Hunger is a gatekeeper of pain in the brain

A neuronal population has now been found that regulates two competing needs — hunger and pain. Urgent pain overrides hunger, but appetite-inducing neuronal activity dampens long-term pain responses to enable feeding.
lexey Ponomarenko is at the Institute of Clinical Neuroscience and Medical Psychology, Medical Faculty, Heinrich-Heine-University, Düsseldorf 40225, Germany.

Search for this author in:

Tatiana Korotkova is at the Max Planck Institute for Metabolism Research, Cologne 50931, Germany.

Search for this author in:

The body’s basic needs include a timely supply of nutrients and the avoidance of tissue damage, which are signalled in the brain by hunger and pain, respectively. But these needs cannot be fulfilled simultaneously, because their resolution involves mutually exclusive behaviours. How does the brain prioritize the more urgent need? Writing in Cell, Alhadeff et al.1 report that the brain’s priorities are set depending on the type of pain involved. Hunger-mediating neurons suppress long-term inflammatory pain, but acute pain, which signals an immediate threat, dampens the activity of these neurons and thus deprioritizes feeding.

Alhadeff and colleagues deprived mice of food for 24 hours, and analysed how the hungry animals responded to pain. The researchers found that responses to long-term inflammatory pain — of the type associated with chronic disease and recovery from injury — were reduced in the food-deprived animals compared with controls. By contrast, short-term responses to acute pain that was induced by chemicals, heat or force remained intact in hungry mice.

The brain’s hypothalamus contains several structures involved in regulating food intake. One of these, the arcuate nucleus, harbours a population of neurons that express agouti-related protein (AgRP), and help to signal nutritional needs — activation of these neurons evokes voracious feeding2, whereas their ablation leads to starvation3,4. Alhadeff et al. found that stimulation of the AgRP-expressing neurons mimicked the pain-inhibiting effect of hunger in mice. By contrast, silencing of these cells blocked the reduction of inflammatory pain.

AgRP cells send projections to many brain regions. Not all of these projections directly regulate feeding5,6 — some therefore probably have other roles. Alhadeff and colleagues systematically activated AgRP projections to seven brain regions, to search for the projections that mediate the neurons’ pain-relieving effect during inflammation. They found a powerful reduction in inflammatory pain following stimulation of AgRP-cell projections to a single target region in the hindbrain, the parabrachial nucleus (PBN). This structure is part of a central pain-processing circuit that relays pain signals from the spinal cord to various brain regions. Notably, the neurons that receive AgRP inputs, which are found in the lateral portion of the PBN (the lPBN), are activated by painful stimuli7 and inhibited during feeding7,8. Presumably, then, lPBN neurons act to suppress appetite in threatening conditions, when eating might be dangerous, whereas their inhibition by input from AgRP neurons supports feeding in conditions of inflammatory pain.

AgRP neurons produce three neurotransmitter molecules that stimulate feeding9: AgRP itself, γ-aminobutyric acid (GABA) and neuropeptide Y (NPY). Such co-transmission of signals by multiple molecules is widespread in the brain, but breaking down co-transmission into its constituent parts to understand its functions is challenging. Alhadeff et al. overcame this challenge, investigating which of the three molecules were essential for the pain-inhibiting effect of AgRP neurons by injecting each neurotransmitter into the lPBN. NeitherAgRP nor GABA had a pain-relieving effect. But NPY suppressed inflammatory pain by acting through the Y1 receptor on lPBN neurons.

Finally, the authors demonstrated that acute pain led to a sharp decrease in the activity of AgRP neurons. A similar decrease in AgRP activity occurs when an animal first senses food10, and this change in activity is thought to be important for the termination of further food seeking and a transition to food intake, which is then positively reinforced by structures in the hypothalamus other than the arcuate nucleus11. Taking this together with the authors’ data, a picture emerges in which acute pain prompts a behavioural transition by suppressing the activity of AgRP neurons. This inhibition prevents the AgRP cells from activating downstream brain regions involved in feeding, and enables pain signals from the spine to spread from the lPBN to other brain regions, indicating the need to avoid noxious stimuli (Fig. 1a).

Figure 1 | Getting priorities right in the brain. Alhadeff et al.1 have described a population of neurons that express agouti-related protein (AgRP) and regulate the competing needs of hunger and pain in the mouse brain. a, When a mouse is subject to acute pain, AgRP-expressing neurons are inhibited (dashed arrow), and feeding is suppressed. Pain signals from the spinal cord are transmitted throughout the brain via a region called the parabrachial nucleus (PBN). b, By contrast, AgRP-expressing neurons remain active during long-term pain, such as that caused by inflammation. The neurons send signals to the PBN to prevent pain transmission to other brain regions, and so feeding is supported.

By contrast, inflammatory pain does not require rapid behavioural responses and is filtered out by active AgRP cells, which might reduce the activity of lPBN neurons to prevent spreading of pain information to other brain regions and so maintain food seeking (Fig. 1b). This previously unknown mechanism for the management of competing needs provides insights into how hypothalamic computations use both the neurochemical properties and the connectivity of neural circuits to make adaptive decisions about behaviour.

Alhadeff and colleagues’ work has several implications. First, it provides evidence that the potency of AgRP-mediated long-term pain relief is comparable to that of opiates — at least, in the authors’ long-term pain test. As they point out, differences in the processing of acute and chronic pain suggest that treatments for the two should be aimed at different target cells or proteins. In addition, designing painkillers that lack the off-target effects of opiates is desirable. Alhadeff and colleagues point to NPY–Y1-receptor signalling in the lPBN as a potential site of action for chronic painkillers.

Second, the authors’ thorough characterization of a pathway in which signals for two negative states (hunger and pain) interact paves the way to understanding the biological mechanisms that define other complex and dynamic hierarchies in human and animal behaviours. Similar principles at work in other brain regions might further support unaltered food intake during inflammatory pain in hungry mice — for example, by promoting pain inhibition during meals. This painkilling effect probably would not rely on AgRP cells, because their activity is reduced after the sensory detection of food10, so the behavioural hierarchy at work during feeding itself is probably controlled by other neuronal populations. One possibility is that this hierachy is mediated by neurons in the lateral hypothalamus, which is connected to the PBN12 and contains several groups of neurons that are active during feeding1315.

As another example, concurrent negative states of hunger and food aversion contribute to eating disorders such as anorexia nervosa: food-related cues elicit aversion, impairing food intake. Delineating interactions between the neurons that mediate hunger and those that control emotional responses to food could shed light on the mechanisms underlying eating disorders.

Nature 556, 445-446 (2018)

doi: 10.1038/d41586-018-04759-0
Nature Briefing

Sign up for the daily Nature Briefing email newsletter

Stay up to date with what matters in science and why, handpicked from Nature and other publications worldwide.

Sign Up


  1. 1.

    Alhadeff, A. L. et al. Cell 173, 140–152 (2018).

  2. 2.

    Aponte, Y., Atasoy, D. & Sternson, S. M. Nature Neurosci. 14, 351–355 (2011).

  3. 3.

    Gropp, E. et al. Nature Neurosci. 8, 1289–1291 (2005).

  4. 4.

    Luquet, S., Perez, F. A., Hnasko, T. S. & Palmiter, R. D. Science 310, 683–685 (2005).

  5. 5.

    Betley, J. N., Cao, Z. F. H., Ritola, K. D. & Sternson, S. M. Cell 155, 1337–1350 (2013).

  6. 6.

    Steculorum, S. M. et al. Cell 165, 125–138 (2016).

  7. 7.

    Campos, C. A., Bowen, A. J., Roman, C. W. & Palmiter, R. D. Nature 555, 617-622 (2018).

  8. 8.

    Carter, M. E., Soden, M. E., Zweifel, L. S. & Palmiter, R. D. Nature 503, 111–114 (2013).

  9. 9.

    Krashes, M. J., Shah, B. P., Koda, S. & Lowell, B. B. Cell Metab. 18, 588–595 (2013).

  10. 10.

    Chen, Y., Lin, Y.-C., Kuo, T.-W. & Knight, Z. A. Cell 160, 829–841 (2015).

  11. 11.

    Sternson, S. M. & Eiselt, A.-K. Annu. Rev. Physiol. 79, 401–423 (2017).

  12. 12.

    Saper, C. B. & Loewy, A. D. Brain Res. 197, 291–317(1980).

  13. 13.

    Jennings, J. H. et al. Cell 160, 516–527 (2015).

  14. 14.

    Nieh, E. H. et al. Cell 160, 528–541 (2015).

  15. 15.

    Carus-Cadavieco, M. et al. Nature 542, 232–236 (2017).

Download references