NMDA receptors: two gates, one current, and a calcium signal

Set the chemistry and the voltage independently and watch what the receptor does. Conductance, current, and calcium influx are reported separately, because they do not peak at the same membrane potential.

Blind trial: readouts are hidden until you record a prediction.

Membrane potential

−70 mV

Mg²⁺ block of NMDA pore

Strong

AMPA conductance

NMDA open / pore unblocked

NMDA current

0 (none)

Ca²⁺ influx through NMDA

Membrane potential over time Voltage clamp

Conductance over time AMPA fast, NMDA slow

Current–voltage relationship at full ligand occupancy

NMDA net current AMPA net current Ca²⁺ influx (scaled)

Reading the ion animation

Each moving circle is one ion crossing the pore. The rate of ions is set by conductance multiplied by that ion's driving force, so ions of different species move at different rates and can move in different directions at the same voltage.

glutamate glycine / D-serine Mg²⁺ blocking the pore Na⁺ Ca²⁺ K⁺

Ca²⁺ carries roughly a tenth of the NMDA current, and is drawn at about that share. Absolute ion numbers are illustrative.

Experiment controls

Preset

Glutamate applied

Glycine / D-serine

AMPA receptors present

Hold voltage

Extracellular Mg²⁺

1.0 mM

Physiological is about 1 mM. Set to 0 to remove the block entirely.

Membrane potential

−70 mV

Drag to set the voltage the cell is held at.

Playback speed

0.20×

1× is real time. Synaptic events last only a few milliseconds, so slow the playback down to see their order.

Stimulus protocols

Pairing interval

+10 ms

Positive: glutamate first, then the action potential.

Train frequency

50 Hz

Number of stimuli

250

Repeated glutamate release. Speed the playback up to run long trains.

Predict before you look

Hide readouts until I answer

With the current settings, will Ca²⁺ enter the spine through NMDA receptors at a meaningful rate?

What to notice

At negative membrane potentials, glutamate can bind the NMDA receptor while Mg²⁺ still obstructs the pore. Ligand binding and membrane voltage are separate requirements.

Calcium and synaptic strength

Optional. The rest of the simulation works whether or not you open this.

Calcium concentration, with the two thresholds that separate weakening from strengthening in this model:

θ_d

θ_p

Moderate, prolonged Ca²⁺

→

calcineurin and protein phosphatase 1 → AMPA receptors removed → weakening

Large, rapid Ca²⁺

→

calmodulin → CaMKII → AMPA receptors added → strengthening

Synaptic strength 1.00×

Let strength change the AMPA response

Which direction a synapse moves depends on the amount and time course of the calcium signal, not simply on whether NMDA receptors opened. Weak activity that produces a small, drawn-out calcium rise can weaken a synapse that strong activity would strengthen. The thresholds here are illustrative, and the biochemistry is compressed into two arrows.

What this model does and does not do

NMDA receptors need glutamate and a co-agonist, glycine or D-serine depending on the synapse and brain region. Both sites must be occupied.
Mg²⁺ block follows 1 / (1 + [Mg]/3.57 · e^(−0.062V)), the standard empirical description. It is graded, and its voltage dependence shifts with Mg²⁺ concentration.
Mg²⁺ occupancy is drawn as flicker rather than as an ion being pushed out of the pore. The ion binds and unbinds rapidly; depolarization changes how much of the time it is bound.
Current is conductance multiplied by driving force. Net current through both receptors reverses near 0 mV. Ca²⁺ influx does not, because the calcium equilibrium potential is far more positive.
AMPA and NMDA conductances use single-exponential deactivation with very different time constants. Real receptors have multiple kinetic states, desensitization, and subunit-dependent kinetics.
When the voltage is not held, one synapse is given enough conductance to depolarize the whole compartment. A real single synapse produces a fraction of a millivolt at the soma.
Currents are in arbitrary units. Concentrations, single-channel conductance, receptor numbers, spine geometry, calcium buffering, and stochastic gating are all omitted.
The plasticity panel is a compressed illustration of the calcium-level account. Real plasticity involves many interacting pathways, other calcium sources including voltage-gated channels and internal stores, and depends on receptor subunit composition.