Lab History


Diagrammatic summary of a hypothesis for synaptic vesicle membrane recycling at the frog neuromuscular junction.

  1. Synaptic vesicles discharge their content of transmitter as they coalesce with the plasma membrane at specific regions adjacent to the muscle.
  2. Equal amounts of membrane are then retrieved when coated vesicles pinch off from regions of the plasma membrane adjacent to the Schwann sheath.
  3. Finally, the coated vesicles lose their coats and coalesce to form cisternae which accumulate in regions of vesicle depletion and slowly divide to form new synaptic vesicles.

From: Heuser, J.E. and T.S. Reese. 1974. "Morphology of synaptic vesicle discharge and reformation at the frog neuromuscular junction", Ch. 4, pp. 59-78, in: Synaptic Transmission and Neuronal Interaction, M.V.L. Bennett, ed., Soc. Gen. Physiol. Series, vol. 28, Raven Press, New York.


Diagram summarizing our current view of vesicle recycling at the frog neuromuscular junction.

A vesicle starts off milling around free in the cytoplasm, until it comes by chance into contact with sticky filaments that compose the pre-synaptic dense tufts. There it is held with a group of other vesicles that quiver close to the presynaptic membrane but cannot quite contact it because of an electrostatic energy barrier. When calcium enters with the nerve impulse, it neutralizes negatively charged spots on the vesicle and plasma membrane and allows the two to contact each other.

In the presence of calcium, and perhaps with the help of intercalated protein molecules, their phospholipids mix at the point of contact until their membranes fuse and a perforation is formed that lets transmitter out of the vesicle. Then the vesicle collapses completely and its membrane diffuses into the presynaptic membrane. An equal amount of membrane is retrieved elsewhere, partly by a bulk uptake of phospholipids into intracellular cisternae when the plasma membrane is expanding fast, and partly by formation of coated vesicles which continue for some time after stimulation to retrieve certain specific vesicle membrane constituents and possibly also some soluble proteins discharged into the cleft.

The retrieved membrane phospholipids and proteins mix inside with each other and with new vesicle constituents supplied by axoplasmic flow to produce a population of variegated membrane compartments. Somehow, these cisternae sort out the important vesicle constituants, begin to accumulate acetylcholine, and partition back to vesicle size again, thus forming a new generation of synaptic vesicles which can enter the cycle once again.

From: Heuser, J.E. 1976. "Morphology of synaptic vesicle discharge and reformation at the frog neuromuscular junction". Chapter 3, pp. 51-115, in: The Motor Innervation of Muscle, ed. S. Thesleff, Academic Press, London.


Summary of our current view of synaptic vesicle recycling at the frog neuromuscular junction, based on the new quick-freezing data reviewed in this report.

After synaptic vesicles undergo exocytosis and collapse into the presynaptic membrane, two different sorts of compensatory endocytosis can occur. Normally, during moderate rates of synaptic activity, the more leisurely pathway on the right is operative: Coated vesicles selectively retrieve specific vesicle components from the presynaptic membrane and directly produce new synaptic vesicles.

On the other hand, when secretion is driven to unusually high levels experimentally, the dotted pathway on the left also comes into play. (Whether these conditions and this pathway ever occur in vivo is presently not known.) Large vacuoles composed of random bits of membrane pinch off rapidly from the presynaptic membrane and gather inside the nerve terminal.

How these vacuoles or "cisternae" give rise to new synaptic vesicles is also not known. They may simply divide down to vesicle size again, or else coated vesicles may pinch off from them, exactly as if they were still part of the presynaptic membrane.

From: Heuser, J.E. 1978. "Synaptic vesicle exocytosis and recycling during transmitter discharge from the neuromuscular junction". Life Sci. Res. Report 11: 445-464. Transport of Macromolecules in Cellular Systems, ed. S.C. Silverstein. Berlin: Dahlem Konferenzen.

Takai and deCamilli's model of synaptic vesicle recycling, in which new synaptic vesicles are generated directly from the uncoating of clathrin-coated vesicles. It suggests that clathrin-coated vesicles bud either from deep invaginations of the surface plasma membrane or from internal vacuoles that derive from the pinching off of these invaginations, or both.

This view differs from our original idea that newly formed clathrin-coated vesicles initially fuse together to form "cisternae" (which now would be called "early sorting endosomes"), and that new synaptic vesicles are formed secondarily via noncoated budding and partitioning of these endosomes.

FROM: Takei, K., Mundigl, O., Daniell, L., and P. deCamilli (1996). The synaptic vesicle cycle: A single budding step involving clathrin and dynamin. J. Cell. Biol. 133: 1237-1250.

NOTE: The primary reason that Takai and deCamilli doubt our original model is the following: In the time that has elapsed since its formulation, other more recent membrane-transport studies have demonstrated that membrane coats are generally shed prior to fusion.If this were generally true, then the clathrin-coated buds we saw connected to endosomes ought to have been budding from them, not fusing with them as we originally suggested!

Indeed, we and others have shown in recent years that clathrin coated vesicles do indeed lose their coats very rapidly upon entering the cell, and on the other hand, that clathrin coated buds can be induced to form on endosomes in vitro as well as in vivo. Hence, the view that new synaptic vesicles are generated directly from uncoating of clathrin-coated vesicles does now seem more and more reasonable. It would be quite consistent with the similar size of synaptic vesicles and the majority of nerve terminal clathrin-coated vesicles.

Furthermore, it would explain why cisternae (or sorting endosomes) are not commonly seen in vivo, despite how rapidly many CNS synapses are likely be turning over their vesicles: Their mechanisms for converting clathrin-coated vesicles directly back in to synaptic vesicles are presumably just as rapid. According to this line of reasoning, cisternae would only be expected to form in nerve terminals when they experienced unusually strong stimulation and discharged their vesicles faster than they could reform clathrin coated vesicles.

In this case, the nerves would indeed be expected to develop deep invaginations of their plasma membrane, and these invaginations might well pinch off directly (by some form of "bulk endocytosis") to form intracellular cisternae. And once this had happened, it would also seem quite likely that the cisternae would be treated just like isolated bits of plasma membrane and that clathrin-coated vesicles would be budded from THEM, again forming new synaptic vesicles directly. Takai and deCamilli argue convincingly that this model would reduce the number of steps required for synaptic vesicle reformation and thereby help to explain the very rapid time course of synaptic vesicle recycling.