I~euron, Vol. 1, 439-448, August, 1988, Copyright 0 1988 by Cell Press Immortalization of Precursor Cells from the Mammalian CNS t(. Frederiksen: P S. Jat,t N. Valtz, D. levy, and R. McKay ilepartment of Biology and Department of Brain and Cognitive Sciences ,vlassachusetts Institute of Technology Cambridge, Massachusetts 02139 fummary Recent studies show that the nervous system contains many molecularly distinct cell types. Clonal cell mark- ing experiments demonstrate that different cell types En some areas of the CNS are products of a multipoten- tial stem cell. The factors controlling the differentiation of vertebrate CNS precursor cells would be more acces- sible to molecular analysis if cell lines with precursor properties could be established. Here we show that cell iines expressing an antigenic marker specific for a ma- ior brain precursor cell population can be established from rat cerebellum. We demonstrate that cell lines ex- press the precursor, neuronal or glial properties de- pending on the growth conditions. This work supports the view that brain precursor cells expressing the marker Rat 401 are multipotential and can differentiate into cells with either neuronal or glial properties. Cell lines capable of differentiation should be useful in defining the signaling systems generating the cell types of the brain. kntroduction ‘3rain function requires a large number of distinct cell :ypes. How this cellular diversity is specified is a central (question of developmental biology. Lineage analysis of ,.etinal development in flies, frogs, and mice shows that different neuronal and glial cell types can be derived :rom a multipotential precursor cell (Ready et al., 1978; awrence and Green, 1979; Turner and Cepko, 1987; iolt et al., 1988; Wetts and Fraser, 1988). These studies ,;upport a model in which the complex cellular organiza- ‘ion of the brain is a consequence of interactions between .,recursor cells and differentiated cells in the neuroepi- helium. To identify the cell types present in the developing nammalian neuroepithelium, we made a set of mono- s:lonal antibodies against embryonic rat nervous system Hockfield and McKay, 1985). One of these antibodies, bt 401, identifies a 200 kd protein encoded by a 6.0 kb nRNA that is only expressed in the developing nervous jystem when neurons are differentiating (Lendahl et al., 1988, Sot. Neurosci. abstract). Rat 401 specifically binds to a large in vivo population of neuroepithelial cells in the b Present address: Department of Pharmacology, Nova Research Aboratory, Alle, 2880 Bagsvaerd, Denmark. 1 Present address: Ludwig Institute, Courtauld Building, 91 Riding -louse Street, London WlP 86T, England. rat with characteristics expected of neuronal and glial precursor cells (Frederiksen and McKay, 1988). It would be useful to have cell lines with embryonic properties to characterize the signaling pathways in brain precursor cells. In this paper we report that Rat 401-positive cell lines can be established from the de- veloping CNS of the rat. These cell lines are stable, and their state of differentiation can be altered by changing the conditions of growth. Results The feasibility of generating immortal cell lines that ex- press the functions of neuroepitheiial precursor cells is influenced by three major factors: the primary cell pop- ulations available for immortalization, the immortaliz- ing method, and the conditions and assays used to ana- lyze the putative precursor cells and their differentiated products. Proliferating Cerebellar Cell Types We chose to use the cerebellum as a source of primary cells because neurogenesis in this region is postnatal, al- lowing easy access to large numbers of cells at early stages of development. As we proposed to use retroviral vectors to establish cell lines from the postnatal cerebel- lum, we first defined, in vivo, the proliferating cell types that are potential targets for retrovirus-mediated gene transfer. The Rat 401 antigen and vrmentin are markers for precursor populations. In addition, a third popula- tion may be identified by the expression of the astrocytic marker glial fibrillary acidic protein (GFAP). Dissociated cerebellar cells were stained with Rat 401, anti-vimentin, and anti-GFAP antibodies (Figure 1). The proportion of Rat 401-positive cells is initially large and declines to zero before postnatal day 15 (PN15). These results on dissociated cells are consistent with previous studies on the distribution of rat 401 in sec:tions of the developing cerebellum (Hockfield and McKay, 1985). The propor- tion of vimentin-positive cells is much larger than the proportion of Rat 401-positive cells, but it also declines over this period. The proportion of GFAP-positive cells is initially lower than the proportion of Rat 401-positive cells, but by the end of the second postnatal week, the number of GFAP-positive cells increases. Figure 1 shows that these three markers are differen- tially expressed in cerebellar cell populations, but fur- ther data are needed to establish whether these cell populations are independent of one another. The over- lap in cell populations was determined by double label immunohistochemistry (Table 1). These results showthat the Rat 401-positive population is a subset of the vimen- tin population. The GFAP-positive population and the vimentin-positive population also overlap. The proliferative status of the antigenically distinct cell types was measured by an in vivo pulse of [jH]thymi- dine followed by immunohistochemistry and autoradi-