OUP user menu

Remyelination protects axons from demyelination-associated axon degeneration

K. A. Irvine, W. F. Blakemore
DOI: http://dx.doi.org/10.1093/brain/awn080 1464-1477 First published online: 18 May 2008


In multiple sclerosis, demyelination of the CNS axons is associated with axonal injury and degeneration, which is now accepted as the major cause of neurological disability in the disease. Although the kinetics and the extent of axonal damage have been described in detail, the mechanisms by which it occurs are as yet unclear; one suggestion is failure of remyelination. The goal of this study was to test the hypothesis that failure of prompt remyelination contributes to axonal degeneration following demyelination. Remyelination was inhibited by exposing the brain to 40 Gy of X-irradiation prior to cuprizone intoxication and this resulted in a significant increase in the extent of axonal degeneration and loss compared to non-irradiated cuprizone-fed mice. To exclude the possibility that this increase was a consequence of the X-irradiation and to highlight the significance of remyelination, we restored remyelinating capacity to the X-irradiated mouse brain by transplanting of GFP-expressing embryo-derived neural progenitors. Restoring the remyelinating capacity in these mice resulted in a significant increase in axon survival compared to non-transplanted, X-irradiated cuprizone-intoxicated mice. Our results support the concept that prompt remyelination protects axons from demyelination-associated axonal loss and that remyelination failure contributes to the axon loss that occurs in multiple sclerosis.

  • ageing
  • axonal degeneration
  • oligodendrocyte progenitor cells
  • multiple sclerosis
  • remyelination
  • stem cells


Multiple sclerosis is an inflammatory autoimmune demyelinating disorder of the CNS, and with a lifetime risk of one in 400 is the most common neurological disorder in young adults. Historically, research has focused on demyelination and remyelination; however, it is becoming increasingly evident that axonal degeneration is the major contributor to the progressive disability seen in multiple sclerosis (Ferguson et al., 1997; Trapp et al., 1998; Bitsch et al., 2000; Kornek et al., 2000; Kuhlmann et al., 2002). The proposed mechanisms of axonal degeneration in multiple sclerosis are considered to vary depending on the stage of the disease. During the early stages of the disease, axonal injury has been correlated with the magnitude of the inflammatory response (Trapp et al., 1998). Activated microglia/macrophages have been consistently found in close association with dying axons and experimental evidence suggests that the axonal injury is caused due to bystander damage associated with inflammatory mediators such as tumour necrosis factor α (TNF-α), proteases, excitotoxins and nitric oxide playing a key role (Anthony et al., 1997; Smith and Lassmann, 2002). In the later stages of the disease when inflammation is no longer a feature of the demyelinated lesions other mechanisms must be contributing to the ongoing axonal damage.

Neuroprotective strategies are an area of current interest in multiple sclerosis research and possible approaches include anti-excitotoxic agents (Pitt et al., 2000; Rosin et al., 2004), nitric oxide and iNOS inhibitors (Okuda et al., 1997), anti-oxidants (Gilgun-Sherki et al., 2003), Ca2+ channel blockers (Brand-Schieber and Werner, 2004, Na+ channel blockers (Waxman, 2002) or Na+/Ca2+ exchanger inhibitors and growth/neurotrophic factors (Webster, 1997; Linker et al., 2002). An alternative or additional neuroprotective strategy could be enhancement of remyelination since Kornek and co-workers (2000) noted that there was significantly less evidence of axonal injury in remyelinated multiple sclerosis shadow plaques compared to inactive demyelinated lesions. An observation suggesting that remyelination had protected axons from further degeneration. In support of this interpretation, transplant-mediated remyelination in the mouse hepatitis virus model of demyelination resulted in axon sparing and a recovery of function of locomotor abilities (Totoiu et al., 2004).

A protective effect of remyelination could involve a number of mechanisms. The myelin sheath may simply act as a ‘barrier’ to toxic mediators during inflammation (Redford et al., 1997). An additional mechanism could be the provision of trophic support for the axon. Such a mechanism is suggested by studies using myelin mutants which show progressive axon degeneration in the absence of inflammation as they age (Griffiths et al., 1998; Lappe-Siefke et al., 2003) and tissue culture studies which identify oligodendrocyte-derived factors such as IGF-1 (Wilkins et al., 2001), GDNF (Strelau and Unsicker, 1999; Wilkins et al., 2003) and NGF (Byravan et al., 1994) which promote the survival of CNS neurons. Oligodendrocytes have also been shown to regulate slow anterograde transport rates (De Waegh et al., 1992; Kirkpatrick and Brady, 1994; Kirkpatrick et al., 2001), and more recently, absence of PLP/DM20 in oligodendrocytes was shown to impair fast axonal transport (Edgar et al., 2004). Alterations in axonal transport have been suggested as a common mechanism for axonal degeneration in a range of neurodegenerative diseases including hereditary spastic paraplegia (Crosby and Proukakis, 2002), Alzheimer's disease (Hiruma et al., 2003) and Huntington's disease (Gunawardena et al., 2003; Szebenyi et al., 2003).

To date there is no direct evidence that remyelination protects demyelinated axons from degeneration. In a previous study (Irvine and Blakemore, 2006), we demonstrated an age and strain effect on the incidence of demyelination-associated axon degeneration in cuprizone intoxication and we speculated that the age effect could be related to prolonged exposure of demyelinated axons to inflammatory mediators resulting from the slowing of the rate of remyelination that occurs with age (Shields et al., 1999; Hinks and Franklin, 2000; Sim et al., 2002). To test this hypothesis, we decided to adopt a two-pronged approach. First, we would examine whether inhibiting remyelination, which can be achieved by exposing tissue to 40 Gy of X-irradiation (Blakemore and Patterson, 1978; Hinks et al., 2001), would be associated with an increase in axon degeneration. Then, since X-irradiation may itself cause axon degeneration we would control for this by restoring the remyelination capacity of the X-irradiated brain by reconstituting the oligodendrocyte progenitor population by transplantation prior to inducing demyelination. Furthermore, by using old mice as recipients and transplanting embryo-derived progenitors which would remyelinate more quickly than the host cells, we could examine whether reducing the length of time demyelinated axons were exposed to the inflammatory environment would decrease the extent of axon degeneration. By adopting this approach, we hoped to show that rapid remyelination protects axons from demyelination-associated axon degeneration.

Materials and Methods

Animals and experimental plan

Sixty-two female ex-breeder C57Bl/6 mice (6 months or older) were obtained from Harlan UK Limited (Bicester, UK) (Fig. 1). Eighteen mice were randomly selected and their brain was exposed to 40 Gy of X-irradiated to deplete OPCs (Irvine and Blakemore, 2007) and thereby inhibit endogenous remyelination of the corpus callosum. To restore the remyelinating capacity removed by the X-irradiation, these mice were transplanted bilaterally with 2 μl of embryonic intact GFP+ neurospheres at a concentration of 150 000 cells/μl into each lateral ventricle one week later. Both the transplanted and non-transplanted mice were placed on a powdered diet containing 0.4% cuprizone. During the third week on the diet, 18 randomly selected non-transplanted mice and four normal mice were cranially X-irradiated with 40 Gy of irradiation. After a further 5 weeks on the cuprizone diet, half the transplanted mice, and 18 non-transplanted mice (nine irradiated and nine non-irradiated mice) were chosen at random from each treatment group and perfused with either 4% glutaraldehyde for resin embedding (three from each group) or 4% paraformaldehyde (PFA) for immunohistochemistry (six from each group). The remaining mice (nine transplanted, nine non-transplanted and X-irradiated, nine non-irradiated) were returned to a normal diet and perfused after a further 7 weeks. Four normal X-irradiated and four control mice were perfused with PFA for immunohistochemistry.

Fig. 1

Experimental design. A flow diagram showing the number, treatment and sacrifice times of the mice used in the experiment.


For X-irradiation, mice were anaesthetized with an i.p. injection of a mixture of 0.75 ml of fentanyl citrate/fluanisone (Hypnorm, Janssen Pharmaceuticals, New Jersey, USA) 0.5 ml of Hyponovel (10 mg/2 ml Roche) and 0.75 ml of injectable water (Norbrook Laboratories Ltd) at a dose rate of 0.05 ml/10 g body weight. When anaesthetized, mice were carefully placed in lateral recumbency and held in place by Durapore tape (3 M UK plc, Bracknell, UK). Five-millimetre thick lead shielding was positioned so as to expose the brain caudal to the eyes, dorsal to the palate and rostral to the ears. The positioning of this shielding was checked by taking a radiograph and appropriate repositioning carried out so that the eyes, ears and palate were protected and that the rostro-caudal width of the unprotected region measured 5 mm. When correct positioning had been achieved, the brain was exposed to 40 Gy using an orthovoltage (225 kV) source (Pantak, Gulmay Medical Ltd, Camberley, UK).

Isolation and cultivation of neurospheres

Mice with enhanced green fluorescent protein (EGFP) under control of a chicken b-actin promoter (C57Bl/6-Tg(ACTB-EGFP)1Osb/J, The Jackson Laboratory, Maine, USA) were mated with wild type C57Bl/6 mice and killed at embryonic Day 14 (E14). Forebrains from EGFP-transgenic mouse embryos were mechanically dissociated and cells were cultured at a density of 250 000 cells/ml in serum-free medium containing epidermal growth factor (EGF, R and D Systems Europe Ltd, Oxon, UK) and basic fibroblast growth factor (bFGF, R and D Systems Europe Ltd, Oxon, UK) (both at 20 ng/ml). After 7 days of culture, cells had grown to free-floating neurospheres. For passaging, cultures were centrifuged, spheres were enzymatically dissociated with accutase and cells were re-suspended in the same medium at a density of 100 000 cells/ml.

Transplantation of neurospheres

Three days following their second passage (a total of 17 days in vitro) EGFP neurospheres were collected and the number and viability of the intact spheres quantified. The spheres were then re-suspended in sterile MEM (Sigma–Aldrich. Poole, UK) to a concentration of 150 000 cells/μl and placed on ice until transplantation. For transplantation the scalp was opened, a hole was drilled and 2 μl of cell suspension was slowly (over a period of 3 min) injected bilaterally into the lateral ventricles at the following positions (coordinates from Bregma in millimetre) anteroposterior 0.0 mm; mediolateral 1.0 mm; dorsoventral 2.0 mm.

Tissue processing

Following fixation with glutaraldehyde, the brain was cut down the midline to separate the two hemispheres. One hemisphere was sampled coronally from the area between 165 and 195 in the mouse brain atlas (http://www.hms.harvard.edu/research/brain) and trimmed to leave a tissue block that contained the corpus callosum in coronal section. The other hemisphere was sampled sagittally by cutting a 1 mm thick section starting from the olfactory bulbs to the brainstem. This block was then trimmed to leave a tissue block that contained the entire corpus callosum in sagittal section. These blocks were processed into resin and sections cut at 1 µm were stained with alkaline toluidine blue for light microscope examination. Following fixation with PFA, the brain was removed and cut along the midline fissure to separate the two hemispheres and post-fixed in PFA for 2 h. The two hemispheres were immersed in 30% sucrose solution (in 0.1 M phosphate buffer) at 4°C until they sunk, following which they were placed in mounting media [a 50:50 solution of 30% sucrose solution and OCT (RA Lamb Lab supplies, Eastbourne, UK)] for 2 h before rapidly freezing one hemisphere for coronal sectioning and the other for sagittal sectioning in a cryomould using dry ice. Starting at the olfactory bulb, 25 μm thick sections were cut from the block mounted for coronal sectioning at −25°C with a cryostat until area 165 of the mouse brain atlas (http://www.hms.harvard.edu/research/brain) was reached. From this point on 10 μm thick transverse sections of the brain were cut and mounted onto polylysine-coated glass slides (VWR international Ltd, Lutterworth, UK). Ten-micrometre sections were collected from the midline surface of the block mounted for sagittal sectioning. The slides were stored at −70°C until used.


To analyse the in vitro differentiation potential of neurospheres a sample of the neurospheres that were transplanted into the mice was cultured for 7 days on poly-l-lysine-coated plastic chamber slides using a medium in which the mitogens were replaced by 1% fetal calf serum (FCS; Sigma–Aldrich, Poole, UK). To identify the different cell types the following primary antibodies were used; monoclonal mouse antibodies to β-Tubulin type III (Sigma–Aldrich. Poole, UK) to identify neurons, polyclonal rabbit antibodies to glial fibrillary acidic protein (GFAP; Dako UK Ltd, Ely, UK) to identify astrocytes and mouse monoclonal antibodies to galactocebroside (Chemicon Hampshire, UK) to identify oligodendrocytes. The primary antibodies were diluted in 2.5% normal goat serum (NGS, Sigma–Aldrich, Poole, UK) in 0.1% Triton in PBS and incubated overnight at RT. Following washing, slides were exposed to fluorochrome-labelled secondary antibodies for 2 h. The slides were washed for a final time before being coverslipped using Vectashield with DAPI (Vector Laboratories, Peterborough, UK). Labelled cells were examined and photographed at appropriate wavelengths using a Nikon fluorescent microscope.

Counting of immunostained neurospheres

A total of 40 digital photographs for each primary antibody (20 of the primary antibody and 20 of the DAPI stained nuclei) were taken of the cells at ×20 magnification. Using Adobe Photoshop, the photo of the immunopositive cells and the DAPI-stained nuclei was overlaid and quantified. The number of cells in which a nucleus was visible was counted in each photograph. A percentage number of positive cells were calculated for each photograph and all the percentages combined and averaged to give a total percentage number for that cell marker.


For immunohistochemistry, frozen sections were brought to room temperature, washed in 0.3% Triton. Sections for SMI 32 staining were pre-treated with 3% hydrogen peroxide in methanol for 20 min and washed. All sections were blocked with 10% normal goat serum (NGS) (Sigma–Aldrich. Poole, UK) for 1 h and then incubated overnight at 4°C with diluted primary antibody. The primary antibodies used were polyclonal rabbit anti-GFAP (1:200), polyclonal rabbit anti-NG2 (1:200) (a kind gift from Professor W. Stallcup, Burnham Institute, USA), polyclonal rabbit anti Olig-2 (1:5000) (a kind gift from Prof. D.H. Rowitch, Dana-Farber Cancer Institute and Harvard Medical School, USA), monoclonal mouse SMI 31 antibody (1:3000) (Sternberger, CA, USA), monoclonal mouse SMI 32 antibody (1:1000) (Sternberger, CA, USA) monoclonal mouse anti-GFP (1:300; Molecular probes, Invitrogen, Paisley, UK) and monoclonal rat anti-CD11b (1:400) (Serotec, Oxford, UK). All sections were rinsed and then incubated for 2 h with the relevant biotinylated secondary antibodies (1:200) (anti-rabbit (Vector Laboratories, Peterborough, UK), anti-rat (Vector Laboratories, Peterborough, UK) or anti-mouse (Southern Biotech, AL, USA). Sections incubated with the anti SMI 32, anti NG2 and anti-GFP primary antibodies were rinsed and then incubated with standard ABC reagent (Vector laboratories, Peterborough, UK), for 1 h at room temperature followed by visualization with diaminobenzidine (DAB, Vector laboratories, Peterborough, UK). Slides were then dehydrated and coverslipped using DPX mounting medium (VWR international Ltd, Lutterworth, UK). Sections incubated with anti-SMI 31, NG2, Olig-2, CD11b, GFP and GFAP antibodies were rinsed after the appropriate biotinylated secondary antibody and incubated for 1 h with a dilution of 1:100 streptavidin–FITC (Serotec, Oxford, UK) or Alexa fluor 594 (1:500, Molecular Probes, Paisley, UK). Sections were rinsed and coverslipped using vectashield with DAPI (Vector Laboratories, Peterborough, UK).

Quantification of the axonal degeneration within the corpus callosum of C57BL/6 mice

Using a ×40 objective, the total number of SMI 32+ axonal spheroids in the callosal radiation (CR) and the medial corpus callosum (MCC), were counted from three sections of each animal. The values for each area were summed and averaged and a group mean obtained. The statistical significance between transplanted and non-transplanted values was calculated using two-tailed t-test, assuming unequal variances for the two samples.

Quantification of the thickness of and density of axons within the corpus callosum

Using a calibrated eyepiece scale, the thickness of the medial region of the corpus callosum in SMI 31 stained sections was measured. Using a confocal microscope and a point grid overlaid on SMI 31-stained sections, the density of axons within the medial region of the corpus callosum was quantified using a Microcomputer Imaging Device (Imaging Research Inc.). For each animal, the mean% density was calculated from four measurements. To gain an index of the number of axons in the medial region of the corpus callosum, the value of the thickness (μm) and density (%) were multiplied for each animal and a group mean obtained. The statistical significance of any difference between an experimental group and age-matched controls were calculated using one-way analysis of variance and Bonferroni's multiple comparison test. Values for callosal axon number from the transplanted mice were compared to the extent of GFP expression in sections taken 100 μm away. To do this, the area of GFP-expression was obtained using Scion Image Beta 4.02 software (Scion Corporation, Frederick, Maryland, USA) on sections stained at the same time with a common set background density. Six images of the corpus callosum were analysed per animal and the average area from each animal was used for this analysis.

Quantification of the identity of the GFP+ cells

Six non-adjacent brain sections from each animal were stained with DAPI and anti-GFP antibody and either antibodies to Olig2 or NG2 (three sections each) and a total of nine digital photographs at ×20 magnification were taken in random areas of the corpus callosum and of the cortex for each section stained with either anti-NG2 or Olig2 antibodies. Using Adobe Photoshop, the anti-GFP, DAPI and anti-NG2 or anti-Olig2 photographs were overlain and the percentage of nucleus-containing GFP+ cells that were NG2+ or Olig2+ determined. The percentages for each area were averaged to give a total percentage number for that cell marker.


X-irradiation during cuprizone intoxication results in remyelination failure

After 8 weeks on the cuprizone diet, both the X-irradiated and the non-irradiated mice showed evidence of demyelination. This was not present throughout the corpus callosum (CC) but was restricted to the lateral aspects of the corpus callosum, CRs (Fig. 2B) and the body to the splenium (Fig. 2A) of the MCC in all the cuprizone-fed mice. The genu remained normally myelinated. In the non-irradiated group, the areas of demyelination were hypercellular compared to the X-irradiated group (Fig. 2C cf. D). The cells in the X-irradiated CR were predominately large myelin-debris and neutral fat-filled macrophages (Fig. 2Di) that were smaller and less numerous in the non-irradiated animals (Fig. 2Ci). Vacuolation of myelin was more noticeable in the X-irradiated group and axonal spheroids were present in the areas of demyelination in both groups of animals. At the recovery stage, the corpus callosum of the non-irradiated mice was remyelinated (Fig. 2E) while in the X-irradiated animals, there was only an occasional myelinated axon and the thickness of the corpus callosum was markedly reduced and many large lipid droplet-filled macrophages separated the demyelinated axons (Fig. 2F). In some animals, these formed large cellular masses often associated with the presence of cholesterol crystals.

Fig. 2

X-irradiation during cuprizone intoxication results in failure of remyelination. Schematic diagrams of the corpus callosum of the mouse brain in sagittal (A) and coronal (B) section. The corpus callosum (CC) of both non-irradiated (C) and X-irradiated mice (D) is demyelinated at 8 weeks. The CC of the non-irradiated mice contains cells with a small round nucleus which are less numerous in the X-irradiated mice (filled arrows; C cf. D). The cells in the CC of the X-irradiated mice are predominately myelin debris filled macrophages (Di) which are smaller (white-filled arrow; Ci) and less numerous in non-irradiated mice. The CC of both groups of mice contains cells with large pale nuclei—astrocytes (black-filled arrow; Ci). At 16 weeks, the CC of the non-irradiated mice are remyelinated (E) while the CC of the X-irradiated mice remain demyelinated and contain large numbers of large lipid droplet-filled macrophages (F). Scale bars; C and D, F and E = 35 μm, Ci and Di = 17.5 μm.

Axonal degeneration associated with demyelination is significantly greater in X-irradiated mice

To examine the extent of axonal damage in the areas undergoing demyelination, coronal sections of the corpus callosum were immunostained with the non-phosphorylated neurofilament antibody SMI 32, to demonstrate axonal spheroids (Fig. 3A). After 8 weeks on the cuprizone diet, there were significantly more axonal spheroids in the X-irradiated mice compared to the non-irradiated mice (Fig. 3B). To demonstrate the consequences of the spheroid formation observed during the demyelinating phase, axon number was evaluated in the MCC of the cuprizone-intoxicated mice after 7 weeks on a normal diet and this revealed a significant loss of axons in both non-irradiated and X-irradiated cuprizone-treated mice compared to control mice (Fig. 3C). Significantly, the extent of axon loss in the X-irradiated animals was significantly greater than that in the non-irradiated mice indicating that preventing remyelination resulted in a significant increase in axon degeneration. There was no evidence of axonal degeneration in X-irradiated mice without cuprizone intoxication (Fig. 3C).

Fig. 3

Inhibiting remyelination by X-irradiation during cuprizone intoxication results in a significant increase in axon degeneration. The callosal radiations (CR) of both the X-irradiated and non-irradiated mice contain evidence of axonal degeneration in the form of axonal spheroids which can be stained with the SMI 32 antibody (arrows; A). Quantification of number of SMI 32+ axonal spheroids at the demyelinating stage (B) reveals a significantly higher incidence of spheroids in mice that had been X-irradiated during cuprizone intoxication compared to non-irradiated mice. Sections of the medial corpus callosum stained with the SMI 31 antibody were used to gain an index of the total number of axons in the CC at 16 weeks and this revealed there was significantly more axonal loss in non-remyelinating X-irradiated mice compared to the remyelinating non-irradiated mice (C). Scale bar: 50 μm.

Effect of transplanted GFP+ neurospheres during cuprizone intoxication

Having shown that inhibiting remyelination was associated with enhancement of axonal loss in the corpus callosum following cuprizone induced demyelination, our next aim was to restore the remyelinating ability to the X-irradiated brain in order to test whether the increased incidence of axon degeneration was caused by X-irradiation or remyelination failure. We have already reported the dynamics of oligodendrocyte progenitor cell loss and transplant-mediated repopulation of the normal mouse brain following exposure to 40 Gy (Irvine and Blakemore, 2007) and here we will concentrate on the behaviour of transplanted cells in cuprizone intoxicated animals and in particular showing that the transplanted cells only contributed cells of the oligodendrocyte lineage to the host brain.

In vitro differentiation potential of EGFP neurospheres

Neurospheres were propagated from cells derived from E14 EGFP+ mouse forebrains in culture medium containing EGF and FGF-2. Seven days after replacing the mitogens with 1% FCS the neurospheres had formed a monolayer in which 68.22% ± 1.5 of the cells were GFAP+, 14% ± 1.8 were positive for the neuronal marker, β-tubulin III and 8% ± 1 were positive for the oligodendrocyte marker, GalC (data not shown). Transplantation experiments were performed with intact neurospheres from the second passage.

Effect of X-irradiation on the endogenous NG2+ cell population

When the normal mouse brain is exposed to 40 Gy of X-irradiation there is rapid loss of OPCs from the cerebral cortex, corpus callosum, septum and hippocampus and a slow protracted reduction in OPC number elsewhere in the brain and spinal cord (Chari et al., 2003; Irvine and Blakemore, 2007). The cerebral cortex, corpus callosum, septum and hippocampus are rapidly repopulated by OPCs from non-irradiated adjacent cortex at a rate of 1 mm/week (Fig. 4A and B, E–G) while the others areas show progressive depletion and only show repopulation when OPC numbers are almost completely depleted (Irvine and Blakemore, 2007). In contrast, although exposing the brain of cuprizone-intoxicated animals resulted in a similar pattern of OPC-depletion (Fig. 4C) there was no evidence of repopulation of the OPC-depleted cortex with time (Fig. 4D).

Fig. 4

Combination of X-irradiated and cuprizone intoxication results in repopulation failure of OPC depleted tissue by endogenous cells. Diagrams showing the distribution of NG2+ cell depletion in sagittal section at different time post irradiation in normal and cuprizone- intoxicated animals. At 7 days post-irradiation in normal mice (A) there is almost complete depletion of NG2+ cells within the cortex, corpus callosum and hippocampus (dark grey area) and the NG2+ cells that remain within these depleted areas have a more ‘bushy’ appearance than normal as depicted in (E). In contrast to the cortex, NG2+ cells in ventral areas caudal to the septum appear normal (white areas: G) and there is a sharp demarcation between the populated and depleted areas of the cortex. At 4 weeks post-irradiation in normal mice, the length of the depleted area of the cortex has significantly decreased (A: dark grey areas cf. Fig. B) and by this time the NG2+ cells within the ventral regions of the brain are present at lower than normal density (Light grey area) and have a bushy appearance as seen in (F). In contrast, X-irradiated cuprizone intoxicated mice at 4 weeks post-irradiation (C) show a failure of OPCs from non-irradiated areas to repopulate NG2+ cell depleted cortex. Analysis of cuprizone-intoxicated animals 12 weeks post-irradiation (D) indicate that there has been a further decrease in the number of NG2+ cells surviving within the irradiated diencephalon and no evidence of repopulation of the completely OPC-depleted cortex (Fig. 4A and B are adapted from Irvine and Blakemore, 2007). Scale bars; E = 75 μm, F and G = 100 μm.

Distribution of the transplanted cells

The distribution and density of the grafted cells in the brains of cuprizone-intoxicated mice was studied at 10 and 18 weeks post-transplantation. At 10 weeks post-transplantation, grafted cells showed widespread integration into various regions of the brain (Fig. 5M) including the cortex (Fig. 5C), hippocampus and thalamus (Fig. 5E), with the highest density being present in the areas undergoing demyelination, the corpus callosum and the fimbria (Fig. 5A and B). Grafted cells were also found in the deep white matter of the cerebellum (a site that can also undergo demyelination in cuprizone intoxication) of two transplanted animals (Fig. 5D). The morphology of the grafted cells as visualized by GFP staining varied in different areas of the brain. Rostrally, where the transplanted cells were interfaced with tissue still populated by endogenous cells, the transplanted cells had fine primary and secondary branches orientated radially around a small cell soma (Figs 5F and 6I) similar to that of endogenous cells. In contrast, transplanted cells at the caudal border of the transplant-populated tissue where they would be entering tissue depleted of its endogenous progenitor population, cells had a long thin cell body with two or three branches extending in one direction giving the cell an elongated appearance typical of a migrating cell (Fig. 5G). In areas like the fimbria and corpus callosum which would be undergoing demyelination and where cells were present at high density some of the GFP+ cells had the morphology of oligodendrocytes since multiple processes extended from the cell body with an orientation that related to the associated axons (Fig. 5H). In some animals, GFP cells were found as spheres attached to the ependyma (Fig. 5I; n = 3) as well as embedded within the choroid plexus and meninges (Fig. 5J).

Fig. 5

Distribution and morphology of transplant-derived GFP+ cells at 10 and 18 weeks post-transplantation. Bilateral intra-ventricular transplantation of GFP+ neurospheres resulted in the migration and integration of GFP+ cells into the brains of X-irradiated cuprizone-intoxicated mice. GFP+ cells were found within white matter tracts including the corpus callosum (A), fimbria (B) and cerebellar white matter (D). In white matter areas that were demyelinating some GFP cells displayed an oligodendrocytic morphology (B and H). Cells found in grey matter had the morphology associated with oligodendrocyte progenitor cells (OPCs): cortex (C and F) and thalamus (E). At the rostral interface between transplanted and non-transplanted areas, where the GFP+ cells were opposed to endogenous OPCs the GFP+ cells had a similar morphology to endogenous cells (F) while at the caudal border of the area populated by GFP+ cells where the cells would be entering OPC-depleted tissue the cells had processes orientated in a longitudinal direction projecting from a small oval nucleus towards the depleted tissue (G). Clumps of GFP+ cells were also found within the ventricle wall (I) and in one animal as an intact neurosphere within the choroid plexus (J). At 18 weeks post-transplantation, GFP cells had a myelinating oligodendrocytic appearance both within the corpus callosum (K) and the fimbria (L). (M) Shows the distribution of GFP+ cells in the six animals examined 10 weeks post-transplantation and (N) is a scanned image of the distribution of GFP+ cells 18 weeks post-transplantation. Scale Bars: A and H = 50 μm, BE and I–L = 100 μm, F and G = 75 μm.

Fig. 6

Identity of the transplant-derived cells at 10 and 18 weeks post-transplantation. Analysis of the brains of mice at 10 and 18 weeks post-transplantation revealed that the GFP cells within the corpus callosum and the cortex had not differentiated into astrocytes (C and D: GFAP) or microglia/macrophages (A and B: CD11b). Immunohistochemistry with the oligodendrocyte lineage marker, Olig2, revealed that all of the transplant-derived cells had differentiated into cells of the oligodendrocytic lineage (E and F). In the cortex all the GFP+ cells were NG2+ (G and I) and had a similar morphology to endogenous NG2+ cells (I) while in the corpus callosum at 10 weeks one-third of the GFP+ cells were NG2 negative (H) and many expressed the oligodendrocyte marker CC1 (J and K) white matter. Scale Bar: 50 μm.

At 18 weeks post-transplantation, the distribution of transplanted cells was more extensive and there was a particularly high density of cells in white matter tracts that had undergone demyelination (Fig. 5N) where many of the GFP+ cells had an oligodendrocytic appearance with multiple processes aligned with the longitudinal orientation of axons (Fig. 5K and L).

Identity of the transplanted cells

To analyse the in vivo differentiation of the transplanted cells immunohistochemistry was performed using antibodies to: GFAP for astrocytes, CD11b for microglia and macrophages and NG2, Olig2 and CCI for oligodendroglia. At 10 weeks post-transplantation there were no GFP+ cells for CD11b (Fig. 6A and B) or GFAP (Fig. 6C and D) and all the GFP+ cells within the brain substance were positive for Olig2 antibodies, indicating that all the cells had differentiated into cells of the oligodendrocyte lineage (Fig. 6E and F). NG2 and GFP double immunostaining revealed that the grafted cells within the cortex were all NG2+ (99.4 ± 0.04%) (Fig. 6G and K). Nearly two-thirds of the GFP+ cells population in white matter tracts undergoing demyelination co-expressed the marker for oligodendrocyte progenitors, NG2 (65 ± 0.5%) (Fig. 6H). The remaining GFP+ cells in the white matter tracts co-expressed Olig2 and had the morphology associated with oligodendrocytes and so can be considered to be cells that had differentiated into premyelinating and/or myelinating oligodendrocytes. In some cases, these cells stained with the CC1 antibody (Fig. 6I and J).

At 18 weeks post-transplantation, all the GFP+ cells within the brains of the transplanted mice were Olig2+ and within the cortex 98.1 ± 0.1% of the GFP+ cells were also positive for NG2. In the corpus callosum the number of GFP+ cell had increased from that seen at 10 weeks but the percentage of GFP+ cells that were NG2+ was lower 44.1% (± 0.6%) compared to 65% (±0.5%) at 10 weeks and where axons were sectioned in their longitudinal plane many of the GFP+ cells had the typical morphology of oligodendrocytes.

Reconstitution of old mice with myelinating cells from embryonic brain results in remyelination and significant axon protection

Plastic sections of the brains of the transplanted mice revealed that in contrast to the non-transplanted mice there was evidence of remyelination and an absence of demyelinated axons within the corpus callosum and the striatal bundles (Figs 2F cf. 7A and C cf. B). SMI 32 staining revealed that there was significantly fewer axonal spheroids in the X-irradiated-transplanted mice compared to the X-irradiated mice (Fig. 7D) indicating that restoring remyelination potential to the X-irradiated mice reduced the incidence of axon degeneration. Interestingly, there were fewer spheroids in the X-irradiated-transplanted mice compared to the nontransplanted non-irradiated mice; however this reduction did not reach significance.

Fig. 7

Transplantation of X-irradiated cuprizone-treated mice results in remyelination and axon survival compared to non-transplanted mice.Semithin sections of the corpus callosum (A) and striatum (C) of transplanted mice show evidence of remyelination that was not present in the X-irradiated-non-transplanted mice (Fig. 2F cf. 7A and C cf. B). Transplanted-X-irradiated cuprizone-intoxicated mice have significantly fewer SMI 32 spheroids than non-transplanted-X-irradiated cuprizone-intoxicated mice (D) and although there is a lower number of SMI32 spheroids in the transplanted mice compared to the non-transplanted-non-irradiated mice, this does not reach significance (D). Quantification of axons in the medial corpus callosum revealed that restoring remyelinating potential to X-irradiated cuprizone-intoxicated animals resulted in a significant reduction in axon loss (E). Scale Bar: 70 μm.

Analysis of axon number in the MCC of animals returned to a normal diet for 7 weeks showed that the X-irradiated transplanted mice had lost significantly fewer axons than the X-irradiated mice (Fig. 7E). A result that would indicate that remyelination failure rather than X-irradiation was the cause of the increased axon degeneration seen in the X-irradiated mice and a conclusion that indicates the ability to remyelinate protects axons from demyelination-associated axon degeneration.

Evaluation of the extent of transplant-mediated remyelination was not possible; however, it was possible to get an index of the extent of transplant-derived remyelination by evaluating the number of transplant-derived cells—the majority of which will be oligodendrocytes—in sections near to those used for the evaluation of axon loss. This determination showed a positive correlation between the amount of GFP staining in the corpus callosum and axon number namely index of GFP staining to index of axon number of 721–17 150; 1314–24 000; 3991–26 100; 3400–2700 indicating that the presence, proliferation and survival of transplant-derived oligodendrocyte lineage cells was associated with axon protection.

Effect of transplanted GFP+ neurospheres on the inflammatory response during cuprizone-induced demyelination

To obtain an indication of the effect of transplanted GFP+ neurospheres on the astrocyte and microglial response to demyelination sections were immunolabelled with the antibody CD11b (Fig. 8D–F), which labels both microglia and macrophages, and with anti-GFAP antibodies for astrocytes (Fig. 8A–C). The astrocyte response was greatest in the non-irradiated non-transplanted animals with the response in the transplanted and non-transplanted X-irradiated animals being similar. The microglia/macrophage response differed between the normal and X-irradiated mice (Fig. D and E) in that the X-irradiated mice had large numbers of foamy macrophages present in the demyelinated corpus callosum. The microglial/macrophage response in the transplanted X-irradiated animals (Fig. F) had an intermediate appearance between the non-transplanted non-X-irradiated (Fig. D), and non-transplanted X-irradiated animals (Fig. E), indicating that while there were no transplant-derived cells of the microglial/macrophage lineage in the tissue the presence of the transplant-derived cells of the oligodendrocyte lineage was having an effect on the response of the endogenous cells.

Fig. 8

Effect of transplanted GFP+ neurospheres on the inflammatory response during cuprizone-induced demyelination. The astrocytic response in the callosal radiation as visualized by fluorescent antibodies to GFAP at the demyelinating stage (A–C) was greater in the non-irradiated/non-transplanted mice compared to both the X-irradiated mice groups of mice regardless of transplantation. The microglial response in similar areas as visualized by CD11b immunoperoxidase staining (D–F) showed increased number of CD11b+ positive cells in all groups. The X-irradiated non-transplanted group had the greatest number and most of the cells had the appearance of fat-filled macrophages while the transplanted-X-irradiated had an intermediate appearance between the non-irradiated non-transplanted and the X-irradiated non-transplanted animals in that the tissue contained a mixture of fat-filled macrophages and processed cells. Scale bar: 50 µm.


In our previous study we showed that in cuprizone intoxication of C57BL/6 mice the axon degeneration that is associated with demyelination increased with age and we proposed that this may either be related to changes in the nature of the inflammatory response that occur with aging (Giulian et al., 1993; Rozovsky et al., 1998; Ou et al., 2002; Yu et al., 2002; Stolzing and Grune, 2003) and/or was a reflection of the slowing of remyelination that occurs with age (Shields et al., 1999; Hinks and Franklin, 2000; Sim et al., 2002); a phenomena that would result in axons being exposed to an inflammatory environment for a longer length of time. The aim of the present experiment was to test the latter hypothesis by abolishing remyelination following cuprizone-induced demyelination. This was achieved by exposing tissue to 40 Gy of X-irradiation to deplete the tissue of its oligodendrocyte progenitor population. The inhibition of remyelination so induced resulted in a significant increase in the incidence of axon spheroids in the corpus callosum and a significant decrease in the number of axons in the MCC compared to the non-irradiated-cuprizone-intoxicated animals of similar age. A result that would indicate that failure of remyelination results in an increase in demyelination-associated axon degeneration and would be consistent with the idea that remyelination protects axons. However, since it was also possible that the increased axon degeneration may be related to an effect of X-irradiation on the brain we excluded this by restoring the remyelinating capacity of the X-irradiated brain by transplanting intact embryonic EGFP neurospheres into the lateral ventricles. This resulted in population of the brain by transplant-derived oligodendrocyte progenitor cells, which subsequently generated oligodendrocytes, that remyelinated the demyelinated axons. In these transplanted animals there was a significant reduction in the number of axon spheroids in the corpus callosum and a greater number of axons in the MCC than in the non-transplanted-X-irradiated animals, indicating that the increase in axon degeneration seen in the X-irradiated-cuprizone-intoxicated animals was the result of remyelination failure and was not caused by the effects of X-irradiation. However, it is also possible that the increased axon protection seen in the transplanted old animals could also have been influenced by the integration of embryo-derived astrocytes and/or embryo-derived microglia/macrophage into the tissue influencing the nature of the inflammatory response to demyelination and thereby modifying the incidence of axon degeneration. We consider this most unlikely as all the GFP+ cells in the brain were Olig2+ and no GFP+/GFAP+ or GFP+/Mac-1+ cells were detected; a result that indicates that only cells of the oligodendrocyte lineage were generated from the transplanted neurospheres.

In addition to providing a rapid production of remyelinating oligodendrocytes one cannot exclude the possibility that the mere presence of the embryo-derived neural precursors, may have been associated with a degree of neuroprotection. Ostenfeld and colleagues (1999) demonstrated that a co-culture of embryonic mesencephalic neurons and neural stem cells (NSCs) derived from the striatum significantly increased the survival of the dopaminergic neurons. Consistent with these findings Ourednik and colleagues (2002) reported a preservation of host dopaminergic neurons following transplantation of NSCs into 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP)-treated mice. The authors suggested that the NSCs were acting as a source of neuroprotective factors such as GDNF. In cuprizone intoxication, a number of factors have been shown to be reduce demyelination (Mason et al., 2000; Emery et al., 2006) and the neuroprotective effect of NSC transplantation in experimental allergic encephalitis is now considered to be a consequence of neuroprotection and immunomodulation rather than enhanced remyelination (Pluchino et al., 2005; Einstein et al., 2003). Thus in addition to remyelination some of the neuroprotection observed in our transplanted animals could be a consequence of the non-specific benefits of introducing embryonic cells into the CNS. In this regard, it is perhaps significant that the microglial response in the transplanted-X-irradiated animals was transformed to one that resembled that present in the non-irradiated cuprizone intoxicated animals. In addition to demonstrating the value of remyelination for protection of axons from demyelination-associated axon damage, the present experiments illustrate a number of phenomena that have relevance to understanding remyelination failure in multiple sclerosis and the potential of transplanted cells to remyelinate areas of demyelination. As has been shown by others (Einstein et al., 2003) we found long-term survival of clumps of transplanted cells in the ventricular system and in the meninges and observed that cells introduced into the lateral ventricles entered distant areas of white matter when demyelination was present. What was of particular interest was that unlike in irradiated normal mouse brain (Irvine and Blakemore, 2007) where endogenous OPCs progressively repopulate the X-irradiation OPC-depleted cortex, in the X-irradiated non-transplanted cuprizone-intoxicated mice there was no repopulation of the OPC-depleted tissue by endogenous cells. In multiple sclerosis, some areas of chronic demyelination contain OPCs while others do not (Wolswijk, 1998). It has been proposed that the primary reason for remyelination failure in multiple sclerosis is concurrent destruction of OPCs and oligodendrocytes (Blakemore et al., 2002; Chari and Blakemore, 2002) which, because of the slow rate with which adult OPCs repopulate OPC-depleted tissue, results in OPCs arriving adjacent to demyelinated axons when the acute inflammatory response required to drive remyelination has subsided. Tissue culture studies have indicated that astrocytes can inhibit OPC migration and process extension (Fok-Seang et al., 1995); however, in a previous study using the taiep rat (Foote and Blakemore, 2005), where there was a significant astrocytosis, it was found that endogenous OPCs repopulated OPC-depleted tissue at a rate that was not significantly different to that in found in normal animals of similar age and it was concluded that the presence of an astrocytosis did not influence OPC-repopulation of OPC-depleted tissue. The current study suggests that in acute demyelinating situations the environment can be altered in such a way that it inhibits endogenous OPCs from repopulating OPC-depleted tissue. In cuprizone intoxication, a marked astrocytosis and microglia response occurs early in the intoxication (Kesterson and Carlton, 1971; Pattison and Jebett, 1971) and this is further enhanced as a consequence of the myelin sheath breakdown. Thus, contrary to our previous conclusion, it would appear and under certain circumstances, the glial responses that are associated with demyelination can be inhibitory for OPC-repopulation of OPC-depleted tissue. In terms of multiple sclerosis, such an inhibition would explain why not all areas of chronic demyelination contain OPCs. Significantly what the present study shows is that tissue that cannot be repopulated by endogenous cells can be readily repopulated by transplanted-embryo-derived precursor cells.


This work was supported by a grant from the MS Society. We gratefully acknowledge the NG2 antibody from Professor William Stallcup and the Olig2 antibody from Professor David Rowitch. K.A.I. would also like to acknowledge Stephen P.J. Fancy for his critical comments on this study.


  • Abbreviations:
    callosal radiation
    medial corpus callosum
    tumour necrosis factor α.


View Abstract