Progenitor cells of the adult mouse subventricular zone proliferate, migrate and differentiate into oligodendrocytes after demyelination.

Identifying a source of cells with the capacity to generate oligodendrocytes in the adult CNS would help in the development of strategies to promote remyelination. In the present study, we examined the ability of the precursor cells of the adult mouse subventricular zone (SVZ) to differentiate into remyelinating oligodendrocytes. After lysolecithin‐induced demyelination of the corpus callosum, progenitors of the rostral SVZ (SVZa) and the rostral migratory pathway (RMS), expressing the embryonic polysialylated form of the neural cell adhesion molecule (PSA‐NCAM), increased progressively with a maximal expansion occurring after 2 weeks. This observation correlated with an increase in the proliferation activity of the neural progenitors located in the SVZa and RMS. Moreover, polysialic acid (PSA)‐NCAM‐immunoreactive cells arizing from the SVZa were detected in the lesioned corpus callosum and within the lesion. Tracing of the constitutively cycling cells of the adult SVZ and RMS with 3H‐thymidine labelling showed their migration toward the lesion and their differentiation into oligodendrocytes and astrocytes but not neurons. These data indicate that, in addition to the resident population of quiescent oligodendrocyte progenitors of the adult CNS, neural precursors from the adult SVZ constitute a source of oligodendrocytes for myelin repair.

Keywords: CNS; gliogenesis; neural stem cells; remyelination; subventricular zone

The subventricular zone (SVZ) of the lateral ventricle of the telencephalon is an important germinal layer that forms during development. In neonates, the SVZ gives rise to astrocytes and oligodendrocytes ([28]) and neurons of the olfactory bulb ([34]). During gliogenesis, glial precursors originating in the SVZ migrate radially to the corpus callosum and cerebral cortex where they differentiate into oligodendrocytes and astrocytes. In the juvenile brain the SVZ, which atrophies progressively after birth in mammals, remains mitotically active ([48]; [20]) and retains the capacity to generate astrocytes and oligodendrocytes in the corpus callosum ([43]; [27]). In the mature brain, the most anterior part of the SVZ (SVZa) still contains multipotential progenitor cells ([7]), giving rise to new granular and periglomerular neurons in the olfactory bulb ([32]). These progenitors elicit long‐distance migration, in a web‐like pattern called the rostral migratory stream (RMS), as defined by the expression of the polysialylated form of the neural cell adhesion molecules (PSA‐NCAM) ([47]). In vitro, this constitutively and rapidly proliferating cell population has the potential to differentiate into neurons and glia ([31]). Their differentiation into astrocytes, oligodendrocytes or neurons is under the control of relevant growth factors ([13]; [17]). In vivo, the fate of these cells seems to be either cell death or neuronal differentiation after migration into the olfactory bulb ([5]; [37]). However, treatment with growth factors induces expansion of the SVZ and promotes differentiation of these cells into neurons and astrocytes ([6]), as well as into oligodendrocytes ([6]; [22]). Nevertheless, the ability of SVZ progenitor cells to generate glial cells seems to be restrained in adulthood and remains undermined in pathological status such as after myelin lesions.

Several reports have demonstrated the existence in the adult CNS of a bi‐potential cell, the adult O‐2A progenitor ([8]; [55]) which, like its perinatal counterpart, can give rise to oligodendrocytes and astrocytes in vitro. These slowly proliferating progenitors can be converted to rapidly proliferating cells after treatment with platelet‐derived growth factor (PDGF) and basic fibroblast growth factor (bFGF) ([54]). They can be detected in the adult rodent brain by their expression of PDGF‐αR ([38]), NG2‐proteoglycan ([40]) or O4 antigen ([45]). Recently, the involvement of these endogenous progenitors in remyelination has been clearly shown. After a demyelinating lesion, these cells are able to generate remyelinating oligodendrocytes ([10]; [4]). Post‐mitotic oligodendrocytes that survive in a demyelinating lesion fail to contribute to remyelination ([18]). Therefore, the resting progenitors are considered to constitute the major source of remyelinating oligodendrocytes in the adult CNS. Furthermore, these cells appear not to migrate in the adult CNS and their recruitment into demyelinated lesions is restricted to a narrow zone around the area of demyelination, suggesting that repetitive demyelinating lesions could exhaust the pool of remyelinating cells in such area ([4]; [19]). Thus, the identification of other cells in the adult CNS with the capacity to give rise to remyelinating oligodendrocytes would help in the development of strategies to promote remyelination. In the present report, we have analysed the ability of cells of the adult SVZ to generate oligodendrocytes in response to lysolecithin‐induced demyelination of the corpus callosum. Using a combination of immunohistochemistry and autoradiography, we demonstrate that neural progenitors in the SVZa proliferate, migrate dorsolaterally into the demyelinated corpus callosum, and there give rise to oligodendrocytes. These observations highlight the role of the adult SVZ in CNS remyelination, and could have important implications for the development of therapeutic approaches of the demyelinating diseases such as multiple sclerosis.

Three‐month‐old OF1 mice were purchased from IFFA‐CREDO (Les Oncins, France). All animal protocols described were in accordance with the guidelines published in the National Institute of Health Guide for the Care and Use of Laboratory Animals.

Demyelination was induced by stereotaxic injection of 2 μL of a solution of 1% lysolecithin (LPC; Sigma, St Louis, MO, USA) in 0.9% NaCl ([14]; [12]). Mice were deeply anaesthetized and positioned in a stereotaxic frame (Stoelting, Wood Dale, IL, USA). The demyelinating agent was injected unilaterally into the corpus callosum with a 5‐μL Hamilton syringe, using the stereotaxic co‐ordinates of 5.5 mm anterior to the lambda, 1 mm lateral to the bregma and 2.5 mm deep from the skull surface. The injection site was labelled with charcoal dust. Scalp incisions were closed with Vetbond thread.

In order to assay cell proliferation of neural progenitors in the SVZ after demyelination, LPC‐injected mice received two intraperitoneal injections of 3H‐thymidine at an interval of 3 h (6 μCi/g each; specific activity 25 Ci/mmol, Amersham, UK). Animals were perfused 3 h after the last injection. These short 3H‐thymidine pulses were chosen to label proliferating cells at the S‐phase ([21]). Labelling with 3H‐thymidine was analysed within the SVZ at 2, 7 and 15 days after demyelination and in unlesioned control mice.

To trace the migration of SVZ cells to the lesion, mice received first two injections of 3H‐thymidine at an interval of 3 h (6 μCi/g each, i.p). Twenty‐four hours following the last 3H‐thymidine injection, they were then injected with LPC as described above and perfused at 1, 2, 3, 7, 15 and 30 days after demyelination. Because the clearing time of free 3H‐thymidine is <  12 h, this schedule was selected to minimize potential 3H‐thymidine uptake in response to demyelination ([32]). Consequently, labelled cells detected in the corpus callosum after demyelination are assumed to originate from the SVZ or RMS.

Mice (n=  4 for each time‐point) were killed 24, 48 and 72 h, and 7, 15 and 30 days after LPC injection by deep anaesthesia with ketamine (150 mg/kg, Imalgene, Rhone‐Merieux, Lyon, France) and acepromazine (Centravet, Plancoet, France). For immunohistochemistry and autoradiography, they were perfused intracardially with 4% paraformaldehyde in phosphate buffered saline (0.1 m PBS, pH 7.4). For β‐galactosidase histochemistry, animals were perfused with the same fixative containing 0.25% glutaraldehyde. Brains were removed, kept in the same fixative for 2 h and stored overnight at 4 °C in a 20% sucrose/0.1 m PBS solution. They were then frozen in isopentane ( − 60 °C) and stored at − 20 °C until use. Sagittal sections (10 mm thickness) were obtained using a Reichert‐Jung cryostat (Leica, Rueil‐Malmaison, France) and collected on gelatin‐coated slides.

The following antibodies were used: a mouse monoclonal anti‐MenB antibody (mouse IgM, Dr Rougon, Marseille, France) which specifically recognizes PSA‐NCAM ([46]; [39]); a mouse monoclonal anti‐myelin basic protein, MBP (IgG1, clone MIG M19; Valbiotec, Paris, France) was used to stain myelin sheaths and mature oligodendrocytes; a rabbit polyclonal anti‐type II carbonic anhydrase (CAII) was used as a marker for oligodendrocytes ([11]; [25], [26]; [24]); and a rabbit anti‐glial fibrillary acidic protein (GFAP; Dakopatts, Glostrup, Denmark) to identify astrocytes. Carbonic anhydrase type II is a general marker for the oligodendrocyte lineage but identifies a subclass of GFAP‐positive cells in the subcortical grey matter ([3]). Double‐labelling of CAII and GFAP allows a clear identification of oligodendrocytes expressing CAII but not GFAP, from astrocytes which may express both markers. The Tuj1 antibody from Dr Frank‐furter ([23]) was used to detect neuroblasts ([35]) and the 2F11 anti‐neurofilament antibody (Sigma, St Louis, MO, USA), to detect mature neurons ([15]).

For immunostaining, sections were rehydrated in 0.1 m PBS (pH 7.4) and variously fixed (100% acetone at − 20 °C for 10 min for PDGF‐αR, 90% acetone at − 20 °C for GFAP and CAII, 90% ethanol at − 20 °C for PSA‐NCAM, 100% ethanol at room temperature for MBP). After saturation with 10% normal swine serum in 0.1 m PBS for 30 min, sections were incubated overnight at 4 °C, with the primary antibodies at the following dilutions: anti‐MenB, 1 : 200; anti‐MBP, 1 : 100. After several washes in PBS, they were incubated for 2 h at room temperature with the respective secondary antibodies: FITC‐conjugated swine antirabbit IgG (Dakopatts, Glostrup, Denmark); TRITC‐conjugated goat antimouse IgG1 (Southern Biotechnology, Birmingham, AL, USA), FITC‐conjugated goat antimouse IgM Fab fragment (Southern Biotechnology). Sections were washed in 0.1 m PBS and coverslipped with Fluoromount medium (Southern Biotechnology). Slides were observed under fluorescence on a Leica DMRB microscope. Sections from control experiments, in which either the primary antibodies were omitted or the normal preimmune serum used, were unlabelled.

Sections were processed as above for indirect immunohistochemistry for PSA‐NCAM, GFAP and CAII, defatted in 70% and 95% ethanol and then dipped into melted Nuclear Track emulsion NTB2 (Kodak, Europeenne d'Imagerie Scientifique, Massy, France). Slides were then stored in the dark at 4 °C for 4 weeks, developed using Kodak D‐19 developer, washed extensively in water, counterstained with Hoechst 33342 (Sigma, St Louis, MO, USA) rapidly dehydrated and mounted in Fluoromount mounting medium.

Cells labelled with 3H‐thymidine were quantified in the lesion, the SVZa and the corpus callosum determining a fixed area box over each structure. The lesion site was identified by the presence of charcoal, and the demyelinated area defined by immunohistochemistry for MBP (see Fig. 1C) on sections 100 μm apart. Moreover, the lesioned area was estimated by computing image analysis using the Visioscan System (Biocom, Les Ulis, France). For each animal, five different levels 100 μm apart were analysed and for each level, three sections 10 μm apart were examined. The number of 3H‐thymidine labelled cells counted in each region were averaged from 15 tissue sections and expressed in numbers of cells/mm2, using the area calculated from the computing image analysis. The data in Fig. 6 represent the mean of counts ± SD per surface unit. Three animals were analysed for each time‐point.

Graph: 1 Demyelination model of the corpus callosum. Sagittal sections of the adult mouse brain. (A) The area of the stereotaxic injection of lysolecithin. (B) The SVZa and RMS after immunohistochemistry for PSA‐NCAM; the injection site (star at level b) is localized in the corpus callosum midway from the SVZa (level a) and the enlargement of the RMS (level c) entering the olfactory bulb. Three days after LPC injection, immunohistochemistry for (C) MBP and (D) GFAP indicates a myelin palor and astrocyte reactivity at the site of injection (arrow). (E) Electron microscopy showing demyelinated axons in the lesion 7 days postinjection. (F) Myelinated axons of the intact white matter. Magnification: B, ×  45; C and D, ×  70; E, ×  23 000; F, ×  25 000.

Focal demyelination was induced in the corpus callosum after stereotaxic injection of 2 μL of 1% LPC into the rostral corpus callosum of 4‐month‐old ABY mice, 1.5 mm rostral to the SVZ (Fig. 1A). As previously reported, neural progenitors located in the SVZa and RMS could be detected by their PSA‐NCAM immunoreactivity (Fig. 1B). Three days after the lesion, immunohistochemical analysis of the brains showed a focal depletion of the MBP immunolabelling at the injection site (Fig. 1C), circumscribed by a slight increase in GFAP‐positive astrocytes at its edge (Fig. 1D). Seven days after LPC injection, ultrastructural analysis of the corpus callosum revealed, as expected, the presence of denuded axons in the lesion (Fig. 1E) contrasting with myelinated axons in the unlesioned corpus callosum (Fig. 1F). Macrophages and a few astrocytes were also seen in the demyelinated area (not shown). As previously described ([14]), the lesion became progressively remyelinated after the first week of LPC injection (not shown).

To determine whether neural progenitors, located in the SVZa, are able to respond to demyelination, sagittal brain sections of mice, killed at various time points after LPC injection, were immunolabelled with the anti‐MenB antibody which specifically recognizes the polysialic acid residues of embryonic NCAM ([46]). In the adult rodent CNS, this antibody is a useful marker of undifferentiated neuroectodermal cells located in the SVZa and RMS. On sagittal sections of the unlesioned adult brain, PSA‐NCAM immunolabelling was detected on a few cells clustered in the SVZa and organized as a chain of bipolar cells migrating to the olfactory bulb (Fig. 2A) as previously reported ([47]). In control mice, PSA‐NCAM‐positive cells were never detected in the corpus callosum or striatum (Fig. 2A). Two days after demyelination, the number of PSA‐NCAM‐labelled cells was slightly increased within the SVZa (Fig. 2B). The expansion of the SVZa revealed by PSA‐NCAM immunohistochemistry was more obvious one week after demyelination (Fig. 2C). By 2 weeks a large increase in PSA‐NCAM‐immunoreactive cells in both the SVZa and along the RMS correlated with the maximal enlargement of these zones (Fig. 2D). At this stage, the increased number of PSA‐NCAM‐positive cells in the SVZa correlated with an increase in 3H‐thymidine incorporation in this structure (Fig. 2E). Likewise, the expansion of PSA‐NCAM‐expressing cells in the RMS was also correlated with an amplification of the cell proliferation in this region (not shown). Sagittal brain sections of unlesioned adult mice showed the presence of a narrow rim of 3H‐thymidine‐positive cells restricted to the SVZa and RMS (Fig. 2F), as previously demonstrated ([48]; [30]; [44]; [37]). In saline‐injected animals, 3H‐thymidine incorporation in the SVZa was not enhanced in comparison with unlesioned animals. These results indicate that demyelination of the corpus callosum enhances cell proliferation in the SVZa and RMS.

Graph: 2 Expansion of neural cells of the adult SVZa in response to LPC‐induced demyelination of the corpus callosum. Immunodetection of PSA‐NCAM on sagittal brain sections through the SVZa. (A) Control. (B–D) Lesioned mice at 2 (B), 7 (C) and 15 (D) days after LPC injection. In A, a few PSA‐NCAM‐positive cells are clustered around the wall of the lateral ventricle (arrow) and in the rostral migratory stream. A progressive expansion of the SVZa is noted from 2 (B) to 7 (C) and 15 days (D) after demyelination. Arrowheads point, in B, to small bipolar cells which are dispersed in the corpus callosum and, in D, to chains emerging from the SVZa into the corpus callosum. (E) 3H‐thymidine‐labelling of proliferating cells within the SVZa at 15 days after demyelination shows an increase in 3H‐thymidine‐labelled cells compared with (F) unlesioned control mice. In B–E, the lesion is located at the left side of the illustrated area. Magnification: A–D, ×  225; E–F, ×  275.

In addition to their progressive accumulation in the SVZa and RMS, PSA‐NCAM‐positive cells were also detected in the striatum, and especially in the corpus callosum. While a few PSA‐NCAM‐positive cells were detected in the corpus callosum close to the SVZ 48 h after LPC injection (Fig. 2B), their number in this structure increased in a spatio‐temporal fashion. By 15 days, PSA‐NCAM‐immunostained cells were detected as small chains emerging dorsolaterally from the SVZa (Figs 2D, and 3A,B) and RMS (Fig. 3C) at all levels between the SVZa and the lesion. Within the corpus callosum, the small chains of closely aposed and intensely stained PSA‐NCAM‐positive cells were frequently detected between myelinated fibers (Fig. 3A) and usually along astrocyte processes stained for PSA‐NCAM and GFAP (Fig. 3B,C). Some coexpressed PSA‐NCAM and the oligodendrocyte marker CAII (Fig. 3B). Cells positive for PSA‐NCAM were detected at the lesion site as early as 72 h after LPC injection (Fig. 3D). Among these cells, a few of them with a round shape were also found to express CAII, thus suggesting their differentiation into oligodendrocytes. Moreover, none of these cells expressed the neuronal neuroblast antigen Tuj1 or the neurofilament antigen 2F11 (not shown). In saline‐injected mice, expansion of the SVZa and emergence of PSA‐NCAM‐positive chains within the corpus callosum did not occur, but few isolated PSA‐NCAM + /GFAP +  positive cells were detected in this structure (not shown).

Graph: 3 Emergence of SVZ cells in the demyelinated corpus callosum, 15 days after demyelination. (A) Double immunolabelling for MBP (red) and PSA‐NCAM (green) at the SVZa level. PSA‐NCAM‐expressing cells emerging from the SVZa are detected between myelinated axon bundles of the corpus callosum. (B) Adjacent section stained for PSA‐NCAM (green), GFAP (red) and CAII (blue); small chains of PSA‐NCAM‐expressing cells emerging in the corpus callosum (facing arrowheads, lower insert) are often associated with astrocytic processes stained both for PSA‐NCAM and GFAP. A few emerging PSA‐NCAM‐positive cells were found to express CAII but not GFAP (arrow, see upper insert). (C) Double immunostaining for PSA‐NCAM (green) and GFAP (red) at the level of the lesion illustrating small chains of PSA‐NCAM cells leaving the RMS and entering the lesion (L). (D) Detection of small PSA‐NCAM‐expressing cells (green) within the lesion; a few of them express the oligodendroglial marker CAII (blue) (arrow, see insert) in the absence of GFAP (red); the arrowhead in (D) points to a doublet of small cells, one of which express PSA‐NCAM and the other CAII. PSA‐NCAM labelling is shown in fluorescein, MBP and GFAP are detected in rhodamine and CAII in coumarine. All views are from sagittal sections. Magnification: A–C, ×  214; D, ×  249; upper inset B, ×  1990, lower inset B, ×  300 inset D, ×  1395.

To trace the migration of SVZa and RMS progenitors towards the lesion, and their possible glial differentiation, mice were injected with 3H‐thymidine 24 h before demyelination. In the absence of demyelination, 3H‐thymidine incorporation was principally confined to the SVZ of the lateral ventricle and the RMS during the first week after injection (see Fig. 2F). The clearing time of free 3H‐thymidine being <  12 h, the presence of cells double‐labelled for 3H‐thymidine and PSA‐NCAM at locations other than the RMS and SVZ in lesioned animals implies that they originate from the RMS or SVZ. When the same 3H‐thymidine pulses were followed by LPC injection, a large number of 3H‐thymidine‐labelled cells were detected at the SVZa 3 days after demyelination (Fig. 4A) but were absent from this structure at 15 days (Fig. 4B), consistent with the demonstration of their migration to the olfactory bulb ([32]). This decrease of labelled cells at the SVZ was correlated with the appearance of rare 3H‐thymidine labelled cells within the lesion as early as 3 days post‐lesion (Fig. 4C) and a larger number of them at 15 days post‐lesion (Fig. 4D) indicating the ability of the SVZ progenitors to migrate into the demyelinated area. Moreover, in some cases, 3H‐thymidine‐labelled emerging chains established a clear connection between the SVZa and the lesion (not shown). Immuno‐characterization of the 3H‐thymidine‐labelled cells with cell‐specific markers showed that at early times some of the 3H‐thymidine‐labelled cells present at the lesion stained for PSA‐NCAM (Fig. 5A,B) while at later times they stained for glial specific markers such as CAII (Fig. 5C,D) or GFAP (Fig. 5E,F) indicating that with time the PSA‐NCAM‐positive cells derived from the SVZ differentiate into either cells of the oligodendroglial or astrocyte lineage.

Graph: 4 3H‐thymidine tracing of SVZa or RMS cycling cells after demyelination. Mice received 3H‐thymidine 24 h prior to demyelination and were killed 3 days (A and C) and 15 days (B and D) after demyelination. (A) 3H‐thymidine‐labelled cells are detected at the SVZa at 3 days after demyelination; (B) they are absent at 15 days after demyelination. (C) Rare 3H‐thymidine‐labelled cells originated from the SVZa or RMS appear in the lesion (L) at 3 days atfer demyelination; (D) their number increased in this area 15 days after demyelination. The arrowheads point to examples of Hoechst‐labelled nuclei which are labelled with 3H‐thymidine. Magnification: A–C, ×  445.

Graph: 5 Immunocharacterization of SVZa‐ or RMS‐traced cells in the lesion. (A) General view showing the RMS, with PSA‐NCAM‐positive cells emerging from the RMS and entering the lesion (L). The arrow in A points to a cell whose Hoechst‐labelled nucleus is labelled with 3H‐thymidine. (C–F) Examples of SVZ‐derived oligodendrocytes stained for CAII (C) and SVZ‐derived astrocytes (arrow) stained for GFAP (E) which are localized within the lesion and are labelled with 3H‐thymidine (D,F) 15 days after demyelination. Magnification: A, ×  257; B, ×  711; C, D, ×  507; E, F, ×  484.

Quantification of the number 3H‐thymidine labelled cells detected in the SVZa, corpus callosum and within the lesion confirmed that in unlesioned animals, 3H‐thymidine‐labelled cells in the SVZa was mainly restricted to the SVZa at early times but progressively disappeared from this structure thereafter. In contrast, for lesions equally distant (1.5 mm) from the rostral SVZ, the number of 3H‐thymidine‐labelled cells present at the lesion was low at 3 days but reached a peak value 7 days after LPC injection (Fig. 6A). At this time, their number was 30 ×  the control value. When the ratio of cells double‐labelled for 3H‐thymidine and glial specific markers over the total SVZ‐recruited cells was evaluated in the lesion (Fig. 6B), the proportion of PSA‐NCAM‐positive cells labelled for 3H‐thymidine decreased with time from 20 ± 3 to 7.8 ± 4% at 3 and 30 days post‐injection, respectively. Meanwhile, CAII‐positive oligodendrocytes labelled for 3H‐thymidine increased from 0 to 18.5 ± 10% for the same time‐points (see Fig. 6B). Astrocytes in the lesion derived from SVZ, double‐labelled for 3H‐thymidine and GFAP, were estimated at 17 ± 0.4 and 15.2 ± 5% at 3 and 30 days after demyelination, respectively. These data further sustain that the immature cycling cells emerging from the SVZa or RMS are recruited to the lesion and differentiate into cells of the glial phenotype at the lesion.

Graph: 6 Quantification of the number of 3H‐thymidine‐labelled cells after LPC injection. (A) Total number of 3H‐thymidine‐labelled cells present in the lesion area. Data are the mean of counts ± SD per unit surface from three animals for each time point. C, unlesioned animal; LPC, lysolecithin lesioned animal; SVZ, subventricular zone; CC, corpus callosum. (B) Number of different cell types recruited by the lesion. Data in percentage represent the number of cells double‐labelled for 3H‐thymidine and a cell‐specific marker over the total recruited population at 3 and 30 days postlesion.

In the present study, we demonstrate that progenitor cells of the adult mouse SVZa contribute to gliogenesis after demyelination of the corpus callosum. Our results show (i) an expansion of the adult SVZ in response to demyelination; (ii) the emergence in the lesioned corpus callosum of small chains of PSA‐NCAM originating from the SVZa and RMS; (iii) cell tracing of the SVZa cells by 3H‐thymidine combined with their immuno‐characterization after the demyelination indicates that these cells are recruited by the lesion where they generate astrocytes and oligodendrocytes.

In vitro, the adult SVZ generates neurons and astrocytes ([31]; [34]) but also oligodendrocytes ([29]). While in the adult mammalian brain, progenitor cells lining the antero‐dorsal wall of the lateral ventricle generate granule neurons of the olfactory bulb ([32]), treatment with epidermal growth factor (EGF), transforming growth factor α (TGFα) or fibroblast growth factor‐2 (FGF‐2) triggers their emigration and maturation into neurons, astrocytes and oligodendrocytes ([6]; [22]). Acute lesions of the CNS ([52]; [50]; [15]) or inflammation ([2]) also modulate these phenomena, and lead exclusively to SVZ‐derived astrogliogenesis ([50]; [15]). While in those studies oligodendrogliogenesis was not investigated, we report that, in response to demyelination, the adult SVZa generates, in addition to astrocytes, oligodendrocytes likely to be involved in the remyelination of the lesion. In the adult CNS, the hippocampus ([17]; [42]) and the spinal cord ([51]) also contain multipotential cells which in vitro give rise to oligodendrocytes, astrocytes and neurons. Interestingly, the presence of small immature cells expressing PSA‐NCAM ([39]) or nestin ([9]) were also observed in the peri‐ependymal area of the demyelinated or injured spinal cord. However, further investigation should establish the precise involvment of these peri‐ependymal cells in spinal cord remyelination. These data suggest that multiple structures through the adult CNS with potential gliogenesis may contribute to the renewal of remyelinating oligodendrocytes.

Several earlier observations in experimental and pathological demyelination have argued that mature oligodendrocytes that survive in demyelinating lesions may generate remyelinating oligodendrocytes (for review see [33]). However, recent studies have clearly demontrated that postmitotic oligodendrocytes that survive within a demyelinating lesion do not proliferate and do not contribute to remyelination in the adult CNS ([18]; [4]). Therefore, the resting progenitor cells that persist in the adult rodent and human CNS ([45]; [38]; [53]) appear to be the major source of remyelinating oligodendrocytes. Recently, [10], using retroviral tracing of cycling cells, provided evidence for the presence of a locally restricted population of oligodendrocyte progenitors and its involvement in remyelination. However, the incapacity of the adult white matter oligodendroglial population to self‐renew and to migrate from areas remote from the lesion ([10]; [4]; [19]) may restrict their contribution to the repair of proximal and single demyelination events. It is worth considering whether the precursors of the adult SVZ could be also a source of remyelinating oligodendrocytes. The apparently unlimited capacity of these cells to self‐renew and to migrate ([7]), combined with their capacity to give rise to newly generated oligodendrocytes, suggest that such cells could repopulate lesions after successive events of demyelination. Our data indicate, however, that only a small proportion of the recruited SVZ cells differentiate into mature oligodendrocytes within the lesion, suggesting their cooperation with the resident oligodendroglial progenitors to the remyelination of the lesion. While this modest contribution may reflect their limited survival or their differentiation into other cell types, the presence of a large number of SVZ‐derived astrocytes within the lesion supports the latter hypothesis.

The spatio‐temporal emergence in the demyelinated corpus callosum of SVZ‐derived astrocytes or oligodendrocytes indicates that SVZ cells migrate toward the lesion. Furthermore, these cells seem to emerge from the SVZa either dorsolaterally directly through the corpus callosum or after rostral migration through the RMS. Dorsolateral migration of SVZ‐derived oligodendrocyte progenitors also occurs during normal rodent postnatal development ([25], [26]) but in juveniles becomes progressively restricted to the unmyelinated subcortical white matter ([24]). In adult animals, the dorsolateral migratory pathway is shut and precursor cells migrate only rostrally to the olfactory bulb ([32]). Our present data show that in response to demyelination of the corpus callosum cells from the antero‐dorsal SVZ or the RMS once again migrate dorsolaterally through the corpus callosum.

In the developing brain, this type of migration probably occurs along radial glia. It may also occur along axons themselves and may be controlled by modulation of adhesion ([41]) and extracellular matrix molecules, integrins (Milner & ffrench[36]) or diffusible factors. While the latter could be up‐regulated after CNS injury ([1]), axophylic migration is unlikely to occur in fully myelinated white matter. Moreover, radial glia do not persist in the adult CNS ([49]). However, this migration may be supported by astrocytes because PSA‐NCAM‐positive cells leaving the SVZa or the RMS were often associated with perpendicularly orientated GFAP‐positive processes projecting into the corpus callosum. In the unlesioned brain, neural precursors migrating toward the olfactory bulb are channeled from the SVZa into the RMS through a furrow of tenascin‐C expressing astrocytes, which could prevent them entering the adjacent tissues ([16]; [49]). The detachment of these small PSA‐NCAM‐expressing cells from the SVZa or RMS suggests that induced demyelination disorganizes the astrocyte furrow and eases them to leave the SVZa and RMS. Recently, expansion of the SVZ and migration of SVZ‐derived cells in the corpus callosum were also reported after experimental allergic encephalomyeltis ([2]).

Our data argue in favour of a novel source of oligodendrocytes for myelin repair. Although the differentiation into oligodendrocytes seems to be limited, the possibility to enhance this potential by growth factor treatment ([6]; [22]) suggests that manipulation of neural precursors of the adult CNS may provide novel therapeutic approaches to demyelinating diseases such as multiple sclerosis.

We thank Alexandra Gonzales for excellent technical assistance in the transplantation studies. We thank Dr Rougon for her kind gift of the anti‐PSA antibody, Dr Gandhour for the CAII antibody, Dr Frankfurter for the Tuj1 antibody and Dr R. McKay for the antinestin antibody. This study was supported by INSERM, the Association pour la Recherche sur la Sclerose en Plaques (ARSEP), and Berlex (CA, USA). During this study, B.N.O. was a fellow of the 'Societe des Amis des Sciences'.

• Abbreviations
• CAII carbonic anhydrase type II
• bFGF basic fibroblast growth factor
• GFAP glial fibrillary acidic protein
• LPC lysolecithin
• MBP myelin basic protein
• N‐CAM neural cell adhesion molecule
• PBS phosphate‐buffered saline
• PDGF platelet‐derived growth factor
• PDGF‐αR platelet‐derived growth factor alpha‐receptor
• PSA polysialic acid
• RMS rostral migratory pathway
• SVZ subventricular zone
• SVZa anterior subventricular zone