Synapse

A synapse is a specialized intercellular junction that mediates functional communication between a presynaptic neuron and a postsynaptic target cell—either another neuron, a skeletal or cardiac myocyte (at a neuromuscular junction), or a secretory gland cell. The human brain contains approximately 8.6 × 10¹⁰ neurons interconnected by an estimated 10¹⁴ to 5 × 10¹⁴ synapses, forming the anatomical and computational substrate for sensory processing, motor coordination, homeostatic regulation, and memory storage. Synapses are divided into two structurally and biophysically distinct classes: electrical synapses, which pass ionic current directly between cytoplasms via gap-junction channels, and chemical synapses, which convert a presynaptic action potential into the Ca²⁺-dependent exocytosis of a neurotransmitter across an extracellular cleft to activate postsynaptic receptors.

The concept of the synapse emerged from the late nineteenth-century controversy between the reticular theory of Camillo Golgi—who argued that the nervous system forms a continuous syncytial network—and the neuron doctrine championed by Santiago Ramón y Cajal, whose silver-chromate (Golgi stain) histological preparations demonstrated that nerve cells are discrete anatomical units separated by microscopic discontinuities. In 1897, the British neurophysiologist Charles Scott Sherrington coined the term synapse (from the Greek synapsis, meaning "conjunction" or "clasp," derived from syn-, "together," and haptein, "to fasten") in Michael Foster's Textbook of Physiology to explain the one-way conduction and temporal delay of spinal reflex arcs.

Direct ultrastructural proof of the neuron doctrine was achieved in 1954–1955 when George Palade, Sanford Palay, and Eduardo De Robertis used transmission electron microscopy to resolve the 20–40 nm extracellular synaptic cleft separating presynaptic and postsynaptic cell membranes, as well as the cluster of 40-nm membrane-bound synaptic vesicles within the presynaptic terminal. Simultaneously, Bernard Katz, Paul Fatt, and José del Castillo (1952–1956) established the quantal hypothesis of neurotransmitter release at the frog neuromuscular junction, while Edwin Furshpan and David Potter (1959) demonstrated the existence of fast rectifying electrical synapses in the crayfish giant motor synapse.

Ultrastructure and Biophysics of Electrical Synapses

At an electrical synapse, the plasma membranes of two adjacent neurons approach within 3.5 nm of one another—compared to the normal 20 nm intercellular space—at specialized plaques termed gap junctions. Each gap-junction channel spans both bilayers and consists of two docked hemichannels (connexons in vertebrates, formed by hexamers of four-pass transmembrane connexin proteins such as Cx36 in mammalian central interneurons; innexons formed by innexin proteins in invertebrates). The central aqueous pore has a diameter of ~1.2–1.5 nm, making it non-selectively permeable to inorganic ions (K⁺, Na⁺, Ca²⁺, Cl⁻) as well as small second messengers and metabolites up to ~1,000 Da, including ATP, cyclic AMP (cAMP), and inositol 1,4,5-trisphosphate (IP₃).

Because ionic current (I_j = g_j[V_pre − V_post]) flows passively down the transjunctional voltage gradient through low-resistance connexon pores, electrical transmission occurs with virtually zero synaptic delay (<0.1 ms) and is typically bidirectional (though certain heterotypic connexin pairings exhibit voltage-dependent rectification). Electrical synapses excel at synchronizing the subthreshold oscillations and spiking of coupled neuronal ensembles—such as GABAergic parvalbumin-positive basket cells in the mammalian neocortex and hippocampus, inferior olivary neurons, and neuroendocrine hypothalamic magnocellular neurons—and mediating rapid escape reflexes in teleost Mauthner cells and crustacean giant fibers.

Architecture of the Chemical Synapse

Chemical synapses constitute the vast majority of synapses in the mature vertebrate nervous system. A canonical central chemical synapse comprises three compartments aligned with nanometer precision by trans-synaptic cell-adhesion molecules (specifically presynaptic neurexins bound to postsynaptic neuroligins, leucine-rich repeat transmembrane proteins [LRRTMs], and cadherins):

1. Presynaptic Bouton (Axon Terminal): Contains mitochondria, hundreds of 35–45 nm clear-core synaptic vesicles loaded with classical neurotransmitters, and an electron-dense specialization of the plasma membrane called the active zone. The active zone cytomatrix—built from scaffolding proteins RIM (Rab3-interacting molecule), Munc13, Bassoon, Piccolo, and ELKS—tethers docked vesicles within 10–20 nm of voltage-gated calcium channels (Ca_V2.1 [P/Q-type] and Ca_V2.2 [N-type]).
2. Synaptic Cleft: A 20–30 nm wide (up to 50 nm at the vertebrate neuromuscular junction) extracellular space filled with an organized basal lamina or matricellular glycoprotein scaffold that restricts lateral diffusion and houses neurotransmitter-degrading enzymes (such as acetylcholinesterase) and astrocytic perisynaptic processes (forming the tripartite synapse).
3. Postsynaptic Specialization: Frequently located on a dendritic spine head (~0.1–1.0 μm³ volume), characterized by an electron-dense submembranous protein meshwork called the postsynaptic density (PSD). In excitatory glutamatergic synapses, the scaffold protein PSD-95 (containing PDZ, SH3, and GK domains), Shank, Homer, and CaMKII anchor AMPA and NMDA receptors directly opposite the presynaptic active zone, forming a trans-synaptic nanocolumn. In inhibitory GABAergic and glycinergic synapses, the scaffold protein gephyrin clusters GABA_A or glycine receptors on the dendritic shaft or soma.

Katz's Quantal Hypothesis and Calcium Cooperativity

While recording intracellularly from the frog sartorius neuromuscular junction in the early 1950s, Bernard Katz and Paul Fatt observed spontaneous, subthreshold depolarizations of ~0.4–0.5 mV occurring randomly at ~1 Hz in the absence of presynaptic stimulation. Termed miniature end-plate potentials (mEPPs), each event reflects the spontaneous exocytosis of a single synaptic vesicle containing a discrete multimolecular packet—or quantum—of approximately 5,000–10,000 acetylcholine molecules. When an action potential invades the motor nerve terminal, the evoked end-plate potential (EPP) (~40–50 mV) is an integral multiple of the unitary quantal amplitude q. Under binomial statistics, the mean number of quanta released per action potential (quantal content, m) is:

m = N · p

where N is the number of release-ready vesicles in the readily releasable pool (RRP) docked at active zones, and p is the average release probability of a docked vesicle. Consequently, the mean evoked postsynaptic response amplitude is V_EPP = N · p · q.

Presynaptic neurotransmitter release exhibits a fourth-power dependence on extracellular Ca²⁺ concentration ([Ca²⁺]_out), first quantified by Dodge and Rahamimoff (1967):

Release ∝ ([Ca²⁺]_in)⁴

When an action potential depolarizes the bouton, Ca_V2.1 and Ca_V2.2 ion channels open during the falling phase (when the inward electrical driving force V_m − E_Ca rapidly increases), creating transient local Ca²⁺ microdomains of 50–200 μM within 20 nm of the channel mouth for <0.5 ms.

SNARE Complex Zippering and Synaptotagmin-1

The molecular machinery that couples Ca²⁺ influx to membrane fusion within 100–200 microseconds was elucidated by James Rothman, Randy Schekman, and Thomas Südhof (2013 Nobel Prize in Physiology or Medicine). Fusion is driven by the assembly of the SNARE (Soluble N-ethylmaleimide-sensitive factor Attachment Protein Receptor) complex, a four-helix coiled-coil bundle formed by:

• Synaptobrevin-2 (VAMP2): A v-SNARE (R-SNARE) anchored in the synaptic vesicle membrane.
• Syntaxin-1: A t-SNARE (Qa-SNARE) anchored in the presynaptic plasma membrane.
• SNAP-25: A peripheral t-SNARE (Qbc-SNARE) attached to the plasma membrane via palmitoylated cysteine residues, contributing two α-helices to the four-helix bundle.

Prior to an action potential, Munc18-1 and Munc13 catalyze the partial N-terminal zippering of the trans-SNARE complex (vesicle priming), while the protein complexin clamps the partially zippered bundle to prevent premature fusion. The vesicular transmembrane protein synaptotagmin-1 acts as the primary fast Ca²⁺ sensor: its two cytoplasmic C₂ domains (C₂A and C₂B) cooperatively bind 4 to 5 Ca²⁺ ions, inducing insertion of hydrophobic loops into the phospholipid bilayer, displacing the complexin clamp, and triggering complete C-terminal SNARE zippering. The free energy released (~35–65 k_BT per SNARE complex) overcomes the electrostatic hydration barrier between the vesicle and plasma membranes, opening a fusion pore. Following exocytosis, the AAA+ ATPase NSF and α-SNAP hydrolyze ATP to disassemble cis-SNARE complexes, and clathrin-mediated or ultrafast endocytosis recycles the vesicle membrane.

EPSPs, IPSPs, and Dendritic Summation

Neurotransmitter molecules diffuse across the 20-nm synaptic cleft in <10 μs and bind postsynaptic ionotropic or metabotropic receptors. Opening of ligand-gated cation channels reverses the postsynaptic membrane potential toward a specific reversal potential (E_rev):

• Excitatory Postsynaptic Potentials (EPSPs): Glutamate binding to AMPA or NMDA receptors (or ACh binding to nicotinic receptors) opens channels permeable to both Na⁺ and K⁺, with E_rev ≈ 0 mV. Because 0 mV is positive to the action-potential threshold (−55 mV), inward Na⁺ current exceeds outward K⁺ current, depolarizing the membrane by 0.2–2.0 mV per unitary central synapse.
• Inhibitory Postsynaptic Potentials (IPSPs): GABA binding to GABA_A receptors or glycine binding to glycine receptors opens Cl⁻-selective pores with E_rev = E_Cl ≈ −70 to −80 mV (maintained in mature neurons by the K⁺/Cl⁻ cotransporter KCC2). Opening Cl⁻ channels either hyperpolarizes the cell or acts via shunting inhibition, clamping V_m near rest and reducing input resistance (R_in) so that concurrent excitatory currents produce smaller voltage deflections (Δ*V* = I_EPSP · R_in).

Because a single central EPSP (~0.5 mV) cannot depolarize a neuron from −70 mV to its −55 mV threshold, the axon initial segment integrates thousands of synaptic inputs via spatial summation (simultaneous activation of multiple synapses across the dendritic tree, attenuated according to the cable space constant λ) and temporal summation (high-frequency activation of the same synapse within the membrane time constant τ_m = R_mC_m ≈ 10–30 ms).

Hebbian Synaptic Plasticity: LTP and LTD

In 1949, Donald Hebb postulated that coincidental activity between presynaptic and postsynaptic neurons strengthens their synaptic connection. In 1973, Terje Lømo and Timothy Bliss discovered Long-Term Potentiation (LTP) at perforant-path synapses in the rabbit hippocampus. At canonical Schaffer collateral–CA1 pyramidal cell synapses, LTP is governed by the biophysical properties of the NMDA receptor functioning as a molecular coincidence detector:

1. At resting potentials (−70 mV), the pore of the NMDA receptor is occluded by an extracellular Mg²⁺ ion, even when presynaptic glutamate is bound.
2. High-frequency presynaptic stimulation (or paired pre- and postsynaptic spiking within a 10–20 ms spike-timing-dependent plasticity [STDP] window) strongly activates AMPA receptors, depolarizing the dendritic spine and electrostatically expelling Mg²⁺ from the NMDA channel pore.
3. Unblocked NMDA receptors conduct a large influx of Ca²⁺ (~10 μM) into the spine head, activating Ca²⁺/calmodulin-dependent protein kinase II (CaMKII). Autophosphorylated CaMKII (at Thr286) phosphorylates AMPA receptor GluA1 subunits (increasing single-channel conductance) and drives the exocytic insertion of additional AMPA receptors into the postsynaptic density, while actin polymerization enlarges the dendritic spine head.

Conversely, prolonged low-frequency stimulation (1 Hz for 10–15 minutes) produces a modest, sustained Ca²⁺ elevation (~1 μM) that selectively activates the higher-affinity protein phosphatase calcineurin (PP2B) and protein phosphatase 1 (PP1), dephosphorylating AMPA receptors and triggering their clathrin-mediated endocytosis to produce Long-Term Depression (LTD).

Why is there a 0.5 to 2.0 millisecond delay at chemical synapses?

Contrary to intuition, diffusion of neurotransmitter across the 20-nm synaptic cleft takes less than 10 microseconds. The bulk of the 0.5–2.0 ms synaptic delay at a chemical synapse occurs presynaptically: roughly 80% of the delay is required for the action potential depolarization to open voltage-gated Ca_V2.1/2.2 channels, allow Ca²⁺ influx during the spike repolarization phase, and drive Ca²⁺ binding to synaptotagmin-1 and SNARE-mediated membrane fusion.

How do botulinum and tetanus toxins paralyze synaptic transmission?

Botulinum neurotoxins (BoNT A–G) and tetanus neurotoxin (TeNT), produced by Clostridium bacteria, are zinc-dependent metalloproteases that enter presynaptic terminals and specifically cleave SNARE proteins. BoNT/A and BoNT/E cleave SNAP-25 at the neuromuscular junction, producing flaccid paralysis, whereas tetanus toxin undergoes retrograde axonal transport to spinal inhibitory interneurons and cleaves synaptobrevin-2 (VAMP2), blocking glycine/GABA release and causing spastic paralysis.

How does the NMDA receptor act as a coincidence detector during learning?

Activation of the NMDA glutamate receptor requires two simultaneous events: presynaptic glutamate release (which binds the ligand-binding domain) and strong postsynaptic depolarization (which electrostatically expels the voltage-dependent Mg²⁺ ion blocking the channel pore). Only when presynaptic activity coincides with postsynaptic depolarization does Ca²⁺ enter through the NMDA receptor to activate CaMKII and induce Long-Term Potentiation (LTP).

1. Katz, B., & Miledi, R. (1965). "The measurement of synaptic delay, and the time course of acetylcholine release at the neuromuscular junction." Proceedings of the Royal Society of London. Series B, Biological Sciences, 161(985), 483–495. https://doi.org/10.1098/rspb.1965.0016
2. Südhof, T. C. (2013). "Neurotransmitter release: the last millisecond in the life of a synaptic vesicle." Neuron, 80(3), 675–690. https://doi.org/10.1016/j.neuron.2013.10.022
3. Pereda, A. E. (2014). "Electrical synapses and their functional interactions with chemical synapses." Nature Reviews Neuroscience, 15(4), 250–263. https://doi.org/10.1038/nrn3708
4. Huganir, R. L., & Nicoll, R. A. (2013). "AMPARs and synaptic plasticity: the last 25 years." Neuron, 80(3), 704–717. https://doi.org/10.1016/j.neuron.2013.10.025