Evidence that perinatal and adult NG2-glia are not conventional oligodendrocyte progenitors and do not depend on axons for their survival

In the adult CNS, cells that express the NG2 chondroitin sulphate proteoglycan are considered oligodendrocyte progenitor cells (OPCs). However, adult NG2-glia are highly complex cells, suggesting they may be a mature glial cell type distinct from conventional OPCs. To test this possibility, we have determined the response of NG2-glia to axon loss in the neonatal rat optic nerve, which has been shown previously to result in an almost complete loss of OPCs. We show here that NG2-glia are not lost following enucleation of the neonatal or juvenile optic nerve and that at all ages NG2-glia become "reactive" in response to axon degeneration, in direct contrast to previous studies on OPCs. We provide evidence that NG2-glia with an adult morphological phenotype develop postnatally, at the same time as oligodendrocytes. In the neonatal nerve, we show that the loss of axons blocks the development of oligodendrocytes, although reactive NG2-glia are abundant and are able to generate oligodendrocytes when placed in culture. Conversely, axon transection did not result in the loss of mature oligodendrocytes, indicating that the dependence of oligodendrocytes on axon-derived survival factors decreases with differentiation. This study shows that NG2-glia are a highly reactive cell type and do not generate oligodendrocytes in vivo in the absence of appropriate axon-derived signals. This may underlie in part the failure of remyelination in chronic multiple sclerosis, despite the abundance of NG2+ expressing glial cells within demyelinated lesions. It is clear either that OPCs do not depend on axons for their survival, in total contradiction to previous studies, or that in vivo a population of NG2 expressing glia are not perinatal or adult OPCs in the accepted sense. We suggest that a population of NG2-glia diverge postnatally into a novel glial type that is functionally and behaviourally distinct from conventional OPCs.     

Different response of astrocytes and Bergmann glial cells to portacaval shunt: an immunohistochemical study in the rat cerebellum.

The present study was performed in order to follow the response of rat cerebellum astroglial cells (Bergmann glial cells and astrocytes) to long-term portacaval shunt (PCS), by means of glial fibrillary acidic protein (GFAP) and vimentin immunoreactivities. Bergmann glia accumulated GFAP in response to PCS, whereas astrocytes decreased GFAP immunoreactivity when compared to control rats. The increase of GFAP occurs in cells located in the cerebellar layer where glutamate is mainly released. Since the vimentin content remained unaltered in response to PCS, when compared to control rats, it can be concluded that only the GFAP filaments are affected by PCS. Nevertheless, GFAP immunoreactivity presents regional differences in the cerebellar astroglial population, and the factors responsible for these variations are still unknown.     

NG2 proteoglycan-expressing cells of the adult rat brain: possible involvement in the formation of glial scar astrocytes following stab wound

Stab wound lesion to the adult central nervous system induces strong proliferative response that is followed by the formation of a dense astroglial scar. In order to determine the origin of those astrocytes composing the glial scar, the cell proliferation marker bromodeoxyuridine (BrdU) was administered to lesioned rats that were fixed 3 h or 6 days later. At 3 h after the BrdU administration, labeled nuclei were frequently associated with either NG2(+) cells or microglia/macrophages, but rarely with astrocytes expressing glial fibrillary acidic protein (GFAP). Six days later, by contrast, numerous BrdU-labeled nuclei were associated with astrocytes located along the lesion borders. After the injection of a viral vector of the green fluorescent protein (GFP) into the lesional cavity, GFP was preferentially detected within NG2- or GFAP-labeled cells when lesioned animals were fixed 1 or 6 days after the injections, respectively. The combined detection of glial markers within cells present in the lesioned area indicated that, although they rarely express GFAP, the marker of mature astrocytes, NG2(+) cells located along the lesion borders frequently express nestin and vimentin, i.e., two markers of immature astrocytes. Lastly, chronic treatment of lesioned rats with dexamethasone was found to inhibit the proliferation of NG2(+) cells present within the lesioned area and to subsequently alter the formation of a dense astroglial scar. Taken together, these data strongly suggest that following a surgical lesion, at least a portion of the astrocytes that constitute the glial scar are issued from resident NG2(+) cells.     

Astroglial cells in the external granular layer are precursors of cerebellar granule neurons in neonates

It is well established that cerebellar granule cell precursors (GCPs) initially derive from progenitors in the rhombic lip of the embryonic cerebellar primordium. GCPs proliferate and migrate tangentially across the cerebellum to form the external granule cell layer (EGL) in late embryogenesis and early postnatal development. It is unclear whether GCPs are specified exclusively in the embryonic rhombic lip or whether their precursor persists in the neonate. Using transgenic mice expressing DsRed under the human glial fibrillary acidic protein (hGFAP) promoter, we found 2 populations of DsRed⁺ cells in the EGL in the first postnatal week defined by bright and faint DsRed-fluorescent signal. Bright DsRed⁺ cells have a protein expression profile and electrophysiological characteristics typical of astrocytes, but faint DsRed⁺ cells in the EGL and internal granule cell layer (IGL) express markers and physiological properties of immature neurons. To determine if these astroglial cells gave rise to GCPs, we genetically tagged them with EGFP or βgal reporter genes at postnatal day (P)3–P5 using a hGFAP promoter driven inducible Cre recombinase. We found that GFAP promoter⁺ cells in the EGL are proliferative and express glial and neural stem cell markers. In addition, immature granule cells (GCs) en route to the IGL at P12 as well as GCs in the mature cerebellum, 30 days after recombination, express the reporter protein, suggesting that GFAP promoter⁺ cells in the EGL generate a subset of granule cells. The identification of glial cells which function as neuronal progenitor cells profoundly impacts our understanding of cellular plasticity in the developing cerebellum.     

GFAP promoter-controlled EGFP-expressing transgenic mice: a tool to visualize astrocytes and astrogliosis in living brain tissue

We have generated transgenic mice in which astrocytes are labeled by the enhanced green fluorescent protein (EGFP) under the control of the human glial fibrillary acidic protein (GFAP) promoter. In all regions of the CNS, such as cortex, cerebellum, striatum, corpus callosum, hippocampus, retina, and spinal cord, EGFP-positive cells with morphological properties of astrocytes could be readily visualized by direct fluorescence microscopy in living brain slices or whole mounts. Also in the PNS, nonmyelinating Schwann cells from the sciatic nerve could be identified by their bright green fluorescence. Highest EGFP expression was found in the cerebellum. Already in acutely prepared whole brain, the cerebellum appeared green-yellowish under normal daylight. Colabeling with GFAP antibodies revealed an overlap with EGFP in the majority of cells. Some brain areas, however, such as retina or hypothalamus, showed only low levels of EGFP expression, although the astrocytes were rich in GFAP. In contrast, some areas that were poor in immunoreactive GFAP were conspicuous for their EGFP expression. Applying the patch clamp technique in brain slices, EGFP-positive cells exhibited two types of membrane properties, a passive membrane conductance as described for astrocytes and voltage-gated channels as described for glial precursor cells. Electron microscopical investigation of ultrastructural properties revealed EGFP-positive cells enwrapping synapses by their fine membrane processes. EGFP-positive cells were negative for oligodendrocyte (MAG) and neuronal markers (NeuN). As response to injury, i.e., by cortical stab wounds, enhanced levels of EGFP expression delineated the lesion site and could thus be used as a live marker for pathology.     

Plasticity of developing cerebellar cells in vitro studied with antibodies against the NG2 antigen

The NG2 antigen, a chondroitin-sulfate proteoglycan, is a cell surface marker for a class of smooth protoplasmic astrocytes found throughout the brain and at high frequency in the cerebellar molecular layer (Levine and Card, 1987). To study the development of the cerebellar astrocytes at the level of the single cell, we have analyzed the distribution of the NG2 antigen by indirect immunofluorescence in dissociated cell cultures prepared from postnatal cerebella and compared that distribution to the distribution of several other cell surface and intracellular antigens that identify specific cell types in cultures of nervous tissue. When cerebellar cells from 5 d rat pups were grown in a medium containing 10% fetal calf serum, the NG2-labeled cells, which constituted 0.1-1.0% of the total glial cells present, contained glial fibrillary acidic protein (GFAP)-immunoreactive filaments and bound monoclonal antibody A2B5, a surface marker for neurons and some astrocytes. Approximately 30% of the NG2-labeled cells were also labeled with tetanus toxin, an additional surface marker for neurons and immature astrocytes. Less than 2% of the cells were labeled with antibodies against galactocerebroside or with monoclonal antibody O1, both of which are surface markers for oligodendrocytes. About half the NG2-labeled cells exhibited high-affinity uptake of 3H-GABA, and this uptake was partially inhibited by both beta-alanine and DABA. Thus, the NG2 antigen is a cell surface marker for a subpopulation of the type II or fibrous astrocytes present in the cultures. When the cerebellar cells were grown in a chemically defined, serum-free medium, the NG2-labeled cells had a stellate morphology and between 50-60% of the cells bound tetanus toxin. Although almost all the cells bound antibody A2B5, less than 5% of the cells expressed either of the oligodendrocyte surface markers or GFAP immunoreactivity. As was the case with cells grown in serum-containing medium, 60% of the NG2-labeled cells had high-affinity uptake of 3H-GABA. However, this uptake was inhibited by DABA but not by beta-alanine. This phenotype may be the in vitro analog of the NG2-labeled, filament-lacking, smooth protoplasmic astrocytes identified in the intact adult cerebellum. The expression of these 2 phenotypes could be reversed by switching the tissue culture medium within 5 d of plating the cells. These results demonstrate that the in vitro environment can influence the phenotypic properties expressed by developing cerebellar astrocytes and suggest that smooth protoplasmic astrocytes may be developmentally related to glial cells of the O-2A lineage.     

Glial fibrillary acidic protein gene promoter is differently modulated by transforming growth factor-beta 1 in astrocytes from distinct brain regions

The expression of glial fibrillary acidic protein (GFAP), the major intermediate filament protein of mature astrocytes, is regulated under developmental and pathological conditions. Recently, we have investigated GFAP gene modulation by using a transgenic mouse bearing part of the GFAP gene promoter linked to the beta-galactosidase reporter gene. We demonstrated that cerebral cortex neurons activate the GFAP gene promoter, inducing transforming growth factor-beta 1 (TGF-beta 1) secretion by astrocytes. Here, we report that cortical neurons or conditioned medium derived from them do not activate the GFAP gene promoter of transgenic astrocytes derived from midbrain and cerebellum suggesting a neuroanatomical regional specificity of this phenomenon. Surprisingly, they do induce synthesis of TGF-beta 1 by these cells. Western blot and immunocytochemistry assays revealed wild distribution of TGF receptor in all subpopulations of astrocytes and expression of TGF-beta 1 in neurons derived from all regions, thus indicating that the unresponsiveness of the cerebellar and midbrain GFAP gene to TGF-beta 1 is not due to a defect in TGF-beta 1 signalling. Together, our data highlight the great complexity of neuron-glia interactions and might suggest a distinct mechanism underlying modulation of the GFAP gene in the heterogeneous population of astrocytes throughout the central nervous system.     

A potential role for bone morphogenetic protein signalling in glial cell fate determination following adult central nervous system injury in vivo

Bone morphogenetic proteins (BMPs) and their endogenous inhibitors, including noggin, chordin and follistatin, have roles in pattern formation and fate specification of neuronal and glial cells during nervous system development. We have examined their influence on glial reactions in the injured central nervous system (CNS). We show that penetrating injuries to the brain and spinal cord resulted in the upregulation of BMP-2/4, BMP-7, and noggin, with the latter being expressed almost exclusively by reactive astrocytes at the injury site, and we show that astrocytes in vitro produce noggin. As BMPs have been shown to drive cultured NG2-positive oligodendrocyte precursors (OPCs) towards a multipotential phenotype (type II astrocytes), we investigated the effects of inhibiting noggin with a function-blocking antibody (noggin-FbAb). In vitro, BMP-driven conversion of OPCs to type 2 astrocytes was inhibited by noggin, an effect that was reversed by noggin-FbAb. Noggin-FbAb also increased the number of type 2 astrocytes generated from cultured OPCs exposed to an astrocyte feeder layer, consistent with astrocytes producing both BMPs and noggin. In knife cut injuries in vivo, noggin-FbAb treatment resulted in an increase in the number of NG2-positive cells and small GFAP-positive cells in the injury site, and the appearance of glial cells with the morphological and antigenic characteristics of type 2 astrocytes (as generated in vitro), with coexpression of both GFAP and NG2. This potential conversion of inhibitory OPCs to type 2 astrocyte-like cells in vivo suggests that endogenous BMPs, unmasked by noggin antagonism, might be exploited to manipulate cell fate following CNS trauma.     

Reduced glial fibrillary acidic protein and glutamine synthetase expression in astrocytes and Bergmann glial cells in the rat cerebellum caused by delta(9)-tetrahydrocannabinol administration during development

In this study we analyzed the responses of cerebellar astroglial cells to pre- and perinatal delta(9)-tetrahydrocannabinol (THC) exposure in three postnatal ages and both sexes. To determine whether THC during development directly modifies astroglial growth, this study investigated the effects of THC on astroglial morphological changes and on the expression of specific astroglial markers (glial fibrillary acidic protein: GFAP and glutamine synthetase: GS). Our results demonstrated that the administration of THC during development has deleterious effects on astroglial maturation in the cerebellum. These results also indicate that THC might interfere with astroglial differentiation in a way dependent on sex. The effect of cannabinoids on the development of cerebellar astroglial cells (astrocytes and Bergmann glial cells) is to reduce protein synthesis, since both GFAP and GS decreased in astroglial cells, not only during THC exposure but also in adult ages. Our data suggest that pre- and perinatal THC exposure directly interferes with astroglial maturation by disrupting normal cytoskeletal formation, as indicated by the irregular disposition of GFAP and the lower GFAP expression observed at all the ages studied. THC exposure during development may also modulate glutamatergic nervous activity since GS expression is reduced in THC-exposed brains. GS expression increased progressively after THC withdrawal, but GS expression had still not reached control values two months after THC withdrawal. This indicates that glutamate uptake is lower in glial cells exposed to THC, since GS expression is lower than in older controls. Consequently, glutamatergic neurotransmission may be affected by cannabinoid exposure during gestation. Therefore, cannabinoids exert developmental toxicity, at least on astroglial cells, which could contribute to fetal brain growth retardation.     

Injuring neurons induces neuronal differentiation in a population of hippocampal precursor cells in culture

A novel population of hippocampal precursor cells (HPCs) that can be induced to differentiate into astrocytes and oligodendrocytes can be derived from hippocampal cultures grown in serum-free media. The HPCs are PDGF-responsive, do not proliferate with bFGF, and grow as sheets of cells rather than gathering into neurospheres. The HPCs share many markers (A2B5, GD3, poly-sialylated neuronal common adhesion molecule (PSA-NCAM), and NG2) with oligodendrocyte precursor cells (OPCs). The HPCs do not express markers for mature neurons, astrocytes, or oligodendrocytes. Like OPCs, the HPCs differentiate into glial fibrillary acidic protein (GFAP)+ astrocytes and GalC+ oligodendrocytes with the addition of bone morphogenetic protein-4 (BMP-4) and triiodothyronine (T3), respectively. They do not differentiate into neurons with the addition or withdrawal of basic fibroblast growth factor (bFGF), brain-derived neurotrophic factor (BDNF), or retinoic acid (RA). These HPCs can be stimulated to differentiate into neuron-like cells by the induction of neuronal injury or cell death in nearby cultured neurons or by conditioned medium from injured neuronal cultures. Under these conditions, HPCs grow larger, develop more extensive dendritic processes, become microtubule-associated protein-2-immunoreactive, express large voltage-dependent sodium currents, and form synaptic connections. The conversion of endogenous pluripotent precursor cells into neurons in response to local brain injury may be an important component of central nervous system homeostasis.     

Reactive astroglia-neuron relationships in the human cerebellar cortex: A quantitative, morphological and immunocytochemical study in Creutzfeldt-Jakob disease

In order to investigate the role of neuron-glia interactions in the response of astroglial to a non-invasive cerebellar cortex injury, we have used two cases of the ataxic form of Creutzfeldt-Jakob disease (CJD) with distinct neuronal loss and diffuse astrogliosis. The quantitative study showed no changes in cell density of either Purkinje or Bergmann glial cells in CJ-1, whereas in the more affected CJ-2 a loss of Purkinje cells and an increase of Bergmann glial cells was found. The granular layer in both CJD cases showed a similar loss of granule cells (about 60% ) in parallel with the significant increase in GFAP+ reactive astrocytes. GFAP immunostaining revealed greater reactivity of Bergmann glia in CJ-2 than in CJ-1, as indicated by the thicker glial processes and the higher optical density. Granular layer reactive astrocytes were regularly spaced. In both CJD cases there was strict preservation of the spatial arrangement of all astroglial subtypes—Fañanas cells, Bergmann glia and granular layer astrocytes. Reactive Fañanas and Bergmann glial cells and microglia/macrophages expressed vimentin, while only a few vimentin+ reactive astrocytes were detected in the granular layer. Karyometric analysis showed that the increase in nuclear volume in reactive astrloglia was directly related with the level of glial hypertrophy. The number of nucleoli per nuclear section was constant in astroglial cells of human controls and CJD, suggesting an absence of polyploidy in reactive astroglia. Ultrastructural analysis revealed junctional complexes formed by the association of macula adherens and gap junctions. In the molecular layer numerous vacant dendritic spines were ensheathed by lamellar processes of reactive Bergmann glia. Our results suggest that quantitative (neuron/astroglia ratio) and qualitative changes in the interaction of neurons with their region-specific astroglial partners play a central role in the astroglial response pattern to the pathogenic agent of CJD.     

Development and differentiation of glial precursor cells in the rat cerebellum

The development and differentiation of bipotential glial precursor cells has been studied extensively in tissue culture, but little is known about the distribution and fate of these cells within intact animals. To analyze the development of glial progenitor cells in the developing rat cerebellum, we utilized immunofluorescent, immunocytochemical, and autoradiographic techniques. Glial progenitor cells were identified with antibodies against the NG2 chondroitin-sulfate proteoglycan, a cell-surface antigen of 02A progenitor cells in vitro, and the distribution of this marker antigen was compared to that of marker antigens that identify immature astrocytes, mature astrocytes, oligodendrocyte precursors, and mature oligodendrocytes. Cells expressing the NG2 antigen appeared in the cerebellum during the last 3-4 days of embryonic life. Over the first 10 days of postnatal life, the NG2-labeled cells incorporated ³H-thymidine into their nuclei and their total number increased. At all ages examined, the NG2-labeled cells did not contain either vimentin-like or glial fibrillary acidic protein (GFAP)-like immunoreactivity, suggesting that they do not develop along an astrocytic pathway. NG2-labeled cells of embryonic animals expressed G_D3 ganglioside antigens, a property of oligodendrocyte precursors, whereas NG2-positive cells of postnatal animals did not express G_D3 immunoreactivity. Nevertheless, the NG2-labeled cells of the nascent white matter expressed oligodendrocyte-specific marker antigens. Cells lyingoutside of the white matter continued to express the NG2 antigen. In adult animals, the NG2-labeled cells incorporated ³H-thymidine. Glial cells isolated from adult animals and grown in tissue culture express the NG2 antigen and display the phenotypic plasticity characteristic of 02A progenitor cells. These findings demonstrate that a population of glial progenitor cells is extensive within both young and adult animals.     

Uptake [H]GABA by oligodendrocytes in dissociated brain cell culture: A combined autoradiographic and immunocytochemical study

Uptake of [³H]GABA by dissociated mixed cell cultures of fetal mouse brain was studied using light microscopic autoradiography. Major cell types in the cultures were identified and quantified by immunocytochemical localization of reliable cell type-specific antigenic markers. In 12 days in vitro (DIV) cultures [³H]GABA uptake was predominantly into neurons and oligodendrocytes, whilst at 28 DIV the only surface cells labeled were oligodendrocytes. This was confirmed by complement-dependent antibody-mediated cytotoxicity against galatocerebroside-positive oligodendrocytes. There was a moderate labeling of almost all flat cells, the majority of which were glial fibrillary acidic protein (GFAP)-positive astrocytes. Heavily labeled astrocytes were only occasionally observed. Oligodendrocytes accumulated [³H]GABA more rapidly than astrocytes but slower than neurons. Oligodendroglial labeling was predominantly over the cell body, whereas neuronal labeling was more uniformly distributed over cell body and processes. The uptake was inhibited by diaminobutyric acid (DABA) and nipecotic acid, but not by β-alanine, and thus had similar characteristics to neuronal rather than astroglial uptake. Oligodendrocytes did not accumulate [³H]β-alanine, which labeled only astrocytes. Oligodendroglial [³H]GABA uptake was Na⁺-dependent and sensitive to ouabain, but was only slightly enhanced by aminooxyacetic acid (AOAA), whereas astroglial uptake was not sensitive to ouabain but was markedly enhanced by AOAA. The results indicate that oligodendrocytes, in addition to astrocytes, may also be involved in the modification of neuronal function by the uptake and inactivation of neuroactive substances.     

Role of astroglial cell clones in the survival and differentiation of cerebellar embryonic neurons

To investigate the role of astrocytes in the survival and differentiation of cerebellar neurons during development, we have used astroglial cell clones, derived from 8-day postnatal cerebellar explants and which might be the in vitro equivalents of the 3 main types of cerebellar astrocytes, the Golgi epithelial cells and their Bergmann processes, the velate protoplasmic and the fibrous astrocytes (F. Alliot and B. Pessac, Brain Res., 306 (1984) 283-291). Nearly all single cells, dissociated from 15-day embryonic mouse cerebella and seeded at low density, adhered to layers of each of the cerebellar astroglial cell clones as well as to other glial lines or artificial substrates. However, the cerebellar embryonic neurons survived well only on monolayers of either the 'Golgi-Bergmann'-like or the 'velate protoplasmic'-like clones. On these layers, 60-80% of the neurons were still present after 5 days of co-culture, while only less than 5% survived on the other types of substrates. The differentiation pattern of the neurons surviving on the 'Golgi-Bergmann' and the 'velate protoplasmic' astroglial clones was studied with markers of postmitotic granule cells, the major neuronal population in adult cerebellum. The velate protoplasmic-like clone was the only one able to support the coordinate acquisition by most surviving neurons of the phenotypic characteristics of granule cells, i.e. a distinct morphology, a specific epitope binding the monoclonal antibody 7-8 D2 and immunoreactivity to glutamate. These data show a broad heterogeneity in the capacity of astroglial cell clones to support embryonic cerebellar neurons. In addition, they indicate that neuronal survival per se is not sufficient for the acquisition of a differentiated neuronal phenotype.     

GFAP-expressing cells in the postnatal subventricular zone display a unique glial phenotype intermediate between radial glia and astrocytes

Neural stem cells in the adult subventricular zone (SVZ) derive from radial glia and express the astroglial marker glial fibrillary acidic protein (GFAP). Thus, they have been termed astrocytes. However, it remains unknown whether these GFAP-expressing cells express the functional features common to astrocytes. Using immunostaining and patch clamp recordings in acute slices from transgenic mice expressing green fluorescent protein (GFP) driven by the promoter of human GFAP, we show that GFAP-expressing cells in the postnatal SVZ display typical glial properties shared by astrocytes and prenatal radial glia such as lack of action potentials, hyperpolarized resting potentials, gap junction coupling, connexin 43 expression, hemichannels, a passive current profile, and functional glutamate transporters. GFAP-expressing cells express both GLAST and GLT-1 glutamate transporters but lack AMPA-type glutamate receptors as reported for dye-coupled astrocytes. However, they lack 100 microM Ba2+-sensitive inwardly rectifying K+ (K(IR)) currents expressed by astrocytes, but display delayed rectifying K+ currents and 1 mM Ba2+-sensitive K+ currents. These currents contribute to K+ transport at rest and maintain hyperpolarized resting potentials. GFAP-expressing cells stained positive for both K(IR)2.1 and K(IR)4.1 channels, two major K(IR) channels in astrocytes. Ependymal cells, which also derive from radial glia and express GFAP, display typical glial properties and K(IR) currents consistent with their postmitotic nature. Our results suggest that GFAP-expressing cells in concert with ependymal cells can perform typical astrocytic functions such as K+ and glutamate buffering in the postnatal SVZ but display a unique set of functional characteristics intermediate between astrocytes and radial glia.     

Bone morphogenetic proteins mediate cellular response and, together with Noggin, regulate astrocyte differentiation after spinal cord injury

Bone morphogenetic proteins (BMPs) play a critical role in regulating cell fate determination during central nervous system (CNS) development. In light of recent findings that BMP-2/4/7 expressions are upregulated after spinal cord injury, we hypothesized that the BMP signaling pathway is important in regulating cellular composition in the injured spinal cord. We found that BMP expressions were upregulated in neural stem cells (NSCs), neurons, oligodendrocytes and microglia/macrophages. Increased expression levels of pSmad1/5/8 (downstream molecules of BMP) were detected in neurons, NSCs, astrocytes, oligodendrocytes and oligodendroglial progenitor cells (OPCs). Active astrocytes which form the astroglial scar were probably derived from NSCs, OPCs and resident astrocytes. Since quiescent NSCs in the normal adult spinal cord will proliferate and differentiate actively into neural cells after traumatic injury, we proposed that BMPs can regulate cellular components by controlling NSC differentiation. Neurosphere culture from adult mouse spinal cord showed that BMP-4 promoted astrocyte differentiation from NSCs while suppressing production of neurons and oligodendrocytes. Conversely, inhibition of BMP-4 by Noggin notably decreased the ratio of astrocyte to neuron numbers. However, intrathecal administration of Noggin in the injured spinal cord failed to attenuate glial fibrillar acidic protein (GFAP) expression even though it effectively reduced pSmad expression. Noggin treatment did not block phosphorylation of Stat3 and the induction of GFAP in the injured spinal cord, suggesting that in addition to the BMP/Smad pathway, the JAK/STAT pathway may also be involved in the regulation of GFAP expression after spinal cord injury.