Hepatocyte Fas-associating death domain protein/mediator of receptor-induced toxicity (FADD/MORT1) levels increase in response to pro-apoptotic stimuli.

We examined the regulation of Fas-associating death domain (FADD) protein as an important adaptor molecule in apoptosis signaling and hypothesized that the regulation of FADD could contribute to hepatocyte death. FADD/mediator of receptor-induced toxicity (MORT1) is required for activation of several signaling pathways of cell death. In this study we report the interesting and unexpected result that actinomycin D increased the expression of FADD protein, and we demonstrate that other cellular stresses like ultraviolet irradiation or heat shock could also increase FADD levels in hepatocytes. In cells treated with actinomycin D, FADD levels were elevated homogeneously in the cytoplasm. The increase in cytoplasmic FADD protein by actinomycin D or FADD overexpression alone both correlated with cell death, and specific antisense inhibition of FADD expression consistently diminished approximately 30% of the cell death induced by actinomycin D. These data indicate that FADD protein expression can increase rapidly in hepatocytes exposed to broadly cytotoxic agents.

Acute liver failure is marked by a massive degree of hepatocyte apoptosis induced by exposure to toxic substances or ligation of members of the death receptor family (1)(2)(3). Actinomycin D (Act D) 1 has been used experimentally in models of hepatic damage because it is both directly toxic to hepatocytes and sensitizes the liver to induction of apoptosis by death receptor ligands (4 -6). Although transcriptional inhibition has been proposed as the mechanism for the effects of Act D, it has not been proven that this mechanism alone accounted for the direct toxicity or the sensitization of hepatocytes to receptormediated death.
Fas-associating death domain/mediator of receptor-induced toxicity (FADD/MORT1) is a cytoplasmic adaptor molecule that is required for the initiation of apoptosis signal transduction initiated by ligands that bind to the death receptors (7). FADD is a 29-kDa intracellular protein that contains two proteinprotein interaction domains: a death domain (DD) and a death effector domain (DED) (8,9). When cells are treated with death ligands that bind and activate cell-surface receptors like Fas or TNF receptor, FADD protein is recruited to the plasma membrane (10). There the DD of FADD interacts with the DD of a death receptor or another linker molecule, TNF receptor-associated death domain protein (TRADD) (11)(12)(13). Interaction of the DED of FADD with the tandem DED of procaspase-8 or -10 induces oligomerization, autocleavage, and activation of these initiator proteases in the apoptosis signaling cascade (13)(14)(15)(16)(17). Activated caspases can then cleave downstream substrates that transmit and amplify the programmed cell death signal. Cleavage of procaspase-3 propagates the cellular and nuclear apoptotic signal (16), and cleavage of BID by caspase-8 activates the mitochondrial apoptotic pathway in hepatocytes (17).
Because of the essential role of FADD in the most proximal mechanisms to activate ligand-induced apoptosis, we hypothesized that FADD levels would be a determinant of cell death in hepatocytes. We also tested the possibility that significant elevations in FADD levels alone, even in the absence of death ligands, would be adequate to induce death by cellular stresses. We measured hepatocyte viability and FADD protein levels in response to Act D, cycloheximide (CHX), ultraviolet irradiation (UV), heat shock, and specific expression of wild-type or antisense FADD. This study suggests that the modulation of endogenous FADD consisted of a highly regulated mechanism that can influence the ultimate demise of the cell exposed to toxic stresses.

EXPERIMENTAL PROCEDURES
Materials-Williams Medium E, penicillin, streptomycin, L-glutamine, and HEPES were purchased from Invitrogen. Insulin was purchased from Eli Lilly (Indianapolis, IN). Low endotoxin calf serum was from HyClone Laboratories (Logan, UT). Pan-caspase inhibitor benzyloxycarbonyl-Val-Ala-Asp fluoromethylketone (Z-VAD-fmk) and caspase-8 inhibitor Ac-Ile-Glu-Thr-Asp-CHO (IETD-CHO) were from Alexis Corp. (San Diego, CA). Stock solution of Z-VAD-fmk was prepared at 100 mM in dimethyl sulfoxide, and stock IETD-CHO was prepared in water at 100 mM. Mouse recombinant TNF␣ was obtained from R&D Systems (Minneapolis, MN). Antibodies used for this study * This work was supported by grants from the American College of Surgeons and the Society of Surgical Oncology (to P. K. M. K.) and by National Institutes of Health Grants R01-GM-44100 and R01-GM-50441 (to T. R. B.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF406779.
were purchased from Sigma (for ␣-actin), Santa Cruz Biotechnology, Inc. (Santa Cruz, CA) (for TRADD), and Stressgen (Victoria, British Columbia, Canada) (for FADD, caspase-3, and caspase-8). To ensure the specificity of the antibody used to detect FADD, the rat FADD gene was cloned (GenBank TM accession no. AF406779) and expressed by transient transfection in 293 human kidney cells using LipofectAMINE (Invitrogen). A 29-kDa rat FADD protein was detected by Western blotting with the anti-mouse FADD antibody (data not shown). Rabbit polyclonal antibody for BID was a generous gift from Dr. Xiao-Ming Yin. Bicinchoninic acid (BCA) protein and Supersignal TM chemiluminescent protein detection reagents were from Pierce. Unless otherwise indicated, all other reagents were purchased from Sigma.
Preparation of Primary Hepatocytes and Cell Culture-Primary hepatocytes were isolated and purified from male Sprague-Dawley rats (Harlan, Indianapolis, IN) or C57BL/6 mice (Charles River Laboratories, Wilmington, MA) by a collagenase perfusion method as described previously (18,19). Highly purified hepatocytes (Ͼ98% purity and Ͼ95% viability by trypan blue exclusion) were suspended in Williams' E medium supplemented with 10% calf serum, 1 M insulin, 2 mM Lglutamine, 15 mM HEPES, pH 7.4, 100 units/ml penicillin, and 100 g/ml streptomycin. The cells were plated on collagen-coated tissue culture plates at a density of 2 ϫ 10 5 cells/ml/well in 12-well plates for cell viability analysis, or 5 ϫ 10 6 cells/5 ml/10-cm dish for Western blotting assays. Cells were cultured overnight at 37°C in 5% CO 2 .
Apoptosis was induced by incubating the hepatocytes with culture medium containing 2000 units/ml TNF␣ and 200 ng/ml Act D for 8 -12 h. Hepatocytes were then scraped off the plates and centrifuged. For cytosolic lysates, cells were washed with cold phosphate-buffered saline (PBS) and resuspended in 5-fold volume of hypotonic buffer A (20 mM HEPES, pH 7.5, 10 mM KCl, 1.5 mM MgCl 2 , 1 mM EGTA, 1 mM EDTA, 0.5 mM phenylmethylsulfonyl fluoride, 5 g/ml aprotinin, 5 g/ml pepstatin, and 10 g/ml leupeptin). After three cycles of freezing and thawing, cell debris was removed by centrifugation at 13,000 ϫ g at 4°C for 20 min. The supernatant was used as a cytosolic lysate for Western blotting analysis. Protein concentration was determined using BCA assay (Pierce) with bovine serum albumin as standard.
Animal Studies-All procedures in this experiment were performed according to the guidelines of the Council on Animal Care at the University of Pittsburgh and the National Research Council's Guide for the Care and Use of Laboratory Animals. Sterile Act D (15 g/kg) or saline was injected via the dorsal penis vein into Sprague-Dawley rats under brief isoflurane anesthesia, and the livers were harvested 24 h later. Serum samples were obtained at the time of death, and the liver function tests were determined for aspartate aminotransferase and alanine aminotransferase using the Opera Clinical Chemistry System (Bayer Co., Tarrytown, NY). Liver samples were minced by cryostat and prepared as for cultured hepatocytes. Isolated hepatocyte protein lysates were assessed by Western blotting and probed for FADD protein.
Immunofluorescent Staining-Acid-cleaned, 3-aminopropyltriethoxysilane, 0.25% glutaraldehyde, collagen-1-coated (50 g/ml; Vitrogen 100, Palo Alto, CA) coverslips were prepared using a modified protocol (20). Hepatocytes were plated on coverslips overnight and treated with TNF␣ (2000 units/ml) and Act D (200 ng/ml). Six hours after treatment, hepatocytes were fixed for 20 min in ice-cold methanol. Apoptosis was measured by using the terminal deoxynucleotidyltransferase-mediated dUTP nick end labeling assay (TUNEL; Roche Molecular Biochemicals) as per manufacturer's instructions. The specimens were washed twice in PBS. Ten microliters of terminal deoxynucleotidyltransferase reaction mixture containing cobalt chloride and biotinylated dUTP were added to the slides and incubated at 37°C for 90 min. Slides were then washed with PBS three times and labeled with streptavidin-conjugated Alexa 488 (Molecular Probes, Eugene, OR) for detection of DNA strand breaks. Cells were washed in bovine serum albumin solution (0.15 g of glycine, 0.5 g of bovine serum albumin in 100 ml of 1ϫ PBS). Coverslips were blocked in 5% goat serum (Sigma) in bovine serum albumin for 45 min and washed. Cells were probed with rabbit anti-FADD polyclonal antibody (1:4000, Stressgen) for 1 h, washed, and then incubated with Cy3-linked goat anti-rabbit antibody (1:3000; Jackson Immunolabs). Cells were washed and nuclei were stained for 30 s with Hoechst 33325 (blue, 2 g/ml in water) or Sytox stain (green; Molecular Probes, Eugene, OR). Coverslips were mounted onto glass slides using watersoluble mounting medium (Gelvatol; Monsanto, St. Louis, MO). Images were collected with a 40ϫ objective together with multipass fluorescence to ensure optimal image registration. The image collection system was a Provis microscope (Olympus, Tokyo, Japan), a high sensitivity, integrating three-chip color camera (Sony Corp, Park Ridge, NJ), a frame grabber board (Coreco Inc., Saint-Laurent, Quebec, Canada), and Optimus software (Media Cybernetics, Silver Spring, MD). All images were collected with the same camera and microscope settings (three frames of integration) with neutral density fields.
Ultraviolet (UV) Radiation-Hepatocytes were washed twice with PBS and exposed to UV using a Philips Westinghouse FS40T12 ultraviolet lamp (peak wavelengths 290 -330) calibrated with a UVX digital radiometer and UVX-31 probe (UVP, San Gabriel, CA). After exposure to UV, fresh medium was added, and the cells were incubated at 37°C with 5% CO 2 .
Cell Viability Assay-Cell viability was determined by the crystal violet method as described previously (21). Briefly, cells were stained with 0.5% crystal violet in 30% ethanol and 3% formaldehyde for 10 min at room temperature. Plates were washed six times with tap water. After drying, cells were lysed with 1% sodium dodecyl sulfate (SDS) solution, and dye uptake was measured at 550 nM using a 96-well microplate reader. Cell viability was calculated from relative dye intensity of the mean for duplicate samples and presented as percentages relative to untreated samples. Cell viability is presented with the representative Western blotting results from the same hepatocyte harvest.
Immunoblotting Analysis-Forty micrograms of protein were separated on 13% SDS-PAGE and transferred onto a nitrocellulose membrane. Loading of equal protein amounts was assessed by staining of nitrocellulose membranes with 0.1% Ponceau S (Sigma) in 5% acetic acid. Nonspecific binding was blocked with PBS-T (14 mM sodium phosphate (monobasic, monohydrate), 88 mM dibasic sodium phosphate (anhydrous), 100 mM NaCl, and 0.1% Tween 20) containing 5% nonfat milk for 1 h of incubation with agitation at room temperature. Anti-FADD (1:4000), anti-caspase-8 (1:1000), anti-BID (1:1000), anti-TRADD (1:1000), or anti-␣-actin (1:1000) antibodies were added. After 1-h incubation at room temperature with agitation, membranes were washed three times with PBS-T. The horseradish peroxidase-conjugated goat anti-rabbit secondary antibody (Amersham Biosciences) was incubated at 1:10,000 for 1 h at room temperature. Following five washes with PBS-T, the protein bands were visualized with Supersignal TM (Pierce) according to the manufacturer's instructions. The bands were visualized on Kodak film (Eastman Kodak) exposed to the membrane to detect chemiluminescence signals. Intensity of bands was assessed by computer analysis using UN-SCAN-IT (Silk Scientific, Orem, UT).
Viral Vectors and Infection-E1-and E3-deleted Ad-FADD-WT and Ad-FADD-AS were constructed through Cre-lox recombination with reagents provided by Dr. S. Hardy (Somatix, Alameda, CA). An XhoI/ XhoI 720-bp murine FADD cDNA coding region (isolated from PCIm-Mort1 FL; generously provided by Dr. Astar Winoto) was inserted into the shuttle vector pAdlox with CMV promoter in forward and reverse direction to create wild type (WT) and antisense (AS) expression vectors, respectively. Recombinant adenovirus was generated by cotransfection of SfiI-digested pAdlox-FADD and Psi5 helper virus DNA into the Ad packaging cell line CRE8 that expressed Cre recombinase as previously described (22). Recombinant adenoviruses were propagated on CRE8 or 293 cells and purified by cesium chloride density gradient centrifugation and subsequent dialysis. Titers of viral particles were determined by optical densitometry, and recombinant viruses were stored in 3% sucrose at Ϫ80°C. The presence of the viral transgene product FADD was confirmed by Western blot analysis of infected hepatocytes. Plaque-forming units (PFU) were estimated at 100 particles/PFU. Ad-EGFP was constructed and obtained as described previously (23), and Ad-Psi5 was a gift from S. Hardy.
For adenoviral FADD gene transduction of hepatocytes, recombinant adenovirus was added to PBS-washed hepatocytes cultures at increasing multiplicities of infection (m.o.i. ϭ number of PFU/cell) in serumfree Opti-MEM. Cells were incubated for 4 h at 37°C and 5% CO 2 . Following this, cells were washed twice with PBS and fresh medium containing 10% calf serum was added.
FADD Northern Blotting-Isolation of total RNA and Northern blot analysis of rFADD mRNA levels in cultured hepatocytes were performed using the 720-base pair rat FADD cDNA coding region as a DNA probe as described previously (24). Northern blots were probed in Ex-pressHyb hybridization solution (CLONTECH), according to the instructions of the manufacturer, using radiolabeled rFADD cDNA (Gen-Bank TM accession no. AF406779) or ␤-actin (CLONTECH). Equal RNA was loaded as assessed by quantitation with UV spectrophotometry and measurement of 28 and 18 S ribosomal RNA.
Other Analyses-Data are presented as means Ϯ S.E. of the mean of at least three separate experiments except where results of blots are shown, in which case a representative experiment is depicted in the figure. Comparisons between values were analyzed using analysis of variance (ANOVA) in SigmaStat. Differences were considered significant at p values Յ 0.05.

Act D Induced Hepatocyte Death and Increased Levels of FADD Protein-Previous
studies have shown that hepatocyte death cannot be induced by TNF␣ unless used in conjunction with sensitizing agents such as Act D (25). We tested the hypothesis that Act D influenced cell death by altering FADD protein expression. Act D alone induced a concentration-dependent decline in hepatocyte viability that was associated with a dose-dependent increase in FADD protein levels (Fig.  1A). FADD levels were also up-regulated by Act D in a timedependent manner, which inversely correlated with viability (Fig. 1B). The increase in FADD levels was observed as early as 4 h after Act D (200 ng/ml) treatment, with up to 5-fold increase in FADD protein levels that seemed to plateau after 8 h. Ex-posure to TNF␣ alone had minimal effects on hepatocyte viability, but when added with Act D, TNF␣ further reduced viability from 49 Ϯ 5.2% to 37 Ϯ 11.2% (Fig. 1C). TNF␣ alone did not significantly increase FADD levels and did not alter the Act D-induced increases in FADD. As expected, cell death by Act D was associated with a decrease in procaspase-8 and increase in caspase-8 and caspase-3 cleavage bands without changes in TRADD or ␣-actin levels (Fig. 1C), although there was a slight increase in TRADD levels with Act D and TNF␣ treatment in combination.
TNF␣ with the transcriptional inhibitor D-galactosamine has been reported to increase FADD mRNA and protein levels in hepatocytes in vivo (26). To determine whether Act D would increase FADD in vivo, FADD protein levels were measured in FIG. 1. Act D induced dose-dependent and time-dependent cell death and up-regulation of FADD protein, but not mRNA, expression in rat hepatocytes in vitro. A, rat hepatocyte cultures were exposed to increasing doses of Act D. Eight hours later FADD expression was assessed by Western blotting, and viability was assessed at 12 h by crystal violet staining. B, hepatocytes were exposed to increasing durations of Act D at 200 ng/ml. Decrease in hepatocyte viability measured by crystal violet staining correlated with increase in FADD protein expression by Western blotting. C, rat hepatocytes stimulated for 8 h with Act D (200 ng/ml) with or without TNF␣ (2000 units/ml) show a 5-fold increase in FADD protein, decrease in procaspase-8, increase in active caspase-8 and caspase-3 cleavage bands, and no change in ␣-actin or TRADD levels by Western blotting. Cell viability was measured for each treatment at 12 h by crystal violet staining of adherent cells. Cell viability was decreased by treatment of rat hepatocytes with Act D, TNF␣/Act D, but not by TNF␣ alone. D, FADD protein was increased in the liver of rats injected with Act D. Rats were injected via penis vein with sterile normal saline or Act D (15 g/kg) in normal saline and sacrificed 24 h later. FADD protein was measured by Western blotting of whole liver lysates. E, Act D decreased RNA expression of rat FADD in hepatocytes. Rat hepatocytes were treated with Act D (200 ng/ml) for 6 h and harvested for total RNA. RNA was probed by Northern blotting with rat FADD cDNA, stripped, and re-probed with ␤-actin cDNA. Act D decreased FADD and ␤-actin mRNA levels. 28 and 18 S ribosomal RNA was assessed to measure equal loading of total RNA. liver lysates from rats injected intravenously with Act D (15 g/kg). Fig. 1D demonstrates a marked increase in FADD levels 24 h after Act D injection. At this time point, no changes in liver morphology or serum liver enzymes were observed (data not shown).
To assess whether Act D also affected FADD mRNA levels, Northern blots were performed on total RNA isolated from hepatocytes treated with Act D. We used the 720-bp rat FADD cDNA coding region as a probe for Northern blotting. In contrast to protein levels, hepatocyte FADD mRNA levels were decreased 6 h after treatment with Act D (200 ng/ml; Fig. 2). This result suggested that FADD protein up-regulation occurred at the post-transcriptional level. ␤-Actin levels were also decreased in hepatocytes treated with Act D consistent with the action of Act D as a transcriptional inhibitor, but 28 and 18 S RNA levels in each lane showed equivalent total RNA loading.
Subcellular Localization of FADD in Hepatocytes-To determine whether the up-regulation of FADD was associated with changes in the subcellular localization of FADD, light and immunofluorescent microscopic imaging studies of the cells were undertaken. Unstimulated hepatocytes remained adherent to collagen-coated plates as a semiconfluent monolayer ( Fig. 2A). Hepatocytes stimulated for 12 h with Act D displayed characteristics of apoptotic cell death, including cell shrinkage and nuclear condensation, and detached from the plate (Fig.  2B). Unstimulated hepatocytes analyzed by immunofluorescent staining for FADD protein exhibited a cytoplasmic distribution of FADD protein (Fig. 2C). A homogeneous increase in FADD immunofluorescent staining was observed in the cytoplasm in hepatocytes treated with Act D for 6 h (Fig. 2D) that was consistent with the increase in FADD protein observed by Western blotting (Fig. 1A). These localization studies demonstrated that the increase in cytoplasmic FADD levels did not involve translocation of protein from a sequestered, intracellular source.
These studies demonstrated that Act D treatment of hepatocytes would result in increased FADD expression, but the role of elevated FADD protein in the activation of cell death pathways in primary hepatocytes was unclear. FADD overexpression has been shown to induce death in the B lymphoma cell line BJAB and MCF-7 breast cancer cells, as well as embryonic fibroblasts (7,27). To evaluate whether overexpression of FADD protein alone would activate death pathways in primary hepatocytes, we constructed a replication-deficient adenovirus that expressed wild type mouse FADD (Ad-FADD-WT). Fluorescent analysis of hepatocytes infected with a control virus that expressed enhanced green fluorescent protein (Ad-EGFP) at m.o.i. of 10 resulted in Ͼ90% infectivity without induction of cell death (Fig. 2E). Ad-FADD-WT infection of hepatocytes resulted in an increase in FADD immunofluorescent staining in the cytoplasm and condensed, apoptotic nuclear and cellular morphology (Fig. 2H) compared with infection with control virus (Fig. 2F) or medium alone (Fig. 2G). Higher magnification of hepatocytes treated with Act D showed nuclei with condensed apoptotic morphology (Fig. 2J). To assess for DNA fragmentation, cells treated with Act D (Fig. 2L) showed an increase in TUNEL-positive staining compared with untreated hepatocytes (Fig. 2K). These results demonstrate that Act D treatment or overexpression of FADD could induce an apoptotic cytoplasmic and nuclear morphology in hepatocytes.

FADD Overexpression Induced Cell Death in Hepatocytes, and FADD Antisense Expression Blocks Hepatocyte Death by
Act D-To assess the molecular effects of FADD overexpression, we examined the cleavage of procaspase-8 and Bid, which are molecules immediately downstream of FADD in the apoptotic signaling pathway. Eight hours following Ad-FADD-WT infection at m.o.i. 10, FADD protein expression was significantly increased in hepatocytes, as detected by Western blotting (Fig. 4A). Cleavage products of caspase-8 and BID (tBID) were seen 18 h after Ad-FADD-WT infection but not after Ad-EGFP infection (Fig. 3A). To quantitate the cell death induced by FADD overexpression, infection of hepatocytes with Ad-FADD-WT resulted in a dose-dependent decrease in viability of hepatocytes (Fig. 3B), which was similar to results after treatment with Act D (Fig. 1A). Ad-FADD-WT infection at m.o.i. 10 induced a reduction in cell viability to ϳ58% compared with controls (Fig. 3C, lane 3). As expected, death by Ad-FADD-WT infection was inhibited by co-culture with pancaspase inhibitor Z-VAD-fmk (100 M) or by caspase-8 inhibitor IETD-CHO (100 M) (Fig. 3C, lanes 5 and 6, respectively; n ϭ 4, ANOVA p Ͻ 0.05). TNF␣ alone did not induce death in hepatocytes, nor did it augment the cytotoxic effects of Ad-FADD-WT at m.o.i. of 10.
To determine whether the hepatotoxicity induced by Act D was dependent on FADD protein expression, an adenoviral vector expressing FADD antisense mRNA (Ad-FADD-AS) was constructed. Infection of hepatocytes with Ad-FADD-AS decreased endogenous FADD protein levels at day 3 after infection (Fig. 4A), which demonstrated the efficacy of the Ad-FADD-AS construct to abrogate FADD expression. Cell viability was not significantly changed in cells treated with Ad-FADD-AS at an m.o.i. of 10 (Fig. 4B, lane 3). Ad-FADD-AS infection inhibited cell death by TNF␣/Act D (Fig. 4B, lane 7; n ϭ 4, ANOVA p Ͻ 0.05) or Act D alone (Fig. 4B, lane 6; ANOVA p Ͻ 0.05) by ϳ30%. These results suggested that death induced by Act D alone was at least partially FADD-dependent.
Ultraviolet Irradiation and Extreme Heat Shock Increased FADD in Hepatocytes-We wanted to determine whether other cytotoxic stimuli could also affect FADD expression levels. Ultraviolet (UV) irradiation has been shown to induce apoptosis in several cell types via a FADD-dependent mechanism that involves cross-linking of death receptors such as Fas or TNF receptor (28,29). To assess whether FADD levels also correlated with UV exposure in hepatocytes, FADD protein levels were assessed after UV treatment and found to be increased as soon as 6 h after exposure to 200 mJ/cm 2 UV (Fig. 5A). Hepatocytes exhibited a dose-dependent increase in FADD protein measured at 8 h after exposure to UV (Fig. 5B). The lowest exposure associated with a rise in FADD levels (50 mJ/cm 2 ) was associated with a 50% decrease in cell viability by 8 h after UV exposure.
UV and Act D might have induced hepatocyte death by activation of FADD-mediated pathways of apoptosis, but cells can also respond to stress by the activation of cytoprotective mechanisms that include heat shock proteins (30). We postulated that the level of stress would dictate whether hepatocytes activated pro-death pathways through FADD or pro-survival mechanisms. This hypothesis was tested in cultured hepatocytes exposed to heat shock, a stimulus that, depending on severity, is known to induce either the stress response or death (31). After modulating the duration of 42°C heat exposures from 20 to 120 min, induction of the cytoprotective protein Hsp70 was assessed as a marker of the heat shock response in hepatocytes exposed to varying degrees of stress (32,33). Hsp70 has been shown to block recruitment of caspase-9 to Apaf-1 (34), and elevation of Hsp70 by nitric oxide in hepatocytes has been shown to inhibit apoptosis (32). Cells treated for up to 120 min with heat shock showed no immediate change in viability by morphology or crystal violet staining (data not shown). FADD levels, however, were markedly elevated after 120 min of heat shock (Fig. 5C). After cells were cultured at 37°C for 18 h, minimal changes in viability were seen with up to 40 min of initial heat shock, although 120-min exposures to heat shock resulted in global cell death. Coincident with the minimal mortality of the 40-min heat exposure was the absence of early FADD expression (Fig. 5C) and the marked up-regulation of Hsp70 at 18 h (Fig. 5D). In contrast, 120-min heat exposure was associated with a marked early up-regulation of FADD protein (Fig. 5C), failure to up-regulate Hsp70 (Fig. 5D), and ϳ93% cell death at 18 h. The kinetics of FADD protein elevation observed as a response to extreme heat shock was notably more rapid than the responses observed in hepatocytes exposed to UV (Fig. 5A) or Act D (Fig. 1B). These experiments showed that FADD protein could be increased by stresses known to induce cell death through ligand-independent mechanisms including ultraviolet irradiation and thermal injury. Moreover, at levels of heat shock that led to cell death, upregulation of the cytoprotective protein hsp70 was replaced by significant elevations in FADD protein.
Finally, to assess whether other metabolic inhibitors could also increase FADD protein levels, hepatocytes were treated with cycloheximide (CHX), a translational inhibitor. CHX has been used to sensitize hepatocytes to cell death via receptor ligation, and CHX can induce hepatocyte death in vitro. We found that CHX (100 M) also increased FADD in a time-dependent manner (Fig. 5E). Taken together, these studies demonstrate that levels of FADD protein were increased by several inducers of hepatocyte death. DISCUSSION The essential role of FADD in the initiation of many apoptotic signaling pathways led us to examine the role of changes in FADD protein levels in apoptosis. In these studies we have demonstrated that FADD levels were modulated under conditions associated with increased apoptosis and that cell death could be controlled by intracellular changes in FADD levels. Constitutive expression of FADD has been shown in several different cell types (9). However, it was unknown whether basal levels are adequate to initiate ligand-dependent apoptosis and whether elevation in endogenous FADD levels represents a mechanism to activate apoptotic signaling. In the studies reported here, FADD levels correlated with hepatocyte death following treatment with Act D (Fig. 1), CHX (Fig. 5E),

FIG. 5. Cellular stress differentially increases FADD and Hsp70 protein expression in hepatocytes.
A, hepatocytes were exposed to increasing doses of UV and harvested after 8 h. Protein lysates were assessed for FADD by Western blotting. B, ultraviolet radiation B (280 -330 nm) up-regulated FADD and increased cell death in rat hepatocytes. Hepatocytes were exposed to 200 mJ/cm 2 UV and harvested at 0, 2, 4, 6, and 8 h after exposure. C and D, cultured rat hepatocytes were treated with medium exposed to 42°C medium for 0, 20, 40, 60, and 120 min to induce heat shock. C, cells were harvested immediately after heat shock and assessed for FADD protein levels. ultraviolet radiation (Fig. 5A), or heat shock (Fig. 5C). These data indicate that FADD can be rapidly induced in cells exposed to a broad range of cytotoxic stimuli. Furthermore, similar increases in specific FADD expression in hepatocytes induced caspase-dependent cell death in the absence of other external insults. Conversely, depletion of FADD protein by expression of antisense FADD diminished cell death induced by Act D alone or by TNF␣ in combination with Act D by ϳ30%. Unlike FADD protein expression, steady-state levels of FADD and actin mRNA were decreased by 6-h treatment with Act D (Fig. 1E). It is likely that hundreds of proteins are decreased in expression by the transcriptional inhibitor Act D. These data, in conjunction with the observation that Act D up-regulated FADD protein, indicate that increased FADD levels contributed significantly to Act D-induced death. More importantly, changes in FADD levels could serve as a mechanism to activate cell death signaling pathways following exposure to significant stress.
Nevertheless, the mechanisms for the up-regulation of FADD are not yet clear. We have treated HepG2 hepatocellular carcinoma and A549 lung cancer cells with Act D (200 ng/ml), and we did not observe an increase in FADD protein expression above baseline for these cells (data not shown). This raises interesting questions because these tumor cell lines responded differently from primary hepatocyte cultures. One possibility is that immortalized cells lose the ability to up-regulate FADD that exists in primary cells in culture. Tumor necrosis factor-␣ alone does not readily induce hepatocyte death in vitro or in vivo unless used in conjunction with Act D or D-galactosamine (5). Transcriptional inhibitors D-galactosamine and ␣-amanitin also increased FADD protein levels in cultured hepatocytes (data not shown). Others have shown in HeLa and SHEP neuroblastoma cells that Act D decreased short-lived inhibitors of apoptosis like FLIP and XIAP (35,36). We did not observe changes in FLIP or IAP protein levels in hepatocytes exposed to Act D (data not shown). Taken together, these findings lend support to the idea that up-regulation of FADD by transcriptional inhibitors may also act to sensitize hepatocytes to cell death by TNF␣, and we postulate that another rapidly metabolized factor exists that may regulate FADD protein expression.
Other investigators have reported the appearance of "death effector filaments" in HeLa cells induced for apoptosis via death receptors (37,38). We did not observe cytoplasmic filaments in treated or untreated hepatocytes immunostained for FADD. No change in subcellular localization was observed when hepatocytes were treated with Act D alone, but FADD levels were higher in the cytoplasm. Although expression levels of ligands and death receptors were not manipulated or measured in this study, it would appear that FADD could act to precipitate apoptosis in the absence of direct receptor ligation and that this process becomes more efficient as FADD levels increase. Chen et al. (39) have shown that Ras-dependent apoptosis appears to signal through FADD via a receptor-independent mechanism, and chemotherapeutic agents induce cell death in U937 monocytes via FADD-dependent, ligand-independent mechanisms. However, whether FADD up-regulation occurs in these situations is not known and warrants further investigation. Our findings suggest that elevations in cytoplasmic FADD expression by cellular stresses may play an important regulatory role in the induction of apoptosis.
Because hepatocytes are difficult to transiently transfect in vitro and in vivo, to express FADD wild-type and antisense genes, we generated adenoviral vectors that readily infect primary hepatocytes. Both the full length and truncated forms of FADD have been shown to mediate apoptotic and necrotic cell death pathways. Expression of dominant-negative FADD induced non-apoptotic death in NIH3T3 cells treated with TNF␣ (40). In Jurkat T cells and L929 fibroblasts, necrotic death was induced by signaling through Fas receptor that appeared to be FADD-dependent (41)(42)(43). We found that Act D induced cell death with apoptotic morphology in hepatocytes. Our results in conjunction with other studies suggest that FADD may be involved with multiple death pathways including caspase-dependent apoptosis as well as caspase-independent necrosis (44). Therefore, in our studies, we used crystal violet staining of adherent cells as a simple and accurate viability assay for hepatocytes. Surprisingly, the dominant-negative, mutant FADD protein has been shown to retain some pro-death activity in normal prostate cells (45). For these reasons we justified the use of antisense technology to prevent murine FADD protein expression and inhibit cell death instead of expressing the dominant-negative human FADD protein. Preliminary results indicate that Ad-FADD-AS may protect against death induced by heat shock or UV in hepatocytes, but it was unclear in those experiments whether adenoviral infection itself introduced additional hepatotoxicity. We decided to pursue that question in a physiologically and biologically relevant model; we have demonstrated in human keratinocytes, as in hepatocytes, that UV increases FADD protein levels and induced apoptosis which was inhibited by antisense inhibition of FADD protein expression by Ad-FADD-AS. 2 Our findings of endogenous augmentation in FADD expression support the pursuit of future studies to determine its biological relevance. FADD expression was absent in PLC/ PRF/5 hepatocellular carcinoma cells (46), and also shown to be decreased 10-fold in mantle cell lymphomas (47). Therapeutically, activation of FADD/MORT1 function by gene therapy has been used to induce cell death in cancers of the brain (48,49). In our studies, severe cellular stress in the form of heat shock increased FADD protein levels that correlated with depressed cell viability (Fig. 5C). By contrast, mild to moderate amounts of stress was associated with increases in the protective protein Hsp70. FADD levels were not increased immediately after 60 min of heat shock (Fig. 5D). Because of the stress induced by heating cells, hepatocyte death by heat shock is likely more complex and complete than death induced by the other agents like Act D. In preliminary experiments, death induced by heat shock was not blocked by caspase inhibitors (data not shown) and may represent death by necrosis. In the setting of moderate heat, we expect additional unknown, pro-necrotic mecha-2 P. K. M. Kim, R. Weller, and T. R. Billiar, submitted for publication. nisms other than FADD up-regulation would also influence cell death.
These data lend support to a model of a "severe stress sensor" whereby the cell has a mechanism to detect and respond to the degree of stress to which it is exposed. Cells exposed to mild injury appear to respond with elevations of cytoprotective proteins like hsp70 that might inhibit cell death by various mechanisms and allow cellular repair (34); after exposure to lethal stresses, FADD may be preferentially up-regulated to ensure eradication of incapacitated or mortally injured cells (Fig. 6). This reciprocal expression of the two proteins raises the interesting possibility that Hsp70 may regulate FADD expression, a hypothesis that is not yet tested. Because this response mechanism to toxic stresses may modulate cell death after severe injury, we and others have proposed the manipulation of FADD-mediated hepatocyte death as an alternative modality to protect the liver from fulminant hepatic failure in vivo (1,50). Finally, because FADD can effectively mediate multiple death pathways, one should not be surprised to discover that other human diseases may be caused, modified, and treated by inducing changes in the expression of this apoptotic signaling molecule (51,52).