The neurons in our brain communicate with one another through specialized structures called synapses. In certain types of neuron, synapses sit on spiny protrusions. These spines are more than just structural protuberances: it seems that spine growth can be induced by stimulating neurons, indicating that changes in spine density may reflect changes in the synaptic activity -- and hence the computational power -- of individual neurons. Spines are associated mainly with excitatory postsynaptic specializations on neurons that communicate with glutamate. Spines are connected to dendrites by a neck, and they form small, semi-isolated compartments that segregate postsynaptic responses. This compartmentalization of the thousands of synaptic inputs on each excitatory neuron is widely believed to be critical for the neurons' computational power and hence for higher brain functions -- a view supported by the finding that abnormalities in spine density or shape are associated with human cognitive disorders, such as Down's syndrome and fragile X syndrome . So there is considerable impetus to identify the molecular pathways responsible for spine morphogenesis. But what are the molecular mechanisms underlying spine growth? Passafaro et al. identify a key player in spine production -- and surprisingly, it is a protein already known for its important role in synaptic function.
Many molecules have been implicated and the list comprises both signalling and structural proteins, which typically exert their functions through direct or indirect regulation of filamentous actin -- the principal structural component of the spines'intracellular skeleton . Overoproduction of either GluR2 lacking the amino-terminal portion, or a closely related AMPA-receptor subunit (GluR1), had no effect on spine growth. In contrast, a GluR2 produced spine alterations similar to those produced by the intact GluR2 subunit. Furthermore, if GluR2 production was inhibited in neurons, using a technique called RNA interference, these neurons developed fewer spines than normal. Finally, the authors showed that overproduction of GluR2 in neurons, that normally have low levels of this subunit, induced spines on otherwise spine-free branches of these cells. So it appears that amino-terminal portion of GluR2 is both necessary and sufficient for the induction of spines. Passafaro et al. have shown that the extracellular amino-terminal domain of GluR2 also has an important function. How does the amino-terminal region of GluR2 work its wizardry? The authors offer three potential models of how it might activate spine growth. First, it might bind with a secreted factor; second, it might interact with a specific postsynaptic membrane protein; third, it might bind with a protein present on the presynaptic membrane. What is unique about the findings of Passafaro and colleagues is that the effect of the GluR2 amino-terminal domain seems to be independent of ion flux through AMPA-receptor channels containing GluR2. The work of Passafaro et al. allows us a glimpse of how changes in synaptic activity might be connected with structural changes and how an increase in the number of GluR2 containing AMPA receptors could cause parallel increases in both synaptic efficacy and spine density.