Normal transmission across an excitatory synapse relies on the neurotransmitter glutamate, the glutamate-specific AMPA receptor (AMPAR), and calcium ions. Calcium ion entry into the presynaptic terminal causes the presynaptic release of glutamate which diffuses across the synaptic cleft binding to glutamate receptors on the postsynaptic membrane. There are four main types of glutamate receptors: AMPA receptors (AMPARs), NMDA receptors (NMDARs), Kainate receptors, and quisqualate receptors, some of which are also metabotropic receptors. Most research has been focussed on the AMPARs and the NMDARs. When glutamate binds to AMPARs located on the postsynaptic membrane, they permit a mixed flow of Na+ and K+ to cross the cells membrane, causing a depolarization of the postsynaptic membrane. This depolarization is called the excitatory postsynaptic potential (EPSP).
Silent synapses release glutamate as do "normal" synapses, but they lack AMPARs on the surface membrane of the postsynapse. Only NMDARs (and perhaps metabotropic receptors) are found in the surface postsynaptic membrane where they can bind synaptically released glutamate. AMPARs are not completely absent from silent synapses, they are simply located inside the postsynaptic cell, where they cannot detect extracellular glutamate. The NMDAR is functionally similar to AMPAR except for two major differences: NMDARs carry ion currents composed of Na+, K+, but also (unlike most AMPAR) Ca2+; NMDARs also have a site inside their ion channel that binds magnesium ions (Mg2+). This magnesium binding site is physically located in the channel at a place within the electrical field generated by the membrane potential. Normally, current will not flow though the NMDAR channel, even when it has bound glutamate. This is because the ion channel associated with this receptor is plugged by magnesium, acting like a cork in a bottle. However, since the Mg2+ is charged and is bound within the membrane's electric field, depolarization of the membrane potential above can dislodge the magnesium, allowing current flow through the NMDAR channel. This gives the NMDAR the property of being voltage-dependent, in that it requires strong postsynaptic depolarization to allow ion flux.
The stimulation of a silent synapse does not elicit EPSPs when the postsynaptic cell is clamped at -60 mV. Stimulation of a silent synapse will elicit EPSPs when the postsynaptic cell is depolarized beyond -40 mV. This is because they lack surface AMPAR to pass current at hyperpolarized potentials, but do possess NMDARs that will pass current at more positive potentials (because of relief of magnesium block) Moreover, the EPSPs elicited with depolarized membrane potentials can be completely blocked by APV, a selective NMDAR blocker.
Silent synapses are activated via the insertion of AMPARs into the postsynaptic membrane, a phenomenon commonly called "AMPA receptor trafficking".
When glutamate binds to a strongly-depolarized postsynaptic cell (e.g., during Hebbian LTP), Ca2+ quickly enters and binds to calmodulin. Calmodulin activates calcium/calmodulin-dependent protein kinase II (CaMKII), which — among other things — acts on AMPAR-containing vesicles near the postsynaptic membrane. CaMKII phosphorylates these AMPARs, which serves as a signal to insert them into the postsynaptic membrane. Once AMPARs are inserted, the synapse is no longer silent; activated synapses no longer require simultaneous pre- and postsynaptic activity in order to elicit EPSPs.
Evidence suggests that dendrite arborization and synapse maturation 1 (Dasm1),an Ig superfamily member, is involved in the maturation of synapses, essentially "awakening" the silent synapses.
The characterization of silent synapses is an ongoing field of research and there are many things about them that are not yet known. Some of what is currently accepted about the properties of silent synapses may still prove to be incorrect in whole or in part. Some controversies about silent synapses have however, been settled. For example, until recently, there were four competing hypotheses for the mechanisms of synapse silence (see Voronin et al., 2004).
All four of these hypotheses had their adherents, but the first three were largely ruled out as a mechanism for synapse silence by work published since 2002 (e.g. Montgomery et al. 2002).