News & Views | Published:


Lipid signals in pain control

Cyclooxygenase enzymes produce lipid messenger molecules whose roles in health or disease depend on their context. The discovery of cyclooxygenase-3 should enhance our knowledge of such events.

Some of the most widely used medicines today are aspirin and other such nonsteroidal anti-inflammatory drugs (NSAIDs), which are well known for their pain-relieving, fever-reducing and anti-inflammatory effects. Simply put, these drugs work mainly by inhibiting the formation of prostaglandins — potent lipid messenger molecules — and they do this by blocking cellular cyclooxygenase enzymes, of which two have previously been discovered. It has long been a puzzle, however, that acetaminophen, otherwise known as paracetamol, works well to relieve pain and reduce fever, yet is not an effective anti-inflammatory medicine (although it is often classified as an NSAID). Also, it does not markedly inhibit the two known cyclooxygenases. As they describe in Proceedings of the National Academy of Sciences, Daniel Simmons and colleagues1 might have found part of the solution: they show that acetaminophen inhibits a third cyclooxygenase, which is strongly expressed in the brain and may be involved largely in mediating pain.

The membranes of our cells are a Pandora's box of lipid messenger molecules, many of which have powerful effects — some good, some bad — on the body. Enzymes such as phospholipase A2 play the role of Pandora in releasing these messengers, by splitting a certain chemical bond in phospholipids, the predominant membrane components. Phospholipase A2, for instance, releases arachidonic acid from phospholipids; arachidonic acid in turn serves as a substrate for the cyclooxygenases, which generate a short-lived intermediate, prostaglandin H2. This lipid is the precursor to a variety of derivatives, which share family resemblances in their chemical structures, but do different things. For example, some make nerve endings hypersensitive and others lead to inflammation or fever. Some activate platelets, thereby contributing to blood clotting, and others help protect the stomach lining. Yet more participate in birth, bone remodelling, and sleep induction; some modulate neurotransmitter release; some are anticonvulsive; and others might participate in Alzheimer's disease.

Back in 1971 only one cyclooxygenase was known, and Vane2 discovered that it is inhibited by NSAIDs. This finding provided a unifying explanation for the fever-reducing, pain-relieving and anti-inflammatory actions of NSAIDs, and for the fact that these drugs have qualitatively similar side-effects, such as preventing blood clotting and harming the gastrointestinal tract. The results also firmly established the prostaglandins as critical mediators of inflammatory disease.

But the discovery could not explain why different NSAIDs, at similar doses, have quantitatively different effects and side-effects. A mystery surrounded acetaminophen in particular. Hugely popular, with a history almost as venerable as that of aspirin, this drug reduces pain and fever but has little anti-inflammatory activity, and has virtually no side-effects on the stomach and platelets (although there have been concerns about its effects on the liver). It was suggested that acetaminophen specifically affects the brain, which mediates pain and controls fever, and the general idea developed that there are various — perhaps tissue-specific — cyclooxygenases that are affected differently by different NSAIDs3.

That idea lay dormant until the early 1990s, when, amid great excitement, researchers discovered4,5 a second cyclooxygenase, COX-2. The original enzyme, now dubbed COX-1, is expressed constitutively, including in platelets and the stomach (Fig. 1). By contrast, COX-2 is generally expressed under specific circumstances, largely as a result of inflammation: it is induced by cytokine proteins associated with inflammation, and by injury, seizures, and so on6. However, in the brain, kidneys and some other tissues, COX-2 is expressed constitutively, illustrating the multifunctional nature of these enzymes. For example, the brain COX-2 gene is expressed as a result of synaptic activation in a totally physiological event7.

Figure 1: The growing family of cyclooxygenases.

The figure shows the known cyclooxygenases (COXs), some of their known sites of expression and functions, and the drugs (including nonsteroidal anti-inflammatory drugs, NSAIDs) that inhibit them. Simmons and colleagues1 have discovered that several enzymes are encoded by the COX-1 gene, including COX-3 and two small partial COX-1 proteins, PCOX-1a and PCOX-1b. COX-3 is very sensitive to acetaminophen.

It also soon became clear that different NSAIDs have different effects on these two cyclooxygenases, and that this could explain the drugs' quantitatively different effects. For instance, drugs that inhibit COX-2 strongly but COX-1 weakly show potent anti-inflammatory activity, with fewer gastrointestinal side-effects8. Some of these drugs — etodolac, meloxicam and nimesulide — already existed, but were now found to inhibit COX-2 preferentially. Others were developed as selective COX-2 inhibitors, including rofecoxib and celecoxib.

Yet the discovery of COX-2 did not solve the acetaminophen mystery. At the concentrations used therapeutically, this drug inhibits both COX-1 and COX-2 only weakly. This suggested the existence of yet more cyclooxygenases, and a few years ago Simmons and colleagues9 discovered, in cell lines, some forms of COX-2 with slightly different properties. Now the Simmons group1 has discovered an entirely new cyclooxygenase, COX-3, which they believe to be the acetaminophen-sensitive species. They have also found two smaller forms of COX-1.

All of these are derived from the COX-1 gene by 'alternative splicing' of the COX-1 messenger RNA. So, to make the mature mRNA encoding COX-1 protein, cells remove a particular stretch of sequence, intron 1. To make the mature mRNA for COX-3, intron 1 is included. The same is true of the mRNA encoding one of the small COX-1 derivatives, although here some further sequences are removed. COX-3 therefore has the same sequence (except for intron 1) as COX-1, and both are bound to the cell membrane. The retention of intron 1 might change the way the enzyme molecule is folded, or the conformation of its active site.

Simmons and colleagues1 also found that, in humans, COX-3 is most abundant in the heart and the cerebral cortex, where it is present at about 5% of COX-1's concentration. In vitro, COX-3 is more sensitive than COX-1 or COX-2 to inhibition by acetaminophen, and is much more sensitive to diclofenac, indomethacin, ibuprofen and aspirin — the NSAIDs tested so far. So it seems plausible that the pain-relieving effects of these drugs are mediated largely through the nervous-system-located COX-3. It's still not clear, however, how acetaminophen reduces fever. When the COX-2 gene is knocked out in mice, the drug no longer has fever-reducing effects, which might suggest that it works through the COX-2 protein to reduce fever. Yet concentrations of acetaminophen that reduce fever in vivo do not appear to inhibit COX-2. Perhaps, like the COX-1 gene, the COX-2 gene can also produce different enzymes, one of which is affected by acetaminophen. Simmons and colleagues1 also suggest that there may be other COX enzymes.

Why are there several different cyclooxgenases? They all seem to have broadly similar structures, and catalyse the same reaction. And how can they have such different effects? COX-2, for instance, is expressed normally in several tissues, whereas elsewhere its production is increased rapidly in pathological conditions7. In the brain, prostaglandin E2 produced by COX-2 specifically regulates the synaptic plasticity that is relevant to memory10. And phospholipase A2 and arachidonic acid modulate excitatory neurotransmission and neuronal survival11, through mechanisms that may involve a cyclooxygenase. This diversity could stem from the fact that cyclooxygenases have substrates other than arachidonic acid, and act in different cellular locations and signalling pathways. For example, COX-2 converts anandamide and 2-acylglycerol — lipids that bind to cannabinoid receptors on specific nerve cells — into prostaglandin ethanolamines12,13; this may be involved in pain mechanisms. COX-2 also selectively oxygenates N-arachidonoyl glycine, resulting in amino-eicosanoids14, whose exact function is unclear.

Simmons and colleagues' discovery1 of a third cyclooxygenase will no doubt herald yet more twists and turns to this story. There is still much to learn about these enzymes, but the effort will be well worthwhile given their many roles in health and disease.


  1. 1

    Chandrasekharan, N. et al. Proc. Natl Acad. Sci. USA 99, 13926–13931 (2002).

  2. 2

    Vane, J. R. Nature New Biol. 231, 232–235 (1971).

  3. 3

    Flower, R. J. & Vane, J. R. Nature 240, 410–411 (1972).

  4. 4

    Xie, W. L., Chipman, J. G., Robertson, D. L., Erikson, R. L. & Simmons, D. L. Proc. Natl Acad. Sci. USA 88, 2692–2696 (1991).

  5. 5

    Fletcher, B. S., Kujubu, D. A., Perrin, D. M. & Herschman, H. R. J. Biol. Chem. 267, 4338–4344 (1992).

  6. 6

    Marcheselli, V. L. & Bazan, N. G. J. Biol. Chem. 271, 24794–24799 (1996).

  7. 7

    Bazan, N. G. Nature Med. 7, 414–415 (2001).

  8. 8

    Warner, T. D. et al. Proc. Natl Acad. Sci. USA 96, 7563–7568 (1999).

  9. 9

    Simmons, D. L., Botting, R. M., Robertson, P. M., Madsen, M. L. & Vane, J. R. Proc. Natl Acad. Sci. USA 96, 3275–3280 (1999).

  10. 10

    Chen, C., Magee, J. & Bazan, N. J. Neurophysiol. 87, 2851–2857 (2002).

  11. 11

    Kolko, M., DeCoster, M. A., Rodriguez de Turco, E. B. & Bazan, N. G. J. Biol. Chem. 271, 32722–32728 (1996).

  12. 12

    Dinh, T. P. et al. Proc. Natl Acad. Sci. USA 99, 10819–10824 (2002).

  13. 13

    Walker, J. M., Huang, S. M., Strangman, N. M., Tsou, K. & Sañudo-Peña, M. C. Proc. Natl Acad. Sci. USA 96, 12198–12203 (1999).

  14. 14

    Prusakiewicz, J. J., Kingsley, P. J., Kozak, K. R. & Marnett, L. J. Biochem. Biophys. Res. Commun. 296, 612–617 (2002).

Download references

Author information

Rights and permissions

Reprints and Permissions

About this article

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.