Response: astrocyte mediated control of cerebral microcirculation.

We thank Christopher Anderson and Maiken Nedergaard for their balanced, constructive comments on our recent paper describing the central role of astrocytes in the control of cerebral microcirculation [1]. Anderson and Nedergaard raised the intriguing possibility that propagation of intercellular Ca2+ waves between astrocytes, whether mediated by gap junction communication, ATP and/or glutamate [2 and 3], might be involved in the control of arteriole tone by astrocytes. Although in our study we did not specifically address this issue, we often observed that the increase in the concentration of intracellular Ca2+ ([Ca2+]i) triggered by neuronal stimulation in a distinct astrocyte was often followed by [Ca2+]i elevations in other astrocytes surrounding the same arteriole. The timing for the Ca2+ response in these astrocytes could be compatible with that for a [Ca2+]i wave propagating between these cells. Depending on the absence or presence of [Ca2+]i waves, the following two diverse models can, thus, be conceived. In the first, [Ca2+]i elevations triggered by neuronal activity in perisynaptic astrocytic processes spread intracellularly to processes in contact with blood vessels, and there trigger the release of vasoactive agents. Accordingly, the neuronal control of blood flow could be exerted through astrocytes that have processes in contact with both synapses and vessels. These astrocytes might represent a subpopulation of cells specifically dedicated to neurovascular coupling. In the second model, the activation of astrocyte processes at the synapse is transferred through a propagating [Ca2+]i wave to a number of perivascular astrocytes, including some astrocytes that are remote from the site of activation. In such a case, the resulting upstream vasodilation might have a major impact on the overall conductance of the vascular network in that region. Gap-junction communication certainly adds a further step of complexity in the mechanism that underlines astrocytic control of arteriole tone. Interestingly, this form of communication has been observed also between astrocytes and endothelial cells, at least in cultures [3]. If such signalling exists in situ, astrocytes could transfer [Ca2+]i elevations to endothelial cells which, in turn, could release vasodilating agents, such as prostaglandin I2. Astrocytes can also activate [Ca2+]i signals in endothelial cells trough an ATP-mediated signalling pathway [4]. Endothelial cell activation might, thus, represent the final step in the cellular mechanism at the basis of the astrocyte control of cerebral microcirculation. Given that endothelial cells are also coupled to one another, the [Ca2+]i signal might ultimately involve many of these cells from upstream and downstream vessels. Coordination of large and small arteriole behaviour by this signal might be pivotal for the adaptation of blood flow to altered energy demands of active neurons. Anderson and Nedergaard also raised the question of potential differences in the astrocyte response between the stimulation that is normally used in brain slices and that used in most of our experiments, which is more prolonged and of higher intensity. The rationale behind our stimulation paradigm (a high-intensity and prolonged stimulus) was that of activating long-lasting [Ca2+]i oscillations in as many astrocytes as possible. We would like to underline that [Ca2+]i elevations in astrocytes evoked by this type of stimulus were drastically reduced in the presence of antagonists of group I metabotropic glutamate receptors [1, 5 and 6] and were abolished by 1 small mu, GreekImage tetrodotoxin [6] (G. Carmignoto et al., unpublished). It seems, therefore, unlikely that the stimulus we used could have triggered tetrodotoxin-insensitive [Ca2+]i elevations in astrocytes, subsequent a Ca2+-dependent release of glutamate and, finally, activation of metabotropic glutamate receptors in other astrocytes. However, we fully agree with Anderson and Nedergaard that there is an enormous amount of experimental and theoretical work yet to be done to characterize neurone-to-astrocyte signalling at the synaptic level. A better understanding of the rules governing neuronal activity-dependent activation of astrocytes is, indeed, crucial to dissection of the distinct role of these cells as modulators of neuronal transmission (through the release of glutamate, in addition to other neuroactive compounds) on the one hand, and as mediators of the neurovascular coupling (through the release of vasoactive agents) on the other.