Above is a model of a generic synapse, showing the (upper) pre-synaptic nerve terminal and the (lower) post-synaptic neuron. The pre-synaptic neuron is constantly releasing vesicles that contain neurotransmitters (NT) into the synapse that diffuse to the post-synaptic neuron, and when a pre-synaptic nerve terminal receives an action potential, the rate of this release is altered. Upon secreting its store of NT molecules, the extra membrane from the vesicle is usually incorporated into the pre-synaptic plasma membrane.
In order to ensure that the plasma membrane stays at a steady-state level, endocytosis at the nerve terminal must compensate for this by retrieving the plasma membrane back into the cell, where it can re-form vesicles. This compensation come in a few varieties:
1) Kiss and run. Small synaptic vesicles make transient contact with the plasma membrane forming a short-lasting fusion pore, through which the neurotransmitter is released, then it is released.
This seems to be the least common mechanism and it is somewhat controversial.
2) Fuse and collapse. This is a common endocytosis mechanism in all cells. It’s slower than the other mechanisms because it requires the recruitment of the clathrin complex by the adaptor protein-2 (AP-2) complex. Clathrin, a protein, is recruited to form a lattice around the membrane. It is removed after it collapses the vesicle from the plasma membrane (PM).
In at least some terminals this clathrin-mediated endocytosis is triggered by concentrations of >10 μm calcium. A recent study shows that inhibiting the calcium-sensing phosphotase calcineurin slows endocytosis significantly, suggesting that calcineurin is involved in the calcium regulation:
3) Bulk endocytosis. When the above mechanisms become “rate limiting” and the plasma membrane size is larger than a certain comfortable range, the nerve terminal often resorts (excuse the anthropomorphizing) to endocytosing a large area of presynaptic membrane at once, in a clathrin-independent fashion. This is faster, in part because it is non-specific. And it involves its own suite of proteins, some of which you can see in the two-step visualization below:
In pre-synaptic nerve terminals bulk endocytosis leads to the formation of endosomes, from which synaptic vesicles can bud off. This mech often occurs during episodes of high-frequency action potentials.
Note that although all of the above drawings are in 2d, the brain is 3d so it will be more useful if you can visualize it like that. Models of endocytosis are attracting and should continue to attract attention because of the key role they likely play in synaptic plasticity.
Sun T et al, 2010, The Role of Calcium/Calmodulin-Activated Calcineurin in Rapid and Slow Endocytosis at Central Synapses, J Neuro, doi: 10.1523/JNEUROSCI.1481-10.2010
Shupliakov O, et al. Synaptic Endosomes, NCBI bookshelf ID NBK6352, link.
Jarousse N et al, 2001. Endocytotic mechanisms in synapses. Current Biology doi:10.1016/S0955-0674(00)00237-4
Ryan T, 2003 Kiss-and-run, fuse-pinch-and-linger, fuse-and-collapse: The life and times of a neurosecretory granule doi: 10.1073/pnas.0530260100
Kim Y et al, 2005. Interaction of SPIN90 with Dynamin I and Its Participation in Synaptic Vesicle Endocytosis , J Neuro, DOI:10.1523/JNEUROSCI.1643-05.2005
Kessels M et al, 2004. The syndapin protein family: linking membrane trafficking with the cytoskeleton. J of Cell Science, doi: 10.1242/jcs.01290