Mitochondrial immobilization mediated by syntaphilin facilitates survival of demyelinated axons

Results

The purpose of this study is to compare mitochondrial distribution/shape and axonal pathology in WT and syntaphilin-deficient demyelinated axons. We first investigated whether the number of myelinated fibers and the distribution/size of mitochondria in myelinated axons were similar in WT and syntaphilin-null corpus callosa. We chose the corpus callosum because it is a common target of demyelination in multiple sclerosis brains (26) and is extensively demyelinated in cuprizone-fed mice. By using serial block face scanning EM, we quantified myelinated axons and compared the 3D size and distribution of individual mitochondria in myelinated and demyelinated corpus callosa from WT and syntaphilin-null mice (Fig. 1 A–G). The size and density of myelinated fibers (Fig. 1 A and C) was similar in WT and syntaphilin-deficient corpus callosa. Mitochondrial number per myelinated axonal volume (Fig. 1K) and mitochondrial volume per myelinated axonal volume (Fig. 1L) were similar in WT and syntaphilin-null corpus callosa; we used total axonal volume instead of axonal length as the denominator to allow for any differences in axon diameter among sample sets. Volumes (Fig. 1H) and lengths (Fig. 1I) of individual axonal mitochondria were also similar. Compared with WT myelinated axons, there was a slight but significant decrease in the diameter of mitochondria (Fig. 1J) in syntaphilin-null myelinated axons. Similar densities and volumes of axonal mitochondria and the absence of major neurological phenotypes (24) indicate that the mouse neuron/axon can compensate for loss of syntaphilin during the first four postnatal months.

We next investigated how WT and syntaphilin-null axons responded to demyelination induced by cuprizone/rapamycin treatment. Cuprizone is a toxin that kills oligodendrocytes (27). During a 6-wk cuprizone feeding paradigm, spontaneous remyelination of the corpus callosum occurs by 4 wk of cuprizone-only treatment. Cuprizone, however, does not kill these remyelinating oligodendrocytes. This remyelination can, however, be inhibited by daily concurrent injections of rapamycin (28). The combination of cuprizone feeding and daily rapamycin injections for 6 wk demyelinates the corpus callosum (Fig. 1B). All control mice received daily rapamycin injections in these experiments. Demyelination was extensive in WT and syntaphilin-null corpus callosa (Fig. 1C). In WT demyelinated axons, the volumes (Fig. 1H) and lengths (Fig. 1I) of individual mitochondria and mitochondrial number per axonal volume (Fig. 1K) were significantly increased, and mitochondrial volume per axonal volume (Fig. 1L) was doubled compared with myelinated WT axons. In contrast, neither the volume of individual mitochondria (Fig. 1H) nor mitochondrial numbers per axonal volume (Fig. 1K) were increased in demyelinated syntaphilin-null axons in which mitochondria were ∼50% shorter (Fig. 1I) and ∼50% wider (Fig. 1J) compared with myelinated syntaphilin-null axons. Thus, unlike WT demyelinated axons, syntaphilin-null axons did not increase mitochondrial volume per axonal volume following demyelination (Fig. 1L). High doses of cuprizone delivered to weanling mice have profound effects in hepatocytes, producing swollen “giant” mitochondria characterized by fragmented short cristae concentrated at the periphery of the swollen mitochondria (29, 30). Following 6 wk of cuprizone/rapamycin treatment, giant mitochondria with fragmented cristae were not found in WT (Fig. S1A) or syntaphilin-null axons (Fig. S1B), including regions with axonal swellings. Swollen axonal mitochondria with altered cristae were not detected in previous studies using similar cuprizone dosing paradigms (31, 32). Mitochondria in hepatocytes in our cuprizone/rapamycin-treated mice, however, were slightly swollen with altered cristae distributions (Fig. S1 C and D). This differential effect on hepatocyte mitochondria reflects the detoxifying role of liver. These observations demonstrate that the cuprizone treatment in the present study does not alter mitochondrial structure in CNS axons and indicate that syntaphilin is required for the increase in mitochondrial size and density that occurs after demyelination of CNS axons.

We next asked whether the lack of syntaphilin promotes axonal degeneration in the demyelinated corpus callosa. Degeneration of demyelinated axons was assessed by comparing the number of nonphosphorylated neurofilament-positive axonal ovoids in demyelinated corpus callosa from WT (Fig. 2A) and syntaphilin-null (Fig. 2B) mice. These ovoids represent the cut ends of axons that are still connected to viable neurons (2) or axonal swellings that precede axonal transection (33, 34). Axonal ovoids positive for nonphosphorylated neurofilament were not observed in myelinated corpus callosa of WT and syntaphilin-deficient mice. Axonal ovoids were significantly increased and twice as common in syntaphilin-null demyelinated corpus callosa compared with WT (Fig. 2C), and 70.0 ± 5.5% (n = 5 corpus callosa) of axonal ovoids contained detectable levels of the amyloid precursor protein (APP; Fig. 2 D–G), a marker for compromised axonal transport (3, 4, 35). Axonal ovoids were readily identified in EM images of corpus callosum (Fig. 2H) and contained densely packed axoplasmic organelles, including intact and fragmented mitochondria. Axonal ovoids were more frequent and approximately three times larger in EM images of demyelinated syntaphilin-KO mice compared with demyelinated WT mice. In EM analyses, axonal ovoids were not found in myelinated WT and syntaphilin-KO mice (Fig. S2). These results show that axonal ovoid formation is increased in demyelinated syntaphilin-deficient axons.

We next compared axonal DsRed2-labeled mitochondrial distribution and transport in myelinated and demyelinated living cerebellar slice cultures from WT and syntaphilin-deficient mice (Fig. 3A). In these slice cultures, the majority of axonal mitochondria were stationary, and the size of mitochondrial stationary sites was similar in WT and syntaphilin-deficient myelinated axons (Fig. 3B). The number of motile mitochondria, however, was increased in syntaphilin-deficient myelinated axons compared with myelinated WT axons (Fig. 3C and Fig. S3), but their mean velocity of mitochondrial transport was similar (Fig. 3C and Fig. S3). Mitochondrial stationary site size (Fig. 3B), the number of motile mitochondria (Fig. 3C), and the velocity of mitochondrial transport (Fig. 3D) were significantly increased in demyelinated WT axons compared with WT myelinated axons. In contrast, mitochondria stationary site size was similar in myelinated and demyelinated syntaphilin-deficient axons (Fig. 3B). The number of motile mitochondria (Fig. 3C) and the mean velocity of motile mitochondria translocation (Fig. 3D) were significantly decreased in demyelinated syntaphilin-deficient axons compared with myelinated syntaphilin-deficient axons.

To examine axonal pathology in the slice cultures, myelinated and demyelinated WT and syntaphilin-deficient cerebellar cultures were immunostained for neurofilaments and APP (3, 4, 35). Although the number of APP-positive axonal ovoids in myelinated WT slices was comparable to that in myelinated syntaphilin-deficient slices, demyelinated syntaphilin-deficient slices had significantly more APP-positive axonal ovoids and transections than demyelinated WT slices and myelinated syntaphilin-deficient slices (Fig. 4). Increased axonal pathology in demyelinated syntaphilin-deficient axons was also found in demyelinated corpus callosa of syntaphilin-KO mice (Fig. 2 and Fig. S2). Na⁺ channel blockers can reduce axonal pathology in conditions that compromise axonal mitochondrial energy production (36, 37). If reduced mitochondrial density in syntaphilin-deficient axons leads to insufficient energy production for axonal survival, blocking Na⁺-channel activities may be beneficial for the survival of demyelinated syntaphilin-deficient axons. Flecainide is a Na⁺ channel blocker that can reduce axonal loss following NO treatment in vitro and during inflammatory demyelination in vivo (37, 38). Indeed, when syntaphilin-deficient slice cultures were treated with flecainide, APP-positive axonal ovoids and transections were significantly decreased in a dose-dependent manner (Fig. 4G). These in vitro results collectively demonstrate that syntaphilin plays an important role in increasing the volume of axonal mitochondrial stationary sites upon demyelination, and provides further support to the hypothesis that increased mitochondrial stationary site size protects against degeneration of demyelinated axons, which is caused in part by the increased energy demands of nonsaltatory nerve conduction.

Discussion

The present study investigated mitochondrial behavior in acutely demyelinated CNS axons. We described changes in mitochondrial distribution in demyelinated axons in vivo by using 3D EM and visualized mitochondrial motility and distribution in living myelinated and demyelinated CNS axons in vitro. As we previously described for demyelinated PNS axons in vitro (17), CNS demyelinated axons compensate for loss of myelin by increasing (i) the volume of mitochondrial stationary sites and (ii) the number and velocity of motile mitochondrial transport. By using syntaphilin-deficient mice, we extend these fundamental mitochondrial responses shared by CNS and PNS demyelinated axons by establishing that (i) syntaphilin-mediated immobilization of stationary mitochondria is essential for the increased volume of mitochondrial stationary sites in demyelinated CNS axons, (ii) deleting axonal syntaphilin significantly increases degeneration of demyelinated CNS axons, and (iii) the incidence of this axonal degeneration can be reduced by blocking Na⁺ channel activity. Collectively, our studies indicate that acutely demyelinated axons increase the volume of mitochondrial stationary sites to meet the increased energy demands of nonsaltatory nerve conduction and thereby prevent or slow axonal degeneration.

Lack of significant behavioral or neurological phenotypes (24) and no change of mitochondrial density and volume in syntaphilin-null mice at 2–3 mo of age indicate that myelinated axons can compensate for the loss of syntaphilin during development and early postnatal life. The mechanism by which this compensation occurs is unknown. In contrast, we establish that syntaphilin-null axons cannot fully compensate for loss of myelin. Specifically, demyelinated syntaphilin-null axons do not increase the size of stationary mitochondria, but they contain more ovoids and they degenerated at a greater rate than demyelinated WT axons. The frequent presence of axonal ovoids positive for APP support that axonal ovoid formation in demyelinated syntaphilin-null axons accompanies impaired axonal transport (3, 4, 35). Mitochondrial biogenesis and/or redistribution in addition to an imbalance between mitochondrial fusion/fission are likely to be responsible for the increase of stationary mitochondria volume in demyelinated axons. Syntaphilin immunoreactivity is increased in demyelinated lesions in postmortem brains obtained from individuals with multiple sclerosis, and this increased staining is associated with mitochondria (14). A similar increase in syntaphilin-positive mitochondria was found in demyelinated axons in the present study (Fig. S4). Mitochondria in syntaphilin-null demyelinated axons were shorter and thicker than those in demyelinated WT axons. Syntaphilin appears to be required for the rapid increase in stationary mitochondrial volume that occurs through interactions with microtubules following demyelination. Further studies are necessary to address how syntaphilin is regulated in demyelinated axons and how syntaphilin-mediated mitochondrial immobilization indirectly affects mitochondrial fusion/fission to facilitate mitochondrial elongation.

An important question in the present study is whether cuprizone alters mitochondria function in demyelinated axons. To address this question, we used 3D EM to investigate whether mitochondrial pathologic conditions described in liver (29, 30) were present in demyelinated axons in cuprizone/rapamycin-treated mice. Consistent with previous studies that used similar concentrations of cuprizone (0.2–0.3%) to demyelinate CNS axons (31, 32), we did not detect swollen mitochondria with peripherally located cisternae in WT or syntaphilin-null demyelinated axons. The cuprizone doses used to cause giant mitochondria in liver were 60–300% greater than that used in the present study, and the mice were much younger and thus more susceptible to cuprizone toxicity (27, 39). In contrast to the CNS, we did detect abnormal mitochondria in the liver from our cuprizone/rapamycin-treated mice (Fig. S1). Hepatocyte mitochondria from cuprizone/rapamycin-treated mice were swollen and contained abnormally distributed cisternae compared with mitochondria in hepatocytes from mice that received rapamycin only (Fig. S1 D and E). Their diameter never exceeded 2 µm and never approached diameters reported in the literature (>10 µm diameter) from younger mice receiving higher cuprizone doses. The cuprizone dose used in the present study does not cause changes in mitochondrial structure in CNS axons. Although we cannot rule out cuprizone-induced changes in mitochondrial function, increased mitochondrial size/volume in demyelinated axons has been reported in multiple studies regardless of the cause of demyelination (14, 17, 40–42). The mitochondrial changes described following cuprizone-induced demyelination in vivo were also found following lysolecithin-induced demyelination in vitro. The observation that lysolecithin did not alter mitochondrial behavior in unmyelinated axons provides additional support for demyelination (not cuprizone-induced mitochondria changes) as the cause of the changes in axonal mitochondria described in the present paper.

Studies of postmortem multiple sclerosis brains have described increased mitochondria volume in axons that are likely to be demyelinated for years or decades (14, 42). We establish here that increased axonal mitochondrial volume occurs within days (slice cultures) or weeks (mice) of demyelination, and that this increase is caused by increased sizes of mitochondrial stationary sites. Further studies are necessary to elucidate the mechanisms that simultaneously regulate mitochondrial stationary site size and mitochondrial motility in demyelinated axons. To compensate for loss of myelin, the demyelinated axon distributes voltage-gated Na⁺ channels along its entire surface (43). In contrast to nerve impulses that jump from node to node in myelinated axons, nerve impulses are propagated by continuous depolarization of the axolemma in demyelinated axons. It has been hypothesized that this increases the energy demand for repolarization of the axolemma, which occurs by exchange of axoplasmic Na⁺ for extracellular K⁺ in an energy-dependent manner. In addition to increasing the energy demand of nerve conduction, demyelination may also decrease the availability of ATP precursors. Recent studies have raised the possibility that the monocarboxylase transporter-1 located in myelin transfers lactate from the oligodendrocytes to myelinated axons (44, 45). Axons are at risk to degenerate when they cannot generate sufficient ATP to exchange Na⁺ for K⁺, as excessive axoplasmic Na⁺ is exchanged for extracellular Ca²⁺ by the Na⁺/Ca²⁺ exchanger in an energy-independent manner (12, 46). Increased axoplasmic Ca²⁺ is thought to be a major cause of pathology and degeneration of demyelinated axons (11). We propose, therefore, that failure to increase mitochondrial volume in syntaphilin-deficient demyelinated axons reduces ATP production, increases axonal Na⁺ and Ca²⁺, and augments axonal degeneration. This interpretation is consistent with previous studies that have shown that the lack of syntaphilin-mediated mitochondrial docking in presynaptic terminals induces rapid buildup of intracellular Ca²⁺ during intensive stimulation without significantly affecting basal synaptic transmission (24).

ATP output in individual demyelinated axons cannot be reliably measured. We rationalized that the increased energy demand would be proportional to the rate of axonal firing in demyelinated axons. This hypothesis is supported by our observation that blocking Na⁺ channels of syntaphilin-null demyelinated axons significantly reduced axonal degeneration. This is consistent with previous reports that Na⁺ channel blockers reduced axonal degeneration in the immune-mediated demyelinating mouse models of experimental autoimmune encephalomyelitis and neuritis (38, 47, 48) and in myelinated peripheral axons treated with NO (37). Given that glutamate-mediated excitotoxicity has been implicated in damage of demyelinated axons (49), stationary mitochondria may also have a direct beneficial role to buffer axoplasmic Ca²⁺ in demyelinated axons, and this buffering capacity may be compromised in syntaphilin-deficient axons. In addition, blockage of axonal Na⁺ channels can also reduce Ca²⁺ influx and Ca²⁺ signaling pathways that alter axonal stability. Our studies have described mitochondrial changes in demyelinated axons that could involve mitochondrial biogenesis, redistribution, and dynamics, and can render the demyelinated axon more susceptible to axonal degeneration. They raise the possibility that therapeutic stabilization of mitochondria–microtubule docking interactions may prevent or reduce degeneration of demyelinated axons.

Materials and Methods

Supplementary Material