A unified model of the presynaptic and postsynaptic changes during LTP at CA1 synapses

Long-term potentiation (LTP) has been studied extensively at CA1 synapses of the hippocampus, and there is evidence implicating both postsynaptic and presynaptic changes in this process. These changes include (i) addition of AMPA channels to the extrasynaptic membrane and diffusional equilibrium of extrasynaptic receptors with synaptic receptors, (ii) sudden addition of AMPA channels to the synapse in large groups, (iii) a change in the mode of glutamate release (presumably from kiss-and-run to full fusion), and (iv) a delayed increase in the number of vesicles released. However, it remains unclear whether (or how) these changes work together. We have incorporated all of these processes into a structural model of the synapse. We propose that the synapse is composed of transsynaptic modules that function quasi-independently in AMPA-mediated transmission. Under basal conditions, synapses are partially silent; some modules are AMPA-silent (but contribute to NMDA-mediated transmission), whereas others are functional (and contribute to both AMPA- and NMDA-mediated transmission). During LTP, there is both a rapid change in the mode of vesicle fusion and a rapid insertion of a postsynaptic complex (a hyperslot) containing many proteins (slots) capable of binding AMPA channels. The combined effect of these pre- and postsynaptic changes is to convert AMPA-silent modules into functional modules. Slot filling is transiently enhanced by a rapid increase in extrasynaptic GluR1, a form of the AMPA-type receptor. A slower transsynaptic growth process adds AMPA-silent modules to the synapse, enhancing the number of vesicles released and thereby enhancing the NMDA response. This model accounts for a broad range of data, including the LTP-induced changes in quantal parameters. The model also provides a coherent explanation for the diverse effects of GluR1 knockout on basal transmission, LTP, and distance-dependent scaling.