Injury to the adult mammalian CNS results in reactive changes among the glial cells surrounding the site of damage. Recently, an unusual class of glial cells has been identified within the intact adult rat cerebellum on the basis of the expression of the NG2 chondroitin-sulfate proteoglycan (Levine and Card, 1987). To determine whether the cells that express the NG2 proteoglycan show reactive changes after injury, small puncture lesions were made into the cerebelli of adult rats, and changes among astrocytes, microglia and NG2-positive cells were examined using immunohistochemical staining with cell type-specific marker antibodies. Beginning at 24 hr after lesion, NG2-positive cells immediately adjacent to the lesion site bound the anti-NG2 antibodies more heavily than cells within the undamaged areas of the cerebellum. This increase in anti-NG2 immunoreactivity was transient, reaching a maximum at 7 d postlesion and declining slowly thereafter. The increase in anti-NG2 immunoreactivity was accompanied by an increase in the levels of mRNA encoding the NG2 core protein as demonstrated by in situ hybridization. NG2-positive cells adjacent to the lesion site incorporated 3H-thymidine into their nuclei beginning at 24 hr postlesion and increased in number. Concurrent with these changes, microglia became activated and increased in number, monocytes invaded the damaged tissue, and an astrocytic scar formed. These observations demonstrate that the cells that express the NG2 proteoglycan are a reactive cell type that responds to brain injury. The increased expression of the NG2 chondroitin-sulfate proteoglycan may contribute to the failure of damaged CNS axons to regenerate successfully.

Immunofluorescence and immunoperoxidase techniques were used to localize a cell surface chondroitin-sulfate proteoglycan antigen, termed NG2, in the developing and adult rat cerebellum. In the adult, both polyclonal and monoclonal anti-NG2 antibodies labeled cells throughout the cerebellar cortex, with the labeled cells being especially prominent in the molecular layer. The labeled cells had small, irregularly shaped cell bodies from which thin highly branched processes radiated in a stellate array. The NG2-labeled cells were not labeled with antibodies against glial fibrillary acidic protein (GFAP), vimentin, or S-100 protein, intracellular markers for astrocytes. However, electron microscopic immunocytochemical analysis of NG2 immunoreactive cells revealed a cell morphology consistent with that of protoplasmic astrocytes. Labeled cell bodies contained a thin rim of organelle-poor cytoplasm surrounding a euchromatic nucleus. Thick processes originating from the cell soma tapered to form thin branches with highly irregular surface contours that extended between adjacent neuronal elements. The labeled processes did not form synapses in the neuropil, and no synaptic profiles onto anti-NG2-labeled cell bodies or processes were observed. Thus, we conclude that the NG2 antigen is a cell surface marker for a class of smooth protoplasmic astrocytes. Immunoreactive cells were seen in the developing cerebellum beginning at embryonic day 16. The number of labeled cells increased during the early stages of cerebellar development, reaching a peak at about postnatal day (PND) 4 or 5 and declining thereafter. In the developing cerebellum, labeled cells lying within the forming molecular layer resembled the cells seen in the adult, whereas cells lying deeper within the folia had an immature appearance with fewer processes and less branching. This apparent gradient of morphological maturation suggests that an interaction with parallel fibers in the developing molecular layer may play a role in the terminal cytodifferentiation of the NG2-labeled smooth protoplasmic astrocytes.

The NILE (nerve growth factor-inducible large external) glycoprotein is a 230,000-dalton molecule found on the surface of PC12 cells. immunologically cross-reactive glycoproteins in the molecular weight range of 215,000 to 230,000 have been found on many types of neurons in culture. Using immunohistochemical methods, we have shown that NILErelated glycoproteins are present in neuronal fiber tracts of the developing rat brain. Antibody against the NILE glycoprotein specifically labels processes that appear identical to those recognized by antibodies against the neurofilament triplet of proteins. These processes are clearly distinct from the radial glial fibers recognized by antibody against the intermediate filament protein vimentin. NILE glycoprotein is not distributed uniformly over the entire neuronal surface but is concentrated on neurites and is much less abundant on cell bodies. NILE-positive fiber tracts are first seen in the spinal cord and rhombencephalon on embryonic day 11 and over the next 2 days appear in the mesencephalon and diencephalon.Staining in the telencephalon is not seen until embryonic day 15. The appearance of NILE immunoreactivity in these various regions closely parallels the appearance of neurofilament polypeptides, suggesting that NILE-related glycoproteins are present during the early phases of fiber tract formation. This idea is supported by the finding that the NILE glycoprotein can be found postnatally in parts of the nervous system such as the cerebellar cortex and olfactory bulb which undergo major histogenesis during the postnatal period. In the cerebellum the appearance of NILE immunoreactivity in the two major fiber zones, the molecular layer and the white matter, parallels the development of the fiber structure of these layers. These findings support tissue culture studies which suggest a role for the NILE glycoprotein in mediating nerve fiber fasciculation.

A monoclonal antibody against a cell surface ganglioside present on neuroepithelial cells was produced by immunizing mice with the B49 cell line, a clonal line with properties of both neurons and glial cells. The expression of this antigen, designated as D1.1, was analyzed in the developing rat cerebellum. The D1.1 antigen was localized by the immunofluorescent staining method to germinal cells of the external granule layer (EGL). Fluorescent labeling of cells comprising the EGL was apparent on embryonic day 18 when the EGL first forms, and the labeling was present throughout the period of postnatal cerebellar development. No cells within the adult cerebellum were labeled with the anti-D1.1 antibody. The D1.1-labeled cells of the EGL synthesize DNA, as demonstrated by [3H] thymidine autoradiography. However, within 48 hr after their final mitosis, nascent cerebellar cells that had migrated away from the external granule layer were no longer labeled with antibody. Some of the neurons and some of the astrocytes in cerebellar cell cultures were fluorescently labeled with the anti-D1.1 antibody. The number of anti-D1.1-labeled neurons in the cultures decreased over the first 10 days in vitro in agreement with the findings that in vivo the fluorescent labeling of the D1.1 antigen disappears from postmitotic cells. The antibody recognizes a ganglioside that in thin layer chromatographic experiments has a mobility between that of the GM1 and GM2 ganglioside. These data suggest that the D1.1 ganglioside antigen is a cell surface marker for germinal cells and that the acquisition and subsequent loss of this antigen is an aspect of the biochemical maturation of neurons and glial cells.

Antibodies raised in rabbits and in guinea pigs against nerve growth factor-treated PC12 cells were absorbed exhaustively with three non-neuronal cell lines. In immunofluorescent staining experiments, these absorbed sera, designated anti-PC12, labeled specifically several different types of cultured neurons, including cerebral, cerebellar, spinal, dorsal root, and superior cervical neurons. Non-neuronal cells in these cultures, such as astrocytes, oligodendrocytes, and fibroblasts, were not labeled. However, neural crest-derived adrenal chromafhn cells and Schwann cells were stained by anti-PC12. Further absorption with adrenal tissue rendered anti-PC12 specific for neurons and Schwann cells, while still further absorption with Schwann cells yielded an antiserum that was reactive only with dorsal root and superior cervical neurons. Absorption of anti-PC12 with adult brain resulted in the loss of activity against all cells except PC12 cells.These absorption experiments suggested that anti-PC12 recognized at least four distinct cell surface components. Some anti-PC12-reactive components were identified by using the anti-PC12 serum to prepare immune precipitates from detergent extracts of 12"1-labeled cells. The immune precipitates were analyzed by polyacrylamide gel electrophoresis. Anti-PC12 precipitated two components with apparent molecular weights of 190,000 and 140,000 that were common to Schwann cells, chromaffin cells, and all types of neurons. Smaller quantities of these two components could also be precipitated from cerebellar glial cells. In addition, anti-PC12 precipitated a glycoprotein of greater than 200,000 daltons from the cultured neurons, the neuronal cell lines, and the Schwann cells, but not from chromaffin cells or from cerebellar glia. Each neuronal cell type expressed a characteristic form of this large glycoprotein, as judged by electrophoretic mobility. For example, the molecules found on four cell lines, PC12, B35, N18, and /3HC, had apparent molecular weights of 235,000, 225,000, 220,000, and 215,000, respectively. The 235,000-dalton glycoprotein from PC12 cells is likely to be identical to the NILE glycoprotein described by McGuire et al. (McGuire, J., L. Greene, and A. Furano (1978) Cell 15: 357-365), while the 220,000-dalton component from N18 cells is probably similar to that described by Akeson and Hsu (Akeson, R., and W. Hsu (1978) Exp. Cell Res. 115: 367-377). Adrenal-absorbed anti-PC12 did not precipitate the 190,000-and 140,000-dalton components from any of the cell types but continued to precipitate the 215,000-to 235,000-dalton glycoproteins from the neurons and Schwann cells. These high molecular weight glycoproteins thus appear to be cell surface markers for neurons and Schwann cells.The PC12 cell line (Greene and Tischler, 1976) shares many properties with sympathetic neurons and has thus provided an extremely useful model system for studying several types of neuronal mechanisms, including the function of neuronal acetylcholine receptors (Patrick and