Keeping the home fires burning: AMP-activated protein kinase

Abstract

Living cells obtain energy either by oxidizing reduced compounds of organic or mineral origin or by absorbing light. Whichever energy source is used, some of the energy released is conserved by converting adenosine diphosphate (ADP) to adenosine triphosphate (ATP), which are analogous to the chemicals in a rechargeable battery. The energy released by the conversion of ATP back to ADP is used to drive most energy-requiring processes, including cell growth, cell division, communication and movement. It is clearly essential to life that the production and consumption of ATP are always maintained in balance, and the AMP-activated protein kinase (AMPK) is one of the key cellular regulatory systems that ensures this. In eukaryotic cells (cells with nuclei and other internal membrane-bound structures, including human cells), most ATP is produced in mitochondria, which are thought to have been derived by the engulfment of oxidative bacteria by a host cell not previously able to use molecular oxygen. AMPK is activated by increasing AMP or ADP (AMP being generated from ADP whenever ADP rises) coupled with falling ATP. Relatives of AMPK are found in essentially all eukaryotes, and it may have evolved to allow the host cell to monitor the output of the newly acquired mitochondria and step their ATP production up or down according to the demand. Structural studies have illuminated how AMPK achieves the task of detecting small changes in AMP and ADP, despite the presence of much higher concentrations of ATP. Recently, it has been shown that AMPK can also sense the availability of glucose, the primary carbon source for most eukaryotic cells, via a mechanism independent of changes in AMP or ADP. Once activated by energy imbalance or glucose lack, AMPK modifies many target proteins by transferring phosphate groups to them from ATP. By this means, numerous ATP-producing processes are switched on (including the production of new mitochondria) and ATP-consuming processes are switched off, thus restoring energy homeostasis. Drugs that modulate AMPK have great potential in the treatment of metabolic disorders such as obesity and Type 2 diabetes, and even cancer. Indeed, some existing drugs such as metformin and aspirin, which were derived from traditional herbal remedies, appear to work, in part, by activating AMPK.

Keywords: AMP-activated protein kinase; adenosine triphosphate; cell signalling; energy homeostasis; mitochondria.

Figure 1.

Reactions catalysed by (a) ATPases or ATP synthases and (b) adenylate kinases.

Figure 2.

Schematic of the major catabolic pathways (and selected anabolic pathways) in a typical animal cell, showing how they are distributed between the cytoplasm and the mitochondria. Key to intermediates: G6P, glucose-6-phosphate; G1P, glucose-1-phosphate; UDPG, UDP-glucose; F6P, fructose-6-phosphate; FBP, fructose-1,6-bisphosphate; triose-P, dihydroxyacetone phosphate or glyceraldehyde-3-phosphate; 3PG, 3-phosphoglycerate; 2PG, 2-phosphoglycerate, PEP, phosphoenolpyruvate; 2OG, 2-oxoglutarate; OAA, oxaloacetate. (Online version in colour.)

Figure 3.

(a) Transmission electron micrograph and (b) fluorescence micrograph of cells showing mitochondria. The image of a human lung cell in (a) (placed in the public domain by Louisa Howard) reveals the double membrane surrounding mitochondria, and the invaginations of the inner membrane known as cristae. The fluorescence micrograph in (b) (by Simon Troeder, CC BY 4.0), which is at much lower magnification, shows mitochondria in two human cells expressing a mitochondrially targeted green fluorescent protein, and reveals that mitochondria actually form a branching network of tubules. The image in (a) is of two tubules sampled as thin sections, which therefore appear to be spherical or ovoid. (Online version in colour.)

Figure 4.

The distinct reactions catalysed by protein kinases and protein phosphatases.

Figure 5.

Tripartite mechanism for activation of AMPK by increases in the cellular AMP : ATP ratio, and by the non-canonical Ca²⁺-CaMKK pathway. Binding of AMP to the AMPK-γ subunit causes activation by (1) promoting phosphorylation by LKB1; (2) inhibiting dephosphorylation by protein phosphatases and (3) allosteric activation. Binding of ADP at higher concentration can mimic effect (2), whereas biding of ATP antagonizes all three effects. Increases in intracellular Ca²⁺ activate CaMKK2, which phosphorylates the same site on AMPK (Thr172) as LKB1. (Online version in colour.)

Figure 6.

Schematic of changes in domain disposition in the AMPK heterotrimer when AMP (a) rather than ATP (b) is bound to the CBS3 site on the γ subunit. The model at the top is based on the structure of the human α1β2γ1 complex, crystallized with Thr172 phosphorylated in the presence of AMP and cyclodextrin (PDB file 4RER) [39]; the two views are rotated by 70° with respect to each other around the y axis. The model at the bottom (of the same two views) is a hypothetical structure in which the catalytic and nucleotide-binding modules have shifted apart upon displacement of AMP by ATP, with consequent release of the α-linker from the CBS3 site.

Figure 7.

Model for the non-canonical, AMP-independent activation of AMPK upon glucose starvation. In the presence of glucose, a high flux through glycolysis means that aldolase, which associates with the v-ATPase on the lysosomal membrane, has fructose-1,6-bisphosphate (FBP) bound, preventing interaction of the Axin : LKB1 complex, which is cytoplasmic, with the v-ATPase and the Ragulator. AMPK may already be, at least partly, located on the lysosome due to N-myristoylation of the β subunit. On removal of glucose, aldolase is now largely unoccupied by FBP and a conformational change allows the Axin : LKB1 complex to bind to the v-ATPase and the Ragulator. LKB1 now phosphorylates and activates AMPK: whether this causes dissociation of the active AMPK from the lysosomal membrane remains unclear at present. Diagram courtesy of Chensong Zhang and Shengcai Lin, based on [69]. (Online version in colour.)

Figure 8.

AMPK recognizes its targets by means of complementary interactions between amino acid side chains surrounding the target serine/threonine residues and side chains from the kinase domain of AMPK. The images were based on a model of the kinase domain with the sequence around Ser79 on the downstream target ACC bound to it [41]. The kinase domain is in ‘spacefilling' representation, while the ACC sequence is in ‘cartoon' representation with the polypeptide backbone represented as a narrow green ribbon, and specific side chains shown with carbon atoms in green, nitrogen in blue and sulfur in orange. (a) The basic side chain (Arg75) at P − 4 (i.e. four residues N-terminal to the target serine) binds to an acidic patch (red) formed by Asp103 and Glu100 from the kinase domain, while the hydrophobic side chain at P − 5 (Met74) interacts with a hydrophobic pocket (orange) containing Leu212 from the kinase domain; (b) the basic side chain (His73) at P − 6 interacts with an acidic patch (red) formed by Asp215, Asp216 and Asp217; (c) the basic side chain at P + 3 (His82) interacts with Asp56 from the kinase domain, while the hydrophobic side chains at P + 4 (Leu 83) interacts with a hydrophobic pocket (orange) on the surface of the α-KD; (d) the α-helix running from P − 16 to P − 5 binds in a hydrophobic groove (the target protein-binding groove, figure 6), with the hydrophobic side chains of Ile63 (P − 16), Leu66 (P − 13), Leu70 (P − 9) and Met74 (P − 5) binding in the groove. (Online version in colour.)

Figure 9.

Summary of catabolic pathways (green) switched on and anabolic pathways (red) switched off when AMPK is activated. Green arrows indicate activation and red lines with bars across the end indicate inhibition. Key to acronyms: GLUT1/GLUT4, glucose transporter-1/-4; F26BP, fructose-2,6-bisphosphate; CD36, cluster of differentiation 36, a fatty acid transporter; ACC1/ACC2, acetyl-CoA carboxylase-1/-2; PGC1α, peroxisome proliferator-activated receptor co-activator-1α; ULK1/2, UNC-51-like kinase-1/-2; MFF, mitochondrial fission factor; HMGR, HMG-CoA reductase; GPAT, glycerol phosphate acyltransferase; SREBP1c, sterol response element-binding protein-1c; TSC2, tuberous sclerosis complex protein-2; EF2, elongation factor-2; TIF-IA, transcription initiation factor-IA. (Online version in colour.)