The Chemical Synapse: Structure and Overview
Picture a single Purkinje cell in the cerebellum. It receives inputs from roughly 175,000 parallel fibers. Each contact lasts less than a millisecond. Yet collectively they shape a single motor decision. How does one cell listen to so many others without their signals simply blurring together? The answer lies in a specialized junction barely 20 nanometers across — the chemical synapse. Here, an electrical signal in one neuron is translated into a chemical message. That message is ferried across a gap. It is then reconverted into an electrical or biochemical signal in the next cell. This translation step is not a minor technical detail. It is the site where the nervous system does almost all of its computation, its plasticity, and, when it fails, much of its pathology. The chemical synapse matters because it is the fundamental unit of neural communication and neural change. Action potentials, covered in the study of Saltatory Conduction, are stereotyped, all-or-none events. They carry information only in their timing and frequency. The synapse is where that timing gets converted into something rich and modifiable. It becomes a graded, chemically encoded signal. Its strength depends on how much transmitter is released, how many receptors are present, and how the postsynaptic cell is built to respond. Learning and memory trace back to the molecular architecture described here. So do the actions of nearly every psychiatric and neurological drug. So does the wiring logic of the entire brain. This article builds on Neuron Structure and Morphological Types. There, the neuron was described as a polarized cell with dendrites, soma, axon, and terminal. Here we zoom into that terminal and its target. We treat the synapse as a three-part apparatus — presynaptic terminal, synaptic cleft, and postsynaptic density. We also contrast it with the older, simpler solution to intercellular communication: the electrical synapse. Understanding this architecture sets up the mechanistic detail of vesicle fusion and receptor gating that follows in later articles. By the end, a student should be able to draw a synapse from memory. They should label its parts with the correct molecular residents. They should explain why chemical transmission is slower but far more flexible than electrical transmission. They should also recount the classic experiments — from Katz's quantal analysis to Furshpan and Potter's discovery of electrical synapses. These experiments established this dual architecture as fact rather than inference.
- A chemical synapse has three compartments: presynaptic terminal, synaptic cleft (~20 nm), and postsynaptic density.
- Presynaptic terminals contain synaptic vesicles, active zones, and voltage-gated Ca²⁺ channels that trigger transmitter release.
- Neurotransmitter diffuses across the cleft and binds receptors embedded in the postsynaptic density, a protein-dense scaffold beneath the membrane.
- Electrical synapses (gap junctions) pass current directly and instantaneously; chemical synapses impose a synaptic delay (~0.3–5 ms) but allow amplification, sign inversion, and plasticity.
- Katz's quantal analysis at the neuromuscular junction proved that transmitter is released in discrete packets (quanta), the physical correlate of synaptic vesicles.
- Furshpan and Potter's crayfish giant synapse experiments proved rectifying electrical transmission exists, establishing electrical synapses as a distinct, real category alongside chemical ones.
The chemical synapse converts an electrical signal into a diffusible chemical one and back again, trading speed for enormous gains in amplification, flexibility, and plasticity.
The core idea
A chemical synapse is a specialized asymmetric junction between two neurons — or between a neuron and a muscle or gland cell — at which the presynaptic cell releases a chemical neurotransmitter that diffuses across an extracellular gap and binds receptors on the postsynaptic cell. The word "synapse," coined by Charles Sherrington in 1897 from Greek roots meaning "to clasp together," originally described a purely physiological gap inferred from the fact that reflex responses were slower and more one-directional than a continuous nerve fiber would predict. Only decades later did electron microscopy reveal the actual anatomical structure: a presynaptic bouton packed with small membrane-bound vesicles, a narrow extracellular cleft, and a postsynaptic membrane thickened by a dense protein meshwork. Three structural compartments define every chemical synapse. The presynaptic terminal (or bouton) is a swelling at the axon's end specialized for regulated secretion — it contains synaptic vesicles clustered near an active zone, a patch of presynaptic membrane facing the cleft where vesicle fusion is spatially restricted and tightly coupled to calcium influx. The synaptic cleft is a roughly 20-nanometer extracellular space, far too narrow to be a passive gap; it is filled with adhesion molecules and extracellular matrix proteins that physically align the release machinery with the receptor field. The postsynaptic density (PSD), most extensively studied at excitatory glutamatergic synapses, is an electron-dense thickening of the postsynaptic membrane packed with neurotransmitter receptors, scaffolding proteins, and signaling enzymes that anchor receptors precisely opposite the presynaptic active zone. Chemical synapses are conventionally contrasted with electrical synapses, in which adjacent cells are connected by gap junction channels — hexameric assemblies of connexin protein (in vertebrates) that directly bridge the two cytoplasms. At an electrical synapse, ionic current from one cell flows directly into the other with no chemical intermediate, no cleft to cross, and (in most cases) negligible delay. Chemical synapses trade this speed for something electrical synapses cannot offer: amplification of a small presynaptic signal into a large postsynaptic one, the ability to be either excitatory or inhibitory regardless of the presynaptic action potential's polarity, and a substrate for long-lasting plasticity — themes elaborated later in the study of Neurotransmitter Release and the Vesicle Cycle and Neurotransmitter Receptors: Ionotropic and Metabotropic.
Mechanism
Step 1 — Depolarization arrives at the terminal. An action potential propagates down the axon and invades the presynaptic terminal, depolarizing its membrane. Unlike the axon shaft, whose depolarization is dominated by voltage-gated Na⁺ and K⁺ channels, the terminal membrane is dense with voltage-gated Ca²⁺ channels (primarily P/Q-type and N-type at central synapses) concentrated at the active zone.
Step 2 — Calcium influx triggers vesicle fusion. Because the extracellular Ca²⁺ concentration (~1–2 mM) vastly exceeds the resting intracellular concentration (~100 nM), depolarization opens a brief, steep inward Ca²⁺ gradient. Local Ca²⁺ concentration near an open channel can transiently reach 10–100 μM within nanometers — a microdomain sensed by synaptotagmin, a calcium-sensing protein on the vesicle membrane. Calcium-bound synaptotagmin triggers full fusion of docked, primed synaptic vesicles with the presynaptic membrane, a process mediated by the SNARE complex (syntaxin and SNAP-25 on the plasma membrane, synaptobrevin on the vesicle). This entire electrical-to-chemical conversion, from Ca²⁺ entry to transmitter appearing in the cleft, takes under a millisecond — but the total synaptic delay (including diffusion and receptor binding) is typically 0.3–5 ms, distinctly slower than the near-instantaneous transmission at electrical synapses.
Step 3 — Diffusion across the cleft. Fusion releases the vesicle's contents — typically a few thousand to tens of thousands of neurotransmitter molecules — into the 20 nm cleft. Diffusion over this distance is extremely fast (microseconds), so the rate-limiting steps are vesicle fusion and receptor binding kinetics, not diffusion itself.
Step 4 — Receptor activation in the postsynaptic density. Neurotransmitter binds receptors held in precise apposition to the release site by the PSD scaffold. Two broad receptor classes exist, to be treated fully elsewhere: ionotropic receptors, ligand-gated ion channels that open directly upon binding (fast, ~1–2 ms signaling), and metabotropic receptors, G-protein-coupled receptors that trigger second-messenger cascades (slower, tens of milliseconds to seconds, but often longer-lasting and amplifying). The PSD is not passive scaffolding — proteins like PSD-95 cluster glutamate receptors, link them to the actin cytoskeleton, and organize the signaling enzymes that implement synaptic plasticity.
Step 5 — Termination. Signaling ends through neurotransmitter reuptake by presynaptic transporters, enzymatic degradation (as with acetylcholinesterase at the neuromuscular junction), or simple diffusion away from the cleft — ensuring each release event produces a discrete, temporally bounded signal rather than a persistent one.
The evidence
Bernard Katz and quantal transmitter release (1950s, frog neuromuscular junction). Before the vesicle hypothesis was confirmed structurally, physiologists needed proof that transmitter release was packaged rather than continuous. Katz and colleagues, particularly Paul Fatt and José del Castillo, recorded intracellularly from frog skeletal muscle fibers at the neuromuscular junction. Even with no presynaptic stimulation, they observed small, spontaneous depolarizations of about 0.4–1 mV, which they termed miniature end-plate potentials (mEPPs). When a nerve impulse was evoked, the resulting end-plate potential was not smoothly graded but its amplitude, across many trials at low external Ca²⁺, jumped in discrete steps that were always integer multiples of the mEPP size. Statistical analysis showed the number of quanta released per stimulus followed a Poisson distribution, exactly as predicted if release consists of a random, independent number of fixed-size packets. This quantal hypothesis, for which Katz shared the 1970 Nobel Prize in Physiology or Medicine, proved that neurotransmitter is released in discrete units — later identified anatomically as the contents of single synaptic vesicles once electron microscopy (Palade, De Robertis, Bennett) revealed the vesicle-packed presynaptic terminal. This is the direct physiological ancestor of the vesicle cycle described in later articles.
Furshpan and Potter and the discovery of electrical synapses (1959, crayfish giant fiber system). Before this experiment, virtually all synaptic transmission was assumed to be chemical, and the small class of unusually fast reflexes were attributed to unusually efficient chemical synapses. Edwin Furshpan and David Potter recorded simultaneously from the presynaptic lateral giant axon and the postsynaptic giant motor axon at the crayfish tail-flip escape synapse. They found that depolarizing current injected presynaptically produced an almost instantaneous postsynaptic depolarization with no measurable synaptic delay — far too fast for a diffusion-and-receptor-binding chemical mechanism. Crucially, transmission was rectifying: current passed efficiently from the presynaptic to postsynaptic side but not the reverse, meaning this could not be a chemical synapse (which are typically unidirectional but for different biophysical reasons) nor a simple passive electrical short. Their result directly demonstrated a low-resistance, direct cytoplasmic pathway — the physiological signature of a gap junction, later resolved structurally as arrays of connexon channels by electron microscopy and freeze-fracture studies. This experiment established electrical synapses as a real, distinct category of neural communication, not merely a theoretical curiosity, and forced the field to accept that the nervous system uses both mechanisms in parallel. Together, these two lines of evidence — Katz's demonstration that chemical release is quantal and vesicular, and Furshpan-Potter's demonstration that some synapses instead pass current directly — established the two-category framework (chemical vs. electrical) that anchors this entire article.
Worked examples
Example 1 — Distinguishing synapse type from recording data. A physiologist records simultaneously from a presynaptic neuron and a postsynaptic neuron in an invertebrate ganglion. Injecting a depolarizing current pulse into the presynaptic cell produces a postsynaptic depolarization with the following features: onset delay of 0.15 ms, no failure at high stimulation frequency (up to 200 Hz), and a postsynaptic response that persists (attenuated) even when extracellular Ca²⁺ is replaced with Co²⁺ (which blocks voltage-gated Ca²⁺ channels). What type of synapse is this, and why? Reasoning: Chemical synapses require Ca²⁺ influx to trigger vesicle fusion; blocking Ca²⁺ channels with Co²⁺ should abolish transmission entirely. Since transmission persists under Co²⁺, the synapse cannot depend on Ca²⁺-triggered vesicle release. The near-zero delay (0.15 ms is consistent with axonal conduction time, not the added ~0.3–5 ms chemical synaptic delay) and the lack of high-frequency depression (chemical synapses often show vesicle depletion at very high frequencies) both point to direct electrotonic current flow through gap junctions. Conclusion: this is an electrical synapse. This exercise mirrors exactly the logic Furshpan and Potter used, substituting a Ca²⁺ channel blocker for their frequency and rectification tests.
Example 2 — Quantal content calculation. At a neuromuscular junction, the average miniature end-plate potential (mEPP) amplitude is 0.5 mV. A single presynaptic action potential evokes an end-plate potential (EPP) of 25 mV under conditions of reduced extracellular Ca²⁺ (to avoid saturation/nonlinear summation). Assuming linear summation of quanta, estimate the mean quantal content (m), the average number of vesicles released per action potential. Reasoning: Quantal content is simply the evoked response divided by the unitary (single-quantum) response: m = EPP amplitude / mEPP amplitude = 25 mV / 0.5 mV = 50 quanta released per impulse. If the probability of release per vesicle in the readily releasable pool is estimated independently (say, from variance-mean analysis) at p = 0.2, and quantal content follows m = N·p (N = number of release-ready vesicles), then N = m/p = 50/0.2 = 250 release-ready vesicles participate in this synapse's response to a single action potential. This is the same statistical logic — comparing evoked responses to the unitary spontaneous event — that Katz used to establish the quantal hypothesis, now used quantitatively to estimate release-site number and probability, concepts that become central in the study of the vesicle cycle.
Common misconceptions & traps
| Misconception | Correction |
|---|---|
| "The synaptic cleft is empty space." | It is filled with adhesion molecules and matrix proteins that physically align release sites with receptors; it is a structured, not empty, compartment. |
| "All synapses are chemical." | A substantial minority, especially in invertebrate escape circuits, cardiac tissue, and some vertebrate CNS regions (e.g., inferior olive, retina), are electrical, using gap junctions. |
| "Electrical synapses are simpler, more 'primitive' versions of chemical synapses." | They are a distinct, independently useful mechanism (used for speed and synchrony), not an evolutionary stepping-stone toward chemical synapses. |
| "One vesicle = one neurotransmitter molecule." | A single vesicle contains several thousand transmitter molecules; "quantum" refers to one vesicle's contents, not one molecule. |
| "Postsynaptic density is just structural scaffolding." | It also contains signaling enzymes (e.g., CaMKII) central to synaptic plasticity, not merely anchoring proteins. |
| "Synaptic delay is mostly diffusion time across the cleft." | Diffusion across 20 nm is on the microsecond scale; most of the delay comes from Ca²⁺ channel opening, vesicle fusion kinetics, and receptor binding. |
The scientist behind it
Bernard Katz spent his career at University College London reconstructing, with unusual rigor, exactly how a nerve terminal talks to a muscle fiber. Trained originally in Germany and Britain, he built on the work of his collaborators Paul Fatt and José del Castillo, applying careful statistical reasoning — treating transmitter release as a stochastic, countable process — to what had been a purely descriptive physiological phenomenon. His insistence on quantitative, falsifiable models of biological variability (Poisson statistics applied to a living synapse) exemplifies how a genuinely mechanistic answer to "how do neurons talk?" required abandoning intuitive, continuous notions of chemical release.
Origins · 1 minSir Bernard Katz and how this was discovered
history
- 1897 — Sherrington coins "synapse" to describe an inferred physiological gap between neurons.
- 1950s — Fatt, Katz, and del Castillo discover miniature end-plate potentials and formulate the quantal hypothesis at the frog neuromuscular junction.
- 1954 — Palade and colleagues use electron microscopy to visualize synaptic vesicles, giving quantal release a physical structure.
- 1959 — Furshpan and Potter demonstrate rectifying electrical transmission at the crayfish giant synapse, establishing electrical synapses as real and distinct.
- 1970 — Katz awarded the Nobel Prize for elucidating neurotransmitter storage, release, and inactivation mechanisms.
Where this matters today
Nearly every neuroactive drug in clinical use targets some component of the chemical synapse described here. Selective serotonin reuptake inhibitors block presynaptic reuptake transporters to prolong serotonin's action in the cleft; botulinum toxin cleaves SNARE proteins to prevent vesicle fusion entirely, exploited both as a lethal toxin and, in controlled micro-doses, as a treatment for dystonia and cosmetic muscle relaxation; myasthenia gravis is an autoimmune attack on postsynaptic acetylcholine receptors at the neuromuscular junction, directly targeting the PSD-equivalent structure of that synapse. Anesthetics, anticonvulsants, and most antipsychotics act by modulating receptors embedded in postsynaptic membranes or the presynaptic release machinery upstream of them. Electrical synapses have their own clinical and research relevance: connexin mutations cause inherited deafness and cardiac arrhythmias (since gap junctions also synchronize heart muscle contraction), and gap-junction blockade is studied as a way to interrupt pathological synchrony in epilepsy. In neuroscience research, optogenetic and pharmacological tools that manipulate specific synaptic components — blocking particular receptor subtypes, tagging PSD scaffolding proteins with fluorescent markers, or measuring quantal parameters at identified synapses — remain the standard experimental toolkit for probing learning-related synaptic changes such as long-term potentiation, connecting the basic anatomy taught here directly to the study of memory, addiction, and neurodegenerative disease.
Ask the tutor: How does the postsynaptic density change its receptor number during long-term potentiation, and how does that relate to the structural picture described here?
Why is there a delay at chemical synapses but not electrical ones?
Chemical transmission requires several sequential biochemical steps: Ca²⁺ channel opening, calcium-triggered vesicle fusion, transmitter diffusion across the cleft, and receptor binding. Each step takes real time, together producing a synaptic delay of roughly 0.3 to 5 milliseconds. Electrical synapses bypass all of this: ionic current flows directly through gap junction channels connecting the two cytoplasms, with no chemical intermediary and no diffusion step, so the postsynaptic response begins essentially simultaneously with the presynaptic one. The trade-off is that chemical synapses gain enormous flexibility — amplification, sign inversion, and plasticity — that the faster electrical synapse cannot offer.
Are electrical synapses common in the human brain, or are they mostly found in invertebrates?
Electrical synapses are less numerous than chemical synapses in the mammalian brain but are far from rare or purely invertebrate phenomena. They occur between interneurons in the neocortex and hippocampus, coupling inhibitory cells to synchronize their firing; in the retina, where they link photoreceptors and other cell types to pool signals; and in the inferior olive, where they help generate synchronized oscillations important for motor timing. Cardiac muscle also relies heavily on gap junctions to synchronize contraction. Invertebrates famously use electrical synapses in fast escape circuits, but the mechanism is a general vertebrate feature too, not an evolutionary relic.
What exactly is the postsynaptic density, and why does it matter for plasticity?
The postsynaptic density (PSD) is an electron-dense protein meshwork just beneath the postsynaptic membrane, most prominent at excitatory glutamatergic synapses. It contains neurotransmitter receptors, scaffolding proteins (such as PSD-95) that hold receptors in place opposite the presynaptic active zone, and signaling enzymes such as CaMKII. Because the PSD can dynamically add or remove receptors and reorganize its scaffold in response to activity, it is the physical substrate for long-term potentiation and depression — the synaptic weight changes thought to underlie learning and memory, a topic developed further once receptor mechanisms are introduced.
How did scientists know synaptic vesicles existed before electron microscopy could see them?
Physiological evidence preceded direct visualization. Fatt, Katz, and del Castillo recorded spontaneous miniature end-plate potentials and showed that evoked responses came in integer multiples of this unitary size, following Poisson statistics — strong indirect evidence that release was packaged into discrete quanta rather than continuous. This quantal hypothesis predicted a physical, countable unit of release. When electron microscopy became available in the 1950s, researchers such as George Palade observed exactly the predicted structure: small, uniform membrane-bound vesicles clustered at the presynaptic terminal, confirming the physiological inference with anatomical evidence.
Can a single presynaptic terminal release more than one type of neurotransmitter?
Yes. While classical teaching described one neuron as releasing a single transmitter ("Dale's principle"), it is now well established that many neurons co-release a fast-acting classical transmitter (such as glutamate or GABA) alongside a neuropeptide or a second small-molecule transmitter, often from different vesicle populations within the same terminal. This co-release adds another layer of signaling flexibility beyond the basic architecture described here, since different transmitters can act on distinct receptor populations with different kinetics.
Why does the postsynaptic density matter more at chemical synapses than electrical ones have an equivalent structure?
At a chemical synapse, precise molecular alignment is essential: transmitter released at the active zone must reach specific receptors within microseconds before diffusing away or being cleared, so the postsynaptic cell needs a dense, organized receptor field held in exact register with the presynaptic release site. Electrical synapses do not need this because current simply flows through gap junction channels that directly and continuously couple the two cells; there is no diffusing ligand to "aim" receptors at, so no equivalent scaffolding structure is required on either side of the gap junction plaque.
Is the synaptic cleft the same width at every chemical synapse?
Roughly, yes — around 20 nanometers is a widely cited approximate figure for typical central and neuromuscular chemical synapses, though exact measurements vary somewhat by synapse type and fixation method used in electron microscopy. This narrow, consistent spacing is not accidental: it is maintained by transsynaptic adhesion molecules (such as neurexins and neuroligins) that span the cleft and mechanically couple the presynaptic active zone to the postsynaptic density, ensuring transmitter released from a vesicle reaches its target receptors reliably and rapidly.