© 2003 Nature Publishing Group

In A, researchers used a fluorescent protein (synaptopHluorin) to visualize synaptic vesicle movement. Some vesicles stay open briefly before retrieval (kiss-and-run). Others stay open longer but also don't collapse fully into the plasma membrane (compensatory). Still others collapse and are not retrieved until another stimulus is delivered (stranded). In B, another group used a dye FM1-43, to study vesicle retrieval and found that single vesicles can undergo many rounds of fusion. Single electrical stimuli caused only partial loss of dye. Further release from the same vesicle could sometimes be evoked, but not until a dead time of around 23 seconds had passed. (From S.O. Rizzoli, W.J. Betz, Nature, 423:591–2, 2003.)

Communication between neurons – the stuff of our senses, emotions, and memories, as well as motor and visceral control – relies on chemical messengers. Packaged into tiny membrane-enclosed vesicles, neurotransmitters are delivered...


In the early 1970s, electron microscopy studies showed coated-pit endocytosis within nerve terminals at the neuromuscular junction. Recycling time was initially estimated at about a minute; recent data have more than halved that. Still, many consider this implausible. "If you are delivering nerve impulses at the rate of 10 per second," says Chuck Stevens, professor of molecular neurobiology of the Salk Institute for Biological Studies in San Diego, "and it takes 10 seconds to take one in, then you'd have a bunch of vesicles stuck in the membrane all the time." The kiss-and-run mode was proposed, also in the 1970's, to account for the speed required by neurons.

Using fluorescence technology to label vesicles, the game has moved to the small synapses of the central nervous system. Three research groups are currently studying the kinetics of vesicle turnover. One is led by Stevens, the others by Richard Tsien, professor of molecular and cellular physiology at Stanford University, and Tim Ryan, associate professor of biochemistry at Weill Medical College of Cornell University. They all use similar methods and preparations, yet they have three different stories to tell.

Stevens tagged vesicle associated membrane protein (VAMP) with a pH-sensitive form of green fluorescent protein (GFP). Because synaptic vesicles maintain an acidic lumen within, the fluorescence is quenched as long as the vesicle remains intact. During release, the vesicle fuses with the cell surface, exposing the so-called synaptopHluorin to a neutral pH and causing it to fluoresce.1 "What we found was that the increase in fluorescence was quantized so we could identify a single vesicle's fluorescence," says Stevens. "We could tell how many were there."

Subsequent decreases in fluorescence indicated reuptake of the tagged vesicle proteins, as new vesicles and protons are pumped in to reacidify the contents. The amount of synaptopHluorin taken in was equal to the amount put out. Further, Stevens says, "The same amount of fluorescence corresponded to one vesicle across synapses and across experiments." Synaptic vesicles vary quite a bit in size, so having a constant number of synaptopHluorins implies "some mechanism for counting proteins put into the vesicle," he says, and perhaps for keeping those proteins aggregated.

Basically the Stevens study identified different kinetic profiles of vesicle uptake. In the standard mode, a released vesicle would stay on the surface membrane for 10 to 15 seconds and then get taken back in. "If it wasn't taken in by 15 seconds, it was stuck," says Stevens. The fastest mode observed, which Stevens interpreted to be kiss-and-run, took from 1/2 to 3/4 of a second.

Tsien used a different fluorescent tag, the styryl dye FM1-43, to stain neuronal membranes. The nerves were stimulated five times to induce dye uptake into synaptic vesicles retrieved from the cell surface. After washing excess stain away, vesicles containing the dye were monitored through subsequent stimuli.2 By staining lightly, Tsien says, "We were able to track one vesicle through many seconds of its life cycle."

A single action potential could induce a full loss of the dye, which they interpreted as being consistent with complete insertion of the vesicle into the plasma membrane. More frequently, Tsien and his colleagues observed only a partial loss of dye, which they interpreted to be consistent with kiss-and-run. Additional stimuli promoted further dye loss in these same vesicles.

Tsien says in addition to providing a fast mode of recycling, kiss-and-run also gives neurons a clever way to regulate the amount of neurotransmitter that comes out. This may sound like heresy to textbook descriptions that neurotransmitters are released in quantal units corresponding to the amounts packaged in the vesicles. "That's certainly the classic way of thinking," says Tsien, but reports of subquantal release are in the literature. "They're considered oddities, even though they've been repeatedly observed," he says. So in addition to the normal mode of neurotransmission, with quantal release and vesicular collapse, Tsien proposes "a whispering mode, where the presynaptic terminal speaks in such a quiet voice that only those with particularly sensitive ears can hear it."


Ryan uses both synaptopHluorin and styryl dyes to monitor the balance between exocytosis and endocytosis. Accumulation of dye on the cell surface indicated vesicles amassing on the plasma membrane. By increasing the rate of repetitive stimulation, Ryan pushed the system until endocytosis was unable to keep up with exocytic release.3 "You can stimulate up to a certain frequency and show that membrane doesn't accumulate," he says. "As you go above that frequency, the machinery for endocytosis can't keep up anymore."

Ryan doesn't invoke a different mode for endocytosis to explain the changes in behavior under different stimulus conditions. "It's not that the synapse is slow, it just has limited capacity," he says, comparing it to a bathtub drain. "The amount of time it takes to drain the bathtub depends on how much water you have in total."

Clathrin's reputation for slowness has biased people's thinking, says Ryan, leading to "the strong hypothesis that there must be other modes of recycling." As soon as kiss-and-run was hypothesized, people started looking for it and finding evidence consistent with it. But given that data, Ryan says, "There's no reason to think it's more complicated than one mechanism that has a finite capacity."

Other researchers are inclined to label themselves according to the division, either on the kiss-and-run side with Stevens and Tsien or the not-enough-evidence side with Ryan.

Jane Sullivan, assistant professor of physiology and biophysics at the University of Washington in Seattle, says a majority of researchers in the field would agree that kiss-and-run can occur under certain conditions. But many question "whether it's occurring normally under physiological conditions," she says.

If multiple mechanisms for vesicle retrieval exist, one might expect neurons to capitalize on this flexibility. "One notion is that synapses that have a very high probability of release might use one mode of cycling, and synapses with a low probability of release would use another," says Sullivan. The Stevens study shows that synapses with a low probability of release are more likely to use kiss-and-run. That relationship seems "a little odd," says Sullivan, as one might expect the fast mode to operate with higher demand.

"Maybe it's a moot question," says Schweizer, agreeing with Ryan. "Maybe endocytosis through the classical clathrin-mediated mechanism is fast enough already."


Inspired by kinetics, the kiss-and-run model has come to include mechanistic predictions. The debate is no less ardent on topics such as efficient recovery of vesicle components and the existence of a fusion pore.

What happens to vesicle components on the plasma membrane remains one of the least understood aspects, says Bill Betz, professor and chair of physiology and biophysics at the University of Colorado in Denver. "Are they flowing uniformly together to some site a few hundred nanometers away where clathrin is waiting to gobble them up? Or are they dispersed?" Ryan has shown that at the synapse lots of vesicle protein on the surface just sits there at rest, and if that's the usual case, says Betz, it would suggest that some dispersion occurs. But Betz notes that Stevens, studying the same synapses, has evidence that they go back in together. Betz has no doubt that clathrin is involved. "The question is: Is it always involved?"

The counterargument is that kiss-and-run "allows that vesicle to be used more efficiently," says Tsien. "It doesn't have to go to the garage for a complete overhaul every time it needs to have a change of oil." Tsien's study indicated that a vesicle might undergo "three or four oil changes without a complete tune-up."

Sullivan points to "big arguments about what is opening up to allow neurotransmitter to come out," whether it be vesicle proteins forming a channel or perhaps vesicle lipids arranging a lipidic fusion pore. It is a leap, says Ryan, to hypothesize a fusion pore based on evidence of different release amounts, whether of dye or transmitter. Others question the physical feasibility of such a construct.

Schweizer characterizes opening times of up to one second, called kiss-and-run in the Stevens study, as a "very, very, very long hovering time," and questions the physical feasibility of that. "There are massive forces, osmotic gradients and so on, that might all be kept in check as long as the vesicle is a vesicle," he says. An enormously strong mechanism would be necessary to maintain a semifused state.

"Kiss-and-run implies a certain amount of reversibility," says Reinhard Jahn, professor and director of neurobiology at the Max Planck Institute for Biophysical Chemistry in Göttingen, Germany. "You reach a certain stage, but you can revert from that stage. As far as biochemists are concerned, it looks like that forward reaction is irreversible." Jahn posits that reversal, if it does occur, might be accomplished by coordinating the fusion and fission pathways, such that "immediately as the fusion happens, you have another reverse reaction being triggered somehow."


Scientists are right to focus on more than timing, says Schweizer. But with research branching off in diverse directions, the core controversy – whether kiss-and-run occurs, at least at central synapses – is muddled. "What is it we're talking about? Is it a different endocytic mechanism? Is it a different fusion mechanism? Is it a functional definition for emptying less than the full contents of the vesicle? I think different people have different definitions of it and that makes it hard," says Schweizer.

For a field in flux, the debate is surprisingly civil. The three senior investigators, Stevens, Tsien, and Ryan, speak respectively of each other's work. They do not question methods used or data gathered; rather, the disagreement comes on the interpretation.

Tsien says that Ryan is studying kiss-and-run in his experiments; he just doesn't want to call it that. Comparing Ryan to an attorney on the opposite side of the courtroom, Tsien says, "I can see he's performing certain legalistic maneuvers that he's certainly entitled to perform. Are they aimed at getting the truth? No, they're aimed at getting the best possible proof in front of the jury."

For his part, Ryan says, "I view the challenge as trying to pinpoint the molecular mechanisms. I know from history that the pure biophysical signatures can mislead us."

Jill U. Adams juadams@the-scientist.com

Interested in reading more?

Magaizne Cover

Become a Member of

Receive full access to digital editions of The Scientist, as well as TS Digest, feature stories, more than 35 years of archives, and much more!
Already a member?