It is a general belief that the establishment of long-term memory is associated with its long-lasting storage in synapses, which is accomplished either by an addition of new synaptic contacts or an increase in the strength of existing ones. Numerous attempts have been made to define structural correlates of these processes with the aid of two basic approaches. One of these, electron microscopy, has a serious limitation because it provides static images of synapses fixed at a given moment after a new behavior has been learned. The other approach involves the use of two-photon laser scanning microscopy in combination with molecular probes for time-lapse imaging of dendritic spines on living neurons in slices and in vivo. The latter technical developments have produced a flurry of studies demonstrating that spines, which typically receive a single excitatory synapse, are highly dynamic structures (reviewed in refs. 1-3). Although some spines constantly outgrow from and are retracted into their parent dendrites, there are spine subpopulations that remain stable over a period of months and may represent loci of long-term information storage. In spite of its advantages, however, two-photon laser scanning microscopy lacks the resolution necessary to visualize synapses. Therefore, electron microscopy, with the aid of modern stereological techniques, remains the ultimate method for obtaining rigorous estimates of synapse number and for examining structural synaptic changes. The earlier studies on changes in synapse number and structure associated with learning and memory were extensively reviewed previously4-8. In this chapter, we focus our discussion on the data obtained over the last decade by quantitative electron microscopic analyses of the mammalian brain following acquisition of newly learned behaviors.
Was this article helpful?