Tumor Necrosis Factor Induces Bcl-2 and Bcl-x Expression through NFκB Activation in Primary Hippocampal Neurons*

Emerging data indicate that tumor necrosis factor (TNF) exerts a neuroprotective effect in response to brain injury. Here we examined the mechanism of TNF in preventing neuronal death in primary hippocampal neurons. TNF protected neurons against hypoxia- or nitric oxide-induced injury, with an increase in the anti-apoptotic proteins Bcl-2 and Bcl-x as determined by Western blot and reverse transcriptase-polymerase chain reaction analysis. Treatment of neurons with an antisense oligonucleotide to bcl-2 mRNA or that to bcl-x mRNA blocked the up-regulation of Bcl-2 or Bcl-x expression, respectively, and partially inhibited the neuroprotective effect induced by TNF. Moreover, adenovirus-mediated overexpression of Bcl-2 significantly inhibited hypoxia- or nitric oxide-induced neuronal death. To examine the possible involvement of a transcription factor, NFκB, in the regulation of Bcl-2 and Bcl-x expression in TNF-treated neurons, an adenoviral vector capable of expressing a mutated form of IκB was used to infect neurons prior to TNF treatment. Expression of the mutant NFκB completely inhibited NFκB DNA binding activity and inhibited both TNF-induced up-regulation of Bcl-2 and Bcl-x expression and neuroprotective effect. These findings indicate that induction of Bcl-2 and Bcl-x expression through NFκB activation is involved in the neuroprotective action of TNF against hypoxia- or nitric oxide-induced injury.

Acute and chronic neurodegenerative disorders are characterized by increased levels of a variety of inflammatory cytokines. Among them, the role of tumor necrosis factor (TNF) 1 (1) in injury responses has been indicated (1,2). Cell culture studies have shown that TNF can protect cultured neurons against glucose deprivation-induced injury and excitatory amino acid toxicity, by stabilizing intracellular calcium concentration (1). Moreover, experiments using mice deficient in TNF receptors demonstrated that endogenous TNF protects neurons against ischemic and excitotoxic insults, with induction of an antioxidative pathway (2). However, the exact mechanisms by which TNF protects neurons are largely unknown.
Nuclear factor B (NFB) is a ubiquitous transcription factor that is activated by a variety of cytokines including TNF (3,4) and is thought to be a key regulator of genes involved in inflammation, response to infection, and stress. Conventional NFB is a heterodimer that consists of p50 and p65 subunits. The activity of NFB is strictly regulated by one of the IB inhibitory proteins, such as IB␣ or IB␤, which forms a complex with NFB and keeps NFB in the cytoplasm (3,4). When cells receive signals that activate NFB, IBs are phosphorylated and degraded through a ubiquitin/proteasome pathway (5). Coincident with IB degradation, activated NFB translocates to the nucleus and participates in the transactivation of a variety of genes, including genes for inflammatory cytokines, chemokines, and cell adhesion molecules (5).
Recently, dominant negative mutations of IB␣ that are not phosphorylated and therefore not proteolyzed have been reported (6,7). One such protein is a mutant IB␣ that contains serine-to-alanine mutations at amino acids 32 and 36, which inhibit signal-induced phosphorylation and subsequent proteasome-mediated degradation of IB␣. This form of mutant IB has been used to demonstrate that inhibition of NFB induces apoptosis through a variety of cancer agents (8,9).
Several pathways are emerging that depend on external stimuli to signal key molecules in the induction or the prevention of apoptosis. One such pathway is regulated by the expression of the proto-oncogene bcl-2 family, whose members include bcl-2, bcl-x, and bax. Bcl-2 is an intracellular protein that localizes to mitochondria, endoplasmic reticulum, and the nuclear envelope (10,11) and has been shown to enhance cell survival by inhibiting apoptosis under diverse conditions in a variety of cell types (12)(13)(14). The bcl-x gene also functions to regulate cell death. Bcl-x transcripts are alternatively spliced into a long and short form or a form lacking the transmembrane domain (15,16). The long form (bcl-x L ) suppresses cell death, whereas the short form (bcl-x S ) acts directly as a dominant interfering bcl-2 and bcl-x L antagonist, favoring apoptosis (15,16). Bax forms a dimer with Bcl-2 or Bcl-x L and prevents the death repressor activity of these anti-apoptotic proteins (17,18). In the nervous system, enforced expression of Bcl-2 prevents axotomy-induced (19) and ischemia-induced neuronal death (20), and enhanced expression of Bcl-2 and Bcl-x L has been documented in neurons that are destined to survive after ischemic (21) brain injury. On the other hand, Bax expression is increased in degenerating neurons in the ischemic brain (22). However, the mechanism by which expression of the bcl-2 family members is controlled is unknown.
In this report, we found increased expression of Bcl-2 and Bcl-x upon stimulating primary neurons with TNF. Furthermore, inhibition of NFB activation by adenovirus-mediated overexpression of the mutant form of IB abolished the upregulation of Bcl-2 and Bcl-x expression and the neuroprotective effect induced by TNF.

EXPERIMENTAL PROCEDURES
Cell Culture-Cultures of rat primary hippocampal neurons were prepared from embryonic day 18 Sprague-Dawley rat embryos as described previously (23) with slight modifications. Fetal hippocampi were dissected and digested with calcium/magnesium-free Hank's balanced salt solution (Life Technologies, Inc., Scotland) containing 0.25% trypsin for 20 min at room temperature. Tissues were dissociated by repeated trituration. The cells were seeded at a density of 1 ϫ 10 6 cells/cm (2) on poly-L-lysine (10 g/ml)-coated plates (Falcon, Lincoln Park, NJ) and maintained in growth medium at 37°C in a humidified atmosphere of 5% CO 2 and 95% room air. The growth medium consisted of Dulbecco's modified Eagle's medium supplemented with 10% inactivated fetal calf serum, 30 mM glucose, and 0.5% (v/v) penicillin/streptomycin. To prevent growth of glial cells, cytosine arabinoside (10 M) was added to the cultures 48 h after seeding. All experiments were performed in 8 -10-day-old cultures.
Experimental Treatments-The growth medium was replaced with serum-deprived medium that consisted of Dulbecco's modified Eagle's medium supplemented with 30 mM glucose and 0.5% (v/v) penicillin/ streptomycin immediately before exposure of cultures to hypoxia or a NO donor, sodium nitroprusside (SNP; Sigma). Hypoxic treatment was performed using an incubator attached to a hypoxia chamber (Coy Laboratory Products, Ann Arbor, MI), which maintained a humidified atmosphere with low oxygen tension (pO 2 ; 12-14 torr) as described previously (24). Human recombinant TNF was purchased from Sigma. Unless otherwise stated, TNF was added to cultures 24 h prior to the onset of hypoxia or SNP treatment. Oligodeoxynucleotides (ODN) were purchased from Yuki Gosei Kogyo Co. (Tokyo, Japan). The sequences of the Bcl-2 and the Bcl-x antisense ODN were 5Ј-TCCCGGCTTGCGC-CAT-3Ј (25) and 5Ј-CTGAGACATTTTTAT-3Ј (26). Three different control ODN were used as follows: the Bcl-2 sense ODN, the Bcl-x sense ODN, and mismatch control ODN, 5Ј-CTGTCGCGCGCTCGACTC-3Ј.
Recombinant Adenovirus-The human Bcl-2 cDNA in pSFFV-neo expression vector, number 3088 (donated by Dr. Korsmeyer, Washington University), was digested with EcoRI, obtaining the 1.9-kilobase pair cDNA fragment. The obtained bcl-2 gene was cloned into a cassette cosmid, pAxCALNLw (donated by Dr. Saito, University of Tokyo) (27). The gene should be kept silent because of the stuffer of the neo-resistant gene and is expected to be activated by Cre-mediated excisional deletion of the stuffer when a sufficient amount of Cre recombinase is expressed. The cosmid was then transfected to 293 cells together with the EcoT221-digested adenoviral DNA-terminal protein complex (28) using the calcium phosphate precipitation method. The desired recombinant adenovirus (Ad) generated through homologous recombination, designated as AxCALNLBcl-2, was purified through a CsCl 2 gradient followed by extensive analysis.
The recombinant Ads AxCANCre, which efficiently produces a nuclear localization signal-tagged Cre recombinase under control of the CAG promoter (29,30), and AxCALNLZ, which can express LacZ when a sufficient amount of Cre recombinase is expressed (27), were donated by Dr. Saito (University of Tokyo).
All the viruses described above were grown in 293 cells and purified by CsCl 2 gradient centrifugation. Virus titers were determined by plaque assay, and concentrated virus was stored at Ϫ80°C.
Infection was carried out by adding recombinant Ad(s) to serumcontaining medium. The cells were incubated at 37°C for 60 min with constant agitation. The medium was changed, and the cells were incubated at 37°C for 24 h before treatment with TNF or exposure to hypoxia or SNP, unless otherwise stated.
Assessment of Cell Viability-Neuronal cell viability was assessed by the release of lactate dehydrogenase (LDH) into the culture medium, which indicates loss of membrane integrity and cell death. LDH activity was measured using a commercial kit (Kyokuto Chemical Co., Tokyo, Japan), in which a colorimetric assay measures the pyruvate-mediated conversion of 2,4-dinitrophenylhydrazine into a visible hydrazone precipitate. Percent neuronal viability was expressed as (1 Ϫ experimental value/maximum release) ϫ 100, where the maximum release was obtained after exposure of untreated control cultures to 0.2% Triton X-100 for 15 min at 37°C.
Analysis of DNA Fragmentation in Agarose Gel-Cortical cells (3 ϫ 10 7 ) were lysed in 1 ml of DNA extraction solution containing 20 mM Tris-HCl, pH 7.4, 0.1 M NaCl, 5 mM EDTA, and 0.5% sodium dodecyl sulfate. The lysates were incubated with 100 g/ml proteinase K at 37°C for 16 h. After incubation, 1 ml of phenol/chloroform (1:1, v/v) was mixed well with the cell lysates, which were then centrifuged at 20,000 ϫ g for 10 min. DNA in the aqueous phase was incubated with 5 g/ml DNase-free RNase A at 37°C for 1 h and extracted with phenol/chloroform again, and then with chloroform. DNA was collected by precipitation with 2 volumes of absolute ethanol in the presence of 5 M NaCl. After centrifugation, the DNA pellets were washed with 70% ethanol and air-dried. The DNA was dissolved in 10 mM Tris-HCl and 1 mM EDTA, and its concentration was determined at 260 nm by spectrophotometry. DNA was separated on 1.8% agarose gel containing 1 g/ml ethidium bromide, and DNA fragments were visualized by exposing the gel to UV light.
Western Blot Analysis-Cell extracts for Western blot analysis were prepared by washing the cells three times with PBS and lysing them in sample buffer (50 mM Tris-HCl, pH 8.0, 20 mM EDTA, 1% SDS, and 100 mM NaCl). The samples were boiled for 5 min before subjecting 20 g of them to electrophoresis on 12.5% SDS-PAGE gels. After the proteins were transferred onto polyvinyl difluoride membrane (Millipore Corp., Bedford, MA), the membrane was incubated in blocking buffer (1ϫ PBS, 5% nonfat dried milk) for 1 h at room temperature and then probed with a primary antibody in blocking buffer overnight at 4°C. The membrane was washed four times in PBS containing 0.3% Tween 20, probed with the secondary antibody in blocking buffer for 1 h at room temperature, and washed again in PBS containing 0.05% Tween 20. Detection of signal was performed with an enhanced chemiluminescence detection kit (Amersham Corp., Little Chalfont, UK). The primary antibodies used were anti-rat Bcl-2 monoclonal antibody (MBL, Nagoya, Japan), rat Bcl-x monoclonal antibody (Transduction Laboratories, Inc., Lexington, KY), anti-mouse Bax polyclonal antibody (P-19, Santa Cruz Biotechnology, Santa Cruz, CA), anti-mouse IB polyclonal antibody (Santa Cruz Biotechnology), and anti-mouse ␤-actin monoclonal antibody (Sigma).
RNA Isolation and RT-PCR-Total RNA (derived from 2 ϫ 10 7 cells) was extracted from rat primary neurons by means of the acid guanidinium-thiocyanate phenol chloroform method (32). For RT-PCR, 1 g of total RNA was reverse-transcribed using oligo(dT) (Pharmacia Biotech, Uppsala, Sweden) and Moloney murine leukemia virus reverse transcriptase (Applied Biosystems, Foster City, CA) in a volume of 25 l. For PCR amplification, specific oligonucleotide primer pairs (0.5 M each) were incubated with 5 l of cDNA template in a 50-l PCR reaction mixture containing 1.5 mM MgCl 2 , 25 mM KCl, 10 mM Tris, pH 9.2, 1 l of deoxynucleotides (1 mM each), and 1 unit of Taq polymerase. The sequences of primers used in this study were as follows: Bcl-2 sense primer, 5Ј-CTGGTGGACAACATCGCTCTG-3Ј, and Bcl-2 antisense primer, 5Ј-GGTCTGCTGACCTCACTTGTG-3Ј; Bcl-x sense primer, 5Ј-AGGCTGGCGATGAGTTTGAA-3Ј, and Bcl-x antisense primer, 5Ј-CG-GCTCTCGGCTGCTGCATT-3Ј; Bax sense primer, 5Ј-TGGTTGCCCTT-TTCTACTTTG-3Ј, and Bax antisense primer, 5Ј-GAAGTAGGAAAGG-AGGCCATC-3Ј; ␤-actin sense primer, 5Ј-TGCCCATCTATGAGGGTTA-CG-3Ј, and ␤-actin antisense primer, 5Ј-TAGAAGCATTTGCGGTGCA-CG-3Ј. Dilutions of cDNA were amplified for 26 -28 cycles at 94°C for 30 s, 55°C for 30 s, and 72°C for 30 s. The amplified PCR products were analyzed by 5% polyacrylamide gel electrophoresis and ethidium bromide staining. The product of constitutively expressed ␤-actin mRNA served as the control. All the products were assayed in the linear response range of the RT-PCR amplification process; the cycle number used was determined by finding the midpoint of linear amplification on a sigmoid curve for both amplification products with cycle numbers of 24 -30 plotted against band density. The identity of each PCR product was confirmed by subcloning the amplified cDNAs into the pGEM-T vector (Promega) and sequencing.
Electrophoretic Mobility Shift Assay-Nuclear extracts were prepared from primary neuronal cultures according to published methods (33,34) with some modifications. In brief, cells were harvested by scraping and washing in 0.5 volumes of cold PBS. The cells were then washed once in 0.1 volumes of cold buffer A (10 mM HEPES, pH 7.9, 1.5 mM MgCl 2 , 10 mM KCl, and 0.5 mM dithiothreitol). The washed cell pellets were then suspended in 50 l of buffer A plus 0.1% Nonidet P-40 supplemented with 1 g/ml leupeptin and aprotinin and were incubated on ice for 10 min. After incubation, the pellets were mixed briefly by vortexing and were centrifuged at 10,000 rpm at 4°C for 5 min in a microcentrifuge. The supernatant was carefully removed, and the nuclear pellet was resuspended in 20 l of cold buffer C (20 mM HEPES, 25% glycerol, 0.42 M NaCl, 1.5 mM MgCl 2 , 0.2 mM EDTA, 0.5 mM dithiothreitol, and 0.5 mM phenylmethylsulfonyl fluoride) containing 1 g/ml leupeptin and aprotinin and incubated on ice for 15 min with intermittent vortexing. The extracts were then centrifuged at 10,000 rpm at 4°C for 10 min, and the supernatant was divided into aliquots and frozen at Ϫ70°C. Protein concentrations were determined using the Bio-Rad protein assay kit.
A double-stranded oligonucleotide containing the sequence corresponding to the classical NFB consensus site (5Ј-AGTTGAGGG-GACTTTCCCAGCC-3Ј, Santa Cruz Biotechnology) was end-labeled with [␥-32 P]ATP using T4 kinase (Life Technology, Inc.). Unincorporated nucleotides were removed using two Sephadex G-50 columns (Amersham Pharmacia Biotech). Binding reactions were carried out in a final volume of 25 l consisting of 10 mM HEPES, pH 7.9, 4 mM Tris-HCl, 60 mM KCl, 1 mM EDTA, 1 mM dithiothreitol, 10% glycerol, 1.5 mg/ml bovine serum albumin, 2 g of poly(dI-dC), 2-10 g of nuclear extract, and 0.5 ng of 32 P-labeled oligonucleotide probe (50,000 cpm). Reactions were incubated for 20 min at room temperature. Binding reactions were subjected to nondenaturing polyacrylamide electrophoresis through 4% gels in a 1ϫ Tris borate-EDTA buffer system. Gels were dried and subjected to autoradiography.
Statistical Analysis-Results are presented as mean Ϯ S.E. Experimental groups were compared by one-way or two-way ANOVA, followed by Scheffe's post hoc test. p values less than 0.05 were considered significant.

TNF Protects Neurons against Hypoxia or Nitric Oxide-
Previous studies have shown that cultured neurons are vulnerable to killing by hypoxia or nitric oxide (NO), with features of apoptosis (35,36). Exposure for 24 h to hypoxia or 50 M sodium nitroprusside (SNP) resulted in neuronal survival of 30 -50% (Fig. 1), in agreement with previous reports (35,36). To examine the effect of TNF, neurons were pretreated with increasing concentrations of TNF for 24 h and then exposed for 24 h to hypoxia or 50 M SNP. Pretreatment with TNF dosedependently increased survival of neurons exposed to hypoxia or SNP; 100 ng/ml TNF rescued 54 or 50% of cells from death induced by hypoxia or SNP, respectively (Fig. 1A). Consistent with these results, pretreatment with TNF (100 ng/ml) attenuated DNA laddering induced by hypoxia or SNP (Fig. 1B).
TNF Induces Expression of Bcl-2 and Bcl-x-We previously characterized neuronal death induced by hypoxia and nitric oxide and demonstrated decreased Bcl-2 protein levels in neurons accompanied by neuronal apoptosis (35,36). In addition, Furukawa et al. (25) demonstrated that cycloheximide exerted a neuroprotective action by inducing Bcl-2 expression. There-FIG. 1. TNF protects cultured hippocampal neurons against hypoxia and SNP. A, cultured neurons were preincubated for 24 h with vehicle (saline) or the indicated concentrations of TNF and then were exposed to either normoxia (control), hypoxia, or 50 M SNP. Neuronal survival was examined 24 h later by LDH assay. Data are mean Ϯ S.E. for three independent experiments performed in triplicate. *p Ͻ 0.05 versus TNF ϭ 0 value; two-way ANOVA followed by Scheffe's post hoc test. B, agarose gel electrophoresis of oligonucleosomal DNA fragments (DNA laddering). Neurons pretreated with either vehicle or TNF (100 ng/ml) were exposed to hypoxia or 50 M SNP for 24 h, and then DNA extracted from neuron cultures was subjected to conventional agarose gel electrophoresis. MK, 1-kilobase pair ladder molecular weight marker. fore, we examined whether the neuroprotective effect of TNF involves induction of Bcl-2 and related proteins, Bcl-x and Bax. Western blot analysis showed that TNF induced a concentrationdependent increase in Bcl-2 and Bcl-x protein levels, whereas the expression of Bax protein was not induced by TNF ( Fig.  2A). Bcl-x immunoreactivity was evident as a single band of 26 kDa, the predicted size of Bcl-x L protein, and the antibody did not detect bands at the corresponding size of Bcl-x S or Bcl-x bo protein in any sample tested. RT-PCR analysis revealed increased expression of bcl-2 and bcl-x mRNA upon stimulation with TNF in a dose-dependent manner, whereas expression of bax mRNA was unchanged (Fig. 2B). As for the isoform of bcl-x in rat hippocampal neurons, only a fragment 337 base pairs in size, corresponding to the bcl-x L isoform, was obtained by PCR at 30 cycles in the presence or absence of TNF stimulation. A fragment 150 base pairs in size, corresponding to bcl-x S , was not detected (data not shown). The time course of changes in Bcl-2, Bcl-x, and Bax protein levels after exposure to 100 ng/ml TNF is shown in Fig. 2C. Both the Bcl-2 and Bcl-x protein levels increased within 6 h, peaked by 18 h, and remained elevated at 24 h. The Bax protein level was unchanged by 100 ng/ml TNF up to 24 h of examination (Fig. 2C).
Antisense Oligonucleotides against bcl-2 and/or bcl-x mRNA Inhibit Neuroprotective Effect of TNF-To explore the role of TNF-induced up-regulation of Bcl-2 and Bcl-x expression in neuronal survival, antisense ODN for Bcl-2 and Bcl-x was employed. We showed that treatment of neurons with 10 M Bcl-2 or Bcl-x antisense ODN for 24 h in the presence of TNF almost completely blocked TNF-induced up-regulation of Bcl-2 or Bcl-x expression, respectively, as determined by Western blot analysis (Fig. 3A). Addition of both Bcl-2 and Bcl-x antisense ODN at a concentration of 10 M inhibited the up-regulation of both Bcl-2 and Bcl-x expression induced by TNF (Fig.  3A). Then the effect of Bcl-2 and/or Bcl-x antisense ODN(s) or mismatch control ODN on cell survival in TNF-treated neurons exposed for 24 h to hypoxia or SNP was examined. The Bcl-2 or Bcl-x antisense ODN (10 M) significantly reduced the neuroprotective effect of TNF-␣ in cells exposed for 24 h to hypoxia or 50 M SNP (Fig. 3B). In addition, treatment with both the Bcl-2 and Bcl-x antisense ODN more effectively inhibited TNF-induced neuroprotection than either of the antisense ODN alone (Fig. 3B). Administration of mismatch control ODN (Fig. 3, A and B) or Bcl-2 or Bcl-x sense ODN (data not shown) at 10 M, which did not affect the up-regulation of Bcl-2 and Bcl-x expression induced by TNF, had no effect on the neuroprotective effect of this cytokine.
Adenovirus-mediated Overexpression of Bcl-2 Protects Neurons against Hypoxia or Nitric Oxide-Adenoviral vectors capable of expressing Bcl-2 in hippocampal neurons were obtained to analyze further the effect of Bcl-2 expression on hypoxia-or NO-mediated neurotoxicity. We attempted to create a Bcl-2-expressing adenoviral vector under the control of CAG promoter; however, we were unable to obtain such a vector, perhaps because Bcl-2 expressed in 293 cells after homologous recombination might inhibit cell death of 293 cells and thereby suppress proliferation of the virus. Therefore, Ax-CALNLBcl-2, which is expected to express Bcl-2 only when a sufficient amount of Cre recombinase is expressed, was generated. Primary hippocampal neurons were infected with Ax-CANCre and AxCALNLBcl-2 or AxCALNLZ, a control virus carrying ␤-galactosidase instead of bcl-2, and whole cell extracts were prepared for Western blot analysis. As shown in and AxCALNLZ (10 m.o.i.) did not show increased protein levels. Twenty-four hours after viral infection, the cultures were exposed to hypoxia or 50 M SNP for 24 h, and then neuronal viability was determined. Infection with AxCANCre and AxCALNLZ did not affect cell viability in neurons exposed to either normoxia, hypoxia, or SNP (Fig. 4B). In contrast,

FIG. 2. TNF induces expression of
Bcl-2 and Bcl-x but not Bax. Hippocampal cultures were exposed for 24 h to the indicated concentrations of TNF (A and C) or to 100 ng/ml TNF for the indicated times (B). A and B, extracts of neurons were subjected to SDS-PAGE, and the blots were probed with antibodies to Bcl-2, Bcl-x, Bax or ␤-actin. Visualization of the proteins was performed with ECL. Arrow indicates Bax immunoreactivity at the predicted molecular size. C, total RNA was isolated, and bcl-2, bcl-x, bax and ␤-actin mRNA were amplified by RT-PCR. Resulting PCR products were separated by 5% polyacrylamide gel electrophoresis and visualized by ethidium bromide staining. Results were confirmed in three repeated assays.
neurons infected with AxCANCre and AxCALNLBcl-2 exhibited a 47% reduction of hypoxia-induced cell death and a 38% reduction of cell death induced by SNP, although it did not affect the survival of normoxic neurons (Fig. 4B).
Assessment of Recombinant Adenovirus Expressing IB Super Repressor-The super repressor IB contains serine-alanine mutations in residues 32 and 36, which inhibit its phosphorylation and proteosome-mediated degradation (8,9). An HA-tagged version of this B S32A/S36A driven by the cytomegalovirus promoter enhancer into the replication-deficient adenovirus Ad5 (Ad5IB) was generated by Iimuro et al. (31). Primary cortical neurons were infected with Ad5IB for 24 h at various m.o.i., and whole cell extracts were prepared for Western blot analysis. HA-tagged IB was identified by its slower mobility compared with endogenous IB by Western blotting using an anti-IB antibody, documenting the expression of this cDNA by the adenoviral vector (Fig. 5A).
In order to demonstrate that expression of mutant IB effectively inhibited NFB DNA binding activity in TNF-treated neurons, electrophoretic mobility shift assays were performed on crude nuclear extracts derived from cells that were infected with Ad5IB or control adenovirus Ad5LacZ at m.o.i. of 20 and then 24 h later were treated with TNF for 1 h. Inhibition of NFB DNA binding activity was demonstrated with Ad5IB, but not with Ad5LacZ (Fig. 5B).
These results indicate that expression of the mutant IB protein in neurons potently inhibits TNF-induced NFB DNA binding activity.

IB Super Repressor Inhibits TNF-␣-induced Up-regulation of Bcl-2 and Bcl-x Expression and Neuroprotective Effect-In
order to determine whether NFB activation involves TNFinduced up-regulation of Bcl-2 and Bcl-x expression, we examined the expression of these anti-apoptotic proteins in neurons expressing the mutant IB in response to treatment with TNF. Neurons that were infected for 24 h with Ad5IB or Ad5LacZ at m.o.i. of 20 were treated with 50 ng/ml TNF-␣ for 24 h. Cells were then analyzed for Bcl-2 expression by Western blot. Western blot analysis showed that in neurons expressing mutant IB, TNF-induced up-regulation of Bcl-2 and Bcl-x protein levels was inhibited to a similar level as observed in control untreated neurons (Fig. 5C). In neurons infected with Ad5LacZ, however, TNF up-regulated Bcl-2 and Bcl-x protein levels to a similar level as that in control TNF-treated cells (Fig. 5C).
Next, the effect of mutant IB expression on TNF-induced neuronal survival in comparison with the effect of simultane- ous treatment with Bcl-2 and Bcl-x antisense ODN was examined. Infection with Ad5IB, but not with Ad5LacZ, almost completely inhibited the neuroprotective effect of TNF against hypoxia-and NO-induced toxicity (Fig. 5D). The inhibition of TNF-induced neuroprotective effect by treatment with Ad5IB was significantly more potent than that with Bcl-2 and Bcl-x antisense ODN (p Ͻ 0.05) (Fig. 5D). DISCUSSION Previous studies have demonstrated that overexpression of Bcl-2 in neurons can prevent or delay cell death induced by a variety of stimuli in vitro (37,38) and in vivo (19,20). Overexpression of Bcl-x L also protects sympathetic neurons from apoptosis induced by nerve growth factor deprivation (39). In addition, we previously showed decreased levels of Bcl-2 expression in hypoxia-and nitric oxide-induced neuronal apoptosis (35,36). Furthermore, Bcl-2 and Bcl-x are induced primarily in neurons destined to survive following focal ischemia (22) and global ischemia (21). Thus, it is likely that the endogenous levels of Bcl-2 and Bcl-x regulate neuronal survival. Our present data demonstrated that TNF protects neurons against hypoxia-and nitric oxide-induced toxicity and up-regulates Bcl-2 and Bcl-x expression. Pretreatment of neurons with Bcl-2 or Bcl-x antisense ODN which inhibits up-regulation of Bcl-2 or Bcl-x expression, respectively, significantly inhibits the neuroprotective effect induced by TNF, indicating that either Bcl-2 and Bcl-x induction is involved in the neuroprotective action of this cytokine. Simultaneous addition of both the antisense ODN, which inhibits up-regulation of Bcl-2 and Bcl-x expression, seems to additively but not synergistically inhibit TNFinduced neuroprotection. Evidence taken from previous studies suggests that Bax actively promotes cell death, unless it is bound by either Bcl-2 or Bcl-x L (17,18). Thus, the ratio of Bax to Bcl-2 and Bcl-x seems to be critical for cell survival. Therefore, TNF-induced expression of Bcl-2 and Bcl-x might antagonize the pro-apoptotic function of Bax, whose expression was unchanged by TNF. Moreover, the present finding that overexpression of Bcl-2 protects neurons supports this anti-apoptotic protein as a critical factor for neuronal protection against hypoxia-or nitric oxide-induced injury. Production of TNF has been investigated in an ischemic brain injury model (40), in which the cytokine is rapidly synthesized in the brain. Enhanced expression of Bcl-2 and Bcl-x has also been documented in neurons that are destined to survive after ischemic brain injury (21). Therefore, it might be that TNF produced by resident brain cells up-regulates Bcl-2 and Bcl-x expression in neurons and thereby protects them from ischemic injury.
Most cells including neurons express specific receptors for TNF. Two receptors have been characterized and designated TNF receptor 1 (TNFR1) and TNF receptor 2 (TNFR2). In the present study, human TNF, which activates TNFR1 but not TNFR2 in rodent cells (41), induced Bcl-2 and Bcl-x expression and protected hippocampal neurons against hypoxia-and NOinduced cytotoxicity. TNFR1 is linked to a signal transduction pathway involving a sphingomyelinase that induces release of ceramide, which then results in activation of the transcription factor NFB (42). We here demonstrated that TNF-␣ induced NFB activity in cultured hippocampal neurons as determined by gel shift assay, in agreement with a previous report by Mattson et al. (43). Our experiments using Ad5IB showed that inhibition of NFB activation completely abolished the neuro-  protective effect and inhibited the induction of Bcl-2 and Bcl-x expression in TNF-treated cells. These results indicate that the neuroprotective effect and up-regulation of Bcl-2 and Bcl-x expression induced by TNF are mediated by activation of NFB.
The relation between NFB and apoptosis was elucidated recently by studies demonstrating that inhibition of NFB, either by the IB super repressor (8,9) or in Rel A (p65) knock-out cells (44), results in increased apoptosis. Contrary to these reports, it has been shown that NFB activation induces apoptosis in certain cells (45,46) including bone marrow cells and leukemia cells. It is not known at present if NFB activation is involved in promoting apoptosis in some cells and if it is responsible for preventing apoptosis in other cells. Regarding neuronal cells, it has been reported that activation of NFB protected hippocampal neurons against oxidative stress-induced apoptosis, using B decoy DNA approach (43), and that inhibition of NFB in PC12 cells by exposure to B decoy DNA or pyrrolidinedithiocarbamate induced apoptosis that could not be prevented by nerve growth factor (47). These findings, together with our present observations, indicate that the NFB signaling cascade is closely involved in prevention of apoptosis in neuronal cells.
Based on our current understanding of the roles of NFB activation in the regulation of apoptosis, gene expression of proteins that exert anti-apoptotic effects may be involved. We demonstrated here that up-regulation of Bcl-2 and Bcl-x expression induced by TNF is completely inhibited by IB super repressor, indicating that Bcl-2 and Bcl-x are among the gene targets of NFB. The bcl-2 and bcl-x genes have complex structures, and the promoters have not been well characterized. Recently, the presence of NFB site has been demonstrated in the promoters of human bcl-2 and bcl-x (48). Therefore, the regulation of Bcl-2 and Bcl-x expression that we have observed FIG. 5. Expression of IBa S32A/S36A inhibits TNF-induced up-regulation of Bcl-2 expression and the protective effect of TNF against neurotoxicity induced by hypoxia or SNP. A, extracts from neuronal cultures 24 h after infection with Ad5IB at various m.o.i. were subjected to gel electrophoresis, and the blots were probed with antibodies to IB. HA-tagged IB␣ was identified by its slower mobility compared with endogenous IB␣. B, NFB DNA binding activity was assessed by electrophoretic mobility shift assay using NFB oligonucleotide probe with nuclear extracts prepared from neurons incubated for 1 h with or without 100 ng/ml TNF. Neurons were infected or not infected with Ad5IB or Ad5LacZ at m.o.i. 20 at 24 h before TNF treatment. C, cultures were infected or not infected with Ad5IB or Ad5LacZ (m.o.i. 20) and exposed for 24 h to vehicle or 100 ng/ml TNF. Extracts from the cells were subjected to SDS-PAGE, and the blots were probed with antibodies to Bcl-2, Bcl-x, or ␤-actin. D, at 24 h after infection with Ad5IB or Ad5LacZ (m.o.i. 20), cultures were treated for 24 h with vehicle (saline) or 100 ng/ml TNF in the presence or absence of Bcl-2 and Bcl-x antisense ODN at 10 M each (B2AS ϩ BXAS), and then exposed for 24 h to hypoxia or 50 M SNP. Neuronal survival was then determined by LDH assay. Data are mean Ϯ S.E. for five independent experiments performed in triplicate. *, p Ͻ 0.05 versus vehicle and †, p Ͻ 0.05 versus TNF in each of the three groups (control, hypoxia, and SNP); ANOVA followed by Scheffe's post hoc test. to be mediated by NFB in neurons is probably related to an analogous site in the rat bcl-2 and bcl-x promoter, respectively.
In addition to bcl-2 and bcl-x, expression of other genes may be induced by TNF through NFB activation. Recently, it has been reported that the expression of manganese superoxide dismutase is induced by TNF, which is abolished by B decoy DNA, in hippocampal neurons (43). Moreover, in the present study, the neuroprotective effect of TNF was blocked more potently by IB super repressor than by antisense ODN against bcl-2 and bcl-x, in both of which TNF-induced upregulation of Bcl-2 and Bcl-x expression was completely blocked. This suggests that other proteins that exert a cytoprotective effect may be induced by NFB activation in TNFtreated neurons.
In summary, our findings demonstrated that induction of Bcl-2 and Bcl-x expression through NFB activation is involved in the neuroprotective effect of TNF. In addition, our finding that inhibition of NFB activation by expression of a mutant form of IB more potently abolishes the neuroprotective effect of TNF than inhibition of TNF-induced up-regulation of Bcl-2 and Bcl-x expression by the antisense ODN suggests that other genes responsible for anti-apoptotic activity are induced by NFB activation. Identification of the genes will certainly help to understand the regulation of neuronal survival and to design new therapeutic strategies for acute and chronic neurodegenerative disorders.