AMP-activated protein kinase
created by Tyler Brobst
AMP-activated protein kinase, or AMPK (PDB ID: 2V8Q), is an enzyme that regulates cellular metabolism in mammals in response to changes in energy availability. AMPK senses changes in the intracellular AMP/ATP ratio by binding adenosine monophosphate (AMP) and adenosine triphosphate (ATP) in a competitive manner (1). Increases in AMP concentration lead to AMPK activation. Once activated, AMPK acts to increase ATP-generating processes such as fatty-acid metabolism and to decrease ATP-consuming processes such as lipid and protein synthesis. AMPK carries out this function by phosphorylating various downstream protein targets that are involved in cellular metabolism (2). AMPK is a target for the treatment of type II diabetes because it is a regulator of lipid and glucose metabolism. The diabetes drug metformin is known to activate AMPK, leading to a decrease in glucose production by hepatocytes (3). The molecular weight of AMPK is 64,991.78 Da, and its isoelectric point (pI) is 7.74 (4).
AMPK exists physiologically as a heterotrimeric complex, consisting of a catalytic α-subunit with regulatory β- and γ-subunits. Xiao et al. determined mammalian AMPK’s structure using X-ray diffraction of α- and γ-subunits derived from rat with the β-subunit derived from human. The enzyme’s structure consists of an α + β module that interacts with a γ-subunit module. A β-strand segment in the α-subunit forms hydrogen bonds with a β-strand segment in the β-subunit, linking the two together. The α + β and γ modules are connected by β-sheet formation. The last two strands of the β-subunit form an interdomain β-sheet with a β-strand from the γ-subunit (1).
The γ-subunit contains multiple binding sites for AMP and ATP ligands and is therefore important in AMPK’s biological function. The γ-subunit is shaped like a flattened disk approximately 60 Å across and 30 Å deep. It is composed of four cystathionine β-synthase (CBS) motifs which pack together to form two Bateman domains (1). Each CBS motif displays a secondary structure pattern of βαββα (where β indicates a β-strand and α indicates an α-helix) in which the three β-strands form an antiparallel β-sheet with the two helices to one side (5). The four CBS motifs of the γ-subunit form a ring, creating a solvent-accessible channel in the center. The symmetrical circular arrangement of the four CBS motifs means that there are four potential adenyl-binding sites – one at each interface between two CBS motifs. However, only three of these binding sites display significant electron density for AMP binding. Up to three AMP molecules (AMP-1, AMP-2, and AMP-3) may be bound at one time (1).
The adenine ligands sit in hydrophobic pockets formed at the interface of two CBS motifs. Ligand binding is accomplished by the formation of hydrogen bonds with main-chain groups from two different strands. The ligand’s phosphate group interacts with the basic side chains of several different residues (including Arg-69, His-150, Arg-151, Lys-169, His-297, and Arg-298). These residues are protonated at normal physiological pH, giving them a positive charge that allows for favorable electrostatic interaction with a negatively charged phosphate group. The phosphate group also interacts with the hydroxyl groups of either serine or threonine residues. The 2’ and 3’ hydroxyl groups of the ligand’s ribose group make a bidentate interaction (i.e., an interaction in which two hydrogen bonds are formed) with an aspartic acid residue located in one of the CBS motif α-helices adjacent to the binding site (Asp-244 for AMP-1, Asp-89 for AMP-2, and Asp-316 for AMP-3). The fourth potential binding site fails to bind AMP or ATP due to the presence of an arginine residue at position 170 instead of an aspartic acid residue (1).
The first two binding sites on AMPK’s γ-subunit may be bound with either AMP or ATP depending on intracellular concentrations of the two ligands. ATP and AMP bind to the sites in a similar manner due to their structural similarity. The presence of β- and γ-phosphates on ATP necessitates the rearrangement of the side-chains of a few basic residues within the binding pocket; otherwise, there is no significant difference in the two complexes. AMPK is usually present in its inactive form, with both binding sites occupied by ATP.
The third binding site on AMPK’s γ-subunit is bound by a non-exchangeable AMP molecule (AMP-3). Xiao et al. observed co-purification of this AMP molecule with AMPK, indicating tight binding. This third binding site is structurally similar to the first two binding sites, but its enhanced AMP binding capability is explained by the presence of three amino acid residues that are absent in the other binding sites. Ser-225 and Ser-315 interact with AMP-3’s phosphate group and Thr-199 interacts with AMP-3’s ribose group. Because AMP-3 is non-exchangeable, it does not appear to be involved in AMP/ATP ratio sensing and its exact function is not understood (1).
AMPK’s function as a protein kinase is based on the structure of the catalytic α-subunit. The α-subunit’s N-terminal half contains a typical serine/threonine kinase catalytic domain, while its C-terminal half is responsible for interactions with AMPK’s β- and γ-subunits. AMPK activation is dependent on the activity of upstream protein kinases, which phosphorylate the Thr-172 residue within the activation loop of the kinase domain on the α-subunit. Binding of the AMP ligand serves to enhance phosphorylation by upstream kinases and to inhibit dephosphorylation by protein phosphatases by an allosteric mechanism (6). AMP binding also enhances the basal activity of phosphorylated AMPK. AMP binding triggers the formation of inter-subunit interactions that cannot occur when ATP is bound. This change in AMPK’s quaternary structure increases the activity of the kinase domain and decreases its susceptibility to dephosphorylation (1). Various mutations in the γ-subunit, which decrease its ability to bind AMP, have been shown to cause a loss of AMPK function (6). The diabetes drug metformin is known to activate AMPK by binding to the γ-subunit, inducing a conformational change that increases AMPK’s probability of phosphorylation by upstream protein kinases (7). The structure of metformin in complex with AMPK is unavailable in the Protein Data Bank.
AMPK’s β-subunit interacts with both the α- and γ-subunits and serves to link the heterotrimer together. The β-subunit also contains a central glycogen-binding domain (GBD), allowing AMPK to associate with glycogen, a cellular glucose storage molecule. The residues Trp-100, Lys-126, Trp-133, Leu-146, and Thr-148 are responsible for the β-subunit’s ability to the bind glycogen. Glycogen serves to inhibit AMPK by an allosteric mechanism that prevents phosphorylation of its catalytic kinase domain (8).
A protein that is similar to mammalian AMPK is the AMPK homologue from Schizosaccharomyces pombe, or “fission yeast” (PDB ID: 2QR1). Both mammalian and yeast AMPK function to regulate cellular metabolism in response to changing energy availability. The results from PSI-BLAST (E value = 10-142) indicate that mammalian and yeast AMPK have similar primary structures since an E value of 0.0 indicates identical amino acid sequences (9). The results from the Dali server (Z score = 13.0) indicate that mammalian and yeast AMPK have similar tertiary structures since a Z score above 2 indicates similar folding patterns (10). The yeast enzyme has a hinging movement between its α + β and γ modules, creating a 12° difference in orientation as compared to the mammalian enzyme. In terms of ATP binding, mammalian AMPK can bind both ATP and ATP complexed with Mg2+. In S. pombe, metal ions must be stripped from ATP before binding to AMPK (1). Mammalian AMPK is activated by phosphorylation of Thr-172, while yeast AMPK is activated by phosphorylation of Thr-189 (11). The two enzymes differ primarily in their ligand binding sites; mammalian AMPK has three adenine moiety binding sites, while yeast AMPK has only two (1). There is one binding site in each of the two Bateman domains of yeast AMPK’s γ-subunit. Site A (in Bateman domain A) specifically binds adenosine diphosphate (ADP). ADP binding to AMPK does not occur in humans, and its role in yeast AMPK regulation is not understood. Site B (in Bateman domain B) competitively binds AMP or ATP and shares the same amino acid residues as the mammalian AMPK ligand binding sites. In site A, ADP binding is accomplished by a number of non-covalent interactions. The adenine ring is stabilized by van der Waals forces with the side chains of Ile-35 and Ile-55 and by hydrogen bonding between the 6’-amino group and 1’-aza group and the main-chain nitrogen and carbonyl oxygen atoms of Ile-35. ADP’s α-phosphate forms electrostatic interactions with Arg-142 and Arg-287, and the β-phosphate interacts electrostatically with the side chains of Thr-162, Arg-165, Arg-287, and His-289. The β-phosphate’s interaction with these four residues is stronger than the interactions of the α-phosphate, leading to the ADP-binding specificity of this site (11).