Architecture of myelinated axons

The term myelin as the ensheathment of axons dates back to 1854 by the German pathologist Rudolf Virchow. However, the identification of oligodendrocytes as the cellular source of myelin did not happen until 1922 by Pio Del Rio-Hortega. Since then, the structural, molecular and physiological properties of myelinated axons have been defined quite well (Fig. 1). Each oligodendrocyte forms dozens of myelin segments (internodes), each of which consists of tightly packed cell membranes, that are iteratively ‘wrapped’ around the axon. The consecutive alignment of individual internodes covers the axon along its length, leaving only short unmyelinated gaps between each internode, the nodes of Ranvier, which are highly enriched in voltage gated Na⁺ channels (Fig. 1). The tight stacking of myelin membrane electrically insulates the axon. This insulation increases membrane resistance, so that axon depolarisation spreads over a longer distance with a lower potential drop than along an unmyelinated axon. In consequence, the axon membrane is still sufficiently depolarised at distant nodes of Ranvier, so that resident Na⁺ channels can initiate a new action potential. This way, the action potential seems to ‘jump’ from node to node. This type of saltatory conduction enables action potential propagation with up to 100m/s, whereby thickness and length of the myelin amongst other factors affect conduction speed.

Fig. 1:

Cellular architecture of myelinated axons in the CNS A Cartoon showing a neuron with a myelinated axon (magenta), a myelinating oligodendrocyte (green) and an oligodendrozyte precursor cell (blue). B Schematic cross-sectional view of a myelinated axon. C Comparison of continuous (top) and saltatory (bottom) nerve conduction. In saltatory conduction, action potentials are only initiated at the nodes of Ranvier, which then ‘jumps’ from node to node. D Cartoon showing a longitudinal section through a myelinated axon around the node of Ranvier region.

Adapted and adaptive myelination to regulate nervous system function

It is possible to mathematically determine the optimal parameters for fastest action potential propagation, which are primarily affected by the thickness of the axon, the thickness of the myelin, and the distance between the nodes of Ranvier. In nature, however, myelination patterns can greatly differ from the theoretically optimal parameters in order to precisely coordinate temporal control of information flow. For example, it has been shown that in the auditory system of the gerbil, action potential arrival times in the auditory brainstem are regulated by variation in internodal length. This enables precise control of spatial hearing (Ford et al., 2015), and very nicely exemplifies that myelination patterns can be used to regulate network function.

In addition to highly specific myelination, completely atypical patterns of myelinated axons have been described recently. For example, axons of pyramidal neurons in the adult cortex often show only discontinuous myelination with long, unmyelinated gaps (Tomassy et al., 2014). The functional repercussions of such patchy myelination are currently totally unclear. However, in this context, it is interesting to note that adaptive myelination is involved in forms of learning. Acquisition of motor skills such as juggling involves white matter changes in humans. Similarly, mice fail to learn complex motor tasks when the formation of new myelin was genetically prevented (McKenzie et al., 2014). Together, these findings are indicative that active communication between neurons and oligodendrocytes may represent an additional regulatory element of higher nervous system function.

Support of axon survival

Enabling fast action potential propagation is not the only role of myelinating glia. The tight cellular interaction between an axon and surrounding oligodendrocytes does not only provide electrical insulation, but also generally isolates the axon from surrounding cells like astrocytes, an important intermediate cell type for coupling neurons to vasculature. Neurons can have very long axons so that the corresponding cell body can be over one meter away (in large animals). In order to meet the local energy demands during action potential generation, myelinating oligodendrocytes supply axons with glycolysis metabolites (Saab et al., 2013). This ensures long-term axon survival. Indeed, it is plausible that lack of metabolic supply by oligdendrocytes contributes to axon degeneration in disease, for example when demyelinated axons do not get repaired in MS.

The current state of research shows how important oligodendrocytes are for support and modulation of axon function. However, what controls if and how an axon gets myelinated? How plastic are these processes and what are the causes of their deregulation in disease? Here, we still lack fundamental understanding of the principles that underlie communication between these two cell types.

As the interactions between axon and oligodendrocyte are likely to be under the control of cell intrinsic and extrinsic mechanisms, it is important to study them under physiological conditions in vivo. Myelination is a slow and long lasting process that only starts relatively late during ontogenesis. This provides technical challenges for studying dynamics of cellular interactions and their genetic control in many experimental models. Here, zebrafish represent a model organism with which one can overcome many of the technical limitations, as we will describe in the second part of this article.

Visualising cellular dynamics using in vivo microscopy of fluorescent reporters

The small size and optical translucency of young zebrafish make them ideally suited for in vivo microscopy. Continuous technological advances enable us to look into the nervous system of living animals with unprecedented resolution. This also allows to carry out longitudinal studies of the same cell in the same animal in order to obtain information about dynamics of structural changes (see excursion box for an overview of most relevant microscopy setups, Fig. 3). Subtypes of neurons and glial cells can be labelled with transgenic reporter constructs. A typical transgene consists of a gene regulatory sequence that drives the expression of a reporter protein (Fig. 2). The latter can be fluorescent proteins like the green fluorescent protein GFP and its variants, which fluoresce in different colours of the visual spectrum. However, it is equally possible to express any other protein, for example mutant proteins to manipulate cell function. Microinjection of transgenesis constructs into zebrafish eggs allows visualisation of specific cells and cell types in different colour combinations in the resulting organism. When a transgene is integrated in cells of the germline, breeding from these adults will give rise to a full transgenic animal in the daughter generation and all cells of e.g. a certain cell type are labelled (Fig. 2). This way, we and others have in the past generated a series of transgenic reagents and lines to specifically label myelin, myelinated axons, oligodendrocytes and their precursors with different fluorescent proteins (see (Czopka, 2016) for a detailed overview of published lines used to study oligodendrocyte biology in zebrafish). Using in vivo live cell microscopy in zebrafish, we have been able to address the long-standing question of how oligodendrocytes form a multi-layered myelin sheath. Even though the structure of myelin is well known, the principles of myelin morphogenesis remained totally unclear. When we followed individual nascent internodes, we could show that each new layer of membrane is added at the centre of the myelin sheaths, from where it extends laterally to form the multi-layered internode (Snaidero et al., 2014). This was the first demonstration of myelin morphogenesis in vivo, and also provides a good example of how live cell microscopy in zebrafish – together with genetic and ultrastructural analyses – can help understand fundamental biological questions.

Fig. 3:

Major microscopy setups used for in vivo live cell microscopy

Where does new myelin come from?

Myelination of CNS axons is a long-lasting process. More and more axons can get myelinated over time – in some brain regions in humans up until our mid 30’s. At the same time, each individual oligodendrocyte can generate many internodes (the number seems primarily determined by axonal parameters). This raises the question of where new myelin comes from during long-term development? Elegant in vivo imaging studies in zebrafish have revealed highly dynamic behaviour of oligodendrocyte precursor cells (Kirby et al., 2006). However, dynamics of myelinating cells, particularly when oligodendrocytes determine their internode number, remained elusive. One plausible scenario would be that oligodendrocytes can plastically regulate the myelin they produce to dynamically adjust it to axonal demands. To test this hypothesis, we have carried out high-resolution time-lapse imaging and long-term analyses of oligodendrocytes in the zebrafish. In this work, we could show that individual oligodendrocytes initiate all their future internodes in the very short time window of no more than six hours during their differentiation (Fig. 4) (Czopka et al., 2013). After this time, myelinating oligodendrocytes lose their capability to form new internodes.

Fig. 4:

Insights and open questions regarding the formation, maintenance and repair of myelinated axons

The existence of such a short time window for each individual oligodendrocyte to form new myelin segments has important implications. It means that new internodes during long-term development can only come from newly differentiating oligodendrocyte precursor cells. These cells represent a substantial fraction of about 5% of all cells in the adult brain (in the mouse). Oligodendrocyte precursors are also the cellular source during regeneration of damaged myelin (remyelination). Remyelination is an endogenous regenerative process in the CNS, which is, however, often less efficient than primary myelination. This is also evident in diseases like MS, where remyelination often gets progressively worse and eventually fails. If the same oligodendrocyte precursor cells are responsible for primary myelination and secondary remyelination of axons, why is remyelination less efficient? There is probably no simple answer to this question, particularly because many diseases have multiple causes and influencing factors. However, it is possible to test aspects of such questions in the zebrafish model to obtain insights into fundamental principles of cellular behaviours, their intrinsic molecular control, and to disentangle the roles of surrounding cell types.