Axonal Swellings and Degeneration in Mice Lacking the Major Proteolipid of Myelin

See allHide authors and affiliations

Science  05 Jun 1998:
Vol. 280, Issue 5369, pp. 1610-1613
DOI: 10.1126/science.280.5369.1610


Glial cells produce myelin and contribute to axonal morphology in the nervous system. Two myelin membrane proteolipids, PLP and DM20, were shown to be essential for the integrity of myelinated axons. In the absence of PLP-DM20, mice assembled compact myelin sheaths but subsequently developed widespread axonal swellings and degeneration, associated predominantly with small-caliber nerve fibers. Similar swellings were absent in dysmyelinated shiverer mice, which lack myelin basic protein (MBP), but recurred in MBP*PLP double mutants. Thus, fiber degeneration, which was probably secondary to impaired axonal transport, could indicate that myelinated axons require local oligodendroglial support.

Proteolipid protein (PLP) is a four-helix–spanning membrane protein thought to stabilize the ultrastructure of central nervous system (CNS) myelin by forming the double-spaced intraperiod line (IPL), but neither PLP nor its splice isoform DM20 is required for spiral membrane wrapping and myelin compaction (1). Mutations of the X-linked PLP gene (2) cause PelizaeusMerzbacher disease (PMD) and spastic paraplegia–2 (SPG-2) in humans and related disorders in animal models, such as the jimpy (jp) mouse, characterized by premature death of oligodendrocytes and dysmyelination. However, the severe consequences of spontaneous PLP mutations are explained, at least in part, by the toxicity of the encoded polypeptide when misfolded (3).

PLP-DM20–deficient mice containing a null allele develop normally and assemble compacted myelin of appropriate thickness enwrapping small- and large-caliber axons (1). Although the IPL of myelin in these mutants is frequently condensed, oligodendrocytes are morphologically normal. Unexpectedly, from the age of 6 to 8 weeks, increasing numbers of focal axonal swellings containing organelles were detected throughout the white and gray matter in all regions of the CNS, particularly in areas where small-diameter myelinated axons predominate, such as in optic nerves (4). Most spheroids contained numerous dense bodies, multivesicular bodies, and mitochondria (Fig.1, A and B), often located at the presumptive distal paranode (Fig. 1B). The myelin sheath was preserved over small accumulations of organelles, but became focally attenuated and eventually lost through slippage (5) as swellings enlarged (Fig. 1, C and D). The normal periaxonal space was maintained as predicted by the presence of myelin-associated glycoprotein (1). By 1 year of age, numerous large axonal swellings were present in the optic nerve and spinal cord (Fig. 1, C and D). A smaller proportion of swellings, such as those of Purkinje cell axons (“torpedoes”), were rich in phosphorylated neurofilaments (NFs) (Fig.2A), but failed to stain with antibodies recognizing dephosphorylated NF epitopes (6). There was no apparent increase in the number of degenerating fibers in the distal compared with the proximal optic nerve, suggesting that the presence of axonal spheroids was not followed immediately by fiber breakdown. At this time, PLP-deficient mice were normal when tested for motor performance on a rotarod (1). In animals older than 1 year, axonal degeneration was more prominent in the optic nerve and fasciculus gracilis (Fig. 2B), indicating a slow, progressive loss of fibers with age, although neuronal cell loss was not apparent. Increased numbers of microglia and a mild astrocytosis accompanied these degenerative changes (7). Occasional inner-tongue processes of oligodendrocytes contained degenerate organelles. No swellings were detected in wild-type (WT) littermates or in 129/Sv mice, the strain from which the embryonic stem cells were derived.

Figure 1

(left).Axonal spheroids containing membranous organelles in (A) transverse section of the spinal cord and (B) longitudinal section of the optic nerve from a 4-month-oldPlp−/Y mouse. Secondary attenuation of the myelin sheath as a result of slipping is apparent in (A). Axonal swellings often commenced in the paranodal region (B) adjacent to the node of Ranvier (N). (C and D) Large spheroids in the optic nerve of a 1-year-old Plp−/Y mouse. The swellings in (C) contain predominantly neurofilaments with some organelles, whereas those in (D) contain mainly membranous organelles. As swellings enlarge, the myelin sheath retracts and is eventually lost over the swollen region of the axon [in (C)].(E) MBP*PLP-deficient double mutants display spheroids identical to those of PLP-deficient mice. All axons are dysmyelinated and many retain an association with ensheathing oligodendrocyte processes. No such swellings were detected in MBP-deficient (shiverer) mice. Bars: 1 μm (A, B, and E); 2 μm (C and D).

Figure 2

(right).(A) Cerebellar cortex from a 4-month-oldPlp−/Y mouse immunostained with antibody RM024 recognizing phosphorylated NF (12) revealing swellings (filled arrows) in Purkinje cell axons. The cell body of one Purkinje cell (open arrow) contains phosphorylated NFs, suggesting that its axon has degenerated. (B) Degenerating (filled arrow) or swollen (open arrow) small-caliber fibers are apparent in the fasciculus gracilis of a 22-month-old Plp−/Y mouse. The density of myelinated axons is decreased, indicating axonal loss. In contrast (C), degeneration is not apparent and myelin sheaths are intact in the ventral columns of spinal cord from the same mouse. (D) Proximal region of optic nerve from a 22-month-old Plp−/Y mouse. Low-power micrograph shows the retina and optic disc (O), the area of the lamina cribrosa (between the retina and the large arrow), and the more distal myelinated region of the nerve (to the right of the large arrow). Some large axonal swellings are visible (small arrow). (E) Unmyelinated axon from lamina cribrosa area of the optic nerve in (D). The axon, surrounded by astrocyte processes, is normal. (F) A myelinated axon at a level just to the right of the large arrow in (D) shows accumulation of organelles at the distal paranode and node (N). The retinal end is to the left. Bars (E and F): 1 μm.

At older ages, the motor performance of mutant mice appeared altered when the animals' gait was unusually slow and showed signs of spasticity. To quantify these observations, we performed a rotarod test (8) and studied, in a follow-up, a cohort of mutant and WT mice between 6 and 16 months of age. We observed a sudden impairment of motor functions in 16-month-old mice (Fig. 3) without any corresponding evidence of demyelination, for up to at least 22 months (Fig. 2C). Thus, the behavioral phenotype was most likely associated with a threshold effect in the progressive degeneration of CNS axons.

Figure 3

Loss of motor control in aged PLP-deficient mice. (A to C) A cohort of Plp−/Ymice (n = 6) and age-matched controls of the same sex (n = 6) were tested on the rotarod at 6, 12, and 16 months of age (8). The holding time of 16-month-old mutant mice (n = 5, one animal died) was decreased compared with WT littermates, which is not apparent at younger ages (1). The motor performance of WT mice was not significantly different between age groups. Error bars are standard deviations.

Whereas a null allele of the human PLP gene is associated with a mild peripheral neuropathy (9), our PLP-deficient mice showed no peripheral involvement, even at 22 months of age. Specifically, axons of dorsal root ganglia neurons (which project centrally through the affected fasciculus gracilis) appeared normal. This, and the absence of neurogenic atrophy in skeletal muscle, argues against a major peripheral component of the decreased motor performance.

To demonstrate that the absence of proteolipids was the responsible defect, we complemented the mutants (Plp−/Y) with a WT PLP genomic transgene (10) expressing both PLP and DM20 at nearly normal levels (line tg72). The presence of this transgene completely prevented the axonal pathology inPlp−/Y*tg72 offspring (Fig.4). Next, we examined the ability of each isoform, PLP and DM20, to rescue the null mutant. This was achieved by two PLP and DM20 cDNA-based transgenes under control of the human PLP promoter [expressing 70 and 50% of the normal RNA level (11)]. The presence of PLP or DM20 in myelin reduced the axonal changes by 84 and 60%, respectively (Fig. 4), suggesting that both PLP isoforms serve, in principle, the same function.

Figure 4

Quantitation of axonal changes in the optic nerve of PLP/DM20-deficient mice and rescue by transgenic complementation. The combined number of CNS axons, showing abnormal swellings or degeneration, was quantified in cross sections of the mid-optic nerve. In each experiment, the relative proportion of axonal swellings and degeneration profiles was roughly equal (not shown). Complete transgenic complementation was achieved with an autosomal PLP/DM20 genomic transgene (tg72) (10). A partial rescue was observed with both a PLP cDNA (Plp-tg) and a DM20 cDNA (Dm20-tg) construct, driven by the human PLP promoter (11). Also shown are the changes in female heterozygous mice (Plp+/−), whose optic nerves contain a mosaic of PLP+ and PLP− myelin sheaths. Each bar represents the mean (±SEM) of four mice aged 4 months, with the exception of the heterozygotes (n = three mice).

Myelin-forming glial cells modify axonal morphology (12), but the molecular mechanisms are not known. Schwann cell dysmyelination, caused by mutations of the Pmp22 and P0 genes, has also been associated with perturbations of the axonal cytoskeleton and fiber degeneration (13). However, the major changes in myelin volume and composition make it difficult to evaluate the contribution of individual myelin components. This is markedly different in Plp−/Y mice, in which oligodendrocytes elaborate and maintain appropriately thick myelin sheaths. We examined 4-month-old shiverer (shi) mice, which lack myelin basic protein (MBP), the second most abundant CNS myelin protein (14). Although the CNS of shi/shi mice is poorly myelinated and myelin sheaths lack compaction, no similar spheroids were detected. In contrast, shi/shi*Plp−/Y double mutants were severely hypomyelinated, but in addition contained the axonal spheroids (Fig. 1E). Thus, axonal swellings appeared to be related to the absence of PLP-DM20 and not to morphologically abnormal myelin.

Because we never observed axonal pathology in normal nonmyelinated fibers, we hypothesized that glial ensheathment provides a necessary signal that induces the axonal dependency on glial support. To distinguish between possible signaling mechanisms that could be either localized (glia to axon) or long-range (glia to neuron), we studied the optic nerve in mouse chimeras. Because of random X chromosome inactivation, both jp/+ and Plp+/−females are mosaics for Plp gene expression so that axonal internodes have an equal probability of myelination by WT or mutant oligodendrocytes. In jp/+ mice, mutant oligodendrocytes fail to myelinate, leaving patches of naked axons (15). In contrast, all optic nerve axons in Plp+/− females are normally enwrapped, with about 50% of internodes surrounded by PLP-deficient myelin (1). In adultPlp+/− chimeras, we identified all the morphological signs of axonal swelling and degeneration, although less frequently than inPlp−/Y mice (Fig. 4). Thus, optic nerve axons had become dependent on oligodendrocyte function along their entire length, and WT oligodendrocytes were unable to support adjacent axonal segments enwrapped by PLP-deficient myelin. Oligodendrocytes therefore serve a localized function. In contrast, the nonmyelinated patches of adultjp/+ mice showed no axonal swellings. Thus, unmyelinated axonal segments of the optic nerve were not oligodendrocyte dependent, a conclusion supported by the absence of swellings in the normally unmyelinated proximal regions in Plp−/Y mice (Fig. 2, D to F). Taken together, the glial-to-axon signaling requires intimate cellular interaction and is also localized. The nature of the signaling molecule is currently unknown, but because PLP-DM20–deficient oligodendrocytes generated this signal after myelination, it cannot be PLP or DM20 itself.

The normal development of myelin in PLP-deficient mice followed by progressive axonal degeneration was unexpected. The axonal defect is best explained if we assume a deficiency of an oligodendroglial function that relates to the maintainance of axonal integrity. The accumulation of membranous dense bodies and mitochondria at distal paranodes strongly suggests an impairment of retrograde axonal transport (16), and the NF swellings may indicate that slow anterograde transport is also compromised, but both will require direct confirmation. It is likely that the complete breakdown of axonal transport at sites of swellings progresses to axonal degeneration before other mechanisms, such as neuronal target deprivation, produce this effect.

Although a supportive function of oligodendrocytes is the most likely explanation of the phenotype under study, two related explanations cannot be formally excluded: (i) The absence of PLP-DM20 from myelin could provide a hostile environment in which axons alter their behavior; (ii) PLP-deficient oligodendrocytes could physically constrict small-caliber axons, although axon “crushes” have never been observed. Indeed, frequency distributions of internodal axonal diameter in optic nerve were normal in 4-month-old Plp−/Ymice (7).

The molecular basis of glial-axonal communication is poorly understood. In invertebrates, a direct vesicular transfer of molecules from glia to axon has been described and may also occur in vertebrates (17). PLP-DM20 is localized in both compact and adaxonal myelin, and the proposed “channel-like” functions of proteolipids (18) could be involved in glial-axonal communication. The lack of connexin-32 from peripheral nervous system myelin underlies a peripheral neuropathy with axonal involvement (19).

The low frequency of axonal spheroids in the few myelinated nerve segments of PLP mutants (20), such asjimpy, is probably related to their early death. CNS axonal swellings are also uncommon in adult rumpshaker mice (7), presumably because their myelin sheaths incorporate PLP-DM20 (21). The predilection for small-caliber axons is striking; for example, those of the fasciculus gracilis (diameter, 0.96 ± 0.04 μm, mean ± SEM) are affected, but the contiguous larger axons of the fasciculus cuneatus (2.39 ± 0.18 μm) appear, at least morphologically, intact. The reason (or reasons) for the predilection of small-caliber axons is presently unclear.

We suggest a model for neuron-glia interactions in which CNS axons become dependent on oligodendrocyte support some time after myelination has been completed. Accordingly, oligodendrocytes may provide both myelin for rapid impulse conduction and long-term support for axons, a function for which proteolipids and presumably other proteins are required. Whether this function involves the putative channel-like property of PLP (18) awaits further investigation. The principle finding of an oligodendrocyte-dependent axonopathy is relevant to human demyelinating diseases, including multiple sclerosis, where axonal degeneration is now recognized as a major cause of persistent disability (22).

  • * To whom correspondence should be addressed. E-mail: nave{at}


Stay Connected to Science

Navigate This Article