Myelin: a spike that jumps from node to node

Two axons of the same thin radius, 5 µm, side by side, both stimulated at their left ends. One is bare, its membrane full of Hodgkin and Huxley's channels all along; the other is wrapped in myelin, which leaves only nodes of Ranvier, 1 µm wide every millimetre, to carry channels. The myelinated spike jumps from node to node, over ten times faster. Vary the myelin's thickness, the spacing of the nodes and how densely they're packed with channels. (More about the model below.)

About this simulation

Each axon is 2 cm of axoplasm along x, sealed at both ends, as in the propagation demo: a strictly neutral salt solution whose ions drift in the field along the axon, and a membrane through it with ports to the sea water outside, held at its composition and at φ = 0. The bare axon's membrane is Hodgkin and Huxley's all along (1 µF/cm², their Na⁺, K⁺ and leak channels). The myelinated one is regions of the same axoplasm: each node of Ranvier, 1 µm long, with their channels at several times their density (nodes are packed with Na⁺ channels), and each internode, whose membrane is myelin: a capacitance divided by the number of wraps (200 by default) and no channels at all. Both are at 18.5 °C, with the concentrations and resistivity of the propagation demo, stimulated together every 10 ms by K⁺ injected at the left end (more into the bare one, which spreads it over a millimetre of membrane).

A node that fires drives current along the axoplasm, through the internode, to charge the next node past threshold. The internode carries it passively: being myelin, it barely charges and leaks nothing, so the current reaches far, and the spike jumps, the next node firing some tens of microseconds later. That's clearest in the current along the axon: in the myelinated one it's a sawtooth, stepping up at each node, where the channels drive it in, and draining along each internode into the myelin's small capacitance. (The internodes charge in a few microseconds, faster than the nodes take to fire, so the arrival times along them run smoothly from node to node rather than in steps.) And it takes about a tenth of the current: the bare axon charges all of its membrane as the spike goes, the myelinated one its nodes and a little myelin, which is myelin's other saving, in the ions the pumps must later put back. The bare axon has to charge all of its membrane as it goes, at 2.7 m/s (Hodgkin and Huxley's speed scaled by √a, as the propagation demo shows); the myelinated one runs at some 30 m/s, against a compartmental cable to within 1 µs a node (the cable test).

Thicker myelin (more wraps, less capacitance) is faster: with 50 wraps it runs at 12 m/s, with 800 at 68 m/s, and with 20 the internodes take too much charge and the spike dies. More channels at the nodes are faster too. In this thin axon, nodes closer together are faster (44 m/s at a quarter millimetre, 27 m/s at 3 mm): each internode charges as an RC line, in a time that grows as its length squared, about 7 µs per millimetre here, so a longer jump costs more time than it gains. Thicker fibres with thicker myelin charge faster, which is why their nodes are spaced further apart, about a hundred times the fibre's diameter.