In many regions of the CNS, Ca2+ influx through NMDA receptors can trigger two forms of synaptic plasticity: long-term depression (LTD) and long-term potentiation (LTP) which are believed to resemble some elementary features of memory formation at the neuronal level (Malenka, 1994; Bear and Malenka, 1994; Edwards, 1995; Collingridge and Bliss, 1995; Rison and Stanton, 1995; Baudry, 1996; Jeffery, 1997). The voltage-dependent blockade of NMDA receptors by Mg2+ and their high Ca2+ permeability renders them inherently suited for their role in mediating synaptic plasticity (Herron et al., 1986). NMDA receptor channels are only activated in the presence of a local strong depolarization induced by strong AMPA receptors activation and concurrent GABAergic dis-inhibition via feedback effects of GABA on GABAB autoreceptors. As a result, the Mg2+ blockade of NMDA receptors is transiently fully relieved allowing Ca2+ to flow into the postsynaptic neurone. This Ca2+ influx triggers a cascade of secondary messengers which ultimately activate a number of enzymes such as protein kinase C (PKC), phospholipase A2 (PLA2), phospholipase C (PLC), Ca2+/calmodulin-dependent protein kinase II (CaM kinase II), etc. (Abraham and Tate, 1997) (Grant and Silva, 1994; Lisman, 1994; Pasinelli et al., 1995; Benowitz and Routtenberg, 1997; Lan et al., 2001; Bayer et al., 2001). Consequently, these processes lead to fixation of changes in postsynaptic AMPA receptors such as an increase in their affinity and/or number (Maren et al., 1993; Ambros-Ingerson and Lynch, 1993; Ambros-Ingerson et al., 1993; Benke et al., 1998) but see (Kessler et al., 1991) and, possibly through retrograde signals (arachidonic acid, nitric oxide), modulate presynaptic glutamatergic terminals influencing transmitter release (Lynch, 1989; Odell et al., 1991; Kato et al., 1991; Schaechter and Benowitz, 1993; Kato et al., 1994; Luo and Vallano, 1995).
There is accumulating evidence that LTP and LTD share some common mechanisms, although LTD occurs with increases in postsynaptic Ca2+, that are insufficient to induce LTP (Artola and Singer, 1993; Christie et al., 1994; Cummings et al., 1996; Derrick and Martinez Jr, 1996; Hansel et al., 1996; Kirkwood et al., 1996; Tsumoto et al., 1996; Tsumoto and Yasuda, 1996; Christie et al., 1996; Artola et al., 1996). LTP and LTD have been extensively studied as cellular models of learning and memory. Although hippocampal long-term potentiation and spatial learning are impaired by NMDA receptor blockade see (Jeffery, 1997) learning deficits can be almost completely prevented if rats are pretrained in a different water maze (Bannerman et al., 1995; Saucier and Cain, 1995). NMDA receptors may therefore not be required for encoding the spatial representation of a specific environment but rather in other forms of memory important for learning this task (Morris, 1996). Recent evidence indicates that LTP is not only important for synaptic plasticity in the mature CNS but also in the formation of conducting glutamatergic synapses in the developing mammalian brain (Durand et al., 1996).
One form of hippocampal LTP involves the activation of the NMDA receptors and a rise in postsynaptic Ca2+ in the CA1 region but there is still considerable debate as to the site at which the increase in synaptic strength is expressed e.g. (Stricker et al., 1996; Stricker et al., 1996; Isaac et al., 1996; Isaac et al., 1996). Presynaptic mechanisms should be reflected in a change in release probability. This can be measured at excitatory synapses on cultured hippocampal neurones by analysis of the progressive block of NMDA receptor-mediated synaptic currents by the essentially irreversible open channel blocker dizocilpine ((+)MK-801) (Rosenmund et al., 1993). This technique was used to demonstrate that release probability was not affected after the induction of LTP making a presynaptic mechanism unlikely (Manabe and Nicoll, 1994). Moreover, recent reports indicate that a high proportion of synapses in hippocampal area CA1 transmit with NMDA receptors but not AMPA receptors, making these synapses effectively non-functional at normal resting potentials due to Mg2+blockade (Liao et al., 1995; Nicoll and Malenka, 1995; Montgomery et al., 2001; Montgomery and Madison, 2002). These silent synapses acquire AMPA-type responses following LTP induction. Furthermore, this form of LTP is accompanied by an increase in the conductance of postsynaptic AMPA receptors (Bibb et al., 2001; Bibb et al., 2001). Taken together, these findings challenge the view that LTP in CA1 involves a presynaptic modification, and suggest instead a simple postsynaptic mechanism for both induction and expression of LTP.