Nonobese diabetic (NOD/LtJ or NOD) mice are resistant to doses of LPS and d-galactosamine that uniformly produce lethality in C57BL/6J (B6) mice (p < 0.01). Liver caspase-3-like activity, serum transaminase levels (both p < 0.05), and the numbers of apoptotic liver nuclei were also reduced in NOD compared with B6 mice treated with LPS (100 ng) and d-galactosamine (8 mg). NOD mice were also at least 100-fold more resistant to recombinant human TNF-α and d-galactosamine treatment than B6 mice (p < 0.001). Binding of recombinant human TNF-α to splenocytes from NOD mice was similar to that seen in B6 mice, suggesting that the defect in responsiveness was not due to an inability of recombinant human TNF-α to bind the NOD TNF type 1 (p55) receptor. Because the TNF type 1 (p55) receptor shares a common signaling pathway with Fas (CD95), NOD and B6 mice were treated with the Fas agonist antibody, Jo-2. Surprisingly, NOD mice were as sensitive as B6 mice to Fas-induced lethality and hepatic injury. In addition, primary hepatocytes isolated from NOD mice and cultured in vitro in the presence of d-galactosamine with or without TNF-α were found to be resistant to apoptosis and cytotoxicity when compared with B6 mice. In contrast, Jo-2 treatment produced similar increases in caspase-3 activity and cytotoxicity in primary hepatocytes from NOD and B6 mice. The resistance to LPS- and TNF-α-mediated lethality and hepatic injury in d-galactosamine-sensitized NOD mice is apparently due to a post-TNFR binding defect, and independent of signaling pathways shared with Fas.
The nonobese diabetic (NOD)4 mouse is a well-characterized model that is both pathologically and genetically analogous to human type I diabetes (1). NOD mice spontaneously develop autoimmune diabetes indicated by leukocytic infiltration of the pancreatic islets at 3–5 wk of age and eventual destruction of insulin-secreting pancreatic β cells resulting in hyperglycemia (1). Although insulin-dependent diabetes in the NOD mouse is a complex polygenic trait, the primary genetic element underlying diabetes susceptibility in the NOD mouse appears to be the MHC. The unusual H-2g7 (Kd, I-Ag7, I-Enull, Db) possessed by the NOD mouse along with other non-MHC susceptibility loci have been identified as major mediators of disease development (2, 3). The MHC class II molecules of diabetes-prone humans and mice share a homozygous lack of aspartic acid at position 57 in the MHC class II β chain. Substitution of aspartic acid at position 57 protects NOD mice from diabetes, indicating the defining role that presentation of peptides via class II plays in defining the peripheral T cell repertoire (2, 3). However, disease susceptibility in humans as well as NOD mice appears also to be governed in part by the environment, making the mechanism of disease difficult to elucidate (4, 5). The defects present in the NOD mouse allowing for the breakdown of either central or peripheral tolerance to self-Ag are unknown. However, studies by Serreze and Leiter indicate that defects in Ag-presenting cell differentiation and function may contribute to the lack of tolerance to pancreatic β cell self-Ags (1, 4, 6, 7, 8, 9).
Recent studies have suggested that TNF-α expression is increased in the pancreatic β cells of NOD mice, and this localized TNF-α production plays a critical role in both the initiation of insulitis and the subsequent progression to β cell destruction (10). NOD mice deficient in TNFR1 receptor develop insulitis similar to that of wild-type NOD mice; however, progression to diabetes is not observed (11), indicating that β cell toxicity may occur via a TNFR1-dependent mechanism.
Although TNF-α contributes to the autoimmune predisposition of NOD mice, little is known about their ability to synthesize or respond to TNF-α. Jacob et al. (12) reported that peritoneal macrophages from NOD mice have markedly reduced TNF-α production in response to LPS and IFN-γ. Of 24 mouse strains tested, in vitro TNF-α production by peritoneal macrophages stimulated with LPS and IFN-γ was lowest in NZW and NOD strains.
The purpose of this study was to evaluate whether NOD mice synthesize TNF-α and respond to endogenously produced and exogenously administered TNF-α in a manner similar to that for C57BL/6J (B6) mice. Identifying differences in TNF-α production and responsiveness of NOD mice can provide a powerful genetic tool not only for better understanding of TNF-α-dependent disease progression in autoimmune diabetes but also in endotoxin-induced shock and liver injury.
Several years ago, Galanos et al. (13) demonstrated that transcriptional inhibition in the liver with d-galactosamine increases by several thousand-fold the sensitivity of mice to the lethal effects of LPS and TNF-α. d-Galactosamine selectively blocks transcription in hepatocytes by depleting uridine nucleotides necessary for the production of mRNA transcripts (14). Treatment with LPS or TNF-α in conjunction with d-galactosamine results in acute liver apoptosis and liver failure (15, 16). This model of liver injury and lethality is mediated by TNF-α signaling through the TNFR1 receptor, resulting in activation of caspases and subsequent hepatocyte apoptosis (14, 17).
In this report, we have examined the responsiveness of NOD and B6 mice to lethality and hepatic injury secondary to the administration of LPS and d-galactosamine. To eliminate the possibility that differences in the responsiveness to LPS could be explained by a reduced production of TNF-α, the studies were repeated in both strains of d-galactosamine-sensitized mice with recombinant human TNF-α, which binds predominantly to the mouse TNFR1 receptor (18, 19). Finally, because the TNFR1 receptor signaling pathway converges with Fas signaling pathways (20, 21), NOD and B6 mice were also treated with the Fas agonist, Jo-2. We show here that NOD mice are less responsive to both LPS and d-galactosamine, and TNF-α- and d-galactosamine-mediated liver injury and death when compared with B6 mice. However, no differences were seen between NOD and B6 mice in responsiveness to the Fas agonist Ab, Jo-2. In addition, primary hepatocytes isolated from NOD mice and cultured in the presence of d-galactosamine with TNF-α were also resistant to toxicity and had less active caspase-3 activity than hepatocytes from B6 mice. In contrast, Jo-2-induced toxicity was equivalent in in vitro-treated primary hepatocytes from both B6 and NOD mice.
Materials and Methods
LPS from Escherichia coli 0111:B4 and d-galactosamine were obtained from Sigma (St. Louis, MO). Jo-2, a purified hamster anti-mouse Fas mAb in no azide/low endotoxin format with agonist properties was obtained from PharMingen (San Diego, CA). Purified hamster IgG, group 2, λ monoclonal isotype control Ab (NA/LE) was also obtained from PharMingen. In addition, anti-mouse CD3 allophycocyanin (APC)-labeled Ab was also obtained from PharMingen. Recombinant human TNF-α (rhTNF-α) was the generous gift of Dr. Tadahiko Kohno (Amgen, Thousand Oaks, CA).
Animal experimental protocols
Female B6 and NOD/LtJ (NOD) mice between 5 and 8 wk of age were obtained from The Jackson Laboratory (Bar Harbor, ME) or from the Department of Pathology Animal Resource Facility, University of Florida College of Medicine (Gainesville, FL). NOD and B6 mice bred at the University of Florida were derived from breeding stock obtained from The Jackson Laboratory. All animal protocols were approved by the Institutional Animal Care and Use Committee of the University of Florida College of Medicine.
To assess mortality, animals were treated with i.p. injections of 8 mg d-galactosamine per mouse in combination with LPS or recombinant human TNF-α (rhTNF-α) in 200 μl sterile physiologic saline (155 mM NaCl). An anti-Fas mAb (Jo-2) diluted in 200 μl sterile physiologic saline vehicle was also administered to some animals. Control groups of mice received either physiologic saline alone or 8 mg d-galactosamine in physiologic saline. As a control for the anti-Fas-injected animals, an isotype Ab control was injected at the same concentration. Mice were monitored for mortality every 6 h for the first 24 h and then twice daily for up to 72 h. Alternatively, mice were injected and bled by retroorbital puncture at 90 min for serum TNF-α measurements. At 3 h (post Jo-2 injection) or 6 h (post LPS or TNF-α injection), mice were anesthetized with sodium pentobarbital (50 mg/kg body weight ) and bled by cardiac puncture for quantitation of serum transaminases. Mice were then sacrificed by cervical dislocation, and the livers were harvested, rinsed in cold PBS, and immediately homogenized for caspase-3-like activity assay. In addition, one lobe of the liver was fixed in 10% buffered formalin for paraffin embedding, and 5-μm sections were cut and affixed to slides. One slide was stained with hematoxylin and eosin, and another was used for in situ 3′-TUNEL staining as described below.
Preparation of liver or spleen cell homogenates for caspase-3-like activity assay
Tissues were harvested and rinsed in cold sterile PBS and then homogenized in 3 volumes (v/w) of ice-cold homogenization buffer (25 mM HEPES (pH 7.5), 5 mM MgCl2, 1 mM EDTA, 1 mM PMSF, 1 μg/ml leupeptin, and 1 μg/ml aprotinin). Homogenates were centrifuged at 13,500 rpm in a Beckman J2-HS centrifuge for 15 min, and supernatants were assayed for protein content using a Bio-Rad protein assay kit (Bio-Rad Laboratories, Hercules, CA) according to instructions provided.
Caspase-3-like activity determination
Protein extracts from either primary hepatocytes or from one lobe of liver or whole spleen were assayed for caspase-3-like activity using protocols previously described (22). Briefly, 40 μg of total protein per sample were incubated with the synthetic substrate benzyloxycarbonyl-Asp-Glu-Val-Asp-7-amino-4-trifluoromethylcoumarin (Z-DEVD-AFC) (Enzyme Systems Products, Livermore, CA). The cleavage of substrate was monitored in a fluorescence reader using 400 nm excitation and 505 nm emission wavelengths. Calibration curves were generated using standard concentrations of AFC and the caspase-3-like activity was calculated from the slope of the recorded relative fluorescence and expressed as relative fluorescence units.
Cytokine measurements and serum transaminase determinations
TNF-α bioactivity was determined using the WEHI 164 clone 13 cell line cytotoxicity assay as previously described (23). Specificity of the assay was determined by incubating murine serum samples with anti-TNF-α Abs to block cytotoxicity (23).
Transaminase levels (aspartate aminotransferase (AST)) were determined on serum samples using a commercial kit adapted for the small sample volumes obtained from mice (Sigma) as previously described (23).
In situ TUNEL assays on liver tissue sections
Livers were fixed in 10% buffered formalin and embedded in paraffin. Sections of 5 μm were affixed to slides and then deparaffinized and rehydrated. Slides were either stained with hematoxylin and eosin for analysis of morphologic changes or further prepared for fluorescent 3′-end labeling of genomic DNA fragments using a commercial Apoptag kit (Promega, Madison, WI). Slides underwent proteinase K digestion for 15 min in buffer (0.5 M EDTA (pH 8.0), 200 mM Tris, 1 mg proteinase K stock per ml buffer). Slides were washed three times for five min in PBS and equilibrated with buffer for 15 min. Staining solution containing the FITC-labeled nucleotide mix and TdT in equilibration buffer was added to each slide for 80 min. The enzymatic reaction was stopped, and slides were counterstained with a propidium iodide/anti-fade DNA intercolating counterstain solution (Oncor, Gaithersburg, MD). Slides were photographed using a Zeiss Axioskop2 wide-field fluorescence microscope with appropriate filter set (Zeiss, Welwyn Garden City, U.K.).
Flow cytometric analysis for detection of TNFR1 receptor
Splenocytes were prepared by needle dissection in RPMI 1640 supplemented with 10% FCS and washed once in ice-cold PBS before being resuspended in 5 ml of ice-cold PBS buffer. One hundred thousand splenocytes were stained with 20 ng biotinylated rhTNF-α, biotinylated control protein, or 100 molar excess of unlabeled rhTNF-α followed by biotinylated rhTNF-α for 1 h at 4°C, as detailed in the protocol provided in the Fluorokine kit (R&D Systems, Minneapolis, MN). FITC-conjugated streptavidin was added to all tubes for 30 min, and then cells were washed twice. In addition, cells were stained with APC-labeled anti-mouse CD3 mAb (PharMingen) for 15 min. Cells were then washed once, and samples were stained with the vital dye 7-aminoactinomycin D (7-AAD) (Molecular Probes, Eugene, OR) for 30 min at a concentration of 1 μg/ml. Samples were analyzed using a FACScalibur (Becton Dickinson, San Jose, CA). Data were analyzed by gating on CD3-negative and 7-AAD-negative viable cells using CellQuest software (Becton Dickinson).
Hepatocyte isolation and culture
Murine hepatocytes were isolated using a modification of the method described by Klaunig et al. (24) with viability exceeding 90%, as assessed using trypan blue staining. Briefly, mice were anesthetized with sodium pentobarbital (50 mg/kg body weight) and an incision was made through the skin on the ventral midline. The viscera was displaced to reveal the inferior vena cava and portal vein. A ligature was placed around the vena cava posterior to the renal veins and gently tied after introduction of a 24-gauge catheter into the vena cava distal to the ligature. A perfusion buffer (Krebs Ringer containing 20 mM glucose and 0.2 mM EGTA, pH 7.4) was initiated at 2 ml/min. The portal vein was then quickly severed, and the anterior vena cava was clamped between the diaphragm and heart. The perfusion rate was increased gradually to 7 ml/min for a total volume of 50 ml. The perfusion was continued with digestion buffer (Krebs Ringer containing 20 mM glucose, 1.37 mM CaCl2, 0.72% BSA fraction V (Sigma) and 100 U/ml type I collagenase (Sigma), pH 7.4) for another 50 ml total volume. The liver was then excised, and the capsule was ruptured to release cells. The cells were further dispersed by aspiration through a large bore pipet. Hepatocytes were isolated by centrifugation at 50 × g for 5 min at 4°C. Cells were washed three times with ice-cold culture medium (DMEM with 10% FBS and 1% penicillin and streptomycin). The cells were then plated at a density of 500,000 cells/ml on culture dishes coated with 20 μg per ml of type III acid soluble calf skin collagen (Sigma) in either 100 μl volume for 96-well plates for MTT toxicity assay or 3 ml volume for six-well plates for caspase-3-like activity determination. After resting for 8 h, cells were treated by replacement of original culture media with new media with or without d-galactosamine added at various concentrations. Cells were then incubated for 30 min before addition of 1 μg/ml rhTNF-α. Cells were incubated for 6 h for caspase-3-like activity assay or for 20 h for MTT cytotoxicity assay. Alternatively, some cells were treated with the Fas agonist Ab, Jo-2, or isotype control Ab at a concentration of 1 μg/ml. Cytotoxicity and caspase-3-like activity were determined.
MTT cytotoxicity assay
Formazan production from MTT was measured after incubation of cells with 0.6 mg/ml MTT in 100 μl culture media per well in a 96-well plate for 1.5 h. Medium was then removed, and cells were lysed in 100 μl 2-propanol. Next, 100 μl water were added, and plates were read in an ELISA reader at 560/690 nm.
LPS/d-galactosamine-induced lethality and TNF-α production in NOD and B6 mice
NOD and B6 mice were injected i.p. with d-galactosamine (8 mg) and increasing doses of LPS (1 ng–100 μg), and blood was obtained at 90 min for serum TNF-α determinations. Mice were monitored for up to 72 h, and percent mortality was determined. LPS and d-galactosamine treatment produced mortality in B6 mice in a dose-dependent manner (Fig. 1). Increasing the dose of LPS from 1 ng to 10 ng and higher produced 100% percent mortality. Mice started to die within 8 h after LPS administration, and mortality generally occurred within 12–24 h. However, d-galactosamine-sensitized NOD mice were resistant to at least 100-fold higher concentrations of LPS. Lethality was not seen until doses of LPS approached 1 μg, and even at doses of 100 μg, only 50% mortality was seen. Mortality in NOD mice, when it occurred, followed a similar time course with the majority of animals dying within 24 h. Physiologic saline or d-galactosamine (8 mg) treatments alone produced no mortality in either strain (data not shown).
Jacob et al. (12) reported that peritoneal macrophages from NOD mice have a reduced production of TNF-α when stimulated with LPS and IFN-γ. It was also shown that NOD mice produced lower levels of TNF-α than did most other strains, even though TNF-α gene expression was similar among these strains. To determine whether the differences in survival could be explained by a differential TNF-α production, serum TNF-α concentrations were determined at 90 min, a time period previously shown to represent the peak serum appearance after LPS administration (25). As shown in Fig. 2, peak serum TNF-α concentrations in B6 and NOD mice treated with 100 ng or 1 μg LPS and 8 mg d-galactosamine were not significantly different, although they were modestly reduced in the NOD mouse at the lower dose of LPS.
Apoptosis and liver toxicity in LPS/d-galactosamine treated NOD and B6 mice
Mice injected with 100 ng LPS and 8 mg d-galactosamine were sacrificed at 6 h after blood was collected. This time was chosen, based on our previous work, because it preceded death but was a time period associated with significant liver injury (25, 26). Livers were harvested; one lobe was homogenized for caspase-3-like activity, and another lobe was fixed in 10% buffered formalin for sectioning. Liver injury assessed by 6-h serum transaminase levels revealed ∼75% less liver-specific enzyme appearance (p < 0.05) in the blood of NOD compared with B6 mice (Fig. 3,A). Although caspase-3-like activity at 6 h was increased in NOD mice, the levels were significantly lower (287 ± 11 vs 194 ± 11 relative fluorescence intensity, p < 0.05) than those in livers from B6 mice (Fig. 3 B). d-Galactosamine treatment alone did not appear to increase caspase-3-like activity in the liver. Increased caspase-3-like activity was also not readily observed in spleens from LPS and d-galactosamine-treated or d-galactosamine alone-treated animals (data not shown).
In situ TUNEL assays also revealed that NOD mice had markedly fewer positive staining apoptotic cells in liver sections than were found in livers from B6 mice (Fig. 4). Histopathologic analysis of hematoxylin and eosin-stained liver sections from LPS and d-galactosamine-treated B6 mice revealed pervasive destruction of liver architecture and extensive apoptosis. In contrast, the degree of apoptosis and destruction of liver architecture was reduced, albeit not eliminated, in NOD mice. Livers of d-galactosamine alone-treated B6 and NOD mice were phenotypically comparable to saline-injected animals.
rhTNF-α and d-galactosamine toxicity in NOD and B6 mice
Although peak serum TNF-α concentrations did not significantly differ between NOD and B6 mice, the possibility existed that differences in the production of TNF-α may have explained the differences in survival and apoptotic liver injury. Furthermore, by taking advantage of the property that human TNF-α binds preferentially to the mouse TNFR1 receptor (18, 19), administration of human TNF-α to d-galactosamine-sensitized mice offers the opportunity to identify survival responses dependent primarily on TNFR1 receptor signaling. B6 mice treated with d-galactosamine were sensitive to rhTNF-α with doses as low as 1 μg, producing >50% lethality, and 10 μg human TNF-α, producing 100% lethality (Fig. 5).
In contrast, NOD mice were resistant to human TNF-α and d-galactosamine-induced lethality. Doses as high as 100 μg produced no mortality in NOD mice. Higher doses could not be evaluated because of the large amounts of recombinant protein required.
Liver Injury and apoptosis in rhTNF-α and d-galactosamine-treated NOD and B6 mice
Caspase-3-like activity in livers from B6 mice treated with 10 μg rhTNF-α and d-galactosamine was significantly higher than was seen in NOD mice (Fig. 6,B), and serum transaminases were also significantly elevated (both p < 0.05) (Fig. 6,A). Apoptotic liver injury was also confirmed histologically (Fig. 7), and by in situ TUNEL staining. Although NOD mice were protected from mortality to recombinant human TNF-α and d-galactosamine at the doses evaluated, they did exhibit some hepatic injury, hepatocyte apoptosis, and increased caspase-3-like activity.
Flow cytometric analysis of p55 TNF receptor in NOD and B6 mice
One possible explanation for the differential response between NOD and B6 mice may be the inability of recombinant human TNF-α to bind to the TNFR1 receptor on NOD cells. To directly evaluate this question, binding of biotinylated-rhTNF-α to NOD and B6 splenocytes was analyzed using a FITC-conjugated streptavidin secondary label followed by flow cytometric analysis. Fig. 8 clearly shows comparable binding of rhTNF-α to CD3 and 7-AAD double-negative splenocytes from NOD and B6 mice (mean fluorescence intensity (MFI) = 74.99 and MFI = 100.00, respectively). This binding was effectively blocked using a 100 molar excess of unlabeled rhTNF-α. No significant differences in binding of biotinylated rhTNF-α were seen in various cell populations examined in other studies, including CD4+ splenocytes or peritoneal macrophages (data not shown).
Jo-2-induced hepatic injury and lethality in B6 and NOD mice
Because the TNFR1 receptor shares with Fas (Apo1, CD95) an intracellular signaling pathway that converges at the level of Fas-associated protein with death domain (FADD) and caspase-8, the responsiveness of NOD mice to a Fas agonist was examined. NOD and B6 mice were challenged with increasing doses (1–10 μg) of the Fas agonist Ab (Jo-2) known to induce hepatocyte apoptosis and death (20, 21). Both NOD and B6 mice were sensitive to the lethal effects of Jo-2. In fact, NOD mice appeared to be more sensitive to the lethal effects than B6 mice, although this did not reach statistical significance (note that the x-axis on Fig. 9 is a linear scale, whereas the x-axis in Figs. 1 and 5 are logarithmic scales).
With a dose of 10 μg Jo-2 Ab, mice had to be sacrificed at 3 h, because death occurred in both groups between 4 and 8 h. At 3 h, serum transaminase levels (Fig. 10,A) capase-3-like activity (Fig. 10,B), and histologic evidence of apoptotic liver injury (Fig. 11) were similar in both NOD and B6 mice.
In vitro hepatocyte response to rhTNF-α and d-galactosamine treatment
To determine whether the differences in sensitivity to rhTNF-α/ d-galactosamine treatment between the B6 and NOD mice were due to differences in hepatocyte responsiveness, primary hepatocytes were isolated and cultured in the presence of rhTNF-α and d-galactosamine. Hepatocytes were examined for MTT toxicity and caspase-3-like activity. NOD hepatocytes were less sensitive to the toxic effects of rhTNF-α/d-galactosamine when compared with B6 hepatocytes (Fig. 12,A). At a dose of 0.1 mM d-galactosamine, NOD and B6 hepatocytes showed no difference in cell viability, as determined by the MTT toxicity assay. However, at doses of 1, 2.5, and 5 mM d-galactosamine with 1 μg/ml rhTNF-α, NOD hepatocytes showed little cytotoxicity when compared with B6 hepatocytes. In addition, caspase-3-like activity was also increased in the B6 hepatocytes and was significantly different from the NOD at doses of 2.5 and 5 mM d-galactosamine with 1 μg per ml rhTNF-α (Fig. 12 B).
In vitro hepatocyte response to Jo-2 treatment
The responsiveness of B6 and NOD hepatocytes to a Fas agonist was also examined. Hepatocytes from NOD and B6 mice were isolated and cultured in the presence of 1 μg/ml Jo-2 Ab or isotype control. As shown in Fig. 13, no difference in sensitivity to Jo-2 was seen in hepatocytes from either strain of mice treated in vitro with Jo-2.
The present study documents for the first time the observation that NOD mice sensitized with d-galactosamine are resistant to the lethal effects of LPS and TNF-α. This resistance appears to be due to a postreceptor defect in TNF-α signaling through the TNFR1. LPS and d-galactosamine administration produces a primarily TNF-α-dependent model of fulminant hepatocellular apoptotic injury and lethality, unlike the shock and necrotic liver injury found in high dose LPS models (27, 28, 29). In fact, high dose LPS-induced lethality was mechanistically different from the models using sensitization with transcriptional inhibitors (16). Lethality from LPS and d-galactosamine treatment is secondary to caspase-3-mediated apoptotic liver injury, as demonstrated by the ability of synthetic caspase inhibitors to block the apoptosis and lethality seen in mice treated with LPS and d-galactosamine (16, 27). However, these inhibitors do not prevent lethality to high dose LPS, nor do they block the early inflammatory responses to LPS and d-galactosamine treatment, indicating that toxicity in this model is independent of the systemic inflammatory response (16, 30).
The use of recombinant human TNF-α as opposed to murine TNF-α in experimental models with d-galactosamine allows for the specific targeting of the TNFR1 receptor due to species specificity of human vs murine TNF-α (18, 19, 31). Rothe et al. (32) demonstrated that mice lacking a functional TNFR1 receptor were resistant to LPS only when sensitized with d-galactosamine. It has also been shown that the TNFR1 and not the TNFR2 receptor is necessary and sufficient for TNF-α-mediated hepatic apoptotic injury and lethality in this model (26). Conversely, Morikawa et al. (15) observed that lpr mice lacking a functional CD95/Fas were not protected from LPS and d-galactosamine induced injury or lethality, indicating that the presence of Fas/FasL signaling is not required.
These studies with rhTNF-α, however, cannot localize the specific defect or defects in the TNF signal transduction pathway necessary to explain the resistance in NOD mice. The defect appears to be a postreceptor event, because binding of recombinant human TNF-α to NOD splenocytes appeared normal. Furthermore, because levels of caspase-3-like activity in the livers of NOD mice treated with LPS or TNF-α and d-galactosamine were markedly reduced, but not absent, the defect in apoptosis appears to occur distal to TNF-α binding to the TNFR1, but perhaps proximal to or at the level of activation of caspase-3. These data indicate that the ability of NOD mice to signal via the TNFR1 receptor is not completely absent, in that some increased caspase-3-like activity is apparent even though the animals survived.
However, an unexpected observation was that although NOD mice were resistant to lethality and hepatocyte injury induced by TNFR1 signaling, the animals were sensitive to Fas-mediated lethality and hepatocyte injury. The onset of injury and lethality in NOD and B6 mice treated with Jo-2 were similar, indicating that the Fas pathway was functional in NOD mice. Fas-induced cell death is thought to occur more directly than TNFR1-induced death due to the simultaneous activation of NF-κB and other antiapoptotic proteins that inhibit the TNFR1 death pathway (33).
Additionally, we have shown that primary hepatocytes isolated from NOD mice are less sensitive to in vitro TNF/d-galactosamine-induced apoptosis and toxicity than are B6 hepatocytes, indicating that hepatocytes from NOD mice are indeed resistant to this treatment. Also, NOD hepatocytes isolated and cultured in vitro were equally sensitive to Jo-2 when compared with B6 mice, confirming the absence of a defect in the Fas signaling pathway in hepatocytes from NOD mice. The differential sensitivity to TNF-α and Fas agonists in NOD mice may, however, allow for identification of potential defects in the TNFR1 receptor signaling pathway which explain resistance in NOD mice. The TNFR1 and Fas death receptors both contain an intracellular death domain allowing for recruitment of death effector molecules and execution of the apoptotic cascade in the signaling cell. For instance, the Fas pathway involves binding of the FADD to the death domains of the trimerized Fas receptors. This FADD adaptor protein contains a death effector domain at the N terminus that is responsible for the recruitment of caspase-8 to the death-induced signaling complex (34). Additionally, FADD has also been shown to bind to TNFR-associated protein with death domain (TRADD), a proximal mediator in the TNFR1 receptor signaling pathway, resulting in activation of caspases and other death effector molecules culminating in apoptotic cell death (35, 36). Therefore, the adaptor protein FADD represents the convergence point of Fas and TNFR1 receptor pathways.
The current data indicate that the Fas pathway is functional in livers of NOD mice. Because FADD binds to both the Fas receptor and TRADD via the same death domain, the defect seen in the TNFR1 receptor signaling pathway of NOD mice is likely to be independent of intermediates common to the Fas pathway, but rather proximal to the convergence at FADD and caspase-8. The experimental data are therefore consistent with a defect in either the signaling capacity of the p55 receptor or the ability of its death domain to form a signaling complex with TRADD, FADD, RIP or other intracellular signaling molecules. An alternative explanation for the reduced apoptotic injury in hepatocytes from NOD mice treated with LPS or TNF-α and d-galactosamine is that NF-κB activation is increased, resulting in greater expression of NF-κB-dependent “survival genes” (37). This appears to be less likely, because preliminary evidence indicates no increase in nuclear NF-κB translocation in livers from NOD mice treated with LPS and d-galactosamine, as determined by EMSA (data not shown).
Future investigations need to examine the structure and functional relationships in the TNFR1 death-induced signaling complex as a potential site for the lack of responsiveness in NOD mice. One difficulty in examining these pathways is that although TRADD-FADD and TNFR1-TRADD complexes were found by Hsu et al. to be very stable, the trimer of TNFR1-TRADD-FADD were thought to be only transient and unstable without other proteins such as TNFR-associated factors being present. In addition, although they are members of the same superfamily of molecules, Fas and TNFR1 have both shared and unshared signaling mechanisms. Historically, ideas about the molecular events activated by these receptors have been disputed because techniques used to acquire this information largely involve artificial systems. Therefore, apoptosis via TNFR1 and Fas receptors occurs due to a complex set of molecular events, making it difficult to determine where a defect, when present, may lie.
The observation, however, that NOD mice are resistant to the hepatocyte injury and lethality associated with TNFR1 receptor signaling has significant implications for other TNFR1 receptor-dependent processes, such as apoptosis of lymphoid cell populations. Further studies are required to determine whether T and B cell populations from NOD mice are resistant to TNF-α-mediated apoptosis.
We thank Dr. Tadahiko Kohno from Amgen for the gift of rhTNF-α. Without his generosity, we could not have performed these and future studies. We also thank Dr. Dave Serreze from The Jackson Laboratory (Bar Harbor, ME) for his support, guidance, and helpful discussions. In addition, we would like to thank Perry Bain and Keith Bahjat for assistance with hepatocyte isolation.
This work was supported in part by Grants GM-40586-12 and GM-61807-01 awarded by the National Institute of General Medical Sciences, U.S. Public Health Service.
Abbreviations used in this paper: NOD, nonobese diabetic; rhTNF-α, recombinant human TNF-α; B6, C57BL/6J; 7-AAD, 7-aminoactinomycin D; APC, allophycocyanin; TNFR1, TNF type 1 (p55) receptor; AST, aspartate aminotransferase; FADD, Fas-associated protein with death domain; TRADD, TNF receptor-associated protein with death domain; MTT, 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide; MFI, mean fluorescence intensity.