Stimulated spines do not remain enlarged if the enlargement pool (colored) is quickly extruded through the spine neck.
KASAI/ELSEVIER
Memory is thought to stem from long-term potentiation (LTP)—the lasting enhancement of communication between two neurons at a synapse. On the postsynaptic side, information is collected in tiny bulbous membrane protrusions called spines, whose enlargement helps create LTP. Enlargement requires actin polymerization, prompting the authors to examine spine actin organization in brain slices before and after synaptic stimulation.
Unstimulated spines contained a stable pool of actin filaments near the base of the spine and a more dynamic pool throughout. The behavior of the dynamic set resembled that of actin in axonal growth cones, where actin is assembled by Rac at the leading edge and disassembled further back.
A third actin pool appeared in spines that swelled after repeated stimulation with glutamate. Polymerization of this pool seemed to cause the spine expansion, as its appearance correlated with membrane ruffling at spine edges. Its polymerizing enzyme is not yet known, but the group found that it required calcium and especially high concentrations of actin monomers.
In some spines, the enlargement-associated pool was quickly pushed out en masse into the dendrite body through the bud neck, and the spine shrunk back to its former size. The pushing force probably stems from surrounding glia and other neurons.
Lasting growth required more of the stable actin filaments, which might originate from the enlargement pool. Only spines that held onto their enlargement pool for longer than six minutes were still enlarged an hour later. Confinement of the pool required CaM-dependent kinase II, which the authors hypothesize helps cross-link the new filaments, making them stiffer and more difficult to squeeze through the bud neck. Smaller spine necks also helped hold them in.
The findings explain why larger spines have proportionately more glutamate receptors, since the receptors dock to the ends of actin filaments. The resulting increased responsiveness to glutamate in turn helps LTP set in. Enlargement itself probably assists. “So many enzymes are needed for LTP,” says Kasai. “The spine becomes a sort of incubator, and the extra space allows subsequent events to happen.” 
Reference:
