Expression of non-signaling membrane-anchored death receptors protects murine livers in different models of hepatitis

Authors


  • Potential conflict of interest: Nothing to report.

  • See Editorial on Page 314

Abstract

Fas and tumor necrosis factor receptor 1 (TNFR1) are death receptors involved in various diseases such as hepatitis, sepsis, or graft rejection. Neutralizing antibodies to death ligands or soluble death receptors can inhibit cell death; however, they induce side effects because of their systemic actions. To specifically block death signaling to target cells, we created death domain–deficient (ΔDD) membrane-anchored receptors, delivered to the liver by either recombinant adenovirus or hydrodynamic pressure of nonviral recombinant plasmids. In anti-Fas antibody-induced fulminant hepatitis, mice expressing recombinant Fas-decoy receptors (FasΔDD) in their livers were completely protected against apoptosis and survived fulminant hepatitis. In T-cell–dependent concanavalin A–induced autoimmune hepatitis, FasΔDD antagonist expression prevented hepatocyte damage and mouse death. Finally, TNFR1ΔDD effectively protected mice against LPS-induced septic shock. In conclusion, such ΔDD-decoy receptors act as dominant-negative receptors exerting local inhibition, while avoiding systemic neutralization of apoptosis ligands, and might have therapeutic potential in hepatitis. (HEPATOLOGY 2006;44:399–409.)

Tumor necrosis factor-α (TNF-α), Fas ligand (FasL) and TNF-related apoptosis-inducing ligand (TRAIL) belong to the TNF family. TNF-α is primarily produced by inflammatory cells during the innate immune response.1 In lymphoid tissue, FasL is predominantly expressed on activated T lymphocytes and natural killer (NK) cells, and plays essential roles in their apoptotic activity and the downsizing of the T-lymphocyte pool that expanded during an immune response.2 TRAIL, which is expressed on NK and T cells, is involved in defense against tumors and virus-infected cells.3 These three ligands, whose expressions are tightly regulated, are trimeric molecules able to trigger apoptosis after binding to membrane death receptors constitutively expressed on different target tissues. FasL and TNF-α bind pre-associated Fas receptor4 and TNF receptor 1 (TNFR1)5 trimers, respectively, and recent reports suggest that TRAIL also binds to pre-associated receptors.3, 6 Engagement of these death receptors induces their clustering at the cell membrane and the recruitment, through a cytoplasmic region called the death domain (DD), of Fas- or TNFR1-associated adaptor proteins, respectively. Then, caspase-8 or -10 recruitment and constitution of a death-inducing signaling complex lead to caspase-dependent apoptosis.1

FasL and TNF-α are involved in different pathological processes, such as autoimmune diseases,7 hepatitis,8 sepsis,9 and graft rejection.10 Therefore, several strategies have been applied to inhibit in vivo death receptor– induced diseases: pharmacological agents,11 antisense oligonucleotides,12 neutralizing antibodies to death ligands,13, 14 or soluble death receptors.15 Thus, infliximab, a monoclonal antibody directed against TNF-α or etanercept, a recombinant soluble TNFR2, are currently used to treat rheumatoid arthritis.16 In addition, targeting the pre-ligand assembly domain of receptors belonging to the TNFR superfamily was recently proven to effectively treat arthritis.17 However, all these therapies block cell death systemically, thereby generating side effects. Hence, vasculitis, tuberculosis, and even heart failure were reported in patients given infliximab.18 Recently, small interfering RNA (siRNA) targeting Fas or caspase-8 were used as an alternative therapeutic approach to inhibit hepatocyte apoptosis in mice19, 20; however, incomplete silencing of the targeted gene might constitute a major drawback to their use in humans. Moreover, siRNA may stimulate interferon-α production, resulting in nonspecific down-regulation of the transcription of various genes.21

Apoptosis is regulated in vivo by controlling the expression level(s) of death ligands or death receptors, and inducing expression of decoy receptors. For example, during inflammatory processes, TNF-α–induced TNFR1 and TNFR2 shedding protects against excessive TNF-α synthesis.22 Likewise, once at the cell surface, FasL is rapidly cleaved by a metalloprotease to prevent nonspecific killing by lymphoid cells.23, 24 In addition, a soluble Fas-decoy receptor (DcR3) is able to bind FasL and prevent FasL-induced apoptosis.25 Also, a Fas variant resulting from alternative splicing of Fas mRNA has been reported; it has a complete extracellular domain but is DD-deficient (ΔDD) and thus inhibits death signaling.26 Two TRAIL-decoy receptors have been described: DcR1, which is ΔDD and anchored to the target membrane via a glycophosphatidyl inositol, and DcR2, which has a truncated nonfunctional DD.3 Such surface decoy receptors control apoptosis by competing with their ligands' binding to death receptors and via heteromeric interactions between decoy receptors and death receptors.6

Taken together, those observations prompted us to investigate whether gene transfer of ΔDD receptors could potentially block apoptosis pathways locally in vivo, while avoiding systemic neutralization of apoptosis ligands. Using different models of hepatitis, we examined, after the gene had been delivered to the liver with a recombinant adenovirus vector (Ad) or by hydrodynamic pressure (HP) injection, the abilities of ΔDD-TNFR1 and -Fas death receptors to protect mice against liver failure.

Abbreviations

TNF-α, tumor necrosis factor alpha; FasL, Fas ligand; TRAIL, TNF-related apoptosis-inducing ligand; NK, natural killer; TNFR1, type 1 tumor necrosis factor receptor; ΔDD, death domain-deficient; siRNA, short interfering RNA; Ad, adenovirus; HP, hydrodynamic pressure; PCR, polymerase chain reaction; pfu, particle-forming unit; PARP, poly(ADP-ribose) polymerase; FITC, fluorescein isothiocyanate; PBS, phophate-buffered saline; LPS, lipopolysaccharide; ConA, concanavalin A; TUNEL, terminal deoxynucleotidyl transferase-mediated nick-end labeling; ALT, alanine aminotransferase; AST, aspartate aminotransferase; MOI, multiplicity of infection; NF-κB, nuclear factor-kappaB; HES, hematoxylin-eosin-saffron; D-GalN, D-galactosamine.

Materials and Methods

Ad Production.

Murine FasΔDD or TNFR1ΔDD receptors lacking a C-terminal region containing the DD (amino acids 192–306 or 334–433, respectively) were constructed by polymerase chain reaction (PCR) amplification from pMF1 (provided by Dr S. Nagata, Osaka Bioscience Institute, Japan) with primers 5′-GCGCAAGCTTGCTGCAGACATGCTGTGGATCTGGG-3′ and 5′-GCGCGGATCCTTATATGGTTTCAACGACTGGA-3′ or BALB/c cDNA with primers 5′-GCGCAAGCTTACTGCCGGCCGGACATGGGTC-3′ and 5′-CGCGCTCGAGTCAGTCTGCATTGTCAGGACGTTG-3′. FasΔDD and TNFR1ΔDD cDNA were cloned into pCEP4 vector (Stratagene, Amsterdam, The Netherlands) under the control of the human cytomegalovirus immediate early promoter and simian virus 40 polyA signal, and Ad genomes (Ad5ΔE1ΔE3) were obtained.27 Corresponding viruses AdFasΔDD and AdTNFR1ΔDD were synthesized after transfection of 293 cells (CRL-1573, ATCC, Illkirch, France). AdFas-Fc and AdTNFR1-Fc encode soluble receptors constituted by fusion of the murine Fas or TNFR1 extracellular domain with murine IgG1 Fc region. An Ad expressing no transgene28 (AdControl) served as a control of virus infection. All viruses were purified by two-step CsCl-gradient ultracentrifugation. Viruses were quantified as particle-forming units (pfu)/mL after infection of 911 cells.29

Expression Plasmids.

FasΔDD and Fas-Fc cDNA were cloned into pVax (Invitrogen, Cergy-Pontoise, France) and purified using Nucleobond PC 2000 Endofree Kit (Macherey-Nagel, Hoerdt, France).

Western Blotting.

HeLa cells or liver infected with Ad were lysed in RIPA buffer (150 mmol/L NaCl, 1% NP-40, 0.5% DOC, 0.1% SDS, 50 mmol/L Tris-HCl, pH 7.8) and anti-proteases (Roche Diagnostics, Meylan, France). Protein lysates were run in a NuPAGE (Novex, San Diego, CA), transferred onto nitrocellulose membranes, and probed with antibodies to murine Fas (R&D Systems, Abinghom, UK), murine TNFR1 (R&D Systems), or poly(ADP-ribose) polymerase (PARP, R&D Systems).

Immunofluorescence.

Cytospun slides of AdFasΔDD- or AdTNFR1ΔDD-infected HeLa cells [multiplicity of infection (MOI): 250 pfu] were fixed with 4% paraformaldehyde and labeled with goat anti-Fas or -TNFR1 antibodies, followed by a fluorescein (FITC)-conjugated anti-goat IgG (Jackson ImmunoResearch, Cambridgeshire, UK). Slides were mounted in Vectashield mounting medium (Vector Laboratories Inc., Burlingame, CA) containing 4′,6-diamidino-2-phenylindole.

FACS Analysis.

HeLa or Hepa 1-6 cells were infected at different MOI (range: 10-400 pfu/cell) with AdFasΔDD, AdTNFR1ΔDD or AdControl. To determine Fas and FasΔDD levels, cells were labeled with phycoerythrin-conjugated anti-murine (BD Pharmigen, San Diego, CA) or -human (Beckman Coulter, Villepinte, France) antibodies and their isotype-control antibodies (BD Pharmingen). To determine TNFR1ΔDD levels, cells were labeled with anti-TNFR1 antibodies or anti-mouse IgG (Southern Biotechnology Associates Inc., Birmingham, AL), followed by an FITC-conjugated anti-goat IgG (Jackson ImmunoResearch). Human TNFR1 levels were detected using an FITC-conjugated anti-human TNFR1 (R&D Systems) or with an isotype-control antibody (BD Pharmingen). FACS analysis was performed using a FACScan flow cytometry instrument with CellQuest software (BD Biosciences, Le Pont de Claix, France).

Cell Viability Assay.

HeLa or Hepa 1-6 cells were incubated with different AdFasΔDD or AdTNFR1ΔDD MOI for 24 hours, and then challenged with different concentrations of human FasL (Alexis, Lausanne, Switzerland) or mouse TNF-α (Perbio Sciences, Brebières, France), in the presence of actinomycin D (1 μmol/L, Sigma-Aldrich, Lyon, France) for 20 hours at 37°C. Cell viability was assessed by MTT assay (Sigma).

Nuclear Factor-κB Luciferase Reporter Assay.

293T cells (4 × 105 cells) stably transfected with a luciferase reporter construct regulated by 6 copies of the NF-κB response element (293T/NF-κB-luc cells, Panomics, Redwood City, CA) were not infected or infected with AdTNFR1ΔDD or AdControl at the MOI of 1 and 10 pfu/cell. TNF-α was added 24 hours later, and the cells were incubated for an additional 12 hours. Cells were lysed, and luciferase activities were measured according to the Luciferase Assay System (Promega, Madison, WI) on LUMAT LB9501 (Berthold, Thoiry, France). Protein content was determined using the Bio-Rad Protein Assay (Bio-Rad, Hercules, CA).

Murine Models of Hepatitis and Septic Shock.

All animal studies respected institutional guidelines. Mice (13–17-week-old BALB/c females, Janvier, Le Genest-Saint-Isle, France) were injected with 2.5 × 109 Ad pfu intravenously or received 50 μg of expression plasmids by HP.30 Five or 2 days later, respectively, hepatitis was induced by either intravenous injection of 0.25 to 0.5 μg/g Jo2 monoclonal antibody (BD Pharmigen) in phosphate-buffered saline (PBS) or 10 to 15 μg/g concanavalin A (ConA, Sigma) in PBS. For the sepsis model, mice were injected intraperitoneally with 800 μg/g D-GalN (Sigma) plus 0.05 μg/g lipopolysaccharide (LPS; Salmonella typhimurium, Sigma). In both models, mice were observed for 24 hours and livers from killed mice were frozen for protein analysis and fixed in GlyoFix (Shandon, Pittsburgh, PA) for immunohistological analyses. Sera were obtained at different times after hepatitis or sepsis induction.

ELISA Quantification of Soluble Antagonists.

Serum Fas-Fc and TNFR1-Fc concentrations were determined by sandwich ELISA. Plates were coated with anti-Fas or -TNFR1 antibody, then incubated with serum dilutions, followed by alkaline phosphatase-conjugated anti-mouse IgG (Promega). Fas-Fc or TNFR1-Fc purified from serum-free culture supernatants of AdFas-Fc- or AdTNFR1-Fc–infected HeLa cells were used as standards.

Real-Time Quantitative RT-PCR.

Livers were homogenized in lysing matrix tubes containing Trizol reagent (Invitrogen) using the FastPrep system (Qbiogene, Illkirch, France). After RNase-free DNase treatment (Qiagen, Courtaboeuf, France), total liver RNA (1 μg) was reverse transcribed (Reverse Transcription System, Promega). Quantitative real-time PCR was performed using murine Fas-specific TaqMan MGB probe (FAM dye-labeled) and unlabeled primers (Assays-on-Demand, Applied Biosystems, Courtaboeuf, France). The reaction conditions were as follows: 2 minutes at 50°C (1 cycle), 10 minutes at 95°C (1 cycle), and 15 seconds at 95°C and 1 minute at 60°C (40 cycles). Gene-specific (Fas and FasΔDD) PCR products were continuously measured with the ABI PRISM 7900HT detection system (Perkin-Elmer, Boston, MA), normalized to control ribosomal 18S RNA and expressed relative to Fas level measured in PBS-treated mice.

Liver Histology.

Sections (5-μm thick) of livers fixed and embedded in paraffin were stained with hematoxylin-eosin-saffron (HES). Fas was immunodetected using a goat anti-Fas antibody and a biotin-streptavidin-peroxidase–conjugated rabbit anti-goat antibody. Terminal deoxynucleotidyl transferase-mediated nick-end labeling (TUNEL) labeling was performed with in situ Cell-Death Detection Kit (Roche Diagnostics).

Aminotransferase Assay.

Serum alanine aminotransferase (ALT) and aspartate aminotransferase (AST) activities were determined using Enzyline ALT/GPT or AST/GOT Kits (bioMérieux, Marcy l'Etoile, France).

Statistical Analysis.

A one-way ANOVA with Fisher's least significant difference test was conducted.

Results

Blockage of Cell Death by Membrane Expression of Dominant-Negative ΔDD Receptors.

After modifying murine Fas and TNFR1 cDNA by truncating their cytoplasmic DD regions to the C-terminal end, we produced recombinant adenoviruses AdFasΔDD and AdTNFR1ΔDD, respectively. As documented by Western blot analysis and compared to not infected (NI) or control Ad (AdControl)-infected HeLa cells, lysates of HeLa cells infected with these vectors showed the expression of FasΔDD and TNFR1ΔDD proteins at their respective molecular masses of 22 and 37 kd (Fig. 1A). Moreover, immunofluorescence labeling detected strong surface expression of FasΔDD or TNFR1ΔDD on HeLa cells after transduction with AdFasΔDD or AdTNFR1ΔDD, respectively. Finally, FACS analyses indicated that HeLa cells, bearing constitutively human Fas and TNFR1 receptors (Fig. 1C, upper panel), expressed decoy receptors at their cell surface on infection with AdFasΔDD or AdTNFR1ΔDD (Fig. 1C, lower panel). In both Western blot and FACS analyses, ΔDD-receptor expression increased with MOI.

Figure 1.

Cell surface expression of death domain–deficient Fas (FasΔDD) and TNFR1 (TNFR1ΔDD) antagonists. HeLa cells were not infected (NI) or infected at different multiplicity of infection (MOI, expressed in particle-forming unit [pfu] per cell) with recombinant adenovirus (Ad) AdControl, AdFasΔDD, or AdTNFR1ΔDD. FasΔDD or TNFR1ΔDD expression was assessed by Western blotting (A), immunolabeling (B, original magnification, ×600, MOI = 250 pfu), and FACS analysis using murine specific anti-Fas or -TNFR1 antibodies (C, lower panel). Endogenous Fas and TNFR1 expression was detected by human specific anti-Fas or -TNFR1 antibodies (C, upper panel).

To assess the ability of either decoy to protect against apoptosis, HeLa cells were infected with AdFasΔDD or AdTNFR1ΔDD at different MOI or not infected and, 24 hours later, they were subjected to increasing doses of FasL or TNF-α. AdFasΔDD-infected HeLa cells survived the FasL challenge but AdTNFR1ΔDD- or not infected HeLa cells did not (Fig. 2A). In contrast, AdTNFR1ΔDD-infected HeLa cells were protected against the TNF-α challenge, whereas AdFasΔDD- or not infected HeLa cells were not (Fig. 2B). Increasing the MOI allowed an enhanced cell survival, thereby demonstrating its direct relationship with FasΔDD or TNFR1ΔDD expression levels. Using murine hepatoma cell line, Hepa 1-6, which expresses high levels of Fas (Fig. 2C) and TNFR1 receptors (not shown), we confirmed the ability of FasΔDD or TNFR1ΔDD to protect against FasL- and TNF-α–induced cell death (Fig. 2D and data not shown). Because TNF-α acting as a proinflammatory cytokine can signal through TNFR1 to activate the nuclear factor-κB (NF-κB) pathway, we examined the ability of TNFR1ΔDD to inhibit NF-κB signaling. Using human embryonic kidney 293T cells stably transfected with a luciferase gene under the control of NF-κB response elements, we observed, on addition of TNF-α, that TNF-α–induced luciferase transcriptional activation by NF-κB was markedly lower after infection with AdTNFR1ΔDD (Fig. 2E). Thus, FasΔDD or TNFR1ΔDD antagonists acted as dominant-negative receptors because they effectively competed with endogenous Fas receptor and TNFR1 and inhibited their signaling.

Figure 2.

Inhibition of FasL- and TNF-α–induced signaling by FasΔDD and TNFR1ΔDD antagonists. HeLa cells were infected at different multiplicity of infection [MOI, expressed in particle-forming unit (pfu) per cell] with AdFasΔDD or AdTNFR1ΔDD, or not infected (NI). One day later, twofold serial dilutions of either FasL (A) or TNF-α (B) were added, and the percentages of viable cells were determined by MTT assay. (C) FACS analysis of Fas and FasΔDD expression in Hepa 1-6 infected at different MOI with AdFasΔDD or AdControl. (D) Hepa 1-6 infected with AdControl or AdFasΔDD (MOI 200) were subjected 1 day later to FasL (400 ng/mL), and the percentages of viable cells were determined by MTT assay. (E) 293T cells expressing a reporter plasmid containing a luciferase gene under the control of NF-κB response elements were infected at different MOI with AdTNFR1ΔDD or AdControl, or not infected (NI), and challenged 24 hours later with TNF-α. Cell luciferase activity was measured after 12 hours. Results, representative of three experiments, are means ± SD.

FasΔDD Expression Protects Mice Against Jo2-Induced Fulminant Hepatitis.

We took advantage of the strong tropism of intravenously injected Ad for mouse liver to assess whether liver expression of FasΔDD antagonist could protect mice against Fas-dependent hepatitis. First, we used the model of anti-Fas monoclonal antibody (Jo2)-induced fulminant hepatitis.31 BALB/c mice were injected intravenously with PBS, AdControl, AdFasΔDD, or AdFas-Fc, a recombinant Ad driving the expression of a soluble fusion protein consisting of murine Fas extracellular domain and the mouse IgG1 Fc region. Five days later, mice were challenged with intravenous injection of Jo2, 5 or 10 μg, a dose currently used to induce fulminant hepatitis.19 All control mice (PBS and AdControl) died of fulminant hepatitis (liver failure) within 2 to 10 hours, whereas AdFasΔDD- and AdFas-Fc–treated mice were still alive 28 hours after Jo2 injection (Fig. 3A), and none died over the next few days (not shown). However, the slightly prolonged survival of AdControl-infected mice (vs. PBS) led us to include AdControl-treated groups in all experiments using recombinant Ad delivery to obtain a more accurate control setting.

Figure 3.

FasΔDD effectively protected mice against Jo2-induced fulminant hepatitis. Mice were treated with PBS, AdControl, AdFas-Fc, or AdFasΔDD 5 days before intravenous injection of monoclonal Jo2 antibody. (A) Mouse survival (n = 6-7/group) was monitored for 28 hours (one of two experiments is shown). (B) Representative liver histology of mice killed 3 hours after Jo2 challenge (HES staining, original magnification, × 100, arrows indicate hemorrhagic zones). (C) Immunohistochemistry detection of Fas-antagonists expression in liver sections from mice killed at the time of Jo2 injection (original magnification, ×200, arrows indicate labeled cells) or (D) Western blotting. (E) Serum alanine aminotransferase (ALT) activity (one of two experiments is shown, n = 2-4/group; means ± SD shown, *P < .05) measured 2 hours after Jo2 injection. (F) Liver sections from mice killed 3 hours after Jo2 injection were subjected to TUNEL assay (original magnification ×200). (G) Poly(ADP-ribose) polymerase (PARP) processing in liver lysates prepared 3 hours after Jo2 injection was analyzed by immunoblotting (upper line, 116-kd native form; lower line, 23-kd cleaved form; for each condition, results from two mice are given).

In a parallel experiment, mice were sacrificed 3 hours after Jo2 injection, and their livers were excised for histological analysis (Fig. 3B). HES staining showed several focal hemorrhages and necrotic zones in AdControl-treated mouse livers, whereas livers expressing AdFas-Fc or AdFasΔDD were protected against Jo2-induced liver damage. Protection against hepatitis correlated with high expression of soluble (mean ± SD: 304.5 ± 245.7 μg/mL in serum, as assessed by ELISA, n = 5) or membrane-anchored antagonists in the livers of AdFas-Fc– or AdFasΔDD-treated mice, as documented by immunohistochemistry (Fig. 3C) and Western blotting (Fig. 3D). Real-time reverse transcriptase polymerase chain reaction was used to quantify the Fas and FasΔDD mRNA levels in the livers of AdFasΔDD-injected mice at the time of Jo2 challenge: relative Fas mRNA levels were increased 11-fold in AdFasΔDD-treated mice versus AdControls (mean ± SD: 18.6 ± 7.6 vs. 1.7 ± 0.5, respectively, n = 4, P < .01). Even when the AdFasΔDD dose was 10-fold lower (2.5 × 108 pfu plus AdControl to obtain the same total virus dose injected), mice produced high levels of Fas mRNA (mean ± SD: 6.0 ± 1.6, P < .01), providing complete protection against Jo2-induced hepatitis.

Protection Against Hepatitis Reflects Inhibition of Hepatocyte Apoptosis.

To determine whether inhibition of apoptosis was responsible for mouse survival, serum ALT was measured. AdFasΔDD- and AdFas-Fc–treated mice, had no ALT increases (mean ± SD: 319 ± 235 U/L for AdFasΔDD), in contrast to AdControls (1841 ± 1479 U/L; n = 2-4/group, P < .05) (Fig. 3E), indicating the absence of hepatocyte injury. Moreover, TUNEL assay on liver sections showed very few apoptotic nuclei in AdFasΔDD- and AdFas-Fc–treated mice; however, apoptotic nuclei were abundant in AdControl-treated mouse livers (Fig. 3F). In accordance with TUNEL data, cleavage of poly(ADP-ribose) polymerase (PARP), a well-known substrate of activated caspases, was markedly lower, as assessed by Western blotting on liver lysates prepared from AdFasΔDD- and AdFas-Fc–treated mice, than PBS- or AdControl-treated mice (Fig. 3G), thereby demonstrating effective hepatocyte protection against Fas-induced death.

FasΔDD Expression Protects Cells Against Concanavalin A–Induced Hepatitis.

Concanavalin A (ConA), a T-cell mitogen, triggers hepatic damage, considered a model of human autoimmune hepatitis,32 by recruiting inflammatory cells in the liver. We examined the protective ability of FasΔDD in this model. BALB/c mice were first injected with AdFasΔDD, AdFas-Fc or AdControl and, 5 days later, ConA (10 μg/g body weight) was injected intravenously. Aminotransferase determinations and liver histology were carried out 24 hours later. AdControl-injected mice developed massive liver necrosis (Fig. 4A, encircled by dotted lines), which was not seen in mice expressing FasΔDD or Fas-Fc. The significantly lower serum ALT and AST concentrations in mice expressing FasΔDD and to a lesser extent Fas-Fc (8- and 3-fold lower for ALT, 6- and 4-fold lower for AST) confirmed Fas-antagonist protection against ConA-induced acute hepatitis (Fig. 4B). Notably, hepatocyte FasΔDD expression provided better protection than the soluble Fas-Fc antagonist. On challenge with a higher ConA dose (15 μg/g), all AdControl-treated mice died, whereas AdFas-Fc– and AdFasΔDD-treated mice survived (not shown).

Figure 4.

FasΔDD antagonists prevented concanavalin A (ConA)-induced acute hepatitis. ConA was injected 5 days after intravenous injection of AdControl, AdFas-Fc, or AdFasΔDD into mice (one of two experiments is shown). (A) Histological analysis of HES-stained liver sections 24 hours after ConA administration (original magnification, × 100). Dotted lines encircle the areas of liver necrosis. (B) Serum alanine (ALT) and aspartate (AST) aminotransferase activities 24 hours after ConA challenge (n = 4/group; means ± SD shown, *P < .05 and **P < .01).

Membrane-Anchored TNFR1 Decoys Protect Mice Against Septic Shock.

Septic shock is induced by bacterial release of endotoxins, which trigger intense production of proinflammatory cytokines, such as TNF-α that ultimately lead to multiorgan failure, including liver destruction.9 We showed (Fig. 2B) that TNFR1ΔDD inhibited TNF-α–dependent apoptosis of HeLa cells. After verifying that intravenous inoculation of AdTNFR1ΔDD into mice generated sufficiently high levels of TNFR1ΔDD antagonists 3 and 7 days later, as assessed by Western blot analysis (Fig. 5A), we determined the ability of this decoy to protect mice against TNF-α–dependent death. Mice were pretreated with AdTNFR1ΔDD, AdTNFR1-Fc (encoding a soluble TNFR1 antagonist), AdControl, or PBS; 5 days later, they were injected intraperitoneally with LPS (0.05 μg/g) and D-galactosamine (D-GalN, 800 μg/g), a selective transcriptional inhibitor in hepatocytes, to induce septic shock.33 Although 100% of mice expressing TNFR1ΔDD (Fig. 5A) or TNFR1-Fc (mean ± SD: 17.2 ± 8.5 μg/mL, detected in serum by ELISA, n = 8) survived the LPS challenge, the majority of AdControl- and PBS-treated mice died within 10 hours (Fig. 5B). Histological examination of organ sections taken 7 hours after LPS-D-GalN challenge from AdControl-injected mice showed vascular congestion of the red pulp in the spleen, recruitment of leukocytes into the lung (not shown), and destruction of liver architecture with presence of hemorrhagic zones (Fig. 5C). As expected, these mice had high ALT levels. In contrast, no liver destruction was seen in mice expressing TNFR1ΔDD or TNFR1-Fc, thereby attesting to the significantly lower ALT in TNFR1ΔDD- or TNFR1-Fc–treated mice (Fig. 5D). Taken together, these observations demonstrate that liver TNFR1ΔDD expression is able to fully protect mice against LPS-induced hepatitis.

Figure 5.

TNFR1ΔDD decoy receptors protected mice against endotoxin-induced death. Mice were injected with PBS, AdControl, AdTNFR1-Fc, or AdTNFR1ΔDD 5 days before challenge with intraperitoneally injected LPS and D-galactosamine (D-GalN). (A) Detection of TNFR1ΔDD expression by Western blotting on liver lysates 3 and 7 days after adenovirus administration. (B) Monitoring of mouse survival (one of two experiments is shown, n = 4-9/group). (C) Histological analysis of HES-stained liver sections 7 hours after LPS-induced sepsis (original magnification, ×100). (D) Serum alanine aminotransferase (ALT) activity 7 hours after challenge with LPS and D-GalN (n = 3-5/group; means ± SD, *P < .05 and **P < .01; data from two experiments).

Protection Against Hepatitis by Nonviral HP Delivery of Plasmids Encoding a Membrane-Anchored Decoy Receptor.

Using Ad-gene delivery, we demonstrated that membrane-anchored antagonists were able to specifically inhibit apoptotic pathways in different hepatitis models. To confirm that this protection against apoptosis would be effective regardless of the gene delivery route, we examined whether intravenous HP30 delivery of recombinant plasmids encoding FasΔDD could be protective. Mice received recombinant plasmids encoding FasΔDD, Fas-Fc or a control plasmid (pVax) by HP and, 2 days later, were challenged with Jo2.31 Whereas control mice injected with pVax died within 16 hours, 100% of the mice expressing FasΔDD or Fas-Fc survived the first 24 hours and 75% of FasΔDD-expressing mice were still alive 3 days later (Fig. 6A). Expression of soluble (mean ± SD: 15.9 ± 0.4 μg/mL in serum, as assessed by ELISA, n = 3) or membrane-anchored antagonists in the livers of mice injected with recombinant plasmids encoding Fas-Fc and FasΔDD was confirmed at the time of Jo2 injection, as documented by immunohistochemistry (Fig. 6B). Thus, using two gene-delivery routes, we showed that membrane-anchored antagonists protect against the development of hepatitis.

Figure 6.

Hydrodynamic pressure delivery of FasΔDD to the liver protected mice against hepatitis. (A) Mice were injected with recombinant plasmids encoding FasΔDD or Fas–Fc, or a control plasmid (pVax) 2 days before challenge with intravenous injection of Jo2, and their survival was monitored (one of two experiments is shown, n = 8/group). (B) Immunohistochemistry detection of Fas-antagonists expression in liver sections from mice killed at the time of Jo2 injection (original magnification, × 100)

Discussion

Apoptosis induced by death receptors belonging to TNFR family is involved in many pathological conditions, including hepatitis, ischemia/reperfusion, sepsis, or graft rejection. Most of the strategies developed to counteract apoptosis rely on the administration of soluble antagonists, for example, antibodies or soluble decoy receptors.14, 15 For example, systemic administration of anti–TNF-α antibody or soluble TNFR attenuated rheumatoid arthritis symptoms34; however, because such molecules neutralize apoptosis ligands throughout the body, they can have harmful side effects by interrupting physiological processes, as demonstrated by tuberculosis reactivation after anti–TNF-α therapy.18, 35 Also, on binding to transmembrane death ligands, these soluble antagonists may induce macrophage apoptosis by reverse signaling.36

To avoid such undesirable effects, we examined the potential of membrane-anchored decoy receptors to inhibit apoptosis. To do so, we constructed two antagonists, FasΔDD and TNFR1ΔDD, corresponding to ΔDD death receptors. Membrane expression of these receptors provided dose-dependent protection against in vitro FasL- or TNF-α–induced signaling (via the apoptosis pathway but also the inflammatory pathway for TNF-α) demonstrating their dominant-negative effect. We then determined that these decoys were able to inhibit hepatitis in different mouse models. Taking advantage of Ad's ability to target the liver after intravenous injection, we showed that hepatocyte FasΔDD expression completely prevented death in a murine fulminant hepatitis model. Moreover, FasΔDD prevented the massive liver damage associated with ConA-induced autoimmune hepatitis. In parallel, we demonstrated that mice expressing TNFR1ΔDD survived LPS-induced septic shock. In our hands, FasΔDD and TNFR1ΔDD receptors, whose expression is targeted to the hepatocyte membrane, protected mice against apoptotic liver damage as effectively as soluble antagonists (Fas-Fc and TNFR1-Fc). Our results clearly show that an approach based on DD deletion of TNFR-family receptors could effectively inhibit apoptosis in vivo; such a strategy could be a general way to obtain highly competitive death-receptor decoys exerting local inhibition.

However, the use of ΔDD-receptors raises several concerns. First, what level of receptor expression is required to block apoptosis signal transduction? In our experiments, using Ad-mediated gene delivery, we showed that 3-4–fold more FasΔDD mRNA than endogenous Fas mRNA effectively protected mice against apoptosis. However, high ΔDD-receptor expression may be not necessary because death receptors belonging to TNFR family must trimerize to transduce apoptosis. Indeed, Fas heterotrimers containing at least one decoy monomer were unable to signal.37 Moreover, at least for the Fas receptor, such heterotrimers may trigger the NF-κB pathway38 and thus promote not only hepatocyte survival but also liver regeneration.

The second concern is to design the most potent ΔDD receptor. Notably, a TNFR1 mutant whose entire cytoplasmic region was deleted reached the cell surface more efficiently and possessed greater dominant-negative activity than a mutant containing a DD point mutation.39 Hence, the extent of the deletion (from only the DD to the entire cytoplasmic region) could influence ΔDD-receptor potency.

Finally, the optimal way to deliver ΔDD-decoy receptors remains to be determined. Using intravenous Ad delivery, we demonstrated that membrane-anchored ΔDD-decoy receptors protected mice against hepatitis. However, although that technique achieved high levels of decoy expression, it is not amenable to treatment of human hepatitis because Ad possesses an intrinsic immunogenicity that induces both innate and specific immune responses.40 Recruitment of inflammatory cells into the liver leads to elimination of transduced hepatocytes,41 which should impair long-term ΔDD-receptor expression. Moreover, such immune responses could be deleterious and enhance ongoing hepatocyte damage. In an attempt to develop a safer delivery route, we used HP to deliver recombinant plasmids in a large volume as a bolus injection.42 Our results demonstrated that naked plasmids encoding membrane-anchored ΔDD-decoy receptors represent an efficient alternative to Ad delivery to prevent hepatitis. Plasmid recipients were protected against hepatitis, showing the efficacy of low death-decoy receptor expression. HP delivery route can achieve a transient high level of gene expression43 but can be repeated several times with only minor liver toxicity.42 However, because of the large volume injected, HP induces transient cardiac irregularities, which present a risk of heart failure.44 Nevertheless, regional HP gene delivery to the liver using the portal vein instead of the tail vein should constitute a safer route that could be amenable to clinical use.45

HP-injected siRNA represents another alternative because siRNA were able to lower the level of Fas-receptor20, 46 or caspase-819, 46 expression in the liver and thereby to protect against hepatitis19, 20 or septic shock.46 Recently, covalent binding of cholesterol to siRNA enabled transduction into different tissues after low-pressure injection.47 However, siRNA selectivity and potency remain uncertain.48 For example, siRNA can activate the interferon pathway, leading to nonspecific shut-down of protein synthesis and RNA degradation21; however, they also have interferon-independent “off-target activity,” that is, they drastically lower expression even when they present few mismatches with their mRNA target sequence.49 Furthermore, specific targeted delivery must be achieved before the therapeutic potential of siRNA can be envisaged.

Using different models of hepatitis, we showed that membrane-anchored ΔDD-decoy receptors are able to inhibit the progression of apoptosis signaling in vivo. Although Ad-gene transfer can achieve therapeutic levels of these decoy receptors, recombinant plasmids also can, and do so more safely. Such ΔDD-receptors have a well-defined ligand specificity, and they exert their blocking effects only in cells that express them, thereby avoiding systemic reaction. Moreover, their expression might be finely controlled by using tissue-specific promoters. Finally, different applications could be envisaged for ΔDD decoys. First, we propose that ΔDD decoys might be applicable in a therapeutic setting to control apoptosis in acute and chronic liver diseases and also to prevent the development of fibrosis. Second, they can be used in transgenic studies to induce cell-type–specific inhibition of death receptor function to achieve a better understanding of role of death receptors during liver physiological or pathological processes.

Acknowledgements

The authors thank all the staff of the animal facilities and particularly Monique Stanciu and Désiré Challuau for their excellent technical assistance. We also thank Philipo Rosselli (CNRS FRE 2939, Institut Gustave-Roussy, France) for providing 293T/NF-κB-luc cells, and Pierre Bobé (CNRS, UPR 9045, Villejuif, France) for his critical reading of the manuscript.

Ancillary