Tumor necrosis factor–related apoptosis-inducing ligand (TRAIL; also known as APO-2L) was cloned in 1995 through a systematic search of a library of expressed sequence tags to identify proteins similar to tumor necrosis factor-α (TNF-α) and was also found to have a close sequence homology to the Fas ligand.1 Early studies demonstrated the ability of TRAIL to induce apoptosis or programmed cell death of cancer cell lines in vitro and to reduce cancer xenograft burden in vivo.2 Data emerged on the differential sensitivity of transformed cells to TRAIL-induced cell death, whereas nontransformed cells were resistant to its apoptosis-inducing effects. Unlike the other death ligands, Fas ligand (which induces fulminant hepatic failure) and TNF-α (which induces systemic inflammatory response syndrome), the systemic administration of TRAIL was found to be safe in mice and nonhuman primates.3 In the era of rational biological cancer therapy, this brought the early promise of TRAIL-induced selective cancer cell death without bystander damage to normal tissues.
Five TRAIL receptors have been identified: TRAIL-R1 (DR4 [death receptor 4]), TRAIL-R2 (DR5 [death receptor 5]/TRICK-2/KILLER), TRAIL-R3 (DcR1 [decoy receptor 1]/TRID/LIT), TRAIL-R4 (DcR2 [decoy receptor 2]/TRUNDD), and osteoprotegerin.4–7 Of these receptors, only DR4 and DR5 transmit proapoptotic signals and induce cytotoxicity. DcR1 and DcR2 also have extracellular domains that bind TRAIL but do not activate intracellular death signals. Osteoprotegerin has a bone modulatory function and several ligands, including TRAIL, and may not be a pathophysiologically relevant TRAIL receptor. On the basis of this receptor heterogeneity and promiscuity, TRAIL toxicity toward tumor cells was thought to be restricted by death receptor expression. However, TRAIL and its receptors are widely expressed in healthy tissues, and resistance to TRAIL-induced apoptosis in normal tissues is likely an intracellular phenomenon.
Cellular FLICE-like inhibitory protein (cFLIP) is one such protein that checks TRAIL signaling proximally, at the level of the receptor complex.8 cFLIP shares structural homology with caspase 8, an apical effector protease in the cytotoxic signaling cascade, and is recruited to the receptor complex, preventing the activation of caspase 8. The inhibition of caspase 8 thus inhibits downstream apoptotic signaling. Additional intracellular control mechanisms consist of the Bcl-2 family of proteins.8 Antiapoptotic members of the Bcl-2 family (Bcl-2, Bcl-XL, Mcl-1, A-1, Bcl-w, and Boo) and proapoptotic members (Bax, Bak, Bok, Bcl-XS, Bim, Bik, Bad, Bid, Bmf, Hrk, Puma, and Noxa) act as controllers of the cellular fate via complex interactions that remain to be fully defined. In addition to the activation of early caspases, nuclear factor κ-B and c-jun N-terminal kinase (JNK) are also activated by TRAIL. Nuclear factor κ-B leads to the transcriptional activation of inflammatory, antiapoptotic, and prosurvival proteins in apoptosis-resistant cells. JNK, on the other hand, can be proapoptotic or antiapoptotic, depending on the cellular context and duration of activation.9 It is also interesting that JNK can transcriptionally up-regulate the expression of DR5, conferring increased sensitivity to TRAIL.
Although cancer cells are sensitive to TRAIL, emerging data indicate that other cellular perturbations can also confer TRAIL sensitivity. In normal human liver, TRAIL, DR4, and DR5 messenger RNA is expressed at low levels, and receptor protein expression may be difficult to identify; however, emerging data indicate that their expression is up-regulated in diseased states.7, 10 In animal models of acute hepatitis (Listeria-induced and concanavalin A–induced acute hepatitis), TRAIL expression in liver mononuclear cells alone is sufficient to induce liver injury.11 In experimental acute adenoviral hepatitis, DR5 expression is up-regulated, and subsequent TRAIL administration leads to steatosis and apoptosis.12 In contrast, alcohol intake sensitizes only to the steatotic effect of TRAIL12; no increase in DR5 expression or apoptosis has been observed. In liver specimens from patients with chronic hepatitis C, TRAIL expression is up-regulated independent of the extent of lymphocyte infiltration.13 In a hepatitis B viral infection, soluble TRAIL is up-regulated and correlates with liver injury.14 Bile acids can up-regulate DR5 expression and inhibit cFLIP function, thereby sensitizing hepatocytes to TRAIL-mediated cell death.15 DR5 is also up-regulated in the livers of patients with nonalcoholic steatohepatitis and in tissue culture models of hepatic steatosis.16 Thus, infectious, metabolic, chemical, or genotoxic stress in hepatocytes confers TRAIL sensitivity by DR5 up-regulation that is either JNK-dependent or p53-dependent. Thus, the safety of TRAIL administration to humans with underlying liver disease is unclear and has significant implications for the development of TRAIL as a biological anticancer agent.
In this issue of HEPATOLOGY, Volkmann et al.17 explore the safety of TRAIL in normal, steatotic, and chronic hepatitis C–infected livers. Volkmann et al. have examined the hepatotoxicity of 3 distinct forms of TRAIL in isolated primary human hepatocytes and organotypic liver slices.18 The latter overcomes the rapid dedifferentiation characteristic of isolated primary cells. With respect to isolated hepatocytes, the organotypic model better approximates the in vivo environment, retaining the natural complexity of cellular polarity, interactions, and proportions. Although it is conceivable that circulating cell populations, such as natural killer cells, natural killer T cells, and red blood cells, are mechanically removed from the surface of 5-mm cubes, sinusoidal endothelial cells, Kupffer cells, and stellate cell components are intact. The short-term (24-hour culture) morphologic integrity and utility in pharmacologic studies of the uptake, toxicity, and metabolism of such culture systems is well described. Indeed, the authors demonstrate the viability of the system by demonstrating the absence of basal cell death, using caspase activity and lactate dehydrogenase release. They observe that isolated human hepatocytes develop sensitivity to TRAIL toxicity independent of the form of synthetic TRAIL used. In contrast, organotypic liver cultures from corresponding normal human liver remained resistant to TRAIL. In this regard, Volkmann and coworkers' data support the concept that normal liver is resistant to TRAIL and highlight the misinformation that may originate from the use of primary culture systems. The authors are to be congratulated on exploring and validating the organotypic model!
In contrast to normal human liver, Volkmann and coworkers17 demonstrate that steatotic and HCV-infected livers are indeed sensitive to TRAIL cytotoxicity. The mechanism is due to the up-regulation of DR5. These results dovetail with recent work by Malhi et al.,16 in which Huh7 and HepG2 cells and primary rodent hepatocytes rendered steatotic by oleic acid also up-regulated DR5 and became sensitive to TRAIL killing.16 These data suggest that patients entering TRAIL-based anticancer clinical trials should be screened for steatosis and chronic viral hepatitis. In all likelihood, these patients are at increased risk for potential TRAIL hepatotoxicity and should not receive this agent.
Tumor cells can be primarily resistant to TRAIL or develop TRAIL resistance during oncogenic transformation through, for example, the up-regulation of cFLIP or Bcl-2 family proteins.8 These cells can be sensitized to TRAIL by combination therapy with traditional chemotherapeutic agents, radiation treatment, or newer chemotherapeutic agents, such as proteasomal inhibitors or histone deacetylase (HDAC) inhibitors. Depsipeptide, an HDAC inhibitor, leads to the reversal of oncogenic epigenetic changes, resulting in the derepression of tumor-suppressor genes and apoptotic genes. It is known to activate the intrinsic or death receptor–independent pathway of apoptosis by increasing cellular Bid and Bim levels (proapoptotic Bcl-2 proteins) and also to enhance TRAIL sensitivity by increasing DR5 expression. Both TRAIL agonistic antibodies and HDAC inhibitors are in clinical trials, and understanding their synergistic toxicity is key to their combined clinical utilization. The data of Volkmann et al.17 are provocative in that normal liver tissue, presumably with normal levels of histone deacetylation, demonstrates depsipeptide-induced gene transcription that favors apoptosis. In this regard, depsipeptide, similar to metabolic and inflammatory stress, sensitizes the liver to TRAIL toxicity. Furthermore, gene expression analysis shows that similar changes are induced by depsipeptide, metabolic stress, and inflammatory stress. DR5 expression is enhanced, proapoptotic Bax and Puma expression is enhanced, and antiapoptotic cFLIP and Bcl-2 expression is suppressed.
In summary, TRAIL is emerging as a key mediator of hepatic injury during inflammatory disorders of the liver. It is also clear that TRAIL is not toxic to normal human liver on systemic administration. On the basis of these observations, one option would be to restrict the use of TRAIL in cancer chemotherapy to individuals with normal livers. This will likely exclude many, especially patients with underlying hepatic steatosis (a very common clinical scenario). TRAIL, in itself, is no magic bullet for cancer, and its use in combination with other agents is necessary to induce tumor cell death. The potential hepatotoxicity of these combinations must be considered before we engage in large-scale studies. Another intriguing possibility, based on DR5-dependent TRAIL signaling in liver injury, is the development of selective DR4 agonistic antibodies for use in cancer therapy. Perhaps this approach would maximize cancer efficacy and negate potential hepatotoxicity. In this regard, Volkmann et al.17 have presented a reliable in vitro assay that closely recapitulates the in vivo condition for the evaluation of the hepatic safety of combination chemotherapy. 1