Suppression of calcium release from inositol 1,4,5-trisphosphate-sensitive stores mediates the anti-apoptotic function of nuclear factor-kappaB.

The activation of the transcription factor nuclear factor-kappaB (NF-kappaB) by growth factors, cytokines, and cellular stress can prevent apoptosis, but the underlying mechanism is unknown. Here we provide evidence for an action of NF-kappaB on calcium signaling that accounts for its anti-apoptotic function. Embryonic fibroblasts lacking the transactivating subunit of NF-kappaB RelA (p65) exhibit enhanced inositol 1,4,5-trisphosphate (IP(3)) receptor-mediated calcium release and increased sensitivity to apoptosis, which are restored upon re-expression of RelA. The size of the endoplasmic reticulum (ER) calcium pool and the number of IP(3) receptors per cell are decreased in response to stimuli that activate NF-kappaB and are increased when NF-kappaB activity is suppressed. The selective antagonism of IP(3) receptors blocks apoptosis in RelA-deficient cells, whereas activation of NF-kappaB in normal cells leads to decreased levels of the type 1 IP(3) receptor and decreased calcium release. Overexpression of Bcl-2 normalizes ER calcium homeostasis and prevents calcium-mediated apoptosis in RelA-deficient cells. These findings establish an ER calcium channel as a pivotal target for NF-kappaB-mediated cell survival signaling.

Activation of the transcription factor NF-B 1 by a variety of signaling pathways provides protection against apoptotic insults both in vitro and in vivo (1). The kinds of death signals that NF-B counteracts include cytokines such as tumor necrosis factor-␣ (TNF␣) and Fas ligand (2), trophic factor deprivation (3), overactivation of ionotropic glutamate receptors (4), and various types of oxidative stress (5). The gene targets of NF-B that mediate its anti-apoptotic function have not been established, although several candidates have been identified including Bcl-2 family members (6), manganese superoxide dismutase (5), and members of the inhibitor of apoptosis protein family (7). These proteins may block the apoptotic process by stabilizing mitochondrial membranes, decreasing oxyradical levels, and inhibiting caspases. Recent findings suggest that NF-B can also repress pro-apoptotic proteins such as GADD153 (8) and JNK (9,10).
Calcium is a second messenger that mediates cellular responses to various stimuli; examples include cell proliferation, motility, secretion, and neurotransmission (11)(12)(13). Calcium is also a trigger of apoptosis in physiological and pathophysiological processes. For example, increases of intracellular calcium levels mediate Fas ligand induction and killing by cytotoxic T lymphocytes (14), ischemic death of neurons (15), and the deaths of cells induced by cytotoxic chemicals (16). Recently, interest has focused on the role of the endoplasmic reticulum (ER) in cell deaths that occur in a variety of physiological and pathological conditions (17,18). Calcium release from the ER can trigger apoptosis in many different types of cells including fibroblasts, neurons, and tumor cells (18 -20). Moreover, data suggest that pro-apoptotic proteins such as presenilins (21), Bax (22), and caspase-12 (23), and anti-apoptotic proteins such as Bcl-2 (24) may exert their actions by modulating ER calcium release. It was recently reported that activation of inositol 1,4,5-trisphosphate (IP 3 ) receptors regulates NF-B activity (25), suggesting a possible role for NF-B in modulation of cell survival/death decisions under conditions in which ER calcium release plays a role. Here we identify suppression of calcium release through IP 3 receptors as a key mechanism whereby NF-B prevents apoptosis.
Decoy DNA and Antisense Treatments-NF-B decoy DNA was synthesized and prepared as described previously (5). IB␣ antisense and scrambled control DNA oligodeoxynucleotides were synthesized and prepared as described previously (26). Decoy and antisense were added to neuronal cultures in Neurobasal medium. The cell uptake was evaluated by using fluorescence-tagged decoy DNA and confocal microscopy analysis. The efficacy of the decoy DNA and antisense approaches to decrease and increase NF-B binding activity, respectively, was assessed by electrophoretic mobility shift assay.
Cell Cultures  (WT) immortalized mouse embryonic fibroblasts were a gift from H. Nakshatri, D. Baltimore, and A. Hoffmann. Primary MEFs were prepared from E16 WT and p50 knock-out mouse embryos according to standard procedures (27). Fibroblasts were maintained at 37°C (5% CO 2 atmosphere) in Dulbecco's modified Eagle's medium supplemented with 2 mM L-glutamine, 10% heat-inactivated fetal bovine serum, and 1% penicillin-streptomycin mixture. Stable cell lines expressing p65EGFP and overexpressing Bcl-2 were generated by transfecting RelA Ϫ/Ϫ MEFs with 1-2 g of cDNAs using CLONfectin according to the manufacturer's instructions (Clontech). Selection was carried out for 4 weeks in Dulbecco's modified Eagle's medium containing either G418 (pEGFP and p65EGFP) or puromycin (pBabe and pBabe-Bcl2).
Dissociated cell cultures of primary rat cerebral cortical neurons were prepared using methods similar to those described previously (28  Cerebral cortices were removed from embryonic day 18 Harlan Sprague-Dawley rats (Harlan Sprague-Dawley, Inc., Indianapolis, IN). Cells were dissociated by mild trypsinization and trituration and seeded on 22-mm coverslips or 60-mm polyethyleneimine-coated culture dishes containing Eagle's minimal essential medium supplemented with 26 mM NaHCO 3 , 40 mM glucose, 20 mM KCl, 1 mM sodium pyruvate, 10% (v/v) heat-inactivated fetal bovine serum, and 0.001% gentamycin sulfate. After a 3-5-h incubation period to allow for cell attachment, the medium was replaced with Neurobasal medium with B27 supplements (Invitrogen). Experimental treatments were performed on 7-9-day-old neuronal cultures in which ϳ95% of the cells were neurons, and the remaining cells were astrocytes.
Fluorometric Measurements of [Ca 2ϩ ] c -Cytosolic Ca 2ϩ was measured by ratiometric (F 340 /F 380 ) imaging of Fura-2 fluorescence as described previously (29). Briefly, MEFs were loaded for 45 min with 2 M Fura-2/AM in the presence of 0.3% pluronic acid in HBCS (120 mM NaCl, 5.4 KCl, 0.8 mM MgCl 2 , 2 mM CaCl 2 , 15 mM glucose, 0.5% phenol red, and 20 mM Hepes, pH 7.3). [Ca 2ϩ ] c measurements were obtained using a Zeiss AttoFluor system with a ϫ40 oil objective. The system was calibrated using stock solutions containing either no calcium or a saturating concentration of calcium (1 mM) using the formula [ The experiments in the absence of calcium were performed using HBCS lacking CaCl 2 and supplemented with 1 mM EGTA; the ATP solution was also prepared in Ca 2ϩ -free medium. Analysis of [Ca 2ϩ ] er was performed as described previously (30). Briefly, Fura-2FF/AM-loaded fibroblasts were washed with Ca 2ϩfree buffer and permeabilized using a protocol that preserves the functional integrity of the ER Ca 2ϩ stores by incubation for 3 min with 15 ng/ml digitonin in intracellular medium (120 mM KCl, 10 mM NaCl, 1 mM KH 2 PO 4 , 20 mM Tris-Hepes, pH 7.2, and 2 mM ATP). The specificity of compartmentalization of Fura-2FF/AM within the ER was assessed in pilot studies by addition of 10 M IP 3 to the permeabilized cells and detection of a decrement of the Fura-2FF/AM ratio.
Preparation of Microsomal Fractions-Microsomes from WT and RelA Ϫ/Ϫ cells were prepared as described previously (25). Briefly, cells on 150-mm plates were rinsed twice with phosphate-buffered saline, detached, and pelleted by centrifugation at 500 ϫ g for 10 min. The pellet was then resuspended in ice-cold buffer containing 1 mM EDTA, 0.32 M sucrose, 0.1 mM dithiothreitol, 1 mM Hepes, pH 7.4, and protease inhibitors. Cells were homogenized with 40 strokes in a Dounce homogenizer with type A pestle. The cellular debris were removed by centrifugation at 500 ϫ g for 10 min, and supernatants were centrifuged at 20,000 ϫ g for 20 min to pellet intact mitochondria and nuclei. The supernatant was then transferred to a polycarbonate tube and centrifuged at 100,000 ϫ g for 1 h to obtain the microsomal fraction. Microsomes were resuspended in ice-cold buffer, and protein concentration was measured by Bradford assay using bovine serum albumin as the standard. Protein concentration was adjusted to 2 g/l, and aliquots were prepared for Western blot analysis or snap-frozen in liquid nitrogen and stored at Ϫ80°C for sarcoplasmic endoplasmic reticulum calcium ATPase (SERCA) activity determination.
Electrophoretic Mobility Shift Assay and Western Blots-Preparation of cell extracts and electrophoretic mobility shift assays were performed as reported previously (31). For Western blot analysis, solubilized proteins were separated by electrophoresis on a 10% polyacrylamide gel for whole cell extracts (50 g) and on a 4% Tris-glycine gel for microsomes (20 g). After transfer to a nitrocellulose sheet, the membrane was blocked at room temperature for 1 h in 5% non-fat milk in Tris-buffered saline containing Tween 20 and immunoreacted overnight at 4°C with the different primary antibodies. The immunoreacted bands were visualized by incubation with a horseradish peroxidase-conjugated secondary antibody and a chemiluminescence detection kit (Amersham Biosciences). The following antibodies were used: mouse monoclonal anti-NF-B p65 (Sigma), mouse monoclonal anti-Bcl-2 (Transduction Labo-ratories, Lexington, KY), rabbit polyclonal anti-actin (Sigma), rabbit polyclonal anti-SERCA2b (a gift from S. Chan), rabbit polyclonal anticalreticulin (Affinity BioReagents, Golden, CO), mouse monoclonal anti-IP 3 R3 (BD Transduction Laboratories), rabbit polyclonal anti-IP 3 R1 (Affinity BioReagents), and rabbit polyclonal anti-IP 3 R2 (Santa Cruz Biotechnology, Santa Cruz, CA).
Measurements of SERCA Activity-SERCA activity was measured using an enzyme-coupled spectrophotometric assay in which hydrolysis of ATP is coupled with the oxidation of NADH (32). The depletion of NADH was detected by a decrease in absorption at 340 nm using a Beckman DU7500 spectrophotometer. The assay buffer contained 120 mM KCl, 2 mM MgCl 2 , 1 mM ATP, 1.5 mM phosphoenolpyruvate, 1 mm dithiothreitol, 0.45 mm CaCl 2 , 0.5 mM EGTA, 25 mM MOPS/KOH, pH 7.0, 0.32 mM NADH, 10 units/ml pyruvate kinase, 11 units/ml lactate dehydrogenase, and 2 M of the calcium ionophore 4-bromo-A23187. The reaction was started with the addition of 10 g of microsomal membranes and monitored at 30-s intervals for 10 min. Treatment with the SERCA inhibitor thapsigargin was used as negative control.
Analysis of mRNA Encoding IP 3 R1 and GAPDH by Reverse Transcription-PCR-Total RNA was isolated from cells using TRIzol reagent (Invitrogen). Two g of RNA was reverse-transcribed using random primers with SuperScript First-Strand Synthesis System according to the manufacturer's protocol (Invitrogen). PCR amplification of cDNA encoding for IP 3 R1 and GAPDH was performed for 38 and 28 cycles, respectively, of 30 s at 94°C, 30 s at 55°C, and 45 s at 72°C. Gel electrophoresis with PCR products was performed using ethidium bromide-stained 1.2% agarose gels, and PCR products were identified via UV transillumination at their predicted size by using a 100-bp DNA ladder as a molecular weight marker. Images of UV-transilluminated gels were acquired with a FLA 3000 (Fujifilm, Tokyo, Japan). The primer sequences used in this study were as follows: IP 3 R1 forward, GTGGACGTTTCATCTGCAAGC; IP 3 R1 reverse, GCTTTCGTGGAA-TACTCGGTC; GAPDH forward, CCAAAAGGGTCATCATCTCC; and GAPDH reverse, GTAGGCCATGAGGTCCACCA.
Quantification of Cell Survival-Fibroblasts were seeded in 96-well plates, allowed to adhere for 24 h, and then treated with 50 or 100 ng/ml hTNF␣ at 37°C for 24 h. Experiments were also performed in the presence of pharmacological inhibitors of IP 3 R (xestospongin C, 1 M), ryanodine receptors (dantrolene, 50 M) and serine-palmitoyltransferase (ISP-1, 1 M). The inhibitors were added 1 h prior to the addition of hTNF␣. At the end of the incubation period, cell survival was assessed using CellTiter 96® Aqueous Non-Radioactive Cell Proliferation Assay (Promega) following the manufacturer's instructions. The assay measures the conversion of a tetrazolium compound (MTS) by dehydrogenase enzymes found in metabolically active cells. The quantity of soluble formazan product, as measured by the amount of 490 nm absorbance, is directly proportional to the number of living cells in culture. Results were expressed as a percentage compared with control.
Assessment of apoptotic cell death was performed by staining fixed MEFs or neurons with the fluorescent DNA binding dye bisbenzamide (Hoechst stain). Nuclei were visualized and photographed under epifluorescence illumination (340 nm excitation and 510-nm barrier filter) using a ϫ20 oil immersion objective. Cells in which nuclear staining was diffuse were considered viable, and cells in which nuclear staining was condensed and fragmented were considered apoptotic.
Immunocytochemistry-Following experimental treatments, cells were fixed for 30 min in 4% paraformaldehyde in phosphate-buffered saline and then permeabilized by incubation in 0.2% Triton X-100 in phosphate-buffered saline. Cultures were then incubated for 1 h in 5% normal goat serum and stained for indirect immunofluorescence using a rabbit polyclonal anti-IP 3 R1 antibody (Affinity BioReagents) and a biotinylated goat anti-mouse secondary antibody (Molecular Probes). At designated time points, cell extracts were prepared, and electrophoretic mobility shift assays were performed using specific oligonucleotides for NF-B. The autoradiogram shows NF-B binding activity in extracts from cells that had been exposed scope. All images were acquired using the same laser intensity and photodetector gain to allow quantitative comparisons of relative levels of immunoreactivity among cells. The average pixel intensity per cell body was determined using the software provided by the manufacturer (Zeiss, Thornwood, NY).
Lipid Extraction and Ceramide Measurements-Lipid extraction from WT and RelA Ϫ/Ϫ cells and electrospray-tandem mass spectrometry analyses were performed using methods described previously (33). Briefly, equal amounts of cell pellets were homogenized in deionized water, and lipids were extracted by sequential addition of methanol: ammonium acetate and chloroform. After centrifugation, the chloroform phase was collected and used for the lipid analysis. Samples were injected using a Harvard Apparatus pump at 15 l/min into an electrospray ionization Sciex API 3,000 triple stage quadrupole tandem mass spectrometer from Sciex Inc. (Thornhill, Ontario, Canada) operated in a positive mode. Precursor ion scanning or neutral loss scanning of purified standards of C16:0, C18:0, C18:1, C20:0, C24:0, C24:1 ceramides and C16:0-C18:1 phosphatidylethanolamine were used to identify each species. Quantification of ceramides in the samples was accomplished by summing the total mass counts accumulated under each peak after a 3-min injection period.
Statistical Analyses-Data are presented as means Ϯ S.D. (or S.E., where indicated). One-way analysis of variance combined with Fisher's protected least significant difference post hoc test was used for pairwise comparisons.

Lack of RelA Results in Increased Endoplasmic Reticulum
Calcium Release-To investigate the role of NF-B in calcium homeostasis, we monitored cytosolic free calcium levels ([Ca 2ϩ ] c ) in WT MEFs and in MEFs lacking the transactivating subunit of NF-B RelA/p65 (RelA Ϫ/Ϫ ) during exposure to different agonists that alter [Ca 2ϩ ] c (Fig. 1A). To activate the IP 3 calcium signaling pathway, cells were stimulated with the surface receptor agonists bradykinin (34) (Fig. 1B) and ATP (35) (Fig. 1, CϪE). A dramatic enhancement of calcium release from IP 3 -sensitive stores was documented in RelA Ϫ/Ϫ MEFs compared with WT cells. In both WT and RelA Ϫ/Ϫ MEFs we were unable to detect responses elicited by caffeine or ryanodine (data not shown), suggesting a lack of functional ryanodine receptors in these cells. The calcium response to ATP was also significantly enhanced in WT MEFs in which NF-B activity was suppressed by a 24-h pretreatment of the cells with B decoy DNA (Fig. 1C). Treatment with APB, a specific inhibitor of IP 3 receptor-mediated calcium release from the ER (36), completely abolished the [Ca 2ϩ ] c response to ATP in both WT and RelA Ϫ/Ϫ cells (Fig. 1E). Stable transfection of RelA Ϫ/Ϫ MEFs with a fusion protein of green fluorescent protein and RelA (p65EGFP) significantly attenuated the calcium response to ATP, whereas the expression of EGFP alone did not (Fig.   1F), confirming a specific effect of NF-B on calcium signaling.
NF-B Modulates IP 3 -gated, Thapsigargin-sensitive Calcium Stores-To determine whether the sustained elevation in [Ca 2ϩ ] c resulted from calcium influx across the plasma membrane or release from intracellular calcium stores, we stimulated the cells with ATP in calcium-deficient medium. Removal of extracellular calcium resulted in a transient calcium response to ATP, a response that was dramatically enhanced in RelA Ϫ/Ϫ MEFs ( Fig. 2A). Therefore, the enhanced calcium response in cells lacking RelA can be attributed to an effect of NF-B on calcium release from intracellular stores rather than calcium influx. In most cell types IP 3 -gated Ca 2ϩ stores overlap with thapsigargin-sensitive Ca 2ϩ stores (37). We sought to determine whether an increased ER Ca 2ϩ pool storage capacity was responsible for the enhanced [Ca 2ϩ ] c responses of RelAdeficient cells to agonists that trigger calcium release from IP 3 -sensitive stores. We monitored [Ca 2ϩ ] c in WT and RelA Ϫ/Ϫ cells before and during exposure to thapsigargin, an inhibitor of ER Ca 2ϩ -ATPases, in Ca 2ϩ -free medium. Cells lacking RelA displayed a significantly greater rise in [Ca 2ϩ ] c than did WT MEFs (Fig. 2B), suggesting that constitutive NF-B activity reduces the total pool of ER calcium available for release. Notably, the magnitude of the capacitative Ca 2ϩ entry, analyzed by the addition of 3 mM Ca 2ϩ following the return of [Ca 2ϩ ] c to basal levels, was essentially indistinguishable in WT and RelA Ϫ/Ϫ cells (Fig. 2B). This experiment demonstrated that the activity of Ca 2ϩ release-activated Ca 2ϩ channels is not altered in RelA Ϫ/Ϫ cells. The Ca 2ϩ content of the intracellular stores, as assessed by equilibrium loading experiments with 45 Ca 2ϩ , also demonstrated an increased ER calcium pool in cells lacking RelA (data not shown). We further assessed the ER luminal Ca 2ϩ concentration ([Ca 2ϩ ] er ) by monitoring the fluorescence of compartmentalized Fura-2FF/AM in permeabilized cells (Fig. 2, C and D). In agreement with a previous report (30), the basal ER Ca 2ϩ concentration was estimated to be ϳ790 nM in normal cells. In contrast, RelA Ϫ/Ϫ fibroblasts exhibited a higher [Ca 2ϩ ] er of ϳ1700 nM (Fig. 2, C and D). Moreover, when 10 M IP 3 was added to the Fura-2FF/AM permeabilized cells, a greater decrease in [Ca 2ϩ ] er was observed in the RelA Ϫ/Ϫ cells (data not shown).
Considering the significant [Ca 2ϩ ] er difference observed in WT as compared with RelA Ϫ/Ϫ MEFs, we measured the SERCA activity in isolated microsomes. SERCA activity was higher in RelA Ϫ/Ϫ cells compared with WT cells (Fig. 2E), suggesting that the constitutive ER calcium overload observed in RelA Ϫ/Ϫ cells is linked to increased active uptake. We then analyzed the . *, p Ͻ 0.05 compared with control. C, total RNA was isolated from cortical neurons that had been exposed for 16 h to the indicated treatments, and it was subjected to reverse transcription-PCR analysis to determine relative levels of IP 3 R1 mRNA. The two neuronal splice variants of IP 3 R1 were detected as 535-and 410-bp PCR products. The relative band intensities were normalized using GAPDH as an internal control. *, p Ͻ 0.05 compared with control. D, representative confocal laser scanning microscope images showing IP 3 R1 immunoreactivity in a control culture and in a culture that had been exposed to hTNF␣ for 24 h. levels of proteins that reside in the ER that are involved in cellular calcium homeostasis. The steady-state levels of SERCA type 2b, as well as the expression of the calcium-binding protein calreticulin and of IP 3 R type 2, were comparable in the two cell types (Fig. 2F). However, IP 3 R1 and IP 3 R3 levels were 3.2 and 2 times higher, respectively, in RelA Ϫ/Ϫ MEFs compared with WT cells (Fig. 2F). Because Bcl-2 has been reported to be a gene controlled by NF-B (38), and because it was previously shown that overexpression of Bcl-2 reduces the [Ca 2ϩ ] er and that this action of Bcl-2 is part of its anti-apoptotic mechanism (39), we also measured the levels of Bcl-2. Levels of Bcl-2 were similar in WT and RelA Ϫ/Ϫ cells (Fig. 2F), consistent with the previous observation that the levels of bcl-2 mRNA in B cells generated from WT and RelA Ϫ/Ϫ mice are similar (6). Nevertheless, overexpression of Bcl-2 did counteract the [Ca 2ϩ ] c destabilizing effect of RelA deficiency. Indeed, the RelA Ϫ/Ϫ MEF clones overexpressing Bcl-2 exhibited a markedly decreased [Ca 2ϩ ] c response to ATP compared with cells transfected with the empty vector, such that the magnitude of the calcium response in RelA Ϫ/Ϫ MEFs overexpressing Bcl-2 was comparable to the response observed in WT MEFs (Fig. 2F). These data indicate that the lack of NF-B activity as the result of RelA deletion causes a profound alteration of ER calcium signaling that involves increased SERCA activity, ER calcium uptake, and IP 3 -mediated calcium release and that this alteration can be counteracted by Bcl-2.
Inhibition of IP 3 -induced Calcium Release Protects RelA Ϫ/Ϫ Cells from TNF␣-induced Cytotoxicity-In order to link the abnormal ER calcium homeostasis in cells lacking RelA to the cell survival-promoting function of NF-B, we quantified cell death induced by TNF␣. As expected, RelA Ϫ/Ϫ MEFs exhibited increased vulnerability to killing by TNF␣ compared with WT cells (Fig. 3, A and B). When RelA Ϫ/Ϫ MEFs were pretreated for 8 h with TNF␣, their [Ca 2ϩ ] c response to ATP was greatly enhanced compared with RelA Ϫ/Ϫ MEFs not treated with TNF␣ (Fig. 3C). In contrast, pretreatment of WT cells with TNF␣ did not result in an increased [Ca 2ϩ ] c response to ATP; rather, an opposite trend was observed. We next determined the role of ER calcium release in the enhanced sensitivity of RelA Ϫ/Ϫ MEFs to apoptosis by blocking calcium release from IP 3 and ryanodine receptors with xestospongin C (40) and dantrolene (41), respectively. Because these experiments required a 24-h exposure time, we opted for the use of xestospongin C because it has a similar ability to decrease IP 3 -induced calcium release (compare Figs. 1E and 3D), without the cellular toxicity observed with long-term exposures to ABP (data not shown). Xestospongin C completely prevented killing of RelA Ϫ/Ϫ cells by TNF␣ (Fig. 3, A and B), whereas dantrolene did not modify the calcium response (Fig. 3D) or the survival of RelA Ϫ/Ϫ cells (Fig. 3A). Thus, opening of IP 3 R channels is essential for the cell death-promoting effect of RelA deficiency.
In our studies we used human recombinant TNF␣, which in rodent cells activates the p55 TNF␣ receptor and signaling pathways that generate ceramide (42,43); ceramide has been shown to trigger intracellular calcium release through IP 3 receptors (44). It appears that one factor controlling the cell susceptibility to ceramide-induced apoptosis is the [Ca 2ϩ ] er (39). Generation of ceramide requires the production of the precursor sphinganine arising from the condensation of serine and active palmitate through the catalysis of the enzyme serinepalmitoyltransferase. We studied the involvement of ceramide generation in the sensitivity of RelA Ϫ/Ϫ MEFs to TNF␣-induced apoptosis by using the inhibitor of serine-palmitoyltransferase ISP-1 (32). Blockage of de novo ceramide synthesis was very effective in preventing apoptosis (Fig. 3E). Notably, we also found that the activity of neutral sphingomyelinase was about two times higher in RelA Ϫ/Ϫ MEFs compared with WT cells under basal culture conditions (Fig. 3F). In addition, tandem mass spectrometry analysis of lipid extracts from WT and RelA Ϫ/Ϫ cells demonstrated that under basal conditions, RelA Ϫ/Ϫ cells contain a 2-fold higher level of the long chain ceramides C16 and C24 (Fig. 3G). When we analyzed the effect of ISP-1 treatment on the calcium response to ATP, we found a reduced average fold increase in RelA Ϫ/Ϫ cells (Fig. 3H). Szalai et al. (45) recently demonstrated that the Ca 2ϩ sensitivity of the permeability transition pore in mitochondria increases in cells exposed to ceramide. In such "ceramide-primed" cells, IP 3 -induced calcium spikes (which are normally buffered by mitochondrial uptake) cause the opening of permeability transition pores, cytochrome c release, and apoptosis. Our data suggest that inhibition of NF-B activity places cells in a similar apoptosis-prone state with regard to sensitivity to IP 3induced calcium release.
Activation of NF-B Leads to Decreased Expression of IP 3 R Type 1-It was recently reported that activation of TNF␣ receptors can decrease levels of IP 3 receptors as well as intracellular calcium mobilization in lymphoma cells (46). Because our data pointed to a specific effect of NF-B activation on IP 3 receptor channels, we determined whether IP 3 R levels are regulated by NF-B. For these experiments, we employed primary embryonic rat cortical neurons, a cell type with a relatively high level of constitutive NF-B activity (47). NF-B activity was either inhibited by introducing NF-B decoy DNA into the cells or enhanced by antisense-mediated reduction in the levels of IB␣, the inhibitory subunit of NF-B (Fig. 4A). We determined the effects of reducing or increasing NF-B activity on the expression of IP 3 R1, the major IP 3 R isoform in neurons (48) and the one primarily affected by RelA deficiency (Fig. 2F). We found that NF-B decoy DNA increased the amount of IP 3 R1 protein (Fig. 4B) and IP 3 R1 messenger RNA (Fig. 4C) in the neurons, whereas IB␣ antisense had the opposite effect. Correspondingly, exposure of neurons to TNF␣ also resulted in a decrease in the levels of IP 3 R1 protein and mRNA (Fig. 4, D and E); these effects of TNF␣ were prevented in neurons co-treated with NF-B decoy DNA (Fig. 4, D and E), demonstrating a requirement of NF-B activation in modulation of IP 3 R1 by TNF␣. Increases in intracellular calcium levels are closely linked to excitotoxicity (49). We therefore analyzed the effect of NF-B activity on calcium mobilization and sensitivity to apoptosis in response to glutamate. Pretreatment with TNF␣ significantly protected neurons from glutamate-induced apoptosis (Fig. 4F), an effect mimicked by IB␣ antisense treatment and completely prevented by co-treatment with NF-B decoy DNA (Fig. 4F). In agreement with a previous report (46), treatment with TNF␣ resulted in diminished levels of calcium release, an effect abolished by co-treatment with NF-B decoy DNA (Fig. 4G). Notably, neurons treated with NF-B decoy DNA exhibited an enhanced calcium response to glutamate (Fig. 4G), thus substantiating our findings in RelA Ϫ/Ϫ cells. Previous studies have shown that neurons lacking the p50 subunit of NF-B exhibit enhanced elevations of intracellular calcium levels and increased sensitivity to death after exposure to glutamate (4). We found that cultured primary fibroblasts (Fig. 4H), as well as cortical and hippocampal neurons (data not shown), established from p50-deficient mice did not exhibit a decrease in IP 3 R1 levels following treatment with TNF␣, in contrast to cells from WT mice. Moreover, cells lacking p50 exhibited an increased basal level of IP 3 R1 protein, supporting a role for NF-B in determining the basal level of IP 3 R1 (Fig. 4H). DISCUSSION Increasing evidence points to events occurring in the ER as pivotal determinants of cell death and life decisions. For example, ER stress responses such as the unfolded protein response can trigger apoptosis (50), and ER protein chaperones can prevent apoptosis (51). Our data identify the ER (and specifically, ER calcium signaling) as a pivotal subcellular target that mediates the cell survival-promoting function of NF-B. Our analysis of cells lacking RelA and of cells in which NF-B activity was increased or decreased suggests that NF-B prevents cell death by reducing calcium release from IP 3 -sensitive ER stores. The down-regulation of IP 3 R1 expression by NF-B appears to be a pivotal action of NF-B that promotes cell survival by preventing excessive elevations of cytoplasmic calcium levels.
Abnormalities in ER calcium release have been implicated in diseases that involve aberrant apoptosis. For example, mutations in presenilin 1 that cause early-onset Alzheimer disease may act, in part, by enhancing calcium release from ER stores (21). Perturbations in NF-B activity have been documented in studies of Alzheimer disease (52), but a possible link between the calcium signaling defect and altered NF-B activation has not been explored. Beyond establishing a key role for modulation of calcium release from IP 3 -sensitive stores in the cell survival-promoting function of NF-B, our findings suggest that NF-B may modify calcium signaling events involved in various other physiological processes. Intracellular calcium release channels play important roles in regulating cell proliferation (53) and differentiation during embryogenesis (54), in lymphocyte signaling (55), and in modulation of synaptic plasticity in the nervous system (56). It was recently reported that calcium release from IP 3 -sensitive stores could trigger activation of NF-B in cultured neurons (25). NF-B may therefore serve in a dynamic transcription-dependent feedback mechanism to down-regulate ER calcium release responses.
Signaling pathways that affect transcription of the genes encoding IP 3 receptors are being identified. The promoter of IP 3 R1 contains regulatory elements responsive to several different transcription factors including AP-1, estrogen receptors, and NF-B (57). In addition, it has been shown that calciumresponsive signaling pathways can control IP 3 R1 expression. For example, the calcium/calmodulin-dependent protein phosphatase calcineurin enhances IP 3 R1 expression in neurons (58). Our data show that NF-B down-regulates expression of IP 3 R1 in fibroblasts and neurons and further suggest that this action of NF-B on IP 3 R1 expression plays an important role in the anti-apoptotic action of NF-B. Interestingly, another link between mechanisms regulating apoptosis and IP 3 Rs was identified in a study showing that the anti-apoptotic protein Bcl-xL affects calcium homeostasis in lymphoid cells by altering the expression of IP 3 Rs (59). Interactions between calcium signaling pathways and proteins that regulate apoptosis are therefore likely to play important roles in various physiological and pathological conditions that involve apoptosis. Our findings reveal a novel cell survival-promoting feedback pathway in which activation of IP 3 R in the ER stimulates NF-B (25), which, in turn, negatively regulates expression of IP 3 R1.
NF-B is often activated in cells in response to potentially lethal stimuli including exposure to TNF␣, Fas ligand, oxidative stress, and calcium influx (60,61). Although it was initially thought that NF-B plays a role in the cell death process, subsequent studies showed that selective inhibition of NF-B activity exacerbated cell death, whereas enhanced activation of NF-B prevented cell death (5, 26, 62, 63). Indeed, RelA-null mice die during embryonic development as the result of massive hepatocyte apoptosis related to the deregulated TNF␣/ tumor necrosis factor receptor 1 signaling (64,65). Recently, it has become clear that the pro-survival activity of NF-B includes both activation of anti-apoptotic genes and repression of pro-apototic effectors. Thus, during TNF␣-induced cytotoxicity, NF-B activation causes down-modulation of JNK activity, and, accordingly, JNK activity in RelA Ϫ/Ϫ MEFs is persistent and higher than that in WT cells (9,10). Our findings identify a specific mechanism whereby NF-B can stabilize cellular calcium homeostasis and thereby prevent cell death by acting at the level of the ER.
It is notable that in dorsal root ganglion neurons, TNF␣ engages tumor necrosis factor receptor 1, mobilizing Ca 2ϩ from the ER and activating JNK, a mechanism involving sphingolipid signaling (66). Furthermore, RelA Ϫ/Ϫ cells have also been shown to be highly sensitive to ER stressors such as tunicamycin, which blocks glycosylation causing protein accumulation, and the calcium ionophore A23187, which depletes the ER calcium pool (8). It is of considerable interest that NF-B targets several genes that encode proteins that exert their effects on mitochondria and/or the ER including Bcl-2 (38), manganese superoxide dismutase (5) and IP 3 receptors (the present study). The latter proteins share an ability to modulate cellular calcium homeostasis. Bcl-2 can decrease the calcium concentration in the ER by reducing its loading without affecting capacitive calcium entry (67)(68)(69). By decreasing the levels of superoxide and peroxynitrite, manganese superoxide dismutase indirectly stabilizes ER calcium homeostasis, an action that is particularly important in cells subjected to oxidative stress (70,71). The differential modulation of anti-apoptotic (increased transcription) and pro-apoptotic (decreased transcription) proteins by NF-B suggests a coordinated effort of NF-B to stabilize calcium homeostasis through multiple mechanisms.