Reassessing the AMPK-MTORC1 balance in autophagy in the central nervous system

Autophagy is a cellular degradation and recycling system, indispensable for cellular and organ development, homeostasis, and function. This cellular process is evolutionarily highly conserved to quality control of many proteins and dysfunctional organelles, which finally recycle components as amino acids. This process is effective during normal physiology as part of anabolism and plays an additional important role during starvation (Dikic and Elazar, 2018). Different types of autophagy have been characterized based on their dynamic, mechanism of action, target substrates, and protein markers. Some of them are macroautophagy (hereafter called “autophagy”), microautophagy, and chaperone-mediated autophagy (Fleming et al., 2022).

In this perspective, we will focus specifically on the seemingly competing regulatory roles of AMP-activated protein kinase (AMPK) versus Ser/Thr kinase mechanistic target of rapamycin (MTORC1) in autophagy after considering some newly published data.

This type of autophagy starts with the initiation of the autophagosome and ends with its fusion with the lysosome and the degradation of the autophagic cargo via lysosomal hydrolases. This dynamic and highly complex process has been termed autophagic flux (Ariosa and Klionsky, 2016) and can be divided into several steps, recently reviewed in detail (Yamamoto et al., 2023)

Induction of autophagy is mediated in mammals by the ULK complex, which contains ULK1/2, FIP200, and the autophagy-related (ATG) proteins ATG13 and ATG101. ULK1/2 kinase activates the class III PI3K complex (class III phosphatidylinositol 3-kinase). The activation of this lipid kinase recruits other proteins such as Beclin-1, ATG14, and Vps15, among other proteins, and the phagophore is generated. The elongation of this structure generates the autophagosome, regulated by the ATG5-12, ATG7, MAP1LC3B/LC3B) and ATG8 family.

The cytosolic form of LC3B (LC3BI) is phosphatidylethanolamine (PE) modified to generate LC3B-PE (LC3BII). In the final process, several protein adapters such as SQSTM1/p62, NBR1, Optineurin, and LC3BII pathway (Gatica et al., 2018; Yamamoto et al., 2023), select substrates to be degraded after fusion with the lysosome. This dynamic process called “autophagic flux” is complex and some guidelines are reported to measure specific autophagy (Klionsky et al., 2021).

Modulation of autophagy-related pathways: In the balance between anabolism and catabolism, two key players stand out: the MTORC1 and AMP-activated protein kinase (AMPK) pathways. Both systems are present in almost all cell types and are phylogenetically conserved. In fact, it is generally accepted as a “prototypical mechanism” that mTORC1 is a repressor of autophagy, while AMPK, which is activated under energy stress, is a promoter of autophagy (Steinberg and Hardie, 2023).

MTORC1: The MTORC1 complex shares the same MTOR catalytic serine-threonine kinase with the MTORC2 complex. Both contain the inhibitory subunit DEPTOR and MLST8/GβL; furthermore, MTORC1 contains RAPTOR and PRAS40, while MTORC2 contains RICTOR, MSIN1, and PROTOR1. The assembly of FKBP12 to the mTORC1 complex can be inhibited by rapamycin, limiting the access of mTORC1 to substrates. Several intra- and extracellular signals converge on MTORC1, for example, under normal conditions, growth factors, amino acids, glucose, and oxygen keep the MTORC1 complex activated at the lysosomal membrane through several regulators (Figure 1). The deficiency of these nutrients can modify, among others, the ATP/ADP ratio, generating a homeostatic response that increases AMPK activity in parallel to the inhibition of MTORC1.

Figure 1.

The scheme represents some basic regulatory elements of the MTORC1 and AMPK pathways.

In red, we highlight the most inhibitory elements and in green and black those that are mostly activators. As prototypical nutrients, it points out more the pathways activated or used by glucose, amino acids (for example Leucine and glutamine), and growth factors such as insulin. The MTORC1 pathway is regulated by the activity of many growth factors through, among others, the class I lipid-kinase PI3K, Akt, and the GTPase Rheb, which maintain the active MTORC1 complex associated with the lysosome membrane. This activity will be reflected in the activation and maintenance of, for example, protein synthesis through the p70 S6K kinase. This pathway activity maintains low levels of autophagy. The AMPK pathway can be activated by the kinases LKB1, CaMKK2, and TAK1, in response to some stress situations, such as low oxygen and/or glucose, mitochondrial dysfunction that causes an ATP/AMP imbalance, or pharmacologically through metformin, phenformin, or AICAR. The new data would indicate that contrary to the classical model, non-obligatory activation of AMPK activates autophagy (marked in dashed blue). This activation of AMPK reflected in ULK1/2 or ATg14, among others, seems to keep the system “prepared but not active,” waiting for a second signal, yet to be defined. Created with Microsoft PowerPoint.

AMPK: AMPK is a heterotrimeric complex with serine-threonine activity, composed of three subunits: the catalytic alpha (α1/α2), and two regulatory subunits, beta (β1/β2) and gamma (γ1/γ2/γ3), which implies a complex regulation (Herzig and Shaw, 2018).

Primary role of AMPK is the stimulation of energy production through the catabolic metabolism of glucose and lipids, while inhibiting energy-consuming anabolic functions such as protein, fatty acid, or cholesterol synthesis. Additionally, it suppresses inflammation and cellular proliferation or induces mitochondrial renewal by inducing mitophagy. Regulation by phosphorylation of AMPK has been described by at least three kinases: LKB1, CaMKK2, and TAK1. This activation may be followed by phosphorylation of its prototypical target ACAC/ACC (Acetyl-CoA carboxylase) at S79 (Steinberg and Hardie, 2023).

MTORC1/AMPK dysfunction in Alzheimer’s disease (AD): AD is characterized by two initial features: senile plaques generated by the extracellular deposition of amyloid β peptide (Aβ) and neurofibrillary tangles caused by the aggregation of the hyperphosphorylated Tau protein. Both characteristics are generated by overexpression or defect in proteostasis/degradation. Many data suggest that there is a strong link between mTOR and autophagy. This relationship could be explored as a potential mechanism for AD therapy. Furthermore, AMPK is mainly expressed in reactive neurons and astrocytes. It has been described in patients with AD that AMPK is hyperactivated in neurons containing tangles, which could be involved in the degradation of Aβ peptides (Marinangeli et al., 2016).

Some reports have found that somatic mutations are enriched in AMPK pathway genes in the brains of AD patients (Park et al., 2023). Furthermore, changes in energy metabolism and inflammation associated with AD may change the physiological AMPK/MTORC1 balance (Marinangeli et al., 2016).

However, data indicate that treatment with metformin (an activator of AMPK activity) aggravates neurodegenerative processes in some mouse models. More recently, it has been reported that metformin does not reduce the risk of patients with AD and may instead increase the risk in the Asian population (Luo et al., 2022).

MTORC1 and AMPK as therapeutic targets: According to the prevailing model, mTOR inhibitors or AMPK activators should be able to regulate autophagy and, in consequence, have therapeutic effects through the clearance/prevention of protein aggregates in the brain.

The impact of the MTORC1 inhibitor rapamycin on autophagy and Aβ secretion remains controversial (Carosi and Sargeant, 2023). However, our data demonstrated that inhibiting MTORC1 in the APP/PSEN1 in vivo led to a reduced human-amyloid beta accumulation in the cortex and plasma due to autophagy increase (Benito-Cuesta et al., 2021). This therapeutic effect has been replicated in APP/PSEN1 primary neurons, clearly showing that the reduction of h-Aβ secretion/production is autophagy-dependent (Benito-Cuesta et al., 2021). More recently this mechanism was confirmed in APP/PSEN1 primary astrocytes (García-Juan et al., 2024). However, the role of AMPK remains controversial in distinct cell types (Park et al., 2023). Indeed, it has been described that AMPK hyper-activation could induce neuronal and synaptic loss, whereas its inhibition could be protective against neurotoxic Aβ oligomers in a mouse model of AD (Marinangeli et al., 2016).

Our ex vivo (in cell culture) data showed that even in the case of high mTORC1 inhibition by rapamycin the reduction h-Aβ was never higher than 50%. Thus, we analyzed whether AMPK activation may increase this therapeutic effect or even could have a synergistic effect with rapamycin. Surprisingly, the activation of AMPK using AICAR or Metformin generated a notable activation, as inferred from pACC, whereas the levels of autophagy markers and the autophagy flux indicated opposite effects. AICAR generated a modest induction of autophagy, without affecting MTORC1activity, whereas Metformin produced a blockage of autophagy similar to the ULK1/2 inhibitor (MRT67307). Neither of the two AMPK activators was able to generate an autophagy-mediated anti-amyloidogenic effect in APP/PS1 neurons or astrocytes.

Amyloid levels were reduced in the presence of 2-deoxy-D-glucose (in both neurons and astrocytes) and metformin (just in neurons), although this effect was not reversed by the AMPK inhibitor CoC, suggesting AMPK-independent mechanisms. On the contrary, inhibition of AMPK with CoC led to decreased amyloid-beta levels in both neurons and astrocytes and prevented the increased secretion of Aβ40 observed in neurons treated with AICAR (García-Juan, et al., 2024; Benito-Cuesta et al., 2021).

The autophagy-independent pro-amyloidogenic effect of AMPK was corroborated by the expression of the WT or CA-AMPK versions. The conclusion from our experiments using primary neurons and astrocytes strongly supports the hypothesis that AMPK activation is not sufficient to activate autophagy. Furthermore, mCherry-GFP LC3B transduction data strongly supports that AICAR and metformin blocked autophagy at a certain point of its induction (Benito-Cuesta et al., 2021; García-Juan, et al., 2024).

Recently, an excellent and extensive analysis was presented by Do-Hyung Kim´s group, using several cell and cell lines showing that AMPK negatively regulated ULK1 activity and autophagy by phosphorylation of ULK1. They showed that AMPK phosphorylation of ULK1 at Ser 556 and Thr 660 prevents the rapid induction of autophagy. This report demonstrated that either after glucose, or amino acids deprivation or Phenformin addition, the final mechanism of ULK1-phosphorylation AMPK-mediated inhibits autophagy (Park et al., 2023). Furthermore, they indicate that this phosphorylation protects ULK1 from caspase-mediated degradation. They proposed that the activation of AMPK in some substrates contributes more to maintaining the autophagy machinery in a state of readiness, primed for activation, rather than directly enhancing autophagy” (Ji-Man Park et al., 2023).

Therefore, considering these new data, obtained in different cell types (primary neurons, astrocytes, hepatocytes, and MEFs), it is tempting to propose that the mechanism of AMPK activation is a general negative regulator of autophagy rather than an activator; see the attached schematic (Figure 1). Perhaps we should consider that phosphorylation of AMPK on proteins such as ULK1/2 or in the MTORC1 complex does not necessarily activate autophagy, but rather keeps it blocked but ready for activation. In other words, they are “necessary but not sufficient” for the initiation of autophagy, in the absence of a second impulse yet to be precisely defined.

More work must be done to understand what is the second step that coordinates AMPK with autophagy. In the case of AD pathology because will be important to understand whether an additional boost of autophagy may enhance the Aβ elimination. In our opinion, it is critical to investigate how autophagy is regulated in the brain and its contribution to Aβ degradation; what is the role of each cell type (neurons, astrocytes, microglia, and endothelial cells) in redefining the “therapeutic opportunity” of MTORC1/AMPK pathway.

We thank Sergio Rivas (PhD; CBM-UAM & CIBERNED, Spain) and Maria Jose Perez (PhD; UAM-CBM, Spain) for the work and their helpful discussions along these years.

This work was supported by Grants from Spanish FEDER/Science and Innovation Ministry I + D + i-RETOS-PID2021-124801NB-I00; Centro de Investigación Biomédica en Red sobre Enfermedades Neurodegenerativas (CIBERNED; an initiative of the ISCIII) [PI2016/01]; and institutional grants from the Fundación Ramón Areces and Banco Santander to the CBMSO (to FW).