| Synaptic dynamics are critical to the function of neuronal circuits on multiple timescales. In the first part of this dissertation, I tested the roles of action potential timing and NMDA receptor composition in long-term modifications to synaptic efficacy. In a computational model I showed that the dynamics of the postsynaptic [Ca2+] time course can be used to map the timing of pre- and postsynaptic action potentials onto experimentally observed changes in synaptic strength. Using dual patch-clamp recordings from cultured hippocampal neurons, I found that NMDAR subtypes can map combinations of pre- and postsynaptic action potentials onto either long-term potentiation (LTP) or depression (LTD). LTP and LTD could even be evoked by the same stimuli, and in such cases the plasticity outcome was determined by the availability of NMDAR subtypes. The expression of LTD was increasingly presynaptic as synaptic connections became more developed. Finally, I found that spike-timing-dependent potentiability is history-dependent, with a non-linear relationship to the number of pre- and postsynaptic action potentials. After LTP induction, subsequent potentiability recovered on a timescale of minutes, and was dependent on the duration of the previous induction.;While activity-dependent plasticity is putatively involved in circuit development, I found that it was not required to produce small networks capable of exhibiting rhythmic persistent activity patterns called reverberations. However, positive synaptic scaling produced by network inactivity yielded increased quantal synaptic amplitudes, connectivity, and potentiability, all favoring reverberation. These data suggest that chronic inactivity upregulates synaptic efficacy by both quantal amplification and by the addition of silent synapses, the latter of which are rapidly activated by reverberation. Reverberation in previously inactivated networks also resulted in activity-dependent outbreaks of spontaneous network activity. Applying a model of short-term synaptic dynamics to the network level, I argue that these experimental observations can be explained by the interaction between presynaptic calcium dynamics and short-term synaptic depression on multiple timescales. Together, the experiments and modeling indicate that ongoing activity, synaptic scaling and metaplasticity are required to endow networks with a level of synaptic connectivity and potentiability that supports stimulus-evoked persistent activity patterns but avoids spontaneous activity. |
