Synaptic Vesicle Release: How a Neuron Fires a Chemical Message

From spike to trigger: the action potential arrives

The story begins when an action potential, a wave of membrane depolarization propagating down the axon at up to ~100 m/s in myelinated fibers, reaches the axon terminal. Here the plasma membrane balloons into a bouton studded with specialized secretory zones called active zones, each facing a matching patch of receptors on the target cell. The membrane's resting voltage of about −65 mV swings positive during the spike, and that voltage change is the only cue the terminal needs.

Clustered in the active zone are voltage-gated calcium channels (chiefly the P/Q-type Ca_V2.1 and N-type Ca_V2.2 in the mammalian brain). Depolarization flips their gates open in a few hundred microseconds. Because calcium is roughly 10,000-fold more concentrated outside the cell (~1–2 mM) than in the resting cytosol (~50–100 nM), it rushes inward down a steep electrochemical gradient the instant the channels open. This calcium influx is the master switch: block these channels, and the electrical signal arrives but no message is sent.

The calcium nanodomain: a microsecond, hyper-local spike

Calcium is a terrible long-range messenger because cytosolic buffers and pumps mop it up within nanometers. Evolution turned this weakness into a feature. Each open channel produces a calcium nanodomain, a plume in which the local concentration leaps from ~100 nM to tens of micromolar (~10–25 µM) within microseconds, but only within ~10–100 nm of the channel mouth. The release machinery is tethered right inside that plume.

The calcium sensor is synaptotagmin-1, a vesicle protein whose two C2 domains together bind roughly five Ca²⁺ ions. Its affinity is deliberately low (in the tens of micromolar), so it fires only inside the nanodomain and only during a spike, not from background calcium. This cooperative binding is why release rate scales as roughly the fourth to fifth power of calcium concentration, [Ca²⁺]^~4–5. A small rise in calcium produces an enormous, switch-like jump in release probability, giving the synapse both a sharp threshold and exquisite tunability.

The SNARE machine: zippering two membranes into one

Fusing two lipid bilayers is thermodynamically expensive. Bare membranes repel each other, and forcing them through the strained hemifusion intermediate to a full fusion pore costs on the order of 40–50 k_BT of energy. The cell pays this bill with the SNARE proteins. Three of them, synaptobrevin/VAMP2 on the vesicle and syntaxin-1 plus SNAP-25 on the plasma membrane, interlace into an extraordinarily stable four-helix coiled-coil bundle.

SNAREs assemble like a zipper, from the membrane-distal end inward toward the two bilayers. As the bundle zippers, it drags vesicle and target membranes together, and the mechanical strain of the final turns is what tips the lipids into fusion. Each fully assembled complex liberates on the order of tens of k_BT (~35 k_BT by common estimates), and a small number of complexes acting cooperatively (~1–3, still debated) supply the energy to breach the barrier. Before firing, the assembly is held in a spring-loaded, half-zippered state: the protein complexin clamps it, preventing premature fusion. When calcium binds synaptotagmin, synaptotagmin displaces the clamp, plunges into the membrane, and releases the latch, letting zippering snap to completion. The result is a fusion pore that opens in tens of microseconds; the entire calcium-to-fusion step can take under 100 µs at fast synapses.

Supporting proteins set the stage beforehand. Munc18 and Munc13 template and prime the SNAREs, while Rab3 and RIM dock vesicles at the active zone within nanometers of the calcium channels, minimizing the diffusion distance and shaving microseconds off the delay.

Quantal release: neurotransmitter comes in countable packets

The single most important discovery about synapses is that they release neurotransmitter in fixed, discrete units. In the 1950s Bernard Katz and colleagues, recording at the frog neuromuscular junction, saw tiny spontaneous voltage blips of ~0.5–1 mV even with no incoming spike. These miniature end-plate potentials were all about the same small size, and del Castillo and Katz reasoned that each blip was the fusion of a single vesicle, one quantum. Under weak stimulation, evoked responses came in whole-number multiples of that unit — an evoked signal was simply many quanta released at once.

The numbers behind a quantum are startling. A synaptic vesicle is only ~40 nm across, yet it concentrates neurotransmitter to ~100–150 mM inside, packaging roughly 1,000–5,000 molecules per vesicle (about 5,000–10,000 acetylcholine molecules at the neuromuscular junction). A vesicular transporter, powered by a proton gradient built by a V-type ATPase, loads this cargo against a steep concentration gradient. Each active zone keeps only a small readily releasable pool of ~5–10 primed vesicles, so release is quantized not just chemically but by supply. The probability that any one vesicle fuses per spike ranges from ~0.1 to nearly 1, and that probability is itself a knob the brain adjusts during learning.

Across the cleft: diffusion, receptors, and clearance

Once the pore opens, neurotransmitter escapes into the synaptic cleft, a gap of only ~20–40 nm in the central nervous system. Diffusion over such a distance is astonishingly fast. Using t ≈ x²/2D with a diffusion coefficient of ~4×10⁻¹⁰ m²/s, a glutamate molecule crosses a 20 nm gap in roughly 0.5 microseconds. The transmitter binds ionotropic receptors (ligand-gated ion channels such as AMPA, NMDA, or nicotinic acetylcholine receptors), which open their own pores within a fraction of a millisecond, letting ions flow and generating a postsynaptic potential. Other transmitter binds metabotropic G-protein-coupled receptors for slower, modulatory effects.

The message must also be switched off, and quickly, or the synapse would smear together. Two clearance strategies dominate. Reuptake transporters (EAATs for glutamate, plus DAT, SERT, and NET for dopamine, serotonin, and norepinephrine) pump transmitter back into neurons or surrounding astrocytes. Enzymatic destruction is used for acetylcholine: acetylcholinesterase is one of the fastest enzymes known, with a catalytic efficiency approaching the diffusion limit (~10⁸ M⁻¹s⁻¹) and a turnover of ~10⁴ molecules per second, chopping acetylcholine into acetate and choline almost as fast as it arrives. The vesicle membrane, meanwhile, is retrieved by clathrin-mediated endocytosis (or rapid "kiss-and-run") and recycled, completing the synaptic vesicle cycle so the terminal can fire again seconds later.

The look-alike cousin: electrical synapses

Not every synapse is chemical. At electrical synapses, the two cells are bridged by gap junctions: paired channels called connexons (hexamers of connexin proteins) that align across a mere ~3.5 nm gap and connect the two cytoplasms directly. Ionic current, and even small molecules, flow straight through with essentially no delay and usually in both directions.

The tradeoff is stark. Electrical synapses are fast and reliable but rigid: they transmit roughly one-to-one, cannot amplify a weak signal, and offer little room for the plastic strengthening that underlies learning. Chemical synapses spend a precious half-millisecond of delay and considerable metabolic energy on their vesicle machinery precisely to buy computation: gain, sign inversion (excitation or inhibition), integration of many inputs, and long-lasting modification. This is why electrical synapses cluster where speed and synchrony matter most, cardiac muscle, the retina, and escape-reflex circuits, while the flexible, learning brain is overwhelmingly chemical.

How we know, and why it matters: toxins, disease, and open questions

Much of what we know comes from perturbing the machinery. Botulinum and tetanus toxins are proteases that cleave specific SNARE proteins; a few molecules of botulinum toxin can silence a neuromuscular junction by shredding SNAP-25 or synaptobrevin, which is exactly why "Botox" paralyzes muscle and why the toxin is a therapy for spasticity and dystonia. Black widow spider venom (α-latrotoxin) does the opposite, forcing massive, uncontrolled vesicle fusion. In myasthenia gravis, antibodies attack the postsynaptic acetylcholine receptor, and many antidepressants (SSRIs) work by blocking the reuptake transporters that clear the cleft. High-speed imaging, capacitance measurements that detect single-vesicle fusion as a step in membrane area, and optical reporters like pHluorins that light up when a vesicle's acidic interior meets the neutral outside have let researchers watch these events one vesicle at a time.

Open questions remain. Exactly how many SNARE complexes are needed per fusion event is still argued. The molecular structure of the earliest fusion pore, and whether it flickers open and shut (kiss-and-run) or commits to full collapse, is debated. And the fusion problem itself has become a template for biomimetics: engineers are designing synthetic SNARE-like peptides and DNA-based fusion machines to trigger membrane fusion on demand, for targeted drug delivery that mimics the cell's own millisecond, quantized, calcium-triggered secretion.

Chemical synapses vs. electrical (gap-junction) synapses: two ways neurons pass a message

Feature	Chemical synapse	Electrical synapse
Signal carrier	Neurotransmitter released by vesicle fusion	Ionic current through connexon (gap-junction) channels
Structural gap	~20–40 nm cleft, no cytoplasmic continuity	~3.5 nm, cytoplasms directly connected
Delay	~0.5–1 ms (Ca²⁺ channels + fusion + diffusion)	Virtually none (~0.1 ms or less)
Directionality	One-way (pre → post)	Usually bidirectional
Signal gain	High — one spike can be amplified or filtered	~1:1, no amplification
Plasticity	Highly modifiable (learning, memory)	Limited; favors fast synchronization
Examples	Neuromuscular junction, cortical synapses	Cardiac muscle, retina, escape reflexes