Targeting an energy sensor to treat diabetes

See allHide authors and affiliations

Science  04 Aug 2017:
Vol. 357, Issue 6350, pp. 455-456
DOI: 10.1126/science.aao1913

Obesity occurs when whole-body energy intake exceeds energy expenditure for prolonged periods. This is a major public health issue because obesity increases the risk of disorders such as type 2 diabetes mellitus (T2DM). The liver and muscle store excess energy in the form of fat, leading to resistance to the hormone insulin. Released when blood glucose rises after meals, insulin normally promotes glucose uptake by muscle and represses glucose production by the liver, thus rapidly returning blood glucose to normal. However, this process is impaired in insulin-resistant individuals, who may eventually develop persistently elevated blood glucose (i.e., T2DM), which can cause debilitating or life-threatening complications. Because the energy sensor AMPK (adenosine monophosphate-activated protein kinase) promotes muscle glucose uptake by insulin-independent mechanisms, it was proposed in 1999 that AMPK-activating drugs might represent a novel approach to treating T2DM (1). Representing the culmination of more than 15 years of developing this concept, a study by Myers et al. (2) on page 507 of this issue and a study by Cokorinos et al. (3) show that compounds that bind to a unique site on AMPK can promote glucose uptake by muscle, and hence reverse elevated blood glucose in animal models of T2DM.

AMPK senses cellular energy status by monitoring the levels of AMP, ADP, and ATP (adenosine mono-, di-, and triphosphates) (4). ATP and ADP can be likened to the chemicals in a rechargeable battery, with a high ATP:ADP ratio being equivalent to a fully charged battery; AMP is a breakdown product of ADP. AMPK is activated by rising AMP:ATP and ADP:ATP ratios (which signify falling cellular energy levels) and acts to restore energy balance by switching on metabolic pathways that generate ATP (catabolism) while switching off processes that consume ATP, including anabolism.

The potential of AMPK activation for treatment of T2DM was reinforced by findings that it was activated by metformin (5). Metformin is the frontline drug treatment in T2DM, prescribed to more than 150 million patients worldwide. It activates AMPK by inhibiting mitochondrial ATP synthesis, thus increasing cellular AMP and ADP concentrations. Given this indirect mechanism, it is not surprising that metformin has multiple effects, some (6) but not all (7) of which are AMPK-independent. Nonetheless, the effects of metformin are largely confined to the liver and gut, so drugs that activate AMPK in muscle might have additional benefits.

Lowering blood glucose

Activating AMPK by targeting the ADaM site in β2 complexes induces skeletal muscle glucose uptake and thus lowers blood glucose.


AMPK exists as complexes comprising three subunits—a catalytic α subunit and regulatory β and γ subunits, with multiple isoforms (α1, α2, β1, β2, γ1, γ2, and γ3)—giving rise to 12 possible complexes that have tissue-specific expression patterns (4). Repeated sequences in the γ subunits generate three binding sites for AMP, ADP, and ATP, and some AMPK activators target these. However, a new regulatory mechanism was discovered when a new activator, A-769662, was identified (8). Although it mimicked two of the activating effects of AMP on AMPK, A-769662 clearly bound at a different site (9), which was identified when an AMPK complex was crystallized in the presence of A-769662 or another activator, 991 (10). This site is unique to AMPK and is called the allosteric drug and metabolite (ADaM) site (11). Although this site is currently only known to bind synthetic activators, it is widely assumed that there are unidentified cellular metabolite(s) that bind there [one natural product, salicylate, binds the site (12) but only occurs in plants].

The new ADaM site activators MK-8722, reported by Myers et al., and PF-739, reported by Cokorinos et al., are remarkably similar. Previous activators that bind the ADaM site, such as A-769662, are more specific for complexes containing the β1 isoform, but MK-8722 and PF-739 activate both β1 and β2 complexes. This is important because β1 is mainly expressed in the liver (at least in rodents), whereas muscle expresses mainly β2. MK-8722 is a potent activator of all 12 human AMPK complexes. Consistent with its ability to activate β2 complexes, MK-8722 enhanced glucose uptake by skeletal muscle in vitro and improved glucose tolerance in several rodent models of T2DM in vivo; in diabetic nonhuman primates (rhesus macaques), MK-8722 improved glucose tolerance and lowered hemoglobin A1c (HbA1c), a marker of persistently elevated blood glucose (2).

Cokorinos et al. (3) showed that, like MK-8722, PF-739 activated muscle AMPK and lowered blood glucose in mice with diet-induced obesity. PF-739 did not affect glucose production by the liver but increased peripheral glucose disposal in vivo. Consistent with this finding, the absence of α1 and α2 in the liver had no impact on blood glucose reduction by PF-739, whereas their absence in muscle attenuated it. PF-739 also lowered blood glucose and insulin concentrations in nondiabetic cynomolgus monkeys. Therefore, blood glucose reduction requires a “pan-β” activator that acts on the β2 isoform expressed in muscle (see the figure). However, β1-selective activators may still be useful. For example, a β1-selective activator, PF-06409577, activated AMPK in the kidney and reduced urinary protein content and other markers of kidney damage in a rat model of diabetic kidney disease (13).

The results with MK-8722 and PF-739 are promising, because the new compounds exert effects on blood glucose via actions on muscle and would therefore be expected to have beneficial effects beyond that of metformin. However, concerns about adverse effects of AMPK activators were raised by previous studies of humans with mutations in the gene that encodes the γ2 subunit. These mutations increase basal AMPK activity and are associated with increases in cardiac glycogen content and hypertrophy (increased heart weight) as well as potentially life-threatening cardiac arrhythmias (14). Indeed, Myers et al. found that long-term use of MK-8722 in rats and rhesus macaques was associated with increased cardiac glycogen content and hypertrophy, although not with arrhythmias. Cardiac effects of PF-739 were not reported, so it remains unclear whether they are common to all pan-β activators.

Thus, MK-8722 and PF-739 were effective in reversing elevated blood glucose concentration in rodents and nonhuman primates, and this was due to activation of AMPK in muscle, not the liver. This might make them valuable adjuncts to metformin, which acts primarily on the liver. One caveat is their potential to cause cardiac hypertrophy, which is likely to raise concerns with regulatory authorities. However, as Myers et al. point out, the cardiac hypertrophy observed with MK-8722 treatment is reminiscent of that found in elite athletes (which may result from AMPK activation in cardiac muscle during training) and does not necessarily have adverse consequences. Currently, there are several AMPK activators that bind the ADaM site; whether there is a natural metabolite that binds there remains unclear, although those in the field may now be stimulated to look.


View Abstract

Stay Connected to Science

Navigate This Article