Concept

Long-term potentiation

T-073Home BU-302Threads regulation · systems

Statement

A cellular mechanism of learning and memory.

Why it matters

synaptic-transmission already describes how a signal crosses a synapse via neurotransmitter release and receptor binding, but that description alone treats synaptic strength as fixed. Long-term potentiation (LTP) is the best-understood cellular mechanism by which a synapse's strength can be persistently increased by its own pattern of use, giving concrete mechanistic content to the Hebbian idea that neurons which fire together strengthen the connection between them, and providing a plausible physical substrate for learning and memory at the level of a single synapse.

neural-integration's account of how a neuron sums many simultaneous inputs explains why coincident activity at nearby synapses, even from separate presynaptic neurons, can supply the depolarisation LTP's induction requires.

Hypotheses

LTP specifically requires coincident, correlated activity: strong, repeated presynaptic firing occurring together with sufficient postsynaptic depolarisation at approximately the same time.A synapse's strength change depends on the temporal relationship between pre- and postsynaptic activity, not simply on total activity at either side in isolation. The NMDA receptor acts as a coincidence detector: its channel requires both glutamate binding and sufficient postsynaptic depolarisation to expel a voltage-dependent magnesium ion blocking it, before current can flow through it.Without both conditions met simultaneously, the channel remains blocked even with glutamate bound. LTP is classically divided into an early phase, lasting roughly one to a few hours and not requiring new protein synthesis, resulting from modification of existing synaptic proteins, and a late phase, persisting far longer and requiring new gene transcription and protein synthesis; the mechanism described in the Proof is primarily the early-phase mechanism.

Proof

\[ \text{Glutamate binds both AMPA and NMDA receptors; AMPA activation alone produces ordinary depolarisation, but the NMDA channel remains Mg}^{2+}\text{-blocked at resting potential.} \]

The NMDA receptor's channel requires more than glutamate binding alone to open (Hypothesis 2). A

\[ \text{Sufficient postsynaptic depolarisation expels the Mg}^{2+}\text{ block, unblocking the channel.} \]

This satisfies both of the NMDA receptor's gating requirements simultaneously — glutamate bound, and the electrostatic block relieved. A

\[ \text{The unblocked NMDA channel is substantially permeable to Ca}^{2+}\text{ as well as Na}^+\text{, producing a local rise in postsynaptic calcium specifically at the activated synapse.} \]

Unlike the AMPA receptor, the NMDA channel's calcium permeability is what converts coincident electrical activity into a chemical (second-messenger) signal. A

\[ \text{Calcium influx activates calcium-dependent kinases (notably CaMKII), which phosphorylate existing AMPA receptors and promote insertion of additional AMPA receptors from an intracellular reserve pool.} \]

This is the direct molecular step converting a transient calcium signal into a lasting change in the postsynaptic membrane's receptor content. B

\[ \text{Subsequent presynaptic glutamate release at that synapse now produces a larger postsynaptic depolarisation than before.} \]

More, and more sensitive, AMPA receptors are now present at that specific synapse, while other, non-coincidently-active synapses on the same neuron are left largely unaffected. A

Result

\[ \text{coincident activity} \to \text{NMDA unblocked} \to \text{Ca}^{2+}\text{ influx} \to \text{more/added AMPA receptors} \to \text{strengthened synapse} \]

Reading. LTP converts a specific pattern of correlated neural activity into a lasting, synapse-specific increase in signal transmission strength.

Scope. The mechanism described is specific to NMDA-receptor-dependent LTP, the best-characterised form, particularly well studied in the hippocampus; other LTP-like mechanisms not requiring NMDA receptors exist at some synapses elsewhere in the brain.

Corollaries & converses

synaptic-transmission's basic account of neurotransmitter release and receptor binding is the substrate LTP modifies, not something LTP replaces: LTP works entirely by changing how many, and how sensitive, postsynaptic receptors are present, using the same release-and-binding machinery already established there.
neural-integration's account of how a neuron sums simultaneous inputs explains why coincident activity at nearby synapses can supply the depolarisation LTP requires (Hypothesis 1).
Converse: repeated low-frequency stimulation, insufficient to produce strong postsynaptic depolarisation, commonly produces the opposite, complementary change — long-term depression, a persistent weakening of the synapse — via smaller, more prolonged calcium signals activating phosphatases rather than kinases.

Fails without

Drop coincidence (Hypothesis 1): presynaptic activity alone, without sufficient simultaneous postsynaptic depolarisation, leaves the NMDA receptor's magnesium block in place regardless of how much glutamate is released, so no calcium influx and no potentiation occurs — synaptic transmission proceeds normally via AMPA receptors, but leaves no lasting trace of that activity.
Drop the NMDA receptor's calcium permeability specifically (Step3): if the coincidence-detecting channel conducted only \(\text{Na}^+\), as the AMPA receptor does, there would be no intracellular second messenger to trigger the kinase cascade of Step4, and coincident activity would produce only an ordinary, transient depolarisation with no mechanism to convert that coincidence into lasting change.

Common errors

Assuming LTP is a single, uniform mechanism throughout the brain; the NMDA-receptor-dependent form described here is the best studied, but NMDA-receptor-independent forms of long-lasting synaptic strengthening exist at other synapses.
Confusing the roles of the AMPA and NMDA receptors; AMPA carries most fast, everyday excitatory current and is not calcium-permeable, while NMDA's distinctive doubly-gated, calcium-permeable channel is what makes it a coincidence detector.
Treating LTP as occurring at every active synapse on a neuron simultaneously; the calcium signal and resulting strengthening are confined largely to the synapse where coincident activity actually occurred.
Assuming early-phase and late-phase LTP are the same process at different time points rather than mechanistically distinct phases requiring different downstream events (Hypotheses, t3).

Discussion

Terje Lømo first observed long-lasting synaptic strengthening following brief high-frequency stimulation in the rabbit hippocampus in 1966, work developed further with Timothy Bliss and published in 1973, establishing LTP as a defined experimental phenomenon. Donald Hebb had, in 1949, already proposed on purely theoretical grounds that persistent, repeated correlation between a presynaptic and a postsynaptic cell's firing should strengthen the connection between them — a predictive theoretical proposal that LTP later supplied with concrete molecular machinery, decades after Hebb's original, largely conceptual argument.

While LTP is the most extensively studied candidate cellular mechanism for memory formation, establishing that LTP actually occurs during natural learning in an intact, behaving animal, and that blocking it impairs memory formation specifically, required a further, separate line of behavioural and pharmacological evidence beyond LTP's basic cellular mechanism alone.

Common misconception: that LTP is itself a memory, or that a single LTP event at one synapse stores one specific memory. LTP is a synaptic mechanism, a way individual synaptic strengths can be persistently modified by activity; an actual memory is thought to be represented by the coordinated pattern of many such synaptic changes distributed across a much larger population of neurons and synapses.

Worked examples

\[ \text{Repeated, correlated activation of a hippocampal synapse (high-frequency presynaptic stimulation strongly depolarising the postsynaptic cell) unblocks NMDA receptors, producing a substantial calcium transient.} \]

This triggers AMPA receptor insertion as described in the Proof (Step2–4). A

\[ \text{A subsequent, single, ordinary-strength test stimulus at the identical synapse now evokes a measurably larger postsynaptic response than it did before, and this enhancement persists without further strong stimulation.} \]

An untouched, non-coincidently-active neighbouring synapse on the same postsynaptic cell shows no such change, confirming the synapse-specific character of the effect. A

\[ \text{brief correlated activity} \to \text{lasting, synapse-specific increase in transmission strength} \]

Reading. A short episode of coincident activity leaves a persistent physical trace at exactly the synapse where it occurred.

Scope. This synapse-specificity is a hallmark distinguishing LTP from any general, whole-cell change in excitability.

Further example. In a hippocampal CA1 pyramidal neuron, the resting membrane potential is \(V_m = -70 \text{ mV}\). The reversal potential for the NMDA receptor is \(E_{\text{NMDA}} = 0 \text{ mV}\). The maximum conductance of a single NMDA receptor channel is \(g_{\text{max}} = 10 \text{ pS}\). The magnesium (\(\text{Mg}^{2+}\)) block of the NMDA channel is voltage-dependent; at \(V_m = -70 \text{ mV}\), the block factor is \(B = 0.99\) (meaning 99% of the conductance is blocked). At \(V_m = -30 \text{ mV}\) (due to strong postsynaptic depolarisation), the block factor is \(B = 0.10\) (meaning 10% of the conductance is blocked). Calculate the ratio of the ionic current through the NMDA receptor at \(V_m = -30 \text{ mV}\) to the current at \(V_m = -70 \text{ mV}\).

\[ I = g_{\text{eff}} (V_m - E_{\text{NMDA}}) \]

The ionic current is the product of the effective conductance and the driving force (voltage difference). A

\[ g_{\text{eff}} = g_{\text{max}} (1 - B) \]

The effective conductance is the maximum conductance reduced by the fraction blocked by magnesium. A

\[ I_{-70} = 10 \text{ pS} \times (1 - 0.99) \times (-70 \text{ mV} - 0 \text{ mV}) = -7 \text{ fA} = -0.007 \text{ pA} \]

At rest, the driving force is 70 mV, but the conductance is only 1% of maximum due to the strong Mg\(^{2+}\) block. A

\[ I_{-30} = 10 \text{ pS} \times (1 - 0.10) \times (-30 \text{ mV} - 0 \text{ mV}) = -270 \text{ fA} = -0.27 \text{ pA} \]

Upon depolarisation, the driving force decreases to 30 mV, but the conductance increases to 90% of maximum as the Mg\(^{2+}\) block is relieved. A

\[ \text{Ratio} = \frac{|I_{-30}|}{|I_{-70}|} = \frac{0.27 \text{ pA}}{0.007 \text{ pA}} \approx 38.6 \]

The relief of the voltage-dependent block causes a much larger increase in current than the decrease in driving force causes a decrease. A

\[ \text{Current ratio} \approx 38.6 \]

Reading. This demonstrates the coincidence detection mechanism: significant NMDA receptor current only flows when the postsynaptic membrane is sufficiently depolarised to expel the magnesium ion.

Scope. This calculation assumes a single channel and linear I-V relationship, which is a standard approximation for introductory analysis.

Problems

Explain, in terms of the NMDA receptor's two-part gating requirement, why weak, infrequent presynaptic stimulation alone typically fails to produce LTP.
Solution
Weak, infrequent stimulation produces insufficient postsynaptic depolarisation to expel the NMDA receptor's magnesium block (Step2); glutamate binds but the channel remains closed, so no calcium influx occurs and no potentiation follows (Fails without, first bullet).
A drug blocks NMDA receptors specifically without affecting AMPA receptors. Predict its effect on ordinary synaptic transmission and on the induction of new LTP.
Solution
Ordinary, moment-to-moment synaptic transmission is largely unaffected, since it is carried predominantly by AMPA receptors; induction of new LTP is blocked, since it specifically requires the NMDA receptor's coincidence-detecting, calcium-permeable channel (Step1–3).
Why is LTP described as "synapse-specific" rather than as a general increase in a neuron's overall excitability?
Solution
Because the triggering calcium signal, and the resulting AMPA receptor changes, are confined to the specific synapse where coincident pre- and postsynaptic activity actually occurred (Step3–5); other synapses on the same neuron that were not simultaneously, coincidently active are not potentiated.
Field EPSP slopes recorded in hippocampal area CA1 average \(0.50\ \text{mV ms}^{-1}\) before and \(0.80\ \text{mV ms}^{-1}\) one hour after high-frequency stimulation, while a second pathway to the same cells that was not stimulated stays at \(0.45\ \text{mV ms}^{-1}\). Calculate the potentiation and explain what the unchanged control pathway shows.
Solution
Potentiation \(= (0.80 - 0.50)/0.50 = 0.60\), a 60% increase. The unchanged second pathway shows input specificity: LTP is confined to the synapses that were active during the induction, rather than being a general change in the excitability of the postsynaptic cells. This follows from the NMDA receptor mechanism, because only synapses that release glutamate during the depolarisation admit the calcium that triggers potentiation.
Explain why the NMDA receptor acts as a "coincidence detector" for presynaptic and postsynaptic activity, referencing the roles of glutamate binding and membrane depolarisation.
Solution
The NMDA receptor requires two simultaneous events to open and conduct current. First, glutamate must bind to the receptor (presynaptic activity). Second, the postsynaptic membrane must be sufficiently depolarised to expel the magnesium ion (\(\text{Mg}^{2+}\)) that blocks the channel pore (postsynaptic activity). If only glutamate binds (weak presynaptic activity), the channel remains blocked by \(\text{Mg}^{2+}\) at resting potential. If the membrane is depolarised but no glutamate is present, the channel remains closed. Only when both conditions are met does the channel open, allowing calcium influx and initiating LTP.