Action Potential Diagram: Labeled Phases, Ion Channels and 3D Axon Model

An action potential is the electrical signal that neurons, muscle cells and some other cells use to send messages. It is a rapid, self-renewing change in membrane potential: the inside of the cell flips from negative to positive and back in a few milliseconds, and the change travels along the cell membrane without getting smaller.

The whole event is caused by ions moving through protein channels. Sodium ions (Na+) rush into the cell to make the inside positive (depolarization), then potassium ions (K+) flow out to make it negative again (repolarization). The two ions are kept at very different concentrations inside and outside the cell by the Na+/K+ pump, and those concentration gradients are the battery that powers every action potential.

These are the numbers most textbooks use. They are averages for teaching: real neurons rest anywhere from about −40 to −80 mV and have thresholds from about −40 to −55 mV. The table near the end of this page lists how the values change between cell types.

A labeled action potential diagram plots membrane potential (mV) against time (ms) and shows seven parts: the resting potential, the stimulus, the threshold, depolarization (the rising phase), the peak or overshoot, repolarization (the falling phase) and hyperpolarization (the undershoot), followed by the return to resting potential. The two refractory periods are usually drawn as bars under the curve.

The dashed hump on the left of the labeled diagram is a failed initiation: a small graded depolarization that does not reach threshold, so no action potential follows. It is a common item on labeled diagrams because it shows that threshold, not stimulus size, decides whether the neuron fires.

The steps of an action potential are: (1) the membrane rests at −70 mV; (2) a stimulus depolarizes it to threshold; (3) voltage-gated Na+ channels open and Na+ rushes in; (4) at the peak, Na+ channels inactivate and voltage-gated K+ channels open; (5) K+ flows out and the membrane repolarizes; (6) K+ channels close slowly, so the membrane overshoots into hyperpolarization; (7) the membrane returns to rest.

1. Rest. The voltage-gated Na+ channel has two gates. At rest its activation gate is shut and its inactivation gate is open, so it is closed but ready. The voltage-gated K+ channel is closed. K+ leaks out through leak channels.
2. Stimulus. Neurotransmitter opens ligand-gated channels on the dendrites. Positive ions enter and the depolarization spreads to the axon initial segment. If the summed depolarization reaches about −55 mV, the neuron fires.
3. Depolarization. At threshold, the activation gates of many Na+ channels open. Na+ is pulled in by both its concentration gradient and the negative inside. Each Na+ ion that enters depolarizes the membrane further and opens more channels.
4. Peak. A fraction of a millisecond after opening, each Na+ channel's inactivation gate swings shut. At about the same time, the slower voltage-gated K+ channels finish opening.
5. Repolarization. K+ leaves the cell, taking positive charge with it, and the inside becomes negative again.
6. Hyperpolarization. The K+ channels close with a delay, so for a moment more K+ leaves than at rest and the voltage dips below −70 mV. During repolarization the Na+ channels reset: the activation gate closes and the inactivation gate reopens.
7. Return to rest. When the K+ channels have closed, the leak channels and the Na+/K+ pump hold the membrane at −70 mV, ready for the next action potential.

A common mistake is to say the Na+/K+ pump repolarizes the membrane. It does not. Repolarization is caused by K+ flowing out through channels, which is thousands of times faster than pumping. One action potential moves only a tiny fraction of the cell's ions, so the pump works in the background to keep the gradients topped up.

The ionic basis of the action potential is the unequal distribution of Na+ and K+ across the membrane, plus channels that let each ion through at a different time. Na+ is about 12 times more concentrated outside the cell, and K+ is about 40 times more concentrated inside.

Concentrations are typical mammalian values from OpenStax Biology 2e (table 35.1). Every ion has an equilibrium potential, the voltage at which its electrical pull exactly balances its concentration gradient (given by the Nernst equation; a tenfold concentration difference is worth about 60 mV). The membrane potential at any moment lies between the K+ and Na+ equilibrium potentials, closer to whichever ion the membrane is more permeable to. At rest the membrane is far more permeable to K+, so it sits near E_K; at the peak it is far more permeable to Na+, so it swings toward E_Na. That shift in permeability is the whole action potential.

The resting membrane potential is the voltage across the membrane of a cell that is not firing, about −70 mV in a typical neuron. It is negative because the resting membrane is much more permeable to K+ than to Na+: neurons have far more K+ leak channels than Na+ leak channels. K+ leaks out, down its concentration gradient, and the large negative ions it leaves behind cannot follow.

The leak stops getting bigger because the inside becomes negative and starts to pull K+ back. If K+ were the only ion that could cross, the membrane would settle at E_K, about −75 to −90 mV. A small Na+ leak drags it a little more positive, to about −70 mV. In the classic example from Neuroscience Online the resting K+ permeability is about 100 times the Na+ permeability.

The Na+/K+ pump uses one ATP to move 3 Na+ out and 2 K+ in. It sets up the gradients in the first place and keeps them there. Because it moves more positive charge out than in, it also makes the inside a few millivolts more negative directly. At the muscle end-plate, for example, pump activity makes the membrane 2–4 mV more negative than the rest of the fibre.

Resting potential is not the same in every cell: ventricular heart muscle rests near −90 mV, atrial muscle near −80 mV, skeletal muscle fibres around −60 to −90 mV, and SA node cells never rest at all. See the cardiac sections below.

A graded potential is a small, local change in membrane potential whose size depends on the strength of the stimulus; an action potential is a large, fixed-size, all-or-none signal that travels the length of the axon without fading. Graded potentials happen in dendrites and the cell body; action potentials start at the axon initial segment.

All-or-none means that once threshold is reached, the action potential always runs to the same size, however strong the stimulus. A stronger stimulus does not make bigger action potentials; it makes them come more often. That is how neurons code stimulus strength: by firing rate, not by spike size.

Excitatory postsynaptic potentials (EPSPs) move the membrane toward threshold and inhibitory ones (IPSPs) move it away. Inputs at different places add up (spatial summation) and inputs that arrive close together in time add up (temporal summation). In OpenStax's example, inputs that together depolarize the membrane by 15 mV, from −70 to −55 mV, reach threshold and fire the neuron.

An action potential propagates because the Na+ entering one patch of membrane spreads sideways inside the axon and depolarizes the next patch to threshold, which then fires its own action potential. In unmyelinated axons this happens patch by patch along the whole membrane (continuous conduction). In myelinated axons it happens only at the nodes of Ranvier, and the signal seems to jump from node to node (saltatory conduction).

Under the myelin, the current spreads passively and gets weaker, but it is still strong enough to bring the next node to threshold, where a fresh, full-size action potential is made. Each node is like a relay station. The spacing matters: if nodes were too far apart, the current would fall below threshold before reaching the next one; if they were too close, conduction would be slower than it needs to be.

Conduction speed depends on axon diameter and myelin. Wider axons conduct faster, but the gain is poor: doubling the speed needs about four times the diameter. Squid solve this with a giant axon; vertebrates use myelin. Measured speeds range from less than 0.1 m/s to about 100 m/s. In one recent study, single nodes of a rat touch-sensory nerve conducted at about 48 m/s.

Action potentials travel only one way, away from the cell body, because the membrane they have just left is in its absolute refractory period. Its Na+ channels are inactivated and cannot be reopened by the current spreading back toward them.

The refractory period is the time after an action potential during which the membrane cannot fire normally. It has two parts. In the absolute refractory period no stimulus, however strong, can start a second action potential. In the relative refractory period a stronger-than-normal stimulus can.

Refractory periods do two jobs. They make propagation one-way, and they set a maximum firing rate: a neuron with a 1–2 ms absolute refractory period cannot fire faster than roughly 500–1,000 times a second, and most fire far more slowly. Na+ channels need about 3–4 ms to recover fully from inactivation. In the heart the refractory period lasts about 250 ms, almost as long as the contraction, which is why heart muscle cannot be tetanized.

In a neuron, inputs arrive at the dendrites and cell body as graded potentials, add up at the axon hillock and initial segment, where the action potential starts, and the action potential then travels down the axon, jumping between nodes of Ranvier under the myelin sheath, to the axon terminals and synaptic boutons. The 3D model above shows each of these, plus a magnified slice of the axon membrane with its channels, pump and ions. Here is every labeled structure of the model, by group.

When an action potential reaches an axon terminal, it opens voltage-gated Ca2+ channels. Ca2+ flows in and makes synaptic vesicles fuse with the membrane and release neurotransmitter into the synaptic cleft. The transmitter binds receptors on the next cell and produces a graded postsynaptic potential there, about 1 ms after the action potential arrived. The transmitter is then broken down by enzymes or taken back up by the neuron or by glia.

So the signal changes form twice: electrical (action potential) to chemical (neurotransmitter) at the terminal, and chemical back to electrical (graded potential) at the next cell. Electrical synapses, where ions pass directly through gap junctions, skip the chemical step and have almost no delay.

The cardiac action potential of a ventricular muscle cell has five phases: phase 4 (rest at about −90 mV), phase 0 (rapid depolarization by Na+ entry), phase 1 (a small notch as transient K+ channels open), phase 2 (a long plateau as L-type Ca2+ channels let Ca2+ in) and phase 3 (repolarization as Ca2+ channels close and K+ flows out). The whole event lasts 250–300 ms.

The plateau is the key difference from a neuron. The Ca2+ that enters during the plateau supplies about 20% of the Ca2+ the cell needs to contract; the rest comes from the sarcoplasmic reticulum. Because the fast Na+ channels stay inactivated until the membrane has fully repolarized, the absolute (effective) refractory period lasts about 200 ms, and the relative refractory period about 50 ms more. A new action potential cannot start until the heart has nearly finished contracting, so the heart always relaxes and refills between beats.

The SA node pacemaker potential is a slow, spontaneous depolarization from about −60 mV to a threshold of about −40 mV, followed by a Ca2+-driven upstroke to about +15 mV and K+-driven repolarization. Because SA node cells have no stable resting potential, they fire on their own, and they set the heart rate.

• Phase 4 (pacemaker potential or prepotential). Starting near −60 mV, a slow inward current called the funny current (I_f) begins. It flows through HCN channels (mainly HCN4), which open when the membrane is hyperpolarized, the opposite of most voltage-gated channels; that is why it was called "funny". The net inward current is carried mostly by Na+. Late in phase 4, T-type Ca2+ channels open at about −50 mV and speed the climb.
• Phase 0 (upstroke). At about −40 mV, L-type Ca2+ channels open and Ca2+ entry drives the upstroke. It is much slower than a neuron's or a ventricular cell's, because Ca2+ channels are slower than fast Na+ channels.
• Phase 3 (repolarization). Ca2+ channels inactivate and K+ channels open; K+ leaves until the cell is back near −60 mV, and the cycle restarts.

Pacemaker cells have no phase 1 or phase 2. Left alone, the SA node fires about 80–100 times a minute; the vagus nerve slows it and sympathetic nerves speed it up by changing the slope of phase 4.

A skeletal muscle action potential starts at the neuromuscular junction. The motor neuron releases acetylcholine, which opens ligand-gated channels in the muscle fibre's motor end-plate. Positive ions flow in and make a local graded depolarization, the end-plate potential (EPP). The EPP opens voltage-gated Na+ channels next to the end-plate, and a muscle action potential spreads along the sarcolemma and down the T-tubules.

In the T-tubules the action potential triggers Ca2+ release from the sarcoplasmic reticulum, and Ca2+ starts the sarcomeres contracting (see our sarcomere diagram). Acetylcholinesterase breaks down acetylcholine in the cleft so that each nerve impulse gives one muscle action potential.

The end-plate potential is normally 2–5 times bigger than it needs to be to reach threshold. This margin is the safety factor of neuromuscular transmission, and it falls with fatigue, injury and diseases such as myasthenia gravis. A single vesicle (quantum) of acetylcholine gives only a tiny, about 1 mV, miniature end-plate potential; a normal EPP is the sum of many quanta released together.

Neurons, skeletal muscle fibres and heart cells all use action potentials, but the shapes differ because each cell uses a different set of channels. The table compares the textbook values.

The textbook numbers on this page, −70, −55 and +30 mV, are averages. Different books and different cells give different values, and students often meet more than one set.

Invertebrate axons such as the squid giant axon, used in the classic experiments of Hodgkin and Huxley, have different ion concentrations again (for example about 400 mM K+ inside and 20 mM outside). The principles are the same in all of them: rest near E_K, swing toward E_Na, and return.

Most textbook diagrams say the action potential starts at the axon hillock, the cone where the axon leaves the cell body. Measurements in real neurons put the trigger zone a little further along, in the axon initial segment (AIS), the first 10–60 µm of the axon before the myelin begins. The AIS has a lower threshold than the cell body because it holds far more voltage-gated Na+ channels: about 30–50 times more Nav1.6 channels than the soma. (Solé & Tamkun (2019): Trafficking mechanisms underlying Nav channel subcellular localization in neurons, Channels 14:1.)

These channels are not scattered at random. A large scaffolding protein, ankyrin-G, holds them in place, and the same ankyrin-G system clusters Na+ channels at every node of Ranvier. That is why the AIS and the nodes look so alike: both are short stretches of axon packed with Na+ channels, built to make action potentials. OpenStax describes the same idea in simpler terms: summation happens at the axon hillock or initial segment, where voltage-gated Na+ channels are dense. (OpenStax Anatomy & Physiology 2e, §12.5 Communication Between Neurons.)

In most diagrams, every patch of membrane that fires an action potential has voltage-gated Na+ channels and voltage-gated K+ channels side by side. That is true for unmyelinated axons and for the classic squid giant axon, but mammalian myelinated axons are organized differently. The node itself is packed with Na+ channels. The fast voltage-gated K+ channels, Kv1.1 and Kv1.2, sit under the myelin on each side of the node, in a region called the juxtaparanode, separated from the node by the paranodal junctions where the myelin seals onto the axon. (Pinatel & Faivre-Sarrailh (2020): Assembly and function of the juxtaparanodal Kv1 complex, Life 11:8.)

So what repolarizes a mammalian node? A 2019 study recorded directly from nodes of rat sensory nerves and found that blocking voltage-gated K+ channels had little effect. Instead, two leak-type channels from the two-pore-domain (K2P) family, TREK-1 and TRAAK, were clustered at the nodes at more than 3,000 times their density on the cell body. Knocking them down slowed repolarization and cut the conduction speed from about 48 m/s to 25–30 m/s. (Kanda et al. (2019): TREK-1 and TRAAK are principal K+ channels at the nodes of Ranvier, Neuron 104:960.)

Nodes also contain slow voltage-gated K+ channels of the KCNQ (Kv7) family. KCNQ2 sits with Na+ channels and ankyrin-G at nodes and initial segments throughout the brain, spinal cord and peripheral nerves. Drugs that open KCNQ channels make axons less excitable, and mutations in KCNQ2 cause neonatal epilepsy and myokymia, a rippling twitch of muscles caused by extra firing in motor axons. The 3D model shows all three: Na+ channels and KCNQ channels at the node, K2P leak channels at the node, and Kv1 channels hidden under the myelin. (Devaux, Kleopa, Cooper & Scherer (2004): KCNQ2 is a nodal K+ channel, Journal of Neuroscience 24:1236.)

The SA node sets the heart rate, but the rate it sets is not fixed. Sympathetic nerves release norepinephrine, which acts on beta-adrenoceptors and steepens phase 4, mainly by boosting the funny current and the slow inward Ca2+ currents, so threshold is reached sooner. The vagus nerve releases acetylcholine, which acts on muscarinic receptors, increases K+ conductance and hyperpolarizes the cell, and reduces the funny and Ca2+ currents, so the cell takes longer to reach threshold. (Klabunde, CV Physiology: Regulation of Pacemaker Activity.)

Without any nerve input the SA node fires at about 100–110 beats per minute in Klabunde's figures (80–100 in OpenStax); resting vagal tone is what brings a typical resting heart rate down to 60–80. The channels that carry the funny current are HCN4 channels. They are concentrated in the SA node and conduction system, and inherited loss-of-function HCN4 mutations mainly cause a slow heart rate (sinus bradycardia). One mutation that shifts their activation only about 5 mV more negative was calculated to slow the heart by about 29%. (DiFrancesco (2015): HCN4, sinus bradycardia and atrial fibrillation, Arrhythmia & Electrophysiology Review 4:9.)

In psychology courses the action potential is taught as the way a neuron "fires". The key points are:

• At rest, the inside of a neuron is negative, about −70 mV: the resting potential.
• Messages from other neurons push the charge up or down. If it reaches the threshold of excitation, about −55 mV, the neuron fires.
• Firing is all-or-none: the action potential is either fully sent or not sent at all. Stronger stimuli make a neuron fire more often, not more strongly.
• During firing, Na+ rushes in and the inside briefly becomes positive (about +30 mV); then K+ flows out and the charge drops, briefly overshooting to about −80 mV, before returning to rest.
• The myelin sheath, made by glial cells, insulates the axon and speeds the signal, which jumps between gaps called nodes of Ranvier.
• At the end of the axon, the terminal buttons release neurotransmitters from synaptic vesicles into the synaptic cleft. Leftover transmitter is removed by reuptake into the neuron that released it.
• The sodium–potassium pump moves 3 Na+ out for every 2 K+ it brings in, keeping the neuron ready to fire.

Test yourself with this unlabeled action potential diagram. Write the name of each numbered part, and the main ion movement where there is one, then open the answer key.

Extra questions:

1. Which ion enters the cell during depolarization, and through which channel?
2. Why can't a second action potential start during the absolute refractory period?
3. Why does a stronger stimulus not produce a bigger action potential?
4. Name two ways an action potential in a ventricular muscle cell differs from one in a neuron.
5. What carries the upstroke of an SA node action potential?

Related pages: sheep brain dissection, heart blood flow and human anatomy.