OUP user menu

Oligodendrocyte survival, loss and birth in lesions of chronic-stage multiple sclerosis

Guus Wolswijk
DOI: http://dx.doi.org/10.1093/brain/123.1.105 105-115 First published online: 1 January 2000

Summary

One of the hallmarks of the human demyelinating disease multiple sclerosis is the inability to compensate adequately for the loss of myelin and of oligodendrocytes, the myelin-forming cells of the CNS. Oligodendrocyte precursor cells, a potential source of oligodendrocytes, have been identified in lesions of chronic multiple sclerosis, but it is not known whether they develop into new, fully differentiated oligodendrocytes, capable of remyelination. Sections of post-mortem multiple sclerosis tissue were therefore immunolabelled with antibodies to galactocerebroside (GalC), the first oligodendrocyte-specific molecule to be expressed by differentiating oligodendrocyte precursor cells, and myelin oligodendrocyte glycoprotein (MOG), a marker for mature oligodendrocytes. In total, 23 lesions from 15 subjects with chronic progressive multiple sclerosis were analysed. The immunolabelling revealed that chronic multiple sclerosis lesions contain only small numbers of immature, process-bearing, GalC-positive oligodendrocytes (0–2 cells/mm2 in 10 μm thick sections); they had a relatively large, pale nucleus (maximum diameter: 9.9 ± 0.9 μm). Although they appeared to make contact with surrounding demyelinated axons, most immature oligodendrocytes appeared not to be engaged in myelination. These findings suggest that oligodendrocyte differentiation of precursor cells is a rare event in chronic multiple sclerosis, which is consistent with the general failure of myelin repair during the later stages of this disease. The lesions in the collection, in particular those with recent demyelinating activity, contained another distinct population of oligodendrocytes. It consisted of small, round cells with a small, dense nucleus (maximum diameter: 6.8 ± 0.8 μm) that expressed both GalC and MOG but lacked processes, suggesting that these cells were mature oligodendrocytes that had survived the loss of their myelin sheaths, i.e. they were demyelinated oligodendrocytes. In the most recent lesions in the collection, the demyelinated oligodendrocytes were found in large numbers throughout the centre of the lesion (up to 700 cells/mm2), while in the older lesions they were found only at the edges. Moreover, when the borders of these older lesions still contained numerous macrophages, they tended to contain more demyelinated oligodendrocytes than those lacking macrophages. These findings suggest that mature, demyelinated oligodendrocytes gradually disappear from lesion areas with increasing age of the lesion. The present study thus suggests that the failure of myelin repair in at least some cases of chronic multiple sclerosis is due to (i) the loss of demyelinated oligodendrocytes from lesion areas and (ii) the failure of the oligodendrocyte precursor population to expand and generate new oligodendrocytes. Gaining further insight into these processes may prove crucial for the development of remyelination promoting strategies.

  • GalC
  • MOG
  • multiple sclerosis
  • oligodendrocyte precursor cell
  • remyelination
  • GalC = galactocerebroside
  • GFAP = glial fibrillary acidic protein
  • LCA = leucocyte common antigen
  • MBP = myelin basic protein
  • MOG = myelin oligodendrocyte glycoprotein
  • mAb = monoclonal antibody
  • PDGF = platelet-derived growth factor

Introduction

The local loss of the insulating sheath of myelin around axons in the CNS of patients with multiple sclerosis is generally not adequately compensated for during the chronic stage of the disease, resulting in the formation of confluent areas of persistent demyelination and neurological disability due, at least in part, to impaired conduction along demyelinated fibres (Prineas and McDonald, 1997; Lassmann, 1998); recent evidence suggests that some of the disability in multiple sclerosis may result from damage to and loss of axons (Ferguson et al., 1997; Trapp et al., 1998). Spontaneous remyelination does appear to be a prominent feature in lesions that develop during the early course of multiple sclerosis (Lassmann, 1983; Prineas, 1985; Prineas et al., 1993a, b; Raine and Wu, 1993; Brück et al., 1994; Ozawa et al., 1994), although the extent to which this occurs may depend on disease subtype (Lucchinetti et al., 1996; Lassmann et al., 1997; Lassmann, 1998). The basis for this decline in regenerative potential with disease progression remains elusive, but the decline is thought to reflect the extent of loss of oligodendrocytes, the myelin-forming cells of the CNS, from lesion areas (Prineas and McDonald, 1997; Raine, 1997). Thus, lesions that develop during the chronic stages of multiple sclerosis may lack sufficient oligodendrocytes to carry out myelin repair.

The mature human CNS contains an additional potential source of new remyelinating oligodendrocytes, i.e. a population of oligodendrocyte precursor cells (Armstrong et al., 1992; Gogate et al., 1994; Scolding et al., 1995; Wolswijk, 1998a), but their fate in the demyelination process in multiple sclerosis has been revealed only recently (Wolswijk 1998b). Using a collection of appropriately fixed multiple sclerosis tissue specimens, this study showed that lesions obtained from subjects who died during the chronic stage of multiple sclerosis contained significant numbers of oligodendrocyte precursor cells, identified as cells that bound the O4 monoclonal antibody (mAb), but not antibodies to galactocerebroside (GalC) and glial fibrillary acidic protein (GFAP) (Wolswijk, 1998b). Subsequent analysis of the O4-positive oligodendrocyte precursor cells in chronic-stage multiple sclerosis lesions suggested that they were relatively quiescent, as none expressed the nuclear proliferation antigen recognized by the Ki-67 antibody (Wolswijk, 1998b). Demyelinated lesions derived from patients with chronic-stage multiple sclerosis also contain cells that bind antibodies to the α-type receptor for platelet-derived growth factor (PDGF) (Scolding, 1998; Wolswijk, 1998b), a putative marker for oligodendrocyte precursor cells (Pringle et al., 1992; Nishiayama et al., 1996; Oumesmar et al., 1997). These cells were observed not only in the centre of the lesions, but also in the periplaque white matter and in control white matter (Scolding et al., 1998); Gogate and co-workers had shown earlier that PDGF-α receptor mRNA expressing cells are present in normal-appearing temporal lobe white matter from patients with epilepsy (Gogate et al., 1994). There may be some overlap between the two populations of oligodendrocyte precursor cells in the adult human CNS, as Gogate and colleagues have shown, using in situ hybridization, that human O4-positive cells grown in tissue culture contain transcripts for the α-type PDGF receptor (Gogate et al., 1994).

The presence of oligodendrocyte precursor cells in lesions of chronic-stage multiple sclerosis suggests that they survive the demyelination process and/or repopulate lesion sites after their local destruction. The first possibility appears the most likely, as even the centre of the lesions with the most recent demyelinating activity contained significant numbers of oligodendrocyte precursor cells, as they were found throughout the demyelinated area of all lesions studied, including the very large lesions in the collection, and as they were not concentrated near lesion borders (Wolswijk, 1998a, b). If the oligodendrocyte precursor population is indeed largely unaffected by the myelin destruction process in multiple sclerosis, this suggests that the disease process specifically targets oligodendrocytes and/or their myelin sheaths, and that the limited success of myelin repair during the chronic stage of multiple sclerosis is the result of the failure of the local oligodendrocyte precursor population to expand and generate new myelin-forming cells. Oligodendrocyte precursor cells may, however, generate new oligodendrocytes in lesions that develop during the early phases of multiple sclerosis, as such lesions frequently contain oligodendrocytes that do not yet express markers for fully differentiated oligodendrocytes (Prineas et al., 1989; Brück et al., 1994; Ozawa et al., 1994).

Following the recent identification of oligodendrocyte precursor cells in chronic-stage multiple sclerosis lesions, the present study examined the potential origin of two distinct types of GalC-positive oligodendrocytes previously identified in chronic-stage multiple sclerosis lesions, i.e. a population of large, process-bearing oligodendrocytes and a population of small, round, non-myelinating oligodendrocytes (Wolswijk, 1998b). The data of the present study suggest that the first population is derived from the pool of oligodendrocyte precursor cells in the lesions, whereas the second population comprises mature oligodendrocytes that have lost their myelin-forming processes during episodes of demyelinating damage.

Material and methods

Preparation of sections of post-mortem CNS tissue

Blocks of post-mortem CNS tissue from subjects with a clinical diagnosis of chronic progressive multiple sclerosis containing one or more macroscopically visible lesions were fixed in 4% paraformaldehyde in PBS (phosphate-buffered saline; pH 7.4) and then processed as described in detail previously (Wolswijk, 1998b). The clinical diagnosis of multiple sclerosis was confirmed neuropathologically by Dr W. Kamphorst, Department of Pathology, Academic Hospital of the Free Hospital, Amsterdam, The Netherlands. The average age of the multiple sclerosis subjects at death was 54 years (range 32–82 years), the average disease duration was 19 years (range 8–49 years), and the average delay between the death of the subject and the fixation of the tissue was 6 h 15 min (range from 3 h 45 min to 9 h 35 min).

Immunohistochemistry

Sections 10 μm thick were immunolabelled using indirect immunofluorescence or immunoperoxidase procedures, as described in detail in a previous paper (Wolswijk, 1998b). The primary antibodies used were: (i) a rabbit anti-human CD3 antiserum (DAKO, Denmark); (ii) a mouse IgG3 anti-GalC mAb [the Ranscht mAb; Ranscht et al., 1982)], a rabbit antiserum to cow GFAP (glial acidic fibrillary protein) (DAKO); (iii) a rabbit antiserum to human Ki-67 (DAKO); (iv) a mixture of two mouse IgG1 anti-human leucocyte common antigen (LCA) mAbs (mAbs 2B11 and PD7/26; DAKO); (v) a rabbit antiserum to myelin basic protein (MBP) (a gift of Dr H. van Noort, TNO, Leiden, The Netherlands); (vi) a mouse IgG1 anti-MOG (myelin oligodendrocyte glycoprotein) mAb [the Y10 mAb (Piddlesden et al., 1993), a gift of Dr S. Piddlesden]; (vii) a mouse IgG1 anti-neurofilament mAb (the RT97 mAb; Boehringer Mannheim, Germany); (viii) the mouse IgM O1 mAb (Sommer and Schachner, 1981; an anti-GalC mAb); (ix) the mouse IgM O4 mAb (Sommer and Schachner, 1981); and (x) a mouse IgG1 anti-vimentin mAb (Boehringer Mannheim). The binding of the primary antibodies was visualized using appropriate class- and species-specific FITC (fluorescein isothiocyanate)-, TRITC (tetramethyl rhodamine isothiocyanate)- and biotin-labelled conjugates (all from either Southern Biotechnology, Birmingham, Ala., USA or Jackson ImmunoResearch, Westgrove, Pa., USA); the binding of the biotinylated antibodies was visualized with either Cy5-conjugated (Jackson ImmunoResearch) or horseradish peroxidase-conjugated streptavidin (Vectastain ABC kit, Vector Laboratories, Burlinghame, Calif., USA); 3,3′-diaminobenzidine was used as substrate for the peroxidase. Nuclei were visualized by incubating sections in 1 mg/ml Hoechst 33258 (Sigma, St Louis, Mo., USA) (for conventional fluorescence microscopical analysis), in 1 mM TO-PRO-3 iodide (Molecular Probes) in PBS (for confocal laser scanning microscopical analysis) or in a haematoxylin solution (for bright-field microscopical analysis). Sections were examined on a Zeiss Axiophot microscope with filters optimized for distinguishing between FITC, TRITC and Hoecht 33258 emission and with bright-field and phase-contrast optics and on a Zeiss 410 confocal laser scanning microscope (CLSM) with three lasers emitting at 488, 543 and 633 nm to excite FITC, TRITC and Cy5/TO-PRO-3 iodide, respectively.

Results

Identification of oligodendrocytes in chronic-stage multiple sclerosis lesions

The first oligodendrocyte-specific molecule to be expressed by differentiating oligodendrocyte precursor cells is the glycolipid GalC (Raff et al., 1978, 1983; Curtis et al., 1988; Reynolds and Wilkin, 1988; Warrington and Pfeiffer, 1992; Pfeiffer et al., 1993; Wolswijk, 1995). These cells subsequently start synthesizing a number of other oligodendrocyte-specific molecules, including proteolipid protein and MBP (Reynolds and Wilkin, 1988; Pfeiffer et al., 1993). One of the last molecules to be expressed by maturing oligodendrocytes is MOG (Pfeiffer et al., 1993; Piddlesden et al., 1993; Ludwin, 1990). In addition to changes in antigenic phenotype and an increase in the complexity of their morphology, the nucleus of maturing oligodendrocytes changes from large and pale to small and dense (Mori and Leblond, 1970). Therefore, to distinguish between immature and fully differentiated oligodendrocytes in chronic-stage multiple sclerosis lesions, sections were labelled with antibodies to GalC and MOG and with the nuclear dyes Hoechst 33258 or TO-PRO-3 iodide. Using this labelling procedure, it was expected that immature oligodendrocytes would be GalC-positive, MOG-negative cells with a large, pale nucleus, while mature oligodendrocytes would express both GalC and MOG and possess a small, dense nucleus.

The success and the specificity of stainings involving antibodies to GalC is dependent on the fixation of the tissue and the immunolabelling procedure employed (Prineas et al., 1989; Wu and Raine, 1992; Wolswijk, 1998b). For example, no or only weak labelling with anti-GalC antibodies is observed in sections cut from unfixed, snap-frozen multiple sclerosis tissue, whereas the cytoplasm of reactive astrocytes is stained prominently in sections of formalin-fixed, paraffin-embedded material. As shown in a previous study, excellent oligodendrocyte lineage-specific labelling with anti-GalC antibodies (and the O4 mAb) is obtained in sections of 4% paraformaldehyde-fixed, 30% sucrose-treated, frozen chronic-stage multiple sclerosis tissue (Wolswijk, 1998b).

Collection of chronic-stage multiple sclerosis lesions

The expression of GalC and MOG was examined in a series of 23 appropriately fixed, macroscopically visible brain lesions (with a diameter of ≥3 mm) derived from 15 subjects with chronic progressive multiple sclerosis (with a disease history of 8–49 years) (Table 1); most of the lesions in the collection were taken from white matter areas around the ventricles, a common site for lesions (Prineas and McDonald, 1997; Lassmann, 1998). As shown previously (Wolswijk, 1998b), detailed immunohistochemical analysis of the lesions with antibody markers for oligodendrocytes/myelin (antibodies to GalC, MBP and MOG), astrocytes (antibodies to GFAP), axons (anti-neurofilament antibodies) and leucocytes (anti-human LCA antibodies), i.e. (activated) microglia, macrophages and lymphocytes, indicated that the collection consisted of four distinct types of demyelinated lesions (Table 1; see also Fig. 1); these findings were based on the analysis of a minimum of 10 sections per lesion. Four of the lesions in the collection, i.e. lesions D, G, H and I, were hypercellular and contained large numbers of macrophages laden with myelin debris (>400 cells/mm2, 10 μm thick sections) and oligodendrocyte cell bodies (see later), but little myelin (type A lesions). Sixteen lesions had a centre that was hypocellular and almost free of myelin (<1% myelinated axons) and oligodendrocytes. Eleven of these had a wide border region and contained significant numbers of debris-laden macrophages in the centre [between 7 ± 6 (lesion J, subject 96-040) and 324 ± 52 (an area of lesion F, subject 96-039) cells/mm2 section (data from Wolswijk, 1998b)]; they were concentrated predominantly, but not exclusively, at the edges of the lesions, which were often hypercellular (type B lesions). The five other lesions with an oligodendrocyte- and myelin-free centre had a very sharp lesion border and were almost totally lacking in macrophages both in the centre and at the border, suggesting that they were relatively old (type C lesions); it can take several months for macrophages laden with myelin degradation products to disappear from lesion sites (Prineas et al., 1989; Brück et al., 1995; Lassmann, 1998). Finally, lesions P, R and S contained significant, but variable, numbers of myelinated axons throughout the centre (from <5% to >50%); these lesions were hypocellular and lacked macrophages and were thus also likely to be relatively old (type D lesions); with the tools available it was not possible to determine whether these three lesions were partly demyelinated or partly remyelinated. Macrophages containing debris that was still strongly immunoreactive for MBP were seen in only two of the lesions in the collection (the hypercellular lesions H and I, both from subject 96-040); their numbers were small and they were observed only in the borders of these two lesions. As MBP staining of myelin debris within macrophages is lost within 10 days of uptake by macrophages (Brück et al., 1995; Lassmann, 1998), this finding suggested that very little active myelin breakdown was occurring at the time of death of the multiple sclerosis subject in the lesions studied. Perivascular cuffs with variable numbers of LCA-positive cells (mainly lymphocytes and macrophages) were observed within and around the demyelinated area of all lesions studied (not shown).

View this table:
Table 1

Details of multiple sclerosis subjects and lesions

NBB number of MS subjectAge at death (years)SexDisease duration (years)Post-mortem delay (h : min)LesionLesion diameter (mm)Lesion type*
MS = multiple sclerosis; NBB = Netherlands Brain Bank. *See text for detailed description; classification scheme adopted at a recent conference on the immunopathology of multiple sclerosis lesions (Lassmann et al., 1998).
94-04232M89 : 35A13C (inactive)
95-09556M125 : 24B10C (inactive)
96-02534F106 : 50C>19B (inflammatory)
D>3A(inflammatory)
96-02669F199 : 15E3C (inactive)
96-03957F195 : 45F>12B (inflammatory)
96-04035F115 : 45G4A (inflammatory)
H5A (demyelinating, inflammatory)
I>3A (demyelinating, inflammatory)
J>3B (inflammatory)
K>3B (inflammatory)
L>4B (inflammatory)
M>3B (inflammatory)
96-07440F147 : 00N>11B (inflammatory)
96-07681F494 : 15O9C (inactive)
96-10473M224 : 45P>10D (inactive)
96-12153F187 : 16Q>11B (inflammatory)
97-00662F256 : 45R>14D (inactive)
97-07082F254 : 30S3D (inactive)
T3C (inactive)
97-07750M175 : 40U>4B (inflammatory)
97-12346M233 : 45V14B (inflammatory)
97-16040F107 : 00W12B (inflammatory)
Fig. 1

Expression of GalC in unaffected white matter and in various types of chronic-stage multiple sclerosis lesions. (A) Normal-appearing white matter of multiple sclerosis subject 97-160 labelled with antibodies to GalC. Note that individual myelin segments and oligodendrocyte perikarya are not visible in unaffected white matter. The intensity of labelling with antibodies to GalC in normal white matter is much lower than that of GalC-positive cells and structures in demyelinated areas (BE). (B) Centre of a hypercellular lesion containing large numbers of GalC-positive oligodendrocyte cell bodies lacking processes (type A lesion). The lesion area also contained many debris-laden macrophages but few myelin segments. Detail of lesion H, subject 96-040. (C) GalC-positive myelin segments and cells in the border area of a lesion that had an almost myelin-free and oligodendrocyte-free centre but still contained numerous phase-bright macrophages (type B lesion). Note that the cells and myelin are labelled in the characteristic punctate fashion typical of labelling involving antibodies to GalC; a similar staining pattern is observed with the O4 antibody (Wolswijk, 1998b). Detail of lesion W, subject 97-160. (D) Only small numbers of rounded GalC-positive oligodendrocyte perikarya were observed at the borders of lesions lacking debris-laden macrophages (type C lesion). Detail of lesion B, subject 95-095. (E) GalC-positive myelin sheaths in the centre of a partly demyelinated lesion (type D lesion). Detail of lesion P, subject 96-104. The images shown in this figure and in Fig. 2 were generated using a confocal laser scanning microscope; each image represents a single optical section of thickness 1–2 μm. Scale bars = 25 μm.

GalC expression in chronic-stage multiple sclerosis lesions

The intensity of the labelling with antibodies to GalC was higher at the lesion border than in the white matter surrounding the demyelinated lesion and in control white matter (Fig. 1). Individual GalC-positive cells and GalC-positive myelin segments were clearly visible in lesion areas, but not in the white matter surrounding the lesion and in control white matter (Fig. 1); however, single oligodendrocytes and myelin sheaths immunoreactive for GalC were observed in adult human cortical grey matter (not shown). In some of the chronic multiple sclerosis lesions, the anti-GalC immunoreactivity was found predominantly at the lesion borders, while in others GalC-positive cells and structures were observed throughout the demyelinated area. In addition to myelin and variable numbers of debris-laden macrophages, the anti-GalC antibodies labelled two distinct populations of oligodendrocytes in the lesion sites, i.e. large, process-bearing oligodendrocytes and small oligodendrocytes with few if any processes (Figs 1 and 2).

Fig. 2

Chronic-stage multiple sclerosis lesions contain two distinct types of GalC-positive oligodendrocytes. In addition to oligodendrocyte precursor cells (A), lesions derived from subjects who died during the chronic stage of multiple sclerosis were found to contain two distinct types of GalC-positive cells, i.e. immature, process-bearing oligodendrocytes (BE) and mature, demyelinated oligodendrocytes (FI). (A) Oligodendocyte precursor cell labelled with the O4 antibody (green, FITC) and the nuclear dye TO-PRO-3 iodide (red); this cell had not bound antibodies to GalC (not shown). This figure illustrates that the O4-positive, GalC-negative cells in chronic-stage multiple sclerosis lesions have a fairly simple morphology and that they possess a relatively large nucleus [with a maximum diameter of ~9.8 μm (Wolswijk, 1998b)]. Detail of lesion U, subject 97-077. (B) GalC-positive oligodendrocyte (green, FITC) with an immature morphology and a nucleus (red, TO-PRO-3) that resembles those of oligodendrocyte precursor cells in the lesions both in size and morphology (see A). Detail of lesion N, subject 96-074. (C) Even the processes of non-myelinating, immature, GalC-positive oligodendrocytes (red, TRITC) appeared to make contact (arrowheads) with surrounding, demyelinated, neurofilament-positive axons (green, FITC). Detail of lesion N, subject 96-074. (D) Immature oligodendrocyte attached to some myelin segments. Section from lesion N (subject 96-074) double-labelled with antibodies to GalC (red, TRITC) and MOG (green, FITC). (E) Immature, GalC-positive oligodendrocyte (red, TRITC) connected to MBP-positive (green, FITC) myelin segments. This cell was present in lesion B from subject 95-095. (F) Detail of the centre of lesion G (subject 96-040) double-labelled with anti-GalC antibodies (green, FITC) and the nuclear dye TO-PRO-3 iodide (red) illustrating the small size of both the cell body and nucleus of the rounded oligodendrocyte. (G) The rounded cells in chronic-stage multiple sclerosis lesions are positive for both GalC (red, TRITC) and O4 (green, FITC), and thus appear orange in this figure. The cell on the left (arrowhead) has bound only the O4 antibody and is thus an oligodendrocyte precursor cell (see also A). Detail of lesion W, subject 97-160. (H) Section double-labelled with antibodies to GalC (red, TRITC) and MOG (green, FITC). Note that the rounded, GalC-positive cells at the border of this lesion (lesion N, subject 96-074) are MOG-positive. However, the intensity of the staining is much lower than that of myelin. (I) Rounded oligodendrocytes are frequently found in areas with little or no myelin, clearly indicating that these cells are non-myelinating cells. Section triple-immunolabelled with antibodies to GalC (red, TRITC), MBP (green, FITC) and neurofilament (blue, Cy5). Detail of lesion N, subject 96-074. Scale bars = 25 μm.

Process-bearing GalC-positive oligodendrocytes

The chronic-stage multiple sclerosis lesions in the collection contained small numbers of large, process-bearing, GalC-positive cells (Fig. 2) scattered throughout their demyelinated areas. Their nucleus was relatively large [maximum diameter = 9.9 ± 0.9 μm (n = 13)] and often irregular in shape (Fig. 2) and thus resembled that of the O4-positive, GalC-negative, GFAP-negative oligodendrocyte precursor cells in the lesions [maximum diameter = 9.8 ± 1.5 μm (Wolswijk, 1998b)]. They also bound the O4 mAb [which recognizes an antigen(s) expressed on the surface of both oligodendrocytes and their precursor cells (Sommer and Schachner, 1981; Wolswijk and Noble, 1989)], but their cell body was not labelled with the Y10 anti-MOG mAb. Furthermore, confocal laser scanning microscope analysis showed that these cells lacked the intermediate filament GFAP and vimentin (not shown). Some of the process-bearing oligodendrocytes had a relatively simple morphology and in this respect resembled oligodendrocyte precursor cells, while others had a more complex morphology, i.e. they had many processes and they were often arranged in a more symmetrical manner than those of oligodendrocyte precursor cells (Fig. 2).

To determine whether the processes of these antigenically and morphologically immature oligodendrocytes were connected to myelin sheaths, sections were double-labelled with anti-GalC antibodies and antibodies to either MOG or MBP. These experiments showed that only a few (<10%) of these cells were engaged in myelination (Fig. 2). Confocal laser scanning microscopy of sections triple-immunolabelled with antibodies to GalC, MBP and neurofilament revealed that even the processes of the non-myelin-forming, antigenically immature oligodendrocytes appeared to be in close contact with surrounding, demyelinated axons (Fig. 2).

The density of the immature oligodendrocyte population in the lesions studied was much lower than that of the oligodendrocyte precursor population. While between 2 and 34 cells/mm2 in the chronic-stage multiple sclerosis lesions in the collection were oligodendrocyte precursor cells (Wolswijk, 1998b), the density of the immature oligodendrocyte population was much less than 1 cell/mm2 in most lesions (Wolswijk, 1998b). For example, the demyelinated area present in three sections of a region of lesion B from subject 95-095 (in total 24 mm2) contained 333 O4-positive, GalC-negative oligodendrocyte precursor cells (an average density of 13.9 cells/mm2 section), but only two process-bearing GalC-positive oligodendrocytes (a density of <0.1 cells/mm2 section). The highest density of immature oligodendrocytes was found in lesion Q (2 ± 1 cells/mm2 section).

Rounded GalC-positive oligodendrocytes

The vast majority of the GalC-positive cells in the chronic-stage multiple sclerosis lesions studied were small, round cells with little cytoplasm and few, if any, processes; they were observed not only in the 23 lesions in the collection but also in much smaller lesions (with a diameter of <3 mm) (not shown). The nucleus was small [maximum diameter = 6.8 ± 0.8 m (n = 64)] and generally round or oval; the nucleochromatin often appeared clumped or condensed. Double-labellings of sections with anti-GalC antibodies and the Y10 anti-MOG mAb revealed that the GalC-positive cell bodies were MOG-positive (Fig. 2), and thus appeared to belong to fully matured oligodendrocytes. As illustrated in Fig. 2, these cells expressed much lower levels of MOG than myelin. Moreover, in contrast to previous reports (e.g. Brück et al., 1994; Ozawa et al., 1994), the perikarya of mature oligodendrocytes in periplaque white matter and also in control white matter appeared not to be labelled with antibodies to MOG, even when sections were examined using a confocal laser scanning microscope (not shown). Additional immunolabellings involving the O4 mAb and the O1 mAb [which is another anti-GalC mAb (Sommer and Schachner, 1981)] showed that the rounded oligodendrocytes were both O4-positive (Fig. 2) and O1-positive. As expected, they did not bind antibodies to LCA (a marker for leucocytes), CD3 (a marker for T lymphocytes, which are also small round cells), GFAP or vimentin (not shown).

Confocal laser scanning microscope analysis of sections labelled with antibodies to GalC and either MOG or MBP suggested that most of the rounded oligodendrocytes were not connected to myelin segments, even though they were often found in areas of the lesions that still contained some myelin, such as the lesion edges (Figs 1 and 2). This notion was supported by the observation that the rounded oligodendrocytes were frequently present in areas completely devoid of myelin. Immunolabelling of sections with both anti-GalC and anti-NF antibodies showed that the rounded oligodendrocytes also did not extend processes to surrounding, demyelinated axons (Fig. 2).

The number and distribution of the rounded oligodendocytes in the 23 lesions studied varied considerably. For example, in the hypercellular lesions D, G, H and I they were present in large numbers throughout the lesion area [up to ~700 cells/mm2 (lesion I)], while in the type B and C lesions the rounded cells were present almost exclusively at the lesion borders (Figs 1 and 2); their centres contained in most cases <5 rounded oligodendrocytes per mm2 section. As illustrated in Fig. 1 and Table 2, the margins of the type B lesions, i.e. lesions with many debris-laden macrophages at their border, tended to contain more rounded oligodendrocytes in the border than type C lesions, i.e. lesions with few, if any, phase-bright macrophages at the border (Table 2; Fig. 1). For example, only a single row of rounded oligodendrocytes was observed at the edges of lesion B (subject 95-095) (Fig. 1D), whereas several rows of rounded oligodendrocytes were present in the border of an area of lesion W (subject 97-160) (Fig. 1C). Only a very occasional rounded oligodendrocyte was encountered in the centre of lesions with reduced numbers of myelin sheaths (not shown).

View this table:
Table 2

Relative density of rounded oligodendrocytes in the borders of type B and C chronic multiple sclerosis lesions

Lesion (NBB number of MS subject)Lesion typeRelative number of rounded oligodendrocytes along lesion borderRelative number of phase-bright macrophages along lesion border
The number of oligodendrocytes and debris-laden macrophages present at the borders of type B and C lesions was scored on a scale from ± (occasional cell) to +++ (several rows of cells) (see also Figs 1 and 2). Some variation in the relative numbers of rounded oligodendrocytes and debris-laden macrophages was observed. If the variation was large, this is indicated in the table. MS = multiple sclerosis; NBB = Netherlands Brain Bank.
C (96-025)B+++++
F (96-039)B+/++++/+++
J (96-040)B+/++++/++
K (96-040)B++
L (96-040)B++
M (96-040)B++++
N (96-074)B++++
Q (96-121)B+++++
U (97-077)B++++
V (97-123)B++++
W (97-160)B+/++++/++
A (94-042)C+±
B (95-095)C+±
E (96-026)C±±
O (96-076)C±±
T (97-070)C00

As the density of the rounded oligodendrocytes in the centre and/or borders of the chronic-stage multiple sclerosis lesions studied was often relatively high and as these cells sometimes appeared to be clumped together (Figs 1 and 2), it was checked whether any of the cells expressed a nuclear proliferation-associated protein recognized by the Ki-67 antibody; this molecule is expressed during the late G1, S, G2 and M phases but not in the G0 phase of the cell cycle (Gerdes et al., 1984; Brown and Gatter, 1990). Although all GalC-positive (or O4-positive), rounded cells present in four to six sections were examined (in many cases, >100 cells/section), none had a Ki-67-immunoreactive nucleus; also, none of the immature oligodendrocytes in the lesions was found to contain a Ki-67-positive nucleus. As shown previously (Wolswijk, 1998b), some of the 23 lesions did contain some Ki-67-positive nuclei in the demyelinated centre, but their numbers were small [highest density = 2 ± 1 positive nuclei/mm2 section (lesion N, subject 96-074)].

Discussion

The present study provides evidence that chronic-stage multiple sclerosis lesions contain only small numbers of immature oligodendrocytes. The morphology of those that were examined ranged from precursor-like cells to cells that were engaged in myelinating small numbers of axons. Although these cells could have been derived from dedifferentiated mature oligodendrocytes (Wood and Bunge, 1991; Canoll et al., 1999), it is more likely that the immature oligodendrocytes in the lesions had been recruited from the local pool of oligodendrocyte precursor cells. If this is indeed the case, then this suggests that oligodendrocytic differentiation of precursor cells is generally a rare event in chronic multiple sclerosis lesions or that the newly generated oligodendrocytes die as soon as they reach a certain developmental stage. These findings complement previous studies showing that myelinating oligodendrocytes tend to be present in only very small numbers in the centre of chronic multiple sclerosis lesions (Prineas and McDonald, 1997; Lassmann, 1998; Mews et al., 1998; Wolswijk, 1998b) and are consistent with the limited success of myelin repair during the chronic stage of the disease (Prineas and McDonald, 1997; Lassmann, 1998). Whether oligodendrocyte differentiation of precursor cells is inhibited by factor(s) present in the lesion area or whether intrinsic factor(s) play a role remains unclear. These factor(s) may possibly not play a role early on in the disease, as antigenically immature oligodendrocytes are frequently present in early multiple sclerosis lesions (Prineas et al., 1989; Brück et al., 1994; Ozawa et al., 1994). Further insight into this and into the failure of the oligodendrocyte precursor population to expand may provide clues as to how the repair of demyelinated lesions in multiple sclerosis may be promoted therapeutically, e.g. by supplying lesions sites with putative mitogenic factors for these cells, such as PDGF, basic fibroblast growth factor (FGF-2) and glial growth factor-2 (GGF2) (Wolswijk and Noble, 1989, 1992; Wolswijk et al., 1991; Engel and Wolswijk, 1996; Shi et al., 1998), and/or by reducing the local production of remyelination inhibitory factors.

Ozawa and colleagues examined the generation of new oligodendrocytes in lesions of chronic-stage multiple sclerosis in a different way (Ozawa et al., 1994). By comparing in separate sections the number of oligodendrocytes expressing mRNAs for proteolipid protein, a marker for oligodendrocytes engaged in the synthesis and maintenance of myelin, with the number of MOG-positive, fully matured cells, they found that oligodendrocytes containing proteolipid protein mRNA outnumbered those that expressed MOG. These findings thus suggest that such lesions contained cells that were engaged in myelination, but were not fully differentiated, which is consistent with the possibility that small numbers of new oligodendrocytes are generated in chronic-stage multiple sclerosis lesions, either from the pool of precursor cells or from dedifferentiated, mature oligodendrocytes. It is important to note, however, that the immature oligodendrocytes in the study of Ozawa and colleagues were at a much later developmental stage than those identified in the present study, as the majority of these were non-myelinating.

The vast majority of the GalC-positive oligodendrocytes in the demyelinated area of the chronic-stage multiple sclerosis lesions studied were rounded cells that appeared not to be engaged in myelination. Although such rounded oligodendrocytes or oligodendrocyte-like cells have been observed previously in both early and chronic lesions (e.g. Raine et al., 1981; Prineas et al., 1989; Brück et al., 1994; Ozawa et al., 1994), their origin has remained unclear. The data presented here, together with earlier observations by other investigators, suggest that these cells are probably demyelinated oligodendrocytes, i.e. mature oligodendrocytes that have survived the loss of the myelin sheath during bouts of disease activity, rather than newly generated, non-myelinating oligodendrocytes. First, like mature oligodendrocytes, the rounded cells expressed MOG, a marker for terminally differentiated oligodendrocytes (Ludwin, 1990; Pfeiffer et al., 1993; Piddlesden et al., 1993), and had a small, spherical nucleus (Brück et al., 1994; Ozawa et al., 1994; this study). Secondly, the highest number of rounded oligodendrocytes was found in areas of the lesions with the most recent demyelinating activity, as evidenced by the presence of large numbers of debris-laden macrophages (this study). Thirdly, the lesion environment did not prevent the expression of a process-bearing morphology, as the small numbers of immature oligodendrocytes in the lesion areas did extend processes and, moreover, were sometimes found in the same area of the lesion as the rounded oligodendrocytes (this study). In other words, if these cells had been generated by oligodendrocyte precursor cells, it seems most likely that they would have displayed a process-bearing morphology. If the rounded oligodendrocytes in both acute/early and chronic lesions are indeed demyelinated, mature oligodendrocytes, then this suggests that in at least some cases of multiple sclerosis the myelin sheath and not the oligodendrocyte cell body is the primary and specific target of the disease process.

The distribution and density of the demyelinated oligodendrocytes in the various types of chronic-stage lesions in the collection suggests the following scenario for lesion development in at least some cases of multiple sclerosis. During episodes of demyelinating damage, mature, myelinating oligodendrocytes lose their myelin sheath. These demyelinated oligodendrocytes may perhaps re-form new processes and remyelinate denuded axons during early stages of multiple sclerosis, while during the later stages of the disease factor(s) present in the lesion area may prevent this; it is not yet clear whether demyelinated oligodendrocytes are able to generate new myelin sheaths. Without perhaps the trophic support from axons, or because of the absence of diffusible oligodendrocyte survival factor(s), the demyelinated oligodendrocytes gradually disappear starting from the centre, so that at later stages of lesion formation the rounded oligodendrocytes are found only at the lesion edges. This loss of oligodendrocytes from lesion areas is probably a very slow process, taking weeks or months, as very few oligodendrocytes appear to be undergoing degeneration in chronic-stage multiple sclerosis lesions (Raine, 1997; G. Wolswijk, unpublished observations) and as the demyelinated oligodendrocytes are often found in areas completely devoid of macrophages. This view of lesion development is supported by the results obtained from the analysis of seven lesions derived from multiple sclerosis subject 96-040 (lesions G–M). Lesions H and I, which were the most recent lesions of the seven, as evidenced by the presence of small numbers of MBP-positive macrophages, were packed with large numbers of demyelinated oligodendrocytes and macrophages laden with myelin debris. Lesion G lacked MBP-positive macrophages and had a centre that contained fewer rounded oligodendrocytes and debris-laden macrophages than lesions H and I, while the remaining four lesions had a hypocellular centre lacking myelin and oligodendrocytes. In these lesions, the rounded oligodendrocytes were present only at their (macrophage-containing) borders. If this scenario of lesion development for some types of multiple sclerosis is correct, then myelin repair in multiple sclerosis may be promoted therapeutically by increasing the local concentrations of factors that increase the survival of demyelinated oligodendrocytes and promote the regeneration of their myelin-forming processes.

In contrast to the rounded oligodendrocytes examined in the present study, Schönrock and colleagues found that some early multiple sclerosis lesions (biopsy material) contained small numbers of MOG-positive cells with apparently no processes that expressed the proliferation antigen recognized by the Ki-67 antibody (Schönrock et al., 1998). This observation suggests that mature, demyelinated oligodendrocytes may be proliferatively active in lesions that develop during the early course of multiple sclerosis; however, no mitotic figures were observed in these oligodendrocytes [In fact, mitotic oligodendrocytes have never been observed in multiple sclerosis lesions (Raine, 1997).] The proliferation of oligodendrocyte precursor cells was not analysed in this study, but a proportion of the Ki-67-positive cells were not labelled with markers for oligodendrocytes, astrocytes or microglia/macrophages. It is thus possible that some of the Ki-67-positive nuclei belonged to oligodendrocyte precursor cells in the lesion areas.

It is becoming clear that significant numbers of oligodendrocyte precursor cells may remain in many chronically demyelinated multiple sclerosis lesions and that mature oligodendrocytes may initially survive the loss of their myelin-forming processes. Together with the observation that spontaneous myelin repair is often a prominent feature of lesions that develop during the early course of multiple sclerosis, these findings indicate that the potential for repair of demyelinated multiple sclerosis lesions is more pronounced than previously thought. It remains to be determined whether the oligodendrocyte precursor population can be stimulated therapeutically to expand and generate remyelinating oligodendrocytes and whether demyelinated oligodendrocytes can be stimulated to form new myelin sheaths.

Acknowledgments

The author wishes to thank the team of the NBB (Anke de Boer, Hans Daniëls, Bart Fisser, Mariann Fodor, André Goessen, Stephan Guldenaar, Anne Holtrop, Marina Kahlmann, Wouter Kamphorst, Michiel Kooreman, Elly de Nijs, Sylvia Pindak, Rixt Riemersma, Ahmad Salehi, Dick Swaab, Unga Unmehopa, Paul van der Valk, Michiel Vermaak, Richard Vos, Rob de Vries, Harry Winters and José Wouda) for collecting the multiple sclerosis tissue and for advice, Sara Piddlesden and Hans van Noort for their generous gift of antibodies, Gerben van der Meulen for assistance with the preparation of the figures and Mark Noble, Dick Swaab and Paul Dijkhuizen for critical reading of the manuscript. Financial support for this study came from the Netherlands Foundation `Friends MS Research', the Multiple Sclerosis Society of Great Britain and Northern Ireland, and BIOGEN Inc. (Cambridge, Mass., USA). Human brain tissue was obtained from the Netherlands Brain Bank (NBB) in Amsterdam (Coordinator Dr R. Ravid).

References

View Abstract