Conflict of interest The authors have declared that they have no conflicts of interest.
Dr. J M Schattenberg, Department of Medicine, University Medical Center of the Johannes Gutenberg University Mainz, Mainz 55101, Germany. Email: email@example.com
Hepatocellular carcinoma (HCC) remains a disease with a poor prognosis despite recent advances in the pathophysiology and treatment. Although the disease is biologically heterogeneous, dysregulation of cellular proliferation and apoptosis both occur frequently and contribute to the malignant phenotype. Chronic liver disease is associated with intrahepatic inflammation which promotes dysregulation of cellular signaling pathways; this triggers proliferation and thus lays the ground for expansion of premalignant cells. Cancer emerges when immunological control fails and transformed cells develop resistance against cell death signaling pathways. The same mechanisms underlie the poor responsiveness of HCC towards chemotherapy. Only recently advances in understanding the signaling pathways involved has led to the development of an effective pharmacological therapy for advanced disease. The current review will discuss apoptosis signaling pathways and focus on apoptosis resistance of HCC involving derangements in cell death receptors (e.g. tumor necrosis factor-alpha [TNF], CD95/Apo-1, TNF-related apoptosis-inducing ligand [TRAIL]) and associated adapter molecules (e.g. FADD and FLIP) of apoptotic signaling pathways. In addition, the role of the transcription factor nuclear factor-kappaB (NFκB) and members of the B cell leukemia-2 (Bcl-2) family that contribute to the regulation of apoptosis in hepatocytes are discussed. Eventually, the delineation of cell death signaling pathways could contribute to the implementation of new therapeutic strategies to treat HCC.
The role of death receptors in hepatocellular carcinoma—starting trouble
Apoptosis is a highly preserved mechanism that contributes to tissue homeostasis through targeted elimination of single cells without disrupting the biological functionality of the tissue (Fig. 1). Depending on the initial site of activation, apoptosis can be initiated through an extrinsic or intrinsic pathway.1 Most prominent among the cytokines that can induce apoptosis of hepatocytes are the members of the TNF receptor superfamily, CD95 (Apo1/Fas), tumor necrosis factor alpha (TNF, CD120), and TNF-related apoptosis inducing ligand (TRAIL). These cytokines exert physiological functions through their cognate receptors, namely the CD95 receptor (Apo1/Fas receptor), TNF receptor type 1 (TNF-R1, p55/65, CD120a) and type 2 (TNF-R2, p75/80, CD120b), TRAIL receptor type 1 and type 2. The role of this cytokine family in hepatocarcinogenesis varies according to the subsequent intracellular signaling events (see Table 1). Failure of transformed cells to undergo apoptosis severely disrupts tissue homeostasis and allows proliferation of the resistant clone, a phenomenon that is frequently observed in HCC, and such failure correlates with decreased expression of the CD95 receptor.2,3
Table 1. Apoptosis receptor and ligands, their role in hepatocellular carcinoma (HCC)
Expression in HCC
CD95 ligand (Fas ligand)
• Elimination of HCC-reactive T-lymphocytes • Decreased sensitivity towards receptor-mediated apoptosis (immune evasion) • Increased resistance towards cytotoxic drugs
Secreted by inflammatory and immune-competent cells
Lymphotoxin α and β (LTα, LTβ), and LTβ receptor (LTβR)
Increased expression and signaling in HCC
• Inflammation and HCC • Upregulation in hepatocytes from viral hepatitis • Activation of NF-κB • Activation of proinflammatory lymphocytes
Constitutive expression in HCC
• Activation of NF-κB • Tumor-related angiogenesis • Proliferation of transformed cells
Paracrine and autocrine signaling
Anti-tumor activity and hepatocyte injury
• Apoptosis in transformed and non-transformed hepatocytes
In addition to downregulation of apoptosis receptors in HCC, increased expression and secretion of the CD95-ligand has been found.4 Thus the threshold to undergo apoptosis in transformed cells is increased and the malignant tissue is capable of inducing apoptosis in lymphocytes that are directed against HCC cells, thereby evading a potential immunological control mechanism. Decreased sensitivity towards the CD95 signaling pathway is closely related to the malignant phenotype of HCC and has been linked to a poor response to treatment with cytotoxic drugs, as well as the clinical outcome following resection.4–6
In contrast to the CD95 signaling pathway, TNF is a pleiotropic cytokine involved not only in apoptosis, but also with inflammation, hepatocyte protection and proliferation. Although TNF was initially identified as a factor that induces cell death in sarcoma, and polymorphisms of the TNF gene have been linked to the emergence of HCC, the role of TNF in hepatocarcinogenesis not clearly defined.7–9 The response of a cell towards TNF signaling is determined by the transcription factor NF-κB. If NF-κB is activated, hepatocyte survival and proliferation commences. Conversely, cells undergo apoptosis when NF-κB is transcriptionally inactive (see below). The proinflammatory cytokines lymphotoxin alpha (LTα) and beta (LTβ) activate the TNF receptor as well as the membrane bound LTβ receptor (LTβR). In this way, they contribute to the activation of NF-κB through both the canonical and non-canonical pathway.
Physiologically, LTα and LTβ are expressed on activated lymphocytes and NK T-cell types, especially in response to viral hepatitis. Recently, it was shown that these receptors can be induced in hepatocytes and promote the development of HCC in viral hepatitis or when overexpressed in mice.10 The mechanisms involved in hepatocarcinogenesis from increased signaling through LTβR are related to chronic inflammation, activation of NF-κB and proliferation of hepatic progenitor (oval) cells, independent of the TNF receptor.10 Thus, in the context of viral hepatitis, LT signaling is considered to be a crucial contributor to oncogenic transformation. TNF-like weak inducer of apoptosis (TWEAK) is another member of the TNF superfamily. TWEAK is constitutively expressed in HCC, but not in non-transformed hepatocytes. The role of TWEAK in hepatocarcinogenesis is still controversial. However, proliferation, activation of NF-κB and tumor-related angiogenesis have been linked to both autocrine and paracrine signaling in HCC.11
The TRAIL receptor family has received special attention in the context of hepatocarcinogenesis. Early reports found a selective sensitivity of transformed cells towards the cytotoxic effects of TRAIL while non-transformed hepatocytes appear to be resistant towards TRAIL-induced apoptosis.12,13 However, the initial enthusiasm about a selective and potent anti-tumor compound was lost when TRAIL was found to be cytotoxic to non-transformed human hepatocytes.14,15 As a consequence, further studies have predominantly focused on selectively increasing the sensitivity and overcoming the resistance of transformed hepatocytes towards TRAIL-induced cytotoxicity. This has been partly achieved by histone deacetylase inhibitors,16,17 proteasome inhibitors,18 sorafenib,19 or inhibition of the c-Jun N-terminal kinase (JNK) signaling pathways.20 Thus, the anti-tumoral activity of TRAIL could be useful in the treatment of HCC if sensitization can be achieved selectively in transformed cells and tumor-directed delivery is available.
In summary, the role of members of the TNF-R superfamily in hepatocarcinogenesis is heterogeneous and influenced by the levels of expression and degree of NF-κB activation in response to receptor-ligand interaction. While CD95 and TRAIL are predominantly cytotoxic, TNF-R, LTβR, and TWEAK potentially promote cellular proliferation involving activation of the transcription factor NF-κB.
Death receptor adapter molecules—a link to malignancy?
The apoptosis signal in hepatocytes is transmitted through complex interaction of intracellular proteins following binding of the ligand to the corresponding receptor (Fig. 2).21 Upon activation of a cell death receptor member of the TNF receptor superfamily, the early signaling events are similar. In CD95- and TNF-mediated apoptosis, the Fas-associated death domain (FADD/MORT1) is recruited through protein-protein interactions of corresponding death effector domains (DED). Subsequently an early intracellular signaling complex forms and dissociates from the receptor to mediate activation of caspase 8. This complex has been termed the death-inducing signaling complex (DISC), and the mode of activation is referred to as the “induced proximity” model of activation.22 The central role of the adapter molecules has been demonstrated in mice expressing dominant negative (dn) FADD protein. Following activation of CD95 and TNF signaling pathways, these mice are protected from liver injury.23 Beyond the loss of cell surface receptors in HCC, transformed hepatocytes appear to downregulate proteins that are required to form this signaling complex.24 For example, in HBV-associated HCC decreased levels of FADD can be observed in vitro and in vivo.25 Additionally, disruption of FADD, using a dominant-negative protein, inhibited chemotherapy-associated apoptosis in HCC cells and prevented the release of proapoptotic factors from mitochondria. Thus, impaired function of FADD in hepatocytes inhibits activation of both the extrinsic and intrinsic signaling pathways of apoptosis, while defects in FADD signaling potentially contribute to the chemo-resistant phenotype of HCC.26 In the context of liver regeneration, FADD is required in non-transformed hepatocytes. Thus, mice that expressed a dominant-negative FADD protein exhibited severe impairment in hepatocyte proliferation in response to hepatectomy.27
The cellular FLICE inhibitory protein (cFLIP) is another prominent member of adapter molecules that regulates caspase 8 activity in the context of cell death and hepatocyte regeneration.28 cFLIP is a caspase 8 homologue that lacks the catalytic active domain required for procaspase 8 processing and, when recruited to the DISC, blocks its activation.29 To the present time, a number of different spicing variants and three isoforms of cFLIP have been described: cFLIPlong (cFLIPL), cFLIPshort (cFLIPs) and cFLIPRaji (cFLIPr). However, the roles of the different isoforms in cell death and hepatocarcinogenesis have not yet been fully resolved. Deletion of cFLIP in mice results in embryonic lethality at day 10.5 from impaired development of the heart and vascular endothelium, a phenotype similar to FADD- or caspase 8-defective mice. This stresses the importance of these regulators of apoptosis signaling during organogenesis and tissue homeostasis.30,31 cFLIPs was shown to act as a dominant-negative and anti-apoptotic regulator in the context of caspase 8 activation. Downregulation of cFLIP in response to activation of the TNF-R1, via its proteasomal degradation, results in apoptosis involving JNK and ubiquitin ligase E3.32
The role of cFLIPL is more complex. Although it can prevent apoptosis in some cases, cFLIPL allows some degree of autoprocessing and activation of caspase 8 from cFLIPL : caspase 8 heterodimers. This explains the in vitro findings when overexpression of cFLIPL increased caspase-dependent apoptosis.33 In HCC, high levels of cFLIP are frequently observed; such expression negatively influences recurrence-free survival in patients who have undergone HCC resection.34 The role of cFLIP in resistance towards apoptosis is not restricted to liver cancer, but has also been observed in breast, ovarian, prostate, and colorectal cancer.35–38
Beyond the inhibition of caspase activation, a cleavage product of cFLIPL, p43cFLIP, was shown to activate NF-κB and Erk signaling. This process involves the kinases, RIP and Raf-1 (see below), and enhances canonical Wnt signaling through reduced ubiquitination of beta-catenin.39–41 This explains a potential role of cFLIP in cellular regeneration, and the observation that activation of CD95 promotes regeneration following partial hepatectomy.42 Thus, in addition to direct regulation of death receptor signaling, cFLIPL can exert oncogenic effects via activation of NF-κB and growth factor signaling pathways. Overall, intracellular cell death signaling adapter molecules critically regulate the sensitivity of transformed cells to apoptosis. Beyond the transmission of cell death signals, these molecules also contribute to proliferation and activation of NF-κB. Further studies are needed to address the tissue-specific role of these proteins and their isoforms in cell apoptosis and carcinogenesis.
NF-κB signaling in hepatocarcinogenesis—a flame to malignancy?
Inflammation occurs with a variety of chronic liver diseases and is thought to contribute to hepatocarcinogenesis at multiple levels. In the liver, infiltrating immune cells and residential macrophages are capable of releasing proinflammatory cytokines that act on hepatocytes. The transcriptions factor NF-κB is critically involved in both inflammation and regeneration of hepatocytes and has been shown to be activated in HCC.43 Its role on hepatocarcinogenesis remains controversial, however, with interpretations depending on the model studied. NF-κB is a dimeric transcription factor. Depending on the mode of activation, the subunits p50 (NF-κB1), p52 (NFκB2), p65 (RelA), c-Rel, or RelB can bind each other. Inactive NF-κB is located in the cytoplasm in a complex with inhibitory IκB proteins; these mask the NF-κB nuclear localization signal. The IκB-kinase (IKK) complex phosphorylates IκB proteins allowing their ubiquitination and targeting for proteasomal degradation; this liberates unbound NF-κB dimers to be taken up by the nucleus where they exert their transcriptional activity. Activation of the IKK complex in response to activation of the TNF receptor promotes NF-κB signaling and transcriptional activity in this way (Fig. 2). This pathway is referred to as canonical activation during which a p50 : p65 dimmer promotes transcription of potent antiapoptotic proteins, e.g. Bcl-XL.44 The subunits of the IKK complex are required to activate NF-κB transcription, and loss of IKK isoforms promotes the development of HCC. However, the relative importance of the IKK isoforms differs. Deletion of NEMO (IKKγ) in hepatocytes resulted in the most pronounced phenotype with spontaneous development of HCC, hepatic inflammation and increased lipid deposition. Liver cancer appears to result from increased amounts of oxidative stress, hepatocytes apoptosis, and compensatory regeneration.45 In contrast, mice that lack the subunit IKKβ (IKK2) only develop HCC in the presence of a mitogenic stimulus, presumably due to compensatory activation of IKKα (IKK1).46
Bcl-3 (B cell leukemia-3) is a nuclear member of the IκB family which was initially identified in chronic lymphatic leukemia and has been implicated in cell proliferation and inhibition of apoptosis.47 In contrast to the cytoplasmatic IκB family members, Bcl-3 can act as both an inhibitor or co-activator of the NF-κB subunits.48In vivo, Bcl-3 was shown to regulate the differentiation and development of lymphoid organ tissues and to contribute to the differentiation of Th1 and Th2 lymphocytes. Loss of Bcl-3 increases susceptibility to infectious pathogens, resembling a phenotype observed in mice lacking p52.49 Induction of Bcl-3 in response to proinflammatory cytokines and inhibits NF-κB binding activity in macropahges.50 In an HCC cell line, Bcl-3 was found to be located in the cytoplasm in addition to its classical nuclear location; it contributed to an autoregulatory loop controlling the subcellular location and activity of p50.51
Little is known about the function of Bcl-3 in hepatocarcinogenesis. In HCC, overexpression of Bcl-3 has been observed frequently in addition to nuclear expression of the NF-κB subunits, p52 and p50.52 In an HCC cell line, hepatitis Bx (HBx) antigen mediated upregulation of cyclin D1 through increased transcriptional activity of a p52 : Bcl-3 heterodimer, thus accentuating the oncogenic characteristic of these cancer cell lines.53 Overexpression of Bcl-3 in and increased transcription of anti-apoptotic factors has also been observed in breast and colorectal cancer.54,55 In addition to the induction of potent antiapoptotic genes, Bcl-3 has been shown to suppress the tumor suppressor gene, p53.56 Another interesting link to the oncogenic potential of Bcl-3 was shown in a study of insulin signaling pathways. Interaction of nuclear insulin receptor substrate (IRS)-3 and Bcl-3 augmented transcriptional activity of p50 to the NF-κB DNA binding site, as well as TNF-induced transcriptional activity of NF-κB.57 These findings are interesting in light of the clinical observation that the relative risk of patients with diabetes and obesity for developing HCC is dramatically increased. Thus, further research on the potential link between insulin resistance and NF-κB/Bcl-3 signaling in hepatocytes is warranted.58
CYLD is a de-ubiquitinase and tumor suppressor gene that regulates NF-κB activity through inhibition of the IKK complex, thereby promoting retention of NF-κB in the cytoplasm. The CYLD protein contains binding sites for TRAF2 and NEMO (IKKγ). Inhibition of CYLD expression increased the activity of NF-κB signaling.59 Additionally, CYLD has been implied in regulation of the subcellular location of Bcl-3 involving deubiquitination and retention of Bcl-3 in the cytoplasm. Deletion of CYLD mice sensitizes them towards the development of skin cancer.60 In line with these observations, decreased levels of CYLD have been observed in HCC.61,62 Overexpression of full length CYLD augmented TRAIL-mediated cytotoxicity in HCC cells by negatively regulating NF-κB activity via a process which involved deubiquitination of TRAF2 and interaction with NEMO.63 Genetically engineered mice exhibiting a deletion of exon 7 and 8 of the CYLD gene were shown to retain deubiquitinating capacity in the absence of TRAF2 and NEMO binding sites. This splicing variant of CYLD exerted a strong anti-apoptotic effect on B-cells through increased expression of Bcl-2 caused by activation of NF-κB.64 The central role of CYLD in the regulation of cellular survival and proliferation could make this deubiquitinase an important target to augment anti-oncogenic therapies in HCC.
Mitochondrial amplification of apoptosis—turning up the heat!
The intrinsic pathway of apoptosis involves mitochondria and caspase 9 activation through the apoptosome (Fig. 2). Cleavage of the pro-apoptotic Bcl-2 family member, Bid, by caspase 8 results in truncated Bid (tBid) which triggers oligomerization of Bax and Bak.65 These molecules then insert into the mitochondrial membrane, resulting in mitochondrial permeability transition (MPT) and release of mitochondrial proteins including cytochrome c, Smac/DIABLO, and apoptosis-inducing factor (AIF).66 Smac/DIABLO proteins inactivate the anti-apoptotic proteins, among them X-linked IAP (XIAP), which is a direct XIAP caspase inhibitor.
In hepatocytes, TNF-mediated apoptosis depends on the function of Bid and Bid-deficient hepatocytes exhibit increased resistance to TNF- and CD95-induced cell death, as well as toxic liver injury in parallel to decreased mitochondrial depolarization and cytochrome c release.67,68 This dependency of hepatocytes on the mitochondrial signaling pathway is due to XIAP. During inhibition of XIAP in hepatocytes, apoptosis commenced independently of Bid, a phenotype similar to the apoptotic process in lymphocytes, so-called type 1 cells.69,70 Concordantly, increased expression levels of XIAP have been shown to correlate with a poor prognosis in patients with HCC.71 Following the release of cytochrome c into the cytosol, the apoptosome, which contains apoptosis protease activating factor-1 (APAF-1) and procaspase 9, assembles and caspase 9 becomes activated. In turn, this activates caspase-3 to cause degradation of structural proteins, a key event in apoptosis.72 In addition to XIAP, Bcl-xL and Mcl-1 have been identified as major antiapoptotic Bcl-2 proteins in the liver. Further, overexpression of Bcl-2 or Bcl-XL in hepatocytes ameliorated TNF-induced liver injury.73,74
Mcl-1 is an antiapoptotic member of the Bcl-2 family which contributes to tissue homeostasis in hepatocytes.75 In HCC and colorectal cancer, increased amounts of Mcl-1 contribute to the malignant phenotype with increased activation of proliferative signaling pathways, and resistance towards apoptosis and chemotherapeuticals.76,77 In non-malignant tissue, induction of Mcl-1 can protect hepatocytes from CD95-induced apoptosis,78 while deletion of Mcl-1 in hepatocytes or HCC cell lines sensitizes them towards CD95-induced apoptosis.75,79 Additionally, increased cell injury and turnover promoted the development of HCC in mice.80 Thus, increased amounts of apoptosis in the context of the loss of the anti-apoptotic MCl-1 protein promoted the development of HCC by increasing regeneration and presumably activating progenitor cells. In contrast to the observation in mice exhibiting liver-specific deletion of NEMO45, Mcl-1 induced hepatocarcinogenesis occurs in the absence of significant inflammation. These observations stress the importance of increased liver cell apoptosis in the development of HCC, which was observed similarly in both mouse models.
Summary—dying too much or too little?
The role of apoptosis in hepatocarcinogenesis is dependent on the hepatic microenvironment. Decreased sensitivity towards CD95 signaling pathways contributes to the malignant phenotype including chemoresistance and immune evasion. Inhibition of the apoptosis signal in hepatocytes through decreased expression of adapter molecules that are involved in the formation of the DISC or increased expression of anti-apoptotic factors that block activation of caspases constitutes another commonly encountered mechanism by which pathogens or transformed cells avoid cell death. Other members of the TNF-receptor superfamily have been shown to contribute to inflammation during chronic liver disease and thus promote hepatocarcinogenesis.
The transcription factor NF-κB is of critical importance in regulating inflammation and cell death in hepatocytes. Failure to activate NF-κB transcription in mice with mutations of the IKK complex promotes inflammation and HCC. Factors that modulate NF-κB transcriptional activity are the oncogenic Bcl-3 protein and the tumor suppressor and deubiquitinase, CYLD. Failure to activate NF-κB and the resulting oncogenic potential is closely related to increased cell turnover from inflammation, oxidative stress, and increased apoptosis. In contrast, loss of the antiapoptotic factor, Mcl-1, results in increased cell turnover and hepatocarcinogenesis even in the absence of hepatic inflammation.
In summary, induction of apoptosis constitutes a mechanism by which a cell protects itself against transformation, and blockade of the apoptotic machinery represents a potential mechanism for a cell to survive neoplastic transformation. However, in spontaneous tumor formation increased apoptosis can lead to hepatocarcinogenesis with or without inflammation. To translate these findings to the complex situation in a patient with HCC, an individual evaluation of the hepatic microenvironment and causative agents will be critical. Advances in tumor-directed and selective cytotoxic therapies will have to adapt these findings to benefit our patients with HCC.