Fas and TNFR1, but not cytolytic granule-dependent mechanisms, mediate clearance of murine liver adenoviral infection


  • Conflict of interest: Nothing to report.


After intravenous injection of replication-deficient adenovirus, hepatocytes are transduced and express high levels of adenovirus-encoded genes. However, adenovirally encoded gene expression is ablated rapidly by CD8+ T-cell–dependent mechanisms. Thus, this model is suitable for examining intrahepatic cytotoxic T lymphocyte (CTL) effector mechanisms. In the present studies, recombinant adenoviruses encoding secreted (human apolipoprotein A-I) or intracellular (β-galactosidase) gene products were infused into mice with genetic deficiencies affecting the granule exocytosis-, Fas-, or tumor necrosis factor receptor 1 (TNFR1)-mediated pathways of CTL and natural killer cell effector function; the rates of clearance of adenovirus-encoded gene products were assessed. Clearance of secreted or intracellular adenoviral gene products was not delayed in perforin-deficient mice or dipeptidyl peptidase I-deficient mice, which fail to process and activate granzyme A or granzyme B. TNFR1-deficient mice also exhibited no delay in clearance of adenoviral gene products. However, adenoviral clearance from Fas-deficient mice was delayed, and such delays were much greater in mice deficient in both TNFR1 and Fas. In contrast, chimeric mice lacking both hepatic Fas and lymphocyte perforin function exhibited no greater delay in adenoviral clearance than chimeras deficient only in hepatic Fas expression. In conclusion, Fas-dependent mechanisms are required for efficient clearance of virally infected hepatocytes and, in Fas-deficient animals, TNFR1-dependent mechanisms provide an alternative mechanism for hepatic adenovirus clearance. In contrast, perforin- and granule protease–dependent cytotoxicity mechanisms play no apparent role in clearance of adenovirus from the liver. (HEPATOLOGY 2005;41:97–105.)

Hepatocytes are a primary or secondary site of viral replication during the course of many naturally acquired infections. In addition, viral vectors such as recombinant, replication deficient-adenoviruses efficiently infect and transduce hepatocytes, and thus have been explored as vehicles for liver-directed gene therapy.1, 2 Immune responses directed at virally infected hepatocytes are significant causes of liver disease3, 4 and have been found to limit efficacy of liver-directed gene therapy.5 The major viruses responsible for chronic viral hepatitis, hepatitis B virus and hepatitis C virus, are largely noncytopathic viruses that induce liver injury indirectly by eliciting host cytopathic immune responses.3, 4, 6 A variety of experimental observations in hepatitis B virus– and hepatitis C virus–infected chimpanzees suggest that natural killer and CD8+ T cells play major roles both in viral clearance and in liver injury.7, 8 Similarly, in murine models of hepatic infection/transfection by recombinant, replication-defective adenoviruses, CD8+ T cells play a major role in rapid clearance of adenoviral genomes from the liver by destroying adenovirally infected hepatocytes.9

Cytotoxic lymphocytes kill target cells by two major pathways: one dependent on exocytosis of granule effector molecules including perforin and granzymes A and B, and the other dependent on engagement of target cell TNFR1 family death receptors (e.g., Fas, tumor necrosis factor receptor 1 [TNFR1], DR4, DR5) by effector cell–expressed Fas ligand (FasL), membrane-bound or secreted tumor necrosis factor (TNF), or TNF-related apoptosis-inducing ligand.10–12 Mice with targeted mutations in single or multiple cytotoxic T-lymphocyte (CTL) effector molecules have been used to analyze the role of cytotoxic effector functions in antiviral responses. Perforin-deficient mice are defective in clearing multiple noncytopathic viruses such as lymphocytic choriomeningitis virus,13 ectromelia,14 murine cytomegalovirus,15 and Theiler's virus.16 Isolated deficiency in granzyme A expression also is associated with defects in clearance of the poxvirus ectromelia17 and with increased spread of herpes simplex virus in the peripheral nervous system.18 Moreover, mice deficient in both granzyme A and granzyme B exhibit a more profound defect in ectromelia clearance than granzyme A-only–deficient mice.19

Despite the prominent role that the perforin- and granzyme-mediated granule exocytosis pathway plays in the clearance of a variety of extrahepatic viral infections, mice with isolated perforin deficiency do not exhibit deficiency in clearance of hepatic adenoviral infection.20 Moreover, despite readily apparent delays in clearance of murine cytomegalovirus from the spleen21 and salivary gland22 of perforin-deficient mice, there is no delay in clearance of murine cytomegalovirus from the liver21 of these mice. In contrast, marked delay in hepatic adenoviral infection is noted in mice deficient in Fas, FasL, or TNF expression.13, 20, 23–25 However, mice with isolated defects in expression of TNFR1, the major mediator of TNF-induced apoptosis, clear hepatic adenoviral infection at rates similar to those in wild-type mice, whereas mice with defects in either TNF or tumor necrosis factor receptor 2 expression exhibit delays in adenoviral clearance that seem to be related to diminished activation of intrahepatic FasL-expressing CTL.24 Thus, results of prior studies suggest that among the various CTL effector pathways, FasL/Fas-mediated cytotoxic effector mechanisms play the most prominent role in clearance of hepatic viral infections,20, 24 whereas the potential role of other cytotoxic effector mechanisms remains unclear.

The lack of data implicating a prominent role for other cytotoxicity mechanisms in the clearance of hepatic viral infections may relate to the fact that hepatocytes are uniquely sensitive to Fas-induced apoptosis,26 and thus in mice with isolated defects in other CTL effector pathways, the FasL/Fas pathway readily can compensate for such defects. Indeed, in other murine viral infection models, it has been necessary to use experimental models with simultaneous defects in both perforin- and FasL-dependent CTL effector pathways to elucidate fully the role of individual CTL effector mechanisms in the clearance of viral infections.27, 28 The presence of multiple, redundant pathways for immune-mediated clearance of viral infections seems to be an important feature of host antiviral defenses, because multiple viruses have evolved mechanisms to evade individual immune effector pathways selectively.29–31 Thus, in the liver as in other organs, the presence of alternative or compensatory CTL effector pathways may have great importance in determining host ability to limit persistent viral infections.

The present studies were designed to examine the potential for clearance of virally infected hepatocytes via Fas-independent CTL and natural killer cell effector mechanisms. Replication-deficient, recombinant adenoviruses encoding either secreted or intracellular protein products were infused into mice with single or multiple deficiencies in CTL and natural killer cell effector pathways. Mice with lymphocyte defects in perforin alone or in combination with hepatocyte defects in Fas expression, and dipeptidyl peptidase I knockout mice deficient in both granzyme A and granzyme B activation,32 were used to explore the separate role of granule proteases in clearing adenovirally infected hepatocytes. In addition, mice deficient in both TNFR1 and Fas expression were used to assess whether TNFR1-dependent effector mechanisms serve as an alternative pathway for clearing virally infected hepatocytes when the FasL/Fas CTL effector pathway is impaired.


NFR1, tumor necrosis factor receptor 1; FasL, Fas ligand; TNF, tumor necrosis factor; CTL, cytotoxic T lymphocyte; B6, C57BL/6J; B6.pfp−/−, C57BL/6-Pfptm1Sdz; B6.TNFR1−/−, B6.129-Tnfrsf1atm1Mak; B6.lpr, B6.MRL-Tnfr6lpr; AdCMV-lacZ, β-galactosidase-encoding adenovirus; AdCMV-HuApoA-I, human apolipoprotein A-I encoding adenovirus; mRNA, messenger RNA; PCR, polymerase chain reaction.

Materials and Methods


C57BL/6J (B6), C57BL/6-Pfptm1Sdz (B6.pfp−/−), B6.129-Tnfrsf1atm1Mak (B6.TNFR1−/−), and B6.MRL-Tnfr6lpr (B6.lpr) mice were obtained from The Jackson Laboratory (Bar Harbor, ME). B6.Tnfr1−/−lpr mice were developed in our laboratory by breeding B6.lpr with B6.TNFR1 mice to produce B6.Tnfr1−/−lpr mice homozygous for both a targeted mutation of the TNFR1 gene and the lpr mutation in the Fas gene. Genotyping of offspring was performed following the protocol published on The Jackson Laboratory web site. Dipeptidyl peptidase I-deficient mice (B6.dppI−/−) were provided by Dr. Christine Pham32 (Washington University, St. Louis, MO). Mice in individual experiments were age- and sex-matched and were used before 12 weeks of age. All experimental protocols were approved by the UT Southwest Institutional Animal Care and Use Committee in accordance with criteria outlined in “Guide for Care and Use of Laboratory Animals” (NIH publication 86-23).

Adenovirus Vectors.

The E1-deleted, replication-deficient, β-galactosidase–encoding (AdCMV-lacZ33) or human apolipoprotein A-I encoding (AdCMV-HuApoA-I34) recombinant adenoviruses were propagated in 293 cell cultures and purified on cesium chloride gradients as previously described.33, 34 Mice were injected intravenously with 109 to 1010 plaque-forming units of AdCMV-lacZ or AdCMV-HuApoA-I. In each experiment, a constant number of plaque-forming units per gram of body weight was administered to each mouse.

β-Galactosidase Assay.

β-Galactosidase activity in liver tissue homogenates was quantified by measuring the rate of cleavage of 4-methylumbelliferyl-β-D-galactoside to yield the fluorescent product 4-methylumbelliferone as previously described.20, 24 For each sample, the reported value is an average of dilutions falling in the assay linear range.

Protein Assay.

Protein concentrations in tissue homogenates were assayed by the bicinchoninic acid method with reagents purchased from Pierce (Rockford, IL) using bovine serum albumin as a standard.

Hepatic HuApoA-I Messenger RNA (mRNA) Quantification.

Tissue HuApoA-I mRNA levels were determined using a ribonuclease protection assay as previously described.35 Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as an internal control. Species-specific 32P-labeled riboprobes were synthesized using MAXIscript in vitro transcription kits (Ambion Inc., Austin, TX) in the presence of 3 μM (HuApoA-I) or 100 μM (glyceraldehyde-3-phosphate dehydrogenase) labeled nucleotide. Samples of liver were homogenized in RNA STAT-60 (TEL-TEST Inc., Friendswood, TX). Total RNA (40 μg) was hybridized with 32P-labeled riboprobes simultaneously at 68°C using the HybSpeed RPA protocol (Ambion Inc). After RNase digestion, the mRNA-protected 32P-labeled probes were separated on 8 M urea, 5% polyacrylamide gels together with 32P-labeled MspI-digested pBR322 size standards. The radioactivity in each band, as well as background radioactivity, was quantified using a phosphorimaging system (Molecular Dynamics Inc., Sunnyvale, CA).

Serum Levels of HuApoA-I.

Two hundred–microliter aliquots of serum were collected from experimental animals and were used for HuApoA-I assays as previously described,35 with minor modifications. Aliquots (5 μL) of each sample were mixed with 50 μL of the antibody reagent (goat antihuman ApoA-I; Sigma, St. Louis, MO) in 96-well, flat-bottom enzyme-linked immunosorbent assay plates (Corning, Acton, MA) and incubated for 15 minutes at room temperature. The absorbance was read at 340 nm on a Molecular Devices (Sunnyvale, CA) Thermomax plate reader and was analyzed using Softmax software. All HuApoA-I determinations were performed in duplicate. Standard curves were constructed from HuApoA-I standards provided by the kit supplier.

Adenoviral DNA Assay.

Semiquantitative polymerase chain reaction (PCR) of adenoviral DNA and a reference genomic insulysin sequence in liver was performed using a DNA digestion procedure adapted from the Tail DNA for PCR (no organic solvents) protocol from the Jackson Laboratory (http://www.jax.org/imr/tail_nonorg.html). Pulverized, frozen liver tissues were incubated at 50°C with gentle agitation in 0.8 mL of PCR buffer with nonionic detergents (50 mM KCl, 10 mM Tris-HCl, 2.5 mM MgCl2, 0.1 mg/mL gelatin 0.45% Nonidet P40, and 0.45% Tween 20) containing 0.05 mg/mL proteinase K (Qiagen, Valencia, CA) until digestion was visibly complete. The samples then were heat denatured at 100°C, and serial dilutions in diethylpyrocarbonate-treated H2O were subjected to PCR amplification over a range of cycles using insulysin primers to identify concentrations of digest yielding approximately equal amounts of the genomic PCR product. Adenoviral DNA content in appropriately diluted samples then was determined by PCR amplification over a range of cycles with β-galactosidase–specific primers. PCR amplification was carried out in standard reaction mixtures36 of 25 μL containing 1 μL of appropriately diluted liver digest using a cycling program of 94°C, 55°C, and 72°C for 45 seconds each after an initial 5-minute denaturating step at 94°C.

Chimeric Mouse Preparation.

B6.pfp−/−, B6.lpr, and control B6 mice were killed and bone marrow cells were isolated and treated with 250 μM leucyl-leucine methyl ester to prevent graft-versus-host disease.37 B6- and B6.lpr-recipient mice received lethal total body irradiation (900 cGy) and then were infused with 5 × 106 leucyl-leucine methyl ester pretreated bone marrow cells from B6, B6.lpr, or B6.pfp−/− donor mice. The chimeric animals were allowed to recover for 3 months to allow immune reconstitution by donor bone marrow–derived T lymphocytes37 before they were infused with AdCMV-HuApoA-I.

Statistical Analysis.

The data are presented as mean ± SEM. To test for differences among groups, a one-way ANOVA was performed when the data were parametric, and significant results were analyzed further using the Tukey multiple comparison procedure. For nonparametric data, Kruskal-Wallis analysis followed by the Mann-Whitney test was used.


AdCMV-HuApoA-I Clearance Is Delayed in B6.lpr- But Not Perforin-Deficient Mice.

In separate experiments, control B6- and perforin-deficient B6.pfp−/− (Fig. 1A) or fas-defective B6.lpr mice (Fig. 1B) were infected with AdCMV-HuApoA-I virus, and serum levels of the adenoviral transgene product, HuApoA-I, were assessed at different time points. HuApoA-I levels decreased by more than 95% in the serum of both B6.pfp−/− and B6 control mice by day 19 after infection, and by day 29 after infection, levels could no longer be detected in either B6.pfp−/− or B6 mice. (Fig. 1A). In contrast, in Fas-deficient B6.lpr mice, serum HuApoA-I levels at day 14 after infection remained at approximately two thirds of the peak levels seen at day 3 (Fig. 1B). Moreover, by day 21, no HuApoA-I could be detected in control animals, whereas the Fas-defective B6.lpr mice still had detectable levels at day 28 after infection (Fig. 1B).

Figure 1.

Clearance of serum HuApoA-I from AdCMV-HuApoA-I–infected perforin-deficient, Fas-deficient, and control B6 mice. B6.pfp−/−, B6.lpr, and B6 mice were infused intravenously with AdCMV-HuApoA-I. Serum levels of HuApoA-I were measured at the indicated number of days after infection in five animals per experimental group, and results were expressed as percent of peak serum HuApoA-I expression at day 3 after injection. Mean levels of serum HuApoA-I expression at day 3 were 162 mg/dL in experiment 1 and 49 mg/dL in experiment 2.

Similar differences in rates of clearance of HuApoA-I mRNA were noted. At day 28 after infection, hepatic HuApoA-I mRNA levels were significantly higher in B6.lpr mice than in B6 mice (Fig. 2A). In contrast, in B6.pfp−/− mice, hepatic HuApoA-I mRNA levels declined at rates similar to that observed in control B6 mice during the first 19 days after injection (Fig. 2B).

Figure 2.

Clearance of AdCMV-HuApoA-I transgene messenger RNA (mRNA) from perforin-deficient, Fas-deficient, and control B6 mice. B6.pfp−/−, B6.lpr, and B6 mice were infused intravenously with AdCMV-HuApoA-I. Hepatic HuApoA-I and GAPDH mRNA levels were measured at the indicated number of days after infection in three to five animals per experimental group. *P < .05 vs. B6 control liver.

The Granule-Exocytosis Pathway Is Not Required for Liver Adenoviral Clearance.

To examine further the role of individual components of the granule exocytosis pathway in clearing adenovirus from the liver and to assess whether this cytotoxic effector pathway may play a greater role in clearing intracellular viral gene products, clearance of a second recombinant adenovirus, AdCMV-lacZ, was assessed in mice with selected deficiencies in this CTL effector pathway. As detailed in Fig. 3, expression of the intracellular adenoviral transgene product β-galactosidase declined at similar rates in granzyme A– and granzyme B–deficient B6.dppI−/− mice, perforin-deficient B6.pfp−/− mice, and control B6 mice.

Figure 3.

Clearance of AdCMV-lacZ in perforin-deficient, dipeptidyl peptidase I–deficient, and control mice. B6.pfp−/−, B6.dppI−/−, and B6 mice were infused intravenously with AdCMV-lacZ, and hepatic β-galactosidase activity was measured at days 3, 17, and 34 after infection. Values are reported as percent of peak β-galactosidase activity at day 3, and the results represent the mean ± SEM of results of six animals per experimental group.

AdCMV-lacZ Clearance Is Delayed to a Greater Degree in Mice Deficient in TNFR1 Plus Fas Than in Mice Deficient in Fas Alone.

To assess the combined role of FasL/Fas and TNF/TNFR1 pathways in adenoviral clearance, B6.Tnfr1−/−lpr mice deficient in both TNFR1 and functional Fas expression, B6.Tnfr1−/− mice deficient in TNFR1 expression, Fas-deficient B6.lpr and control B6 mice were infected with AdCMV-lacZ and killed at different time points to determine hepatic β-galactosidase activity. Figure 4 displays the residual levels at day 36 after infection expressed as percent of peak β-galactosidase activity at day 4. Although AdCMV-lacZ–infected wild-type B6 and B6.Tnfr1−/− mice exhibited hepatic β-galactosidase activity comparable with that of uninfected B6 control mice, AdCMV-lacZ–infected B6.lpr mice exhibited hepatic β-galactosidase levels 8-fold higher than baseline (P < .01). Residual levels of the AdCMV-lacZ transgene product in livers of B6.Tnfr1−/−lpr mice were even more dramatically elevated (P < .01; see Fig. 4), indicating that simultaneous impairment of these two CTL effector pathways had an even greater impact on the course of immune clearance of hepatic adenovirus infection.

Figure 4.

Clearance AdCMV-lacZ from tumor necrosis factor receptor 1 (TNFR1)-deficient, Fas-deficient, TNFR1- and Fas-deficient, and control B6 mice. The indicated strains of mice were infused intravenously with AdCMV-lacZ, and hepatic β-galactosidase activity was measured at days 4 and 36 after infection. Levels are expressed as percent of peak activity at day 4 and represent the mean ± SEM of seven to eight animals per experimental group. *P < .01 vs. B6.

To determine whether these differences in clearance of adenoviral gene products in mice with varying combinations of intact CTL effector mechanisms reflect differences in elimination of virally infected hepatocytes or result instead from differences in viral gene silencing, liver tissue from individual animals included in the experiments detailed in Fig. 4 were assessed by semiquantitative PCR assays for persistence of adenoviral DNA. Persistence of adenoviral DNA was readily apparent in livers of B6.Tnfr1−/−lpr mice killed 36 days after adenoviral infection but could not be detected in livers of wild-type B6 or B6.Tnfr1−/− mice killed at the same time after infection (Fig. 5). Thus, despite intact perforin and granzyme effector pathways, B6.Tnfr1−/−lpr mice are unable to clear adenoviral DNA from the liver during the first 36 days after infection, whereas hepatic adenoviral infection is cleared rapidly by B6 or B6.Tnfr1−/− mice that also express functional Fas receptors.

Figure 5.

Clearance of AdCMV-lacZ viral DNA from B6, tumor necrosis factor receptor 1 (TNFR1)-deficient (TNFR1−/−), and Fas-defective TNFR1-deficient (TNFR1−/−lpr+/+) mice. The indicated strains of mice were infused intravenously with AdCMV-lacZ, and the hepatic content of AdCMV-lacZ DNA relative to genomic DNA was assessed by semiquantitative polymerase chain reaction (PCR) on days 3 and 36 after infection. Samples of liver extract containing approximately equal amounts of genomic DNA were amplified for the indicated number of cycles with insulysin gene-specific primers (F5′-ACCAAACCAAACCAAACCAA and R5′-TCCCCCACACTGTAATGGAT) and with primers specific for the β-galactosidase sequence of the AdCMV-lacZ construct (F5′-GACGTCTCGTTGCTGCATAA and R5′-CAGACGTAGTGTGACGCGAT). The PCR products were separated by electrophoresis through 2% agarose gels and were visualized by staining with ethidium bromide.

AdCMV-HuApoA-I Clearance Is Not Delayed to a Greater Degree in Mice With Defects in Both the Granule-Exocytosis and Fas-Dependent Cytotoxicity Pathways Than in Fas-Deficient Mice.

To investigate a possible compensatory role for the perforin- and granzyme-dependent granule exocytosis pathway in the absence of the Fas-dependent pathways of CTL effector function, additional experiments were performed in mice with varying combinations of CTL Fas or perforin deficiency and hepatic Fas deficiency. Because of previously documented multisystemic disease and poor fertility of perforin and Fas double knockout mice38 and to allow separate examination of the role of CTL versus hepatic Fas deficiency in delayed adenoviral clearance, we created chimeric mice with either normal or deficient hepatic Fas expression and either normal, Fas-deficient, or perforin-deficient CTL. These chimeric mice were infected with AdCMV-HuApoA-I, and samples of sera and livers were harvested subsequently at different time points to assess serum HuApoA-I protein levels and hepatic HuApoA-I mRNA levels.

Regardless of the presence or absence of lymphocyte deficiency in Fas or perforin, all mice with Fas-deficient B6.lpr livers exhibited similar HuApoA-I serum levels at various time points after infection (Fig. 6A). The same was true for chimeric mice with wild-type livers. In addition, regardless of the source of lymphocyte reconstitution, both B6.lpr and B6 recipient mouse groups exhibited similar HuApoA-I serum levels at day 3. However, serum HuApoA-I levels in B6 → B6.lpr, B6.lpr → B6.lpr, and B6.pfp−/− → B6.lpr mice were significantly higher (P < .01) at day 15 than those of B6 → B6, B6.lpr → B6, or B6.pfp−/− → B6 mice (Fig. 6A). In addition, serum HuApoA-I had been cleared from mice with B6 wild-type livers but not from those with B6.lpr livers at day 40 after infection, indicating that only defects in hepatic Fas expression are associated with delayed clearance of adenoviral transgene products.

Figure 6.

Clearance of AdCMV-HuApoA-I transgene products from mice with varying combinations of hepatic Fas and lymphocyte perforin and/or Fas deficiency. Chimeric mice with varying combinations of hepatic Fas deficiency and cytotoxic T lymphocyte perforin or Fas deficiency were infused with AdCMV-HuApoA-I and serum HuApoA-I levels (Fig. 5A), and Hepatic HuApoA-I messenger RNA (mRNA) expression (Fig. 5B) was measured at the indicated time points in three to five animals per experimental group (Fig. 5A) or three animals per experimental group (Fig. 5B).

The same findings were true for hepatic HuApoA-I mRNA levels. Chimeric mice with Fas deficiency in host hepatocytes exhibited more prolonged expression of HuApoAI mRNA than did mice with wild-type B6 hepatocytes, regardless of the presence or absence of perforin or Fas expression by bone marrow–derived cells (P < .01 at day 40 and day 75). Thus, rates of clearance of AdCMV-HuApoA-I gene products from the serum or liver were similar in B6.pfp−/− → B6.lpr and B6 → B6.lpr chimeras, indicating no apparent role for perforin-dependent CTL effector mechanisms, even in Fas-deficient animals. Of interest, B6 recipient mice reconstituted with B6.lpr lymphocytes known to express greatly increased amounts of FasL39 displayed lower levels of hepatic mRNA at days 29 and 40 (P < .05) compared with the mice with B6 wild-type livers (Fig. 6B) reconstituted with B6 wild-type or perforin-deficient B6.pfp−/− lymphoid cells.


The results of the present studies indicate that Fas-dependent mechanisms are required for efficient clearance of virally infected hepatocytes and suggest that in Fas-deficient animals, TNFR1-dependent mechanisms provide an alternative mechanism for clearance of hepatic adenoviral infection. In contrast, the present findings reveal no requisite role for either perforin or cytolytic granule proteases in adenovirus clearance from the liver. Moreover, even in the absence of functional hepatocyte Fas receptors, deficiency of the perforin-dependent effector pathway in CTL did not impair clearance of adenovirus-infected hepatocytes further. Thus, the perforin-dependent granule exocytosis pathway that is the major CTL cytotoxicity mechanism used in clearance of virally infected cells from other sites of infection12 seems not to play a prominent role in antiviral immune responses in the liver. Rather, clearance of virally infected hepatocytes seems to be mediated by Fas- or TNFR1-dependent death receptor pathways and other cytolytic granule-independent mechanisms.

In hepatocytes and other cells with a type II pattern of Fas signaling,40, 41 induction of Fas-dependent apoptosis is dependent on an amplification loop that relies on caspase-8–mediated cleavage of Bid and subsequent release of mitochondrial proapoptotic factors to facilitate formation of the caspase-9 activating apoptosome.42 Because similar pathways of caspase-8 recruitment and activation are initiated by TNF– TNFR1 interactions,43 it is logical that these two death receptor mechanisms serve as alternative pathways for clearance of adenovirally encoded gene products. In support of this hypothesis, clearance of hepatic adenovirus in B6.Tnfr1−/−lpr mice deficient in both pathways is much more impaired than in B6.lpr mice or B6.TNFR1−/− mice lacking only one of these cytotoxicity pathways. However, the lack of significant delay in clearance of hepatic adenoviral infection in B6.Tnfr1−/− mice suggests that in mice with intact FasL/Fas signaling pathways, this CTL effector mechanism is the dominant pathway for clearance of virally infected hepatocytes and is able to compensate fully for deficiencies in the TNF/TNFR1 pathway. The fact that TNF/TNFR1 interactions induce both proapoptotic and antiapoptotic signaling cascades43 also may account for the absence of a significant net positive or negative effect of TNFR1 deficiency on the early stages of adenoviral clearance from the liver and may explain the lack of significant effect of isolated TNFR1 deficiency on clearance of adenovirus-encoded gene products from the liver in the present studies. Finally, the observation that clearance of hepatic adenovirus-encoded mRNA proceeds in an accelerated manner in B6.lpr → B6 chimeras with Fas-defective CTL but intact hepatocyte Fas receptors likely reflects the effects of the compensatory upregulation of FasL expression that has been observed in Fas-deficient T cells.39 This finding therefore further substantiates the major role of FasL/Fas interactions in efficient clearance of virally infected hepatocytes. In addition, the results of this experiment argue that defects in hepatic adenoviral clearance by B6.lpr mice indeed are related directly to defects in CTL-induced Fas-mediated apoptosis in virally infected hepatocytes and not to indirect effects of other immune abnormalities in these mice.

Recent observations indicate that treatment of cells with perforin and granzyme B leads not only to caspase-8 activation, but also leads to direct granzyme B–mediated cleavage of Bid and induction of apoptosis via the mitochondrial intrinsic pathway.44 Thus, it would be predicted that cells with a type II apoptotic pattern,40 such as hepatocytes, would be vulnerable to perforin and granzyme B–induced apoptosis. However, in the present studies, neither mice deficient in perforin nor mice deficient in the granzyme A and B processing enzyme dipeptidyl peptidase I exhibited a significant delay in clearance of adenovirus from the liver. Moreover, even in chimeric mice with hepatocyte Fas deficiency, additional absence of the perforin-dependent cytotoxicity pathway in bone marrow–derived lymphoid cells does not lead to any greater prolongation of adenovirus-encoded gene expression than is observed in Fas-deficient mice reconstituted with wild-type immune effector cells. These results are in agreement with the observation that perforin-deficient mice exhibit no delay in murine CMV clearance from the liver.21 In contrast, Nakamoto et al.45 found that both FasL- and perforin-dependent cytotoxicity mechanisms are required for generation of high levels of cytotoxicity mediated by CTL clones against some but not all hepatitis B virus transgenic hepatocytes, suggesting that under certain circumstances, highly activated CTL use perforin-dependent mechanisms to kill hepatocytes.

Previous studies from our laboratory indicate that adenovirus-infected hepatocytes, although sensitive to FasL/Fas- and TNF/TNFR1-mediated CTL effector mechanisms, are relatively resistant to perforin- and granzyme-mediated cytotoxicity.20 In addition, recent studies have found that after stimulation with cytokines released as part of the innate immune response during viral infections, the granzyme B inhibitory serpins human proteinase inhibitor 9 and mouse serine proteinase inhibitor 6 are expressed in hepatocytes at levels noted to provide resistance to perforin- and granzyme B–dependent cytotoxicity in other cell types.46 Thus, inhibition of the granzyme B–dependent CTL effector pathway by inducible factors acting before either caspase-8 or Bid cleavage in target hepatocytes likely accounts for the participation of FasL/Fas- and TNF/TNFR1- but not perforin- and granzyme B–dependent cytotoxicity mechanisms in clearance of virally infected hepatocytes. The lack of induction of such cytoprotective mechanisms resulting from the absence of preceding innate cytokine responses also may account for the greater apparent role for perforin-dependent cytotoxicity mechanisms in models of hepatocyte injury induced by transfer of clonal populations of in vitro–activated CTL.45

Of note, among the multiple effector mechanisms used by CTL or natural killer cells, the perforin and granzyme B pathway proceeds with the most rapid kinetics and is almost solely responsible for target cell killing that evolves within the first several hours of effector–target cell interaction.47 In contrast, the FasL/Fas- and TNF/TNFR1-induced cytotoxicity mechanisms, found in the present studies to be the principal pathways for immune clearance of virally infected hepatocytes, proceed with much slower kinetics,12 and therefore likely allow for greater degrees of hepatic regeneration during immune clearance of viral infections from the liver. In conclusion, because CTL and natural killer cell responses are an important determinant of the severity of hepatocellular injury during viral hepatitis,3, 4, 6 it seems plausible that hepatocytes have evolved mechanisms that inhibit the major perforin and granzyme cytotoxic effector pathway to avoid acute liver failure induced by overly rapid clearance of virally infected hepatocytes during viral hepatitis. However, the unique reliance on immune clearance mechanisms mediated via the Fas/TNFR1 family of death receptors may leave the liver more vulnerable to infection by viruses that express products capable of impairing these immune effector mechanisms.