Myelination: Wrapping the Axon

A mutant mouse called shiverer begins to tremble in the second or third week after birth. Its neurons are present and so are its axons. What it lacks is one protein of the wrapping around those axons.

That wrapping is myelin, and in the shiverer brain almost none of it is compact. The mouse raises the three questions of this lesson. How does a glial cell decide which axon to wrap, and how thickly? What holds the layers together? And can experience or disease change the answer later?

The idea: insulation that makes the signal jump

Myelin is glial plasma membrane wound many times around an axon and squeezed until almost no cytoplasm is left between the turns. Like any membrane it is a lipid bilayer with proteins in it (see that lesson), but it is unusually rich in lipid, roughly 70 to 80 percent of its dry weight, with a large share of cholesterol and of galactolipids such as galactocerebroside (see that lesson). In the electron microscope compact myelin shows a repeating pattern. A dark major dense line marks where the cytoplasmic faces of successive turns meet; a fainter intraperiod line marks where their outer faces meet.

Why wrap at all? Think of a leaky garden hose. A bare axon is like that: current leaks out across its membrane, and the action potential has to be rebuilt on every patch it reaches, so it creeps. Myelin stacks many membranes in series. That raises the resistance across the axon wall and lowers its capacitance, so current entering at one spot travels farther and charges the membrane ahead faster.

The axon exploits this by packing its voltage-gated sodium channels into short gaps between sheaths, the nodes of Ranvier. Each wrapped stretch is an internode. The action potential is regenerated only at the nodes and seems to leap from one to the next: saltatory conduction, from the Latin for “to jump”.

The gain is large. In unmyelinated axons, conduction velocity rises roughly with the square root of diameter. In myelinated fibers it rises roughly in proportion to diameter. A classic estimate from mammalian peripheral nerve (Hursh, 1939) is about 6 meters per second for each micrometer of fiber diameter, so the largest mammalian fibers reach tens of meters per second, up to around 100, while thin unmyelinated fibers conduct at about a meter per second or less. Sheath thickness is described by the g ratio:

g ratio = axon diameter ÷ fiber diameter (axon plus myelin). Velocity of a large myelinated mammalian fiber ≈ 6 m/s × fiber diameter in µm.

Healthy peripheral fibers usually have g ratios near 0.6 to 0.7, close to the value theory predicts gives the fastest conduction for a fiber of a given size.

Two kinds of glia do the wrapping. In the brain and spinal cord, an oligodendrocyte sends out many processes, and each one makes an internode, usually on a different axon. In peripheral nerves, a myelinating Schwann cell makes a single internode on a single axon. Schwann cells that meet small axons do not myelinate; they hold several of them in grooves of their membrane. (Where both cell types come from is the subject of that lesson.)

Myelination is late and slow. In humans it begins before birth in the spinal cord and brainstem and continues through childhood and adolescence, and some association areas of the cerebral cortex are still gaining myelin in early adulthood.

Watching it happen, and who decides

Myelin forms inside a living animal, where it is hard to watch. Culture solved that. Edith Peterson and Margaret Murray reported in 1955 that sensory ganglia from chick embryos made myelin in a dish. In 1967 Mary Bunge, Richard Bunge, Peterson and Murray described long-term cultures of dorsal root ganglia from fetal rats. Over several weeks, Schwann cells in these cultures wrapped sensory axons in compact myelin with nodes, one cell to one axon, much as in a real nerve.

Because each ingredient could be changed, the dish became an experiment. The Bunges and their colleagues found that Schwann cells needed contact with an extracellular matrix, and in later purified cocultures that ascorbic acid, which cells need to build collagen, was required for both the basal lamina (the sheet of matrix the Schwann cell lays down around itself) and myelin.

That left the central question of this unit: does the glial cell decide, or the axon? Grafting experiments in the 1970s, by Albert Aguayo’s laboratory and by Weinberg and Spencer, put Schwann cells from a nerve with unmyelinated axons into a nerve with myelinated ones, or the reverse. The Schwann cells followed their new axons. In the periphery, the axon decides.

The axon tells the glia how much

The signal turned out to be a membrane protein on the axon, neuregulin-1 type III. Neuregulin-1 comes in several forms; type III keeps its active, EGF-like part tethered near the axon surface, so a Schwann cell can read it only by touching the axon. The Schwann cell reads it with two receptor tyrosine kinases of the EGF receptor family (see that lesson). ErbB3 binds neuregulin but has a weak kinase. ErbB2 binds no ligand but has a strong kinase. Paired, they signal.

The key experiments measured dose. In 2004 Galin Michailov, Klaus-Armin Nave and colleagues reported that mice with reduced neuregulin-1 expression had thinner peripheral myelin for a given axon size, a higher g ratio. Mice overproducing type III in neurons had thicker myelin. In 2005 Carla Taveggia and colleagues in James Salzer’s laboratory reported that in culture, sensory neurons lacking type III were not myelinated, and that sympathetic neurons, whose axons normally stay bare, became myelinated when made to produce more of it. Mice with one working copy of the type III gene were hypomyelinated. A threshold amount of type III decides whether an axon is myelinated, and above that the amount helps set how thick the sheath is.

Inside the Schwann cell, the signal fans out along routes you have met:

• PI3K and Akt, which in other cells promote growth and survival, support myelin growth (see that lesson).
• The Ras-ERK cascade is activated as well (see that lesson).
• PLC-gamma raises calcium, which activates the phosphatase calcineurin. Calcineurin frees the transcription factor NFAT to enter the nucleus, where in work from Gerald Crabtree’s laboratory it joined Sox10 in turning on Krox20 (also called Egr2), a transcription factor that Schwann cells need in order to myelinate.

Keep the scope. This is the rule for peripheral nerve. In the mouse brain and spinal cord, Nave’s group found that oligodendrocytes still myelinate when neuregulin-1 and ErbB signaling are removed. In culture, oligodendrocytes will even wrap inert plastic fibers of a suitable diameter, which suggests that in the central nervous system the program is partly built into the glial cell and the axon biases it.

One protein put back

Myelin basic protein (MBP) is small, strongly positive and sits on the cytoplasmic side of the membrane. It binds the negatively charged lipids of two facing cytoplasmic leaflets and pulls them together, squeezing out the cytoplasm. The seam it leaves is the major dense line. A second protein family holds the outer faces: proteolipid protein in the central nervous system, and protein zero (P0) in peripheral nerve.

In 1979 a French group reported that shiverer brain myelin lacked the major dense line and nearly all its MBP. In the mid-1980s laboratories including those of Leroy Hood and Richard Lazzarini showed that the mutant MBP gene had lost a large stretch of DNA, about 20 kilobases including most of the gene’s exons. Peripheral nerve in shiverer is far less affected, because P0 can also hold the cytoplasmic faces together there.

Design problem

Shiverer mice lack most of the MBP gene and have almost no compact myelin in the brain. A deletion of 20 kilobases could also have damaged something next to the gene. What experiment would show that the missing MBP explains the whole phenotype, and what result would you expect?

Sheaths that let go, and myelin lost

Once an oligodendrocyte can wrap, which axons does it keep? In 2015 two papers in Nature Neuroscience watched this in living zebrafish larvae, whose spinal cords are transparent. Jacob Hines, Bruce Appel and colleagues found that most early wraps were withdrawn, many within about an hour and a half. When they silenced one labeled axon among normal neighbours, by making it express tetanus toxin, which blocks vesicle release, the silenced axons were less often myelinated (about 32 percent, against 55 percent of control axons) and carried shorter sheaths. Sigrid Mensch, David Lyons and colleagues found that blocking vesicle release throughout the nervous system reduced the number of sheaths each oligodendrocyte made, by about a third, and that raising activity with a convulsant drug increased it unless release was blocked. Activity biased where sheaths stayed.

In mice, Erin Gibson, Michelle Monje and colleagues reported in 2014 that light-driven stimulation of neurons in premotor cortex increased the division of oligodendrocyte precursor cells and the making of new oligodendrocytes, and thickened myelin within the stimulated circuit. These are larval fish and mouse cortex. How much activity-dependent myelin contributes to learning in adult humans is still open.

Multiple sclerosis shows what losing central myelin costs. In its usual account, the immune system attacks myelin and oligodendrocytes in the brain, optic nerves and spinal cord, leaving scattered lesions, and axons are damaged too. Jean-Martin Charcot’s description is usually dated to 1868. It is a leading cause of non-traumatic neurological disability in young adults, is diagnosed most often between about 20 and 40, and affects women two to three times as often as men. Clinicians describe its course in types. Relapsing-remitting disease is by far the most common at onset; many such patients later enter a secondary progressive course; a minority progress from the start (primary progressive). A fourth category, progressive-relapsing, was dropped in the 2013 revision of these descriptions. Risk combines genes, with variants of the immune gene HLA-DRB1 the strongest, and environment, notably earlier infection with Epstein-Barr virus.

The nervous system tries to remyelinate lesions. Oligodendrocyte precursor cells move in, mature and make new sheaths, which are typically thinner with shorter internodes, a developmental program run a second time. Repair often falls short, especially in long-standing lesions and in older patients.