Feeding the brain

Article metrics

In computationally active areas of the brain, the blood flow is increased to provide more energy to nerve cells. New data fuel the controversy over how this energy supply is regulated.

Like all tissues, our brains need energy to function, and this comes in the form of oxygen and glucose, carried in the blood. The brain's information-processing capacity is limited by the amount of energy available1, so, as has been recognized for more than a century, blood flow is increased to brain areas where nerve cells are active2. This increase in flow provides the basis for functional magnetic resonance imaging of brain activity2, but exactly how the flow is increased is uncertain. On page 195 of this issue, Mulligan and MacVicar3 reveal a previously unknown role for non-neuronal brain cells called astrocytes in controlling the brain's blood flow. Intriguingly, the new data contradict a previous suggestion for how astrocytes regulate flow.

Figure 1 shows recent developments in our understanding of how the blood flow in the brain is controlled. Glucose and oxygen are provided to neurons through the walls of capillaries, the blood flow through which is controlled by the smooth muscle surrounding precapillary arterioles. Dedicated neuronal networks in the brain signal to the smooth muscle to constrict or dilate arterioles and thus decrease or increase blood flow2; for example, neurons that release the neurotransmitter molecule noradrenaline constrict arterioles. In addition, the neuronal activity associated with information processing increases local blood flow. This is in part due to neurons that release the transmitter glutamate, which raises the intracellular concentration of Ca2+ ions in other neurons, thereby activating the enzyme nitric oxide (NO) synthase and leading to the release of NO. This in turn dilates arterioles4.

Figure 1: Controlling blood flow in the brain.

Computationally active neurons release glutamate (top left). This activates neuronal NMDA-type receptors, Ca2+ influx through which leads to nitric oxide synthase (NOS) releasing NO, which works on smooth muscle to dilate arterioles. This increases the supply of oxygen and glucose to the brain. Glutamate also spills over to astrocyte receptors (mGluRs), which raise the Ca2+ levels in astrocytes and generate arachidonic acid (AA) via phospholipase A2 (PLA2). Cyclooxygenase-generated derivatives of AA (PGE2) dilate arterioles5, whereas, as Mulligan and MacVicar show3, the CYP4A-generated derivative 20-HETE constricts them. Astrocyte Ca2+ levels can also be raised by noradrenaline — released from dedicated neurons that control the circulation — which works through α1 receptors (bottom left). Dotted lines show messengers diffusing between cells. The detailed anatomy of synapses and astrocytes is not portrayed.

A radical addition to this scheme came with the claim of Zonta et al.5 that glutamate also works through astrocytes in the brain to dilate arterioles. Glutamate raises the Ca2+ concentration in astrocytes, and thus activates the enzyme phospholipase A2, which produces a fatty acid, arachidonic acid. This is converted by the enzyme cyclooxygenase into prostaglandin derivatives, which dilate arterioles. An attractive aspect of a role for astrocytes in controlling blood flow is that, although most of their cell membrane surrounds neurons and so can sense neuronal glutamate release, they also send out an extension, called an endfoot, close to blood vessels: thus, astrocyte anatomy is ideal for regulating blood flow in response to local neuronal activity6. In this scheme, a rise in the Ca2+ levels in astrocytes, just like in neurons, would dilate arterioles and increase local blood flow.

The new data contradict these results. Mulligan and MacVicar3 inserted a ‘caged’ form of Ca2+ into astrocytes in brain slices taken from rats and mice. By using light to suddenly uncage the Ca2+, they found that an increase in the available Ca2+ concentration within astrocytes produces a constriction of nearby arterioles that could powerfully decrease local blood flow (the 23% decrease in diameter seen would increase the local resistance to blood flow threefold, by Poiseuille's law).

They show that this constriction results from Ca2+ activating phospholipase A2 to generate arachidonic acid, as above; the twist is that this arachidonic acid is then processed by a cytochrome P450 enzyme (CYP) into a constricting derivative. The authors propose that this derivative is 20-hydroxyeicosatetraenoic acid (20-HETE), formed by CYP4A in the arteriole smooth muscle7 (but the high concentration of CYP4A blocker used to deduce this might also block other enzymes8). The authors also found that noradrenaline evoked a rise in astrocyte Ca2+ concentration and arteriole constriction. Unexpectedly, therefore, it seems that rather than noradrenaline-producing neurons signalling directly to smooth muscle, as is conventionally assumed, much of their constricting action may be mediated indirectly by astrocytes. In fact this is consistent with the finding that many noradrenaline-release sites on neurons are located near astrocytes9.

Is it possible to reconcile the new data3 (a rise in astrocyte Ca2+ levels constricts arterioles) with those of Zonta et al.5 (a rise in Ca2+ dilates arterioles)? A likely solution is that the increased concentration of Ca2+ in astrocytes leads to the production of both constricting CYP-generated derivatives of arachidonic acid and dilating cyclooxygenase-generated derivatives. The relative importance of these opposing effects may have differed in the two groups' experiments because of a difference in the age of the animals or the brain area studied: Mulligan and MacVicar looked at the hippocampus of 13–18-day-old rodents; Zonta et al. focused on the cerebral cortex at 9–15 days. (Furthermore, outside the brain, whether 20-HETE constricts or dilates depends on both tissue area and age10.)

The relative importance of the constricting and dilating signals will also depend on the initial baseline level of contraction of the arteriole smooth muscle. Mulligan and MacVicar found that inhibiting NO synthase converted the astrocyte-evoked constriction into a dilation, implying that the dilation observed by Zonta et al. might be unphysiological and due to their inhibiting NO synthase. However, a rationale for blocking this enzyme is that, in vivo, blood flow generates a baseline level of contraction of arteriole smooth muscle, which is normally absent in brain-slice experiments but can be mimicked by blocking the dilating effects of NO released from neurons and blood vessels. The change from constriction to dilation that Mulligan and MacVicar saw when NO synthase was blocked may reflect a stronger, perhaps more physiological, prior baseline contraction of arteriole smooth muscle in the absence of NO, because from a more constricted baseline it is harder to evoke further constriction and easier to evoke dilation.

Finally, another explanation for the difference between the two groups' results could be that, whereas Mulligan and MacVicar's Ca2+-uncaging experiments elegantly isolate astrocyte-initiated signalling, Zonta and colleagues' use of a glutamate analogue to raise the astrocyte Ca2+ concentration could also activate non-astrocyte signalling pathways, for example in the endothelial cells that line blood vessels11. These path-ways might also be activated by glutamate released in vivo.

Excitingly, Mulligan and MacVicar's discovery of an astrocyte-mediated constriction pathway may also give insight into how arterioles are dilated, in both normal and pathological circumstances. Although normal dilation occurs partly through NO increasing the levels of the molecule cyclic GMP in smooth muscle, NO also inhibits the production of 20-HETE7. So part of NO's dilating effects may result from a suppression of the astrocyte-initiated signalling discovered by Mulligan and MacVicar. The production of 20-HETE is also inhibited when oxygen levels fall7, possibly contributing to the dilation produced by low oxygen levels in the brain.

Eventual reconciliation of the results of Zonta et al.5 and Mulligan and MacVicar3 may require difficult experiments involving perfusing arterioles in brain slices with an artificial extracellular fluid, to generate a physiological level of baseline contraction12. Nonetheless, it is clear that astrocytes play a starring role in the control of blood flow in the brain.


  1. 1

    Laughlin, S. B. & Sejnowski, T. J. Science 301, 1870–1874 (2003).

  2. 2

    Attwell, D. & Iadecola, C. Trends Neurosci. 25, 621–625 (2002).

  3. 3

    Mulligan, S. J. & MacVicar, B. A. Nature 431, 195–199 (2004).

  4. 4

    Faraci, F. M. & Breese, K. R. Circ. Res. 72, 476–480 (1993).

  5. 5

    Zonta, M. et al. Nature Neurosci. 6, 43–50 (2003).

  6. 6

    Simard, M. et al. J. Neurosci. 23, 9254–9262 (2003).

  7. 7

    Roman, R. J. Physiol. Rev. 82, 131–185 (2002).

  8. 8

    Miyata, N. et al. Br. J. Pharmacol. 133, 325–329 (2001).

  9. 9

    Cohen, Z., Molinatti, G. & Hamel, E. J. Cereb. Blood. Flow Metab. 17, 894–904 (1997).

  10. 10

    Fuloria, M., Eckman, D. M., Leach, D. A. & Aschner, J. L. Am. J. Physiol. Lung Cell. Mol. Physiol. 287, L360–L365 (2004).

  11. 11

    Gillard, S. E., Tzaferis, J., Tsui, H. C. & Kingston, A. E. J. Comp. Neurol. 461, 317–332 (2003).

  12. 12

    Brown, L. A., Key, B. J. & Lovick, T. A. Pflügers Arch. 443S, S260 (2002).

Download references

Author information

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Peppiatt, C., Attwell, D. Feeding the brain. Nature 431, 137–138 (2004) doi:10.1038/431137a

Download citation

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.