Increased hepatotoxicity of tumor necrosis factor–related apoptosis-inducing ligand in diseased human liver


  • Xandra Volkmann,

    1. Department of Gastroenterology, Hepatology and Endocrinology, Hannover Medical School, Hannover, Germany
    Search for more papers by this author
  • Ute Fischer,

    1. Institute of Molecular Medicine, University of Düsseldorf, Düsseldorf, Germany
    Search for more papers by this author
  • Matthias J. Bahr,

    1. Department of Gastroenterology, Hepatology and Endocrinology, Hannover Medical School, Hannover, Germany
    Search for more papers by this author
  • Michael Ott,

    1. Department of Gastroenterology, Hepatology and Endocrinology, Hannover Medical School, Hannover, Germany
    Search for more papers by this author
  • Frank Lehner,

    1. Department of Visceral and Transplantation Surgery, Hannover Medical School, Hannover, Germany
    Search for more papers by this author
  • Marion MacFarlane,

    1. Medical Research Council Toxicology Unit, University of Leicester, Leicester, United Kingdom
    Search for more papers by this author
  • Gerald M. Cohen,

    1. Medical Research Council Toxicology Unit, University of Leicester, Leicester, United Kingdom
    Search for more papers by this author
  • Michael P. Manns,

    1. Department of Gastroenterology, Hepatology and Endocrinology, Hannover Medical School, Hannover, Germany
    Search for more papers by this author
  • Klaus Schulze-Osthoff,

    Corresponding author
    1. Institute of Molecular Medicine, University of Düsseldorf, Düsseldorf, Germany
    • Klaus Schulze-Osthoff, University of Düsseldorf, Institute of Molecular Medicine, Universitätsstrasse 1, D-40225 Düsseldorf, Germany===

      Heike Bantel, Department of Gastroenterology, Hepatology and Endocrinology, Hannover Medical School, Carl-Neuberg-Strasse 1, D-30625 Hannover, Germany===

    Search for more papers by this author
    • fax: (49) 211-8115892

  • Heike Bantel

    Corresponding author
    1. Department of Gastroenterology, Hepatology and Endocrinology, Hannover Medical School, Hannover, Germany
    • Klaus Schulze-Osthoff, University of Düsseldorf, Institute of Molecular Medicine, Universitätsstrasse 1, D-40225 Düsseldorf, Germany===

      Heike Bantel, Department of Gastroenterology, Hepatology and Endocrinology, Hannover Medical School, Carl-Neuberg-Strasse 1, D-30625 Hannover, Germany===

    Search for more papers by this author
    • fax: (49) 511-5326998

  • Potential conflict of interest: Nothing to report.


Tumor necrosis factor–related apoptosis-inducing ligand (TRAIL) induces apoptosis in tumor cells but not in most normal cells and has therefore been proposed as a promising antitumor agent. Recent experiments suggested that isolated primary human hepatocytes but not monkey liver cells are susceptible to certain TRAIL agonists, raising concerns about the use of TRAIL in cancer treatment. Whether TRAIL indeed exerts hepatotoxicity in vivo and how this is influenced by chemotherapeutic drugs or liver disease are completely unknown. Employing different forms of recombinant TRAIL, we found that the cytokine can induce proapoptotic caspase activity in isolated human hepatocytes. However in marked contrast, these different TRAIL preparations induced little or no cytotoxicity when incubated with tissue explants of fresh healthy liver, an experimental model that may more faithfully mimic the in vivo situation. In healthy liver, TRAIL induced apoptosis only when combined with histone deacetylase inhibitors. Strikingly, however, TRAIL alone triggered massive apoptosis accompanied by caspase activation in tissue explants from patients with liver steatosis or hepatitis C viral infection. This enhanced sensitivity of diseased liver was associated with an increased expression of TRAIL receptors and up-regulation of proapoptotic Bcl-2 proteins. Conclusion: These results suggest that clinical trials should be performed with great caution when TRAIL is combined with chemotherapy or administered to patients with inflammatory liver diseases. (HEPATOLOGY 2007.)

Triggering death receptors is an attractive strategy to overcome the apoptosis resistance of tumor cells to conventional chemo- and radiotherapy. Tumor necrosis factor–related apoptosis-inducing ligand (TRAIL), also called Apo2L, has received particular attention because early studies suggested that, in contrast with tumor necrosis factor (TNF) and FasL (CD95L), TRAIL might selectively induce apoptosis in transformed but not in normal cells.1 Human TRAIL is expressed as a type II transmembrane protein that binds to 1 of 4 receptors: TRAIL-R1/DR4, TRAIL-R2/DR5, TRAIL-R3/DcR1, and TRAIL-R4/DcR2. Both TRAIL-R1 and TRAIL-R2 are proapoptotic receptors that contain a cytoplasmic death domain and mediate apoptosis on recruitment of caspase-8 and caspase-10 into a death-inducing signaling complex (DISC).2–4 The other 2 receptors, TRAIL-R3 and TRAIL-R4, lack a functional death domain and are not capable of activating the caspase cascade. The precise physiological function of TRAIL remains obscure. The main function of TRAIL appears to be the negative regulation of the innate immune system, although it may also serve as an effector molecule for the elimination of transformed and virally infected cells.5–8

TRAIL is a promising target for development as a cancer-specific agent because it induces apoptotic cell death in a wide variety of transformed cell lines and tumor cells. Even in tumor cell lines that show resistance to TRAIL, combined treatment with chemotherapeutic drugs or ionizing radiation has revealed induction of tumor cell death.9 In particular, inhibitors of histone deacetylases (HDACs) have been shown to sensitize various tumor cells to TRAIL-induced apoptosis by up-regulating TRAIL receptors or proapoptotic Bcl-2 proteins.10–15 Importantly, TRAIL induces apoptosis in several animal tumor xenograft models. Preclinical studies in mice and nonhuman primates demonstrated that TRAIL induces apoptosis in human tumors, whereas no cytotoxicity to normal tissues was found.16–17 The susceptibility of tumor cells to TRAIL and the apparent lack of activity of TRAIL in normal cells have led to the proposed use of TRAIL in cancer therapy. Phase 1 studies with recombinant TRAIL as well as phase 1 and phase 2 studies with agonistic monoclonal antibodies are in progress in cancer patients.

More recent findings, however, demonstrated that normal human hepatocytes, brain tissue, and certain epithelial cells might be susceptible to recombinant TRAIL proteins.18–20 These findings highlight the possibility that systemic TRAIL administration may be potentially hepatotoxic in humans. Reevaluation of these results suggested that the cytotoxicity of TRAIL might be dependent on the version of recombinant TRAIL used.21 Leucine zipper (LZ)–tagged versions of TRAIL did not show any adverse effects in normal tissue in mice. In addition, an untagged TRAIL, currently being developed as a clinical candidate, was not toxic to isolated human and cynomolgus hepatocytes and was well tolerated in vivo in cynomolgus monkeys and chimpanzees. In contrast, normal human hepatocytes but not hepatocytes from mice, rats, or monkeys underwent apoptosis when incubated with a recombinant polyhistidine-tagged or a crosslinked FLAG-tagged soluble form of human TRAIL in vitro.18, 21 Nevertheless, adverse effects of TRAIL in primary human hepatocytes (PHHs) are still controversial and can only partially be explained by the use of different forms of recombinant TRAIL but may also depend on in vitro conditions, that is, the isolation and culturing of PHHs.22

Potential liver toxicity is currently the major obstacle to the use of TRAIL in cancer treatment. So far, concerns about potential toxicity are mainly based on the use of freshly isolated human hepatocytes. However, there are a number of important limitations to the use of cultured primary hepatocytes, which are markedly different from hepatocytes in situ.23 Thus, whether the TRAIL toxicity observed in hepatocytes in vitro can be extrapolated to the in vivo situation is a critical question. It is also necessary to consider that the liver is a complex organ composed of multiple cell types that interact with each other and may modulate the TRAIL susceptibility of hepatocytes. Furthermore, there have been no reports about whether the cytotoxic effects of TRAIL agonists differ between diseased and healthy liver tissue.

In this study, we investigated the potential liver toxicity of different recombinant forms of TRAIL in organotypic cultures of fresh liver explants. Our results demonstrate that TRAIL alone is relatively harmless to healthy liver explants, whereas it activates caspases in isolated PHHs. However, a combination with HDAC inhibitors renders TRAIL highly toxic in healthy liver explants. Strikingly, compared with healthy liver, the different TRAIL agonists were significantly more toxic to diseased liver, such as steatotic or HCV-infected liver. Our results suggest that clinical trials with TRAIL should be performed with great caution, especially when TRAIL is combined with other chemotherapeutic agents or when it is administered to patients with inflammatory liver diseases or metabolic syndrome.


HDAC: histone deacetylase; PHH: primary human hepatocyte; TRAIL: tumor necrosis factor–related apoptosis-inducing ligand.

Patients and Methods

Culture of Liver Explants and Primary Hepatocytes.

We investigated explants of healthy liver from 11 patients (7 men, 4 women; mean age 65 ± 3.2 years, range 47-83 years) and of steatotic liver from 5 patients (3 men, 2 women; mean age 66 ± 5 years, range 48-78 years) who underwent partial hepatectomy because of single metastasis of nonhepatic origin. From the resected livers only non-tumor-bearing portions were taken. In addition, liver biopsies from 9 patients with chronic HCV infection (5 men, 4 women; mean age 52 ± 3.7 years; range 31-67 years) were analyzed. All patients showed normal values for liver function, that is, normal bilirubin, prothrombin time, and cholinesterase activity. Patients with chronic HCV infection were scored according to Ishak et al.24(mean ISHAK A–D score 3.7 ± 0.4, mean ISHAK F score 2.7 ± 0.9). For the reverse transcription polymerase chain reaction (RT-PCR) analyses, an additional 12 patients with HCV infection (3 women, 9 men; mean age 39 ± 3.3 years, range 21-53 years; ISHAK A–D 4.8 ± 0.6, F 2.2 ± 0.6) were included. Explants from freshly isolated liver of healthy and steatotic individuals as well as liver biopsies of HCV-infected patients were precisely divided into 125 mm3 cubes under sterile conditions25,26and incubated with 1 mL of medium in 24-well plates under gentle shaking. PHHs were isolated from healthy liver as described27and provided by Cytonet (Hannover, Germany). For the culture of isolated hepatocytes and liver tissue, we used modified Eagle's medium (Biochrom, Berlin, Germany) supplemented with 1% human serum, 0.4 IE/mL insulin, 20 mM HEPES [4-(2-hydroxyethyl)-1-piperazine ethane sulfonic acid], 2 mM L-glutamine, 0.2 g/L MgCl2× 6 H2O, 1× vitamin solution, 20 mg/L L-ornithine-HCl, 50 mg/L ascorbic acid, 50 μg/mL gentamycin, and 8 μg/mL dexamethasone. Isolated hepatocytes were maintained in medium for 16 hours at 37°C in 5% CO2 prior to incubation with the apoptotic stimuli, whereas incubation of liver tissue was performed immediately. PHHs and liver tissue were incubated for 6 hours with the different TRAIL agonistic agents or 100 ng/mL Flag-tagged CD95L (provided by Dr. I. Schmitz, University of Düsseldorf, Germany) in the presence of 1 μg/mL anti-Flag M2 Mab (Sigma-Aldrich, Taufkirchen, Germany). His-tagged TRAIL (KillerTRAIL, amino acids 95-281, tested for low endotoxin) was obtained from Alexis (Grünberg, Germany), and LZ-tagged TRAIL (residues 95-281) from Dr. H. Walczak (Heidelberg, Germany). Untagged TRAIL (residues 114-281) was produced and purified to near homogeneity as described.17, 28 The HDAC inhibitor depsipeptide (10 nM, kindly provided by Dr. E. Sausville, Bethesda, MD) was added 2 hours prior to incubation with TRAIL.

Detection of Caspase Activity in Human Liver Tissue and Isolated Hepatocytes.

Following incubation with the respective agents, liver tissue was pulverized in liquid nitrogen. Pulverized liver tissue and isolated hepatocytes were lysed in a cocktail of 0.5% Nonidet P-40, 10 mM Tris-HCl (pH 7.4), 10 mM MgCl2, 150 mM NaCl, 10 mM dithiothreitol, and 1% protease inhibitor (Sigma-Aldrich). The protein concentration of cell extracts was determined by the Bradford assay (BioRad, Munich, Germany). Activity of caspase-3 and caspase-7 was measured by a luminescent substrate assay (Caspase-Glo; Promega, Mannheim, Germany) essentially as described.29 Cell extracts were diluted in buffer containing 50 mM Tris-HCl (pH 7.4), 10 mM KCl, and 5% glycerol to reach a final protein concentration of 0.1 mg/mL of the hepatocyte extract and 1 mg/mL of the liver tissue extract. Then 10 μL of the extracts was incubated with 10 μL of the caspase substrate DEVD-luciferin and luciferase reagent for 3 hours at room temperature. Finally, the luminescence of the samples was measured in a luminometer and calculated as the increase relative to that of the untreated control.


Frozen sections (5 μm) of healthy, HCV-infected, or steatotic liver explants, either untreated or treated with the different apoptotic agents, were double-stained for active caspase-3 and TUNEL reactivity using an anti-cleaved caspase-3 antibody (Cell Signaling Technologies, Beverly, MA) and a TUNEL cell death detection kit (Roche Molecular Biochemicals, Penzberg, Germany), respectively. Sections were permeabilized for 30 minutes at room temperature with 0.1% Triton X-100 in 0.1% sodium citrate (pH 4.7). After washing in phosphate-buffered saline (PBS), the sections were incubated for 1 hour at 37°C in a reaction mixture (200 mM potassium cacodylate, 25 mM Tris-HCl [pH 6.6], 0.2 mM EDTA, and 0.25 mg/mL bovine serum albumin) containing terminal deoxynucleotidyl transferase (0.2 U/μL) and fluorescein isothiocyanate–labeled deoxyuridine triphosphate. After being washed in PBS, sections were further incubated for 1 hour at room temperature with the anti-cleaved caspase-3 antibody. Following washing in PBS, the sections were incubated for 30 minutes at room temperature with PE-conjugated donkey anti-rabbit secondary antibody (Jackson ImmunoResearch Laboratory, Hamburg, Germany). After a final washing in PBS, sections were embedded in fluorescence-mounting medium containing DAPI (4′,6-diamidino-2-phenylindole; Vector Laboratories, Burlingame, CA). Cells positive for active caspase-3 or TUNEL were counted in 4 microscopic fields at 400× magnification. Sections were also investigated for the expression of TRAIL receptors using anti-TRAIL-R1 and anti-TRAIL-R2 antibodies from Immunex (Seattle, WA) and peroxidase-coupled secondary antibodies as described.30

RNA Preparation and Quantitative Real-Time PCR.

Liver tissue was disrupted in liquid nitrogen, and total RNA was prepared using an RNeasy kit (Qiagen, Hilden, Germany) and DNaseI digestion according to the manufacturer's instruction. For analysis of TRAIL-R1, TRAIL-R2, Bcl-2, Bax, and Puma expression, QuantiTect primers were purchased from Qiagen. Expression of FLIPL and FLIPS was analyzed with the primers FLIPL (forward; 5′-cctaggaatctgcctgataatcga 3′), FLIPL (reverse; 5′-tgggatataccatgcatactgagatg-3′); FLIPS (forward; 5′-gcagcaatccaaaagagtctca-3′), and FLIPS (reverse; 5′-atttccaagaattttcagatcagga3′). Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as an internal reference. Real-time RT-PCR was carried out in triplicate with 20 ng of RNA employing a SYBR Green RT-PCR kit (Qiagen) and an Applied Biosystems 7300 system (Darmstadt, Germany) under the following conditions: 50°C for 30 minutes, 95°C for 15 minutes, followed by 40 cycles of 94°C for 15 seconds, 54°C for 30 seconds, and 72°C for 30 seconds. In each cycle, fluorescence was measured at 72°C. To verify the specificity of the PCR products, melting curve analysis was performed from 60°C to 95°C with 0.2°C/sec intervals and stepwise fluorescence acquisition. Data were analyzed employing the comparative (ΔΔCT) method after normalization to GAPDH expression.

Statistical Analysis.

Statistical analysis was performed using the 2-tailed t test for equality of means (SPSS 1.5 software). A P value less than .05 was considered significant.


TRAIL Exerted Marginal Cytotoxic Effects on Liver Explants Compared with Isolated PHHs.

In initial experiments, we assessed the potential cytotoxicity of different TRAIL agonists in freshly isolated, normal PHHs of different donors. We incubated PHHs 16 hours after cell isolation with either untagged soluble TRAIL (0.TRAIL) or TRAIL forms fused with leucine zipper (LZ) and His tags. A luminescent caspase assay that is based on cleavage of the luciferin-coupled caspase substrate DEVD and that allows the early detection of caspase activity was employed as a sensitive measure for cytotoxicity.30 Compared to untreated control cells, a significant increase in caspase activity (4.1- ± 0.7-fold, P < 0.05) was observed when hepatocytes were treated for 6 hours with 50 ng/mL of LZ-TRAIL (Fig. 1A). Higher concentrations of TRAIL did not significantly increase caspase activation further. In view of the reported pharmacokinetics of TRAIL in vivo,31, 32 in all subsequent experiments a therapeutically relevant concentration of 50 ng/mL of TRAIL was used. Also, untagged TRAIL (50 ng/mL) significantly increased caspase activity (2.5- ± 0.5-fold, P < 0.05) over that in the untreated hepatocytes. In addition, toxic effects were observed with 50 ng/mL of monomeric His-tagged TRAIL, which led to a 2.4- ± 0.4-fold elevation of caspase activity (P < 0.05).

Figure 1.

Effect of TRAIL on caspase activation in primary human hepatocytes and liver explants. (A) PHHs from different donors (n = 5) were treated in duplicate 16 hours after isolation with TRAIL agonists including untagged TRAIL (0.TRAIL) or TRAIL tagged with a His and leucine zipper (LZ) sequence. After 6 hours of incubation, caspase activity was assessed in duplicate by a luminometric enzyme assay. The data showed a relative increase in caspase activity (mean ± standard error of the mean [SEM]) compared with that of the untreated control. (B) Comparison of caspase activity in liver explants and PHHs isolated from the same healthy individuals (n = 2). PHHs and liver slices were incubated in duplicate with the different TRAIL versions. Caspase activity was assessed after 6 hours by a luminometric enzyme assay. The mean ± standard deviation of a representative experiment is shown. Compared with untreated control liver or hepatocytes, no or only marginal caspase activity was observed when healthy liver explants were treated with the different TRAIL agonists. (C) Assessment of caspase activity in healthy liver explants isolated from 11 individuals. Liver explants were incubated as described in B with different TRAIL versions and analyzed in duplicate for caspase activation. TRAIL induced only marginal caspase activity.

The elevated caspase activation in TRAIL-treated PHHs also resulted in increased cell death, as assessed by the release of LDH and the formation of cells with hypodiploid DNA (data not shown). In accordance with previous reports,22 however, we noticed that the extent of cell death of isolated PHHs was dependent on the duration and conditions of the culture. We therefore sought to establish a model that more closely mimics in vivo conditions and, to this end, used organotypic cultures of liver explants. The advantage of liver explants is the circumventing of a long preincubation of the cells, which may lead to hepatocyte dedifferentiation, and of the use of collagenase digestion, which can alter cell function and damage cells.33 Furthermore, in organotypic culture, the complexity of liver architecture—cellular heterogeneity, cell–cell interactions and bipolarity of hepatocytes—is largely maintained.34, 35 Using caspase and LDH measurements, we first verified that liver explants can be incubated in vitro for at least 8 hours without a significant loss of cell viability (data not shown).

To directly analyze TRAIL effects in this ex vivo system, we compared the cytotoxicity of the different TRAIL versions in isolated PHHs with the effects in the liver explants isolated from the same individuals. A representative experiment is shown in Fig. 1B. Interestingly, neither untagged TRAIL nor His-TRAIL (50 ng/mL) resulted in elevated caspase activity in liver tissue, whereas with both TRAIL versions, a considerable increase in caspase activity was observed in the corresponding hepatocytes isolated from the same individual (Fig. 1B). These differences were most obvious with LZ-TRAIL, which induced a 4.5- ± 0.5-fold increase in caspase activity in PHHs but almost no caspase activation in the liver tissue.

As TRAIL agonists obviously exert little or no toxicity in liver tissue compared to isolated PHHs, we next analyzed caspase activity in a total of 11 healthy liver explants (Fig. 1C). In all samples, a marginal but not significant increase (1.2- ± 0.08-fold) in liver toxicity was observed when healthy liver explants were treated with the untagged TRAIL version. Similar results were obtained with His-TRAIL or LZ-TRAIL. Thus, TRAIL agonists are not toxic to healthy liver explants in marked contrast to their effects on PHHs.

CD95L But Not TRAIL Induced Caspase Activation and Apoptosis in Normal Liver Explants.

To investigate whether TRAIL-induced caspase activation results in apoptosis, we prepared cryosections of the treated liver explants and performed double staining for caspase-3 activation and DNA fragmentation. As a positive control, we also incubated the liver explants with Flag-tagged CD95L in the presence of crosslinking anti-Flag antibody, a condition known to induce severe liver toxicity. Untreated control livers showed neither caspase activation nor TUNEL reactivity (Fig. 2A-C). In contrast, treatment of healthy liver explants with CD95L strongly increased both caspase-3 activity and TUNEL reactivity (Fig. 2D-F). Interestingly, most hepatocytes with caspase-3 activation revealed TUNEL reactivity, underscoring not only the reliability of the staining but also the applicability of our experimental system. Treatment of liver tissue with untagged TRAIL only resulted in very weak caspase-3 activation and DNA fragmentation (Fig. 2J-L). Only a few single scattered cells positive for active caspase-3 and DNA fragmentation were detected in livers treated with LZ-tagged TRAIL (Fig. 2M-O). Furthermore, similar to that in untreated tissue, almost no caspase-3 and TUNEL reactivity was observed when liver explants were treated with His-tagged TRAIL (Fig. 2G-I).

Figure 2.

Immunohistochemical double-staining of healthy liver explants for active caspase-3 and TUNEL reactivity. Liver explants were either left untreated or treated with Flag-tagged CD95L or the different TRAIL agonists as described in Figure 1. After 6 hours, cryostat sections were prepared and stained with TUNEL or an anti-cleaved caspase-3 antibody. Results of a representative example of the 7 experiments performed with different livers are shown. Compared with the almost no caspase activity or TUNEL reactivity shown in untreated controls (A-C), treatment of healthy liver tissue with CD95L strongly increased DNA fragmentation (D) and caspase activity (E). Most caspase-3-positive hepatocytes showed TUNEL reactivity (F). In contrast, no or only marginal caspase activity and TUNEL reactivity were observed after treatment of healthy liver explants with the indicated TRAIL agonists (G-O). Magnification: 400×.

TRAIL in Combination with HDAC Inhibitors Is Hepatotoxic.

Many compounds sensitize tumor cells to TRAIL-induced apoptosis and are therefore tested in clinical trials.9 One such class of antitumor agents is HDAC inhibitors, which modify chromatin structure and potentiate TRAIL-induced apoptosis by various mechanisms.10–15 To investigate whether the combination of TRAIL with HDAC inhibitors exerts liver toxicity, we treated 7 liver explants with the HDAC inhibitor depsipeptide, alone or in combination with the TRAIL agonists. For these experiments, a concentration of depsipeptide (10 nM) was employed that is commonly used in vitro and is within the range of plasma concentrations achieved in clinical trials.36 Immunohistochemical analysis revealed that both depsipeptide (Fig. 3) and sodium valproate, another HDAC inhibitor (data not shown), induced caspase-3 activation and TUNEL reactivity in healthy liver tissue, although their effects were considerably less than that observed with CD95L. Interestingly, a combination of depsipeptide with either His-TRAIL or untagged TRAIL strongly increased caspase-3 activation and DNA fragmentation compared to that with the respective agents alone (Fig. 3). Most caspase-3 positive cells also displayed TUNEL reactivity, indicating that the different TRAIL versions in combination with depsipeptide induced hepatocyte apoptosis. Staining of serial sections revealed a similar cytotoxicity in the central and peripheral areas of the explants (data not shown), suggesting an equal distribution of the agents in the different regions of the liver.

Figure 3.

HDAC inhibition sensitized healthy liver tissue for TRAIL-mediated caspase-3 activation and DNA fragmentation. Liver explants were preincubated for 2 hours with the HDAC inhibitor depsipeptide (10 nM) followed by treatment with the TRAIL agonists for 6 additional hours, as described in Fig. 2. Control liver explants were either left untreated or treated with depsipeptide alone for 8 hours. Thereafter, immunostaining of cryostat sections prepared from the liver tissues for active caspase-3 and TUNEL reactivity was performed. Results of a representative example of the 7 experiments performed with 7 different livers are shown. Magnification: 400×.

To quantify TRAIL-plus-depsipeptide-induced hepatotoxicity, cells positive for active caspase-3 (Fig. 4A) or DNA fragmentation (Fig. 4B) were counted at 400× magnification in 4 microscopic fields of the treated liver explants. Untreated liver tissue showed a mean of 13 ± 1.6 active caspase-3-positive and 18 ± 1.7 TUNEL-positive cells/mm2. Treatment with CD95L significantly (P < 0.01) increased the number of cells positive for active caspase-3 (174 ± 7.8/mm2) and DNA fragmentation (194 ± 5.5/mm2). Depsipeptide-treated liver showed a mean of 87 ± 5.8 and 86 ± 4.2 cells/mm2 cells positive for active caspase-3 and TUNEL (P < 0.01), respectively. Compared with that in untreated controls, no significant differences in caspase-3 and TUNEL reactivity were observed when liver tissues were incubated with His-TRAIL or untagged TRAIL. Importantly, however, compared with TRAIL or depsipeptide alone, with a combination of TRAIL agonists and the HDAC inhibitor, a strong and significant (P < 0.01) increase in hepatotoxicity was observed. When combined with depsipeptide, caspase activation and TUNEL reactivity were found in approximately 180 cells/mm2 of explants treated with His-TRAIL. A very similar induction of caspase activation and TUNEL reactivity was obtained by a combination of depsipeptide and untagged TRAIL (Fig. 4).

Figure 4.

Quantification of caspase-3 activation and TUNEL staining after treatment of healthy liver explants with TRAIL and depsipeptide. Liver tissues (n = 7) were either left untreated or incubated with CD95L or the indicated TRAIL agonists with and without the presence of depsipeptide, as described in Figures 2 and 3. The number of cells (mean ± SEM) positive for (A) active caspase-3 or (B) TUNEL reactivity was assessed by counting 4 microscopic fields at 400× magnification. Untreated control livers as well as liver explants treated with TRAIL showed almost no caspase-3 activation and DNA fragmentation. Depsipeptide treatment resulted in increased caspase-3 activity and TUNEL reactivity but to a significantly lower extent than Flag-tagged CD95L (P < 0.01). Importantly, the combination of the indicated TRAIL agonists with depsipeptide strongly (P < 0.01) increased caspase-3 activation and TUNEL reactivity compared with the respective agents alone.

TRAIL Alone Induced Cytotoxicity in HCV-Infected and Steatotic Liver.

Death receptor–mediated apoptosis has been implicated in a variety of liver diseases, in particular liver steatosis and HCV infection.37, 38 So far, it is entirely unknown whether the TRAIL sensitivity of hepatocytes or other primary cells is altered in pathological situations. We therefore addressed this question by incubating liver explants from patients with steatosis or HCV infection with TRAIL and CD95L. In contrast with healthy liver (Fig. 3), in steatotic and HCV-infected livers, immunostaining for active caspase-3 and TUNEL reactivity was already considerably pronounced in the absence of a death ligand (Fig. 5A,B). Moreover, treatment with CD95L resulted in increased caspase-3 activity and TUNEL reactivity in the diseased livers, although normal liver explants were also highly sensitive to CD95L.

Figure 5.

Effect of TRAIL agonists on diseased liver tissue. Detection of caspase-3 activity and apoptosis in explants from healthy liver (n = 7), steatotic liver (n = 5), and HCV-infected liver (n = 9). (A) Steatotic and (B) HCV-infected liver tissues were treated for 6 hours with the indicated TRAIL agonists and CD95L and assessed for caspase-3 activity and TUNEL by immunohistochemistry. The number of cells positive for (C) active caspase-3 or (D) TUNEL reactivity was assessed as described in Figure 4. The results showed the mean number ± SEM of active caspase-3- or TUNEL-positive cells per square millimeter of liver tissue. TRAIL-induced caspase activation and apoptosis were considerably more pronounced in steatotic and HCV-infected liver explants (P < 0.01) than in healthy livers.

Most striking, however, was our finding that both His-tagged and LZ-TRAIL, which exerted almost no cytotoxic effect in normal liver, strongly induced caspase activity and TUNEL reactivity in both the steatotic and HCV-infected livers (Fig. 5A,B). Although untreated HCV-infected liver revealed a mean of 74 ± 4 cells/mm2 positive for active caspase-3 and a mean of 86 ± 5 cells/mm2 positive for TUNEL, treatment with His-TRAIL led to a more than 3-fold (P < 0.01) elevation of both caspase activation and DNA fragmentation in these liver tissues (Fig. 5C,D). Incubation of HCV-infected liver with LZ-TRAIL resulted in caspase activation and apoptosis to a very similar extent. Furthermore, apoptosis in the liver explants was almost completely abolished by the caspase-8 inhibitor IETD-fmk and TRAIL-neutralizing antibodies (data not shown), indicating that it was indeed TRAIL mediated and not a result of unspecific effects. In addition to HCV-infected liver, we obtained almost the same findings in explants from steatotic livers, which in the absence of TRAIL showed a mean of 79 ± 3.1 and 94 ± 4.4 cells/mm2 positive for active caspase-3 and TUNEL reactivity, respectively. In contrast, after treatment of steatotic liver tissues with His-tagged TRAIL, an average of 195 ± 6.4 active caspase-3-positive cells/mm2 and of 225 ± 8.3 TUNEL-positive cells/mm2 were detected (P < 0.01 compared to healthy liver). Treatment of steatotic livers with LZ-TRAIL had very similar effects (Fig. 5C,D). In the diseased livers, we also quantified the cytotoxic effect of TRAIL in combination with HDAC inhibitors. Although TRAIL alone was hepatoxic to steatotic livers and increased caspase-3 activation and TUNEL reactivity to an extent almost comparable to CD95L, the combination with depsipeptide increased apoptosis further (data not shown). Thus, these results clearly indicate that TRAIL is potentially cytotoxic to diseased but not to normal liver tissue, whereas CD95L induces apoptosis in both healthy and diseased liver.

Expression of TRAIL Receptors and Bcl-2 Proteins Is Altered by HDAC Inhibition and in Liver Disease.

To obtain insights into potential mechanisms of the TRAIL sensitivity in liver disease and in response to HDAC inhibition, we performed real-time RT-PCR analyses and investigated the expression of several apoptotic regulators. Treatment of normal liver explants with depsipeptide resulted in strong up-regulation of TRAIL-R2, whereas mRNA expression of TRAIL-R1 remained relatively unchanged (Fig. 6A). In contrast, expression of both isoforms of FLIP—FLIPL, and FLIPS—as well as of Bcl-2 were significantly reduced by depsipeptide. HDAC inhibition further resulted in increased expression of proapoptotic Bcl-2 proteins such as Bax, Puma (Fig. 6A), and Bim (data not shown). These data are in line with previous reports demonstrating that HDAC inhibitors can affect both the mitochondrial and death receptor–mediated pathways of apoptosis (reviewed by Bolden et al.39). In steatotic liver both TRAIL receptors were up-regulated, and expression of Puma and Bax also was increased, compared with healthy liver tissue (Fig. 6B). Even more enhanced transcript expression of TRAIL-R1 and in particular TRAIL-R2 was seen in HCV-infected liver that also revealed up-regulation of Puma and concomitant down-regulation of Bcl-2 (Fig. 6C). The increased expression of TRAIL receptors was substantiated by immunohistochemistry. Whereas TRAIL-R1 and TRAIL-R2 were barely detectable in normal liver, HCV-infected liver showed pronounced expression of both receptors (Fig. 6D). Although the limited amount of mRNA obtained from liver biopsies did not allow us to further investigate Bcl-2 proteins, these results show that liver disease is characterized by marked changes in the balance of pro- and antiapoptotic mediators. Therefore, the elevated expression of TRAIL receptors and proapoptotic Bcl-2 proteins might explain the increased sensitivity of steatotic and HCV-infected livers to treatment with recombinant TRAIL.

Figure 6.

Expression of TRAIL receptors, FLIP, and Bcl-2 proteins in response to HDAC inhibition and liver disease. (A) Explants from healthy liver tissue (n = 5) were either left untreated or incubated with 10 nM depsipeptide. Total RNA was isolated after 8 hours and subjected to real-time RT-PCR using primers for TRAIL-R1, TRAIL-R2, the short and long FLIP isoforms, Puma, Bax, and Bcl-2. Transcript expression in depsipeptide-treated livers relative to the untreated controls is given as mean ± SEM. The black bars indicate proapoptotic mediators and the gray bars antiapoptotic mediators. Transcript expression in (B) steatotic and (C) HCV-infected livers relative to healthy liver samples. RNA was isolated from 5 normal livers, 5 livers from steatotic patients, and 5 livers from HCV-infected patients. Expression of the TRAIL receptors was analyzed in 9 healthy and 12 HCV-infected livers. (*P < 0.05; **P < 0.01). (D) Up-regulation of TRAIL receptors in HCV-infected livers. Immunohistochemistry was performed for TRAIL-R1 and TRAIL-R2 using biopsies from HCV-infected and healthy livers. No staining was obtained in the absence of the primary antibodies. The photomicrographs demonstrate a strong up-regulation of TRAIL-R1 and TRAIL-R2 in HCV-infected livers compared with that in normal livers.


TRAIL has been shown not only to induce apoptosis in a variety of cancer cells in vitro but also to exert potent antitumor activity in cancer xenograft models.16, 17 Furthermore, preclinical safety studies in nonhuman primates did not show adverse reactions even when substantial doses of recombinant TRAIL were used. One of the most appealing features of TRAIL is that it does not seem to have the extreme liver toxicity that has precluded in vivo testing of CD95L and TNF, which cause massive hemorrhagic liver necrosis. Therefore, it was initially believed that TRAIL could be safely used as cancer therapy without damaging normal tissue. However, this early view was based on studies in laboratory animals and has now been challenged by reports demonstrating TRAIL-mediated apoptosis in isolated human hepatocytes.18, 40–42 It has been shown that, in contrast to hepatocytes from mice, rats and monkeys, human hepatocytes are sensitive to apoptosis mediated through a His-tagged version of TRAIL, raising the concern that cytokine therapy with TRAIL might cause severe hepatotoxicity. Another report demonstrated differential hepatotoxicity of the various recombinant TRAIL versions.21 Therefore, it is currently unclear whether this toxicity in isolated hepatocytes is dependent on the preparation and version of the recombinant TRAIL used. Moreover, there is virtually no information available whether the biological activity of TRAIL is different in the liver in vivo and in isolated hepatocytes and, more importantly, whether liver disease affects the sensitivity to TRAIL. Thus, our knowledge about the potential side effects and hepatotoxicity of TRAIL is still very limited. Our study was therefore the first report that employed treatment of organotypic liver cultures and that investigated the effects of TRAIL in liver diseases.

In this study we used different versions of recombinant TRAIL, and we observed very similar effects with agonistic TRAIL receptor antibodies (data not shown). As the various TRAIL proteins were prepared in different laboratories, which might have influenced their biological properties and toxicity profiles, it was not the intention of this study to compare the different TRAIL forms in detail. We found that TRAIL was able to induce significant caspase activation in isolated PHHs, reflecting potential cytotoxicity, which was certainly also dependent on the TRAIL version used. This finding is in line with recent reports demonstrating strong apoptosis induction and caspase activation in PHHs after treatment with TRAIL, agonistic TRAIL-R antibodies, or expressed adenoviral TRAIL.40–43 Our findings in isolated PHHs are also in accordance with the observation that, in addition to hepatocytes, other isolated primary human cells, such as prostate and thyroid epithelial cells, microvascular endothelial cells, keratinocytes and brain cells,19,44-46might be susceptible to TRAIL. Thus, the toxic effects of TRAIL observed are obviously not restricted solely to cultured PHHs.

Although PHHs are widely used for liver studies, they have various limitations. PHHs are different from hepatocytes in situ,23, 33 and the conclusion that TRAIL is hepatotoxic solely based on its effect on isolated PHHs is certainly premature. Therefore, an important finding of our study is that, unlike isolated PHHs, TRAIL exerted almost no or only a very moderate cytotoxic effect on caspase activation and apoptosis (i.e., TUNEL reactivity) in healthy liver explants. The lack of cytotoxicity of TRAIL in these liver explants was not a result of the experimental conditions, such as, for instance, weak penetration of the proteins into liver tissue. This caveat can be largely excluded because (1) in contrast to TRAIL, soluble CD95L did induce caspase-3 activation, (2) TRAIL was able to induce hepatocyte apoptosis in combination with HDAC inhibitors and (3) TRAIL hepatotoxicity was detected in diseased liver. Our results therefore suggest that phenotypic differences between hepatocytes, which are known to change rapidly during culture,33, 34 might be responsible for the differences in TRAIL sensitivity. In this context, the expression and intracellular distribution of TRAIL-R2 or the caspase-8 inhibitor FLIP, both of which are short-lived and highly inducible gene products, might rapidly alter during the isolation procedure. It should also be taken into account that differences in the cell cycle between hepatocytes in vivo and in culture47 might influence their susceptibility to TRAIL. Because liver explants retain an intact structure that presumably avoids dedifferentiation of hepatocytes and induction of apoptosis-relevant stress genes, our ex vivo model should more truly reflect the physiological situation than would isolated hepatocytes. It must be emphasized that we observed no background cytotoxicity in the liver explants within an 8-hour incubation period, suggesting adequate oxygen and nutrient exchanges, which is in line with studies using similar liver slices (reviewed in Elaut et al.,33 Berry et al.,34 and Gebhardt et al.35).

Our observation that HDAC inhibitors substantially sensitize normal hepatocytes in situ to apoptosis might have important implications for potential applications of TRAIL in combined antitumor therapies. HDAC inhibitors are being developed by several companies as a specific strategy for cancer treatment.39 It was reported that HDAC inhibitors can sensitize even highly resistant tumor cell lines and primary tumor cells to TRAIL-induced apoptosis.10, 12, 15 Our results therefore suggest that any attempts to combine TRAIL with HDAC inhibitors will have to be approached cautiously. It will also be mandatory to investigate whether conventional chemotherapeutic drugs that sensitize tumor cells to TRAIL-induced apoptosis exert liver toxic effects when used in combination with TRAIL. So far, these important issues have not been adequately addressed.

Another important implication of our study concerns the observed toxicity of TRAIL in diseased versus in normal liver tissue. The sensitivity of hepatocytes in situ was dramatically increased in inflammatory conditions, that is, liver steatosis and HCV infection. We and others have recently observed that patients with liver steatosis exhibit strongly elevated caspase activation, which closely correlates with the severity of their fatty liver disease.30, 48 Moreover, adenoviral expression of TRAIL in mouse liver was shown to induce liver apoptosis and steatosis.49 This is in accordance with experimental models of hepatitis, in which hepatic cell death was strongly reduced in TRAIL-deficient mice or mice treated with a blocking TRAIL receptor.50 Thus, it seems that the ability of TRAIL to induce apoptosis in hepatocytes might strongly depend on the inflammatory context. This view is supported by the recent observation that TRAIL is highly synergistic with CD95L in the induction of hepatocyte apoptosis.51

Our data in liver explants suggest that TRAIL sensitivity of hepatocytes is not only strongly increased in fatty liver but also in chronic HCV infection. In both diseases apoptosis has been recognized as an important feature of liver injury.37, 52–55 The present data and a recent study of steatohepatitis patients56 indicate that the increased TRAIL sensitivity in diseased liver is mediated by the elevated expression of TRAIL receptors. Apoptosis sensitivity might be further enhanced by altered expression profiles of Bcl-2 proteins or other cytotoxic events that might render diseased liver generally more susceptible to injury, regardless of the presence of death ligands. Therefore, it will be highly intriguing to investigate whether TRAIL- or caspase-neutralizing agents influence disease progression. In conclusion, our results suggest that human clinical trials with TRAIL should be performed with great caution, in particular under conditions when TRAIL is combined with other chemotherapeutic agents or when TRAIL is administered to patients with inflammatory liver diseases. However, both the advantages and disadvantages need to be considered because the benefits for cancer patients could potentially outweigh some disadvantages of TRAIL treatment.


The authors thank Drs. I. Schmitz, E. Sausville, and H. Walczak for valuable reagents.