Targeting AMPK for Therapeutic Intervention in Type 2 Diabetes

1.1. AMPK activity is critical to cell physiology in different tissues and organs

AMPK functions as a ser/thr kinase which provides an evolutionary conserved cellular energy sensor. This kinase is a focal point for metabolic control in all eukaryotes, where it regulates many aspects of physiology (Hardie, 2008a; Kim et al., 2009a; Li & McCullough, 2010; Lopaschuk, 2008; Ronnett et al., 2009; Steinberg & Kemp, 2009; Zhang, et al., 2009). It is well-established that AMPK and its yeast ortholog Snf1 control a large number of diverse processes; they include the response to nutrient limitation or other environmental changes, transcription, transport across the nuclear envelope, cell growth, cell cycle progression, mitosis, cell polarity, development, auto- and mitophagy (Amato et al., 2011; Bungard et al., 2010; Egan et al., 2010; Lee et al., 2007; Li & McCullough, 2010; Mirouse et al., 2007; Nagata & Hirata, 2009; Narbonne & Roy, 2009; Quan et al., 2007; Steinberg & Kemp, 2009; Viollet et al., 2009a; Wang et al., 2010; Witczak et al., 2008). As a result of these contributions, AMPK is vital to the function of several organs and tissues in metazoans (Fig. 1).

Owing to its pivotal role in the control of glucose homeostasis, carbohydrate, lipid and protein metabolism AMPK is a key player in many human diseases and disorders (Fogarty & Hardie, 2010; Lage et al., 2008; Towler & Hardie, 2007; Viollet et al., 2009b). In particular, the low activation state of AMPK could contribute to the increase in type 2 diabetes and obesity (Hardie et al., 2006). Moreover, as essential regulator of glucose homeostasis and lipid metabolism, AMPK has become an important therapeutic target in type 2 diabetes and obesity. This is exemplified by metformin and thiazolidinedione derivatives (TZDs); these drugs are used for therapeutic intervention in type 2 diabetes and lead to the activation of AMPK.

2.1. Control of AMPK activity by phosphorylation and changes in AMPK concentration

The importance of AMPK as a key regulator in cellular metabolism requires a tight control of the enzyme. The rapid regulation of AMPK activity is based on at least three mechanisms that contribute to AMPK activation (Oakhill et al., 2010; Sanders et al., 2007; Shackelford & Shaw, 2009; Steinberg & Kemp, 2009). (a) The most important step for AMPK activation is the phosphorylation of Thr172 of the α subunit which can be modified by the upstream kinases LKB1, CaMKKβ and TAK1 (Fig. 2). Thr172 is phosphorylated when the energy state of the cell is low, i.e. when the AMP/ATP ratio rises. Under these conditions, AMP binding to the regulatory γ subunit promotes the subsequent Thr172 phosphorylation. LKB1 is the major upstream kinase for this event in tissues like skeletal muscle. The effect of AMP binding depends on the type of γ subunit (Cheung et al., 2000). Specifically, AMP-binding to γ2 subunits leads to the largest increase in AMPK activity. By contrast, a relative small change is observed for the γ3 subunit which is mostly synthesized in glycolytic skeletal muscle. Recent data suggest that the β subunits also play a crucial role in AMPK activation. It was proposed that β subunit myristoylation provides a switch that is a prerequisite for Thr172 phosphorylation (Oakhill et al., 2010). (b) Aside from changes in the AMP/ATP ratio, a rise in intracellular Ca²⁺ concentrations triggers Thr172 phosphorylation. This modification is mediated by CaMKKβ and particularly important in tissues where LKB1 is not the predominant kinase for Thr172. At present, the role of TAK1 in AMPK activation is not fully understood. (c) AMPK activation can be prolonged by preventing the dephosphorylation of Thr172, a process catalyzed by phosphatases PP2A and PP2C (Kim et al., 2009a; Nagata & Hirata, 2010).

Aside from the dephosphorylation of phospho-Thr172, a negative regulation of AMPK involves the phosphorylation of Ser485/491 by PKC and possibly Akt, whereas the decline in activity by Ser173 phosphorylation was ascribed to PKA (Djouder et al., 2010). Such modification on Ser173 may help to fine tune lipid metabolism in adipose tissue.

The tissue-specific regulation of AMPK activity is likely achieved by the combined effects of upstream activating kinases, inactivating phosphatases as well as the synthesis and degradation of AMPK subunits. For example, LKB1 is particularly important to activate AMPK in skeletal muscle, whereas CaMKKβ is crucial in the brain (Ronnett et al., 2009). On the other hand, TNFα alters AMPK activation by modulating the synthesis of PP2C (Lu et al., 2010; Steinberg et al., 2006). Aside from the rapid control of AMPK activation by phosphorylation, changes in the expression of subunit genes or the turnover of AMPK subunits can help to fine tune AMPK activity in some tissues (Barry et al., 2010; Fukuyama et al., 2007; Hallows et al., 2006; Qi et al., 2008; Niesler et al., 2007; Steinberg et al., 2003).

2.2. Pharmacological compounds and other factors that alter AMPK activity

Previous work established the essential role of AMPK in the regulation of carbohydrate, protein and lipid metabolism; this made AMPK a key target for the treatment of type 2 diabetes, obesity and metabolic syndrome (Gruzman, Babai & Sasson, 2009; Hardie, 2008b; Steinberg & Kemp, 2009; Viollet et al., 2010; Viollet et al., 2009b). Indeed, in a clinical setting AMPK activity is altered with the anti-diabetic drug metformin and other biguanides. The drug-induced activation of AMPK has important consequences for the patient; among these is the improvement of insulin resistance.

Pharmacological drugs have also been critical to define how AMPK mediates metabolic control (see Table 1). These compounds employ a variety of molecular mechanisms that culminate in AMPK activation (Gruzman et al., 2009; Hawley et al., 2010; Mantovani & Roy, 2011). For example, the kinase can be activated by a rise in the AMP/ATP ratio, generation of an AMP mimetic or increase in intracellular Ca²⁺ concentrations (Hawley et al., 2010). Metformin impacts several biological processes that ultimately activate AMPK. These include changes in the respiratory chain, increased synthesis of the protein deacetylase SIRT1 (which activates LKB1) and activation of TAK1 (Caton et al., 2010; Hawley et al., 2002; Hawley et al., 2010; Xie et al., 2006). Phenphormin promotes the LKB1-dependent activation of AMPK by inhibiting mitochondrial complex I (Hawley et al., 2010). Resveratrol prevents the acetylation and concomitant inactivation of the upstream kinase LKB1, this compound also inhibits mitochondrial ATP synthase and may increase the concentration of adiponectin (Hawley et al., 2010; Wang et al., 2011). AICAR generates the AMP mimetic 5-amino-4-imidazolecarboxamide ribotide (ZMP) and causes a drop in cellular ATP and ADP, which leads to AMPK activation (Hawley et al., 2010). Aside from drugs that activate AMPK, compound C serves as an ATP-competitive inhibitor of AMPK that has been used widely. All of the compounds discussed here are established pharmacological tools that alter AMPK activation or enzymatic activity; they have been useful for the analysis of AMPK in vitro, in growing cells and in whole animals. The following table summarizes how AMPK activity can be affected in vitro, growing cells, organs or whole organisms.

4.1. Subcellular distribution of AMPK substrates

The combination and integration of different subcellular events regulated by AMPK enables cells, tissues and organs to coordinate different metabolic pathways in order to achieve and maintain the proper energy balance of the whole organism. Fig. 3 depicts established AMPK substrates according to their presence in different subcellular compartments. Table 3 expands this information and specifies how the AMPK-dependent phosphorylation of individual substrates alters their functions.

It is obvious from Fig. 3 that AMPK phosphorylates a large number of proteins that are associated with distinct organelles or subcellular compartments. Cytoplasmic and mitochondrial substrates of the kinase include enzymes that are involved in fat, protein, glucose and glycogen metabolism. Kinase targets in the plasma membrane consist of ion channels, carriers and receptors, whereas other substrates are linked to the function or trafficking of intracellular membranes. This includes the transport of vesicles containing the glucose transporter GLUT4, because GLUT4 translocation to the plasma membrane is a pre-requisite for efficient glucose uptake in skeletal muscle and other tissues. AS160 and TBC1D1 likely play a role in these processes (Table 3). In the nucleus, the AMPK-mediated modification of transcription factors, transcriptional regulators and a subunit of RNA-polymerase I control the expression of genes that are implicated in specific anabolic and catabolic reactions. The phosphorylation of several of these targets is also critical to the biogenesis of mitochondria and the assembly of ribosomes.

Given the diverse types of AMPK substrates and their presence in different cellular locations, it is helpful to recapitulate their functions (Table 3a). This knowledge is a prerequisite to understand how the dynamic association and action of AMPK in different compartments will impact downstream events.

Substrates that have been established for AMPK heterotrimers that contain the α1 or α2 subunit are shown. For different AMPK substrates the function, effect of AMPK-dependent phosphorylation and the major subcellular localization are depicted. For some substrates, there are cell-type specific differences, and the effect of AMPK-dependent phosphorylation may not be fully understood or controversial. The list of AMPK substrates was compiled from PhosphoSitePlus (phosphosite.org).

4.2. Subcellular distribution and trafficking of AMPK

AMPK is associated with different organelles and subcellular compartments; however, little is known about the dynamic nature of this distribution. Analyses of other signaling pathways have demonstrated that the subcellular localization of kinases is critical for the proper response to extra- and intracellular stimuli, and it is likely that the same scenario applies to AMPK. We will therefore briefly review what is currently known about the subcellular localization and trafficking of AMPK.

AMPK is found in the nucleus and cytoplasm which reflects functions for the enzyme in both locations (Kodiha et al., 2007; Witczak et al., 2008). Although the kinase is associated with multiple compartments, α1 and α2 subunits differ in their nucleocytoplasmic distribution. Under normal growth conditions, α1 is predominantly in the cytoplasm, whereas α2 locates to both the nucleus and cytoplasm (Salt et al., 1998). It was further shown that the nuclear localization sequence (NLS) present in the α2 subunit is required for AMPK nuclear translocation (Suzuki et al., 2007), suggesting that the catalytic α subunit is essential for the proper intracellular localization of the holoenzyme. Our current model of AMPK trafficking proposes that the kinase shuttles between the nucleus and the cytoplasm. Data from our group and others support this idea, as they demonstrated that the nuclear carrier Crm1 is essential for AMPK export from the nucleus to the cytoplasm (Kazgan et al., 2010; Kodiha et al., 2007).

How the β subunit impacts the proper targeting of the holoenzyme is at present not entirely clear. The β1 subunit can be modified posttranslationally, both by phosphorylation and myristoylation, and these modifications were linked to the subcellular targeting of the β1 subunit (Suzuki et al., 2007; Warden et al., 2001). It was proposed that AMPK concentrates in the cytoplasm when the heterotrimeric enzyme contains the β1 subunit (Suzuki et al., 2007). However, this model is difficult to reconcile with the fact that both β1 and β2 subunits can be detected in the nucleus (Kodiha et al., 2007). Furthermore, recent studies suggest that the myristoylation of β1 and β2 subunit is particularly important for AMPK activation, as AMP-dependent myristoylation provides a switch that triggers Thr172 phosphorylation (Oakhill et al., 2010). Although the contribution of β subunits to nuclear and membrane targeting of the holoenzyme is not completely understood at this point, the importance of the β subunit for glycogen binding is well established (Polekhina et al., 2003).

In contrast to the α and β subunits, little is known about the trafficking of AMPK γ subunits. In Drosophila, the single γ subunit migrates into the nucleus of the fat body with the onset of autophagy during normal development, and a potential NLS was detected in this subunit (Lippai et al., 2008). It was speculated that this nuclear accumulation contributes to the expression of genes that are necessary for autophagy.

Several studies support the model that the intracellular distribution of AMPK in human and other cells is dynamic. This is particularly important in the context of disease, because the distribution of AMPK can be modulated by physiological and environmental stimuli. For example, the α2 subunit translocates to the cell nucleus upon exercise or environmental stress (Kodiha et al., 2007; McGee et al., 2003), indicating that the adaptation of skeletal muscle during exercise or metabolic stress is at least in part mediated by the subcellular relocation of AMPK. Examples of the relocation of AMPK α subunits in human cells exposed to oxidative stress or depleted for energy are shown in Fig. 4 (Kodiha et al., 2007).

The relocation of AMPK subunits in response to physiological changes is not restricted to the α subunits; our previous experiments demonstrated that AMPK β subunits accumulate in the nucleus as well when cells are exposed to oxidative and other forms of stress (Kodiha et al., 2007). Moreover, AMPK localization could be regulated by the circadian rhythm. Specifically, changes in the expression of AMPK subunits may depend on the circadian rhythm; this change in expression will then alter the intracellular distribution of AMPK (Lamia et al., 2009). Since these studies were carried out with mice, it has yet to be shown whether the same applies to humans.

Studies from several laboratories, including our group, defined the signals and mechanisms that determine the trafficking and intracellular distribution of AMPK. Our work also suggested crosstalk between other signaling cascades and the localized action of AMPK (Kodiha et al., 2007 and unpublished). Ultimately, such crosstalk will add to the complexity of downstream events that are modulated by AMPK. Taken together, previous research suggests that AMPK subunits move between different subcellular locations, and it can be expected that the compartment-specific actions of the kinase are linked to the physiological response of cells and tissues.

4.3. How does the compartment-specific action of AMPK impact cellular functions that are relevant to type 2 diabetes?

Type 2 diabetes is associated with the increased risk of a growing number of diseases and pathologies. This is exemplified by renal nephropathy, myocardial disease, stroke, Alzheimer’s and Parkinson’s disease (Almdal et al., 2004; Biessels et al., 2006; Burdo et al., 2009; Hallows et al., 2010; Hu et al., 2007; Maher & Schubert, 2009; Schernhammer et al., 2011). Several drugs are currently used in the clinical setting to activate AMPK in patients suffering from type 2 diabetes or obesity. However, it should be kept in mind that AMPK activation can be beneficial as well as harmful in the ischemic heart, and AMPK activation may be linked to neurodegeneration (Lopaschuk, 2008; Spasic, Callaerts & Norga, 2009; Thornton et al., 2011; Vingtdeux et al., 2011). Thus, activation of AMPK throughout the whole organism or the entire cell of a particular tissue may not always be advantageous. As an alternative approach, we put forward the concept of a compartment-specific modulation of AMPK action. Since AMPK activation can be damaging in the context of some of the complications associated with type 2 diabetes, our approach applies both to the localized activation as well as inhibition of the kinase. We believe that the confined action of AMPK will provide a better therapeutic approach in the future that could reduce the side-effects of AMPK modulators. The simplified model in Fig. 5 summarizes the possible changes of cellular functions that will be induced by targeting AMPK in different subcellular compartments.

In the past few years, significant progress has been made with the identification of AMPK substrates and their links to human disease. As shown in Fig. 3 and Table 3, AMPK modifies targets in different subcellular compartments or organelles. We propose that the modification of these substrates relies on (a) the amount of active AMPK and (b) the intracellular distribution of AMPK. The combination of AMPK activation and subcellular localization will then determine the level of phosphorylation of its substrates and the subsequent changes in cell physiology. Such changes will affect both the rapid response to specific stimuli as well as the long-term modification of metabolism and other processes. In the following, we focus on some of the processes that are controlled by AMPK and linked to type 2 diabetes or the complications associated with the disease.