extensive myelination

Proc. Natl. Acad. Sci. USA Vol. 91, pp. 11616-11620, November 1994 Neurobiology

Transplantation of an oligodendrocyte cell line leading to extensive myelination ULRIKE TONTSCH*t, DAVID R.

ARCHERO§, MONIQUE DUBOIS-DALCQ*, AND IAN D. DUNCANt¶

*National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, MD 20892; and tSchool of Veterinary Medicine, University of Wisconsin, Madison, WI 53706

Communicated by Richard L. Sidman, July 14, 1994 (received for review April 4, 1994)

Oligodendrocytes, the myelin-forming cells of ABSTRACT the central nervous system, can be generated from progenitor cell lines and assayed for their myelinating properties after transplantation. A growth-factor-dependent cell line of rat oligodendrocyte progenitors (CG4) was carried through 31-48 passages before being transplanted into normal newborn rat brain or the spinal cord of newborn myelin-deficient (md) rats. In md rat spinal cord, CG4 oligodendrocyte progenitors migrated up to 7 mm along the dorsal columns, where they divided and myelinated numerous axons 2 weeks after grafting. CG4 cells were transfected with the bacterial lacZ gene and selected for high -galactosidase expression. The cell migration and fate of these LacZ+ cells were analyzed after transplantation. In normal newborn brain, LacZ+ oligodendrocyte progenitors migrated along axonal tracts from the site of injection and integrated in the forming white matter. In md rats, extensive migration (up to 12 mm) was revealed by staining for 3- galacsdase activity of the intact spinal cord where many grafted cells had moved into the posterior columns. Similar migration and integration of grafted cells occurred in the spinal cord of normal myelinated rats and after a noninvasive graft procedure. Thus, oligodendrocyte progenitors can maintain their ability to migrate and myelinate axons in vivo after multiple passages in vitro. Such progenitor cell lines can be used to study the molecular mechanisms underlying oligodendrocyte development and the repair of myelin in dysmyelinating diseases. To study mechanisms of neural development, rodent cell lines have been established from precursors of central nervous system (CNS) cells that maintain their ability to differentiate after transplantation into the developing rodent nervous system (1, 2). When grafted into the nervous system, multipotent stem cells can give rise to neurons capable of synapsing with host neurons (1, 2). The therapeutic potential of grafting such cells in damaged CNS tissues has been discussed, as such cell lines circumvent problems raised by limited availability of donor tissue for transplantation studies (3). In addition to cell lines able to generate neurons after grafting, glial progenitor cell lines have been established that can generate oligodendrocytes, the CNS myelin-forming cells. Oligodendrocytes are derived from precursor cells that, in response to specific growth factors, divide, migrate in the developing white matter, and eventually myelinate axons (e.g., see refs. 4-7). As myelin facilitates rapid impulse conduction along most nerve tracts, the lack or loss of myelin can result in important neurological dysfunction (8, 9). Although the development of the oligodendrocyte lineage has been studied extensively in vitro (4-7), the molecular mechanisms controlling the growth and migration of oligodendrocyte progenitors (OPs) in vivo, the recognition of axons, the coordinate synthesis of multiple myelin inter-

nodes, the regulation of myelin protein genes, and the death of superfluous oligodendrocytes are largely unknown. An understanding of these molecular mechanisms could be approached by using a growth-factor-dependent cell line in which gene transfer or targeting could be performed and tested in a transplantation paradigm. In addition, such myelin-forming cell lines could be used to explore ways to experimentally correct genetic defects identified in several inherited dysmyelinating diseases and leukodystrophies of animals and humans (10, 11). Transplanted rodent oligodendrocytes and their precursor cells as well as stem cells can myelinate axons in myelindeficient or demyelinated animals (12-19). Recent studies have shown that transplantation of early passages of OPs (20) expanded in the presence of growth factors or immortalized with a conditional oncogene (21), resulted in myelination of naked axons within an irradiated and chemically induced spinal cord lesion in adult rats. However, the myelinating potential of these grafted cells decreased considerably (20) or even ceased (21) after multiple passages. Immortalization of CNS cells by conditional oncogenes such as a thermosensitive simian virus 40 large tumor antigen is a useful strategy to establish CNS cell lines, as the oncogene is usually turned off after transplantation into rodents (1, 21). However, OP cell lines can simply be expanded in the presence of growth factors specific for this lineage. In this study, we have examined the myelinating properties after multiple passages ofbrain-derived progenitors established as a growthfactor-dependent cell line called CG4 (22, 23). We have transfected these cells with the bacterial lacZ gene, selected clones highly expressing (3-galactosidase, and analyzed cell migration and cell fate after transplantation into normal newborn rat brain and into the spinal cord of normal and myelindeficient (md) rats. Our results demonstrate that these OPs have excellent migratory potential when transplanted into the CNS and can lead to extensive myelination of axons in dysmyelinating rats (24, 25).

MATERIALS AND METHODS Cells and Animals. Newborn Sprague-Dawley rats were purchased from Taconic Farms and the breeding colony of md rats was maintained as described (23, 24). CG4 cells were kindly provided by J.-C. Louis (presently at Amgen Biologicals). Cells were expanded in T25 or T75 flasks in serum-free DMEM (GIBCO/BRL) supplemented with N1, biotin, and 30% B104 neuroblastoma conditioned medium and plated on poly(L-ornithine)-coated dishes as described (22, 23). Cultures passaged 30 to 41 times were used in the transplantation experiments described below. Abbreviations: CNS, central nervous system; OP, oligodendrocyte progenitor; X-Gal, 5-bromo4-chloro-3-indolyl ,(D-galactoside. tPresent address: Bender and Co., Dr Boehringergasse A-1120, Vienna, Austria. §Present address: Department of Pediatrics, Emory University, Atlanta, GA 30322. $To whom reprint requests should be addressed.

The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact. 11616

Neurobiology: Tontsch et al. LacZ Vector and Transfection. The pRSVZ plasmid vector leads to expression of lacZ as a fusion protein with a portion of the Drosophila melanogaster alcohol dehydrogenase protein under the transcriptional control of the avian sarcoma virus long terminal repeat (26). The pSV2neo vector contains the neomycin-resistance gene under control of the early simian virus 40 promoter and was obtained from the American Type Culture Collection. CG4 cells (passage 31) expressing the OP phenotype [as identified by labeling with the monoclonal antibody A2B5 (22)] were plated at 5 x 105 cells per 60-mm dish the day before transfection. Transfection was mediated by a synthetic, cationic lipopolyamine molecule (Transfectam; Promega) as recommended by the manufacturer. CG4 cells were cotransfected with pRSVZ (5 Zg of DNA per 60-mm dish) and pSV2neo (1 ug of DNA per 60-mm dish). Six days after transfection, clones were selected for neomycin resistance in the presence of G418 at 400 jig/ml. Approximately 2 weeks later G418-resistant clones were picked and expanded. Only clones containing more than 50% LacZ+ cells (see below) were expanded. Cultures passaged 43 to 48 times were used in transplantation experiments. Transplantation. In md rats. CG4 cells, untransfected or transfected with pRSVZ and pSV2neo, were grown in the presence of neuroblastoma conditioned medium as described (22) and shipped in flasks overnight. Cells (5 x 104 cells in 1-1.5 jId) were then grafted into the T13/L1 region ofthe spinal cord of 5- to 7-day-old md rats. In some cases, rats were injected i.p. with 0.002 jiCi of [3H]thymidine per g (50-90 Ci/mmol, 1 Ci = 37 GBq) 1.5 h before perfusion (see below). In normal newborn rats. LacZ+ CG4 clones (passages 43-48) were grown as described (22). Cells (5 x 104 cells in 1.5 IlI) were injected in the left subcortical region close to the radiatio corporis callosi with a Hamilton syringe. Control animals were injected with medium alone. Remaining CG4 cells were plated and tested for viability and ,B-galactosidase activity (see below). Fixation and Immnochemisry. md rats. Between 14 and 17 days after receiving a graft ofuntransfected CG4 cells, md rats were anesthetized with sodium barbital and perfused with a modified Karnovsky fixative. The areas of the CNS receiving the grafts were dissected out, processed for epoxy embedding, sectioned at 1 ,am, and stained with toluidine blue as described (24). [3H]Thymidine autoradiography was also performed on selected 1-,um sections. The md rats that received a graft of LacZ+ CG4 cells were perfused 2-3 weeks after injection with 4% paraformaldehyde or with 2% paraformaldehyde and 0.5% glutaraldehyde. The spinal cords at the transplant site were removed, rinsed three times in PBS and then incubated in X-Gal solution (5-bromo4-chloro-3-indolyl ,B-D-galactoside) for 15 h before being washed again in PBS. Postfixation with 2% paraformaldehyde and 2% glutaraldehyde was continued overnight. The incubated tissues showing blue reaction product were photographed before being sliced transversely through the X-Gal+, blue areas. Selected tissue slices were then embedded in Epon and stained with p-phenylenediamine (32). Normal newborn rats. Normal newborn rats injected with LacZ+ CG4 cells were anesthetized with avertin and fixed by cardiac perfusion with 2% paraformaldehyde in 1 M Pipes buffer (pH 6.9) at 3, 5, 7, 14, and 21 days after grafting. The brains were immersed in the same fixative for another 24 h. The fixed brains were then sliced transversely till the charcoal mark was found. Coronal 50 Am vibratome sections of these tissue slices were prepared and incubated in X-Gal solution or immunohistochemically stained with a rabbit anti-,B-galactosidase antiserum (27-30) [kindly provided by Jack Price (27) or obtained from 5 Prime -+ 3 Prime, Inc.] followed by a peroxidase-conjugated second antibody (Jackson ImmunoResearch) and visualized with diaminobenzidine

Proc. Natl. Acad. Sci. USA 91 (1994)

11617

as described (31). Brain slices were mounted with fluoromount and examined with a Zeiss axiophot microscope.

RESULTS In an initial series of experiments, we examined whether lacZ-transfected OPs migrate into the forming white matter of normal newborn syngeneic rats and integrate into the host tissue without abnormal proliferation. CG4 cells at passage 31 were cotransfected with pRSVZ and pSV2neo vectors as described. Stably transfected cells were selected in G418,

cloned, and expanded before being tested for (3-galactosidase expression. Only those clones containing 70-90% LacZ1 cells (detected by X-Gal staining or immunoperoxidase labeling) were used in transplantation experiments. These LacZ+ clones were indistinguishable from nontransfected CG4 cells with respect to proliferation in the presence of growth factors and differentiation into oligodendrocytes after growth factor withdrawal as assayed by expression of galactocerebroside and myelin basic protein (22) (data not shown). OPs (passage 40 and higher) transfected with lacZ were grafted into the left hemisphere of newborn rats. The animals were perfused 3-21 days after grafting, and Vibratome sections of the brains were immunostained with anti-/3galactosidase antibodies to localize the LacZ+ grafted cells (28, 29, 31). Three days after transplantation, CG4 cells in the graft were intensely stained, and some of them appeared to have migrated away from the graft site along nerve tracts and to have regrown processes (Fig. la). Between 3 and 7 days after transplantation, rows of LacZ+ cells often showed a bipolar shape typical of motile OPs in vitro (Fig. la, Inset) and were found aligned along axonal tracts within the corpus callosum. During the next 2 weeks, scattered LacZ+ cells with a more complex arborization pattern were found in myelinating tracts within the striatum in the same hemisphere where the graft had been placed (data not shown). The graft could no longer be found, suggesting that most living cells in the graft had dispersed. There was no evidence of uncontrolled proliferation of the CG4 cells nor of a host inflammatory response within the 3-week period studied. We then examined the myelinating potential of CG4 cells after transplantation into the spinal cord of md rats. As these rats have a point mutation in the major protein ofCNS myelin, proteolipid protein (PLP) (10), only rare, uncompacted myelin sheaths are detected. The majority of axons throughout the CNS of md rats are unmyelinated (24, 25). Previous experiments have shown that mixed glial cell preparations, which contain OPs, from normal littermates of md rats can, upon grafting into spinal cords of newborn md rats, lead to myelination at the site of injection in the thoracic dorsal columns (33). We used the same method to transplant an untransfected CG4 OP cell line expanded for 30-41 passages into md rats (Fig. 1 b-f). As shown in Table 1, 21 of 21 md rats grafted at 5 days of age exhibited myelination 2 weeks later in the dorsal tracts along a distance of up to 7 mm as visualized by the presence of a white streak (Fig. lb). In fact, the dorsal columns along this streak (Fig. lb) were almost entirely myelinated and contained numerous transplanted cells (Fig. 1 c and e, compare with d). When examined with the electron microscope, the newly formed myelin had a normal ultrastructure and the majority of glial cells appeared to be normal and metabolically active oligodendrocytes with prominent rough endoplasmic reticulum and ribosomes (data not shown). None of these cells had distended endoplasmic reticulum, the hallmark of md rat oligodendrocytes (24, 25). Although all rats in these experiments showed myelination by the grafted cells, in some cases, the CG4 cells had apparently migrated for only 2 to 3 mm from the glaft site. Some animals were given [3H]thymidine for 1.5 h to label proliferating glial cells in the CNS regions containig the transplanted cells. When autoradiography was performed on

11618

Proc. Natl. Acad Sci. USA 91

Neurobiology: Tontsch et al.

-

.VI

C

AR -.. ho./

(1994)

.

-

toIa+

-

0 0

..

A

Am4

£ Mob.-

..Moo* fimp

. ~ ~~~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~i

~~~~ h .-mok -Amp j h

;..

;,.~~~~~~~. V;

bAgwokadawk.-. -

.''

..w

i FiG. 1. (a) CG4 cells expressing P-galactosidase migrate into the normal forming white matter of newborn rats. Three days after transplantation, LacZ+ cells stained by the immunoperoxidase technique have migrated away from the graft (strongly stained area at left; arrow) along a nerve tract. Note a group of cells with a bipolar shape in the right corner (short arrows). (Inset) A higher magnification of some of the ming cells and their processes. (b-J) Transplantation of untransfected CG4 cells in md rat spinal cord. (b) Dorsal view of the spinal cord of an md rat transplanted 17 days earlier with nontransfected CG4 cells. The 7-mm-long horizontal white streak in the middle of this dorsal region is myelin made by the transplanted cells. (see c and d). (c and d) Dorsal columns showing numerous cells and extensive myelination from an md rat (c) that received untransfected CG4 cells (see e and f for detailed views) compared to a nongrafted md rat (d). (e and f) At higher power, numerous myelinated fibers (with darkly stained myelin rings around axons) and transplanted cells are detected in e while three cells labeled with [3H]thymidine (arrows) are observed in an autoradiograph inf. The cells undergoing DNA synthesis show several grains over their nuclei, have a higher cytoplasmic to nuclear ratio than differentiated oligodendrocytes, and are probably OPs. Of the total glial cells in the transplanted dorsal columns, 8.7% had incorporated [3H]thymidine, compared to 1.6% in the lateral and ventral columns devoid of grafted cells. (g4) Transplantation of LacZ+ CG4 cells in md rat spinal cord. In all cases the blue stain represents the X-Gal staining for P-galactosidase enzyme activity. (g) The transfected CG4 cells grown in neuroblastoma conditioned medium have a spindle shape typical of OPs. Injection of these precursor-stage cells into the md spinal cord resulted 2 weeks later in extensive migration of the cells throughout the dorsal columns, on one side of the grey matter, and the ventral column. (i) A 1-Mm epoxy section stained with p-phenylenediamine shows a number of CG4 cells with blue, cytoplasmic punctate X-Gal stain associated with myelinated fibers (grey rings around axon). For comparison, a nontransplanted md rat has 0-3 myelinated fibers in the whole dorsal column. Note that the blue X-Gal stain is not diffusing in the oligodendrocyte processes. (j) The dorsal region of an md rat is shown 2 weeks after injection of LacZ+ CG4 cells under the dura. A long (12 mm) streak of LacZ+ cells can be seen. When this cord was cut transversely (k), each segment showed blue reaction product within the dorsal column, indicating that grafted cells had penetrated the CNS parenchyma. (1) Normal newborn rats similarly injected with LacZ+ cells under the dura showed cell spreading more than 10-11 mm 2 weeks later.

toluidine-blue-stained sections, DNA synthesis was more apparent in the numerous cells in the areas of myelination

compared to areas without myelin (Fig. If). The 3H-labeled cells within the graft differed from the host labeled cells

Neurobiology: Tontsch et al.

Proc. Natl. Acad. Sci. USA 91 (1994)

11619

Table 1. Myelination in md rats transplanted with CG4 cells Type of CG4 Passage no. of No. of animals No. of animals cells implanted CG4 cells implanted receiving grafts with myelin Untransfected 30-41 21 md rats* 21 9 md rats IacZ transfected 43-48 8 lacZ transfected 46 3 normal ND ND, could not determine whether grafted cells had produced myelin. *Four animals were injected with CG4 cells that had a multipolar and more differentiated morphology.

outside the graft by their abundant cytoplasm and their round nuclei (Fig. 1 e and f) suggesting that they were cells of the oligodendrocyte lineage. Thus, grafted CG4 cells can proliferate, migrate into, and myelinate the md rat spinal cord. To further analyze the migratory properties of OPs, we transplanted CG4 clones (passages 43-48) stably expressing (3-galactosidase in up to 90% of the cells into md rat spinal cords (Fig. ig). Such LacZ+ CG4 cell clones myelinated dorsal columns of 8 of 9 md rat spinal cords after 2 weeks (Table 1). Fixed spinal cord segments taken from animals 2 weeks after grafting were incubated in X-Gal substrate, a procedure which stained the grafted cells blue and revealed in a particularly striking manner the extent of their migration in the md spinal cords (Fig. 1 h-i). Transplanted LacZ+ cells showed extensive migration similar to that seen previously with untransfected CG4 cells. LacZ+ cells were detected in the dorsal columns in all transverse sections taken along the migration path and had also migrated into the grey matter of the anterior and posterior horns, possibly to associate with some of the fiber tracts crossing to the other side of the spinal cord (Fig. lh). To determine whether such extensive migration also occurs in normally myelinated rats, three normal littermates of md rats received LacZ+ CG4 cells. Interestingly, these cells also showed extensive longitudinal migration in the myelinated dorsal columns (data not shown). To establish whether these migratory LacZ+ CG4 cells could make CNS myelin, we used a rapid epoxy-embedding procedure that preserves the X-Gal staining and counterstained myelin in the epoxy sections (32). A proportion of these grafted CG4 cells showed lacZ staining and a number of these cells were associated with myelinating axons (Fig. ii). However, the number of myelinated fibers in the dorsal columns of md rats transplanted with LacZ+ CG4 cells was less than that seen in md rats transplanted with untransfected CG4 cells (compare Fig. 1 e and i). When normal littermates were similarly grafted with untransfected or transfected cells, numerous CG4 cells were also found in dorsal columns next to myelinated fibers. However, it was not possible to determine whether the grafted cells had formed myelin sheaths because ,B-galactosidase staining did not extend into the processes. Finally, we examined whether transplanted cells could reach their appropriate destination when we used a less invasive method of grafting. LacZ+ CG4 cells were injected under the meninges to allow the cells to gain access to the CNS by disseminating or spreading by active migration. As expected, migration was enhanced in these conditions, extending along a total of 10-12 mm in one md rat and one normal littermate (Fig. 1 j-i). In both normal and md spinal cords, the engrafted CG4 cells had penetrated the dorsal columns at all levels of their migration trail as seen in cross sections of the md rat (Fig. 1k).

DISCUSSION Our results demonstrate that brain-derived OPs maintained as a growth-factor-dependent OP cell line (CG4) are able to survive and differentiate into myelin-forming cells upon transplantation into normal and dysmyelinated neural tissue. Most remarkably, untransfected CG4 cells between passages 30 and 41 maintain their ability to proliferate normally, migrate substantial distances along the dorsal tracts, and myelinate nu-

merous axons when transplanted in the developing spinal cord of the md rat. These observations significantly extend previous demonstrations of myelination by oligodendrocytes and their precursor cells grafted as tissue fragments or after primary culture (12-19, 33). In contrast to previous studies indicating a decreased ability of OP cell lines to myelinate CNS neurons with increasing time in culture (20, 21, 32), CG4 cells showed excellent migratory and myelinating properties even after multiple passages. Our data also suggest that OP cells, which are growth-factor dependent in vitro, may respond to signals leading to their regulated proliferation and differentiation in the intact animal. It seems unlikely that the transplanted cells could have induced the genetically deficient host oligodendrocytes of the md rat to make myelin. The myelin present in the area of grafting had a normal ultrastructure, with an intraperiod line unlike that seen in the sparse myelin produced in the nongrafted host (24). Moreover, the electron microscopic appearance of the oligodendrocytes in the graft area was normal, without the swelling of the rough endoplasmic reticulum characteristic of md rat oligodendrocytes (24). The transfection of CG4 cells with lacZ allowed us to distinguish unequivocally between grafted cells and host cells and to follow the migration and fate of the grafted cells. When clones with a high percentage of LacZ+ cells were grafted, the cells integrated well in the newborn brain and spinal cord and did not show abnormal proliferation after 3 weeks. Moreover, ,-galactosidase expression was sustained in the grafted CG4 cells for at least 3 weeks after transplantation, probably because of the strength of the Rous sarcoma virus promoter in the particular construct we used (26). This suggests that OP cells may allow the long-term expression of a variety ofgenes under the control of heterologous promoters both in vitro and in vivo after gaffing. In addition, (3-galactosidase expression revealed by the presence of a colored reaction product turned out to be particularly useful for determining the extent of migration of grafted LacZ+ OPs in the md rat spinal cord. These cells showed remarkable migratory properties, and, even when injected under the dura, they were able to migrate up to 12 mm from the injection site and integrate into the forming white matter. Moreover, grafted OPs integrated well both in normal and in dysmyelinated spinal cord tissue, suggesting that migration of LacZ+ CG4 cells is not triggered only by a CNS milieu unique to md rats. When metaphase chromosomes were examined in untransfected CG4 cells at passages just preceding and following those used in transplantation, the modal number of chromosomes was 40 instead of the 42 expected for the rat, probably reflecting a chromosome rearrangement that may occur in rat cell lines (ref. 34 and C. Szpirer, personal communication; J. M. Zhou and M.D.-D., unpublished data). Yet, after lacZ transfection and selection, 59% hyperdiploid cells were observed at passage 50. This may explain the decreased myelinating potential of these transfected cells. As CG4 cells are bipotent in vitro (22), there is also the possibility that some of the grafted LacZ+ cells became astrocytes instead of oligodendrocytes (I.D.D., unpublished data). Alternately, CG4 cells overexpressing 3-galactosidase may stay longer at the progenitor stage and myelinate more slowly in spite of their excellent migratory potential. These questions can be addressed if LacZ+ CG4 clones selected at earlier passages are

11620

Neurobiology: Tontsch et al.

transplanted into longer-lived dysmyelinating rats. Nevertheless, our results suggest that gene transfer experiments could be performed in stable CG4 cells (up to passages 30-40) with this transfection/transplantation protocol. One could, for instance, construct antisense (30) or gene-targetting vectors to determine which genes are essential for the migratory and myelinating properties of OPs, enhancing the understanding of oligodendrocyte development in vivo. Our study also opens new horizons on the use of transplantation of OP cell lines to correct a defect or a lack of myelin. The recent discovery that not only a variety of point mutations but also deletions and duplications of myelin genes can cause dysmyelination in animals and humans (35, 36), has raised the interest in transplantation of myelin-forming cells to compensate, at least in part, for these defects. Such approaches would be particularly indicated in the case of dominant negative mutations. Our data demonstrate extensive myelination in rats affected with a dysmyelinating disease, although repair was restricted to one region of the spinal cord. Similarly CG4 cells (at passage 24) transplanted into dysmyelinating shiverer mouse brain produce normal myelin with a major dense line (P. Kuhn and R. McKinnon, personal communication). Transplantation of primary glial cells into md rats has recently been shown to enhance action potential conduction along the tracts myelinated by the grafted cells in a segment of the spinal cord (33). With CG4 cells, one could attempt to transfer genes that may enhance OP migratory and myelinating properties after transplantation into dysmyelinating animals. Alternately, one could establish cell lines derived from affected animals and attempt to correct their genetic defects in vitro before testing the myelination properties of such modified cells after transplantation. Finally, CG4 cells would be particularly appropriate to graft in an accessible site at some distance from a focal demyelinating lesion (37). In this respect, the ability of CG4 cells to migrate in normal developing CNS, even when gently deposited under the dura, is important and reminiscent of the behavior of immortalized Schwann cells (38, 39). Although human oligodendrocytes and preoligodendrocytes can be purified and cultured from adult human white matter derived from fresh surgery specimens, it has not been possible so far to expand these populations with growth factors (40, 41). However, human OPs cultured from fetal tissue showed mitogenic potential (42). The observation that rat OPs can be simply expanded with growth factors (20, 22, 23) and passaged 30-40 times without losing their migrating and myelinating properties after transplantation (this study) suggests that OP cell lines derived from the developing human nervous system could be generated and might display similar properties. Clearly the chromosomal stability of such cell lines will have to be carefully monitored, especially after gene transfer and selection. Such a cell line would be an invaluable tool to study the development and myelinating potential of human oligodendrocytes. U.T. and D.R.A. contributed equally to this work. We thank R.

Rusten, R. Hoffman, and D. Rhode for excellent technical and photographic assistance. Animal care was in accordance with National Institutes of Health guidelines. We thank Drs. L. D. Hudson, A. Messing, E. Milward, A. Warrington, and H. Arnheiter for their critical reading of the manuscript. U.T. was supported by a fellowship of the National Multiple Sclerosis Society, and I.D.D. was supported by the Myelin Project and National Institutes of Health Grant NS 23124. 1. Cattaneo, E. & McKay, R. (1991) Trends Neurosci. 14,338-340. 2. Snyder, E. Y., Deitcher, D. L., Walsh, C., Arnold-Aldea, S., Hartwieg, E. A. & Cepko, C. L. (1992) Cell 68, 33-51. 3. Bjorklund, A. (1993) Nature (London) 362, 414-415. 4. Richardson, W. D., Raff, M. & Noble, M. (1990) Semin. Neurosci. 2, 445-454. 5. McKinnon, R., Matsui, T., Dubois-Dalcq, M. & Aaronson, S. (1990) Neuron 5, 603-614.

Proc. Natl. Acad. Sci. USA 91

(1994)

6. Wolswbk, G. & Noble, M. (1992) J. Cell Biol. 118, 889-900. 7. Dubois-Dalcq, M. & Armstrong, R. (1992) in Myelin: Biology and Chemistry, ed. Martenson, R. E. (CRC, Boca Raton, FL), pp. 81-122. 8. Ritchie, J. M. (1984) in Myelin, ed. Morell, P. (Plenum, New York), pp. 117-145. 9. Dubois-Dalcq, M. & Armstrong, R. (1990) BioEssays 12, 569576. 10. Hudson, L. & Nadon, N. (1992) in Myelin: Biology and Chemistry, ed. Martensson, R. E. (CRC, Boca Raton, FL), pp. 677-702. 11. Aubourg, P. (1993) Nat. Genet. 5, 105-106. 12. Lachapelle, F., Gumpel, M., Baulac, M., Jacque, C., Duc, P. & Baumann, N. (1983) Dev. Neurosci. 6, 325-334. 13. Gansmuller, A., Lachapelle, F., Baron-Van Evercooren, A., Hauw, J. J., Bauman, N. & Gumpel, M. (1986) Dev. Neurosci. 8, 197-207. 14. Duncan, I. D., Hammang, J. P., Jackson, K. F., Wood, P. M., Bunge, R. P. & Langford, L. (1988) J. Neurocytol. 17, 351-360. 15. Gumpel, M. (1991) in Methods of Neurosciences, ed. Conn, P. M. (Academic, New York), Vol. 3, pp. 378-394. 16. Duncan, I. D., Archer, D. R. & Hammang, J. P. (1993) Soc. Neurosci. Abstr. 19, 689. 17. Blakemore, W. F. & Crang, A. J. (1988) Dev. Neurosci. 10, 1-11. 18. Blakemore, W. F. & Franklin, R. J. M. (1991) Trends Neurosci. 14, 323-327. 19. Warrington, A. E., Barbarese, E. & Pfeiffer, S. E. (1992) J. Neurosci. Res. 34, 1-13. 20. Groves, A. K., Barnett, S. C., Franklin, R. J. M., Crang, A. J., Mayer, M., Blakemore, W. F. & Noble, M. (1993) Nature (London) 362, 453-455. 21. Trotter, J., Crang, A. J., Schachner, M. & Blakemore, W. F. (1993) Glia 9, 25-40. 22. Louis, J. C., Magal, E., Muir, D., Manthorpe, M. & Varon, S. (1992) J. Neurosci. Res. 31, 193-204. 23. Louis, J. C., Magal, E., Takayama, S. & Varon, S. (1993) Science 259, 689-692. 24. Duncan, I. D., Hammang, J. P. & Trapp, B. D. (1987) Proc. Natl. Acad. Sci. USA 84, 6287-6291. 25. Duncan, I. D. (1994) in Neuroglial Cells, eds. Ransom, B. R. & Kettenmann, H. (Oxford Univ. Press, Oxford), in press. 26. MacGregor, G. R., Mogg, A. E., Burke, J. F. & Caskey, C. T. (1987) Somat. Cell Mol. Genet. 13, 253-265. 27. Price, J. & Thurlow, L. (1988) Development (Cambridge, U.K.) 104, 473-482. 28. Williams, B. P., Read, J. & Price, J. (1991) Neuron 7, 685-693. 29. Levison, S. W. & Goldman, J. E. (1993) Neuron 10, 201-212. 30. Galileo, D. S., Majors, J., Horwitz, A. F. & Sanes, J. R. (1992) Neuron 9, 1117-1131. 31. Shimohama, S., Rosenberg, M. B., Fagan, A. M., Wolff, J. A., Short, M. P., Breakfield, X. O., Friedmann, T. & Gage, F. H. (1989) Mol. Brain Res. 5, 271-278. 32. Feltri, M. L., Scherer, S. S., Wrabetz, L., Kamholz, J. & Shy, M. E. (1992) Proc. Natl. Acad. Sci. USA 89, 8827-8831. 33. Utzschneider, D. A., Archer, D. R., Kocsis, J. D., Waxman, S. G. & Duncan, I. D. (1994) Proc. Natl. Acad. Sci. USA 91, 53-57. 34. Levan, G., Szpirer, J., Szpirer, C., Klinga, K., Hanson, C. & Quamrul Islam, M. (1991) Genomics 10, 699-718. 35. Raskin, W. H., Williams, C. A., Hudson, L. H. & Bird, T. H. (1991) Am. J. Hum. Genet. 49, 1355-1360. 36. Ellis, D. & Malcom, S. (1994) Nat. Genet. 6, 333-334. 37. Gout, 0. & Dubois-Dalcq, M. (1993) Int. J. Dev. Neurosci. 11, 613-623. 38. Boutry, J.-M., Hauw, J.-J., Gansmuller, A., Di-Bert, N., Pouchelet, M. & Baron-Van Evercooren, A. (1992) J. Neurosci. Res. 32, 15-26. 39. Baron-Van Evercooren, A., Duhamel-Clerin, E., Boutry, J. M., Hauw, J. J. & Gumpel, M. (1993)J. Neurosci. Res. 35, 428-438. 40. Armstrong, R. C., Dora, H. H., Kufta, C. V., Friedman, E. & Dubois-Dalcq, M. (1992) J. Neurosci. 12, 1538-1547. 41. Gogate, N., Verma, L., Zhou, J. M., Milward, E., Rusten, R., O'Connor, M., Kufta, C., Kim, J., Hudson, L. & DuboisDalcq, M. (1994) J. Neurosci. 14, 4571-4587. 42. Satoh, J. & Kim, S. U. (1994) J. Neurosci. Res., in press.