In recent years, astrocytes have emerged as key modulators of neuronal excitability and synaptic transmission. It is known that they can use the chemical transmitter glutamate to 'talk' to neurons and synapses, but the nature of the glutamate storage and release machinery has been unclear. Now, however, Bezzi and colleagues present evidence that astrocytes release glutamate through a SNARE-dependent exocytic mechanism, which is strikingly similar to the mechanism of neurotransmitter release at nerve terminals.

Bezzi et al. showed that astrocytes in the rat hippocampus express transcripts that code for the vesicular glutamate transporters VGLUT1 and VGLUT2, which are responsible for glutamate uptake into synaptic vesicles. Using immunogold cytochemistry and electron microscopy, the authors pinpointed the location of these molecules to vesicular organelles, which were observed in regions of astrocytic processes that impinged on neuronal structures. In some cases, astrocytic vesicles faced neuronal NMDA (N-methyl-D-aspartate) receptors at nearly synaptic distance. The vesicles also expressed cellubrevin — a member of the vesicular SNARE (v-SNARE) family of proteins, which regulate the fusion of exocytic vesicles with the plasma membrane

Next, the authors used cultured astrocytes to examine whether the vesicles underwent regulated exocytosis in response to group I metabotropic glutamate receptor (mGluR) activation. They observed the behaviour of the vesicles using total internal reflection fluorescence imaging, a technique that allows dynamic cellular processes to be visualized at high resolution. They labelled the cells with acridine orange, a dye that accumulates in acidic cellular compartments in a self-quenched state and produces a flash of fluorescence when it is released at the cell surface. The astrocytes were also engineered to express VGLUT1 or VGLUT2 tagged with enhanced green fluorescent protein. The authors showed that the mGluR agonist dihydroxyphenylglycine evoked flashes of fluorescence at the cell membrane, concomitant with the disappearance of some of the VGLUT-expressing vesicles.

So, there was clear evidence that exocytosis was taking place in response to mGluR activation, but were the astrocytes actually releasing glutamate? To find out, Bezzi et al. co-cultured astrocytes with 'glutamate-sniffing' cells — insulinoma-1 (INS-1) cells that respond to glutamate application with a rapid elevation in their intracellular calcium concentration owing to the expression of NMDA receptors. The authors showed that the flashes of fluorescence at the astrocyte membrane were followed by an NMDA receptor-dependent increase in intracellular calcium in the INS-1 cells, confirming that glutamate was being released through regulated exocytosis.

Although the astrocytic glutamate-releasing vesicles are similar to synaptic vesicles in many ways, there are also some intriguing differences. For example, the groups of vesicles were often smaller and less well-organized than the vesicles at presynaptic terminals. Also, they express the v-SNARE cellubrevin, whereas synaptic vesicles mostly express VAMP2 (vesicle-associated membrane protein 2). In the future, it will be interesting to explore the implications of these similarities and differences for neuron–glia communication.