Potential conflict of interest: Nothing to report.
Hepatocellular carcinoma (HCC) ranks among the 10 most common cancers worldwide. The fact that HCC is resistant to conventional chemotherapy and is rarely amenable to radiotherapy leaves this disease with no effective therapeutic options and a very poor prognosis. Therefore, the development of more effective therapeutic tools and strategies is much needed. HCCs are phenotypically and genetically heterogeneous tumors that commonly emerge on a background of chronic liver diseases, most of which culminate in cirrhosis, such as alcoholic cirrhosis and chronic hepatitis B and C infections. This review outlines recent findings on the progression of liver disease, including our knowledge of the role of apoptotic processes, with an emphasis on the tumor necrosis factor–related apoptosis-inducing ligand (TRAIL). The proapoptotic and antiapoptotic properties of TRAIL, its involvement in liver injury, and its potential as a therapeutic agent in fibrosis and HCC are discussed. Several contradictory and confusing data have not yet been resolved or placed into perspective, such as the influence of factors that determine the TRAIL sensitivity of target cells, including the tumor microenvironment or cirrhotic tissue. Therefore, we assess these data from the perspectives of gastroenterologists (P.S. and M.W.B.) and a molecular oncologist (I.H.) with research interests in liver injury, apoptosis, and experimental therapeutics. (HEPATOLOGY 2007;46:266–274.)
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Apoptosis is a fundamental process in tissue homeostasis and is necessary for the balance between cell loss and cell gain in normal tissue. Apoptosis is nature's way of eliminating unwanted, senescent, virally infected, and damaged cells from multicellular organisms.1 The liver is no exception, and tissue repair, inflammation, regeneration, and fibrosis may all be triggered by apoptosis.2 Apoptosis is the first cellular response to many toxic events and accompanies viral hepatitis, alcohol-induced liver disease, nonalcoholic fatty liver disease, cholestatic liver disease, and ischemia/reperfusion injury.3, 4 Moreover, hepatocyte apoptosis is significantly increased in patients with alcoholic hepatitis and nonalcoholic steatohepatitis and correlates with disease severity and hepatic fibrosis.4 Thus, the dysregulation of apoptosis is a principal mechanism contributing to many liver diseases, including both acute liver injury and chronic liver diseases.2
Apoptosis can occur by 2 fundamental pathways: the death receptor or extrinsic pathway and the intracellular organelle–based intrinsic pathway (Fig. 1). The regulation of the apoptotic machinery in liver cells is complex but appears to involve the activation of death receptors.2 These receptors include CD95, tumor necrosis factor (TNF) receptor 1, and tumor necrosis factor–related apoptosis-inducing ligand (TRAIL) receptors 1 and 2 (TRAIL-R1/DR4 and TRAIL-R2/DR5). Upon the binding of ligands, the death receptors oligomerize and form a death-inducing signaling complex (DISC) by recruiting the adapter molecule FADD, caspase-8, and FLIP, which is an inhibitor of caspase-8.5 At the DISC, caspase-8 is autocatalytically activated and activates downstream effector caspases (e.g., caspase-3) by 1 of 2 routes: a mitochondria-dependent pathway and a mitochondria-independent pathway.6 In the former, the caspase-8–mediated cleavage of Bid, a proapoptotic member of the Bcl-2 family, converts Bid into an active form that triggers cytochrome c release from mitochondria. Cytochrome c binds apoptotic protease activating factor 1, thereby activating another initiator caspase, caspase-9, which in turn activates effector caspases. In the mitochondria-independent pathway, caspase-8 is initially activated by the same mechanism and then directly cleaves and activates caspase-3, bypassing mitochondria and cytochrome c release. Dependent on the balance between different cellular signaling pathways such as nuclear factor-κ B (NF-κB) and c-Jun N-terminal kinase/AP-1, a cellular stress signal induces apoptosis or switches the cellular balance to resistance.
Therapeutic Intervention in Liver Disease with TRAIL as a Potential Candidate
To avoid the deleterious side effects of apoptosis stimuli on healthy liver tissue, substances with specific antitumor activity but without high toxicity to normal cells are needed. The death ligand TRAIL may fulfill these requirements because no toxic effects have been observed in either experimental animals or primates; as a result, clinical trials have been initiated with the recombinant molecule.7 Following an initial debate on the potential liver toxicity of TRAIL, it became evident that only highly aggregated recombinant forms of TRAIL are toxic to freshly isolated primary human hepatocytes.8 Dependent on the aggregation status, membrane association, and zinc content, different variants of TRAIL vary in their ability to induce apoptosis.9 Although the recent reports of hepatocyte death after treatment with TRAIL in vitro must be taken seriously, one should keep in mind that isolated hepatocytes differ from hepatocytes in situ in several important ways. The isolation of hepatocytes may have contributed to sensitizing them to TRAIL-induced apoptosis, whereas hepatocytes in the context of an intact liver still were not susceptible to TRAIL-induced apoptosis. Also, trimeric TRAIL does not show significant toxic effects on most other primary cells or after injection in monkeys.10–13 As such hepatocyte-safe variants of TRAIL nevertheless induce cell death in a variety of tumor cell lines,14 TRAIL and specific agonistic antibodies have attracted considerable attention for their potential use for the selective induction of apoptosis in activated stellate cells and virus-infected and malignant hepatocytes without associated collateral damage to normal hepatocytes.
TRAIL is expressed on the surface of activated immune cells such as natural killer (NK) cells, T cells, macrophages, and dendritic cells, where it apparently functions as an immune effector molecule, mediating antitumor cytotoxicity and immune surveillance.15–19 Given the efficacy of NK cells in selectively killing abnormal cells and destroying most solid tumor cell targets, a variety of approaches have been used to selectively augment the NK cell response to tumors.20 Importantly, NK cells are quite abundant in the liver of mice, in contrast to a relatively small percentage in the peripheral lymphatics, and they mediate higher cytotoxic activity against tumor cells than spleen or peripheral NK cells in rodents.21 The underlying reason may be that TRAIL expression is highly inducible by interleukin-2 (IL-2) in liver NK cells, whereas this effect has been barely observed in peripheral NK cells.22 This finding has recently been used in a preclinical approach to enhance the efficacy of liver transplantation, for which recurrent hepatocellular carcinoma (HCC) is one of the most fatal complications. A solution for this problem could be the adoptive transfer of IL-2–stimulated NK cells extracted from donor liver graft perfusates, as suggested by Ishiyama et al.22 IL-2–stimulated donor liver cells inductively express TRAIL and induce apoptosis in HCC without causing toxicity against 1-haplotype identical recipient intact tissues. These findings raise a novel concept of preventing the recurrence of HCC after liver transplantation by the adoptive transfer of IL-2–stimulated and TRAIL-expressing NK cells extracted from donor liver grafts.
Function of TRAIL and Its Receptors in Liver Disease
Both TRAIL-R1 and TRAIL-R2 contain a conserved death domain motif and signal apoptosis. Two so-called TRAIL decoy receptors, decoy receptor TRAIL-R3 (DcR1) and decoy receptor TRAIL-R4 (DcR2),23 exist (Fig. 2), and a fifth soluble receptor called osteoprotegerin (OPG), initially identified as a receptor for RANKL/OPGL, was shown later to bind TRAIL.24 The other 3 receptors appear to act as decoy receptors because a death domain is missing. Therefore, the decoy receptors do not transduce a death signal, but competitive binding of TRAIL to the decoy receptors or the soluble receptor OPG can block TRAIL activity.25 The mRNA encoding TRAIL-R1 and TRAIL-R2 is widely expressed in normal tissues; however, receptor expression on the cell surface in normal cells, including hepatocytes and quiescent stellate cells, is absent or relatively low.26–29 In cell lines derived from numerous cancers, such as carcinomas of the liver, pancreas, ovary, uterus, colon, brain, lung, and breast and lymphoma, R1 and R2 proteins are expressed at high levels.29–31 Although there is convincing evidence that the expression of R1 and/or R2 is required for TRAIL-mediated apoptosis, a large number of studies indicate that the magnitude of receptor expression is not directly related to the induction of apoptosis, suggesting that there are additional determinants of sensitivity.7, 29, 32, 33
The TRAIL protein is expressed on the cell surface as a type II transmembrane protein. The molecule is released into the environment in a vesicle-associated form or is cleaved and exists as a soluble homotrimeric ligand stabilized by a zinc moiety.28, 34, 35 The full biological role of TRAIL has not yet been elucidated; however, it is clear that it plays a role in apoptosis in numerous physiological contexts. Studies of transgenic mice engineered to have TRAIL gene deletions have demonstrated that it plays an important role in antitumor immune surveillance.36 TRAIL seems to be broadly expressed by cells of the immune system as an antitumor agent because it is found in activated NK cells, monocytes, and CD4+ and CD8+ T cells.37–40 The TRAIL gene has upstream interferon (IFN)-responsive transcriptional regulatory elements, and this suggests that it may also be used to destroy virus-infected cells.41 The functional surface expression of TRAIL has often been associated with stimulation by cytokines, such as the IFN-γ treatment of monocytes.40 The up-regulation of the TRAIL protein may thus be an important underlying reason for the successful use of IFN as an approved treatment option for patients chronically infected with the hepatitis B and C viruses. Indeed, cells infected with reovirus, herpesvirus, hepatitis B virus (HBV), and HIV can up-regulate TRAIL to induce autocrine apoptosis.42–45 NK cells can use TRAIL to kill virus-infected cells in an IFN-dependent manner.46 In hepatocytes, TRAIL-mediated apoptosis in vivo has been shown to depend on triggering through viral infection.47 Thus, endogenous overexpression of TRAIL seems to enable the organism to selectively eliminate virally infected hepatocytes. TRAIL and its receptors in mouse hepatitis models display up-regulation in adenovirus infection, and liver cells have been demonstrated to be sensitized toward TRAIL-induced cytotoxicity by HBV replication in vitro. In patients with chronic hepatitis B, HBV infection may increase serum TRAIL,48 and HBV-infected hepatocytes can be killed by up-regulated TRAIL.47 Recently, Higuchi et al.49–51 reported that bile acids up-regulate TRAIL-R2 expression, and Zheng et al.52 demonstrated the essential contribution of TRAIL in concanavalin A–induced and Listeria-induced liver injury, indicating that TRAIL is also an important mediator of apoptosis in nonviral liver disease.
Surprisingly, Mundt et al.53 recently reported that TRAIL expression in virally infected livers of mice not only induced apoptosis of infected hepatocytes but also led to steatosis, a characteristic feature of liver diseases such as chronic hepatitis C, alcoholic liver disease, and nonalcoholic steatohepatitis. The authors started with the observation that the expression of TRAIL was increased in the livers of patients with hepatitis C virus–associated steatosis, and they then proceeded to investigate the functional relevance of these observations in animal models. They used adenoviral gene transfer to express TRAIL in mouse liver and found that the overexpression of TRAIL itself had no effect in the healthy liver. However, when the gene transfer of TRAIL was preceded by a high-dose adenoviral infection, this sensitized the liver to respond to TRAIL expression with steatosis and hepatocyte apoptosis. By looking at the expression of TRAIL receptors in the infected livers, the authors were able to show that viral infection down-regulates a TRAIL decoy receptor while increasing the expression of the death domain–associated R2, thereby providing a mechanism to explain how viral infection can amplify the effects of TRAIL. The induction of steatosis was specific for TRAIL because the overexpression of CD95-L resulted in apoptosis of hepatocytes without steatosis. Moreover, after alcohol intake, TRAIL expression led to hepatic steatosis, without apoptosis of hepatocytes, and this indicated that TRAIL-mediated apoptosis and steatosis may be independently modulated after viral infection and alcohol intake. These observations are of major significance because they show that TRAIL is a novel mediator of fatty liver disease that may provide a mechanism to explain the development of steatosis in hepatitis C virus infection. In addition, these data provide an exciting new insight into the pathogenesis of steatosis but also raise a note of caution for the clinical trials of TRAIL, which, at present, show considerable promise for the treatment of malignant diseases, including HCC.54, 55
Because activated stellate cells have been found to be responsible for the exuberant and unbalanced wound-healing response in cirrhosis, their selective removal would be a potential mechanism to attenuate liver fibrosis. There are 3 phases of stellate cell activation: initiation, perpetuation, and resolution.56 The resolution phase of stellate cell activation is characterized by apoptosis. TRAIL may potentially participate in the resolution phase because marked increases in the expression of R1 and R2 death receptors have been found in vitro and in vivo.57 R1 and R2 proteins are not or are only very weakly expressed in whole human liver, normal hepatocytes, and stellate cells (Table 1).30 Thus, selective agonists to TRAIL death receptors (e.g., the monoclonal antibody TRA 8, binding and thereby activating R2 without inducing apoptosis in normal hepatocytes) may potentially induce apoptosis in activated stellate cells without associated collateral damage to hepatocytes.
Table 1. Protein Expression of TRAIL and Its Receptors in Human Liver
+ = expression; − = no or very low expression; nd = not done. The references are shown as superscripts.
TRAIL-Induced Survival Signaling and Therapeutic Implications
Initially, the differential TRAIL sensitivity between normal and cancer cells was thought to be due to the presence of TRAIL decoy receptors on normal cells.58, 59 Thus, it was assumed that the relative distribution of the death-inducing receptors R1 and R2 versus the decoy receptors might determine whether TRAIL induces apoptosis or not in a given cell type. However, current evidence suggests that TRAIL-induced cell death is also regulated by other intracellular factors.29 TRAIL death receptors and TNF receptor 1 bind the adapter molecule TRADD,60, 61 and this explains the potent activation of NF-κB by TRAIL and TNF-α in contrast to CD95-L. Trauzold et al.62 recently demonstrated in pancreatic ductal adenocarcinoma that TRAIL and its death receptors not only stimulate apoptosis but also can engage nonapoptotic signaling pathways leading to the activation of protein kinase C, NF-κB, and mitogen-activated protein kinases. As these pathways stimulate the transcription of genes encoding antiapoptotic, angiogenic, mitogenic, and cell migration–stimulating factors, the possibility arises that, depending on the clinical situation, TRAIL treatment may exert unfavorable protumoral effects in patients. In line with this assumption, a recent report suggests that TRAIL promotes invasion and metastasis in an apoptosis-resistant pancreatic ductal adenocarcinoma in vitro and in vivo.14 Also, various other tumor cell lines have been found to be TRAIL-resistant,11 and it may well be that TRAIL in the context of resistant tumor cells induces tumor progression rather than apoptosis.
Several scenarios can be envisioned in which TRAIL predominantly activates NF-κB rather than the apoptosis pathway in cancer cells. First, the evasion of apoptosis is a hallmark of cancer, and many cancer cells develop resistance to apoptosis induced by death receptors through genetic and epigenetic mechanisms such as promoter silencing by methylation. In the case of TRAIL-induced cell death, this may involve down-regulation of the various proteins involved in the intrinsic or extrinsic pathways of caspase activation or up-regulation of antiapoptotic proteins of the Bcl-2 and IAP families. Second, recent studies suggest that the induction of apoptosis by death receptors may be dependent on the degree of receptor aggregation/multimerization, which may, in turn, depend on the concentration of the death ligand, its form (i.e., soluble versus membrane-bound), the relative expression of the death receptor(s) on the cell surface, and the array of growth factors and cytokines to which the cells are exposed.63 Thus, it is possible that under the conditions prevalent in the microenvironment of fibrogenic and cirrhotic liver tissue, TRAIL fails to induce apoptosis and predominantly activates the NF-κB pathway.
Sensitization of HCC and Activated Stellate Cells to TRAIL-Induced Apoptosis
One way of sensitizing tumor cells to TRAIL-induced apoptosis is the combination of TRAIL and chemotherapeutic drugs. This treatment synergistically suppresses tumor growth in vitro and in severe combined immunodeficient mice without significant toxic effects.32, 64 Similarly, studies of HCC cells have shown that a cotreatment with chemotherapeutic drugs sensitizes hepatoma cells to TRAIL-induced apoptosis.31, 65
The inhibition of NF-κB has been associated with apoptosis induction in many cancer cells,66 and the activation of rat hepatic stellate cells is associated with the induction of NF-κB.67 Therefore, the inhibition of NF-κB may be a potent mechanism for the direct induction of apoptosis in HCC and activated stellate cells or for their sensitization to apoptosis (e.g., by the ligation of TRAIL). Proteasome inhibitors are a new class of NF-κB inhibitors with great clinical potential in the treatment of different tumor entities.68, 69 NF-κB is normally complexed to an endogenous inhibitor protein, I-kappa B (I-κB), in the cytosol.70 NF-κB–activating stimuli result in the phosphorylation of I-κB, leading to its dissociation from NF-κB and degradation by the proteasome.71 Freed from I-κB inhibition, NF-κB translocates to the nucleus and functions as a transcription factor. NF-κB induces the expression of survival genes, including the antiapoptotic Bcl-2 family proteins Bcl-xL and A1.71 Proteasome inhibition prevents I-κB degradation, and this is followed by a blocked activation of NF-κB, an effect that can culminate in the loss of survival proteins and cell death.72 In addition to preventing NF-κB activation, proteasome inhibitors may also up-regulate TRAIL-R2 and the proapoptotic BH3-only protein, Bim.73, 74 Therefore, the sensitization of carcinoma cells and activated stellate cells to TRAIL-induced apoptosis by proteasome inhibitors may offer a more efficient tumor treatment, provided that the combinatorial treatment is not toxic to normal cells. In this respect, Ganten et al.75 showed that human HCC cells could be sensitized with proteasome inhibitors to TRAIL-induced apoptosis, whereas primary human hepatocytes remained resistant. In addition, Anan et al.76 demonstrated that proteasome inhibitors induce apoptosis in human and rat activated stellate cell lines by the inhibition of NF-κB activity and thereby reduce hepatic fibrosis. This effect includes proteasome-induced up-regulation of TRAIL and R2 protein expression in activated stellate cells.76 Bortezomib, the first proteasome and NF-κB inhibitor used as an anticancer drug, is considered a promising approach in antitumor therapy.75 However, high concentrations of this NF-κB inhibitor also sensitized normal hepatocytes for TRAIL-induced apoptosis by up-regulation of R1 and R2, and this suggests that combination therapy with bortezomib/TRAIL bears the risk of severe hepatotoxicity at high but clinically relevant concentrations of bortezomib.77 Other preclinical results provide a paradigm for the role of NF-κB in cancer.78 Luedde et al.78 showed that the sensitization of hepatocytes to death receptor–mediated apoptosis by the inhibition of the NF-κB activator NEMO/IKKγ (NF-kappa B essential modulator/I-kappa kinase gamma) causes steatohepatitis and HCC in the livers of mice. These results reveal that NEMO-mediated NF-κB activation in hepatocytes has an essential physiological function to prevent the spontaneous development of disease pathogenesis in this model and highlight the need for extensive research on the potential side effects of drugs targeting the IKK/NF-κB pathway. Therefore, before clinical studies are undertaken, the effect of proteasome inhibition on different cell types in the liver must be investigated. The potential inhibition of NF-κB activation in Kupffer cells would be anti-inflammatory and beneficial, whereas NF-κB inhibition in hepatocytes may inhibit survival pathways and promote liver injury.
Influence of the Tumor Microenvironment on TRAIL-Induced Apoptosis
TRAIL resistance might not arise entirely from biological aspects of the tumor cells themselves (Table 2) but instead could be a result of factors within the tumor microenvironment, such as fibroblasts, endothelial cells, and immune cells.79In vivo, TRAIL-sensitive tumor cells may readily become resistant by challenging interactions with TRAIL-expressing, tumor-infiltrating immune cells or stromal fibroblasts. Accordingly, it has been recently shown that a coculture of pancreatic tumor cells with fibroblasts leads to the activation of NF-κB and to apoptosis resistance.80 Thus, the role of the tumor stroma and microenvironment is receiving increasing attention from researchers as a critical factor that influences tumor sensitivity to therapies. Fibroblasts, for example, provide the necessary growth signals for survival81–83 by the secretion of proliferative factors, such as transforming growth factor β, matrix metalloproteinases (MMPs), and epidermal growth factor, that play a role in cancer transformation.84 Another factor secreted by stromal cells and fibroblasts is OPG. This decoy receptor for TRAIL could have a clear role in the modulation of TRAIL resistance because it has been proposed to be involved in the resistance of tumor cells to TRAIL.85, 86
Table 2. Factors That Render Tumor Cells Resistant to TRAIL-Induced Apoptosis
TRAIL-induced NF-κB signaling may lead to resistance, proliferation, and even metastasis in certain tumor microenvironments.14
Inflammation can boost the host responses to the tumor and somehow change the tumor microenvironment for enhanced antigen presentation and tissue access.87 Thus, a recent report suggests that not a lack of tumoricidal activity of T cells but rather the failure of T cells to infiltrate the tumor tissue may prevent tumor rejection.88 In this context, recent data demonstrate that effective tumor therapy requires a proinflammatory microenvironment that permits effector cells to extravasate and reject the tumor.89 Similarly, a large number of studies have shown that IFN-γ production by T cells is essential for tumor rejection and that IFN-γ has to act, for example, on tumor stromal cells.90 These proinflammatory cytokines may lead to greater leukocyte infiltration in the tumor and to IFN-γ–induced expression of TRAIL and its receptors in tumor and immune cells but, paradoxically, may also lead to increased resistance of the tumor by simultaneous activity of antiapoptotic pathways. Likewise, failure of immune cells to eradicate tumors could be the result of tumor heterogeneity, in which some cells in the tumor are resistant and others not. In this case, TRAIL-sensitive cells would be killed by the TRAIL application, whereas the remaining resistant tumor cells would survive and proliferate.91
Hypoxic environments, which are often present in growing solid tumors, are known to induce the up-regulation of hypoxia-inducible factor-1α. This transcription factor binds to hypoxia-response elements, thereby inducing many hypoxia-response genes,92 such as VEGF.93, 94 New blood vessel formation by VEGF provides a mechanism for oxygenation and metastasis of the tumor. A hypoxic environment has been shown to induce resistance to numerous different therapies, including TRAIL-induced apoptosis in human colon cancer cells, which are susceptible to TRAIL under normoxic conditions.95 Very recently, the promotion of migration and invasion in vitro has been reported for apoptosis-resistant cholangiocarcinoma cells and has been explained as a result of TRAIL-induced activation of NF-κB.96 Trauzold et al.14 showed in pancreatic carcinoma cells that TRAIL strongly induces the expression of IL-8 followed by increased distant metastasis of pancreatic tumors in vivo. MMPs, which are mainly produced by nonmalignant stromal cells, may be involved because these enzymes constitute a mechanism of tumor growth, invasion, and metastasis.97 Tissue inhibitors of MMPs have been suggested to be useful in combination therapy with TRAIL because an MMP-2/MMP-9 inhibitor caused reduced tumor growth and angiogenesis in nude mice.98 However, the role of tissue inhibitors of MMPs in combination with TRAIL needs to be further investigated to determine the extent to which these molecules can overcome microenvironmentally induced resistance to TRAIL.
Conclusion and Perspective
Molecules that directly activate the death receptors of TRAIL, such as agonistic monoclonal antibodies and recombinant TRAIL protein, are being developed as monotherapies and as part of combination therapies with existing chemotherapeutic drugs and other therapeutic modalities. Despite the encouraging experimental results to date, it remains to be determined whether these agents are capable of inhibiting fibrosis and tumor growth in patients with liver disease, and many obstacles must be overcome to perform appropriate evaluations of these agents in the clinic. In addition to the usual need to optimize the pharmacologic characteristics of any new class of agents, there are several preclinical and clinical developmental concerns that are uniquely applicable to TRAIL-targeting therapeutics. Perhaps one of the most important concerns is that the preclinical models used to evaluate these agents do not fully recapitulate the complexity of human liver diseases. Thus, to maximize the efficacy of TRAIL, it is important to consider not only the sensitivity of the target cells itself but also the potential of factors in the complex microenvironment to impact the sensitivity of the tumor or cirrhotic tissue as a whole.