Tumour cells face various metabolic stresses as they arise and progress. Early-stage tumours that have not yet recruited new blood vessels to supply them with nutrients will be short of both glucose and oxygen1, and some cells face metabolic stress when they make the transition from a normal to a cancerous state and become detached from the extracellular matrix, their normal environment2. Additionally, some mutations that trigger tumour formation accelerate certain cellular metabolic programs and thereby cause stress. A protein called AMP-activated protein kinase (AMPK) is a central sensor of cellular metabolism. It is activated during situations of metabolic stress that lower the intracellular levels of ATP, a crucial energy-supplying molecule3. On page 661 of this issue, Jeon et al. report4 a mechanism by which AMPK dictates tumour-cell survival, and the surprising implication that it may have a cancer-promoting role in some contexts.

The protein that phosphorylates and activates AMPK under conditions of metabolic stress is encoded by a gene called LKB1, which is classed as a tumour suppressor: mutations that inactivate this gene are among the most common causes of human lung cancer. However, the LKB1 protein phosphorylates 12 kinase proteins in addition to AMPK, and it is not clear whether LKB1's tumour-suppressor function lies in its ability to activate AMPK, some of these other kinases, or both. What is clear is that AMPK activation influences cellular signalling pathways that are involved in cancer, such as the mTOR and p53 pathways, and inhibits the proliferation of cells in culture3,5. LKB1-dependent activation of AMPK also acts as a general metabolic 'rheostat': under conditions of metabolic stress, this activation directs the reprogramming of many cellular metabolic processes5,6,7 to restore energy homeostasis, which is essential for cell survival. But it is also not known which of AMPK's target proteins are most important for this activity.

Jeon et al. provide compelling evidence that AMPK exerts its effect on cell survival by regulating cellular levels of NADPH, a coenzyme that is essential for many biosynthetic reactions and which helps to remove dangerous reactive oxygen species (ROS) from cells (Fig. 1). The researchers show that AMPK indirectly regulates NADPH by inhibiting the metabolic enzymes acetyl-CoA carboxylase 1 (ACC1) and ACC2, the first substrates of AMPK to be discovered8. These enzymes control both fatty-acid synthesis, a process that consumes NADPH, and fatty-acid oxidation, which leads to NADPH regeneration.

Figure 1: Promoting cancer-cell survival during metabolic stress.
figure 1

The signalling pathway that involves the kinase proteins LKB1 and AMPK is activated in many tumour cells during conditions of metabolic stress, such as glucose deprivation and detachment from the normal extracellular matrix, and in response to the activation of some oncogenes. The pathway influences many metabolic and cancer-related processes, including glucose metabolism, autophagy, gene transcription and mTOR signalling. Jeon et al.4 show that the pathway can promote tumour-cell survival by inhibiting two AMPK-regulated enzymes, ACC1 and ACC2, which are involved in fatty-acid metabolism. These enzymes lower cellular levels of the coenzyme NADPH, which acts to attenuate dangerous reactive oxygen species (ROS) that accumulate during metabolic stress. It has also been recently demonstrated10 that the AMPK-related kinase protein ARK5 is involved in cell survival in response to activated oncogenes in some settings.

Jeon and colleagues also demonstrate that control of the ACC enzymes and NADPH by AMPK is required for the survival of glucose-deprived cells, as well as for that of cells detached from the extracellular matrix. These data fit nicely with the findings of a previous study2 that demonstrated that the survival of detached mammary epithelial cells depends on NADPH and the neutralization of cellular ROS. The new results also agree with the finding9 that the death of tumour cells during glucose deprivation is reduced by AMPK activation and fatty-acid oxidation. Although Jeon and colleagues have demonstrated a major role for NADPH in AMPK-dependent survival4, improved molecular tracing methods will be needed in future studies to determine quantitatively where in a cell NADPH is produced and consumed during the various forms of metabolic stress, and the extent to which production or consumption is influenced by ACC or AMPK.

In addition to its role in conditions of low nutrient levels or cell detachment, as observed by Jeon et al., AMPK may be required to maintain metabolic homeostasis in situations of deregulated, 'revved-up' metabolism that occur10 in response to the activity of oncogenes — genes that promote tumorigenesis. One oncogene implicated in many human cancers is c-MYC, which encodes a transcription factor, MYC, that primes cells for proliferation by expanding their biosynthetic capacity and reprogramming their metabolism. A recent study10 found that MYC-induced metabolic acceleration results in the activation of AMPK's function as a metabolic regulator and that AMPK and one of its related kinases (ARK5) are required for cell survival, but only under the conditions of elevated MYC levels. Another recent study11 screened more than 200 metabolism-associated genes for effects on cell viability and found that a gene encoding a subunit of AMPK is one of only two genes for which decreased expression resulted in reduced survival of prostate-cancer cells, but had no effect on normal prostate cells.

This selective death of cancer cells in response to reduced AMPK begs the question of whether the protein normally has a pro-tumorigenic effect or whether it more often acts to keep cancer at bay. AMPK is known to exert antiproliferative effects by suppressing many oncogenic signalling pathways and biosynthetic processes, and it is thought to be a major mediator of the tumour-suppressive effect of LKB1 (ref. 3). So how might these previously observed anticancer effects be reconciled with the idea that AMPK function permits survival of tumour cells? Is it possible that the timing or cell-type specificity of the loss of AMPK activity is pivotal in determining whether this event halts tumours or promotes their success by 'rewiring' growth and metabolic pathways to favour cell proliferation?

This potential dual role for AMPK in tumours parallels the current view of the role of autophagy in cancer. Autophagy is a process by which cells break down and recycle their own components, and is one of the metabolic systems regulated by AMPK. In some cases, inhibition of autophagy promotes tumour initiation, but in many other contexts autophagy is needed to keep later-stage tumours alive, such that its inhibition might be expected to have an anticancer effect12. Indeed, Jeon et al. demonstrate that reducing LKB1 or AMPK expression in a mouse model of breast cancer significantly inhibits the growth of mammary- tumour cells, and that the presence of modified versions of ACC1 and ACC2 that cannot be phosphorylated by AMPK also leads to reduced tumour growth. Such findings suggest that these effects result from inhibition of the survival-promoting pathways that stem from ACC1 and ACC2. However, the opposite effect has been observed when LKB1 is knocked out as an early event, including in some mouse models in which cancer development is driven by the activation of Myc (ref. 13) or another oncogenic protein, Kras (ref. 14) — here, loss of LKB1 strongly accelerates tumour growth.

How can these paradoxical results be resolved? It is possible that cultured tumour cells have a greater need for AMPK for survival than do tumours in vivo because nutrients in cell culture media are exhausted more rapidly than nutrients in an intact animal. Alternatively, the difference might arise from the timing of these events, and therefore depend on whether LKB1 or AMPK is inactivated in early- or late-stage tumours. This situation might explain previous data showing15,16 that LKB1- and AMPK-deficient cells are unexpectedly resistant to artificial transformation from a normal to a cancerous state in vitro, although this observation might also relate to altered metabolic demands in cell culture. Interestingly, another context in which AMPK activation seems to have both tumour-suppressing and tumour-promoting effects was recently reported17 in melanoma cells expressing the oncogene BRAF.

Only time and a more thorough dissection of these possibilities in different cell types and mouse models will reveal what determines the part AMPK plays in different cancers. In the meantime, the prospect that AMPK inhibition might sensitize advanced tumour cells to chemotherapy, or even to their own deregulated metabolic demands, warrants further examination. AMPK may therefore be both friend and foe to cancer, depending on when and why it is needed.