The RelA(p65) Subunit of NF- k B Is Essential for Inhibiting Double-stranded RNA-induced Cytotoxicity*

Double-stranded RNA (dsRNA) molecules generated during virus infection can initiate a host antiviral response to limit further infection. Such a response involves induction of antiviral gene expression by the dsRNA-activated protein kinase (PKR) and the NF- k B transcription factor. In addition, dsRNA can also induce apoptosis by an incompletely understood mechanism that may serve to further limit viral replication. Here we demonstrate a novel role for the RelA subunit of NF- k B in inhibiting dsRNA-induced cell death. dsRNA treatment resulted in caspase 3 activation and apoptotic morphological transformations in mouse embryonic fibroblasts (MEFs) derived from RelA 2 / 2 mice but not from RelA 1 / 1 mice. Such dsRNA-induced killing could be inhibited by expression of either a dominant-nega-tive mutant of PKR or wild-type RelA. Interestingly, caspase 3 activated following dsRNA treatment of RelA 2 / 2 MEFs was essential for apoptotic nuclear changes but dispensable for cytotoxicity. A broader specificity caspase inhibitor was also unable to inhibit dsRNA-induced cytotoxicity, suggesting that caspase activation is not essential for the induction of cell death by dsRNA in MEFs. However, combined inhibition of caspase 3 and reactive oxygen species production resulted in complete inhibition of dsRNA-induced cytotoxicity. These results demonstrate an essential role for NF- k B in protecting cells dissociation in an ion trap mass spectrometer (LCQ, Finnigan MAT). A 1- m l aliquot (5%) of the P150 tryptic digest was loaded onto a 100- m m inner diameter, 360- m m outer diameter, 30-cm length of fused silica capillary packed with 15 cm of POROS 10R2 reverse phase beads (Perspective Biosystems). Peptides were eluted with an acetonitrile gradient at a flow rate of 500 nl/min for 15 min. A data-dependent experiment was performed to obtain structural information for selected peptides. Ions with m/z values corresponding to peptides observed by MALDI-TOF MS were monitored in full mass range scans and auto- matically subjected to collision-induced dissociation as each eluted from the capillary column. Peptide masses and selected b and y series frag- ments were used to search an in-house protein and DNA sequence data base with an enhanced version of the FRAGFIT (31) and the SEQUEST program. The mouse caspase 3 was identified by a data base search of data obtained from a liquid chromatography/MS/MS analysis of a tryp- tic digest of the 18-kDa band. MS/MS analysis of MH 1 1118.9 was found to correspond to residues 65–75 (SGTDVDAANLR) of mouse caspase 3. MALDI MS analysis identified an additional eight masses that matched with the caspase 3 protein.

Virus replication within infected cells results in generation of double-stranded RNA (dsRNA) 1 molecules that can trigger host antiviral responses (1). Such dsRNA-activated responses can be mediated by dsRNA-dependent enzymes such as the interferon-inducible protein kinase (PKR) which phosphorylates key cellular substrates (e.g. eukaryotic protein synthesis initiation factor-2␣) (2). PKR can also activate the NF-B transcription factor (see below) resulting in induction of type I interferon gene expression that can prevent further virus infection (3). In addition, dsRNA can also induce apoptosis in a PKR-dependent manner (4 -6). Thus PKR is required for both dsRNA-mediated induction of gene expression and induction of apoptotic cell death. PKR-mediated apoptosis of virus-infected cells by dsRNA may thus limit virus infection by preventing virus replication within host cells.
Apoptosis is a genetically controlled process that plays an essential role in regulating homeostasis and in protecting the host against microbial infections (7,8). Apoptotic cells manifest characteristic morphological changes, such as nuclear condensation and fragmentation, which are mediated by proteases belonging to the caspase family (9,10). Although caspases are normally present in an inactive form, they can be activated by proteolysis triggered by cell death inducers (10,11). Activated caspases cleave key cellular substrates and thus provide a safe and efficient mechanism for eliminating surplus or infected cells. Although inhibitors of caspase proteases have been shown to prevent apoptosis induced by many different agents (10,12,13), recent studies have demonstrated that in certain cell lines, inhibition of caspase activity induces necrotic killing of cells by TNF␣ (14 -16). Such necrosis can occur in cell lines that are normally resistant to TNF␣ killing, is accompanied by increased production of reactive oxygen species (ROSs), and can be prevented by antioxidants (14 -16). Importantly, similar to TNF␣, dsRNA also induces necrosis in the presence of caspase inhibitors in dsRNA-resistant wild-type fibroblasts (16). Both TNF␣ and dsRNA-induced necrosis may represent host strategies for eliminating cells infected with viruses encoding caspase inhibitors (16).
The NF-B family of transcription factors is a key regulator of genes involved in immune and inflammatory responses (17,18). Recent studies have also demonstrated a critical role for NF-B in regulating apoptotic cell death. Mice deficient in the RelA (p65) subunit of NF-B die prenatally because of massive hepatocyte apoptosis (19), which appears related to the cytotoxic effect of TNF␣ (20). Fibroblasts or macrophages derived from RelAϪ/Ϫ mice or cells overexpressing a super-repressor form of the inhibitory IB protein are also highly susceptible to TNF␣-killing (21)(22)(23)(24). These studies have demonstrated an essential role for NF-B in protecting cells from TNF␣-induced killing. In addition, NF-B can also mediate pro-apoptotic effects. A recent report has shown that NF-B is critically important for mediating p53-induced apoptosis (25). Our recent studies have demonstrated an essential role for RelA in induction of the death receptor Fas expression and in subsequent apoptosis after Fas ligation (26). Taken together, these studies suggest that NF-B-mediated anti-apoptosis and pro-apoptosis may be context-dependent.
The NF-B proteins also play a key role in mediating cellular responses to conserved microbial structures such as dsRNA and LPS (17). Stimulation of cells with dsRNA or LPS results in rapid nuclear translocation of NF-B proteins and induction of NF-B target gene expression. Although both dsRNA-mediated NF-B activation and apoptosis require PKR activity, less is known about mechanisms important for regulating dsRNAinduced apoptosis. Recent studies have suggested, however, that PKR-dependent induction of Fas expression may be involved in dsRNA-induced apoptosis (6,27,28). Since NF-B regulates Fas expression, these studies suggest that NF-B may play a pro-apoptotic role in dsRNA-induced cell death.
To determine a possible role for NF-B in dsRNA-induced apoptosis, we have utilized embryonic fibroblasts and macrophages from mice deficient in the RelA subunit of NF-B. We show here that dsRNA-induced Fas expression was dramatically reduced in RelAϪ/Ϫ MEFs. Surprisingly, dsRNA specifically induced apoptosis in NF-B RelAϪ/Ϫ but not in RelAϩ/ϩ MEFs. In addition, inhibition of macromolecule synthesis also rendered RelAϩ/ϩ MEFs susceptible to dsRNA-induced killing. These results suggest that dsRNA-induced apoptosis may be prevented by NF-B mediated induction of anti-apoptotic gene expression, rather than induced by expression of proapoptotic genes, such as fas. These results demonstrate the existence of a novel anti-apoptotic role for NF-B in the dsRNAinduced cell death pathway.

Cells and Materials
Mouse embryonic fibroblasts, fetal liver macrophages, and 3T3 fibroblasts were derived as described previously (21). Fibroblasts were cultured in high glucose Dulbecco's modified Eagle's medium containing L-glutamate (2 mM), penicillin (100 units/ml), streptomycin (100 g/ml), and calf serum (10%). Human TNF␣ was obtained from R & D Systems and used at concentration of 10 ng/ml in all experiments. dsRNA poly(I-C) was purchased from Sigma and used at a final concentration of 100 g/ml. Caspase inhibitors, Z-DEVD-fmk, Z-VAD-fmk, and biot-VAD-fmk, (Enzyme System Products) were dissolved in dimethyl sulfoxide at 20 mM and used at 100 and 1 M, respectively. Actinomycin D and antioxidant butylated hydroxyanisole (BHA) were obtained from Sigma and used at 2 g/ml and 100 M, respectively. The nuclear dye 4Ј,6Ј-diamidino-2-phenylindole (DAPI) and the reactive oxygen species dye dihydrorhodamine 123 (DHR) were obtained from Molecular Probes. pLPC expression vector was a gift from Dr. S. Lowe (Cold Spring Harbor Laboratory, New York). pRelA was constructed by cloning the mouse RelA cDNA into pLPC. pPKRDN, which has a six-amino acid deletion as described previously (29), was a gift from Dr. A. García-Sastre (Mount Sinai Medical Center, New York).

Analysis of Cell Death
Nucleus Morphology-Cells in tissue culture plates were rinsed with PBS, fixed with 3.7% formaldehyde, and permeabilized with 0.2% Triton X-100 for 5 min. They were then washed and incubated with a DAPI labeling solution (2 g/ml in PBS) for 5 min and examined under a fluorescence microscope.
Cell Viability Experiments-Approximately 2 ϫ 10 5 cells were plated on each well of a 6-well plate 1 day before the experiments. The caspase inhibitors, Z-DEVD-fmk or Z-VAD-fmk, or macromolecule synthesis inhibitor, actinomycin D, was added 1 h before the addition of dsRNA. After the indicated periods, the cells were trypsinized (fibroblasts) or scraped (macrophages), and viable cells were counted by trypan blue exclusion. Four independent readings within a single experiment were used to calculate the S.D.
Transfection Experiments-RelAϪ/Ϫ MEFs were cotransfected with a GFP expression vector (0.5 g) and pLPC, pRelA, or pPKRDN (0.5 g) using Fugene 6 (Roche Molecular Biochemicals). 24 h later, cells were either left untreated or treated with dsRNA for 12 h. Viable GFPpositive cells from four randomly chosen fields were counted and used to calculate S.D.

Determination of Reactive Oxygen Species Levels
DHR was added to a final concentration of 2 M before RelAϪ/Ϫ MEFs were treated with the indicated agents. 12 h later, cells were trypsinized, washed and resuspended in PBS before FACS analysis.

EMSA, Northern Blots, Affinity Blots, and Western Blots
Electrophoretic mobility shift assay (EMSA) was carried out as described previously (19). RelA-specific antisera were purchased from Santa Cruz Biotechnology. Northern blotting was carried out as described (26) with probes generated from cDNA fragments by reverse transcriptase-polymerase chain reaction using gene-specific primers. Affinity blotting was performed essentially as described previously (30). Briefly, ϳ5 ϫ 10 6 cells were harvested after the treatments indicated. The cells were then washed once with PBS and pelleted, and the pellet was snap-frozen on dry ice. An equal volume of 1 M biot-VAD-fmk in MDB buffer (50 mM NaCl, 2 mM MgCl 2 , 5 mM EGTA, 10 mM HEPES, 1 mM dithiothreitol (pH 7)) was added to the cell pellet, and the cells were lysed by three cycles of freezing and thawing. The lysates were incubated at 37°C for 15 min and centrifuged. 20 g of protein lysates from the supernatant were separated by SDS-PAGE and transferred to nitrocellulose membrane. The membrane was blocked with TBST (20 mM Tris, 150 mM NaCl, 0.05% Tween 20) supplemented with 2% nonfat dry milk for 30 min and then incubated in avidin-Neutralite (Molecular Probes) at 1 g/ml in TBST supplemented with 1% nonfat dry milk for 1 h. The membrane was then washed and incubated in biotinylated horseradish peroxidase (Molecular Probes) at 25 ng/ml in TBST for 1 h. The labeled protein was visualized by ECL (Amersham Pharmacia Biotech). For Western blotting, membranes were blocked with TBST supplemented with 0.5% casein. Caspase 3-p17 antibody (New England Biolabs), which specifically recognizes the large subunit of activated caspase 3, was used at a 1 to 10,000 dilution in subsequent steps.

Identification of p18
Fifty 10-cm plates of RelAϪ/Ϫ 3T3 fibroblasts grown to 1-2 ϫ 10 7 / plate were either left untreated or treated with TNF␣ (10 ng/ml) for 6 h. Cells were collected by centrifugation, lysed in the presence of 1 M biot-VAD-fmk to label caspases, and the lysates clarified by centrifugation at 100,000 ϫ g for 1 h at 4°C. The protein concentrations of the pooled supernatant were 4.7 mg/ml for untreated and 2.9 mg/ml for TNF␣-treated samples. To remove unbound biot-VAD-fmk, the lysates were dialyzed with four changes of MDB buffer (see above) supplemented with a proteinase/inhibitor mixture (Roche Molecular Biochemicals). Dialyzed extracts were run through an Immunopure Immobilized Streptavidin Column (Pierce). The column was washed with 150 ml of washing buffer (50 mM sodium phosphate, 0.4 M urea, 50 M phenylmethylsulfonyl fluoride (pH 7)), and bound proteins were eluted by boiling in the washing buffer with 2% SDS. The eluted proteins were precipitated with 0.25 volume of trichloroacetic acid solution (100% trichloroacetic acid, 0.4% sodium deoxycholate) and washed twice with acetone. The precipitates were resuspended in 50 l of the SDS-loading buffer. 5 l of either the untreated or TNF␣-treated samples were resolved by SDS-PAGE. The gel was silver-stained to test for the purity and amount of p18 (approximate 10 ng). The rest of the samples were also subjected to SDS-PAGE and transferred to polyvinylidene difluoride membrane (Problott, Applied Biosystem). p18 was excised and wet with 1 l of methanol. The band was reduced and alkylated with isopropyl acetamide followed by digestion in 20 l of 0.05 M ammonium bicarbonate containing 0.5% Zwittergent 3-16 (Calbiochem) with 0.2 g of trypsin (Frozen Promega Modified) at 37°C for 17 h. Peptides generated from in situ tryptic digests were separated on a C18 0.18 ϫ 150-mm capillary column (LC Packing, Inc.). The high pressure liquid chromatography consisted of a prototype capillary gradient high pressure liquid chromatography system (Waters Associates) and a model 783 UV detector equipped with a Z-shaped flow cell (LC Packing, Inc.). A 30-cm length of 0.025-mm ID glass capillary was connected to the outlet of the Z-shaped cell inside the detector housing to minimize the delay volume. Solvent A was 0.1% aqueous trifluoroacetic acid, and solvent B was acetonitrile containing 0.08% trifluoroacetic acid. Peptides were eluted using a linear gradient of 0 -80% solvent B in 60 min and detected at 195 nm. Fractions were collected automatically by a BAI Protocol onto pre-made spots of matrix (0.5 l of 20 mg/ml ␣-cyano-4-hydroxycinammic acid ϩ 5 mg/ml nitrocellulose in 50% acetone, 50% 2-propanol) on the target plate. Ions were formed by matrix-assisted laser desorption/ionization with a nitrogen laser, 337 nm. Spectra were acquired with a Perspective Biosystems Voyager Elite time-of-flight mass spectrometer, operated in reflector delayed extraction mode. Peptides detected by MALDI-TOF MS were subjected to collision-induced dissociation in an ion trap mass spectrometer (LCQ, Finnigan MAT). A 1-l aliquot (5%) of the P150 tryptic digest was loaded onto a 100-m inner diameter, 360-m outer diameter, 30-cm length of fused silica capillary packed with 15 cm of POROS 10R2 reverse phase beads (Perspective Biosystems). Peptides were eluted with an acetonitrile gradient at a flow rate of 500 nl/min for 15 min. A data-dependent experiment was performed to obtain structural information for selected peptides. Ions with m/z values corresponding to peptides observed by MALDI-TOF MS were monitored in full mass range scans and automatically subjected to collision-induced dissociation as each eluted from the capillary column. Peptide masses and selected b and y series fragments were used to search an in-house protein and DNA sequence data base with an enhanced version of the FRAGFIT (31) and the SEQUEST program. The mouse caspase 3 was identified by a data base search of data obtained from a liquid chromatography/MS/MS analysis of a tryptic digest of the 18-kDa band. MS/MS analysis of MH ϩ 1118.9 was found to correspond to residues 65-75 (SGTDVDAANLR) of mouse caspase 3. MALDI MS analysis identified an additional eight masses that matched with the caspase 3 protein.

RESULTS
The RelA Subunit of NF-B Is Required for dsRNA-induced Gene Expression-Our recent studies have revealed a critical role for the RelA subunit of NF-B in the regulation of TNF␣ and LPS-induced Fas expression (26). Previous studies have also suggested that dsRNA-induced apoptosis could be mediated by induction of Fas expression (6,27). We therefore tested the possible involvement of NF-B in dsRNA-dependent induction of Fas expression and apoptosis. To this end, we first determined whether RelA was a component of NF-B complexes activated by dsRNA. RelAϩ/Ϫ mouse embryonic fibroblasts (MEFs) (19) were treated with dsRNA for 2 h, after which nuclear extracts were tested for B-site binding activity by EMSA. As expected, dsRNA strongly activated NF-B (Fig.  1A). Activated NF-B was supershifted by antisera generated against RelA (Fig. 1A), demonstrating the presence of RelA in dsRNA-activated NF-B complexes. Furthermore, as previously observed following TNF␣ treatment (19), dsRNA-treated RelAϪ/Ϫ MEFs showed a significantly lower level of NF-B activation (Fig. 1A, compare lane 2 and lane 5). To determine whether RelA was important for dsRNA-induced Fas expression, we treated RelAϩ/Ϫ or RelAϪ/Ϫ MEFs with dsRNA for 6 h, after which Fas mRNA expression was determined. Similar to TNF␣ and LPS, dsRNA-induced Fas expression was dramatically reduced in RelAϪ/Ϫ cells (Fig. 1B). In addition, dsRNA-mediated induction of the neutrophil-specific chemokine, MIP-2, was also found to be dependent on RelA (Fig. 1B). These results thus demonstrate an important role for the RelA subunit of NF-B in mediating dsRNA-induced gene expression.
RelA Is Essential for Inhibiting dsRNA-induced Cytotoxicity-We wanted to determine whether Fas expression was responsible for controlling susceptibility of MEFs to dsRNA-induced cell death. Surprisingly, a 12-h dsRNA treatment significantly reduced viability of RelAϪ/Ϫ but not RelAϩ/ϩ MEFs ( Fig. 2A). These results demonstrate the existence of a previously unrecognized function of RelA in inhibiting dsRNAinduced cytotoxicity. Recent studies have demonstrated a critical role for RelA in protecting cells from TNF␣-induced killing (21), through induction of survival gene expression (32)(33)(34)(35)(36). Consistent with a role for survival gene expression in preventing TNF␣-induced killing, inhibition of RNA or protein synthesis sensitizes normally resistant cells to TNF␣-induced killing (37). Similar to TNF␣, treatment of RelAϩ/ϩ MEFs with an RNA synthesis inhibitor actinomycin D (thus resulting in inhibition of RelA-mediated transcription) also rendered them susceptible to dsRNA-induced killing ( Fig. 2A). TNF␣ is cytotoxic to both RelAϪ/Ϫ fibroblasts and macrophages (21). To determine whether dsRNA could also induce cytotoxicity to RelAϪ/Ϫ macrophages, we generated fetal liver macrophages from embryonic day 14 (E14) RelAϩ/ϩ and RelAϪ/Ϫ mice. As seen with MEFs, RelAϪ/Ϫ macrophages readily lost viability in the presence of dsRNA, whereas RelAϩ/ϩ cells were not affected (Fig. 2B). Taken together, these results demonstrate a new role for RelA in preventing dsRNA-induced cytotoxicity, which similar to TNF␣ may also be mediated by regulation of survival gene expression. They also suggest that induction of pro-apoptotic genes, such as Fas, may not play an important role in mediating dsRNA-induced killing.
These results have revealed strikingly similar mechanisms for inhibiting cell death induced by TNF␣ and dsRNA. However, it is also possible that dsRNA-induced killing of RelAϪ/Ϫ cells is somehow mediated by the TNF␣ signaling pathway. This could be accomplished by dsRNA-induced release of presynthesized TNF␣ or by dsRNA-dependent mechanisms, for example, which could lead to activation of TNF receptors. Two TNF␣ receptors have been identified and named TNFR1 and TNFR2 (38,39). Unlike TNFR2, TNFR1 contains a death domain that can induce cytotoxicity in many cell types (40). To determine a possible involvement of TNFR1-mediated signaling in dsRNA-induced cytotoxicity, we tested the sensitivity of both TNFR1ϩ/ϩRelAϪ/Ϫ and TNFR1Ϫ/ϪRelAϪ/Ϫ MEFs to dsRNA or TNF␣-induced cell death (41). As expected, TNF␣ induced significant cytotoxicity in TNFR1ϩ/ϩRelAϪ/Ϫ but not in TNFR1Ϫ/ϪRelAϪ/Ϫ MEFs 2 (Fig. 2C). In contrast, dsRNA treatment efficiently killed both cell types. These results thus suggest that dsRNA induces a cell death pathway that does not depend on TNFR1-induced signaling.
dsRNA-induced Cell Death Requires PKR Activity-To determine whether dsRNA-induced killing of RelAϪ/Ϫ cells was in fact due to the absence of RelA, we tested whether ectopic expression of RelA was sufficient to protect RelAϪ/Ϫ MEFs from dsRNA-induced cytotoxicity. A GFP-expressing vector (to identify transfected cells) was cotransfected with either a control pLPC vector or a vector expressing RelA (pRelA). other 12 h after which the viability of GFP-positive cells was determined. As expected, RelAϪ/Ϫ MEFs transfected with the control pLPC vector readily lost viability after dsRNA treatment (Fig. 3). In contrast, cotransfection of pRelA significantly protected RelAϪ/Ϫ MEFs from dsRNA-induced cytotoxicity (Fig. 3). These results suggest that RelA plays a direct role in inhibiting dsRNA-induced cell death.
PKR activity is thought to be important for induction of cell death by dsRNA (6). We were also interested in determining whether PKR was important for dsRNA-induced killing of RelAϪ/Ϫ MEFs. To this end, we cotransfected RelAϪ/Ϫ MEFs with a construct encoding a PKR dominant-negative mutant, pPKR-DN. Importantly, expression of PKR-DN significantly enhanced viability of RelAϪ/Ϫ MEFs following dsRNA treatment (Fig. 3). These results thus suggest that PKR is required for induction of cell death by dsRNA in RelAϪ/Ϫ MEFs.
dsRNA Induces Apoptosis and Caspase Activation in RelAϪ/Ϫ MEFs-Apoptotic cell death results in characteristic morphological changes such as membrane blebbing, nuclear fragmentation, and chromatin condensation (42). DAPI staining of dsRNA-treated RelAϪ/Ϫ MEF nuclei revealed significant nuclear fragmentation and chromatin condensation (Fig.  4A), suggesting that dsRNA induces apoptotic cell death in RelAϪ/Ϫ MEFs. Activation of cysteine proteinases belonging to the caspase family has been shown to be critically important for induction of apoptosis (43). To determine whether caspase proteases are activated in RelAϪ/Ϫ MEFs following dsRNA treatment, we used a biotinylated caspase inhibitor, biot-Val-Ala-Asp(OMe)-fmk (biot-VAD-fmk). This inhibitor can covalently associate with activated caspases and thus provides a sensitive method for detection of caspase activation (9). dsRNA treatment of RelAϪ/Ϫ MEFs resulted in a time-dependent increase of a biot-VAD-fmk binding activity of ϳ18 kDa (p18) (Fig. 4B). Importantly, the molecular weight of p18 corresponded to that of the larger subunit of most activated caspases (also see be-low). Significantly, p18 could also be induced by TNF␣ treatment of RelAϪ/Ϫ MEFs (Fig. 4B), suggesting that both TNF␣ and dsRNA may induce a common caspase in these cells. In contrast, dsRNA treatment of RelAϩ/ϩ MEFs did not lead to increased levels of p18, but cotreatment with actinomycin D resulted in a dramatic increase of p18 (Fig. 4C). These observations thus demonstrate a critical role for RelA in inhibiting dsRNA and TNF␣-induced caspase activation.
Caspase 3, a Major Caspase Activated by dsRNA and TNF␣, Is Essential for Apoptotic Nuclear Fragmentation but Dispensable for dsRNA-induced Cytotoxicity-To characterize further p18, we used a streptavidin affinity column to purify p18. After extensive washing to remove activities bound nonspecifically to streptavidin, an 18-kDa activity could be detected in TNF␣treated extracts subjected to SDS-PAGE (Fig. 5A). Peptide sequencing and mass spectrometry analysis (not shown) of this 18-kDa protein revealed that the sequence of one peptide was identical to residues 65-75 of mouse caspase 3 (SGTD-VDAANLR). Western analysis using an antibody specific for the p17 subunit of activated caspase 3 confirmed its activation following both TNF␣ and dsRNA treatment (Fig. 5B) (the apparent molecular weight difference between p18 and p17 is likely due to association of biot-VAD-fmk, which has a molecular weight of 672, to p18). These results thus demonstrate that caspase 3 is a major caspase activated by both dsRNA and TNF␣ treatment of RelAϪ/Ϫ MEFs. Caspase 3 has been proposed to be a major downstream effector caspase responsible for executing the apoptotic cell death program (10,44,45). To determine the functional significance of caspase 3 in dsRNA-induced apoptosis, we used the caspase 3 inhibitor, Z-DEVD-fmk (9). Z-DEVD-fmk completely inhibited binding of p18 (caspase 3) to biot-VAD-fmk induced by dsRNA in RelAϪ/Ϫ MEFs (Fig. 6A). Interestingly, Z-DEVDfmk did not affect processing of caspase 3 (Fig. 6B) suggesting that Z-DEVD-fmk does not inhibit upstream caspases that are responsible for proteolytic processing of caspase 3 but rather specifically inhibits caspase 3 activity. Although Z-DEVD-fmk completely inhibited caspase 3 activity, Z-DEVD-fmk did not protect RelAϪ/Ϫ MEFs from dsRNA-induced cytotoxicity (Fig.  6C). These results suggest that caspase 3 activity is not essential for dsRNA-induced killing of RelAϪ/Ϫ MEFs. However, Z-DEVD-fmk completely inhibited nuclear fragmentation (Fig.  6D, compared with Fig. 4A). These results thus demonstrate that caspase 3 activity is essential for apoptotic nuclear changes but dispensable for dsRNA-induced cytotoxicity in RelAϪ/Ϫ MEFs.
Combined Treatment of RelAϪ/Ϫ MEFs with Z-DEVD-fmk and Antioxidants Completely Inhibits dsRNA-induced Cytotoxicity-Z-DEVD-fmk is not a broad specificity caspase inhibitor, raising the possibility that the inability of Z-DEVD-fmk to inhibit killing of RelAϪ/Ϫ MEFs was due to incomplete blockage of caspase activity. We therefore tested the effect of the broader specificity Z-VAD-fmk caspase inhibitor (9) on dsRNAinduced caspase activation and killing of RelAϪ/Ϫ MEFs. Unlike Z-DEVD-fmk, Z-VAD-fmk inhibited caspase 3 processing (Fig. 7A, compared with Fig. 6B), demonstrating its ability to inhibit the activity of upstream caspases responsible for caspase 3 processing and activation. However, Z-VAD-fmk did not inhibit, but rather significantly enhanced, cytotoxicity to RelAϪ/Ϫ MEFs following dsRNA treatment (Fig. 7B). In addition, both Z-DEVD-fmk and Z-VAD-fmk did not affect NF-B activation by dsRNA (data not shown), suggesting that the enhanced cytotoxicity by Z-VAD-fmk was not due to inhibition of NF-B. This is consistent with our recent observations showing that Z-VAD-fmk could sensitize RelAϩ/ϩ fibroblasts to dsRNA-induced necrotic killing (16). These results also indicate that caspase inhibition may not be sufficient to protect RelAϪ/Ϫ MEFs from dsRNA-induced cytotoxicity.
To understand further the mechanisms involved in dsRNAinduced killing of RelAϪ/Ϫ MEFs, we wished to identify noncaspase cytotoxic mechanisms that may be potentially responsible for inducing cell death. One such cytotoxic mechanism involves enhanced generation of reactive oxygen species (ROSs), which is also important in induction of necrotic cell death by TNF␣ (14 -16). We therefore tested whether treatment of RelAϪ/Ϫ MEFs with dsRNA resulted in enhanced ROSs production. However, no significant increase in ROSs production was noticed following dsRNA treatment of RelAϪ/Ϫ MEFs (Fig. 7C). Similarly, dsRNA ϩ Z-DEVD-fmk-treated RelAϪ/Ϫ MEFs also showed no significant increase in ROSs production (Fig. 7C). In contrast, treatment of RelAϪ/Ϫ MEFs with dsRNA ϩ Z-VAD-fmk significantly enhanced ROSs production (Fig. 7C). Furthermore, Z-VAD-fmk, but not Z-DEVDfmk, also enhanced ROSs production and necrotic cell death in RelAϩ/ϩ MEFs (data not shown). Thus in the presence of broad specificity caspase inhibitors such as Z-VAD-fmk, dsRNA treatment can enhance ROSs production and induce necrotic cell death in both RelAϩ/ϩ and RelAϪ/Ϫ MEFs. Nevertheless, our results suggest that such enhanced ROSs production may not be responsible for killing of RelAϪ/Ϫ MEFs by dsRNA or dsRNA ϩ Z-DEVD-fmk. It was, however, possible that dsRNA-induced killing of RelAϪ/Ϫ MEFs may result not from increased generation of ROSs but from enhanced susceptibility to constitutively produced ROSs. Importantly, in the presence of the antioxidant BHA, constitutive generation of ROSs was significantly reduced in both untreated and dsRNA ϩ Z-DEVD-fmk-treated RelAϪ/Ϫ MEFs (Fig. 7C). To determine a role for such constitutively generated ROSs in dsRNA-induced killing, we tested whether BHA could inhibit cell death induced by dsRNA. Interestingly, BHA treatment substantially enhanced survival of RelAϪ/Ϫ MEFs suggesting that dsRNA-induced killing may indeed result from enhanced susceptibility to constitutively produced ROSs (Fig. 7D). Importantly, combined treatments with Z-DEVD-fmk and BHA resulted in almost complete protection from dsRNA-induced cytotoxicity (Fig. 7D). These results indicate that inhibition of dsRNA-induced cytotoxicity requires inhibition of both caspase-dependent and ROSs-dependent mechanisms and raise the possibility that anti-apoptotic functions of NF-B are mediated by simultaneous inhibition of both cytotoxic pathways. DISCUSSION The results presented here provide evidence for a novel function of the RelA subunit of NF-B in inhibiting dsRNA-induced apoptosis. Although the role of NF-B in inhibiting apoptosis by endogenously produced factors such as TNF␣ (i.e. hostderived) is well established, we provide the first evidence for an NF-B-dependent function in inhibiting apoptosis induced by an exogenous agent (i.e. microbe-derived factor). Our results thus suggest that NF-B-mediated inhibition of apoptosis may FIG. 5. Caspase 3 is the major caspase activated in RelA؊/؊ MEFs by dsRNA or TNF␣. A, biot-VAD-fmk-labeled extracts from either untreated (UT) or TNF␣-treated RelAϪ/Ϫ fibroblasts were subjected to purification with a streptavidin affinity column. The purity of the extracts before or after purification was tested by SDS-PAGE followed by silver staining. p18 is indicated with an arrow. B, RelAϪ/Ϫ MEFs were either left untreated (UT) or treated with TNF␣ or dsRNA for 6 h. Whole cell lysates were analyzed by Western blotting with a caspase 3-specific antibody. The caspase 3 p17 is indicated by an arrow.
be an important mechanism for regulating cell survival during viral infection. Interestingly, and similar to TNF␣, we have found that dsRNA also induces caspase activation and apoptotic changes in RelAϪ/Ϫ cells. In addition, and also similar to TNF␣, dsRNA also triggers apoptosis in RelAϩ/ϩ cells in the presence of an RNA synthesis inhibitor. These results reveal a striking similarity in cellular mechanisms responsible for inhibiting cell death induced by TNF␣ and dsRNA. These results thus indicate that dsRNA-induced cell death may be inhibited by NF-B-mediated survival gene expression, rather than enhanced by NF-B activation and induction of Fas expression as previously reported (6,27,28).
dsRNA can be generated during infection with virtually any kind of virus (1). NF-B activated by dsRNA in infected cells allows activation of antiviral gene expression to limit further infection (46). However, our results suggest that NF-B-mediated induction of survival gene expression may be important for inhibiting apoptosis of infected cells and may thus enhance viral infection and virulence. Consistent with such a function of NF-B, a previous study has demonstrated a critical role for NF-B in maintaining virulence of encephalomyocarditis virus by preventing virus-induced apoptosis (47). It will thus be interesting to determine whether NF-B-mediated inhibition of dsRNA-induced apoptosis is a mechanism important for enhancing viral virulence. However, it is also possible that inhibition of NF-B-dependent anti-apoptosis in virus-infected cells may lead to eradication of infected cells. As shown here, inhibition of macromolecule synthesis sensitized cells to dsRNA-induced killing. Since shutdown of host macromolecule synthesis is one of the key events in late stage virus replication (48), it is likely that under these conditions the NF-B-induced protective pathway is also inhibited. Under these conditions, dsRNA may thus mediate cytotoxicity to virus-infected cells. Interestingly, TNF␣ has also been shown to kill vesicular stomatitis virus-infected cells (49,50). Since NF-B also protects cells from TNF␣-induced apoptosis, it is possible that killing of virus-infected cells by either dsRNA or TNF␣ is due to viral inhibition of the NF-B-induced protective pathway. Additional studies aimed at determining the impact of these NF-B-dependent pathways in controlling viral virulence may thus FIG. 6. Caspase 3 activity is essential for dsRNA-induced apoptotic nuclear changes but dispensable for cytotoxicity in RelA؊/؊ MEFs. A, RelAϪ/Ϫ MEFs were either left untreated (UT), treated with dsRNA alone, or treated with dsRNA and Z-DEVD-fmk for 6 h. Extracts were tested for biot-VAD-fmk binding as in Fig. 4B. p18 is indicated by an arrow. B, RelAϪ/Ϫ MEFs were treated as in A. Whole cell lysates were analyzed for caspase 3 activation by Western blotting. C, RelAϪ/Ϫ MEFs were either left untreated (UT), treated with dsRNA alone, or treated with dsRNA and Z-DEVD-fmk for 12 h before cell viability determined. D, RelAϪ/Ϫ MEFs were treated with dsRNA and Z-DEVD-fmk for 6 h before stained with DAPI. Nuclear morphology is indicated by arrows.
be required to understand fully the physiological functions of these key anti-apoptotic pathways.
We have shown here that induction of apoptosis by dsRNA in RelAϪ/Ϫ cells is mediated by activation of caspases. In particular, we have identified caspase 3 as a major caspase activated by dsRNA treatment that was found to be critically important for dsRNA-induced nuclear fragmentation in RelAϪ/Ϫ fibroblasts. These results are consistent with a previous study using caspase 3-deficient cells that demonstrated an important role for caspase 3 in mediating apoptotic nuclear changes by death inducers (44). Importantly, a recent study has identified a caspase 3 substrate, acinus, as the factor responsible for induction of chromatin condensation and fragmentation (51). It will thus be interesting to test whether acinus is also processed and activated in RelAϪ/Ϫ fibroblasts by dsRNA. Caspase activation involves a cascade of proteolytic events. Upstream initiator caspases, such as caspase 8 or 9 (which are generally activated by oligomerization) (52,53), induce proteolysis and subsequent activation of downstream effector caspases, such as caspase 3 (10,11). Although upstream caspase(s) involved in dsRNAinduced apoptosis are still not known, a recent study has demonstrated an essential role for the FADD protein in dsRNAinduced, PKR-mediated apoptosis (54). FADD is an adapter molecule involved in activation of caspase 8 by both TNFR1 and Fas (55)(56)(57)(58)(59). These observations thus suggest a possible involvement of caspase 8 in dsRNA-induced apoptosis of RelAϪ/Ϫ cells and suggest that dsRNA and TNF␣ may induce apoptosis by similar mechanisms that can be inhibited by NF-B.
Interestingly, inhibition of caspase 3 by Z-DEVD-fmk did not prevent dsRNA-induced cytotoxicity to RelAϪ/Ϫ MEFs. However, combined Z-DEVD-fmk and antioxidant treatment resulted in complete protection from dsRNA-induced cytotoxicity, even though dsRNA treatment did not result in a significant increase in ROSs generation. In RelAϩ/ϩ MEFs, dsRNA ϩ actinomycin D-induced cytotoxicity was also not inhibited by Z-DEVD-fmk alone, whereas cotreatment with Z-DEVD-fmk and BHA resulted in significant suppression of cell death (data not shown). We have also found that inhibition of TNF␣-induced killing of RelAϪ/Ϫ MEFs requires simultaneous inhibition of caspase proteases and ROSs generation. 3 Similar to dsRNA, TNF␣ treatment of RelAϪ/Ϫ MEFs did not enhance ROSs production. These results thus suggest that dsRNA or TNF␣ treatment may enhance the susceptibility of RelAϪ/Ϫ MEFs to constitutively generated ROSs. These results also suggest that NF-B-regulated anti-apoptotic genes may include those that can provide protection from ROSs-induced cytotoxicity. Such an NF-B-regulated gene may be manganous superoxide dismutase, which functions as a potent scavenger of superoxide anions and has previously been shown to inhibit TNF␣-induced cell death in certain cell lines (60). It is also possible that caspases activated during apoptosis can inhibit cellular damage induced by constitutively generated ROSs, perhaps to prevent necrotic lysis. In either case, our results suggest that caspase inhibition is not sufficient to inhibit apoptosis of RelAϪ/Ϫ cells and suggest that ROSs may provide an additional mechanism that plays a key role in the regulation of apoptosis induced by dsRNA and TNF␣. Caspase inhibitors are currently being considered for possible therapeutic use in the treatment of tissue-degenerative diseases (61). Our results suggest, however, that combined inhibition of both caspase-dependent and ROSs-dependent pathways may be required for inhibiting cell death induced by certain agents.
In conclusion, we have provided evidence of a novel function for the RelA subunit of NF-B in preventing dsRNA-induced apoptosis by potentially inhibiting both caspase-dependent and ROSs-dependent mechanisms. These results further underscore the key role played by NF-B proteins in inhibiting apoptosis induced by diverse agents. Our results also highlight the importance of NF-B proteins in regulating cellular re-sponses to viral products that may be important for controlling viral virulence and pathogenesis.