Journal of Neuroscience Research 30:601-615 (1991)

Differentiation and Heterogeneity in T-Antigen Immortalized Precursor Cell Lines From Mouse Cerebellum C. Redies, U. Lendahl, and R.D.G. McKay Department of Brain and Cognitive Sciences, Massachusetts Institute of Technology, Cambridge

Recently, various techniques have been developed to transfer oncogenes into brain cells in order to generate immortalized neural cell lines. It is of interest to establish how well such cell lines reflect their cellular origin. Here we report the characterization of sixteen cell lines from mouse cerebellum and, as a control, six cell lines from skin. Lines were established by immortalizing postnatal primary cell cultures with a retrovirus carrying a modified temperature-sensitive variant of SV40 large T antigen. The cell lines reflect many properties of the cell type from which they were derived. All of the sixteen cerebellar lines expressed one or more markers of the neural precursor cells, namely, nestin and epitopes for NG2 and A2B5. In contrast, none of the six skin lines expressed neural precursor markers. Both types of cell lines expressed vimentin and fibronectin. Differentiation occurred in some of the cerebellar lines and was enhanced in defined medium. A small percentage of cerebellar cells, usually less than 5%, was positive for a marker of differentiation, e.g., glial fibrillary acidic protein (GFAP), galactocerebroside (GalC), or L1. Expression of GFAP colocalized with that of nestin at varying levels of intensity, indicating a gradual replacement of nestin by GFAP in the cytoskeleton. Both the cells positive for precursor markers and those positive for differentiation markers tended to be located in clusters, suggesting that stochastic processes or cell-cell interactions are important for the determination of the fate of cells within a clonal cell line in vitro. The degree of differentiation seemed to correlate with a shift from serum-containing to defined medium, but not with a shift from the permissive to the nonpermissive temperature for T antigen expression. The immortalization approach described here thus allows the establishment of cell lines which are “captured” in the precursor state of the developing mouse neuroepithelium

.

Key words: brain development, 02A progenitor, temperature-sensitive oncogene, intermediate filament, cell lineage 0 1991 Wiley-Liss, Inc.

INTRODUCTION During the development of the nervous system, multipotent neuroepithelial stem cells transiently and rapidly divide, and the cell progeny differentiates into neurons and glia. To get better experimental access to different intermediate cell stages and differentiation events, immortalized neural cell lines have been generated by transducing oncogenes into the primary neuroepithelial cells (Giotta et al., 1980; Moura Net0 et al., 1986; Bartlett et al., 1988; Evrard et al., 1986, 1988, 1990; Frederiksen et al., 1988; Bernard et al., 1989; Trotter et al., 1989; Birren and Anderson, 1990; Galiana et al., 1990; Ryder et al., 1990) or by targeting the expression of oncogenes to different brain regions in transgenic mice (Hammang et al., 1990; Mellon et al., 1990). The resulting clonal cell lines are proposed to be useful in the investigation of many questions in neurobiology , for example in the study of cell lineage, neural growth factors, neural differentiation, and transplantation biology (reviewed in Cepko, 1989; Lendahl and McKay, 1990). In this study, a recently produced hybrid SV40 T temperature-sensitive oncogene (SV40T U 19/tsA58) (Almazan and McKay, 1988) was used to immortalize cells from neuroepithelial and nonneural control tissues by retroviral transfer into dividing cells. Some of the central premises on which this technology is based were evaluated. The following questions were asked: (1) How many of the characteristics of the primary (precursor) cell population are maintained upon immortalization? (2) What is the degree of heterogeneity that arises within a clonal cell line? (3) How different from one another are cell lines obtained from the same population of primary

Received September 12, 1990; revised March 22, 1991; accepted May 9 , 1991. Address reprint requests to C . Redies, presently at Department of Biochemistry, Max Planck-Institute for Developmental Biology, Spemannstrasse 35/11, D-7400 Tubingen, Germany. U. Lendahl is presently at the Department of Molecular Genetics, Karolinska Institute, S-104 01 Stockholm, Sweden.

602

Redies et al.

cells? (4)To what extent can the cell lines differentiate at both the permissive and nonpermissive temperatures for the oncogene? ( 5 ) How do cell lines immortalized with the modified SV40 T antigen vector compare to those immortalized with other agents, namely, wild-type SV40 T (Moura Net0 et al., 1986; Mellon et al., 1990; Evrard et al., 1990; Hammang et al., 1990), SV40 T tsA58 (Frederiksen et al., 1988), E1A (Evrard et al., 1986), cand N-myc (Bartlett et al., 1988; Bernard et al., 1989; Birren and Anderson, 1990; Ryder et al., 1990), Rous sarcoma virus (Giotta et al., 1980), v-src (Trotter et al., 1989), and polyoma T antigen (Evrard et al., 1988; Galiana et al., I990)? To address these questions, cell lines derived from postnatal day 2 (P2) mouse cerebellum were compared to each other and to cell lines derived in the same way from a control tissue: postnatal mouse skin. Cultures were grown past the confluence stage for 3 weeks to promote the development of cell-cell interactions and differentiation. Immunocytochemistry with antibodies against neural precursor and differentiation markers was used to study the differences between the cell lines. The marker profiles of the neural cell lines were related to known, immunocytochemically defined, cerebellar cell lineages. The pattern of differentiation and degree of heterogeneity within the clonal cell lines were visualized in vitro both at the permissive and at the nonpermissive temperatures for T-antigen expression.

pg/ml, Sigma) in Dulbecco's modified Eagle's culture medium (DMEM). Cells were grown at 39°C in DMEM supplemented with 10% (v/v) horse serum (donor herd, Sigma), penicillin, and streptomycin (Gibco). The same substrate and medium were also used for passaging the cell lines (see below).

Immortalization Procedure Twenty-four hours after plating, primary cultures were shifted to 33°C and subsequently infected with retrovirus as described previously (Frederiksen et al., 1988), by replacing the culture medium with cell-free supernatant from transfected 'P2cells (Cone and Mulligan, 1984) and adding 8 pg/ml polybrene (Aldrich). The retrovirus construct was based on the pZipneo vector (Cepko et al., 1984), into which a hybrid SV40 T oncogene was inserted, made by assembling the immortalizing region from the more efficiently transforming mutant U19 (Paucha et al., 1986) and the temperature-sensitive region from SV40TtsA58 (Jat and Sharp, 1986, 1989). The temperature-sensitive mutation results in instability of the protein at elevated temperatures. The hybrid SV40 T construct was kindly provided by Dr. G. Almazan (Almazan and McKay, 1988). Two hours after infection, the retroviral supernatant was replaced by fresh medium and cells were subsequently grown and passaged at 33°C. One day after infection, 0.5 mg/ml G418 (Geneticin, Gibco) was added to the cultures to select for neomycin-resistant colonies. The selection medium was changed every 3-5 days. MATERIALS AND METHODS Cell lines from skin were derived in a similar way. Animals Small pieces of skin from the transgenic P2 mice were Primary cultures of mouse cerebellum from transexcised, minced with scissors, trypsinized, and plated genic mice carrying copies of the human neurofilament under the same conditions as the cerebellar cells. Re(NF-L) gene (Julien et al., 1987) were made. These mice maining clumps of cultured skin were trypsinized again were provided by Drs' J'p' Julien and A' Peter- two to three times before infection Over 10 days to ensure son. The use of the NF-L transgenic mice allows for the that enough cells had grown onto the bottom of the dish. human neurofilament to indicate potential neuronal difThree to four weeks after infection, macroscopic ferentiation of the cells and to mark them. Southern blot colonies of cells had formed in the infected cultures from analysis revealed that all mice used in this study carried both cerebellum and skin, but not in noninfected control the transgene (data not shown). cultures. Colonies were picked with cloning rings and expanded over 2-6 weeks into gradually larger dishes Preparation of Primary Cerebellar Cultures until about 5 X lo7 cells were available for freezing. All Primary cerebellar cultures were made from one cell lines were passaged in the horse-serum-containing litter of nine pups killed at postnatal day 2 (P2). Ceremedium described above. To decrease the likelihood of bella were dissected into ice-cold, Ca/Mg-free Hank's of the lines, cultures for immunefurther balanced salt solution and stripped of most meninges. staining were made from earlier passages (1-3 weeks of After a 20-min incubation at 37°C in 0.15% trypsin dipassaging). luted in Eagle's basal medium, cells were dissociated into a single-cell suspension by gentle tituration with a 1-ml pipetteman. About 1 X lo6 dissociated cells were Southern plated onto each of several 35-mm plastic cell culture High molecular weight genomic DNA was predishes, which had been sequentially coated with poly- pared from each cell line according to standard proceD-ornithine in water (15 pg/ml, Sigma) and laminin (10 dures (Maniatis et al., 1982). After complete cleavage of

Mouse Cerebellar Precursor Cell Lines

603

10 p,g DNA with restriction enzyme (BamHI, BglII, TABLE I. Scoring System for Immunocstochemical Staining EcoRI, and XbaI), the DNA was size-fractioned in 0.8% Symbol Meaning Percent Positive cells agarose gels and transferred to nylon filter (Gene Screen All cells negative 0% positive Plus, Dupont) (Southern, 1975). The filterbound DNA + Cells rarely positive 95% positive ber of integration sites, and the clonal relationship between the cell lines to be determined (Frederiksen et al. , Cultures and Immunostaining 1988). Cells were plated on 12-mm diameter glass coverslips coated with poly-D-ornithine and laminin as deAntibodies scribed above. The medium was the same as the passagThe following antibodies were used for immuno- ing medium. Cultures to be stained with anti-fibronectin staining: anti-SV40 T antigen monoclonal antibody (pool antibody were grown in the same medium but the horse no. 419; gift of Dr. E. Harlow); rabbit anti-fibronectin serum was passed over a gelatin sepharose column (Pharantiserum (dilution 1:1,000; gift of Dr. R . Hynes); goat macia LKB, cat. no. 17-0956-01) to remove exogenous anti-vimentin antiserum (1 :40; ICN ImmunoBiologicals, fibronectin from the medium. To increase cell-cell intercat. no. 64-740- 1); anti-A2B5 monoclonal antibody (hy- actions, cells from each cell line were plated to an initial bridoma supernatant; American Type Culture Collection, density of 100-180% confluence, and kept at 33°C for cat. no. CRL 1520); rabbit anti-NG2 antiserum (1:200; 3-4 weeks. Parallel cultures were first grown at 33°C for gift of Dr. W. Stallcup); anti-glial fibrillary acidic pro- 2-5 days and then at 39°C for 3-4 weeks. Medium was tein (GFAP) monoclonal antibody (1:250; ICN Immuno- replenished every 2-4 days. Biologicals, cat. no. 69-1 10); anti-neurofilament monoAfter rinsing coverslips with PBS, cells were fixed clonal antibody, phosphorylated epitope of heavy chain for 30 min with freshly prepared 4% paraformaldehyde (1:200; Sternberger-Meyer Immunocytochemicals, cat. in sodium borate buffer (pH 9.5) (neurofilament, NG2, no. SM131); anti-neurofilament monoclonal antibody, and human NF-L) or sodium phosphate buffer (pH 7.2) unphosphorylated epitope of heavy chain (1:200; Stern- (nestin, fibronectin, GFAP, and troponin T). For vimenberger-Meyer Immunocytochemicals, cat. no. SMI32); tin staining, cells were fixed for 2 min in an ethanol/ anti-neurofilament monoclonal antibody, heavy chain (1: acetic acid solution (95%/5%) (v/v), and, for T antigen 160; clone NE14; ICN ImmunoBiologicals, cat. no. 69- staining, for 5-10 min in a methanol/acetone solution 705); anti-human light chain neurofilament monoclonal (70%/30%) (v/v). Staining with antibodies against L1, antibody (1:lOO; obtained from Dr. D. Paulin); rabbit A2B5, 0 4 , and galactocerebroside (GalC) was carried anti-L1 antiserum (1:25; gift of Dr. M. Schachner); anti- out at 4°C for 30 min on live cultures before fixation. 0 4 monoclonal antibody (1:25; gift of Dr. M. Each cell line was stained with all antibodies, except Schachner); anti-galactocerebroside monoclonal anti- against S-100 and myelin basic protein, after growth at body (hybridoma supernatant; gift of Dr. B. Ranscht); both 33°C and 39°C. Double-staining for A2B5 and NG2 anti-troponin T monoclonal antibody (1 :100; clone JTL- was performed by first staining live with anti-A2B5 an12; Sigma Chemical, cat. no. T-6277); rabbit anti-my- tibody and then, after fixation with 4% paraformaldeelin basic protein antiserum (1: 100; Dako Japan, cat. no. hyde in the borate buffer, with anti-NG2 antibody. After A623); and rabbit anti-S-lOOa,b antiserum (1 :200; Dako incubation with the appropriate secondary fluorescent Japan, cat. no. 2311). Two rabbit anti-nestin antisera antibodies, coverslips were mounted and viewed under a were generated in our laboratory (M. Marvin and R . fluorescent microscope. Appropriate controls were used. McKay, unpublished results) against a synthetic 20 The percentage of positive cells for each condition was amino acid peptide from the extreme carbox y terminus estimated using the scoring system given in Table I. The effect of defined medium on the differentiation and against the entire carboxy terminus of the intermediate filament protein nestin, formerly called Rat.401 of ten cerebellar and four skin-derived cell lines was also (Lendahl et al., 1990). These antisera were used at a assessed. The cells were grown in serum-containing medilution of 1: 1,000 and 1:500, respectively. Appropriate dium at 33°C for 3-5 weeks, then shifted to the defined fluorescein-conjugated and rhodamine-conjugated sec- N2 medium by Bottenstein and Sat0 (1979). Insulin and ondary antibodies were purchased from Cappel and ICN putrescein were omitted from the N2 medium since preImmunoBiologicals. All antibodies were tested on ap- liminary experiments showed that they were not required for the in vitro survival of the cell lines. The cells were propriate, positive mouse primary cultures.

604

Redies et al.

cultured in defined medium for 5-10 days at 33"C, then fixed and stained with antibodies against SM13I , GFAP, NG2, and L1 as described above. Using the same culture protocol, four cell lines that showed signs of differentiation (H35, H91, H94, and H98), were shifted to the complete N2 medium, also including insulin (5 pg/ml) and putresceine (16 pg/ml) (Bottenstein and Sato, 1979). Parallel cultures were shifted to the modified N2 medium described above. These cultures were single- and double-stained with antibodies against GFAP, myelin basic protein, S- 100, GalC, and nestin, to more closely study the types of differentiated cells present after shifting to the defined media. Double-staining for nestin and GFAP, and for S-100 and GFAP, was performed on fixed cells by incubation with the rabbit antisera, and then, after postfixation with ethanol/acetic acid solution (95%/5%) (viv), with GFAP antibody. Appropriate controls were used.

RESULTS Sixteen cell lines were immortalized from about 3 lo7 primary cerebellar cells. Six cell lines were obtained from skin. These cell lines have grown continuously at 33°C under passaging conditions for three months and longer without showing signs of crisis or decreasing proliferation rates. No cell lines were obtained from noninfected control cultures. X

Immortalization and T Antigen Expression Table I1 shows that 12 of the 16 cerebellar lines had a single integration of the retroviral construct including the hybrid SV40 T oncogene. The remaining cerebellar lines had two or three integrations. The skin-derived lines had up to seven integrations. Southern blot analysis of cell line DNA revealed that none of the lines was clonally related to another, with the exception of lines H36 and H37, which were sibling lines (Fig. 1). In all of the cerebellar lines with two or more integrations, the DNA restriction fragments corresponding to each of the integrations hybridized to the SV40 T antigen probe with similar intensity (Fig. l), even after many passages in culture. However, in the skin cell lines SKI, SK3, and SK8, the hybridization bands were of slightly different intensity, raising the possibility that these lines may be mixtures of different clones. All lines except H93 contained full-length copies of the retroviral construct. Line H93 contained a slightly shorter fragment (Fig. l), but with a full-length SV40 T antigen gene (data not shown). At the permissive temperature for T antigen expression (33"C), immunostaining with the T antigen antibodies revealed the characteristic nuclear T antigen localization in all cell lines (Table 11). Typically, cells in a given line were heterogeneous with respect to the intensity of

TABLE 11. Number of Integrations and T-Antigen Expression in Cell Lines Number of integrations

H12 H3 1 H32 H34 H35 H36a H37a H8 1 H83 H9 1 H92 HV3 HV4 H95 H96 HV8 SKlb SK2 SK3b SK4 SK7 SK8b

1 1 1 1 3 1 1 2 3 1 2 1 1 1 1 '1

3 1

3 6-7 2 4

T-Antigen Expression

33°C

++++ ++++ +++++ +++++ ++++ ++++ ++++ ++++ +++++ +++++ +++++ +++++ +++++ +++++ +++++ +++ ++++ +++++ +++++ +++++ +++++ +++++

39°C

-

++ ++ ++ +++++ ++ ++ .-

+ -t+ .-

.-

++t

See Table I for explanation of symbols. 3 a m e clone. bPossibly not clonal.

nuclear staining; some cells were even negative (Fig. 2A). During or immediately after mitosis, cells were usually intensely positive. Cells in multilayered regions were generally more intensely stained than cells in monolayer regions of a given culture. At the nonpermissive temperature (39"C), nuclear T antigen staining was totally absent in 14 lines (Table 11). In eight of the remaining lines, faintly positive cells were found but at a low frequency. In line H83, there was no difference between the two temperatures (Table 11).

Growth Characteristics and Morphology at 33°C and at 39°C In serum-containing medium at 33"C, cells continued to proliferate on the glass coverslips used for immunostaining. Cells were flat and polygonal, and in most cases extended processes that were three to five cell diameters in length and ended in growth cones (Fig. 3 ) . Typically, cells tended to clump spontaneously in cultures where cells had grown to confluence, temporarily leaving empty spaces in between aggregates. These empty spaces were then overgrown again from the edges of the aggregates. After 3 weeks, cultures were usually densely confluent with many multilayered regions (see, for example, Figs. 2A,C, and 4). The morphologies of

Mouse Cerebellar Precursor Cell Lines

Xbal

605

Eco RI

412.2

4

12.2

4

6.1

+6.1

4

3.1

-3.1

Fig. 1. Results from the Southern analysis using 32P-labeIled SV40 T antigen probe. The left and right panels show genomic DNA from ten cerebellar lines, cleaved with the restriction enzymes XbaI and EcoRI, respectively. Cell lines are indicated above each lane. Fragment size (in kilobases) is indicated on the right of each panel. The enzyme XbaI cuts at both ends of the retroviral insert. Note that each line, except line H93, has a full length 6.6 kb insert of the retroviral construct. The enzyme EcoRI cuts once inside the construct. Note that the frag-

individual lines differed from each other. Some lines had a larger diameter or had longer processes than other lines (Fig. 3). After a change to serum-free N2 medium, cells grown at subconfluence became phase-bright with thin processes (data not shown). For experiments at the nonpennissive temperature, cells were initially grown to confluence at 33°C and then shifted to 39°C. The cells often clumped and had fewer processes compared to 33°C. After 3 weeks at 39"C, the cultures contained sparsely seeded cells with few cellcell contacts. Individual cells had a larger diameter at 39°C than at 33°C. Only line H83, which retained SV40 T antigen expression, continued to grow rapidly at 39°C. No systematic differences in growth characteristics were observed between cerebellar and skin-derived cell lines.

ments are of different lengths indicating that the cell lines are clonally different, each having its unique integration site, except for lines H36 and H37. Lines H92 and H83 have two and three integrations, respectively, but the corresponding restriction fragments hybridize with the probe at similar intensity in both lines, suggesting multiple integrations of the retroviral construct in the same cell rather than nonclonality of the lines. These results were confirmed by the enzyme, BglII, which also cleaved once, but at a different site, within the insert.

variation in the frequency of positive cells between the lines (from no cells positive to most cells positive). There were also differences between 33°C and 39"C, but these differences were not systematic. In all cultures, cells positive for A2B5 typically occurred in clusters (Fig. 5A) of a few cells to several hundred cells. Cells were either loosely clustered or formed tight groups of cells (Fig. 5A). Only in one line, H37, were positive and negative cells apparently mixed. A2B5-positive cells were usually flat or spindle shaped with few or no long processes. NG2 staining assumed a speckled appearance; it typically covered most but not all of the cell surface (Stallcup et al., 1983). NG2-positive cells also occurred in clusters (Figs. 2C and 5B) but these clusters were often less obvious than the A2B5 clusters. Both A2B5Immunostaining of Cerebellar Lines and NG2-positive cells had similar morphologies. DouPrecursor markers (A2B5, NG2, and nestin). ble-label immunocytochemistry for NG2 and A2B5 was All of the cerebellar lines were positive for at least one of performed in the H3 1 and H96 lines. Half of the positive the precursor markers A2B5 (Eisenbarth et al., 1979; cells were positive for both markers (Fig. 5), whereas the Raff, 1989), NG2 (Stallcup et al., 1983; Levine, 1989), rest were positive for only one of the two markers. Cells positive for nestin were apparently not clusand nestin (Frederiksen and McKay, 1988; Lendahl et tered. In all positive lines, the staining was perinuclear, al., 1990) at both 33°C and at 39°C (Table 111). Table IV as expected for an intermediate filament (Fig. 6D). shows that most lines were positive for more than one precursor marker. For all markers, there was a large Dense nestin fiber bundles were also observed but only

606

Redies et al.

Fig. 2. Cerebellar cell line cultures maintained for three weeks before immunostaining (left) and corresponding phase contrast images (right). The medium was serum-containing in all cases. (A) Anti-SV40 T antigen staining of line H36 at 33°C. Note the heterogeneity of the nuclear T antigen staining in

different cells. (B) Anti-fibronectin staining of line H81 at 33°C reveals an extensive fibrillar matrix of fibronectin. (C) Anti-NG2 staining of line H96 at 33°C. The staining is punctuate and covers the cell surface. Positive cells form a cluster surrounded by unstained cells. The scaling bar is 40 p m .

where cells had clumped and formed more than one cell layer, or in the processes of the phase-bright cells found in defined medium (see below) (Fig. 6A,D). Markers of differentiation. A number of markers were used to identify differentiated neuronal and glial phenotypes within the cultures (Table 111). On all cell lines, we used markers for early premigratory neurons (Ll) (Rathjen and Schachner, 1984); fully differentiated neurons (neurofilaments) (Sternberger and Stern-

berger, 1983; Debus et al., 1983; Julien et al., 1987); astrocytes (GFAP) (Debus et al., 1983); and oligodendocytes (GalC, 0 4 ) (Ranscht et al., 1982; Sommer and Schachner, 1981). Moreover, to define the cell types found in defined medium more closely, other markers for glial cells (S-100) (Moore, 1965) and for myelin-producing cells (myelin basic protein) were used on some lines. In addition, a muscle marker (troponin T) (Squire, 1981) was used since a previously reported SV40 T an-

Mouse Cerebellar Precursor Cell Lines

607

Fig. 3. Morphology of immortalized cell lines under passaging conditions at 33°C in serumcontaining medium (phase-contrastimages). (A) Cerebellar line H95. (B) Cerebellar line H32. (C) Cerebellar line H81. (D) Skin-derived line SK1. The scaling bar is 100 pm. tigen immortalized cell line from rat cerebellum (STI 5A) (Frederiksen et al., 1988) can differentiate into muscle (Valtz, Liu, Hayes, and McKay; unpublished observation). Positive immunoreactivity was observed with three differentiation markers: GFAP, GalC, and L1, and only in a few of the cell lines. The frequency of positive cells was generally less than 5% (Table IV). GFAP-positive cells had the same filamentous staining pattern as cells seen in primary cultures; the staining was strong in defined medium but relatively faint in serum-containing cultures. The cells were found in groups of a few cells to hundreds of cells (Fig. 6B,E). GalC-positive cells occurred in clusters of up to a dozen cells. In some cases, cells were flat with few processes, in which case their entire surface was homogeneously stained (Fig. 4C); in other cases, the staining was more speckled. At 39”C, “ghosts” of cells without nuclei were sometimes positive. None of the lines was positive for 0 4 (Table 111). Cells staining with antiserum against L1 were seen only at 33°C. The cells were found in clusters ranging from a few cells to several hundred cells. In serumcontaining medium, the cells were flat and appeared to have extended relatively long branched processes (Fig.

4A). None of the lines was positive for mouse or human neurofilament (Table 111). Defined medium. After an additional 5-10 days at 33°C in the two defined media (with and without insulin and putresceine, respectively), lines H35 and H91 showed an increase in L1 expression. There were clusters of dozens to hundreds of bipolar or multipolar, phase-bright, L1-positive cells that extended fine processes (Fig. 4B). All cells showing this characteristic morphology were L1-positive. Typically, they were growing on top of flat cells in the vicinity of cell clumps. Flat cells positive for L1, like those present in serumcontaining medium, were no longer present in defined medium. In lines H35 and H9 1, large clusters of GFAP positive flat cells were also seen in the two defined media (Fig. 6B,E). Like the phase-bright cells, GFAP-positive cells were associated with cell clumps. However, the GFAP-positive cells were present mostly in the immediate vicinity and within the clumps whereas the phasebright cells were found over a larger area surrounding the clumps (Fig. 6A-C). Double-staining for nestin and GFAP (Fig. 6) revealed that the phase-bright cells were strongly positive

608

Redies et al.

Fig. 4. Cerebellar cell lines cultures maintained for three weeks before immunostaining (left) and corresponding phase contrast images (right). Medium was serum-containing in all cases except for panel B . (A) Anti-L1 staining of line H98 at 33°C. There are four stained cells (arrows) surrounded by unstained cells. The L1-positive cells are flat but have highly branched processes. The cells in this cluster were the only L1-positive cells in this culture. (B) Anti-Ll staining of line

H35 at 33°C in serum-free medium (see Materials and Methods). The phase-bright L1 -positive cells form a network of thin processes on top of other cells (different plane of focus). The cells were part of a cluster comprising several hundreds of similar cells. (C) Anti-galactocerebroside staining of line H98 at 33°C. This cluster of positive cells was one of very few clusters in this culture. The scaling bar is 40 p,m.

for nestin, with prominent staining up into the tips of the processes. Most of these cells were GFAP-negative. On the other hand, GFAP-positive cells showed weak or moderate levels of staining for nestin. Regardless of the level of expression, the nestin staining always strictly colocalized with the GFAP staining (Fig. 6D-F). Tran-

sitions between the phase-bright and the GFAP-positive cell types were frequent. Almost all cells in the four cell lines tested (H35, H91, H94, H98) expressed varying amounts of S-100, mostly confined to their nuclei (Fig. 5D). Doublestaining for S-100 and GFAP showed that phase-bright

Mouse Cerebellar Precursor Cell Lines TABLE 111. Number of Cell Lines Positive Immunocytochemically (33" or 39OC) Antibody Vimentin Fibronectin Nestin A2B5 NG2 GFAPa Galac tocerebroside L1 Neurofilament' 04 Troponin T

Cerebellum-Derived (1 6 lines)

Skin-Derived (6 lines)

I6 16 14 13 9 2 5 5 -

6 6 -

-

2b

-

aGlial fibrillary acidic protein. bLess than 0.1% of cells positive. 'Against phosphorylated and unphosphorylated epitopes of mouse 200-kD neurofilament and human light chain neurofilament (see Materials and Methods).

cells were generally more strongly positive for S- 100 whereas GFAP-positive cells were sometimes only weakly positive (Fig. 5D-F). Of the four cell lines tested after an additional 5 days in defined medium at 33"C, none expressed GalC or myelin basic protein. The morphology and immunostaining in the cell lines was similar regardless of whether or not insulin and putrescein were present in the defined medium. All ten cerebellar cell lines tested remained negative for neurofilament, and no marked change in NG2 expression occurred in defined medium.

Immunostaining of Skin-Derived Cells The skin cell lines were tested with the same set of antibodies as the cerebellar cell lines. Table 111 shows that all cerebellar and skin-derived lines stained for vimentin and fibronectin at 33°C and at 39°C. Virtually all cells in all cultures were positive for vimentin. The fibronectin staining pattern of the skin-derived lines was similar to that of the cerebellar lines (Fig. 2B). None of the skin-derived cell lines stained for any of the other markers listed in Table 111, except lines SK2 and SK4, in which a very small percentage of cells stained with antiGalC antibody. DISCUSSION The developing nervous system proceeds through a number of transient stages of cellular differentiation, many of which are complicated to analyze. One way to facilitate the analysis is to immortalize cell lines from various stages and to study their properties, e.g., the potential for differentiation in vitro. The availability of

609

new methods to generate immortalized neural cell lines by oncogene transfer has therefore generated much enthusiasm in recent years and, as a result, a number of cell lines have been made (reviewed in Cepko, 1989; Lendahl and McKay, 1990). In this study, we have used a recently engineered retrovirus vector carrying a hybrid version of a tsSV40 T antigen oncogene (Almazan and McKay, 1988) to generate immortalized cell lines from developing mouse cerebellum and, as a control, from skin. The hybrid U19/ tsA58 oncogene vector worked efficiently in immortalizing cells from both tissues. As expected, most cell lines had a single integration site. In the few cases with more than one integration site, all integrations in the cerebellar lines appeared to be of equal strength on Southern blots, even after long periods in culture. Multiple integrations have also been observed in other clonal cell lines (Frederiksen et al., 1988).

Cerebellar Cell Lines Are Similar to Cerebellar 02A-like Precursors To analyze the extent to which the immortalized cerebellar and skin cell lines reflect their distinct origins, we used a number of antibody probes for markers expressed in these two cell types. All 16 cell lines derived from cerebellum expressed at least one of the neural precursor markers NG2 (Stallcup et al., 1983), A2B5 (Eisenbarth et a]., 1979), or nestin (Frederiksen and McKay, 1988). Most cell lines expressed more than one marker'. For example, at 33"C, seven out of eight NG2-positive cell lines were also positive for A2B5, and, when analyzed for coexpression, about half of the cells were positive for both markers. Coexpression of NG2 and A2B5 in the cerebellar lines makes them very similar to a proliferating 02A-like progenitor cell seen in cultures of postnatal cerebellum (Stallcup and Beasley, 1987; Levine, 1989). This progenitor, which gives rise to oligodendrocytes or type I1 astroglia, also coexpresses the NG2 and A2B5 antigens. At present, it is not known whether cerebellar 0 2 A progenitors express nestin protein, but the facts that nestin is expressed at high levels in cerebellum at P2 and that its expression is strongly correlated with the stem cell state in the central nervous system (Frederiksen and McKay , 1988; Lendahl et al., 1990) further support the similarity of the cell lines and cerebellar precursor cells. Vimentin expression is also a feature expected of cells derived from early postnatal Cerebellum (Bovolenta et al:, 1984). However, fibronectin expression by all cerebellar cell lines is puzzling. Fibronectin has been found in the germinal layer of developing cerebellum but its origin is unclear (Stallcup et al., 1989). Neurons and GFAP-positive cells in primary culture are usually fibronectin-neg-

610

Redies et al.

TABLE IV. Immunocytochemical Staining of 16 Cerebellum-Derived Cell Lines Precursor markers Nestin

H12 H3 1 H32 H34 H35 H36” H37a H8 1 H83 H91 H92 H93 H94 H95 H96 H98

Differentiation markers

NG2

A2B5

33°C

39°C

33°C

+++ ++++ +++ +++ ++++ ++ +++ +++ ++ +++ +++

++++ + ++++ +++ ++ +++

+ +++ ++ +++ +++

++ +++

++++ ++ +++ +++ ++ +++

++

39°C -

GalC

GFAP

33°C

39°C

++ ++ ++ +++ ++

++ ++

L1

-

33°C

39°C

33°C

39°C

33°C

-

-

-

-

-

39°C -

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

++++

+ +++ ++ +++ +++ +++

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

++

+++

-

-

-

+

-

-

-

-

-

-

-

+++ +++ ++ ++

-

-

+

+ +

+ +++ +++ ++

-

-

-

-

+++ + ++ -

+(++)b -

-(

+ -

-

+)b

-

+(++)b

-

+ -(++)b

-

-

-

++ +++

-

-

-

-

-

+

+

-

-

-

-

-

-

-

-

-

-

-

-

-

-

++

-

-

-

-

-

+

-

+

++

-

See Table I for explanation of symbols. aSame clone. bResult for defined medium in parentheses.

ative but some astroglial cell lines are fibronectin-positive or secrete fibronectin (reviewed in Hynes, 1989). In marked contrast to the cerebellar cell lines, none of the six cell lines from skin expressed NG2, A2B5, or nestin. As expected, they did express vimentin and fibronectin. It thus appears that immortalization with the modified SV40 T retrovirus vector results in cell lines that reflect many characteristics of their particular cellular origin. The exclusive expression of neural precursor markers in the cerebellar, but not in the skin-derived, cell lines opens up new possibilities for a molecular biological dissection of the cellular machinery resulting in the expression of the precursor-specific antigens.

Cerebellar Cell Lines Are “Captured” in the Precursor State We tested the cell lines for several markers that identify differentiated cell types in the nervous system (Tables 111 and IV). In the cerebellar cell lines, cells positive for three of these markers (Ll, GFAP, GalC) were found. In contrast, the skin cell lines were negative for L1 and GFAP. Surprisingly, two of the skin cell lines contained a few GalC positive cells (Table 111). The expression of L1, GFAP, or GalC in the cerebellar cell lines indicate that some of the cerebellar lines harbor progenitor cells capable of differentiating along astrocytic (GFAP), oligodendrocytic (GalC), or early neuronal pathways (L 1). The potential of astrocyte/oligodendrocyte differentiation is supported by the demonstration of 02A-like, A2B5/NG2 positive progenitors in the cell lines. However, the differentiation capacity ap-

pears to be limited since not all cerebellar cell lines were positive for either L1, GFAP, or GalC, and, in the lines that were positive, usually less than 5% of the cells stained (Table IV). The presence of L1-positive but neurofilament-negative cells, and the persistence of nestin expression in the GFAP-positive cells, indicate that only partial differentiation occurred in the clonal cultures. In vivo, L1 is expressed by early, premigratory, neurons in the external granular layer of the cerebellum (Rathjen and Schachner, 1984) whereas neurofilaments are markers for mature, fully differentiated, neurons (Sternberger and Sternberger, 1983; Debus et al., 1983). Our results are thus similar to results with the v-src oncogene which can block glial cell differentiation (Trotter et al., 1989).

Differentiation Occurs in a Heterogeneous, Clustered Fashion Cells positive for either the precursor markers (A2B5 and NG2) or for the differentiation markers (L1 , GFAP, or GalC), occurred in clusters, sometimes in the vicinity of or within cell clumps, and were surrounded by negative cells (Figs. 2, 4-6). One explanation for this observation is that the commitment to a particular precursor or differentiated cell type is a stochastic process, and that, once a decision is made, for example by a change in the DNA methylation pattern, a particular phenotype is propagated by cell division producing a cell cluster. An alternative explanation is that even a clonal cell culture contains microenvironments that locally support differentiation to a particular fate. In both cases, the cluster of differentiated cells could be stabilized by cell-

Mouse Cerebellar Precursor Cell Lines

611

Fig. 5. Double-label immunohistochemistry of cerebellar cell lines. A-C shows staining of cerebellar line H96 with antiA2B5 antibody (A) and anti-NG2 antiserum (B) and the corresponding phase-contrast image (C). This is a culture maintained for three weeks at 33°C in serum-containing medium. The cluster of A2B5/NG2 double-stained cells is surrounded by unstained cells. D-F shows staining of cerebellar line H91

with anti-S-100 antiserum (D) and with anti-GFAP antibody (E), and the corresponding phase contrast image (F). This is a culture maintained for 3 weeks in serum-containing medium and then shifted for 5 days to defined medium. Note that the phase-bright cells generally show more intense S-100 staining than the GFAP-positive cells. Both scaling bars are 40 k m .

cell interactions or secretion of factors that promote or maintain differentiation only over very short distances. Heterogeneity in clonal neural cell lines has been observed previously for other oncogenes (Ryder et al., 1990), and was seen in the progeny of single cultured neuroblast cells (Temple, 1989). The present immunostaining results are from cell lines passaged in culture for 6-13 weeks after retroviral infection, before culturing for three or more weeks under the various experimental conditions. It cannot be excluded that the heterogeneity in the lines becomes more pronounced after longer periods in culture. In defined medium, phase-bright cells positive for the markers L1, nestin, and S-100 emerged. The same markers are also expressed in the superficial layers of

postnatal mouse cerebellum (Rathjen and Schachner, 1984; Hockfield and McKay, 1985; Landry et al., 1989) suggesting that the phase-bright cells may represent an early neuronal cell type similar to that found in these layers. S-100 is a marker for glia in adult brain, but bipotential precursor cells in the developing mouse brain can coexpress S-100 and neuronal markers (DeVitry et al., 1980). It should be noted that L1, S-100, and nestin are also expressed in Schwann cells (Martini and Schachner, 1986; Scarpini et al., 1986; Friedman et al., 1990), which raises the possibility that we obtained Schwann cells derived from neural crest derivatives (meningeal cells) that may have contaminated the primary cerebellar cultures used for immortalization. However, in view of the fact that the phase-bright cells do not

612

Redies et al.

Fig. 6. Double-label immunohistochemistry of line H35 with anti-nestin antiserum (A, D) and with anti-GFAP antibody (B, E). The corresponding phase contrast images are shown (C, F). In A-C, the edge of a cell clump is located on the top left of the panels (arrow in C). Note the wider spread of the nestinpositive, phase-bright cells from the clump. In D-F, note the

colocalization of nestin and GFAP expression in cells that coexpress both markers. This is a culture maintained for 3 weeks at 33°C in serum-containing medium, and then shifted for 5 days to defined medium (including insulin and putresceine). The scaling bar is 100 ym in A-C and 20 ym in D-F.

express the Schwann cell markers GalC and myelin basic protein, this possibility seems rather unlikely. In defined medium, we also identified cells expressing high levels of the intermediate filament GFAP, which is a marker of the astroglial lineage in mammalian brain (Debus et al., 1983). Interestingly, there were var-

ious degrees of transition between GFAP expression and expression of the precursor marker nestin, which is also an intermediate filament (Lendahl et al., 1990). Moreover, the two intermediate filaments were strictly colocalized suggesting the possibility of a developmental transition in the cytoskeleton during which nestin is grad-

Mouse Cerebellar Precursor Cell Lines

ually replaced by GFAP as precursor cells differentiate into astroglia.

Inactivating the SV40 T Antigen: No Systematic Effect on Differentiation As expected, inactivating the tsSV4O T antigen by a shift to the nonpermissive temperature resulted in a considerable reduction in cell growth in the cell lines (Frederiksen et al., 1988; Jat and Sharp, 1989). The only exception to this was the cerebellar line H83, which continued to express SV40 T antigen at 3 9 T , proliferating at the same rate as at the permissive temperature. This can be explained by a genetic reversion, within the T antigen gene, to the wild type. Interestingly, no systematic appearance of antigens indicative of differentiation and no decrease in the frequency of the precursor cells was observed after a shift to the nonpermissive temperature (Table IV). This is in contrast to another tsA58 T antigen immortalized neural line (Frederiksen et al., 1988) and to other lines immortalized with temperature-sensitive oncogenes (e.g., Fiszman and Fuchs, 1975). The nature of this difference is not clear, but it may partly be a consequence of the modified vector construct. The oncogene used in the present experiment was a hybrid between the C-terminal portion of the SV40TtsA58 mutant, which confers temperature sensitivity, and the N-terminal portion of the SV40T U19 mutant. The U19 mutant is more efficient in immortalizing cells than the wild-type counterpart (Paucha et a]., 1986), and was therefore included in the construct (Almazan and McKay, 1988). It is possible that the increased immortalization activity could result in more profound changes in the cell machinery than the wild-type protein exerts, making the cells less prone to differentiate. Secondary genetic and epigenetic changes can be induced by SV40 and other DNA tumor viruses (reviewed in Linder and Marshall, 1990). For example, the endogenous clock driving the differentiation of 0 2 A progenitors (Raff et al., 1985) may have been disturbed in the cell lines. Alternatively, the hybrid U19/tsA58 oncogene may be more resistant to total inactivation at the nonpermissive temperature than tsA58 alone. Our observation of decreased, but still detectable, levels of T antigen at the nonpermissive temperature in some lines (Table 11) supports this idea. However, cells from the developing rat optic nerve immortalized with the same construct can resume differentiation upon temperature shift (Almazan and McKay , 1988). Likewise, differentiation upon shift to a higher temperature was obtained with neural cell lines immortalized with temperature-sensitive Rous sarcoma virus (Giotta et al., 1980). Moreover, differentiation to glial or neuronal fates can be observed in precursor cell lines

613

with nonconditional expression of T antigen (Evrard et al., 1990; Hammang et al., 1990) and other oncogenes (Giotta et al., 1980; Bartlett et al., 1988; Bernard et al., 1989; Birren and Anderson, 1990; Galiana et al., 1990; Hammang et al. , 1990; Mellon et al., 1990; Ryder et a1. , 1990). A third explanation for the limited differentiation observed in the present study is that the appropriate external signals for differentiation may not have been reproduced in our culture system. For example, proliferation and differentiation of single CNS precursor cells grown in vitro depend on short range soluble factors provided by cocultured neural cells (Temple, 1989). Similarly, the differentiation of some myc-immortalized neural cell lines requires exogenous or autologously secreted growth factors (Bartlett et al., 1988; Bernard et al., 1989). Taken together, the data presented here and those previously published underline the importance of using various vector constructs to immortalize different brain cell populations, in order to find optimal combinations for a particular experimental question. The present study, using the U19/tsA58 SV40 T construct, demonstrates that the immortalized cell lines from cerebellum and skin reflect characteristics of their cellular origin, and that the precursor status of the primary cell can be “captured” in an immortalized cell line. The neural differentiation capacity, although limited, was specific to the cerebellar lines. Differentiation was subject to stochastic or cell-cell interaction phenomena and could be triggered by a shift to defined medium rather than by a switch to the nonpermissive temperature.

ACKNOWLEDGMENTS The authors thank Ms. M. Marvin for providing the anti-nestin antiserum, Dr. Pat Renfranz for comments on the manuscript, and members of the laboratory for valuable discussions and suggestions. Dr. M. Takeichi generously supported the revision of this manuscript while C. Redies was a postdoctoral fellow in his laboratory. Gifts of the vector construct SV40 T U19/tsA58 from Dr. G. Almazan, of transgenic mice from Drs. J.P. Julien and A. Peterson, and of antibodies from Drs. E. Harlow, R. Hynes, B. Ranscht, M. Schachner, and W. Stallcup are gratefully acknowledged. This work was supported by the National Institutes of Health and the Pew Charitable Trust (grants to R.D.G. McKay). C. Redies was a recipient of scholarships from the German Academic Exchange Service, Sonderprogramm Gentechnologie, and the Rita Allen Foundation. U. Lendahl was a recipient of an EMBO Long Term Fellowship. R.D.G. McKay is a Rita Allen Scholar.

614

Redies et al.

REFERENCES Almazan G, McKay RDG (1988): Immortalization of an oligodendrocyte precursor cell via a retrovirus carrying a temperature sensitive oncogene. SOCNeurosci Abstr 14:1130. Bartlett PF, Reid HH, Bailey KA, Bernard 0 (1988): Immortalization of mouse neural precursor cells by the c-myc oncogene. Proc Natl Acad Sci USA 85:3255-3259. Bernard 0, Reid HH, Bartlett PF (1989): Role of the c-myc and N-myc protooncogenes in the immortalization of neural precursors. J Neurosci Res 24:9-20. Birren SJ, Anderson DJ (1990): A v-myc-immortalized sympathoadrenal progenitor cell line in which neuronal differentiation is initiated by FGF but not NGF. Neuron 4:189-201. Bottenstein JE, Sat0 GH (1979): Growth of a rat neuroblastoma cell line in serum-free supplemented medium. Proc Natl Acad Sci USA 76:514-517. Bovolenta 0, Liem RKH, Mason CA (1984): Development of cerebellar astroglia: Transitions in form and cytoskeletal content. Dev Biol 102:248-259. Cepko CL, Roberts BE, Mulligan RC (1984): Construction and application of a highly transmissible murine retrovirus shuttle vector. Cell 37: 1053-1062. Cepko CL (1989): Immortalization of neural cells via retrovirus-mediated oncogene transduction. Annu Rev Neurosci 12:47-65. Cone RD, Mulligan RC (1984): High-efficiency gene transfer into mammalian cells: Generation of helper-free recombinant retrovirus with broad mammalian host range. Proc Natl Acad Sci USA 81:6349-6353. Debus E, Weber K, Osborn M (1983): Monoclonal antibody specific for glial fibrillary acidic (GFA) protein and for each of the neurofilament triplet polypeptides. Differentiation 25: 193203. DeVitry F, Picart R, Jacque C, Legault L, Dupouey P, Tixier-Vidal A (1980): Presumptive common precursor for neuronal and glial cell lineages in mouse hypothalamus. Proc Natl Acad Sci USA 7714165 -41 69. Eisenbarth GS, Walsh FS, Nirenberg M (1979): Monoclonal antibody to a plasma membrane antigen of neurons. Proc Natl Acad Sci USA 76:4913-4917. Evrard C, Galiana E, Rouget P (1986): Establishment of “normal” nervous cell lines after transfer of polyoma virus and adenovirus early genes into murine brain cells. EMBO J 5:3157-3162. Evrard C, Galiana E, Rouget P (1988): Immortalization of bipotential glial progenitors and generation of permanent “blue” cell lines. J Neurosci Res 21:80-87. Evrard C, Borde I, Marin P, Galiana E, Premont J, Gros F, Rouget P ( I 990): Immortalization of bipotential and plastic glio-neuronal precursor cells. Proc Natl Acad Sci USA 87:3062-3066. Fiszman MY, Fuchs P (1975): Temperature sensitive expression of differentiation in transformed myoblasts. Nature 254:429-43 1. Frederiksen K, McKay RDG (1988): Proliferation and differentiation of rat neuroepithelial precursor cells in vivo. J Neurosci 8: 1144-1151. Frederiksen K, Jat PS, Valtz N, Levy D, McKay R (1988): Immortalization of precursor cells from the mammalian CNS. Neuron 11439-448. Friedman B, Zaremba S, Hockfield S (1 990): Monoclonal antibody Rat 401 recognizes Schwann cells in mature and developing peripheral nerve. J Comp Neurol 295:43-5 I . Galiana E, Borde I, Marin P, Rassoulzadegan M, Cuzin F, Gros F, ‘Rouget P, Evrard C (1990): Establishment of permanent astroglial cell lines, able to differentiate in vitro, from transgenic mice carrying the polyoma virus large T gene: An alternative

approach to brain cell immortalization. J Neurosci Res 26: 269-277. Giotta GJ, Heitzmann J, Cohn M (1980): Properties of two teniperature-sensitive Rous sarcoma virus transformed cerebellar cell lines. Brain Res 202:445-458. Hammang JP, Baetge EE, Behringer RR, Brinster RL, Palmiter RD, Messing A (1990): Immortalized retinal neurons derived from SV40 T-antigen-induced tumors in transgenic mice. Neuron 41775-782. Hockfield S, McKay RDG (1985): Identification of the major cell classes in the mammalian nervous system. J Neurosci 5:33103328. Hynes RO (1989): “Fibronectins.” Springer Series in Molecular Biology. New York: Springer. Jat PS, Sharp PA (1986): Large T antigens of simian virus 40 and polyomavirus efficiently establish primary fibroblasts. J Virol 591746-750. Jat PS, Sharp PA (1989): Cell lines established by a temperaturesensitive simian virus 40 large-T-antigen gene are growth restricted at the nonpermissive temperature. Mol Cell Biol 9: 1672-1681, Julien JP, Tretjakoff I, Beaudet L, Peterson A (1987): Expression and assembly of a human neurofilament protein in transgenic mice provide a novel neuronal marking system. Genes Dev 1: 10851095. Landry CF, Ivy GO, Dunn RJ, Marks A, Brown IR (1989): Expression of the gene encoding the P-subunit of s-100protein in the developing rat brain analyzed by in situ hybridization. Mol Brain Res 6:251-262. Lendahl U, Zimmerman LB, McKay RDG (1990): CNS stem cells express a new class of intermediate filament protein. Cell 60: 585-595. Lendahl U, McKay RDG (1990): The use of cell lines in neurobiology. Trends Neurosci 13:132-137. Levine JM (1989): Neuronal influences on glial progenitor cell development. Neuron 3:103-113. Linder S, Marshall H (1990): Immortalization of primary cells by DNA tumor viruses. Exp Cell Res 191:1-7. Maniatis T, Fritsch EF, Sambrook J (1982): “Molecular Cloning (A Laboratory Manual).” Cold Spring Harbor: Cold Spring Harbor Laboratory. Martini R, Schachner M (1986): Immunoelectronmicroscopic Iocalization of neural cell adhesion molecules (LI, N-CAM, and MAG) and their shared carbohydrate epitope and myelin basic protein in developing sciatic nerve. J Cell Biol 103: 2439-2448. Mellon PL, Windle JJ, Goldsmith PC, Padula CA, Roberts JL, Weiner RI (1990): Immortalization of hypothalamic GnRH neurons by genetically targeted tumorigenesis. Neuron 5: 1-10. Moore BW (1965): A soluble protein characteristic of the nervous system. Biochem Biophys Res Commun 19:739-744. Moura Net0 V, Mallat M, Chneiweiss H, Premont J, Gros F, Prochiantz A (1986): Two simian virus 40 (SV40)-transformed cell lines from the mouse striatum and mesencephalon presenting astrocytic characters. I. Immunological and pharmacological properties. Dev Brain Res 26:ll-22. Paucha E, Kalderon RW, Harvey RW, Smith AE (1986): Simian virus 40 origin DNA-binding domain on large T antigen. J Virol 57:50-64. Raff MC, Abney ER, Fok-Seang J (1985): Reconstitution of a developmental clock in vitro: A critical role for astrocytes in the timing of oligodendrocyte differentiation. Cell 42:61-69. Raff MC (1989): Glial cell diversification in the rat optic nerve. Science 243:1450-1455.

Mouse Cerebellar Precursor Cell Lines Ranscht B, Clapshaw PA, Price J , Noble M, Seifert W (1982): Development of oligodendrocytes and Schwann cells studied with a monoclonal antibody against galactocerebroside. Proc Natl Acad Sci USA 79:2709-2713. Rathjen FG, Schachner M (1984): Immunocytological and biochemical characterization of a new neuronal cell surface component (L1 antigen) which is involved in cell adhesion. EMBO J 3: 1-10, Ryder EF, Snyder EY, Cepko CL (1990): Establishment and characterization of multipotent neural cell lines using retrovirus vector-mediated oncogene transfer. J Neurobiol 21:356-375. Scarpini E, Meola G , Baron P, Beretta S , Velicogna M, Scarlato G (1986): S-100 protein and laminin: Immunocytochemical markers for human Schwann cells in vitro. Exp Neurol 93: 77-83. Sommer I, Schachner M (1981): Monoclonal antibodies (01to 0 4 ) to oligodendrocyte surfaces: An immunocytological study in the central nervous system. Dev Biol 83:311-327. Southern EM (1975): Detection of specific sequences among DNA fragments separated by gel electrophoresis. J Mol Biol 98: 503-525.

615

Squire J (1981): “The Structural Basis of Muscular Contraction.” New York: Plenum Publishing Corp. Stallcup WB, Beasley L, Levine JM (1983): Cell surface molecules that characterize different stages in the development of cerebellar interneurons. Cold Spring Harbor Symp Quant Biol 48: 761-774. Stallcup WB, Beasley L (1987): Bipotential glial precursor cells of the optic nerve express the NG2 proteoglycan. J Neurosci 7:27372744. Stallcup WB, Pytela R, Ruoslahti E (1989): A neuroectoderm-associated ganglioside participates in fibronectin receptor-mediated adhesion of germinal cells to fibronectin. Dev Biol 132:212229. Sternberger LA, Sternberger NH (1983): Monoclonal antibodies distinguish phosphorylated and nonphosphorylated forms of neurofilament in situ. Proc Natl Acad Sci USA 80:6126-6130. Temple S (1989): Division and differentiation of isolated CNS blast cells in microculture. Nature 340:47 1-473. Trotter J, Boulter CA, Sontheimer H, Schachner M, Wagner EF (1989): Expression of v-src arrests murine glial cell differentiation. Oncogene 4:457-464.