Neuroscience
Long-Term Potentiation
Synapses that strengthen to store a memory
Long-term potentiation (LTP) is the lasting strengthening of a synapse that follows brief, correlated pre- and postsynaptic activity — the most widely accepted cellular mechanism for learning and memory. In the hippocampus, glutamate released onto a dendritic spine opens AMPA receptors, but the decisive step is the NMDA receptor: only when the postsynaptic membrane is already depolarized does its magnesium block pop out, letting calcium pour in. That calcium pulse activates CaMKII, which drives more AMPA receptors into the membrane and enlarges the spine, so the same input now generates a bigger response — a change that can last from hours to weeks.
- Coincidence detectorNMDA receptor (Mg²⁺ block)
- Trigger ionCa²⁺ (~50 nM rest → µM transient)
- Classic induction100 Hz tetanus, 1 s
- Memory switchAutophosphorylated CaMKII
- Early vs late LTP~1–3 h vs hours–days (needs new protein)
- Main siteHippocampal CA1 dendritic spines
Interactive visualization
Press play, or step through manually. The visualization is yours to drive — try it before reading on.
Watch the 60-second explainer
A condensed visual walkthrough — narrated, captioned, under a minute.
What long-term potentiation actually is
Long-term potentiation is a property of a single synapse: after the connection is driven in a particular pattern, it transmits more strongly than it did before, and the increase outlasts the stimulus by hours or longer. Experimentally, you record the postsynaptic response to a fixed test pulse, deliver a brief burst of strong activity, and then watch the same test pulse produce a larger response — often a doubling of the field excitatory postsynaptic potential (fEPSP) slope — that fails to decay over the recording session. That stable, input-specific gain is LTP.
The concept matters because it gives memory a physical address. Donald Hebb proposed in 1949 that "cells that fire together wire together" — that correlated activity should strengthen connections. In 1973, Bliss and Lømo showed exactly this in the rabbit hippocampus: a high-frequency tetanus to the perforant path produced a potentiation lasting hours. LTP is the molecular embodiment of Hebbian plasticity, and it is studied most heavily in the hippocampus because that structure is essential for forming new episodic and spatial memories.
The mechanism, step by step
Picture a single excitatory synapse in hippocampal area CA1. The presynaptic terminal releases the neurotransmitter glutamate; the postsynaptic side is a tiny dendritic spine studded with two kinds of glutamate-gated channel.
- AMPA receptors are the everyday workhorses. Glutamate opens them, sodium flows in, and the spine depolarizes within a millisecond. They carry baseline synaptic transmission.
- NMDA receptors are the gatekeepers of plasticity. They are permeable to calcium, but at the resting potential of about −65 to −70 mV a magnesium ion (Mg²⁺) sits in the pore and blocks ion flow even when glutamate is bound.
To unblock the NMDA receptor, the spine membrane must depolarize to roughly −40 mV or more. That happens when the presynaptic terminal fires strongly and repeatedly (a tetanus) or when the postsynaptic neuron is independently depolarized at the same moment glutamate is present. Depolarization pushes the positively charged Mg²⁺ out of the pore, and now — and only now — calcium streams through the open NMDA channel into the spine.
This is why the NMDA receptor is the brain's coincidence detector: it conducts calcium only when presynaptic glutamate release and postsynaptic depolarization overlap in time. It is a molecular AND-gate, and it is precisely the requirement Hebb predicted.
The calcium signal is the message. Resting spine calcium sits near 50–100 nanomolar; a successful induction drives it to micromolar levels for a few hundred milliseconds. A large, sharp rise switches on calcium/calmodulin-dependent protein kinase II (CaMKII), an enzyme so abundant in the postsynaptic density that it makes up a substantial fraction of spine protein. CaMKII autophosphorylates at threonine-286, which locks it into an active state that persists after calcium has returned to baseline — a self-sustaining molecular switch that converts a transient calcium pulse into a durable change.
Active CaMKII then does two things that strengthen the synapse:
- It phosphorylates existing AMPA receptors (at the GluA1 subunit), increasing their single-channel conductance.
- It drives trafficking of additional AMPA receptors from intracellular pools and the spine membrane periphery into the synapse, so more receptors face the incoming glutamate.
The spine often physically enlarges as its actin cytoskeleton reorganizes, providing more surface for receptors. The net result: the same quantity of released glutamate now opens more current. The synapse has been potentiated.
Early versus late LTP: the consolidation step
LTP comes in phases that map remarkably well onto short- versus long-term memory. Early LTP lasts roughly one to three hours and uses only post-translational machinery — phosphorylation and receptor trafficking of proteins that already exist. It does not require new gene expression. Late LTP persists for many hours to days and depends on new protein synthesis: calcium and cAMP activate protein kinase A (PKA), which signals to the nucleus and activates the transcription factor CREB, switching on plasticity-related genes whose products stabilize the enlarged synapse.
The experimental signature is clean and clinically resonant. Apply the protein-synthesis inhibitor anisomycin during induction and early LTP appears normally but decays after a couple of hours — late LTP never forms. This is the cellular echo of memory consolidation: disrupting protein synthesis after learning blocks long-term but not short-term memory.
| Feature | Long-term potentiation (LTP) | Long-term depression (LTD) |
|---|---|---|
| Net effect | Synapse strengthened | Synapse weakened |
| Typical induction | High-frequency tetanus (~100 Hz) or strong coincident activity | Prolonged low-frequency stimulation (~1 Hz, ~15 min) |
| Calcium signal | Large, fast rise to micromolar | Modest, slow, sustained rise |
| Dominant enzyme | CaMKII (kinase) — adds phosphate | Calcineurin / PP1 (phosphatases) — remove phosphate |
| AMPA receptors | Inserted into the synapse | Internalized / removed |
| Spine morphology | Enlarges | Shrinks or retracts |
LTP and LTD are not separate systems but two outputs of the same NMDA-receptor calcium signal read out by its amplitude and timing — a continuum of bidirectional plasticity that lets circuits both write and erase.
Clinical correlations
Because LTP sits at the heart of memory, its failure or hijacking shows up across neurology, psychiatry, and pharmacology.
- Alzheimer's disease. Soluble amyloid-beta oligomers suppress LTP and enhance LTD before plaques or neuron loss are evident, partly by stripping synaptic AMPA and NMDA receptors and by over-driving extrasynaptic NMDA receptors. Synaptic dysfunction correlates with cognitive decline better than plaque burden does. The drug memantine is a low-affinity NMDA antagonist that preferentially blocks pathological extrasynaptic signaling while sparing the synaptic activity normal plasticity needs.
- Ketamine and depression. As an NMDA channel blocker, ketamine acutely impairs the calcium signal for LTP and causes amnesia at anesthetic doses. Yet at sub-anesthetic doses it triggers a glutamate surge and downstream AMPA-receptor-dependent synaptic strengthening and spine growth that underlies its rapid antidepressant action — the same target blocking plasticity acutely and promoting it over hours.
- Anti-NMDA-receptor encephalitis. Autoantibodies against the NMDA receptor (often paraneoplastic) cause receptor internalization, producing severe memory loss, psychosis, seizures, and movement disorder — a striking demonstration of what happens when the coincidence detector is removed.
- Stress and PTSD. Glucocorticoids and chronic stress reshape hippocampal and amygdala plasticity; pathologically strong LTP in fear circuits is one model for the intrusive, over-consolidated memories of post-traumatic stress disorder.
- Pain and addiction. LTP-like strengthening in spinal dorsal horn synapses contributes to chronic pain sensitization, and drug-evoked plasticity in reward circuits reinforces addiction — both targets of considerable therapeutic interest.
This article is educational and is not medical advice. For diagnosis or treatment of any condition, consult a qualified clinician.
Common misconceptions
- "LTP is a memory." LTP is a synaptic change that almost certainly contributes to memory storage, but a single potentiated synapse is not a memory — memories are distributed patterns across many synapses.
- "AMPA receptors do the detecting." AMPA receptors carry routine transmission; the NMDA receptor is the coincidence detector that decides whether to change the synapse.
- "More calcium always means stronger synapse." A large fast calcium rise strengthens (LTP); a modest slow rise weakens (LTD). The same ion produces opposite effects depending on amplitude and timing.
- "LTP lasts forever." Early LTP decays within hours unless it is consolidated by new protein synthesis into late LTP, mirroring how unconsolidated memories fade.
- "It only happens in the hippocampus." The hippocampus is the classic model, but NMDA-dependent LTP occurs in cortex, amygdala, striatum, and spinal cord.
Frequently asked questions
What is long-term potentiation in simple terms?
Long-term potentiation (LTP) is a long-lasting increase in the strength of a synapse after the connection is used in a specific way — typically brief high-frequency stimulation or near-simultaneous firing of the sending and receiving neurons. After LTP, the same incoming signal produces a larger postsynaptic response, and that change can persist for hours, days, or longer. Because experience that drives correlated firing strengthens connections, LTP is the most widely accepted cellular mechanism for learning and memory, especially in the hippocampus.
Why is the NMDA receptor called a coincidence detector?
The NMDA receptor opens only when two things happen at once: glutamate must bind it (a presynaptic signal), and the postsynaptic membrane must already be depolarized (a postsynaptic signal). At resting potential a magnesium ion sits inside the channel pore and blocks it; depolarization to roughly −40 mV electrostatically expels that Mg²⁺. So the channel passes calcium only when presynaptic release coincides with postsynaptic activity. This molecular AND-gate is the physical basis of Hebb's rule — cells that fire together, wire together.
How does calcium turn a brief signal into a lasting change?
Calcium entering through NMDA receptors raises spine calcium from roughly 50–100 nM at rest to micromolar levels for a fraction of a second. A large, fast rise activates CaMKII, which autophosphorylates and stays active after calcium falls — a molecular memory switch. Active CaMKII phosphorylates AMPA receptors and the proteins that anchor them, driving more AMPA receptors into the synapse. A smaller, slower calcium rise instead favors phosphatases and weakens the synapse (long-term depression), so the same ion can strengthen or weaken a connection depending on its amplitude and timing.
What is the difference between early and late LTP?
Early LTP lasts roughly 1–3 hours and depends only on modifying existing proteins — phosphorylating AMPA receptors and inserting them into the membrane. It does not need new gene expression and is blocked by neither transcription nor translation inhibitors. Late LTP lasts many hours to days or longer and requires new protein synthesis driven by cAMP, PKA, and the transcription factor CREB. Blocking protein synthesis with anisomycin during induction leaves early LTP intact but prevents late LTP — mirroring how disrupting consolidation can erase long-term but not short-term memory.
How is LTP linked to Alzheimer's disease?
Soluble amyloid-beta oligomers impair LTP and enhance long-term depression even before plaques or neuron death appear, partly by promoting removal of synaptic AMPA and NMDA receptors and by causing excess extrasynaptic NMDA receptor activation. This synaptic failure tracks cognitive decline more closely than plaque count does. The NMDA receptor antagonist memantine, used in moderate-to-severe Alzheimer's, is thought to dampen pathological extrasynaptic NMDA signaling while sparing the synaptic activity that supports normal plasticity.
Why does the NMDA antagonist ketamine affect memory and mood?
Ketamine blocks the NMDA receptor channel, which acutely impairs the calcium signal needed for LTP and disrupts the formation of new explicit memories — part of why dissociative anesthesia produces amnesia. Paradoxically, at sub-anesthetic doses ketamine triggers a burst of glutamate and downstream synaptic strengthening (including AMPA receptor insertion and new spine growth) that underlies its rapid antidepressant effect. The same target can therefore block plasticity acutely and promote it over the following hours, showing how tightly LTP machinery is tied to both memory and mood.