The brain relies on blood to supply oxygen and glucose for energy. Surprisingly, neuronal activity, acting via supporting cells called astrocytes, can increase or decrease blood flow depending on whether oxygen levels are low or high.
The air entering our lungs contains 21% oxygen, and by the time the oxygen reaches our brain this value has fallen to 4%. Yet physiological experiments on isolated brain tissue traditionally use solutions containing 95% oxygen. On page 745 of this issue, Gordon et al.1 demonstrate how this experimental oversupply of oxygen may mislead us about the mechanisms that are physiologically important in the brain, and at the same time they provide a notable advance in our understanding of how the brain regulates its energy supply.
When an area of the brain is active, it needs more energy to power information processing. Information flow along and between neurons is based on the movement of ions across the cell membrane, and, to generate the energy required for pumping ions out, neurons depend largely on glucose and oxygen supplied by the blood.
One way in which neurons control their energy supply is by molecular signalling to surrounding support cells called astrocytes. This signalling is mediated by the neurotransmitter glutamate, which raises the calcium ion (Ca2+) concentration in astrocytes, thereby generating arachidonic acid from membrane lipids. Surprisingly, arachidonic acid can either dilate2,3 or constrict4 the blood arterioles to increase or decrease blood flow, respectively. Dilation follows the conversion of arachidonic acid to prostaglandin E2 and epoxygenase derivatives, which diffuse to the smooth-muscle cells encircling nearby arterioles and relax them. Constriction, by contrast, is caused by arachidonic acid itself diffusing to the smooth muscle, where its derivative 20-HETE causes arteriole constriction by increasing Ca2+ entry through voltage-gated channels4.
What determines whether arterioles dilate or constrict in response to a high Ca2+ concentration in astrocytes? Gordon et al.1 find that the prevailing oxygen level is a key determinant of whether astrocytes increase or decrease blood flow.
Two mechanisms seem to mediate this oxygen-dependent switch. First, at low oxygen levels, increased energy generation in neurons and astrocytes by a pathway that does not use oxygen (glycolysis) results in the extracellular accumulation of lactate. Because transporter proteins in the cell membrane carry prostaglandins into cells in exchange for lactate5, lactate levels affect how prostaglandins are removed from the extracellular space. A high extracellular concentration of lactate inhibits prostaglandin removal, enhancing extracellular prostaglandin levels and so dilating arterioles1.
Second, because at low oxygen levels there is less energy available to make the high-energy compound ATP, intracellular concentrations of the ATP precursor adenosine rise1. Adenosine is then transported out of neurons and astrocytes, and inhibits Ca2+ entry into the arteriole smooth muscle through voltage-gated channels, preventing the constriction that its entry would induce. Thus, low oxygen levels promote increased blood flow by simultaneously potentiating the dilation pathway and inhibiting the constriction pathway in astrocytes1.
As with all major advances, Gordon and colleagues' study raises as many questions as it answers. For example, how much lactate is needed to promote dilation? The authors' data1 suggest that physiological levels of extracellular lactate (about 1 millimolar) in the brain can inhibit prostaglandin uptake, yet when prostaglandin transporters are expressed in a cancer cell line5, much higher lactate concentrations (48 mM) are needed to produce 50% inhibition. Conceivably, the transporter's affinity for lactate is higher in astrocytes than in the cell line. It is noteworthy that lactate can also dilate arterioles by a prostaglandin-independent mechanism6.
Gordon et al. demonstrate astrocyte-mediated dilation of arterioles in solutions containing 20% oxygen, which produces approximately physiological levels of oxygen in brain slices7. They also find that astrocyte-mediated constriction occurs in solutions containing oxygen levels well above the physiological (95%). Why has evolution produced the latter pathway, in which neural activity decreases blood flow? And will physiological tissue concentrations of oxygen ever be high enough to activate this pathway? It turns out that 20-HETE-mediated arteriole constriction is inhibited8 by nitric oxide (NO), a molecule that is released by neurons in response to glutamate secreted by neighbouring neurons (and which can also directly dilate arterioles). The 20-HETE-mediated pathway may therefore be better viewed as a mechanism producing a basal constriction of arterioles that can then be modulated by NO to provide another pathway for activity-dependent dilation.
Future work is likely to focus on how changes in the levels of lactate, adenosine, oxygen and NO interact to coordinate blood flow and hence the brain's energy supply. Some clues can be found in previous data. For example, NO released by neurons inhibits the conversion of arachidonic acid to epoxygenase derivatives that evoke dilation9. As NO production in neurons requires oxygen10, at low oxygen levels this mechanism will be inhibited, promoting dilation. Moreover, oxygen is needed for the synthesis of both constricting (20-HETE) and dilating (prostaglandin E2 and epoxygenase) derivatives of arachidonic acid. At low oxygen levels, however, the production of 20-HETE is inhibited more strongly than that of prostaglandin E2 and epoxygenase derivatives8, increasing dilation. Finally, in blood capillaries, where contractile cells called pericytes may regulate blood flow11, lactate causes constriction at high oxygen levels, but dilation at low levels12. There is, therefore, an array of switching mechanisms that promote brain energy supply when oxygen levels fall.
In a wider context, Gordon and colleagues' observations raise questions for both cognitive neuroscientists and neurologists. Could the initial dip in local oxygen concentration that accompanies neural activity13 affect astrocyte signalling rapidly enough to contribute to the increase in blood flow that generates the signals seen in functional imaging of the brain? And could our new understanding of astrocyte signalling lead to better therapies for correcting disorders of blood flow in the brain, such as those that occur after stroke and in vascular dementia?
Gordon et al.1 have opened a fresh chapter in our investigation of how blood flow is regulated in the brain. But their work has a broader implication: physiological studies using solutions bubbled with 95% oxygen may be altering the operation of signalling pathways in the brain, producing misleading results.