Regulation of nerve growth factor mRNA by interleukin-1 in rat hippocampal astrocytes is mediated by NFkappaB.

Cytokines such as interleukin-1β (Il-1) are produced in the brain during development and during inflammatory processes that result from lesions or disease. One function of Il-1 in the brain appears to be the stimulation of astrocytes to proliferate and produce a variety of cytokines and trophic factors, including nerve growth factor. The mechanisms by which Il-1 exerts its actions on astrocytes remain poorly defined. We present evidence that this cytokine elicits activation of the NFκB transcription factor and that this transcription factor mediates effects of Il-1 on nerve growth factor mRNA expression. Elucidation of the processes by which cytokines activate astrocytes and influence trophic factor expression may provide insight into mechanisms governing inflammatory processes within the central nervous system.

Cytokines such as interleukin-1␤ (Il-1) are produced in the brain during development and during inflammatory processes that result from lesions or disease. One function of Il-1 in the brain appears to be the stimulation of astrocytes to proliferate and produce a variety of cytokines and trophic factors, including nerve growth factor. The mechanisms by which Il-1 exerts its actions on astrocytes remain poorly defined. We present evidence that this cytokine elicits activation of the NFB transcription factor and that this transcription factor mediates effects of Il-1 on nerve growth factor mRNA expression. Elucidation of the processes by which cytokines activate astrocytes and influence trophic factor expression may provide insight into mechanisms governing inflammatory processes within the central nervous system.
Astrocytes in the adult brain are normally quiescent; however, proliferation and activation of these cells occur in response to a number of stimuli, such as cytokines, associated with inflammation and disease. One of the many functions subserved by astrocytes may involve a role in mediating immunological events in the brain. Reactive astrocytes can express MHC antigens (1), cytokines such as Il-1, 1 TNF␣, Il-6 (2), and colony-stimulating factors (3,4). Thus, the stimulation of astrocytes by inflammatory molecules may initiate a chain of events leading to increased production of cytokines and growth factors and the ability of astrocytes to mediate immune responses in the brain (5).
Cytokine activation of astrocytes also induces elevated levels of nerve growth factor (NGF) expression (6,7,8). NGF is a member of a family of neurotrophic factors, called neurotrophins, which includes brain-derived neurotrophic factor (BDNF), neurotrophin-3 (NT-3), and neurotrophin-4/5 (NT-4/ 5). Each of these factors is expressed in distinct cellular populations in the brain, both neuronal and glial, and influences specific responsive target cells. NGF is most highly expressed in the hippocampus (9 -12). Under quiescent conditions, expression of NGF in the rat hippocampus is primarily detected in neurons (13,12). However, when activated by dissociation in tissue culture (6,7), or by a lesion in vivo (14), astrocytes express high levels of NGF mRNA and protein. The induction of glial NGF expression by lesions in the peripheral nervous system was shown to be dependent on interleukin-1 (15). This cytokine was subsequently demonstrated to influence NGF expression in CNS glial cells as well (6 -8).
The transduction mechanisms by which Il-1 influences NGF expression have remained largely undefined. The signaling pathway does not appear to involve cAMP, protein kinase C, phosphatidylinositol turnover, or mobilization of intracellular calcium (16). Activation of c-fos (16), and the presence of an AP-1 site in the NGF promoter (17,18) suggested a potential role for this transcription factor in mediating Il-1 stimulation of NGF. Alternatively, the recent identification of the sphingomyelin pathway, leading ultimately to activation of NFB transcriptional activity by Il-1 and TNF (19,20), suggested that this pathway might mediate actions of Il-1 in astrocytes.
Cytokine stimulation of gene expression in a variety of cell types is mediated, at least in part, through activation of NFB transcription factors (21). This has been particularly well studied in cells of the immune system, where the NFB/rel transcription factor family plays a major role in lymphoid differentiation and activation (reviewed in Refs. 22 and 23). Transcriptionally active protein dimers in the nucleus activate gene transcription by binding to a B sequence motif in the promoter of responsive genes. Inactive proteins are retained in the cytoplasm as unprocessed precursor proteins (p100 or p105) or as part of an inactive complex containing inhibitor molecules such as the IB's. Upon stimulation, the inhibitory molecules are degraded by the ubiquitin-proteosome pathway, freeing the NFB dimers to translocate to the nucleus and activate transcription. The transcriptionally active NFB dimer may consist of hetero-and homodimers of the five NFB/rel proteins or may include novel tissue-specific proteins that recognize B binding sequences (24). Multiple combinations may exist even within a single cell type. Since numerous cellular genes are transcriptionally regulated by NFB, the specificity of target gene activation by a given stimulus in a particular cell may be determined in part by the dimer composition. Genes activated by the NFB/rel transcription factors include those encoding cytokines (Il-1, Il-2, Il-6, Il-8, TNF␣, and ␤-interferon), histocompatibility antigens, and other cellular proteins (e.g. macrophage NO synthase (25)). NFB transcription factors have been shown to play an important role in mediating lymphocyte and macrophage cellular responses to a variety of cytokines (22,23). Recently, Il-1 has also been shown to induce NFB in astrocytes (2) and astrocytoma cell lines (26). In these studies we have examined whether the NFB transcription factors play a role in mediating astrocyte activation and nerve growth factor induction by cytokines.

EXPERIMENTAL PROCEDURES
Hippocampal Astrocytes-Hippocampi from embryonic day 21 (E21) rats were dissected under sterile conditions, dissociated by trituration and plated on polylysine-coated 75-cm 2 flasks. The cells were grown for 1 week in Eagle's minimal essential medium supplemented with glucose (6 mg/ml), penicillin-streptomycin (0.5 unit/ml-0.5 g/ml), and 15% fetal calf serum to enhance astrocyte proliferation. Confluent type 1 astrocytes were then separated from other cell types by their differential adhesive properties according to previously published protocols (27,28). Cultures were subjected to several shaking steps on an orbital shaker to eliminate neurons, microglia, oligodendrocytes, and type 2 astrocytes. Cells were then maintained for 3 days in the presence of cytosine arabinoside (0.1 mM) to prevent proliferation of any remaining contaminating populations. Confluent astrocytes were then trypsinized and replated on polylysine-coated 150-mm dishes. When the replated astrocytes were 80 -90% confluent, cultures were stimulated by treating with Il-1␤ (Boehringer Mannheim, 10 units/ml) for 4 h. Cells were then lysed with guanidinium isothiocyanate for RNA preparation or harvested for nuclear extract preparation or Western blot analysis.
For NFB competition experiments, double-stranded, phosphorothioate oligonucleotides were synthesized on an Applied Biosystems 392 DNA synthesizer with the following sequences: wild type (5Ј-TCA GGG ACT TTC CGC TGG GGA CTT TCC-3Ј) and mutant (5Ј-TCA CTC ACT TTC CGC TGC TCA CTT TCC-3Ј). Each oligonucleotide was annealed to the complementary DNA sequence and was added to astrocyte cultures at a concentration of 7.5 M 0.5 h prior to Il-1 treatment. For p50-p65 antisense experiments the sequences used were as follows: p50 antisense, 5Ј-GGG ATC ATC GTC TGC CAT GGT-3Ј; p65 antisense, 5Ј-GAG GGG AAA CAG ATC GTC CAT GGT-3Ј; p50 sense, 5Ј-ACC ATG GCA GAC GAT GAT CCC-3Ј; and p65 sense, 5Ј-ACC ATG GAC GAT CTG TTT CCC CTC-3Ј. Oligonucleotides were purchased from Oncor linked to the Penetratin peptide to facilitate entry into cells (29). Combinations of p50-p65 antisense or p50-p65 sense oligonucleotides were provided to astrocyte cultures overnight at a concentration of 250 nM.
Electrophoretic Mobility Shift Assays-Nuclear protein was prepared using the extraction method of Dignam et al. (59). 5 g of nuclear protein was incubated with 32 P-labeled oligonucleotides (10,000 -20,000 cpm) containing the sequence of the NFB binding sites present in the HIV long terminal repeat (LTR) (33). Competition experiments were performed by inclusion of a 50-fold molar excess of unlabeled specific or nonspecific competitor DNA in the binding reaction. Supershift assays included specific antibodies (Santa Cruz Biotechnology, Inc.) directed against each member of the NFB/rel family. DNA-protein complexes were resolved by electrophoresis through a 4.5% polyacrylamide gel under nonreducing conditions.
Immunocytochemistry-Astrocytes were grown as described above except that after trypsinization cells were plated onto polylysine-coated 12-well plates. Cells were fixed with 4% paraformaldehyde and washed with PBS. Antisera to p50 and p65 (Santa Cruz Biotechnology, Inc.) were diluted 1:1000 in PBS with 0.3% Triton X-100. Cells were preincubated for 10 min in PBS, 0.3% Triton with 5% goat serum and then exposed to the antisera overnight. Labeled cells were visualized using the avidin-biotin method for peroxidase staining (ABC kit, Vector Laboratories).
Western Blot Analysis-Cultured astrocytes were harvested in PBS, heated in sample buffer consisting of Tris, SDS, and ␤-mercaptoethanol and subjected to electrophoresis through a 10% polyacrylamide gel. Proteins were transferred electrophoretically to polyvinylidene difluoride membrane, exposed to anti-p65 antiserum (Santa Cruz), and detected by chemiluminescence (Boehringer Mannheim). Blots were stripped of antibody in 2% SDS, 100 mM mercaptoethanol, 62.5 mM Tris at 70°C, and re-probed with anti-p50 antiserum (Santa Cruz).

RESULTS
Il-1 Induction of NFB mRNAs in Astrocytes-The mechanisms by which Il-1 signaling is transduced within a cell to influence gene expression remain poorly defined. This cytokine has been shown to act via an AP-1-dependent mechanism to influence a variety of genes in different cell types (34,35). However, in contrast to the AP-1 pathway, recent studies have provided evidence for an Il-1 signaling pathway similar to that described for TNF␣, involving sphingomyelin metabolism and stimulation of a ceramide-dependent kinase leading to activation of the NFB transcription factor (19). To determine the possible role of this pathway in mediating Il-1 actions in astrocytes, we initially examined whether providing the cells with ceramide would induce an increase in NGF mRNA as effectively as Il-1. A membrane-permeable form of ceramide, C2ceramide, was provided to the astrocyte cultures and compared with Il-1 for the ability to induce NGF mRNA. Treatment with ceramide resulted in a strong elevation of NGF mRNA (Fig. 1), consistent with the possibility that a ceramide-dependent pathway may mediate regulation of NGF expression by Il-1.
Il-1 Induction of NGF mRNA Does Not Require Novel Protein Synthesis-To determine whether the Il-1-stimulated increase in NGF expression was a direct effect of cytokine treatment, or whether synthesis of additional proteins was required, protein synthesis was inhibited during Il-1 treatment of astrocyte cultures. The presence of cycloheximide (CHX) in the medium did not prevent the Il-1-induced elevation in NGF expression (Fig.  2), suggesting that the proteins necessary to mediate Il-1 actions were already present in the cell. CHX alone elicited a slight induction of NGF mRNA (Fig. 2), which is characteristic of NFB-dependent genes, since CHX is known to activate nuclear transport of NFB, presumably by inhibiting synthesis of the relatively labile IB molecules (36).
To further assess the involvement of an NFB-mediated pathway, we examined the potential involvement of reactive oxygen intermediates. NFB is activated by oxygen radicals (37), which are commonly produced during the inflammatory process. The production of oxygen radicals may be a common pathway by which many different stimuli activate NFB (37). Treatment of cells with an antioxidant, pyrrolidine dithiocarbamate (PDTC), blocks NFB activation (38). Astrocyte cultures were treated with PDTC prior to stimulation with Il-1, and the antioxidant prevented induction of NGF mRNA (Fig.  3). Induction of NFB-1 and NFB-2 mRNAs, both autoregulated NFB-inducible genes (39,40), was also prevented by the antioxidant (Fig. 3), similar to the effect on NGF. Thus, a correlation was observed between regulation of NGF expression and that of two NFB subunits.
Il-1 Induces Nuclear NFB Translocation and Binding Activity in Hippocampal Astrocytes-To directly address whether Il-1 treatment of astrocytes leads to NFB activation and nuclear translocation, nuclear extracts were prepared from Il-1treated or untreated astrocytes. These extracts were incubated with labeled oligonucleotides containing the B binding sequence from the HIV LTR and were examined by electrophoretic mobility shift assay for NFB binding activity. A spe- FIG. 1. Northern blot analysis of hippocampal cultures. Total RNA was prepared from hippocampal neurons or astrocytes, subjected to electrophoresis, and transferred to nitrocellulose. Astrocytes were treated with either Il-1 (10 units/ml) or with C2-ceramide (10 M) and probed for NGF mRNA.
cific band was detected only in the extracts prepared from cytokine-treated cells, indicating that Il-1 treatment induced a nuclear B binding complex. In competition experiments, this band was eliminated by the presence of excess unlabeled NFB oligonucleotide, but not by the presence of a nonspecific oligonucleotide competitor, demonstrating the specificity of binding to the NFB recognition sequence (Fig. 4A).
Supershift experiments using specific antibodies to each of the NFB/rel proteins were performed to determine the identity of the proteins in the Il-1-induced NFB binding complex. Incubation with antibodies to either NFB-1 (p50) or relA (p65) either elicited a supershifted complex (NFB1) or inhibited binding to the NFB oligonucleotide (relA, Fig. 4B). Antibodies to the other proteins of the NFB/rel family did not alter binding. These data suggest that acute exposure of astrocytes to Il-1 yielded a binding complex containing the classic p50-p65 NFB heterodimer.
To further demonstrate activation of the p50 and p65 NFB proteins, nuclear translocation was examined immunocytochemically (Fig. 5). Untreated or Il-1-treated astrocytes were fixed and stained with antibodies to p50 or p65. Under basal conditions no nuclear staining was observed for either protein.
However, after Il-1 exposure all astrocyte nuclei were labeled with the p65 antibody, and a subset of nuclei showed p50 staining, indicating that a subpopulation of hippocampal astrocytes possessed p50 and p65 in the nucleus.
NFB Mediates NGF Regulation by Il-1 in Astrocytes-To demonstrate directly that regulation of NGF expression was dependent on NFB activation, competing B binding oligonucleotides were added to the astrocytes (41, 42). These doublestranded, phosphorothioate oligonucleotides contained a tandem repeat of the NFB binding sequence from the HIV LTR   FIG. 4. A, electrophoretic mobility shift assay. Nuclear extracts from hippocampal astrocytes (H.A.), which were either untreated or treated for 4 h with Il-1, were incubated with a labeled oligonucleotide probe consisting of the NFB binding sequence from the HIV LTR. The shifted band detected after Il-1 treatment, designated NF-B, was eliminated by competition with a 50-fold molar excess of unlabeled specific B oligonucleotide (S), but not by a nonspecific oligonucleotide (NS). In contrast, the faint band detected in untreated astrocyte nuclear extracts was competed by the nonspecific as well as the specific oligonucleotide and therefore does not represent specific NFB binding. B, supershift assay. Nuclear extracts from Il-1-treated astrocytes were incubated with the labeled B oligonucleotide in the presence of nonspecific (NS) or specific (S) competitor. Antisera against the individual NFB/rel proteins, or Sp1 as a negative control, were included in the incubation, and the resulting complexes were examined by gel electrophoresis. Arrow indicates NFB complex, and the arrowhead indicates supershifted complex with the p50 antiserum.
(see "Experimental Procedures" for sequence). In the presence of these competing oligonucleotides, induction of NGF mRNA by Il-1 was prevented (Fig. 6A). In contrast, the presence of oligonucleotides containing a mutated B binding sequence did not prevent the increase in NGF mRNA elicited by Il-1. To confirm that these oligonucleotides effectively blocked NFBinduced transcriptional activation, regulation of a known NFBdependent gene was examined. Interleukin-6 (Il-6) expression is induced by Il-1 in astrocytes via an NFB-dependent mechanism (4,2). Induction of Il-6 mRNA was severely attenuated in the presence of the wild type, but not the mutated, competing B oligonucleotide (Fig. 6B).
Further evidence for a role for NFB in regulating nerve growth factor expression was provided by the use of antisense oligonucleotides (43). Astrocyte cultures were pretreated overnight with either antisense or sense oligonucleotides to both NFB-1 (p50) and relA (p65). The cells were subsequently treated with Il-1 for 4 h and expression of NGF mRNA was examined. The antisense oligonucleotides prevented induction of NGF mRNA by Il-1, while the sense oligonucleotides did not influence NGF expression (Fig. 7A). Cultures grown and treated in parallel with the sense and antisense oligonucleotides were analyzed by Western blot for the presence of the p50 and p65 proteins. The antisense oligonucleotides completely eliminated p50 from the cultures and decreased the level of p65 (Fig. 7B). These data further indicate that the NFB transcription factor plays an important role in Il-1 regulation of NGF expression.

DISCUSSION
In these studies we have shown that cytokine stimulation of hippocampal astrocytes increased nerve growth factor expression via an NFB-dependent pathway. Il-1 treatment induced nuclear NFB translocation and DNA binding activity. The binding complex consisted of p50/p65 heterodimers as shown by supershift experiments. Several lines of evidence suggested that regulation of NGF expression was correlated with NFB activation. The inability of cycloheximide, as well as the ability of the antioxidant PDTC, to prevent the Il-1-induced elevation of NGF mRNA, were both consistent with an NFB-dependent signaling pathway. Moreover, the slight induction of NGF mRNA by CHX alone may be due to the inhibition of IB synthesis by CHX and is characteristic of NFB-dependent genes. We investigated the possible role of this transcription factor directly by blocking NFB activity, either by competing for B binding or with antisense oligonucleotides to p50 and p65. Inhibition of NFB prevented the Il-1-induced increase in NGF mRNA. Thus, we have shown that a critical transcription factor, known to mediate cytokine actions in a variety of cell types, is activated in brain astrocytes and mediates induction of trophic factors as well as cytokines. The NGF gene has a complex structure and the promoter has not been well characterized. However, the presence of an AP-1 site in the first intron (17,18) has been demonstrated and has implicated this transcription factor in mediating regulation of NGF expression (17,18). Specifically, c-fos seems to be critically involved in the lesion-induced elevation in NGF mRNA in fibroblasts of the  7. A, astrocyte cultures were pretreated overnight with either sense (S) or antisense (AS) p50 and p65 oligonucleotides before treatment with Il-1. Northern blot analysis revealed that the antisense (AS) oligonucleotides blocked Il-1 induction of NGF mRNA, while the sense (S) oligonucleotides did not. This blot is representative of five independent experiments. B, cultures of hippocampal astrocytes grown and treated with antisense or sense oligonucleotides were subjected to Western blot analysis using antibodies to p50 and p65. The p50 protein was completely eliminated by the antisense treatment, while levels of the p65 protein were decreased. sciatic nerve (17). However, the ability of the antioxidant PDTC to prevent the Il-1 induction of NGF in astrocytes does not support an AP-1-dependent pathway. AP-1 is not activated by reactive oxygen intermediates (37), therefore an AP-1-dependent pathway would not be blocked by the antioxidant treatment. Moreover, the activation of transcription factors and regulation of gene expression may differ in distinct cell types and in response to different stimuli. Recently, the presence of an NFB site has been demonstrated in the promoter of the human NGF gene. 2 An analogous NFB site is present in the mouse NGF promoter (GCG data base, locus MUSNGFX, accession number M33683, base pairs 1163-1172). The B sequence from the human NGF promoter has exhibited binding activity in electrophoretic mobility shift assays and shows supershifted bands in the presence of p50 and p65 antibodies. 2 Thus, the regulation of NGF mRNA which we have observed to be mediated by NFB in astrocytes is probably due to an analogous site in the rat NGF promoter.
The NFB/rel family of transcription factors plays a critical role in mediating cellular activation by cytokines. The functional importance of these transcription factors has been well demonstrated in lymphocytes (22,23,44). In the brain, response to injury and disease is not mediated by the immune system as in the periphery. Lymphocytes and macrophages cannot enter the CNS unless the blood-brain barrier is compromised. Specialized glial cell types within the brain appear to have functions analogous to peripheral macrophages and lymphocytes. Microglia are phagocytic cells within the CNS. These cells proliferate and aggregate at the site of a lesion, release a variety of cytokines such as Il-1, and are considered to be the macrophages of the brain. Astrocytes also become activated by injury and can express MHC antigens (4) and produce a variety of cytokines and trophic factors. This ability may provide an important beneficial environment for neurons consequent to a lesion or disease process. The mechanisms governing astrocyte activation by cytokines such as Il-1 remain poorly understood, although Il-6 induction in astrocytes is mediated by NFB (2). Our studies suggest that the NFB transcription factors may critically mediate additional astrocyte responses to cytokines, including, importantly, induction of trophic factors. Thus, NFB may function as a common pathway in the pleiotropic astrocyte response to cytokines in the CNS.
The NFB/rel transcription factors may serve a number of different roles in the CNS. In addition to mediating the effects of cytokines on astrocytes, constitutive nuclear NFB has been identified in neurons (45) and shown to be responsible for the high levels of HIV LTR promoter activity that can be seen in these cells (46,47). Neuronal NFB has been proposed to play a role in normal neuronal gene expression (45), although induction of this transcription factor may also be associated with CNS pathology. The neuronal toxicity of the amyloid ␤ protein was found to correlate with its ability to induce hydrogen peroxide and NFB (48). NFB has also been proposed to play a role in other CNS disorders whose pathogenesis involves damage due to oxygen free radicals. Based on our results and those of others (2), it is tempting to speculate that some of the pathologic changes associated with CNS inflammatory processes, such as astrocyte activation, may also be the result of NFB induction. Although some of these changes, such as NGF induction, may be beneficial in the short term, a prolonged inflammatory process may lead to severe CNS pathology.
The production of trophic factors consequent to a lesion may have a critical impact on surrounding neuronal populations. Lesions of the fimbria-fornix, the pathway connecting the basal forebrain and hippocampus, normally results in extensive cell death in the basal forebrain. Several groups have shown that infusion of NGF after a fimbria-fornix lesion can rescue cholinergic basal forebrain neurons (49 -51) Although fimbria transection results in a local increase in hippocampal NGF mRNA and protein (52,53), the levels of trophic factor that are induced may not be sufficient to prevent the extensive cell death that occurs after a lesion. One possibility is that under conditions of mild trauma, NGF induction in astrocytes may be sufficient to maintain neuronal survival and function; however, upon excessive damage, such as transection of the pathway, additional trophic support is required. Other conditions that compromise the NGF-responsive basal forebrain cholinergic population include progressive diseases such as Alzheimers, in which this population is among the earliest to degenerate, leading to severe memory deficits (54). Animal models using aged rats with memory deficits have shown that NGF is depleted in these animals (55,56) and that NGF treatment improves performance on memory tasks (57). Furthermore, NGF is currently in clinical use for treatment of Alzheimer patients (58). Therefore, availability of trophic factors to responsive neurons when they are compromised by lesion or disease may critically influence continued survival and function. Thus, it is necessary to understand the mechanisms that mediate NGF induction and to be able to exploit these mechanisms to maximize trophic factor production under appropriate circumstances.