|Home | About | Journals | Submit | Contact Us | Français|
Multiple sclerosis involves demyelination and axonal degeneration of the central nervous system. The molecular mechanisms of axonal degeneration are relatively unexplored in both multiple sclerosis and its mouse model, experimental autoimmune encephalomyelitis. We previously reported that targeting the axonal growth inhibitor, Nogo-A, may protect against neurodegeneration in experimental autoimmune encephalomyelitis; however, the mechanism by which this occurs is unclear. We now show that the collapsin response mediator protein 2 (CRMP-2), an important tubulin-associated protein that regulates axonal growth, is phosphorylated and hence inhibited during the progression of experimental autoimmune encephalomyelitis in degenerating axons. The phosphorylated form of CRMP-2 (pThr555CRMP-2) is localized to spinal cord neurons and axons in chronic-active multiple sclerosis lesions. Specifically, pThr555CRMP-2 is implicated to be Nogo-66 receptor 1 (NgR1)-dependent, since myelin oligodendrocyte glycoprotein (MOG)35–55-induced NgR1 knock-out (ngr1−/−) mice display a reduced experimental autoimmune encephalomyelitis disease progression, without a deregulation of ngr1−/− MOG35–55-reactive lymphocytes and monocytes. The limitation of axonal degeneration/loss in experimental autoimmune encephalomyelitis-induced ngr1−/− mice is associated with lower levels of pThr555CRMP-2 in the spinal cord and optic nerve during experimental autoimmune encephalomyelitis. Furthermore, transduction of retinal ganglion cells with an adeno-associated viral vector encoding a site-specific mutant T555ACRMP-2 construct, limits optic nerve axonal degeneration occurring at peak stage of experimental autoimmune encephalomyelitis. Therapeutic administration of the anti-Nogo(623–640) antibody during the course of experimental autoimmune encephalomyelitis, associated with an improved clinical outcome, is demonstrated to abrogate the protein levels of pThr555CRMP-2 in the spinal cord and improve pathological outcome. We conclude that phosphorylation of CRMP-2 may be downstream of NgR1 activation and play a role in axonal degeneration in experimental autoimmune encephalomyelitis and multiple sclerosis. Blockade of Nogo-A/NgR1 interaction may serve as a viable therapeutic target in multiple sclerosis.
Multiple sclerosis is a severe neurological disorder that involves inflammation in the brain and spinal cord, axonal damage and demyelination. Experimental autoimmune encephalomyelitis (EAE) is an animal model that mimics many of the pathophysiological hallmarks of multiple sclerosis. The induction of EAE by immunization with the myelin oligodendrocyte glycoprotein (MOG) peptide encompassing the 35–55 amino acid sequence (MOG35–55) is a commonly used animal model for the neurodegenerative changes observed in multiple sclerosis (Johns et al., 1995). It has long been believed that axonal damage, degeneration and loss are secondary to demyelination. However, data from our laboratory have demonstrated that axonal degeneration can precede demyelination in EAE (Wang et al., 2005). Notably, Trapp et al. (1998) have reported the appearance of newly transected axons in active and hypocellular chronic-active multiple sclerosis lesions from brains obtained at autopsy, suggesting a relationship between axonal damage and permanent neurological deficits.
It is likely that multiple sclerosis may display similar pathogenic mechanisms and features as observed in spinal cord injury, whereby degenerating axons may attempt regeneration, but fail due to the inhibitory environmental cues (Jurewicz et al., 2007). The factors in the adult CNS that inhibit these compensatory regenerative responses include myelin proteins and extracellular matrix constituents (Profyris et al., 2004). Chief among these factors are integral components of myelin known as the myelin-associated inhibitory factors, such as the potent inhibitor of neurite outgrowth, Nogo-A (GrandPre et al., 2000; Fournier et al., 2001). Expression of Nogo-A in oligodendrocytes has been reported to be upregulated in active demyelinating lesions of multiple sclerosis (Satoh et al., 2005). In addition, our laboratory has demonstrated that immunization against Nogo-A or deletion of the nogo gene ameliorates the effects of EAE (Karnezis et al., 2004). While Nogo-A is known to be a potent inhibitor of axonal regrowth, a direct relationship with bona fide axonal pathology, an important feature of both EAE and multiple sclerosis lesions (Trapp et al., 1998; Wang et al., 2005), is as yet undefined. Although recent evidence has demonstrated that deletion of the co-receptor for Nogo-A, LINGO-1, improves axonal integrity and protects against neurological decline in MOG-induced EAE (Mi et al., 2007), the mechanism by which this eventuates is not known. Furthermore, small interfering RNA silencing of Nogo-A in both MOG35–55- and myelin basic protein-induced EAE models has been shown to limit the clinical severity of the disease and promote repair (Yang et al., 2010) but again it is unclear how this therapeutic effect is achieved. This raises the question of whether in EAE and multiple sclerosis, Nogo-A can signal axonal degeneration through its axonally localized cognate receptors.
In this study, we show for the first time that phosphorylated collapsin response mediator protein 2 (CRMP-2) (pThr555CRMP-2, a key molecule that regulates microtubule assembly) is abundant in degenerating spinal cord neurons and their axons of EAE-induced mice and active multiple sclerosis lesions. We are able to define further that the increase in axonally specific pThr555CRMP-2 is limited in EAE-induced ngr1−/− mice, thereby preventing significant axonal and myelin degeneration characteristic of MOG35–55 EAE. Moreover, the introduction of a site-specific T555A mutation in CRMP-2 through a recombinant adeno-associated virus 2 (rAAV2) delivery system to retinal ganglion cells, limits axonal degeneration in the optic nerve during the peak stage of EAE. Finally, we highlight the contribution of pThr555CRMP-2 during the neurodegenerative phase of EAE by reducing the spinal cord levels and associated axonal pathology through the passive transfer of anti-Nogo-A antibodies. Thus, reducing the NgR1-dependent signalling capacity during EAE may limit the activation of the phosphorylation of CRMP-2, thereby preventing axonal degeneration and neurological decline. By extension, these data advocate for targeting NgR1 signalling in multiple sclerosis.
Female C57Bl/6 mice (aged 10−16 weeks) were bred and maintained at Monash University Animal House. Experiments were performed in accordance with the Australian code of practice for the care and use of animals for scientific purposes, approved by the Monash University Animal Ethics Committee and Office of the Gene Technology Regulator of Australia. The ngr1 exon 2 gene mutation is a doubly targeted allele on a C57Bl/6 background and backcrossed more than eight generations as previously described (Kim et al., 2004). To induce EAE, a total of 200µg of the encephalitogenic peptide, MOG35–55 (GenScript) emulsified in 200µl of complete Freund’s Adjuvant (Sigma-Aldrich), supplemented with 4µg/ml Mycobacterium tuberculosis was injected subcutaneously into the lower flanks, then followed with an intraperitoneal injection of 350ng pertussis toxin (Sigma-Aldrich). Mice were injected with a second dose of pertussis toxin 48h later (Karnezis et al., 2004). Female ngr1−/− (n=22) and ngr1+/+ (n=14) mice were graded daily post-MOG35–55 injection (days post-injection) for disease progression (Bernard et al., 1997) corresponding to pre-onset (7 days post-injection), onset (12 days post-injection), peak (18 days post-injection) and chronic (30 days post-injection) stages of disease (in wild-type C57Bl/6 mice, n=53), respectively. Severely paralysed mice (Score 3–4) were killed with CO2 inhalation; serum and tissues were then collected. The brain (including optic nerves) and spinal cord were removed and immediately snap frozen in tubes using liquid nitrogen. For histological experiments, the animals were initially transcardially perfused with 0.1M phosphate-buffered saline (pH 7.4), followed with 4% paraformaldehyde. The tissue was dehydrated in 30% sucrose for 48h. For toluidine blue staining and microscopy, animals were processed with 2.5% gluteraldehyde and 4% paraformaldehyde as previously described (Azari et al., 2006).
Human CNS tissue was obtained through the Department of Anatomical Pathology, The Royal Melbourne Hospital, in accordance with the National Health and Medical Research Council guidelines and the Monash University Human Research Ethics Committee for the use of human specimens in medical research. Post-mortem delay did not exceed 24h and biopsy delay was <1h. The following tissues were used in the current study (Table 1). Multiple sclerosis cases: four secondary progressive (67 plaques analysed from brain and spinal cord), three hyperacute (three plaques analysed from brain); other neurological diseases: epilepsy (one case from temporal lobe), progressive multifocal leukoencephalopathy (three cases from brain), Alzheimer’s disease (two cases from frontal lobe), meningitis (one case from frontal lobe); non-neurological diseases: death by natural causes (10 cases from brain). Brain tissues were immersion fixed in 10% buffered formalin and paraffin embedded, then cut at 10µm using a conventional microtome and mounted on Superfrost Plus® slides (Menzel-Gläser). Histological analysis of de-waxed human CNS tissue was performed as described below.
Histological evaluation was performed on 4% paraformaldehyde-fixed, O.C.T-embedded 10-µm sections of cerebellum, optic nerve and spinal cord. Sections were stained using Luxol fast blue and Periodic acid Schiff (to demonstrate demyelination and inflammatory cell infiltration, respectively) and Bielschowsky silver stains (to demonstrate axonal degeneration/loss), as described previously (McQualter et al., 2001). A trained histopathologist semiquantitatively scored more than three sections per mouse, per genotype or per antibody-treated group [see anti-Nogo(623–640) antibody treatment section below] in a blinded manner. The following criteria were used to define histopathological inflammation and demyelination events respectively: 0, no inflammation; 1, cellular infiltrate only in the perivascular areas and meninges; 2, mild cellular infiltrate in parenchyma; 3, moderate cellular infiltrate in parenchyma; and 4, severe cellular infiltrate in parenchyma. The myelin breakdown was assessed by pale staining with Luxol fast blue (no evidence of remyelination) and scored in a blind fashion as follows: 0, no demyelination; 1, mild demyelination; 2, moderate demyelination and 3, severe demyelination. Images were captured in bright-field under a UPlanApo ×40 1.20 NA objective lens of an Olympus Provis A×70 microscope. The captured 16-bit images were then converted to TIFF files and the images were assembled and formatted using Adobe Photoshop vCS3 software.
The following primary antibodies were used: rabbit polyclonal anti-pThr555CRMP-2 affinity purified (generated by the authors), mouse monoclonal anti-CRMP-2 affinity purified clone (Immuno-Biological Laboratories Co. Ltd.), rabbit polyclonal anti-pThr514CRMP-2 (Cell Signaling Technology), mouse monoclonal anti-tau-5 (Calbiochem), mouse monoclonal anti-hyperphosphorylated-tau (clone AT8; Pierce-Thermo Scientific), rabbit polyclonal anti-pSer199/202-tau (Biosource), mouse monoclonal anti-NF200 (Sigma-Aldrich), mouse monoclonal anti-βIII-tubulin (clone Tuj-1; Millipore-Chemicon), rat monoclonal anti-α-tubulin (Millipore-Chemicon), mouse monoclonal anti-β-actin (Sigma-Aldrich), mouse anti-human CD3 (BD Pharmingen). The following secondary antibodies were used: sheep polyclonal anti-mouse horseradish peroxidase (HRP)-conjugated affinity purified IgG, goat polyclonal anti-rabbit HRP-conjugated affinity purified IgG, goat polyclonal anti-mouse Alexa Fluor® 555 affinity purified IgG, goat polyclonal anti-rabbit Alexa Fluor® 488 affinity purified IgG (Molecular Probes). For immunoprecipitation experiments, the control peptides included full-length CRMP-2 and the pThr555CRMP-2 peptide (Millipore).
A rabbit polyclonal antibody was generated against the pThr555CRMP-2 peptide sequence [Cys-Ile550-Pro-Arg-Arg-Thr-Thr(P)-Gln-Arg-Ile-Val-Ala560] as previously reported (Arimura et al., 2000). Characterization of anti-pThr555CRMP-2 antibody was by enzyme-linked immunosorbent assay, demonstrating phospho-peptide reactivity with the affinity purified polyclonal antibody. Immunoreactivity was at 1:10000 dilution of the anti-pThr555CRMP-2 antibody using 5µg of the pThr555CRMP-2 peptide (Millipore) as substrate. Mouse spinal cord lysates reacted with the anti-pThr555CRMP-2 antibody by western blot showing a 62kDa band, which was subsequently demonstrated to be CRMP-2 by mass spectrometry. The 62kDa band immunoreactivity was blocked by preincubation of the antibody with the pThr555CRMP-2 peptide (20µg). Preincubation with the same CRMP-2 peptide without phosphorylation produced the same immunoreactive 62kDa band demonstrating the phospho-specificity of the antibody. Incubation with the pre-bleed antiserum showed no reactivity with the 62kDa band.
Spinal cord and optic nerve tissues were ground and lysed using RIPA buffer with protease inhibitor (Sigma-Aldrich) and PhosSTOP™ phosphatase inhibitor cocktail (Roche Applied Science) in a dounce homogenizer. Homogenates were centrifuged at 13000×g for 20min and protein concentrations of the supernatants determined using the bicinchoninic acid protein assay reagent kit (Pierce) as previously described (Petratos et al., 2008).
SH-SY5Y cells were cultured and differentiated with retinoic acid as previously described (Petratos et al., 2008). Cells were treated with the 10µM amyloid-β peptide diluted in Dulbecco’s modified Eagle medium: nutrient mixture F-12 (Invitrogen) for 24h. The cells were then lysed using cell lysis buffer and then run on a 12% sodium dodecyl sulphate polyacrylamide gel. Transgenic Tg2576 female mice at 18 months of age, along with human Alzheimer’s disease frontotemporal brain tissue were used as positive controls for the tau phosphorylation and hyperphosphorylation western blot studies. Sarcosyl-insoluble tau preparations were prepared as previously described (Anderson et al., 2008).
We loaded and ran 5µg of protein on a 4–12% graded sodium dodecyl sulphate polyacrylamide gel (Invitrogen). Proteins were then electrophoretically transferred onto polyvinylidene fluoride membranes (Millipore). The membranes were blocked with 5% skimmed milk powder and the primary antibodies diluted in this blocking buffer (polyclonal anti-pThr555CRMP-2, 1:1000; monoclonal anti-CRMP-2, 1:1000; anti-α-tubulin, 1:1000; anti-β-actin, 1:5000; anti-pThr514CRMP-2, 1:1000; anti-tau5, 1:1000) were incubated overnight at 4°C. After washing the membranes in 0.1% v/v Tris-buffered saline-Tween, the secondary anti-rabbit (1:1000), anti-sheep (1:1000), anti-mouse (1:1000) or anti-rat (1:1000) HRP-conjugated antibodies were incubated for 2h at room temperature. Immunoreactive proteins were detected using an ECL Plus™ chemiluminescence method (Amersham). The level of immunostaining was determined by image analysis after scanning the exposed films using AlphaImager® (Alpha Innotech) and then processing the 16-bit monochrome images through the ImageQuant TL™ v2003 software (Nonlinear Dynamics, Ltd.) to measure pixel intensity relative to background. The level of staining was normalized with β-actin or α-tubulin and expressed as arbitrary units. Percentages of band levels were also derived by setting the controls as 100% and the test samples were determined to be elevated or reduced from this basal level. All protein bands were analysed from at least four separate samples and statistics then performed (as described below).
To measure the level of total CRMP-2 or α-tubulin prior to immunoprecipitation of CRMP-2, 5% of the starting amount was run on a 4–12% graded sodium dodecyl sulphate polyacrylamide gel. All immunoprecipitation assays were performed from samples containing 100µg of total protein in 1ml of lysate buffer. Immunoprecipitation was performed by adding 2µg of polyclonal anti-pThr555CRMP-2 or monoclonal anti-CRMP-2 (IBL) capture antibody to each sample. Rabbit IgG or mouse IgG was used as negative controls and the positive control included pThr555CRMP-2 peptide with the anti-pThr555CRMP-2 antibody. Samples were incubated overnight at 4°C followed by the addition of 70µl of a 50% slurry of Protein A-Sepharose beads (Millipore) for incubations with polyclonal anti-pThr555CRMP-2 antibody or 70µl of 50% Protein G Sepharose beads (Millipore) for incubations with monoclonal anti-CRMP-2 antibody. The immunoprecipitation and co-immunoprecipitation procedures were then performed as previously described (Petratos et al., 2008) and samples were loaded onto a 4–12% sodium dodecyl sulphate-polyacrylamide gel and then transferred onto a polyvinylidene fluoride membrane for immunoblotting.
Sarcosyl-insoluble tau preparation was done as previously described (Anderson et al., 2008). Briefly, lumbosacral spinal cords were homogenized in a 10× volume of high salt buffer of 10mM Tris (pH 7.4), 0.8M sodium chloride, 1mM EGTA, 10% sucrose (w/v). The extracts were centrifuged at 45000g for 30min, supernatants collected and pooled, then subjected to another extraction with a 5× volume of the same buffer. The pooled supernatant fraction was incubated with 1% sarcosyl for 1h then centrifuged at 260000g for 1h at 4°C. The resulting pellet represented the sarcosyl insoluble tau preparation.
Following transcardial perfusion of the mice with 4% paraformaldehyde, the lumbar enlargements of the spinal cords were removed and embedded in O.C.T (Tissue-Tek® Sakura Finetek Inc.). Serial 10-µm thick longitudinal sections were then cut on a cryostat (CM 1900, Leica Microsystems) and mounted on Superfrost Plus® slides (Menzel-Gläser). The tissue was initially incubated with blocking buffer (phosphate-buffered saline supplemented with 3% goat serum, 3% mouse serum and 0.3% Triton X-100) for 2h at room temperature. The sections were incubated with primary antibodies in blocking buffer overnight at 4°C. The samples were washed three times in phosphate-buffered saline (pH 7.4) for 10min, followed by 2-h incubation with secondary antibodies (goat anti-mouse Alexa Fluor® 488, goat anti-rabbit Alexa Fluor® 555; Invitrogen) at room temperature. After three washes with phosphate-buffered saline, the sections were stained with 4′,6-diamidino-2-phenylindole (DAPI; Molecular Probes) for 10min, washed and cover-slipped using fluorescent mounting medium (Dako). Primary antibodies used were mouse anti-βIII-tubulin (1:500), mouse anti-NF200 (1:200) and rabbit anti-pThr555CRMP-2 (1:200). Images were captured by fluorescence with an UPlanApo ×40 1.20 NA objective lens of an Olympus Provis Ax70 microscope.
Sections were microwaved for antigen retrieval (two 5min washes in 0.1M citrate buffer, pH 6), thoroughly washed (three 10min washes in phosphate-buffered saline, pH 7.4) and incubated with proteinase K (20µg/ml, 1h at 37°C). Following further phosphate-buffered saline washes and post-fixation with 4% paraformaldehyde (30min at room temperature), the sections were incubated with blocking buffer (10% foetal calf serum and 0.3% Triton X-100 in phosphate-buffered saline) overnight at 4°C. Human CNS tissues were then incubated with primary antibodies [mouse anti-NF200 (1:200), mouse anti-AT8 (1:50); mouse anti-human CD3 (1:50); rabbit anti-pThr555CRMP-2 (1:200)] for 48h at 4°C then followed by secondary anti-mouse (Alexa Fluor® 488) and anti-rabbit (Alexa Fluor® 555), diluted at 1:200, incubated with DAPI, mounted with fluorescent mounting medium and cover-slipped. All human tissue specimens were captured by three channel fluorescence imaging using a confocal Leica TCS NT upright or inverted microscope with the following objective lenses: UPlanApo ×20 0.75 NA, ×40 1.20 NA or ×60 1.40 NA. The captured 16-bit images were then converted to TIFF files and the images were assembled and formatted using Adobe Photoshop vCS3 software.
Mice perfused with 2.5% gluteraldehyde and 4% paraformaldehyde were dissected and the optic nerves and lumbosacral spinal cords post-fixed. Optic nerves were embedded in resin and prepared for ultrastructural analysis of demyelination by calculating G-ratios as previously described (Azari et al., 2006). Toluidine blue staining was performed and photomicrographs were then obtained from these regions using a light dotSlide™ BX51 microscope (Olympus) with a digital camera (2.3.0, RT Slider, Diagnostic Instruments, Inc.). Thickness of axons and their surrounding myelin were measured using the MetaMorph® software. At least 300 axons were measured from serial sections (n=3 sections per animal and n=3 animals per group) at a distance of 100µm apart, with values expressed as mean±SEM. Percentage of demyelination in the optic nerve was calculated from the G-ratio data by converting the basal level of the mean ratio in the adjuvant injected ngr1+/+ control to 100%, then converting the EAE-induced mouse optic nerve ratios to a percentage relative to that basal level (Azari et al., 2006).
The construction of rAAV2-Flag-CRMP2T555A-GFP and control rAAV2-GFP plasmids was generated through Vector BioLabs. Briefly, site-directed mutagenesis was performed on complementary DNA of human CRMP-2 (a gift from Professor Christina Mitchell, Monash University) as previously described (Arimura et al., 2000) using the Stratagene Quikchange® kit (Agilent Technologies) and subloned in the pRK-5-Flag vector. The Flag-CRMP2T555A complementary DNA was subcloned into the pAAV cis-plasmid, with green fluorescent protein (GFP) co-expression. An endotoxin-free, mega-prep of the pAAV cis-plasmid was then followed by transient co-transfection of the cis-plasmid with Ad Helper and RepCap plasmids into 10 plates (150cm) of 293 cells to demonstrate the fidelity of the plasmid vector. Purification of the rAAV was completed through two rounds of CsCl centrifugation and viral titres (measuring genomic copies, i.e. intact virus) were accomplished rendering 2.1×1013 genomic copies/ml for rAAV2-Flag-CRMP2T555A-GFP and 1.0×1013 genomic copies/ml for rAAV2-GFP.
Female C57Bl/6 mice (aged 10 weeks) were anaesthetized by intraperitoneal injection of 2% xylazine (7mg/kg body weight) and 10% ketamine (95mg/kg body weight). Using a finely pulled micropipette, the left eyes of mice were injected from the temporal quadrant with 1.5µl of either the rAAV2-GFP or the rAAV2-Flag-CRMP2T555A-GFP (n=8 per virally injected group, total viral particles injected per eye were 0.75×1010), directly into the vitreous (Leaver et al., 2006; Hellstrom and Harvey, 2011). Following AAV injection, all mice were left for 7 days then immunized with MOG35–55 and pertussis toxin as described above, to induce EAE. A separate group of mice (n=5), not injected with AAV, were also EAE-induced as a control group. The mice were then left to progress clinically to peak stage of disease, which did not exceed clinical score 3 and/or Day 22 post-EAE induction.
All mice (including uninjected control EAE-induced mice) were then perfusion fixed with 4% paraformaldehyde and the optic nerves and retinae collected for immunohistochemistry and axonal counts. Briefly, optic nerves were post-fixed in 4% paraformaldehyde for 24h and eyes for 3 days at 4°C, then both tissues placed in 20% sucrose overnight. Retinae were dissected from eyes and incubated as free-floating tissue with Alexa Fluor® 488 conjugated rabbit anti-GFP (1:200 dilution, Molecular Probes) in blocking buffer and mouse anti-βIII tubulin antibody (1:2000) according to immunofluorescence protocols defined above. Optic nerves were paraffin embedded, then 14µm serial longitudinal sections cut on a microtome, collecting every section. Three sections per optic nerve, at least 28µm apart, were immunostained using goat anti-GFP biotinylated antibody (1:200) then followed with either streptavidin conjugated with Alexa Fluor® 488, mouse anti-Flag antibody (1:200), rabbit anti-pThr555CRMP-2 antibody (1:500) or mouse anti-amyloid precursor protein antibody (1:500, Merck-Millipore). Secondary antibodies consisted of either goat anti-mouse Alexa Fluor® 488 or goat anti-rabbit Alexa Fluor® 555 and the optic nerve sections were processed as defined for immunofluorescence above.
Axonal degeneration was defined according to morphological abnormalities in rAAV2-tranduced, GFP-positive axons. GFP-positive clearly transected axons with retraction bulbs/spheroids were expressed as a percentage of the total number of GFP-positive axons per square millimetre of the left optic nerve.
Anti-Nogo(623–640) affinity purified antibody was generated as previously described (Karnezis et al., 2004) and concentrated at 1mg in 0.2ml for in vivo use. Three intravenous administrations of this antibody at Days 8, 11 and 14, and two intraperitoneal injections on Days 9 and 16 were performed in post-MOG35–55 EAE-induction in adult C57Bl/6 mice (n=4). A control administration of non-specific affinity purified rabbit IgG was also performed at the same time-points and routes of administration (n=4) as the anti-Nogo(623–640) antibody-treated group. Another group received no antibody administration (n=4). Clinical progression data are expressed as mean±SEM.
Data were analysed using Graph Pad Prism v3.02 software. A one-tailed Student’s t-test or a one-way ANOVA with Tukey’s post hoc test was used to determine statistical significance (P<0.05) at a 95% confidence level for optical density levels obtained from immunoblot analysis. Friedman’s non-parametric repeated measures and Dunn’s multiple-comparison were used for analysis of differences in the clinical progression of EAE between different genotypes, rAAV-injected or different antibody treatment groups of mice.
The progression of axonal degeneration and demyelination (Supplementary Fig. 1B) directly correlates with the extant motor deficit in C57Bl/6 female mice after induction of MOG35–55-EAE (Supplementary Fig. 1A). During the most disabling time-points of EAE (by Days 18 and 30, characterized as peak and chronic forms of the disease respectively; Supplementary Fig. 1A), the greatest levels of inflammation (score of 2.5±0.26 for peak, 2.75±0.43 for chronic, in the lumbosacral spinal cord; Supplementary Table 1), demyelination (score of 1.25±0.22 for peak, 2.0±0.37 for chronic in the lumbosacral spinal cord; Supplementary Table 1) and axonal degeneration/loss (Supplementary Fig. 1B) were demonstrated. Since the molecular mechanism(s) involved in axonal degeneration following spinal cord injury has been shown to involve an increase in CRMP-2 phosphorylation (Mimura et al., 2006), we investigated the levels of phosphorylated CRMP-2 during EAE-dependent axonal degeneration (Fig. 1). Following EAE induction, there was an incremental increase of pThr555CRMP-2B in spinal cord lysates when compared with adjuvant-injected controls (Fig. 1A and C). Intriguingly, there was also an increase in the level of the pThr555 N-terminal variant of CRMP-2, i.e. CRMP-2A, in EAE-induced mice that could not be demonstrated in the adjuvant injected controls (Fig. 1C).
Phosphorylated CRMP-2 was localized to axons around inflammatory lesions within spinal cord sections of EAE-induced mice, using the polyclonal pThr555CRMP-2 antibody (Fig. 1E). There were increased levels of pThr555CRMP-2 immunofluorescence co-localized with βIII-tubulin-labelled degenerative axons near inflammatory lesions (Fig. 1E).
Given that increased levels of pThr555CRMP-2 could be demonstrated in EAE mice particularly at peak of disease, we next assessed whether this specific phosphorylation event could also be demonstrated in human active demyelination and axonal degeneration as pathologically depicted in patients with multiple sclerosis. We obtained brain and spinal cord paraffin-embedded tissue samples from four chronic-active and three acute-active multiple sclerosis autopsy cases (Table 1). As shown in Fig. 2A, demyelination along with axonal degeneration (Fig. 2B and C) was apparent in the dorsolateral tracts of the lumbar spinal cord. Degenerating ventral horn motor neurons (Fig. 2D and E) were also demonstrable near inflammatory perivenular lesions in the grey matter (Fig. 2F). In acute-active demyelinating brain lesions, there were numerous tissue areas of active macrophage phagocytosis of myelin (Supplementary Fig. 2A) with extracellular myelin debris and axonal damage (Supplementary Fig. 2B). All of the active multiple sclerosis lesions were immunopositive for pThr555CRMP-2 in the brain and spinal cord (Table 1). In contrast, pThr555CRMP-2 immunoreactivity could not be demonstrated in non-neurological disease control samples or most brain samples with other neurological diseases, including demyelinating disease from progressive multifocal leukoencephalopathy specimens (Table 2). We did, however, detect immunopositivity for pThr555CRMP-2 in cerebral cortex and subcortical white matter from a sample that was collected at post-mortem with meningitis along with chronic neurodegeneration of two patients with Alzheimer’s disease. The rationale for including Alzheimer’s disease brain tissue in this study derives from the fact that hyperphosphorylation of CRMP-2 has been previously reported in brain tissue from patients with Alzheimer’s disease and specifically localized to neurofibrillary tangles (Yoshida et al., 1998; Gu et al., 2000; Cole et al., 2007; Takata et al., 2009; Williamson et al., 2011). Furthermore, we have reported that small molecular weight amyloid-β species can potentiate the phosphorylation of CRMP-2 at the Thr555 site in human SH-SY5Y cells (Petratos et al., 2008).
Acute and chronic-active demyelinated multiple sclerosis lesions in the brain and spinal cord all showed pThr555CRMP-2 positive axonal staining in the plaque core and periplaque white matter (Table 1). No pThr555CRMP-2 immunostaining was detected in chronic-inactive plaques, which showed a lack of macrophage/microglial activity with a sharp lesion border (Table 1). Subsets of grey matter NF200-positive neurons and axons present in the lumbar spinal cord from individuals with chronic-active multiple sclerosis lesions showed pThr555CRMP-2 staining (Fig. 2G–N). Axonal retraction bulbs could be seen immunostained for pThr555CRMP-2 (Fig. 2L–N). Furthermore, degenerative AT8 hyperphosphorylated tau-positive axons also showed co-localization with pThr555CRMP-2 (Fig. 2Q–Si). No co-staining could be detected for either NF200- and AT8-positive axons and neurons in the spinal cord with pThr514CRMP-2 (GSK3β phosphorylation site; Supplementary Fig. 2C–I). Hence, collectively the data suggest a direct relationship of inflammatory-mediated demyelination and neurodegeneration with pThr555CRMP-2 in neurons and axons.
We next investigated the possibility that limiting the phosphorylation of CRMP-2 during EAE will limit the progressive neurodegenerative changes that occur in these mice. We first demonstrated that following EAE-induction, the ngr1−/− mice displayed a shift in the time-point of disease onset when compared with the ngr1+/+ mice (Fig. 3A and B; Supplementary Table 2A). The ngr1−/− mice showed induction of disease by 15–16 days post-MOG35–55 immunization compared with the ngr1+/+ mice, which displayed onset of EAE by 12–13 days post-injection (Fig. 3A and Supplementary Table 2). Importantly, the majority of the ngr1+/+ mice manifested significant disease (12 of 14 mice) by 18 days post-injection, corresponding to peak neurological deficit with a clinical score of 1.93±0.23 (Fig. 3A and Supplementary Table 2). However, of the ngr1−/− EAE-immunized mice, only 6 of the 16 mice displayed disease symptoms. Moreover, peak severity of the clinical signs occurred at clinical score 1.0±0.34 by 19 days post-injection (Fig. 3A and Supplementary Table 2). This was associated with a reduction in the histopathological parameters for the ngr1−/− mice, which coincided with the progressive neurological decline post-MOG35–55-immunization (Fig. 3B and Supplementary Table 2B). Notably, there was a reduction in the score for inflammatory cell infiltrates, demyelination and axonal degeneration in the EAE-induced ngr1−/− mice (Fig. 3B and Supplementary Table 2). These parameters were associated with a reduction in the molecular marker of axonal degeneration, abnormally phosphorylated tau (hyperphosphorylated tau, AT8; Supplementary Fig. 3A). Moreover, following the isolation of the insoluble proteins from spinal cord lysates of EAE-immunized mice (Anderson et al., 2008), we found that the MOG35–55-induced ngr1−/− mice had reduced levels of AT8-positive tau, as compared with the EAE-induced ngr1+/+ mice (Supplementary Fig. 3A). These findings were associated with an increase in the soluble forms of tau in the ngr1+/+ compared with the ngr1−/− mice, immunostained with tau-5 (for total tau levels) and normal phosphorylated tau stained with an anti-pSer199/202 tau antibody (Supplementary Fig. 3A). Thus it would appear that in ngr1−/− mice, a modest increase in the phosphorylated and hyperphosphorylated forms of tau correlates with axonal degeneration and disease progression.
To further assess the extent of axonal degeneration, we performed ultrastructural analysis on lumbosacral spinal cords of both ngr1−/− and ngr1+/+ mice. As illustrated in Fig. 3C, lesions appearing in the white matter tracts of the spinal cord of ngr1+/+ mice showed significant axonal degeneration and demyelination with inflammatory infiltrates (Fig. 3C). However, in the spinal cord white matter tracts of ngr1−/− mice, there was preservation in axo-myelin integrity along with limited inflammatory infiltrates (Fig. 3C).
The reduced axonal degeneration together with the abrogated neurological decline in the ngr1−/− mice following EAE-induction, prompted us to investigate whether spinal cord axons in these mice displayed a reduced propensity to initiate phosphorylation of CRMP-2. Immunofluorescence analysis showed that in the ngr1−/− mice there were reduced levels of pThr555CRMP-2 in axons in the longitudinal white matter tracts during EAE (Supplementary Fig. 3B). We showed that neurofilament NF200-positive axons (particularly those showing axonal-spheroids) co-localized significantly with pThr555CRMP-2 immunofluorescence at peak and chronic stages of EAE in ngr1+/+ mice (Supplementary Fig. 3B). These data suggest that during the progression of EAE there are greater levels of pThr555CRMP-2 in degenerative axons in the spinal cords of ngr1+/+ compared with ngr1−/− mice.
In order to quantitate these changes, we performed co-immunoprecipitation and western blotting from spinal cord lysates of ngr1+/+ and ngr1−/− mice during EAE. We found that the ngr1+/+ mice had an incremental increase in pThr555CRMP-2 levels, compared with ngr1−/− mice during the progression of EAE (Fig. 3D and E). This increase was also evident in the 75kDa N-terminal variant of the protein, which did not appear in the ngr1−/− mice during EAE (Fig. 3D). Importantly, the levels of CRMP-2-associated tubulin were reduced in the ngr1+/+ but not in the ngr1−/− mice during EAE (Fig. 3F and G). The data can be interpreted as increased dissociation of monomeric tubulin from CRMP-2 in the ngr1+/+ [from 19.75±2.83% at pre-onset (P<0.001) to 33.77±3.187% at peak (P<0.001); Fig. 3G] as the levels of CRMP-2 phosphorylation increase during the progression of EAE disease [from 39.16±49.62% in adjuvant injected ngr1+/+ mice to 382±94.41% at peak (P<0.05); Fig. 3E]. However, deletion of the ngr1 gene limits the EAE-dependent phosphorylation (Fig. 3D and E), replenishing the tubulin association levels (Fig. 3F and G).
To obtain a direct and quantitative measurement of demyelination and axonal pathology, we stained semi-thin optic nerve cross-sections with toluidine blue and assessed both axon/myelinated fibre diameter and degeneration/loss, between EAE-induced ngr1−/− and ngr1+/+ mice. Toluidine blue staining clearly demonstrated axo-myelin degeneration in a number of axons from pre-onset and onset stages of EAE in the ngr1+/+ but not in the ngr1−/− mice (Fig. 4A and Supplementary Fig. 5). As indicated by the most severe axonal degeneration/loss, the greatest ultrastructural abnormalities were found at peak and chronic stages of EAE, in ngr1+/+ mice (Fig. 4A). Optic nerve cross-sections were also investigated for lamellated myelin structure, relative to ensheathed large and small calibre axons, by calculating G-ratios (Fig. 4B). There was a significant, incremental increase in the optic nerve G-ratios of EAE-induced ngr1+/+ mice with disease progression (from 0.768±0.004 at pre-onset to 0.799±0.004 at peak; Fig. 4B). However, in ngr1−/− mice, there was a moderate but comparable increase in the G-ratio at pre-onset of disease (from 0.761±0.003 at pre-onset), but neared basal levels by peak (0.705±0.003; Fig. 4B). The overall increase in demyelination was 11.43±0.597% in adjuvant injected controls (basal level) from the optic nerves of EAE-induced ngr1+/+ mice by peak of disease, compared with a −1.651±0.472% change in demyelination in EAE-induced ngr1−/− mice at the same time-point (Fig. 4B). However, there was demonstrable demyelination in the optic nerves of ngr1−/− mice at the earliest investigated time-point (Day 7 post-EAE induction), resembling that seen in the ngr1+/+ mice (6.123±0.437% versus 7.096±0.490%, respectively). Interpretation of the data would be that there is an initial thinning of lamellated myelin surrounding axons in the optic nerves in both genotypes that is eventually repaired in the ngr1−/− mice by Days 12 to 18 (onset and peak, respectively).
However, we did find an increase in the axonal diameter without compensatory increases in the thickness of myelin that normally occurs during development, with larger calibre axons (ffrench-Constant et al., 2004). To exclude the possibility that the optic nerves of ngr1−/− mice exhibited thinly myelinated but larger calibre axons, we plotted the axonal diameter measurements during EAE, of the ngr1+/+ and ngr1−/− mice (Fig. 4B). We found that in ngr1−/− mice, the axonal calibre was at basal levels at Day 7, post-EAE-induction (Fig. 4B). Thus it would appear that the increase in G-ratio in the ngr1−/− mice at this early time-point of EAE is related to myelin thinning of axons but is then repaired by 12–18 days post-injection (Fig. 4B). There was an incremental rise in the axonal diameter in the ngr1+/+ mice as EAE progressed, with a sharp rise by peak of disease (from 25.23±0.21 (Arbitrary Units) at Day 7 to 27.42±0.41 by Day 18 post-EAE induction; Fig. 4B). However, in the optic nerves of ngr1−/− mice, there was only a modest increase in axonal diameters by peak stage of EAE, which was just above baseline (26.06±0.19 (Arbitrary Units); Fig. 4B). These data suggest limited axonal swelling in the optic nerves of the ngr1−/− mice during EAE.
Highlighting the neurobiological role of the ngr1 gene, there was an enhanced viability of axons in the optic nerves of ngr1−/− mice during EAE (Fig. 4C). We showed that there were no differences in axonal loss in the optic nerves of ngr1+/+ and ngr1−/− mice during the pre-onset and onset stages of EAE (Fig. 4C), although axons had swollen in the ngr1+/+ EAE-mice (Fig. 4B). However, by peak stage of EAE, we saw a dramatic drop-out of optic nerve axons in the ngr1+/+ mice (16048±2316 myelinated axons per square millimetre) that was not replicated in the ngr1−/− mice (30818±3141 myelinated axons per square millimetre, P<0.01; Fig. 4C). Thus we can conclude that there is a preservation of axons in the ngr1−/− mice at the peak stage of EAE, correlating with myelin maintenance and a decrease in axonal swelling (Fig. 4B).
We further demonstrated that there was an increase in the phosphorylation of CRMP-2 in the optic nerves of ngr1+/+ mice by peak stage of EAE (Fig. 4D). These data correlated with significant levels of axo-myelin degeneration and loss in the ngr1+/+ mice (Fig. 4B and C), showing an ~2-fold increase of CRMP-2 phosphorylation when compared to the ngr1−/− mice at peak stage of EAE (Fig. 4D). The emergence of the 75kDa pThr555-CRMP-2A band in the ngr1+/+ optic nerves was again present with EAE-induction but absent in ngr1−/− mice (Fig.4D). The levels of tubulin association with CRMP-2 were also reduced in the optic nerves of the ngr1+/+ mice, correlating with the phosphorylation status of CRMP-2 (Fig. 4D). These data implicate the phosphorylation of CRMP-2 with axonal degeneration during the progression of EAE in the optic nerve of ngr1+/+ mice and suggest that deletion of the ngr1 gene limits this molecular mechanism of axo-myelin degeneration from ensuing, in the context of neuroinflammation.
Having established that the phosphorylation of CRMP-2 could be reduced in ngr1−/− mice at peak stage of EAE, we then addressed the question whether an abrogation in this phosphorylation may be protective against neuroinflammatory-mediated axonal damage. A rAAV2 expressing both a site-specific mutant form of CRMP-2 (Flag-T555ACRMP2) and GFP or AAV2-GFP control (Fig. 5A) were intraocularly injected into the left temporal eye quadrant of female wild-type C57Bl/6 mice (n=8 per group) and then EAE induced after 1 week to allow for efficient transduction of retinal ganglion cells. The mice developed disease onset and progression to peak stage of disease at least by Day 21, post-EAE induction (Fig. 5B). We achieved an average of ~40% rAAV2-transduction of each retina with either vector (Fig. 5C–E). GFP-labelled optic nerve axons near inflammatory lesions (Fig. 5Fiii and Supplementary Fig. 6B) could be observed to be degenerative in EAE-induced mice, injected intraocularly with rAAV2-GFP (Fig. 5Fi). Many of these degenerative, GFP-expressing axons also demonstrated pThr555-CRMP-2 co-localization (Fig. 5Fii, iii and Supplementary Fig. 6B). In the rAAV2 Flag-T555ACRMP2-GFP transduced retinal ganglion cells, optic nerve axons did not display significant degeneration (Fig. 5G and Supplementary Fig. 6C) or pThr555CRMP-2 co-labelling (Fig. 5G and Supplementary Fig. 6C) even though these axons resided near inflammatory lesions (Fig. 5G and Supplementary Fig. 6C). However, non-transduced axons near inflammatory lesions did display pThr555CRMP-2 localization (Fig. 5G). Analysis of the numbers of transduced axons demonstrating extant axonal degeneration (axonal spheroids/retraction bulbs) (Fig. 5I) showed that there were an estimated 185±24 degenerative axons of the total number of 1106±2258 transduced axons per square millimetre in the rAAV2-GFP optic nerves (Fig. 5H). This compared to an estimated 28±10 degenerative axons of the total number of 1557±383 transduced axons per square millimetre in the rAAV2 Flag-T555ACRMP2-GFP optic nerves (Fig. 5H). These data translate to an average of 20±3% and 2±1% of degenerative transduced axons in rAAV2-GFP and rAAV2 Flag-T555ACRMP2-GFP optic nerves, respectively (P=0.008, Fig. 5H). Moreover, co-labelling of these axons with an anti-amyloid precursor protein antibody showed uniform amyloid precursor protein localization of rAAV2 Flag-T555ACRMP2-GFP transduced axons (Fig. 5L), suggesting an intact fast axonal transport mechanism in the transduced axons. However, the fidelity of the fast axonal transport of amyloid precursor protein was clearly compromised in the rAAV2-GFP transduced optic nerve axons, with accumulation of amyloid precursor protein in distended axonal segments (Fig. 5K). Furthermore, the accumulation of amyloid precursor protein could also be demonstrated to co-localize with pThr555CRMP-2 in non-transduced optic nerves from EAE-induced mice at peak stage of disease (Fig. 5L, arrow). These data suggest that rAAV2 transduction of retinal ganglion cells with a phospho-mutant form of CRMP-2 (T555ACRMP2), during the peak stage of EAE disease, can limit the axonal degeneration in the optic nerve.
Our previous work has supported the notion that inhibition of Nogo-A in EAE limits axonal degeneration and disease progression (Karnezis et al., 2004). Therefore, we next ascertained whether the mechanism governing the limitation of axonal degeneration in Nogo-A passively vaccinated EAE-induced mice (Karnezis et al., 2004) was because of a limitation in phosphorylation of CRMP-2. We performed the therapeutic administration of the anti-Nogo(623–640) antibody in female wild-type mice following the induction of MOG35–55-EAE (Fig. 6A). A significant delay in the onset of EAE clinical signs could be demonstrated in the anti-Nogo(623–640) antibody treated mice (at 21 days post-induction) when compared with EAE-mice that were treated with a non-specific IgG isotype control or without passive antibody treatment (Fig. 6A). Furthermore, the severity of EAE was also significantly blunted after continuous treatment with anti-Nogo(623–640) antibody (average clinical score of 0.6±0.6 by Day 25 post-induction) compared to the EAE-mice continuously treated with the IgG control antibody (1.8±0.5 by Day 25 post-induction) or without treatment (2.4±0.5 by Day 25 post-induction; Fig. 6A). By the chronic stage of EAE, we also observed a difference in the level of axonal degeneration and demyelination in the spinal cords of the anti-Nogo(623–640) antibody-treated mice compared to the EAE-mice, continuously treated with the IgG control antibody or without treatment (Fig. 6B). Moreover, the abrogation of neurological decline and axonal degeneration demonstrated in the anti-Nogo(623–640) antibody-treated mice correlated with a significant reduction in the level of pThr555CRMP-2 in the spinal cord (Fig. 6C and D). The decrease in the levels of pThr555CRMP-2 correlated with a replenished level of tubulin association with CRMP-2 (Fig. 6C and D). Collectively, these data suggest that the reduction in the neurological decline and associated axonal pathology exhibited by the anti-Nogo(623–640) antibody treatment of EAE-induced mice may be related to a decrease in the phosphorylation status of CRMP-2.
Despite the numerous investigative reports of NgR1 signalling in the CNS during disease and injury, there exists debate as to whether the limitation in axonal damage or enhanced regrowth potential is regulated through this mechanism. The current study supports the contention that limiting the NgR1-dependent phosphorylation of CRMP-2 (which normally functions to regulate axonal microtubule assembly), axonal degeneration and hence neurological impairment can be halted in EAE. We have demonstrated that phosphorylated CRMP-2 is localized only in degenerative neuronal somata and axons in chronic-active multiple sclerosis lesions, where myelin damage and inflammatory cells coincide. Our study advocates that targeting NgR1-signalling in EAE, and by extension multiple sclerosis, may be a feasible therapeutic approach to limit the neurodegenerative effects of these conditions.
Current therapeutics in multiple sclerosis target the inflammatory nature of the disease in an attempt to limit the ‘autoimmune attack’ on CNS myelin and, as a consequence, reduce the devastating neurological complications that characterize this disease. However, comprehensive neuropathological investigations from a vast array of multiple sclerosis brain samples suggest a degree of heterogeneity in the pathogenesis, with some lesions showing tangible neurobiological origins (Lucchinetti et al., 2000). Both the ‘outside-in’ and ‘inside-out’ hypotheses of demyelination and axonal degeneration involve dysregulation of axo-glial signalling, driving axonal degeneration potentiating neurological decline and establishing disability.
Using the exon 2 nogo receptor null mutant mice bred on a C57Bl/6 background (Kim et al., 2004), we report that the signalling mechanism that is operative during the neurodegenerative phase of EAE relies on NgR1-dependent phosphorylation of CRMP-2. A significant increase in the phosphorylation of CRMP-2 at the Thr555 site, specific for Rho-associated, coiled-coil containing protein kinase 2 (ROCK2) (Arimura et al., 2000; Arimura and Kaibuchi, 2005; Mimura et al., 2006), was demonstrated to be operative in EAE. This mechanism has been previously identified to be important in neurite/axon retraction events in culture (Arimura et al., 2000; Arimura and Kaibuchi, 2005). Similarly, pThr555-CRMP-2 has been demonstrated to be responsible for axonal degeneration in an acute model of spinal cord injury (Mimura et al., 2006). Our finding that levels of pThr555-CRMP-2 in the spinal cord and optic nerve of ngr1+/+ but not ngr1−/− mice, incrementally increased during the clinical and pathological progression of EAE, is critical in defining a neurobiological basis for this signalling cascade during neuroinflammatory-mediated degeneration.
There is strong evidence that phosphorylation of CRMP-2, either by GSK-3β, Cdk-5 or ROCK2, inactivates the normal physiological functions of CRMP-2. Physiologically, CRMP-2 can associate with α- and β-tubulin heterodimers (Arimura and Kaibuchi, 2005; Arimura et al., 2005), kinesin-1 motor proteins (Kimura et al., 2005), TrkB and the Sra/WAVE1 complex, to facilitate anterograde axonal transport, and thereby growth, repair, maintenance and communication (Kawano et al., 2005; Arimura et al., 2009). Disruption of transport of this molecular cargo is integral for the interruption in the maintenance of the axonal terminal and by extension, synaptic connectivity (Arimura et al., 2005, 2009; Kawano et al., 2005; Kimura et al., 2005; Petratos et al., 2010). The consequential loss of axo-dendritic connectivity, mediated through various negative signal transduction events, leads to actin and tubulin depolymerization and eventual axonal retraction (Schmidt and Rathjen, 2010). However, the question remains as to whether these events can be initiated during neuroinflammatory diseases such as EAE and multiple sclerosis.
Our data suggest that pThr555CRMP-2 is involved as a neurodegenerative molecule of axons and neurons in active multiple sclerosis lesions and that this signal can be limited by ablation of the NgR1 gene or by therapeutically targeting Nogo-A, during acute neuroinflammation, without any immune-dependent mechanisms being directly responsible for this signalling. However, we did observe that the phosphorylation of CRMP-2 was reduced at the chronic stage of EAE, a stage where neurodegeneration continues to progress. We are currently investigating this finding to delineate whether there is a clearance of extracellular myelin-associated inhibitory factors or NgR1 expression changes at the chronic stage of EAE. However, one line of investigation may derive from data that show that CRMP-2 is cleaved by calpain during overt neurodegeneration, forming a 55kDa truncated form (Taghian et al., 2012). Such forms of CRMP-2 have been linked to neuronal cell death in various models of neurodegeneration (Taghian et al., 2012). Despite the uncertain mechanism occurring during the chronic stage of EAE, we importantly showed that the downstream target of NgR1 signalling, the phosphorylation of CRMP-2 at the Thr555 site, can be abrogated in retinal ganglion cell axons by transduction using a rAAV2 Flag-T555ACRMP2-GFP. The limitation of this signalling cascade during EAE-induced optic neuritis maintains the fidelity of these optic nerve axons and their fast axonal transport mechanisms.
The CNS tissue milieu during injury or disease is rich in axonal outgrowth inhibitors such as Nogo-A, potentiating axonal retraction events and blocking regeneration (Profyris et al., 2004). It has recently been shown that limiting NgR1 signalling in the MOG35–55-induced or the B10.PL models of EAE, by using small interfering RNA knock-down of Nogo-A, can ameliorate the disease severity and enhance axonal repair (through increase in GAP-43-positive axons within the spinal cord) without altering myelin-specific T cell proliferation and cytokine production (Yang et al., 2010). Our study extends these findings by pinpointing the signalling mechanism responsible for axonal degeneration in EAE and has demonstrated that downstream pThr555CRMP-2 is present in degenerative neurons and axons of chronic-active multiple sclerosis lesions.
It was recently discovered that the co-receptor of NgR1, Lingo-1, can promote demyelination in the MOG35–55 EAE murine model (Mi et al., 2007). These investigators showed that in MOG35–55-induced lingo-1−/− mice or wild-type EAE-induced mice treated with a function-blocking anti-Lingo-1 antibody, demyelination was limited thereby blunting the neurological decline. Although these findings implicate the NgR1 signalling complex in the promotion of demyelination during neuroinflammation, our data suggest that targeting the NgR1 high-affinity receptor for myelin associated inhibitory factor-binding limits axonal degeneration during the onset and peak stages of EAE. However, at the early pre-onset stage of EAE, an equivalent proportion of demyelination in the optic nerves of both the ngr1−/− wild-type littermate mice was demonstrated. This initial thinning of myelin (increasing G-ratio) does not persist with progression of disease in the ngr1−/− mice, suggesting that possible remyelination may in fact be preserving axons at the early stage of neuroinflammation (Harrington et al., 2010). Despite these data, we believe that by limiting the axonal phosphorylation of CRMP-2 during neuroinflammation, we can preserve the axons during the peak stage of EAE as we demonstrated in the optic nerve following the rAAV2 Flag-T555ACRMP2-GFP transduction of retinal ganglion cells. These in vivo data for the first time demonstrate that inhibition of the phosphorylation of CRMP-2 can prevent neuroinflammatory-mediated axonal degeneration. We not only demonstrated the limitation of axonal degeneration through the introduction of a site-specific phosphorylation mutant of CRMP-2 but also showed that when the capacity of Nogo-A to bind to its cognate receptor(s) is blocked, the inhibition of axonal degeneration in EAE could limit the phosphorylation of CRMP-2. This suggests that abrogating Nogo-A/NgR1 binding prevents axonal degeneration by blocking CRMP-2 phosphorylation.
Given the recent implication that NgR1 is involved in the efflux of activated macrophages from injured peripheral nerves (Fry et al., 2007), we wanted to investigate the possibility of a putative immune role for NgR1 during CNS inflammation. We found no discernible difference in the proliferation response of T cells isolated from ngr1−/− and ngr1+/+ mice at pre-onset, peak and chronic stages of EAE (7, 18 and 30 days post-injection) in response to the MOG35–55 peptide (Supplementary Fig. 4). Furthermore, we were unable to demonstrate differences in cytokine profiles from EAE-induced ngr1−/− (12 and 18 days post-injection) compared with ngr1+/+ mice. However, when we analysed the populations of CD3+/CD8+, CD3+/CD4+ and B220 lymphocytes, Gr-1+, F4-80+ and CD-11c+ granulocytes in the spleen and CNS of ngr1−/−, we could not demonstrate differences in these populations of immune-lineage cells from ngr1+/+ mice 18 days post-EAE immunization (data not shown). The inference would therefore be that the potential for immune-mediated induction of disease is not altered in the ngr1−/− mice. Thus, it would appear that the limitation of the neurological decline in the ngr1−/− mice is dependent on a neurobiological mechanism. Despite this, recent data show that monocytes, T and B cells isolated from patients with multiple sclerosis express NgR1 (Pool et al., 2009). Given that stimulation of immune cells with Nogo led to an alteration of their adhesion properties, it is possible that in immune cells, NgR1 signalling regulates their cytoskeletal dynamics through RhoA activation (Pool et al., 2009). Therefore, further investigation is required to implicate or exclude a role of NgR1 in immune cell activation and migration in multiple sclerosis.
Current evidence defines pharmacological blockade of myelin associated inhibitory factors as a means of enhancing regeneration in the CNS (McGee and Strittmatter, 2003; Zorner and Schwab, 2010). Blocking the signalling of Nogo-A in CNS disease and injury has been well studied and includes EAE (Karnezis et al., 2004; Mi et al., 2007). In this study, we used the anti-Nogo(623–640) antibody, previously demonstrated to promote axon outgrowth in the presence of the Nogo peptide, as well as the prevention of neurological decline when injected intravenously over the course of EAE (Karnezis et al., 2004). We showed that after four injections of anti-Nogo(623–640) antibodies, the decreased severity of EAE correlated with a decreased amount of axonal degeneration. Importantly, reduced levels of pThr555-CRMP-2 in these animals corresponded with replenished tubulin association.
Function-blocking antibodies of Nogo-A have been successfully used in preclinical experiments enhancing locomotor performance in neurotrauma paradigms in rodents and adult monkeys (Schnell and Schwab, 1990; Freund et al., 2006), improving manual dexterity after a cervical spinal cord injury in the latter (Freund et al., 2006). Over the space of two decades, these significant studies advocating therapeutic targeting of Nogo-A in neurotrauma have encouraged the Phase I clinical trials of anti-human Nogo-A antibody (ATI 355) by Novartis (Kwon et al., 2010). With 45 patients enrolled in this study, intrathecal administration of the antibody in these patients has shown remarkable tolerance with no side-effects manifest in this group (Kwon et al., 2010). With a Phase II trial currently underway, we now propose that one of the major mechanisms by which anti-Nogo-A antibody treatment is clinically effective may well be through the eventual decrease of the phosphorylated levels of CRMP-2 limiting NgR1-dependent axonal degeneration. Interestingly, other pharmacological studies using the ectodomain of NgR1 and co-receptors such as Lingo-1 have recently shown enhanced neurological improvement in injury and disease models of the CNS by blocking the binding of the myelin associated inhibitory factors (Lee et al., 2004; Li et al., 2004; Wang et al., 2006). It is tantalizing to suggest that similar therapeutic effects could be met in the treatment of multiple sclerosis.
This study has provided new insights into the molecular mechanisms that govern axonal degeneration during inflammatory demyelination of the CNS (as occurs in multiple sclerosis) as well as offering novel therapeutic strategies for this disease. Since myelin damage is a dominant event in neurodegeneration, the work presented here may provide novel therapeutic avenues to attenuate axonal abnormalities, which accompany demyelination in a variety of disorders such as spinal cord and brain injury, stroke and multiple sclerosis. We suggest that blockade of NgR1 signalling and reduction of CRMP-2 phosphorylation limits axonal degeneration in EAE, allowing for the normal physiological function of CRMP-2 to ensue, and by virtue, limiting the axonal degeneration and alleviating clinical symptoms of EAE.
National Health and Medical Research Council (NHMRC) of Australia [384157, to S.P., 380832 and 436634 to C.C.A.B.]; National Multiple Sclerosis Society of New York [RG 4398A1/1 to S.P. and R.G., 3844A2/1 to C.C.A.B.]; The Baker Foundation [C.C.A.B.]; NHMRC Allied Health Postdoctoral Training Fellowship [M.F.A.]; Multiple Sclerosis Research Australia Vacation Scholarship [K.T.], McDonald Fellowship multiple sclerosis International Federation [S.L.]; Krembil Chair in Neural Repair and Regeneration [M.G.F.].
Supplementary material is available at Brain online.
Authors would like to acknowledge the assistance of Monash MicroImaging (MMI) for the provision of microscopy instrumentation, training and in particular Stephen Firth, Dr. Judy Callaghan and Dr. Alex Fulcher for their scientific and technical assistance. Authors thank the assistance of Associate Professor Michael F Gonzales, Consultant Neuropathologist, The Royal Melbourne Hospital, Australia, for his assistance in defining the neuropathological cases presented in this manuscript. They would also like to thank Professor Christina Mitchell and Dr. Lisa Ooms (Monash University), as well as Dr. Adam Cole (Garvan Institute of Medical Research) for providing the pRK-5 plasmid CRMP-2 vector. Finally, the authors acknowledge Professor Paul McMenamin for his help in training staff and students in the retinal dissections.