The Myelin Sheath

1. Myelin Geometry

Most myelinated CNS axons are very thin. The critical diameter for myelination in the mammalian CNS is about 0.2 |im, whereas PNS axons myelinate at a diameter of approximately 1 |im (Waxman and Bennett, 1972). Both the radial and longitudinal dimensions of CNS myelin sheaths are coupled to axon diameter. The number of myelin lamellae is related to axon diameter according to a curvilinear function that varies among species (Fig. 4) (Hildebrand and Hahn, 1978). A mature CNS sheath may have up to 160 compacted lamellae. Conduction velocity is maximized when the g ratio (ratio axon diameter/fiber diameter [d/D]) is approximately 0.6, and this ratio is found in most myelinated mammalian CNS fibers (Waxman and Bennett, 1972). Since a single oligodendrocyte may produce sheaths with different numbers of myelin lamellae, the thickness of a myelin sheath seems to be regulated locally by the axon (Friedrich and Mugnaini, 1983; Waxman and Sims, 1984). One and the same oligodendrocyte can myelinate fibers in different tracts (Sternberger et al., 1978), and one cell can produce myelin sheaths spiralling in different directions (Remahl and Hildebrand 1990; Waxman and Sims, 1984).

As in the PNS, the internodal length (L) of myelinated CNS axons increases with D, but CNS axons show a shorter and more variable L than PNS axons (Fig. 5). Older data from various species have been reviewed elsewhere (Blakemore, 1981; Hildebrand et al., 1993). Murray and

2 4 6 8 10 12 14 16 2 4 6 8 10 12 54 16 2 4 6 8 10 12 14 16 2 4 6 8 10 12 14 18

A AXON DIAMETER B AXON DIAMETER C AXON DIAMETER D AXON DIAMETER

2 4 6 8 10 12 14 16 2 4 6 8 10 12 54 16 2 4 6 8 10 12 14 16 2 4 6 8 10 12 14 18

A AXON DIAMETER B AXON DIAMETER C AXON DIAMETER D AXON DIAMETER

2 4 S 6 10 12 14 16 2 4 6 8 10 12 14 16 2 4 6 6 10 12 !4 16 18 20 22

E AXON DIAMETER pm F AXON DIAMETER G AXON DIAMETER ■Um

2 4 S 6 10 12 14 16 2 4 6 8 10 12 14 16 2 4 6 6 10 12 !4 16 18 20 22

E AXON DIAMETER pm F AXON DIAMETER G AXON DIAMETER ■Um

Figure 4 The relation between axon diameter and number of myelin lamellae in spinal cord white matter of various species; (A) cat, (B) rabbit, (C) rat, (D) guinea pig, (E) mouse, (F) frog, and (G) perch. (From Hildebrand and Hahn, 1978.) Electron micrograph showing a transversely sectioned thin myelinated axon with eight myelin lamellae in the canine spinal cord (H). The outer loop (arrow), the inner loop, the spiral nature of the myelin sheath, the alternating major and minor dense lines and the axonal cytoskeleton are apparent. Scale bar = 0.1 |m. (With permission from Raine, 1984a.)

Figure 4 The relation between axon diameter and number of myelin lamellae in spinal cord white matter of various species; (A) cat, (B) rabbit, (C) rat, (D) guinea pig, (E) mouse, (F) frog, and (G) perch. (From Hildebrand and Hahn, 1978.) Electron micrograph showing a transversely sectioned thin myelinated axon with eight myelin lamellae in the canine spinal cord (H). The outer loop (arrow), the inner loop, the spiral nature of the myelin sheath, the alternating major and minor dense lines and the axonal cytoskeleton are apparent. Scale bar = 0.1 |m. (With permission from Raine, 1984a.)

7 8 9 10 11 12 13 14 15 16 17 18 1920)1 Fibre diameter

7 8 9 10 11 12 13 14 15 16 17 18 1920)1 Fibre diameter

800 700 600 500 400 300 200 100 0

y = 30.475x+ 117.52

.. R2 = 0.2852

• #

• * »

•• #

•• # i

m

# ; # •

la^l^^^ M a •

SHEATH DIAMETER (|m)

10 12

SHEATH DIAMETER (|m)

10 12

Figure 5 Regression lines describing the relation between internodal length (L) and fiber diameter (D) in teased preparations of cat spinal cord white matter, as described by different investigators (A). The line indicated by A refers to measurements of distances between intact nodes. The line indicated by D shows the observed relation when internodes bordered by broken nodes (heminodes) were also measured. The latter approach allowed observation of longer large-diameter internodes. (Modified from Murray and Blakemore, 1980, with permission.) (B) Scatterplot illustrating the relation L/D observed in myelinated axons in whole-mounts of the rat anterior medullary velum. Note that this regression line has a smaller slope than those shown in the top graph. (With permission from Ibrahim. et al. 1995.)

Blakemore (1980) noted that cat spinal cord fibers with a diameter of 3 jm have 300 jm long sheaths (range 100-600 jm), while fibers with a diameter of 20 jm possess 1,500 jm long internodes (range 1,100-2,000 jm; Fig. 5A). After dye injection of oligodendrocytes in the optic nerve of young rats, Butt and Ransom (1989) found 150 to 200 jm long myelin segments. In the anterior medullary velum (AMV) of the adult rat L is poorly correlated to D compared to spinal cord fibers. In the AMV the total range of D is 0.4 to 12 jm and L ranges between 50 and 750 jm. For thin fibers (D <1 jm) L varies around approximately 100 jm. For thick fibers (D = 9-10 jm) L varies around approximately 300 jam (Fig. 5B). The different relation L/D in the AMV compared to the spinal cord may be due to a different developmental length growth after myelination (Ibrahim et al., 1995; Butt et al., 1998a).

2. Myelin Fine Structure

Like PNS myelin, CNS myelin exhibits compact and cytoplasmic domains. In EM, the compact domains show alternating electron-dense and electron-lucent layers (Figs. 4H, 6 A, B). The major dense line, or period line, forms where the cytoplasmic surfaces of the plasma membrane of the spiralling oligodendroglial process appose. The fused outer leaflets facing the extracellular space form the intrape-riod line, or minor dense line, which sometimes is split (Figs. 4H, 6D; Hirano and Dembitzer, 1978). One major PNS/CNS difference is that the myelin repeating period (i.e., the distance between the midpoints of two successive major dense lines) is about 16 nm in fresh CNS myelin, but 18 nm in fresh PNS myelin (see Kirschner and Blaurock, 1992). In addition, the average CNS myelin period, but not the PNS counterpart, shrinks differentially during preparation for EM, so that it is 9 nm in thick sheaths, but 11 nm in thin sheaths (Hildebrand, 1972; Hildebrand and Muller, 1974) (Fig. 6D, E). This difference seems to reflect different effects of lipid solvents on thick and thin sheaths during dehydration and embedding. Another PNS/CNS dissimilarity is that myelinated CNS fibers lack a delimiting basal lamina. Since the outer surfaces of adjoining sheaths may appose each other directly, a minor dense line may sometimes form (Figs. 4H, 6B) (Peters et al., 1991).

In cross-sections the string-shaped cytoplasmic domains of a CNS myelin sheath show up as inner and outer loops (Fig. 4H). In thin myelinated CNS fibers, both loops are very small; but in large CNS fibers, and in all PNS fibers (see Peters et al., 1991), the outer loop may include the myelinating glial cell soma (Hildebrand, 1971). The other cytoplasmic domains in CNS sheaths—lateral or paranodal loops and incisures of Schmidt and Lanterman—are best seen in longitudinal sections. The paranodal loops represent expansions of the major dense line at the nodal end of each myelin lamella, which attach to the paranodal axon (Fig. 7B, C). Expansions of the major dense line in internodal parts of the myelin sheath stacked in sequence on top of each other form incisures of Schmidt and Lanterman. Although it has been stated that incisures are lacking in the CNS, many investigators (see Blakemore, 1969; Hildebrand et al., 1993) have noted Schmidt-Lanterman incisures in thick spinal cord fibers. All the cytoplasmic

Figure 6 (A, B) Electron micrographs showing details of thin (left) and thick (right) myelin sheaths in cat white matter. Note the difference in period. (C) Aberrant uniform lamellar pattern that can be found in myelinoid bodies. (x113,000 [A] and x124,000 [B, C]). (With permission from Hildebrand, 1971.) (D) Densitometric graphs made from high-magnification EM plates showing myelinated cat CNS fibers. The left graph shows the lamellar pattern in a thin sheath, and the right graph shows the pattern in a thick sheath. Arrows indicate deflections corresponding to major dense lines. (E) Mean myelin period in myelin sheaths with 0-10 (A), 11-20 (B), 21-30 (C), 31-40 (D), 41-50 (E), 51-60 (F), and >60 lamellae (G). (With permission from Hildebrand, 1972.)

Figure 6 (A, B) Electron micrographs showing details of thin (left) and thick (right) myelin sheaths in cat white matter. Note the difference in period. (C) Aberrant uniform lamellar pattern that can be found in myelinoid bodies. (x113,000 [A] and x124,000 [B, C]). (With permission from Hildebrand, 1971.) (D) Densitometric graphs made from high-magnification EM plates showing myelinated cat CNS fibers. The left graph shows the lamellar pattern in a thin sheath, and the right graph shows the pattern in a thick sheath. Arrows indicate deflections corresponding to major dense lines. (E) Mean myelin period in myelin sheaths with 0-10 (A), 11-20 (B), 21-30 (C), 31-40 (D), 41-50 (E), 51-60 (F), and >60 lamellae (G). (With permission from Hildebrand, 1972.)

strings contain microtubules. Longitudinally oriented tight junctions anchor the inner and outer loops to the myelin and seal the intramyelinic space. Paranodal and incisural junctions follow a transverse course (Raine, 1984a; Rosenbluth, 1984). In cross-cut myelinated CNS axons, the outer and inner loops tend to be located in the same quadrant. In the sector between the outer and inner loops CNS myelin may show a "radial component" (i.e., longitudinally oriented junction strands at the extracellular apposition—an intramyelinic tight junction) (Fig. 7A; Peters, 1964; Dermietzel, 1974; Dermietzel et al., 1980; Shinowara et al., 1980; Kosaras and Kirschner, 1990).

The unrolled myelin sheath has a trapezoidal shape (Fig. 8). Compact myelin forms most of the trapezoid. The outer, inner, and paranodal loops form a continuous cytoplasmic string along the edge of the compact myelin plate.

Figure 7 Electron micrographs showing details from myelinated CNS fibers. (A) Mouse white matter. Note radial component (arrows) below outer tongue (*). The radial component is composed of radially arranged tight junctions seen here as light zones where the minor dense line is narrowed. Scale bar = 0.1 |m. (With permission from Nagara and Suzuki, 1982.) (B) Node-paranode region of myelinated fiber in kitten spinal cord. Each myelin lamella terminates as a lateral loop apposing the axolemma. Arrows indicate nodal axolemma. Scale bar = 1 | m. (C) Paranode in mature small myelinated CNS fiber. Note the presence of regularly spaced densities (arrows), the so-called transverse bands, between the lateral loops and the axolemma. Scale bar = 0.2 |m. (With permission from Raine, 1984a.)

Figure 7 Electron micrographs showing details from myelinated CNS fibers. (A) Mouse white matter. Note radial component (arrows) below outer tongue (*). The radial component is composed of radially arranged tight junctions seen here as light zones where the minor dense line is narrowed. Scale bar = 0.1 |m. (With permission from Nagara and Suzuki, 1982.) (B) Node-paranode region of myelinated fiber in kitten spinal cord. Each myelin lamella terminates as a lateral loop apposing the axolemma. Arrows indicate nodal axolemma. Scale bar = 1 | m. (C) Paranode in mature small myelinated CNS fiber. Note the presence of regularly spaced densities (arrows), the so-called transverse bands, between the lateral loops and the axolemma. Scale bar = 0.2 |m. (With permission from Raine, 1984a.)

In large myelinated CNS fibers, the compact myelin plate is subdivided into subplates by cytoplasmic strings forming the incisures in the rolled sheath.

3. The Myelin-Oligodendrocyte Unit

When Del Rio Hortega studied oligodendrocytes in the early 1900s he recognized four subtypes (Types I-IV) without sharp boundaries, different types being associated to fibers of different diameters (Fig. 9) (reviewed by Bunge, 1968; Friedrich et al., 1980; Wood and Bunge, 1984; Hildebrand et al., 1993; Szuchet, 1995). This view was confirmed by Penfield (1932) and has been supplemented by EM data from the toad (Stensaas and Stensaas, 1968), the cat (Remahl and Hildebrand, 1990), and the rat (Bjartmar et al., 1994a), as well as by light microscopic observations on dye-injected rat oligodendrocytes (Butt et al., 1994a, 1998a, 1998b; Berry et al., 1995; Weruaga-Prieto et al., 1996). The Type I oligodendrocyte emits many thin radially directed processes, and Type II has fewer and thicker processes. Direct observations on stained myelin-oligoden-drocyte units show that rodent optic nerve oligodendrocytes are Type I or II (Butt et al., 1994a). Rat and mouse optic nerve oligodendrocytes are linked to 15 to 20 internodes via 15- to 30-||m long processes (Fig. 10; Butt and Ransom, 1989, 1993; Butt et al., 1994a, 1994c). In the kitten corpus callosum Remahl and Hildebrand (1990) observed Type I/II cells connected to 4 to 11 myelin sheaths. The rat AMV contains thin myelinated axons linked to Type I/II cells as well as thick myelinated trochlear fibers (d = 4-15 |im), supplied by Type III or Type IV oligodendrocytes. In the AMV Type I/II, oligodendrocytes extend thin branching processes long distances (up to over 100 |im) to 5 to 18 fibers with a mean diameter of 1.25 | m. Type III cells are bipolar or tripolar and send processes to two to five fibers with a mean diameter of 3.5 |im (Fig. 11A). The Type IV unit has a Schwann cell-like anatomy, being associated with a single large myelinated fiber with a mean diameter of 6.2 |im (Fig. 11B; see Berry et al., 1995; Butt et al., 1998a, 1998b).

Whether immature oligodendrocytes are intrinsically committed to develop into a particular subtype or induced to develop into a specific phenotype by the axon or other external factors is unknown. Oligodendrocytes exhibit an axon-independent differentiation in vitro, ending up in process elaboration and production of myelin-like sheets (see Szuchet, 1995). Since oligodendrocytes from immature rat spinal cord and cerebrum, respectively, develop different process morphologies in vitro, in the absence of neurons, it was concluded that they follow dissimilar intrinsic programs (Bjartmar, 1998). On the other hand, Fanarraga and coworkers (1998) concluded from transplantation experiments that oligodendrocytes develop a Type I-IV anatomy depending on which axon they contact.

The amount of myelin produced by an oligodendrocyte varies strongly, depending on its subtype. Blakemore (1981) calculated that a single large CNS internode contains 400

Figure 8 Type I/II oligodendroglial cell body (top center) attached to 14 myelin sheaths, two of which have been unrolled to varying degrees to show the arrangement of the cytoplasmic strings in the sheet of myelin. (With permission from Morell and Norton, 1980.)

times as much myelin as a single small internode. He also found that the volume of myelin supported by an oligodendrocyte providing one 11.4-||m thick axon with a 1,250-||m long myelin sheath is 10 times greater than the volume supported by a cell providing 39 thin (d = 1.5 |im) axons with 200-|im long sheaths. According to Butt et al. (1998a), the mean myelin volume amounts to 600 |im3 in Type I/II units and 30,000 | m3 in Type III/IV units in the AMV. If so, a Type III/IV cell maintains 50 times more myelin than a Type I/II cell. A third calculation (Anderson, 2003) indicates that a Type IV oligodendrocyte providing a 16-|im thick axon with a single thick long sheath maintains 100 times more myelin than a Type I cell with 50 thin short sheaths along 0.5-| m thick axons. These calculations show that Type IV cells produce and maintain a greater volume of myelin than Type I cells.

In addition to the occurrence of anatomical subtypes, adult oligodendroglial cells show a molecular heterogeneity. The chemical composition of bovine white matter areas con taining thick and thin myelinated fibers differs from the composition of areas with thin fibers only (Amaducci et al., 1962), and the chemical composition of spinal cord myelin differs from the composition of brain myelin (Norton and Camner, 1984). Further, myelin in rat white matter areas with both thin and thick fibers is more strongly myelin basic protein (MBP) immunoreactive than myelin in areas with thin fibers only. The reverse is valid for proteolipid protein (PLP) immunoreactivity (Hartman et al., 1982). More recently, Butt et al. (1995) showed that carbonic anhydrase II is present in Type I-II rat AMV oligodendrocytes, but not in Type III-IV oligodendrocytes. Moreover, the large isoform of myelin associated glycoprotein is present in all oligodendrocytes, but the small isoform is only present in Type III-IV cells (Butt et al., 1998b). Since various demyeli-nating toxic agents cause greater demyelination where carbonic anhydrase II is low than where it is high, presence or absence of this enzyme may be a key factor in the pathology of white matter (see Butt et al., 1995). Further, it seems pos-

Figure 9 As described by Del Rio Hortega (1928), the oligodendroglial population includes four varieties, Types I-IV in terms of number of processes and branching pattern. (With permission from Wood and Bunge, 1984.)

sible that a molecular heterogeneity of the type described might underlie the fact that tracts myelinating late (Type I/II cells) and early (Type III/IV cells), respectively, are differentially affected by some inborn errors of myelin metabolism, such as Krabbe's disease (Suzuki and Suzuki, 1983).

The cytoplasmic strings of the myelin sheath are used for distribution of myelin molecules. MBP and PLP are major components of CNS myelin. It turns out that MBP is synthesized near its site of assembly in the myelin sheath. This is accomplished by a trafficking of MBP mRNA along the string-related microtubules from the nucleus to the myelin compartment (Barbarese et al., 1999). Indeed, EM examination of oligodendrocytes in vitro and in situ reveals the presence of ribosome clusters in distal oligodendroglial processes (Waxman and Sims, 1984; Barry et al., 1996). As to the biological value of this RNA trafficking, various hypotheses have been discussed (see Barbarese et al., 1999). PLP, on the other hand, is synthesized on endoplasmic retic-ulum-bound ribosomes in the oligodendroglial cell body, shifted to the Golgi system via microtubules, and transferred to the myelin via transport vesicles budding off from transitional ER elements. In the mutant taiep rat, a microtubular defect causes a deficient incorporation of PLP into myelin through a blockage of membrane movement from the ER to the Golgi system, which causes abnormal myelination and progressive demyelination of CNS axons (Couve et al.,

Figure 10 Camera lucida drawings showing an HRP-filled Type I oligodendrocyte in young rat optic nerve. (A) The whole cell. Scale bar = 100 ^m. (B) Drawing at a higher magnification (x1,250) showing the cell body and three selected longitudinal processes. Scale bar = 100 ^m. (C) Schematic interpretation of the parts of this cell seen in (B). (D) The picture shown in (C) would look like this if the myelin sheaths could be unwrapped. Each sheath has been transformed into a trapezoidal sheet bordered by strings of cytoplasm corresponding to the outer, the inner, and the lateral loop (a = axon, etp = outer loop, pl = lateral loop, itp = inner loop, ocb = oligodendroglial cell body, cp = connecting process). (Modified from Butt and Ransom, 1989, with permission.) (E) Camera lucida drawing showing a Type II oligodendrocyte in the adult rat medullary velum visualized by application of the antibody Rip to a velum whole mount. Diagram on the right side shows the relation L/D for the fibers in this unit. Scale bar = 50 ^m. (With permission from Butt et al., 1998a.)

Figure 10 Camera lucida drawings showing an HRP-filled Type I oligodendrocyte in young rat optic nerve. (A) The whole cell. Scale bar = 100 ^m. (B) Drawing at a higher magnification (x1,250) showing the cell body and three selected longitudinal processes. Scale bar = 100 ^m. (C) Schematic interpretation of the parts of this cell seen in (B). (D) The picture shown in (C) would look like this if the myelin sheaths could be unwrapped. Each sheath has been transformed into a trapezoidal sheet bordered by strings of cytoplasm corresponding to the outer, the inner, and the lateral loop (a = axon, etp = outer loop, pl = lateral loop, itp = inner loop, ocb = oligodendroglial cell body, cp = connecting process). (Modified from Butt and Ransom, 1989, with permission.) (E) Camera lucida drawing showing a Type II oligodendrocyte in the adult rat medullary velum visualized by application of the antibody Rip to a velum whole mount. Diagram on the right side shows the relation L/D for the fibers in this unit. Scale bar = 50 ^m. (With permission from Butt et al., 1998a.)

1997). The microtubule defect is present in all oligodendrocytes, but the myelination problem affects mainly thin axons (Lunn et al., 1997; Song et al., 2001). Hence, the logistic functions mediated by the microtubular system might be more important in oligodendrocytes myelinating several thin axons than in cells with few myelin sheaths or a single one.

0 0

Post a comment