Hyperphosphorylation and Aggregation of Tau in Experimental Autoimmune Encephalomyelitis*

Axonal damage is a major morphological correlate and cause of permanent neurological deficits in patients with multiple sclerosis (MS), a multifocal, inflammatory and demyelinating disease of the central nervous system. Hyperphosphorylation and pathological aggregation of microtubule-associated protein tau is a common feature of many neurodegenerative diseases with axonal degeneration including Alzheimer's disease. We have therefore analyzed tau phosphorylation, solubility and distribution in the brainstem of rats with experimental autoimmune encephalomyelitis (EAE), an animal model of MS. Tau was hyperphosphorylated at several sites also phosphorylated in Alzheimer's disease and became partially detergent-insoluble in EAE brains. Morphological examination demonstrated accumulation of amorphous deposits of abnormally phosphorylated tau in the cell body and axons of neurons within demyelinating plaques. Hyperphosphorylation of tau was accompanied by up-regulation of p25, an activator of cyclin-dependent kinase 5. Phosphorylation of tau, activation of cdk5, and axonal pathology were significantly reduced when diseased rats were treated with prednisolone, a standard therapy of acute relapses in MS. Hyperphosphorylation of tau was not observed in a genetic or nutritional model of axonal degeneration or demyelination, suggesting that inflammation as detected in the brains of rats with EAE is the specific trigger of tau pathology. In summary, our data provide evidence that axonal damage in EAE and possibly MS is linked to tau pathology.

Axonal damage is a major morphological correlate and cause of permanent neurological deficits in patients with multiple sclerosis (MS), a multifocal, inflammatory and demyelinating disease of the central nervous system. Hyperphosphorylation and pathological aggregation of microtubule-associated protein tau is a common feature of many neurodegenerative diseases with axonal degeneration including Alzheimer's disease. We have therefore analyzed tau phosphorylation, solubility and distribution in the brainstem of rats with experimental autoimmune encephalomyelitis (EAE), an animal model of MS. Tau was hyperphosphorylated at several sites also phosphorylated in Alzheimer's disease and became partially detergent-insoluble in EAE brains. Morphological examination demonstrated accumulation of amorphous deposits of abnormally phosphorylated tau in the cell body and axons of neurons within demyelinating plaques. Hyperphosphorylation of tau was accompanied by up-regulation of p25, an activator of cyclin-dependent kinase 5. Phosphorylation of tau, activation of cdk5, and axonal pathology were significantly reduced when diseased rats were treated with prednisolone, a standard therapy of acute relapses in MS. Hyperphosphorylation of tau was not observed in a genetic or nutritional model of axonal degeneration or demyelination, suggesting that inflammation as detected in the brains of rats with EAE is the specific trigger of tau pathology. In summary, our data provide evidence that axonal damage in EAE and possibly MS is linked to tau pathology.
Multiple sclerosis (MS) 1 is an inflammatory disease that leads to the destruction of myelin in the central nervous system (1,2). There is increasing evidence that axonal damage is the major morphological correlate of permanent neurological deficits in patients with MS (3,4). Studies using magnetic resonance spectroscopy/imaging have suggested that neurodegeneration starts already at the onset of the disease (5). Accordingly, neuropathological studies revealed significant axonal injury in early disease stages (6 -9). The finding that intralesional axonal damage is related to the degree of inflammation in the lesions has led to the conclusion that inflammation is not only responsible for demyelination but also for axonal injury (10). Axonal degeneration in MS is defined by characteristic morphological signs such as axonal swellings and spheroids (6 -9). Histopathological studies have shown that axonal damage in MS is associated with axonal accumulation of amyloid precursor protein (APP), which is transported in a kinesin-dependent fashion, indicating impairment of axonal transport (6). Because neurons are highly elongated cells, their function depends on efficient transport of proteins and organelles toward synapses. A disturbance in axonal transport would therefore cause energy depletion at synapses, eventually leading to complete transsection and degeneration of axons in MS. The underlying molecular mechanisms of transport impairment and axonal degeneration in MS are so far elusive.
Axonal degeneration is also found in neurodegenerative diseases (i.e. Alzheimer's disease, progressive supranuclear palsy, frontotemporal dementia linked to parkinsonism), which are characterized by pathological hyperphosphorylation and assembly of microtubule-associated protein tau into paired helical filaments (11,12). The physiological function of tau is to bind to and stabilize microtubules in a phosphorylation-dependent way (11). In addition, tau is involved in regulation of anterograde axonal transport by influencing the attachment/ detachment rate of molecular motors along microtubules (13). Pathological hyperphosphorylation of tau as seen in Alzheimer's disease causes detachment of tau from microtubules that might lead to microtubule breakdown and disruption of axonal transport (loss of function). An imbalance of kinases and phosphatases has been proposed to contribute to the pathogenesis of diseases with paired helical filaments (14). A common hypothesis holds that tau hyperphosphorylation and subsequent detachment increases the pool of unbound tau beyond a critical concentration, thereby initiating its aggregation into paired helical filaments (gain of toxic function) (11,12,15,16).
In light of the conspicuous axonal abnormalities in MS, we wondered whether tau abnormalities contribute to neuronal dysfunction and degeneration in experimental autoimmune encephalomyelities (EAE), an animal model of MS. Myelin-oligodendrocyte-glycoprotein (MOG)-induced EAE in rats resembles many characteristic features of MS including multifocal inflammation, demyelination, and axonal loss. Therefore, we characterized tau phosphorylation, solubility, and distribution in rats with acute brainstem EAE.

EXPERIMENTAL PROCEDURES
Induction of EAE-EAE was induced in female LEW.1N rats by intradermal injection of 50 g rat MOG in saline emulsified (1:1) with complete Freund's adjuvant (Sigma) containing 200 g of Mycobacterium tuberculosis (strain H 37 RA; Difco Laboratories, Detroit, MI). Control rats were injected with complete Freund's adjuvant alone. Rats were scored for clinical signs of EAE and weighed daily. Rats were sacrificed 12-13 days after sensitization. The experiments were approved by the regional ethics board.
Treatment of Animals-Rats were treated intraperitoneally with prednisolone (20 mg/kg) starting on day 8 after sensitization as described (17). 8-Week-old C57/Bl6 mice (n ϭ 5) were fed with 0.2% (w/w) cuprizone (bis-cyclohexanone oxaldihydrazone) (Sigma) in ground breeder chow for 5 weeks. Subsequently, brains were snap-frozen in liquid nitrogen for further biochemical analysis (see below). Brains from age-matched animals (n ϭ 5) maintained on a normal diet served as controls.
Isolation of Insoluble Tau-Brainstems were homogenized in lysis buffer (10 mM Tris-HCl, 150 mM NaCl, 20 mM NaF, 1 mM Na 3 VO 4 , 2 mM EGTA, 0.5% Triton X-100, and 0.1% SDS) and protease inhibitor mixture and centrifuged twice at 10,000 ϫ g for 5 min. The supernatant was removed and recentrifuged at 100,000 ϫ g for 30 min. The resulting pellet was re-extracted with 70% formic acid to recover the insoluble material. For isolation of crude paired helical filaments (18) brainstem were homogenized in a buffer containing 10 mM Tris-HCl, 0.8 M NaCl, 1 mM EGTA, 10% sucrose and protease inhibitors, and centrifuged at 27,000 ϫ g for 20 min. The pellet was washed once. Both supernatants were combined, adjusted to 1% Sarkosyl, and incubated for 1 h at 37°C, followed by centrifugation at 100,000 ϫ g for 35 min. The resulting pellet containing crude insoluble tau was resuspended in 8 M urea.
Protein Phosphatase Activity Measurements-Phosphatase activity was determined by phosphatase assay V2460 according to the manufacturer's protocol (Promega).

Hyperphosphorylation of Tau in Rats with EAE-
To study the mechanisms of axonal pathology in MS, we used LEW.1N rats with MOG-induced hyperacute EAE as an experimental model (19). As in acute MS, active demyelination is accompa-nied with extensive axonal damage in these animals (9). In addition, the focal course of the disease with reproducible severe brainstem pathology facilitates the selection and biochemical analysis of lesions. To study expression levels and phosphorylation status of tau in inflammatory demyelinating lesions, LEW.1N rats were immunized with 50 g of recombinant MOG in complete Freund's adjuvant and brainstems were dissected at day 12-13 post-immunization. Immunoblotting analysis with the pan-tau antibody tau-5 did not reveal any significant changes in the expression levels of brainstem tau between rats with EAE, adjuvant only immunized controls, and naïve rats (Fig. 1). To determine whether tau phosphorylation was altered, we used a panel of different phosphorylation-dependent antibodies that are commonly used to detect Alzheimer tau. One can broadly distinguish two classes of paired helical filament tau phosphorylation sites (Table I): 1) Ser-Pro and Thr-Pro (SP/TP) sites in the flanking region of tau phosphorylated by proline-directed kinases such as glycogen synthase kinase 3␤ (GSK-3␤), cyclin-dependent kinase 5 (cdk5), or mitogen-activated protein (MAP) kinase. 2) KXGS motifs in the repeat region phosphorylated by nonproline-directed kinases like microtubule-affinity regulating kinase (MARK/Par-1) and protein kinase A. Tau phosphorylation at KXGS sites has been implicated in the loss of microtubule binding, whereas phosphorylation of the flanking region has only a minor effect on tau microtubule interaction. Replicate blots were stained with PHF-1 (pS396/pS404), AT-8 (pS202/pT205), AT100 (pT212/ pS214), and AT-180 (pT231/pS235) to detect distinct prolinedirected kinase phosphorylation sites in the flanking region of tau. We observed a dramatic increase of immunoreactivity toward the PHF-1, AT-8, AT-100, and AT-180 epitopes in brainstem lysates prepared of rats with EAE compared with those from control animals (Fig. 1). The average increase was ϳ3.5-fold for PHF-1, ϳ4-fold for AT-8, ϳ4-fold for AT-100, and ϳ4.5-fold for AT-180. Next we assessed the phosphorylation status of tau at KXGS motifs in the repeat region phosphorylated by nonproline-directed kinases. Among these phosphorylation sites, Ser-262 is of particular interest, because its phosphorylation is increased in Alzheimer's disease, virtually abolishes tau binding to microtubules and disrupts microtubule stability (20,21). Immunoblotting revealed that phosphorylation at the 12EA (pS262) epitope was strongly increased in rats with EAE ( Fig. 1). Activation of MAPK and p25/cdk5 in EAE-To determine the molecular mechanisms for increased tau phosphorylation, the activity of several known tau-directed protein kinases, including MAPK, GSK-3␤, cdk5, and MARK were studied (Fig. 2, A  and B). Immunoblot analysis using an anti-phospho MAPK (Erk1/2) antibody, which recognizes only the activated form of MAPK, showed a ϳ2.5-fold increase in MAPK activity in EAE brain lysates compared with that in adjuvant alone immunized or naïve rats. The overall expression levels of MAPK were not changed. To assess the activity of cdk5, we determined the levels of cdk5 activators, p35 and its proteolytic fragment p25. The proteolytic conversion of p35 to p25 has been implicated in aberrant cdk5 activation leading to tau hyperphosphorylation, cytoskeleton disruption, and neuronal death (22)(23)(24)(25). There was a striking increase in the p25/p35 ratio in EAE brainstems, suggesting conversion of p35 to p25 in EAE. The activity of the tau-directed kinase GSK-3␤ is negatively regulated by phosphorylation at Ser-9. We found increased phosphorylation at Ser-9, whereas levels of total GSK-3␣/␤ remained unchanged, indicating that GSK-3␤ is not involved in the cascade leading to tau hyperphosphorylation in EAE. Furthermore, we observed that the activity of the non-proline-directed kinase, MARK, which phosphorylates Ser-262, was not altered in EAE.
In addition to kinase activation, inactivation of phosphatases can result in hyperphosphorylation of tau (26,27). We therefore determined the activity of the major tau-directed phosphatase, PP2A, in duplicate samples by a colorimetric assay using a phosphopeptide substrate in the PP2A-specific reaction buffer (27,28). The peptide was dephosphorylated to the same extent by homogenates obtained from rats with EAE compared with controls, indicating that the activity of the tau-directed phos-phatase, PP2A, is not changed (Fig. 2C).
Hyperphosphorylated Tau and p25/p35 in Degenerating Neurons of Rats with EAE-Previous neuropathological studies have demonstrated axonal dilatations and spheroids in rats with MOG-induced EAE as well as in human MS brains (6 -9). It has been shown recently that damaged axons and spheroids stain strongly positive with an antibody against APP (6). Whereas physiological levels of axonal APP are not detected by this method, APP accumulation is found in damaged axons possibly because of failure of axonal transport (6). We detected prominent APP staining in dilated axons and spheroids on paraffin-embedded brainstems of LEW.1N rats with EAE (Fig.  3A). To analyze whether pathologically hyperphosphorylated tau was localized in neurons with axonal injury, we performed immunohistochemical stainings with PHF-1, AT-8, and 12E8 antibodies. All three antibodies prominently stained axons in EAE brains, particularly dilated and irregularly shaped axons and axonal spheroids ( Fig. 3; and data not shown). In contrast, only weak immunoreactivity of AT-8, PHF-1, and 12E8 was observed in adjuvant-immunized control rats. We next evaluated the immunoreactivity with TG-3, an antibody that recognizes phosphorylated tau (pT231) with abnormal conformation and detects early stages of paired helical filaments (29,30). TG-3 displayed extensive staining of abnormal axons and spheroids in rats with EAE, indicating a pathological conformation shift of tau (Fig. 3). Similar results were observed when the conformation-dependent antibody MC-1 was used (data not shown). In addition, an altered compartmentalization of tau with accumulation of amorphous and granular tau deposits in the soma of neurons within demyelinating plaques was detected by PHF-1 and Bielschowsky stainings (Fig. 3B). Previous studies have shown that the TG-3, AT-8, and PHF-1 epitopes can be generated in vivo by cdk5 (31, 32). Because we found a conversion of p35 to p25, indicating an activation of TABLE I Bar diagram of the longest tau isoform tau40 and its phosphorylation sites The repeat region (white bar) consists of 4 repetitive protein sequences and is required for microtubule binding. Phosphorylation of its KXGS motifs leads to detachment of tau from microtubules. In contrast, the phosphorylation of SP/TP motifs in the flanking regions (black bars) have only a modulatory effect on tau-microtubule binding. The table shows phosphoepitopes of tau together with kinases, identified to phosphorylate these sites in vivo and in vitro.

FIG. 2. Analysis of kinase and phosphatase activation in rats with EAE.
A, Western blot analysis of brainstem lysates from EAE rats and controls with antibodies against various tau-directed kinases (samples contained equal amounts of total protein). Note that p-MAPK and p-MARK are directed against active kinase, whereas p-GSK-3␤ recognizes the inactive enzyme. B, quantitative analysis of the expression levels of total and activated kinases. Total expression levels of studied kinases remained unchanged. We observed activation of MAPK, decreased activity of GSK-3␤, no significant change in MARK/Par-1 activity, and increased p25/p35 ratio, indicating aberrant activation of cdk5. Values are mean Ϯ S.D., n ϭ 5 for each value. Two independent experiments showed similar results (*, p Ͻ 0.05; **, p Ͻ 0.01). C, phosphatase activities in brainstem homogenates of rats with and without EAE were determined in duplicate samples using a colorimetric assay in PP2A specific reaction buffer. PP2A activity is shown in % in relation to a phosphate standard curve. The assay was also run in the presence of the phosphatase inhibitor, okadaic acid (OA). The values are mean Ϯ S.D., n ϭ 5 for each value. Two independent experiments showed similar results and no significant difference in phosphatase activity between EAE and control animals. cdk5 toward tau phosphorylation in EAE brains, we analyzed the localization of the regulatory subunits of cdk5 in rats with EAE. Immunostaining with a C-terminal p35 antibody recognizing both p35 and p25 revealed intense staining of axons with abnormal profiles and hyperphosphorylated tau (Fig. 3C). In contrast, we did not detect any significant co-localization of pMAPK and PHF-1 (Fig. 3D). Taken together, these results indicate that cdk5, but not MAPK is involved in hyperphosphorylation of tau in EAE.

Tau Insolubility and Aggregation, but No Formation of Paired Helical Filaments in Rats with EAE-
The accumulation of hyperphosphorylated tau in dilated axons and spheroids as well as its pathological conformation, which is known to precede formation of paired helical filaments, is suggestive of axonal tau aggregations. Because paired helical filaments are highly insoluble, we analyzed the insolubility of tau in 0.1% SDS and 0.5% Triton X-100. EAE brainstems were extracted in lysis buffer, and the detergent-soluble and -insoluble fractions were subjected to immunoblotting analysis. Whereas tau was barely detectable in the detergent-insoluble fraction of control brains we found significant amounts of insoluble tau in EAE brainstems (Fig. 4). The overall levels of tau did not differ between EAE and controls. These data show that in addition to hyperphosphorylation, tau also partially forms detergent-insoluble aggregates in EAE. An established biochemical method for isolation of paired helical filaments is to take advantage of their insolubility in 1% Sarkosyl (18,33). In contrast to its insolubility in 0.1% SDS and 0.5% Triton X-100, tau from diseased animals was soluble in 1% Sarkosyl. Taken together, these findings demonstrate aggregation of tau without the formation of paired helical filaments.
Prednisolone Treatment Reduces Kinase Activation and Tau Phosphorylation in Rats with EAE-High-dosage prednisolone treatment is the standard therapy regime in acute relapses of MS (1, 2). We therefore tested whether the treatment with prednisolone could inhibit the pathological cascade leading to hyperphosphorylation and aggregation of tau in EAE rats. Rats were treated with 20 mg/kg prednisolone intraperitoneally from day 8 post-immunization to day 12. At day 12 animals from treated and untreated groups were sacrificed. As ex-pected, the severity of the disease, measured by the clinical score of each animal, and the number of inflammatory infiltrates were significantly reduced in rats that had been treated with prednisolone. Previous work in a rat model of autoimmune optic neuritis has provided evidence that prednisolone treatment may induce apoptosis in retinal ganglia cells (17). We therefore evaluated the extent of axonal damage by Bielschowsky silver staining on brainstem cross-sections of rats with EAE. However, signs of axonal injury such as dilated axons and spheroids were virtually absent in the brainstem of rats that had been treated with prednisolone.
Next, we analyzed the effect of prednisolone treatment on tau phosphorylation. By immunoblotting brainstem lysates with PHF-1 we found a significant reduction (ϳ50%) of tau phosphorylation in treated animals compared with untreated controls (Fig. 5). In addition, treatment with prednisolone reduced levels of phosphorylated MAPK and p25 significantly.
Tau Is Not Hyperphosphorylated in Other Conditions with Primary Oligodendroglial Dysfunction-The causal relationship of inflammation, demyelination, and axonal degeneration has been difficult to determine in EAE. We therefore analyzed the extent of tau phosphorylation in a model of axonal damage that is caused by oligodendroglial dysfunction and not by inflammation. Mice that are deficient for the oligodendroglial protein, 2Ј,3Ј-cyclic nucleotide 3Ј-phosphodiesterase (CNP), develop a progressive neurodegenerative disorder in the absence of demyelination or inflammation, starting 4 months after birth and are characterized by signs of axonal damage as observed in MS (Fig. 6C and Ref. 34). The development of axonal pathology in CNP-deficient mice was not accompanied by an increase in tau phosphorylation (Fig. 6A). To analyze whether acute demyelination per se can induce tau pathology in axons, we fed mice with the demyelinating toxin cuprizone for 5 weeks. Although cuprizone treatment leads to extensive demyelination ( Fig. 6D and Ref. 35), we did not detect any changes in tau phosphorylation (Fig. 6B). These data suggest that hy- perphosphorylation of tau is part of a specific pathway triggered by an inflammatory attack in the central nervous system of rats with EAE. DISCUSSION Here we show that 1) tau derived out of EAE brains is abnormally hyperphosphorylated at sites that define tau pathology in Alzheimer's disease; 2) that hyperphosphorylation is most likely because of an activation of cdk5 rather than to decreased phosphatase activity; 3) that pathologically hyperphosphorylated tau is localized in injured neurons of EAE brains; and 4) that tau becomes partially insoluble and undergoes a conformational shift that is thought to precede paired helical filament formation, whereas aggregation into paired helical filaments is not observed.
These findings raise the question of how tau could mediate the axonal damage that occurs in active MS lesions. The fact that the binding of tau is regulated by phosphorylation within the repeat region (e.g. Ser-262) has led to the hypothesis that tau hyperphosphorylation leads to a release of tau from microtubules, followed by microtubule breakdown and transport decay (loss of function). In addition, hyperphosphorylated tau could be neurotoxic by itself or in its aggregated form (gain of toxic function). Consistent with the loss of function hypothesis we found significant tau phosphorylation at Ser-262, one site within the repeat domain that strongly inhibits microtubule binding and causes detachment of tau from microtubules when phosphorylated. It is therefore possible that hyperphosphorylation of tau during acute inflammation disengages tau from microtubules, causing their destabilization and impairment of axonal flow. Furthermore, we observed significant abnormal phosphorylation of SP/TP motifs in the flanking regions of tau such as PHF-1 (pS396/pS404), AT-8 (pS202/pT205), AT-180 (pT231/pS235), and AT-100 (pT212/pS214) epitopes.
To identify downstream kinases involved in the sequential toxic hyperphosphorylation of tau at SP/TP motifs we performed immunoblotting with several activity dependent antibodies against active MAPK, inactive GSK-3␤, and upstream activators of cdk5 because all of these have been described to phosphorylate tau in vivo and in vitro. In contrast to GSK-3␤, which was found to be down-regulated we detected an increase in MAPK activity as well as an increase in the ratio of cdk5 activators p25/p35. Considering that MAPK is also expressed in astrocytes and microglial cells it is possible that the detected activated MAPK could originate from glial cells rather than neurons (36). Indeed, we did not detect significant co-localization of pMAPK and PHF-1 in EAE. In contrast, cdk5 and its activators p25/p35 are primarily expressed in neurons (see also Fig. 3C) and have been localized and purified from brain microtubules (37,38). In addition, transgenic mice overexpressing p25/cdk5 display increased tau phosphorylation at SP/TP motifs in the flanking region as well as tau aggregation and neurodegeneration (23)(24)(25). Cdk5 is recruited to the neuronal membrane by its interaction with membrane-anchored p35 (39). Aberrant activation of cdk5 occurs when the myristoylated domain of p35 is cleaved to p25, leading to release of p25-bound cdk5 into the cytoplasmic compartment. Several lines of evidence indicate that only p25 but not p35-bound cdk5 phosphorylates tau in vitro, thereby inducing pathological alterations in neurons (22-25, 40, 41). It is interesting to note that conversion of p35 to p25 is regulated by the calcium-dependent cysteine protease calpain as activation of calpain together with impaired calcium homeostasis has been observed in EAE and MS brains (42,43). It is therefore tempting to speculate that increased calcium influx might trigger a cascade of pathological events that lead to calpain activation, followed by conversion of p35 to p25 and increased and pathological tau phosphorylation at epitopes such as PHF-1, AT-8, and AT-100.
It has generally been assumed that filamentous tau aggregations as they are described in Alzheimer's disease, frontotemporal dementias and other tauopathies might be direct mediators of neuronal toxicity because the clinical progression of Alzheimer's disease correlates with distribution and amount of tau aggregates (44,45). Tau is a highly soluble protein because its sequence consists mostly of hydrophilic residues. The exact molecular mechanisms of its abnormal aggregation into paired or straight helical filaments are not completely understood. In vitro studies propose extrinsic (polyanions, oxidative environment) as well as intrinsic factors (increased tau concentration, tau mutations that promote ␤-structure) as possible reasons for aggregation (46). Hyperphosphorylation has also been assumed to cause pathological aggregation of tau.
The antibody TG-3 recognizes a conformation-dependent epitope that has been reported to precede paired helical filament formation (29). Because we observed a marked TG-3 and MC-1 staining of injured axons, we tested tau solubility and aggregation in the EAE model. We could identify one fraction of tau that had become insoluble in 0.5% Triton X-100, 0.1% SDS, indicating the formation of tau aggregates. Amorphous aggregates in the soma of neurons and axonal spheroids that are stained by PHF-1 and Bielschowsky silver impregnation could be the morphological correlate of the 0.5% Triton X-100, 0.1% SDS-insoluble tau fraction. A widely accepted biochemical method to isolate paired helical filaments of Alzheimer brains is based on their insolubility in 1% Sarkosyl. However, no tau reactivity could be detected in the Sarkosyl-insoluble fraction of EAE brainstem lysates. In line with the absence of Sarkosylinsoluble tau aggregates, no paired helical filaments were observed by Gallyas or Bielschowsky staining. Insolubility and therefore aggregation of tau in the absence of paired helical filaments raises the question of the relevance of these findings for axon degeneration in EAE. It is poorly understood whether hyperphosphorylated tau is toxic by itself or oligomeric tau aggregations or stable fibrils are required to cause cellular dysfunction. Evidence is now accumulating that formation of paired helical filaments is not necessary for tau toxicity. Phosphorylation of tau in a temporally ordered series can already cause neurodegeneration in the absence of aggregation (47). Furthermore, it has been shown that misfolded and oligomerized proteins can cause cellular dysfunction before the deposition of stable fibrils (48). Small protein assemblies that can only be detected biochemically and not by imaging techniques might be in part responsible for cytotoxicity. For example, in mice expressing mutant human APP cell death occurs well before A␤ amyloid deposition (49,50). Similarly, several lines of evidence suggest that tau-related neurodegeneration can occur without or precede paired helical filament formation (47,51). Whether the biochemically detected tau aggregates in EAE brainstems represent toxic intermediates of the fibrillogenic process remains to be established. It is also possible that detergent-insoluble tau is part of larger poorly soluble cellular aggregates (for example, found in axonal spheroids) formed as a consequence of impaired axonal flow and are thus not related to any fibrillogenic process.
Neuropathological studies have shown that axonal damage in MS and EAE lesions correlates with the extent of inflammation, suggesting that the primary insult is an inflammatory attack (6 -9). However, the causal relationship of inflammation, demyelination, and axonal degeneration has been difficult to determine. To address this question we used a genetic model of axonal damage, which is triggered by oligodendroglial dysfunction instead of inflammation. CNP-deficient mice display axonal loss in the absence of demyelination or inflammation. Although the axonal degeneration observed in this model displays similar ultrastructural features such as APP positive axonal swellings and spheroids, hyperphosphorylation of tau is only found in EAE, but not in CNP-deficient mice. In addition, cuprizone-induced demyelination did not lead to tau pathology in axons. These data indicate that demyelination per se is not sufficient to induce changes in tau phosphorylation. Furthermore, one can speculate that inflammation as seen in EAE triggers a specific pathway of axonal damage, which is distinct from the one caused by oligodendroglial dysfunction or demyelination. This raises the question to which extent axonal damage can be reversed by reducing the inflammatory load and how steroid treatment that is the standard therapy for acute relapsing MS influences axonal damage. Our findings suggest that a prednisolone pulse treatment during the active phase of inflammation does not only reduce the amount of inflammatory infiltrates, but also the extent of axonal damage. In summary our data provide evidence that axonal damage in EAE is associated with tau hyperphosphorylation and aggregation. These pathological tau alterations can be partially prevented by early prednisolone treatment. These findings are of particular relevance because the amount of axonal damage is a major determinant of persistent neurological deficits in MS patients. Our results might open new perspectives for understanding molecular pathology and treatment of multiple sclerosis.