeostasis is mainly dependent on the ability of cells to take 3. Discussion precise action with respect to various stimuli (e.g., nutrient deprivation). An evolutionary conserved digestive process,The maintenance called autophagy, of cellular has ahomeostasis key role in recycling is mainly the dependent damaged oron unnecessarythe ability of cellular cells to take componentsprecise of theaction cell. with Autophagy respect is tightlyto various controlled stimuli by two(e.g., sensors nutrient of nutrient deprivation). conditions: An mTORevolutionary and AMPKconserved kinases. digestive mTOR isprocess, the master called regulator autophagy, of proteostasis has a key by integratingrole in recycling different the external damaged or unnecessary cellular components of the cell. Autophagy is tightly controlled by two sensors of

Int. J. Mol. Sci. 2019, 20, x FOR PEER REVIEW 2 of 17

When the ATP level is high, the conditions are optimal for cellular growth, while increasing the level of AMP/ATP quickly blocks the anabolic processes of the cell [2,3]. This mechanism is tightly regulated by the two sensor elements of nutrient conditions and energy supply: one of the mTOR (mammalian target of rapamycin) complexes and AMPK (AMP activated protein kinase) kinase [4– 6]. mTOR is a serine/threonine protein kinase and is the main component of mTOR pathway-regulated cellular proteostasis by integrating different signals such as growth factors, amino acids, glucose, and energy status to control growth and metabolism [7,8]. For proper function, mTOR kinase has to form a complex, mTORC1, with other regulatory subunits such as regulatory-associated protein of mTOR (Raptor), mammalian lethal with SEC13 protein 8 (MLST8), PRAS40, and DEPTOR [9]. Under physiological conditions, mTORC1 becomes active by a phosphatidylinositol 3 kinase (PI3K)–Akt kinase-dependent manner, while the TSC1-TSC2 (hamartin-tuberin) complex is a critical upstream negative regulator of mTORC1 [8,10]. Inhibition of mTORC1 results in a drastic block of protein synthesis by de-phosphorylating the crucial targets of mTOR such as ribosomal protein S6 kinase (p70S6K) and translation initiation factor 4E binding protein-1 (4E-BP1) [8,10]. Besides mTORC1, AMPK tightly controls ATP-consuming processes such as glycogen or protein syntheses, fatty acids and cholesterol syntheses, and upregulates processes that increase ATP (i.e., glycolysis, β-oxidation) [11,12] by sensing energy levels to maintain cellular homeostasis [13]. AMPK is a heterotrimeric protein complex and it is fairly sensitive to the change in the cellular AMP/ATP ratio. When the AMP level is high in the cell, the free AMP directly binds to AMPK and turns it on [2]. AMPK becomes active when it is phosphorylated at the Thr-172 residue [14]. Both AMPK and mTORC1 control cellular homeostasis via an evolutionary conserved digestive process called autophagy [7,15,16]. During autophagy, cellular components become sequestered into a double membrane vesicle, called an autophagosome, whose contents are then delivered to and degraded by the lysosome [17,18]. Cells always have some residual autophagic activity even under physiological conditions; however, the process becomes more efficient during Int. J. Mol. Sci. 2019, 20, 5543 12 of 17 various stress events (i.e., immune signals, starvation, and growth factor deprivation) to promote the degradation of several cellular components [19,20]. Although autophagy seems to be a cyto-protective process by recycling the harmfuland internal and/or signals, unnecessary while AMPK components senses cellular of the cell, energy an status [4–6]. Both kinases control the cellular excessive level of autophagy might induce cellself-digesting death [21,22 process]. One of via the ULK1 key inducers kinase, one of the of the complex inducers of autophagy activator complex [7,15,16]. required for autophagosome formation is UncMany-51 scientificlike autophagy results have activating already kinase revealed (ULK1/2) the important [23]. feedback loops of the mTOR–AMPK–ULK1 Knock-down of ULK1 is able to inhibit autophagyregulatory activation triangle in [13 several]. In addition, cell lines a mathematical[13,23,24]. After model was also built to theoretically describe starvation, ULK1/2, together with Fip200 andautophagy Atg13, forms induction a complex with that respect is able to rapamycin to become treatmentactive (mTOR inhibition) or starvation (AMPK through auto-phosphorylation [13]. This procactivation)ess in turn [33 leads]. However, to translocation we claim of that the an entire essential complex feedback loop is still missing, therefore we re-wired to the pre-autophagosomal membrane and tothe autophagy mTOR–AMPK–ULK1-controlled induction [13]. network (Figure1) and also tested its dynamic characteristics in AMPK and mTORC1 are known to response oppositely to variouscontrol stress the autophagy events incorporating inducing complexboth theoretical and molecular biological techniques. ULK1/2-Fip200-Atg13 [13]. Under nutrient rich Our condition systemss, mTORC1 biological directly approach dow clearlynregulate confirmeds the that AMPK cannot be active when ULK1 is self-eating process by phosphorylating varioussilenced Ser-residues by siRNA on (Figure ULK12 (such and Figure as Ser757) S2) or [13,16 mTOR]. In is hyper-activated (via TSC1 /TSC2 silencing) contrast, AMPK has both direct and indirect(Figure positive3 andeffects Figure on autophagy S2), although induction. the main On inhibitor one hand, of AMPK (i.e., ULK1) is completely diminished AMPK is able to promote the self-eating mechanismfrom the cells.by phosphorylating In both mutant different phenotypes, phosphor only mTORylation was active, suggesting that mTOR has a direct sites on ULK1 (e.g., Ser555) in response to or nutrient indirect limitation negative e[13,25ffect on]. AMPK.On the Here,other wehand also, AMPK revealed that rapamycin-dependent downregulation inhibits the mTOR complex either via the TSC1/2of mTOR pathway resulted or inby AMPK direct phosphorylation (Figureof Raptor4), further supposing that mTOR might have a [13,26–28]. negative effect on AMPK activity directly or indirectly. Since AMPK-dependent inhibition of mTOR is In addition, ULK1 is able to inhibit alreadythe mTOR well-known complex [13 throughout,27], an extra itsdouble Raptor negative subunit, feedback loop might be generated between mTOR generating a so called double negative feedbackand AMPK loop (i.e (see., AMPKULK1 ┤mTOR mTORC1AMPK ┤ ULK1) in Figure [13,29,301). This]. so called bistable toggle switch is a common Recently, it has also been shown that ULK1feature antagonizes of biochemical autophagy regulatory via downregulating networks describing AMPK an irreversible cellular transition between activity [13,31]. This regulatory connectiontwo between well-separated AMPK stable and ULK1 states results in the control in a negative network [38]. Namely, at physiological conditions, feedback loop in the control network. ThroughmTOR is ULK1 active-dependent and AMPK negative is inactive, regulation while AMPK of its quickly becomes activated in the absence of up-stream regulators, the control network helpsnutrients to fine or- energy.tune the In cellular addition, autophagic the induction response of AMPK with results in mTOR inhibition and also promotes respond to cellular stress events [13,32]. autophagy-dependent cellular survival. Since mTOR, the catalytic element of mTORC1, is a Ser/Thr protein kinase, we suppose that mTOR downregulates AMPK via inhibitory phosphorylation. Therefore, we identified potential Ser and Thr

phosphorylation sites on AMPK with NetPhos 3.1 (invented by the Department of Bio and Health Informatics, Technical University of Denmark, Lyngby, Denmark), which is a freely available software, to predict mTOR-dependent serine and threonine phosphorylation sites on AMPK [39]. We found two Ser (Ser-232, Ser-496) and two Thr residues (Thr-356, Thr-488) on AMPK; these phosphorylation motifs represented high similarities to the phosphorylation site preferred by mTOR kinase (for details see Supplementary Materials). This analysis further suggests that mTOR might be able to phosphorylate these Ser and Thr residues on AMPK, supposing a regulatory connection between them. However, this connection must later be proven experimentally. Although AMPK regulation is already well known, as is its effect on autophagy induction, the precise dynamic characteristics of AMPK-P regulation have not been studied yet. Therefore, here we show, with both experimental and theoretical analysis, the exact role of AMPK in enhancing cellular self-cannibalism. Our data clearly verified that AMPK-P is essential (Figure6), but is alone not sufficient (Figure5) to induce autophagy with respect to cellular stress. When rapamycin treatment (i.e., mTOR downregulation) was combined with AMPK-P inhibition (by using Compound C), cells could not turn on autophagy (Figure6), and ULK1 silencing by siRNA completely blocked autophagy induction even when AMPK-P was present in the absence of mTOR (Figure5). Since a transient (but not significant) autophagy induction was observed when the addition of Compound C was combined with rapamycin treatment, but ULK1 remained inactive, we further confirmed that AMPK is crucial to promote autophagy induction via ULK1 phosphorylation. To clarify the essential role of AMPK in ULK1 induction, the combined treatment of Compound C (AMPK inhibition) and rapamycin (mTOR inhibition) should be extended by artificial activation of ULK1. If autophagy is active under these experimental conditions, we will be able to prove whether the main role of AMPK-P is to switch on ULK1 upon cellular stress. Time-dependent mTOR inhibition (via rapamycin treatment) showed that AMPK phosphorylation is observed after mTOR downregulation and always precedes ULK1 induction, while ULK1 is also required to switch on autophagy (Figure4). Namely, when both AMPK-P and active ULK1 are present Int. J. Mol. Sci. 2019, 20, 5543 13 of 17 in the cell, the self-cannibalism becomes active. Interestingly, the active ULK1 quickly downregulates AMPK in the case of rapamycin treatment (see the quick de-phosphorylation of AMPK after 90 min long rapamycin treatment in Figure4), while autophagy remains active. For the proper dynamic characteristics of the control network, this ULK1-dependent downregulation of AMPK assumes a time-delayed negative feedback loop between AMPK and ULK1, otherwise the control system would generate a homeostatic response with respect to cellular stress. Although the negative effect of ULK1 on AMPK activity has already been proven [31], the exact molecular mechanism of this time-delay requires further clarification. We suppose that the de-phosphorylation of AMPK-P is crucial to prevent the hyper-activation of self-cannibalism, which might have a dramatic effect on the cell. Similar results have been already observed in the case of NRF2-dependent downregulation of the AMPK-induced autophagy response during oxidative stress [40,41]. However, these assumptions must be also tested both experimentally and theoretically in the future. By running computer simulations, our new model of the mTOR–AMPK–ULK1 regulatory triangle was able to describe more than 15 mutant phenotypes (Tables S1 and S2), generating the first comprehensive model of the control system. The great advantage of our systems biological approach is that our theoretical analysis does not only describe experimentally already proven cellular treatments, but it is also able to predict the unknown characteristics of the various control elements during treatments. For example, we supposed that siULK1 expression followed by the addition of resveratrol had similar characteristics to the combined treatment of siULK1 + rapamycin (Tables S1 and S2). Our computer simulations also suggest that if TSC1/TSC2 silencing is combined with rapamycin treatment, AMPK is significantly phosphorylated (Figure S2D, Tables S1 and S2) and autophagy also has some activity upon this combined treatment (Figure3). The active AMPK is able to induce ULK1, suggesting that mTOR downregulation is essential to AMPK activation. Interestingly, hyper-activated mTOR kept inactive AMPK, ULK1, and autophagy when TSC1/TSC2 silencing was combined with the addition of resveratrol or nutrient depletion (Figure S2E,F, Tables S1 and S2). We suppose that rapamycin has a stronger effect than resveratrol and nutrient depletion to fight against siTSC1/TSC2-dependent overexpression of mTOR, but this assumption needs further clarification. To understand how the precise molecular balance of mTOR-AMPK regulates cell survival, in particular autophagy, is highly relevant in several cellular stress related diseases such as neurodegenerative diseases (e.g., Parkinson’s disease, Alzheimer’s disease), metabolic diseases, inflammation, and carcinogenesis [42]. Our data suppose that the mTOR–AMPK–ULK1 regulatory triangle, similar to the mTOR–AMPK–NRF2 regulatory module, might be considered as double-edged swords in various diseases, especially in cancer due to their ambiguous role in promoting cell survival in healthy and cancerous cells [43,44]. Our systems biological analysis definitively made some approach toward improving the understanding of the molecular basis of these complex diseases and might also help to promote advanced therapies against these diseases.

4. Materials and Methods

4.1. Materials Rapamycin (Sigma-Aldrich, R0395, St. Louis, MO, USA), DMEM—no glucose, no glutamine (Life Technologies, A14430-01, Carlsbad, CA, USA), and Compound C (Sigma-Aldrich, P5499) were purchased. All other chemicals were of reagent grade.

4.2. Cell Culture and Maintenance As a model system, the human embryonic kidney (HEK293T, ATCC, CRL-3216, Manassas, WV, USA) cell line was used and maintained in DMEM (Dulbecco’s Modiﬁed Eagle Medium) (Life Technologies, 41965039) medium supplemented with 10% fetal bovine serum (Life Technologies, Int. J. Mol. Sci. 2019, 20, 5543 14 of 17

10500064) and 1% antibiotics/antimycotics (Life Technologies, 15240062). Culture dishes and cell treatment plates were kept in a humidiﬁed incubator at 37 ◦C in 95% air and 5% CO2.

4.3. RNA Interference RNA interference experiments were performed using Lipofectamine RNAi Max (Invitrogen Thermo Fisher Scientiﬁc Inc., Waltham, MA, USA) in GIBCO™ Opti-MEM I (GlutaMAX™-I) Reduced-Serum Medium liquid (Invitrogen) and 20 pmol/mL siRNA. The siTSC1 and siTSC2 oligonucleotides were purchased from Ambion (Ambion Thermo Fisher Scientiﬁc Inc., Waltham, MA, USA, 138713, 138741) and the siULK oligonucleotides were purchased from Ambion (118259). A total of 200,000 HEK293T cells were incubated at 37 ◦C in a CO2 incubator in an antiobiotic free medium for 16 h, then the RNAi duplex-Lipofectamine™ RNAiMAX (Invitrogen, 13778-075) complexes were added to the cells overnight. Then, fresh medium was added to the cells and the appropriate treatment was carried out.

4.4. SDS-PAGE and Western Blot Analysis Cells were harvested and lysed with 20 mM Tris, 135 mM NaCl, 10% glycerol, 1% NP40, pH 6.8. Protein content of the cell lysates was measured using the Pierce BCA Protein Assay (Thermo Fisher Scientiﬁc Inc., 23225, Waltham, MA, USA). During each procedure, equal amounts of protein were used. SDS-PAGE was done by using Hoefer miniVE (Hoefer Inc., Holliston, MA, USA). Proteins were transferred onto a Millipore 0.45 µM PVDF membrane. Immunoblotting was performed using TBS Tween (0.1%), containing 5% non-fat dry milk or 1% bovine serum albumin (Sigma-Aldrich, A9647) for blocking membrane and for antibody solutions. Loading was controlled by developing membranes for GAPDH or dyed with Ponceau S in each experiment. For each experiment, at least three independent measurements were carried out. The following antibodies were applied: antiLC3B (SantaCruz, sc-271625 Santa Cruz, PRK, USA), antiULK1-Ser555-P (Cell Signaling, 5869S, Danvers, MA, USA), antiULK1 (Cell Signaling, 8054S), antip70S6K-P (Cell Signaling, 9234S), antip70S6K (SantaCruz, sc-9202), antiAMPK-Thr172-P (Cell Signaling, 2531S), antiAMPK (Cell Signaling, 2603S), antiTSC1 (Cell Signaling, 4906S), antiTSC2 (Cell Signaling, 4308S) and antiGAPDH (SantaCruz, 6C5), anti4-EBP1-P (Cell Signaling, 9459S), anti4-EBP1 (Cell Signaling, 9644S), and HRP conjugated secondary antibodies (Cell Signaling, 7074S, 7076S). The bands were visualized using a chemiluminescence detection kit (Thermo Fisher Scientiﬁc Inc., 32106).

4.5. Immunofluorescence The cells were transferred to a chamber slide (Nunc™ Lab-Tek™ II Chamber Slide™ System, Thermo Fisher Scientific Inc., 154453PK). After silencing and treatment, the cells were washed in phosphate-buffered saline (PBS) and then fixed in 4% paraformaldehyde for 10 min and washed again with ice-cold PBS three times. The fixed cells were permeabilized with 0.1% Triton-X (in PBS) for 10 min and then washed in PBS three times for 5 min. After permeabilization, the cells were blocked with PBS with 0.1% Tween-20 containing and 3% bovine serum albumin for 30 min and incubated overnight with antiLC3A/B (Cell Signaling, 4108S) in 1% bovine serum albumin containing PBS at 4 ◦C. Cells were washed with cold PBS and incubated with Alexa Fluor 488 conjugated anti-rabbit (Cell Signaling, 4412S) for 1 h. Cells were washed in PBS and treated with DAPI (1:10000) for 15 min and washed again twice for 3 min. The slides were mounted with FluorSave Reagent (Millipore, 345789, Burlington, MA, USA) and observed under a fluorescence microscope (Nikon Eclipse Ts2R, Minato, Tokyo, Japan).

4.6. Mathematical Modeling The regulatory network given in Figure1 was translated into a set of nonlinear ordinary di ﬀerential equations (ODEs) and analyzed using the techniques of dynamical system theory [38,45,46]. Dynamical simulations, phase plane, and bifurcation analysis were carried out using the program XPPAUT, which is freely available from http://www.math.pitt.edu/~{}bard/xpp/xpp.html[ 38,46]. ODE describes the Int. J. Mol. Sci. 2019, 20, 5543 15 of 17 time-rate of the change of level or activity of a component (such as AMPK, ULK1, mTOR, and autophagy inducer (ATG)). The initial conditions and parameter values used for the simulations are given in the Supplementary Materials. All variables are dimensionless; rate constants (k’s) have a dimension of min-1, while Michaelis Menten constants are dimensionless. The total level of AMPK, ULK1, mTOR, and ATG were assumed to be constant (one unit). We provide the XPP code (in the Supplementary Materials) that was used to generate all the ﬁgures in the manuscript.

4.7. Statistics For densitometry analysis, western blot data were acquired using ImageJ software (https://imagej. net/). The relative band densities were shown and normalized to an appropriate total protein or GAPDH band used as the reference protein (see Supplementary Materials). For each of the experiments, three independent measurements were carried out. Results were presented as the mean values ± S.D. and were compared using ANOVA with Tukey’s multiple comparison post hoc test. Asterisks indicate statistically significant difference from the appropriate control: ns—nonsignificant; * p < 0.05; ** p < 0.01.

Supplementary Materials: Supplementary materials can be found at http://www.mdpi.com/1422-0067/20/22/ 5543/s1. Author Contributions: O.K., M.H., A.S., and G.B. conceived and designed the models; M.H., T.L., and B.H. performed the experiments; M.H., B.H., and O.K. performed the models; O.K., M.H., T.L., A.S., and G.B. analyzed the data; M.H. and O.K. wrote the paper. Funding: This work was supported by the Baron Munchausen Program of the Department of Medical Chemistry, Molecular Biology, and Pathobiochemistry of Semmelweis University, Budapest by the ÚNKP-18-3-I-SE-19 New National Excellence Program of the Ministry of Human Capacities; by the National Research, Development and Innovation Office by EFOP-3.6.3-VEKOP-16-2017-00009; by STIA_18_KF of Semmelweis University, Budapest; and by a MedInProt grant of the Hungarian Academy of Sciences. Acknowledgments: The authors are thankful to M. Márton. Conflicts of Interest: The authors declare that there are no conflicts of interest regarding the publication of this article.

Abbreviations mTOR Mammalian target of rapamycin AMPK 50 AMP-activated protein kinase rap Rapamycin CC Compound C s6 6 h starvation s24 24 h starvation

References

1. Maiuri, M.C.; Zalckvar, E.; Kimchi, A.; Kroemer, G. Self-eating and self-killing: Crosstalk between autophagy and apoptosis. Nat. Rev. Mol. Cell Biol. 2007, 8, 741–752. [CrossRef][PubMed] 2. Hardie, D.G.; Ross, F.A.; Hawley,S.A. AMPK: A nutrient and energy sensor that maintains energy homeostasis. Nat. Rev. Mol. Cell Biol. 2012, 13, 251–262. [CrossRef][PubMed] 3. Hardie, D.G.; Carling, D.; Gamblin, S.J. AMP-activated protein kinase: Also regulated by ADP? Trends Biochem. Sci. 2011, 36, 470–477. [CrossRef][PubMed] 4. Sarbassov, D.D.; Ali, S.M.; Sabatini, D.M. Growing roles for the mTOR pathway. Curr. Opin. Cell Biol. 2005, 17, 596–603. [CrossRef] 5. Wullschleger, S.; Loewith, R.; Hall, M.N. TOR signaling in growth and metabolism. Cell 2006, 124, 471–484. [CrossRef] 6. Tamargo-Gomez, I.; Marino, G. AMPK: Regulation of Metabolic Dynamics in the Context of Autophagy. Int. J. Mol. Sci. 2018, 19, 3812. [CrossRef] Int. J. Mol. Sci. 2019, 20, 5543 16 of 17

7. Watanabe, R.; Wei, L.; Huang, J. mTOR Signaling, Function, Novel Inhibitors, and Therapeutic Targets. J. Nucl. Med. 2011, 52, 497–500. [CrossRef] 8. Hay, N.; Sonenberg, N. Upstream and downstream of mTOR. Genes Dev. 2004, 18, 1926–1945. [CrossRef] 9. Laplante, M.; Sabatini, D.M. mTOR signaling in growth control and disease. Cell 2012, 149, 274–293. [CrossRef] 10. Zoncu, R.; Efeyan, A.; Sabatini, D.M. mTOR: From growth signal integration to cancer, diabetes and ageing. Nat. Rev. Mol. Cell Biol. 2011, 12, 21–35. [CrossRef] 11. Shaw, R.J.; Kosmatka, M.; Bardeesy, N.; Hurley, R.L.; Witters, L.A.; DePinho, R.A.; Cantley, L.C. The tumor suppressor LKB1 kinase directly activates AMP-activated kinase and regulates apoptosis in response to energy stress. Proc. Natl. Acad. Sci. USA 2004, 101, 3329–3335. [CrossRef][PubMed] 12. Hardie, D.G. AMP-activated/SNF1 protein kinases: Conserved guardians of cellular energy. Nat. Rev. Mol. Cell Biol. 2007, 8, 774–785. [CrossRef][PubMed] 13. Alers, S.; Löffler, A.S.; Wesselborg, S.; Stork, B. Role of AMPK-mTOR-Ulk1/2 in the regulation of autophagy: Cross talk, shortcuts, and feedbacks. Mol. Cell Biol. 2012, 32, 2–11. [CrossRef][PubMed] 14. Woods, A.; Johnstone, S.R.; Dickerson, K.; Leiper, F.C.; Fryer, L.G.; Neumann, D.; Schlattner, U.; Wallimann, T.; Carlson, M.; Carling, D. LKB1 is the upstream kinase in the AMP-activated protein kinase cascade. Curr. Biol. 2003, 13, 2004–2008. [CrossRef] 15. Hoyer-Hansen, M.; Jaattela, M. AMP-activated protein kinase: A universal regulator of autophagy? Autophagy 2007, 3, 381–383. [CrossRef] 16. Kim, J.; Kundu, M.; Viollet, B.; Guan, K.-L. AMPK and mTOR regulate autophagy through direct phosphorylation of Ulk1. Nat. Cell Biol. 2011, 13, 132–141. [CrossRef] 17. Levine, B.; Kroemer, G. Autophagy in the pathogenesis of disease. Cell 2008, 132, 27–42. [CrossRef] 18. Ravikumar, B.; Sarkar, S.; Davies, J.E.; Futter, M.; Garcia-Arencibia, M.; Green-Thompson, Z.W.; Jimenez-Sanchez, M.; Korolchuk, V.I.; Lichtenberg, M.; Luo, S.; et al. Regulation of mammalian autophagy in physiology and pathophysiology. Physiol. Rev. 2010, 90, 1383–1435. [CrossRef] 19. Wirawan, E.; Vanden Berghe, T.; Lippens, S.; Agostinis, P.; Vandenabeele, P. Autophagy: For better or for worse. Cell Res. 2012, 22, 43–61. [CrossRef] 20. Kung, C.P.; Budina, A.; Balaburski, G.; Bergenstock, M.K.; Murphy, M. Autophagy in tumor suppression and cancer therapy. Crit. Rev. Eukaryot. Gene Expr. 2011, 21, 71–100. [CrossRef] 21. Liu, Y.; Levine, B. Autosis and autophagic cell death: The dark side of autophagy. Cell Death Differ. 2015, 22, 367–376. [CrossRef][PubMed] 22. Fulda, S.; Kogel, D. Cell death by autophagy: Emerging molecular mechanisms and implications for cancer therapy. Oncogene 2015, 34, 5105–5113. [CrossRef][PubMed] 23. Ganley, I.G.; Lam du, H.; Wang, J.; Ding, X.; Chen, S.; Jiang, X. ULK1.ATG13.FIP200 complex mediates mTOR signaling and is essential for autophagy. J. Biol. Chem. 2009, 284, 12297–12305. [PubMed] 24. Chan, E.Y.; Kir, S.; Tooze, S.A. siRNA screening of the kinome identifies ULK1 as a multidomain modulator of autophagy. J. Biol. Chem. 2007, 282, 25464–25474. [CrossRef][PubMed] 25. Egan, D.F.; Shackelford, D.B.; Mihaylova, M.M.; Gelino, S.; Kohnz, R.A.; Mair, W.; Vasquez, D.S.; Joshi, A.; Gwinn, D.M.; Taylor, R.; et al. Phosphorylation of ULK1 (hATG1) by AMP-activated protein kinase connects energy sensing to mitophagy. Science 2011, 331, 456–461. [CrossRef][PubMed] 26. Gwinn, D.M.; Shackelford, D.B.; Egan, D.F.; Mihaylova, M.M.; Mery, A.; Vasquez, D.S.; Turk, B.E.; Shaw, R.J. AMPK phosphorylation of raptor mediates a metabolic checkpoint. Mol. Cell 2008, 30, 214–226. [CrossRef] [PubMed] 27. Inoki, K.; Zhu, T.; Guan, K.L. TSC2 mediates cellular energy response to control cell growth and survival. Cell 2003, 115, 577–590. [CrossRef] 28. Meley, D.; Bauvy, C.; Houben-Weerts, J.H.; Dubbelhuis, P.F.; Helmond, M.T.; Codogno, P.; Meijer, A.J. AMP-activated protein kinase and the regulation of autophagic proteolysis. J. Biol. Chem. 2006, 281, 34870–34879. [CrossRef] 29. Jung, C.H.; Seo, M.; Otto, N.M.; Kim, D.H. ULK1 inhibits the kinase activity of mTORC1 and cell proliferation. Autophagy 2011, 7, 1212–1221. [CrossRef] 30. Dunlop, E.A.; Hunt, D.K.; Acosta-Jaquez, H.A.; Fingar, D.C.; Tee, A.R. ULK1 inhibits mTORC1 signaling, promotes multisite Raptor phosphorylation and hinders substrate binding. Autophagy 2011, 7, 737–747. [CrossRef] Int. J. Mol. Sci. 2019, 20, 5543 17 of 17

31. Loffler, A.S.; Alers, S.; Dieterle, A.M.; Keppeler, H.; Franz-Wachtel, M.; Kundu, M.; Campbell, D.G.; Wesselborg, S.; Alessi, D.R.; Stork, B. Ulk1-mediated phosphorylation of AMPK constitutes a negative regulatory feedback loop. Autophagy 2011, 7, 696–706. [CrossRef][PubMed] 32. Lee, J.W.; Park, S.; Takahashi, Y.; Wang, H.G. The association of AMPK with ULK1 regulates autophagy. PLoS ONE 2010, 5, e15394. [CrossRef][PubMed] 33. Szymanska, P.; Martin, K.R.; MacKeigan, J.P.; Hlavacek, W.S.; Lipniacki, T. Computational analysis of an autophagy/translation switch based on mutual inhibition of MTORC1 and ULK1. PLoS ONE 2015, 10, e0116550. [CrossRef][PubMed] 34. Kim, M.; Lee, J.H. Identification of an AMPK phosphorylation site in Drosophila TSC2 (gigas) that regulate cell growth. Int. J. Mol. Sci. 2015, 16, 7015–7026. [CrossRef] 35. Li, H.; Lee, J.; He, C.; Zou, M.H.; Xie, Z. Suppression of the mTORC1/STAT3/Notch1 pathway by activated AMPK prevents hepatic insulin resistance induced by excess amino acids. Am. J. Physiol. Endocrinol. Metab. 2014, 306, E197–E209. [CrossRef] 36. Green, A.S.; Chapuis, N.; Maciel, T.T.; Willems, L.; Lambert, M.; Arnoult, C.; Boyer, O.; Bardet, V.; Park, S.; Foretz, M.; et al. The LKB1/AMPK signaling pathway has tumor suppressor activity in acute myeloid leukemia through the repression of mTOR-dependent oncogenic mRNA translation. Blood 2010, 116, 4262–4273. [CrossRef] 37. Oosterman, J.E.; Belsham, D.D. Glucose Alters Per2 Rhythmicity Independent of AMPK, Whereas AMPK Inhibitor Compound C Causes Profound Repression of Clock Genes and AgRP in mHypoE-37 Hypothalamic Neurons. PLoS ONE 2016, 11, e0146969. [CrossRef] 38. Tyson, J.J.; Chen, K.C.; Novak, B. Sniffers, buzzers, toggles and blinkers: Dynamics of regulatory and signaling pathways in the cell. Curr. Opin. Cell Biol. 2003, 15, 221–231. [CrossRef] 39. Blom, N.; Sicheritz-Pontén, T.; Gupta, R.; Gammeltoft, S.; Brunak, S. Prediction of post-translational glycosylation and phosphorylation of proteins from the amino acid sequence. Proteomics 2004, 4, 1633–1649. [CrossRef] 40. Kosztelnik, M.; Kurucz, A.; Papp, D.; Jones, E.; Sigmond, T.; Barna, J.; Traka, M.H.; Lorincz, T.; Szarka, A.; Banhegyi, G.; et al. Suppression of AMPK/aak-2 by NRF2/SKN-1 down-regulates autophagy during prolonged oxidative stress. FASEB J. 2018, 33, 2372–2387. [CrossRef] 41. Kapuy, O.; Papp, D.; Vellai, T.; Bánhegyi, G.; Korcsmáros, T. Systems-Level Feedbacks of NRF2 Controlling Autophagy upon Oxidative Stress Response. Antioxidants (Basel) 2018, 7, 39. [CrossRef][PubMed] 42. Kensler, T.W.; Wakabayashi, N.; Biswal, S. Cell survival responses to environmental stresses via the Keap1-Nrf2-ARE pathway. Annu. Rev. Pharm. Toxicol. 2007, 47, 89–116. [CrossRef][PubMed] 43. Lau, A.; Villeneuve, N.F.; Sun, Z.; Wong, P.K.; Zhang, D.D. Dual roles of Nrf2 in cancer. Pharm. Res. 2008, 58, 262–270. [CrossRef][PubMed] 44. Kubisch, J.; Türei, D.; Földvári-Nagy, L.; Dunai, Z.A.; Zsákai, L.; Varga, M.; Vellai, T.; Csermely, P.; Korcsmáros, T. Complex regulation of autophagy in cancer-integrated approaches to discover the networks that hold a double-edged sword. Semin. Cancer Biol. 2013, 23, 252–261. [CrossRef] 45. Kaplan, D.; Glass, L. Understanding Nonlinear Dynamics; Springer: New York, NY, USA, 1995. 46. Strogatz, S.H. Nonlinear Dynamics and Chaos; Addison-Wesley Co.: Reading, MA, USA, 1994.

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).