Myelin speeds the conduction of nerve impulses by a factor of 10 compared to unmyelinated fibers of the same diameter. This increases the nervous system's information processing speed and delivery, decreasing reaction times to stimuli, increasing temporal precision, more closely synchronizing spatially-distributed targets (e.g. different regions of a muscle sheet) and providing for shorter delays in feedback loops, increasing their intrinsic stability. Myelin has other advantageous characteristics. It confers a several hundred-fold improvement in metabolic efficiency for recouping the energy cost of nerve impulse traffic. For a nervous system such as ours, which already accounts for an age-dependent 20-50% of the body's resting metabolic energy budget, this is not an inconsequential advantage. Were we a non-myelinated species, we would have greatly diminished problems with being overweight! Another advantage is economy of space. Its speed-up permits a trade-off with size that allows a much more compact nervous system for a given axonal conduction speed, promoting nervous systems such as ours with large numbers of neurons and correspondingly (so we suppose) greater computing power. To attain the same inter-hemispheric travel time for nerve impulses using unmyelinated axons would require scaling up brain dimensions over 100-fold. Somewhat moore abstruse is the fact that the reduced currents surrounding myelinated fivers reduces the "cross-talk" between adjacent fivbers, permitting closer associations without requiring special arrangement to decrease such potentially disruptive interactions. These advantages conferred by myelin provide clear sources of selective pressure for its evolutionary invention.
Myelin has a few disadvantages as well. It costs a significant amount in metabolic and biosynthetic resources to producte the many layers of lipid-rich membrane that comprises myelin. This can be a particularly bothersome problem in environments such as the "oligotrophic" open ocean, which is distant from continent-based sources of nutrients. Further, some of the key components of myelin may be particularly limiting, as in the case of cholesterol, which is not synthesized by protostomes and hence is an essential "vitamin" in their diets.
Like other evolutionary advances that leave no fossil record, we must infer unobservable events in myelin evolution from current observable conditions. To start with, myelin is generally viewed as an exclusively vertebrate innovation. This seems odd, given the number of significant advantages it confers, and were it true, it would make the evolutionary sleuthing much harder. However, not even all vertebrates possess myelinated nervous systems, nor, in fact, are the smaller axons of vertebrates myelinated. The hyperoartia (lampreys) lack myelin, as do the evolutionarily more primitive craniates, the hyperotreti (hag fishes)(Bullock et al 1984). These groups are presumed to represent the primitive condition in the vertebrate line, albeit the possibility for secondary loss by extant taxa has not been adequately investigated.
Neve impulses are generated by ion courrent (carried chiefly sodium ions) flowing into an axon at an "active zone" where the voltage-gated sodium channels present are open. In unmyleinated axons (see figure below) most of the current spreads to regions close to the active zone because it exits the axon the high capacitance and leak-conductance of that membrane. In myelinated fibers, on the other hand (see scond figure below), active zones are restricted to the axonal membrane exposed at the nodes. The multiple layers of dielectric presented by the myelin sheath proportionately reduce the trans-fiber capacitance compared to a non-myelinated fiber. This in turn greatly reduces the radial leakage of transient currents flowing through the sheath during nerve impulses (current flowing through a capacitor is proportional to the time derivative of voltage-change across it) in regions between nodes (the "internodes"). Although sodium channels are concentrated at the nodes at densities well above those of typical unmyelinated fibers, the mean density averaged over the length of the fiber is much less, resulting in a smaller ionic imbalance that must be restored at the expense of metabolic energy (ionic pumps) after an impulse passes. The smaller internodal current loss leaves more current available to raise distant nodes to threshold, which will thus happen more quickly, speeding impulse propagation. Further, the reduced size of exposed nodal membrane, reduces the area of membrane into which this current must flow and increases the rate of change of voltage at the node (technically, the time-constant for charging nodal capacitance is reduced), allowing threshold to be reached faster, further speeding the impulses.
Conduction speed also depends on axial resistance through the interior of the fiber. The larger the diameter of this interior space, the lower the resistance, a principle that holds for unmyelinated as well as myelinated fibers. Changing just the diameter of this inner space (typically mostly filled by the axon) gives a conduction speed that varies as the square root of the inside diameter (1/2 power). This explains the frequent observation of "giant" axons in invertebrates, especially prevalent in circuits involved in rapid escape reactions (e.g. Nicol 1947). The thickness of the myelin sheath, however, varies with the interior diameter, typically maintaining a fairly constant ratio to it (ca 0.7). The result is that internode capacitance per unit area of axon decreases with fiber diameter, adding to the effects from decreased axial resistance and giving the conduction speed a first power dependency on inner (or outer) diameter over a substantial range.
The essential structural features that produce these properties are the restriction of leakage current to cross multiple membrane lamellae in the internode and the reduction of surface area of nodal membrane. If we take these as the defining characteristics of "myelin" then myelin occurs in several taxa of phylogenetically distant invertebrates: among crustaceans of the subclasses malacostraca (including decapod shrimp) and copepoda, and among annelids of the groups polychaeta and oligochaeta. There are several variants in myelin structure seen in invertebrates which still achieve the same functional results.
Vertebrate myelin is spirally wrapped. That is, a continuous double lamella laid down by a Schwann cell or oligodendrocyte winds around the fiber starting against the axon and spiraling outward. Compact myelin is the form most typical of mature vertebrate myelin, with both cytoplasmic and extracellular spaces eliminated. In EM cross section, this gives rise to a regularly banded alternation of thick and thin lines referred to as the "major dense line" (apposed ecytoplasmic membrane leaflets) and the "intraperiod line" (apposed extracellular leaflets). Vertebrate myelin has regions (Schmidt-Lantermann incisures) that retain cytoplasm over short segments and these form continuous spiral intracytoplasmic pathways from just outside of the axon to the outer layer of glial membrane. Spiral wrapping has the disadvantage of requiring specializations to prevent radial current leak following along the spiral path between lamellae.
Reports of myelin in invertebrates are scattered among several phylogenetically diverse groups as shown on the phyletic tree below. Not all of these reports have been confirmed in the electron microscope yet (asterisks in the figure below) and recent EM evidence has failed to confirm its presence in one of the polychaete groups indicated in the figure below (bamboo worms) - see Hartline and Kong (2008).
Myelin of oligochaetes (especially the earthworm) is the best studied of invertebrate myelin at the electron microscope level. It is spirally wrapped, at least in places, as in the vertebrate case (Roots et al. 1991; see figure below). It consists of 20 to 200 layers, often, but not always, compact. The non-compact regions typically have thin layers of cytoplasm sanndwiched between glial cell membranes. Being intracellular and narrow, however, their capability for compromising sheath insulation appears limited. While conduction speed of earthworm myelinated fibers is high compared to that for non-myelinated fibers of the same diameter, the advantage is only a few fold (Gunther, 1976).
All crustacean myelin so far described has proven to be concentrically arranged: lamellae of a given layer encircle the central axon, abutting against corresponding margins of the same layer at the margins. Concentric wraps are electrically more efficient, requiring only that tight seals be made at the margins of the myelinating cells, the "seams," to prevent short circuiting of the insulation. Thus myelin in the decapod shrimp is sometimes compact and sometimes only semicompact, that is, it excludes only the extracellular gap while retaining cytoplasm or vice versa. What is important for its electrical integrity is that the space between layers are sealed from each either by a continuous membranous barrier or by tightly joined appositions at the seams. Two somewhat different forms have been described for different shrimp taxa. In the more "advanced" Caridean shrimp (including the prawns), each myelin layer includes a thin sheet of sandwiched cytoplasm and extends fully around the axon, meeting itself on the opposite side in a seam. The seams of successive layers alternate sides going from inner to outer sheath, producing long electrical paths between seams of adjacent layers (Heuser and Doggenweiler 1966). In contrast, in the more primitive Dendrobranchiata (to which the commercial penaeid shrimp belong), each myelin layer only extends half way around the interior core, with the margins of each half-layer meeting those of a sister layer at the same level coming from the other side (Huang et al 1963). Penaeid fibers are unusual in that the axon occupies only a part of the interior space. The rest is occupied by a glial cell and is termed the "submyelinic space" (Hama 1966). Current entering the axon through voltage-gated channels flows readily out of it again as in non-myelinated nerve, but is trapped and confined in the submyelinic space, as if it were a giant axon filling the space (see figure below). Penaeid axons of 120 microns diameter conduct impulses at the fastest speeds known: over 200 m/s (cf 100 m/s for the fastest myelinated vertebrate axon)(Kusano 1966).
Copepod myelin is compact in the outermost layers of a sheath, but often there is a substantial gap between concentric rings. There is no evidence of seams, so there appear to be no weak spots in the sheath through which current might pass more easily than through the membrane itself (Weatherby et al 2000). Perhaps this is why only cytoplasmic space is consistently eliminated in most parts of copepod myelin. Of the copepods, only superfamilies of more recently-evolved taxa in the order Calanoida appear to have myelinated axons (Davis et al 1999). Calanoids of the more "primitive" superfamilies Augaptiloidea [Arietilloidea] and Centropagoidea [Diaptomoidea] appear to lack myelin (for electron micrographs comparing axons of myelinated and unmyelinated coepods, click here; For more about copepod myelin, click here).
Of the nodes examined in invertebrates closely enough so far, only the Caridean shrimp (Palaemonetes) have circumferential nodes like the vertebrates (Heuser & Doggenweiler, 1966). Earthworms, Penaeids and copepods appear only to have "fenestrated" or "focal" nodes in which only a small piece of the axon is exposed through a gap in the surrounding sheath (Gunther 1976; Hsu and Terakawa 1996; Weatherby et al 2000; see figure below). More than this is not necessary for saltatory conduction to occur, since only a small amount of membrane is needed to accommodate the sodium channels necessary for regeneration of the nerve impulse.
Vertebrate myelin Annelid myelin Caridean shrimp myelin Penaeid shrimp myelin Copepod myelin
A plethora of special molecules have been identified in vertebrate myelin, often found exclusively in the structure. Are there shared homologues in different taxa that suggest a common ancestry for some and maybe all cases of myelin evolution? Light and mystery is shed on the origin of vertebrate myelin by the finding that the genome of a protochordate, Ciona (tunicate) contains homologues for several of the myelin tetra-span proteins (membrane proteins with 4 membrane-traversing regions). Thus in vertebrates, some of the molecular antecedents of myelin seem to have been identified. Anomalously, however, homologues of another highly-important class of proteins, the so-called "non-tetraspan" proteins, appear to be completely absent (Gould et al 2005). Among these are those believed responsible for the tight binding of adjacent membrane lamellae in compact myelin. Homologues for these, especially one forming one of the major protein components of peripheral vertebrate myelin, myelin basic protein (MBP), have failed to turn up in any non-vertebrate groups so far examined. Intriguingly, genes homologous to MBP are found in the adaptive immune system (AIS), another gnathostome invention (see e.g.Klein and and Nikolaidis, 2005).
As with many valuable evolutionary innovations, complicated though they may be and difficult to assemble with all parts functioning correctly, selective pressures have repeatedly "reinvented the wheel" of myelin in several different groups. Knowing that it can indeed be done, we may wonder why this has not occurred in other highly successful groups such as molluscs and insects. However that may be, because invertebrates (especially copepods) are so numerous, it remains a fact that there are more myelinated INVERTEBRATES on this planet than myelinated VERTEBRATES!
As described above, the myelin insulation forces the current generated by a nerve impulse to spread farther and faster down the center of the axon to the next node, where a new impulse is set up. This is true regardless of which species is being considered. However, the problems of making the insulation electrically impervious to current leak that would compromise sheath efectiveness does differ somewhat from species to species. Species with spirally-wrapped myelin such as vertebrates and annelids must prevent curent escape along a spiral path between glial cell margins, as shown in the figure below on the left. This path can be closed by compacting these faces. The problem for concentric myelin is similar, but it can be solved by either compacting adjacent membrane faces or by sealing the connecting paths between layers where glial margins meet (2nd figure from the left). At nodes, all glially-derived myelin appears to need special junctions termed "septate junctions" betwen the margins of the glial sheets and the axon, as shown in the third figire below. Finally, in copepod myelin, the compacting of the myelin around the node itself appears to be sufficient to prevent curent leak (figure on the right).