Art - Johnston (R)

625 Dynamic changes in synaptic strength are thought to provide a cellular basis for information storage in the nervous system. For example, long-term potentiation (LTP) of synaptic transmission is a long-lasting, activity-dependent form of synaptic plasticity that is expressed by all principal neurons in the hippocampus—a brain structure implicated in certain forms of long-term memory1. In the hippocampus (and elsewhere), multiple forms of LTP have been described: those that require activation of the N-methyl-D-aspartate (NMDA) subclass of glutamate receptors for induction, and those that are independent of NMDA receptor activation2. At nearly all glutamatergic synapses in the hippocampus, including commissural/associational (C/A) synapses in CA3, LTP induction has been firmly established to depend on an initial postsynaptic rise in cytosolic Ca2+ concentration1,3,4. This rise can occur via NMDA receptors, voltage-gated Ca2+ channels or release from internal stores5,6, and in turn, triggers a cascade of biochemical events resulting in the expression of an increase in synaptic strength. Although the site of induction of NMDAreceptor-dependent LTP is clearly postsynaptic, the site of expression is more controversial, with reports supporting both pre-7,8 and postsynaptic mechanisms9. The one hippocampal synapse that seems to be an exception to this rule is the mossy fiber input to CA3 pyramidal neurons from dentate granule cells. This synapse has several unusual structural features, including large terminals, multiple release sites and a proximal termination zone along the apical dendrites of CA3 neurons10,11. At this synapse, LTP does not require NMDA receptor activation2 and reportedly can be induced independently of postsynaptic activity; thus, it is thought to be triggered entirely within the presynaptic terminal3,12. Other reports, however, have suggested that similar, to NMDA receptor-dependent LTP, the induction of mossy fiber LTP depends on an increase in postsynaptic [Ca2+]13 and is regulated by the postsynaptic membrane potential14. Apart from several methodological differences, these contrasting results and conclusions have been difficult to reconcile. More recently, it has been suggested that NMDA-receptor-independent LTP at mossy fiber synapses should be subdivided into two forms, depending on the duration of the stimulus train used for induction: brief trains of high-frequency stimulation applied to the mossy fibers (BHFS) induce a form of LTP that depends on an initial postsynaptic step, whereas long trains (L-HFS) elicit presynaptically induced LTP15. Although the induction mechanisms are still unresolved, there is general agreement that the maintenance or expression of mossy fiber LTP is due to presynaptic changes16–18. We have used high-speed fluorescence imaging and improved methods for stimulating and recording from the CA3 region to re-examine the possible contribution of Ca2+ to the induction of these hypothesized multiple forms of LTP, and to determine the specific conditions for their induction. LTP induced with either stimulation protocol similarly required a rise in postsynaptic [Ca2+]. LTP induced with B-HFS or L-HFS was prevented by chelation of postsynaptic [Ca2+], although the concentration of Ca2+ buffer required for blocking LTP was substantially different for the two protocols. Additionally, a rise in postsynaptic [Ca2+], sufficient to induce mossy fiber LTP, occurred even when fast synaptic transmission was blocked by the ionotropic glutamate receptor antagonist kynurenate. These Ca2+ rises resulted from release by internal Ca2+ stores and depended on metabotropic glutamate receptor (mGluR) activation. Finally, LTP was significantly suppressed by postsynaptic inhibition of cAMP-dependent protein kinase. Taken together, these results suggest that mossy fiber LTP shares with other hippocampal synapses the common induction mechanism of an initial rise in postsynaptic [Ca2+].