Death receptors, a subset of the tumor necrosis factor (TNF) receptor (TNFR) superfamily that includes TNFR1, CD95 (Fas, Apo-1), and the TRAIL (TNF-related apoptosis-inducing ligand) receptors, transduce signals capable of engaging apoptosis or necrosis,1 depending on the status of signaling molecules in the cells. Ligation of one such receptor, cluster of differentiation 95 (CD95), has catastrophic consequences in vivo, because this results in lethal, fulminant liver destruction.2 In this issue of HEPATOLOGY, Hikita et al.3 employ a conditional gene deletion model to explore the molecular mechanisms of this liver failure.
Upon ligation, CD95 rapidly recruits an intracellular adapter molecule, Fas-associated protein with death domain (FADD), which in turn binds to and activates the initiator caspase, caspase-8.1 The activation of caspase-8 requires two steps: dimerization of the inactive “pro-form” of the molecule, followed by autocleavage, which generates a stable, active protease.4 (Note that cleavage alone does not activate caspase-8, despite claims of “feedback activation” via cleavage by other enzymes5; there is no biochemical support for such feedback.)
Upon activation of caspase-8, apoptosis can proceed in two ways (Fig. 1). The first route is via caspase-8–mediated cleavage and activation of the executioner caspases, caspase-3 and caspase-7 (unlike caspase-8, executioner caspases exist as preformed dimers, and cleavage alone is sufficient for their activation). This pathway of caspase-8–mediated cleavage and activation of executioner caspases occurs in so-called “Type I” cells.6 However, an endogenous caspase inhibitor, X-linked inhibitor of apoptosis protein (XIAP), if present in sufficient amounts, can bind to and block the activation of the executioner caspases, preventing apoptosis. This occurs in “Type II” cells.7 In this case, apoptosis proceeds when caspase-8 cleaves a BCL-2 family protein, BID, which in turn activates the proapoptotic effectors of the BCL-2 family, BAX and BAK. The latter permeabilize the mitochondrial outer membrane, releasing proteins (e.g., Smac and Omi) that antagonize XIAP, allowing executioner caspase activation, and thus apoptosis, to proceed. Although mitochondrial outer membrane permeabilization also releases cytochrome c, which triggers the activation of caspase-9, this is not required for apoptosis to proceed in this scenario.7
Mice lacking BID are completely resistant to ligation of CD95 in vivo,7, 8 leading to the idea that hepatocytes (and indeed, any cell contributing to CD95-mediated lethality) are Type II cells. This is further supported by the finding that mice lacking both BID and XIAP are highly sensitive to CD95 ligation, and display liver destruction.7
Now, Hikita et al.3 have examined the effects of removing BAX and BAK from hepatocytes by conditional deletion of BAX in a BAK−/− background. BAX and BAK are generally redundant proteins; the presence of one or the other is sufficient for BID-induced cell death in many cells, but in the absence of both, active BID is harmless.9 Liver mainly expresses BAK, and indeed, the authors observed less CD95-induced hepatocyte apoptosis in mice lacking BAK alone than in wild-type mice. In the absence of both BAX and BAK, they observed even less cell death. Nevertheless, the mice underwent catastrophic liver destruction.
Therefore, we have a paradox. BID is required for CD95-induced liver damage, and BAX/BAK are required for BID function. Yet, removal of BAX and BAK did not protect the animals from CD95-induced lethality. How can this be reconciled?
One possibility is that, in the mixed strain background employed in this study, the levels of XIAP were insufficient to protect the liver, and therefore the death proceeded via a “Type I” pathway as opposed to the Type II pathway typically observed in hepatocytes. The authors note that XIAP levels fell several hours after CD95 ligation in these animals. Previous studies have shown that hepatocytes, cultured ex vivo, convert from Type II to Type I cells,10 lending some credence to this idea. However, BID-deficient mice are fully resistant to CD95 ligation, without showing even a delayed liver failure,7 but in those studies, a pure C57Bl/6 background was employed. Strain differences may therefore account for the divergent results.
There is, however, another explanation. Hikita et al.3 deleted BAX only in hepatocytes of the BAK-deficient mice, using a CRE (cyclization recombination) transgene driven by the albumin promoter. Hepatocytes not expressing this CRE would die in response to CD95-induced BID cleavage, as would any cell that does not drive this promoter. Might other cells, dying in this BID-dependent (Type II) manner, cause hepatic injury? In an earlier study, hepatocyte expression of a BCL-2 transgene driven by the albumin promoter (BCL-2 efficiently blocks BID-induced cell death) reportedly blocked hepatocyte apoptosis, but not liver destruction.11 This is completely consistent with the findings of Hikita et al.3
Previously, it was noted that CD95 ligation in vivo induces destruction of vascular endothelium in the liver.12-14 This produces the sinusoidal hemorrhage characteristic of this treatment. As a result, CD95 ligation would be lethal even if hepatocytes were protected. Therefore, although deletion of BID throughout the animal protects hepatocytes, endothelial cells, and the animal as a whole, deletion of BAX and BAK (or expression of BCL-2) specifically in hepatocytes does not. It is an attractive resolution to the apparent paradox.
Hikita et al.,3 however, noted some apoptosis in hepatocytes in their engineered animals upon ligation of CD95. These might be cells that had failed to flox BAX, as mentioned above, or perhaps more intriguingly, may be dying independently of BAX and BAK. The latter possibility is supported by studies showing that metabolic stress (e.g., glucose deprivation) can sensitize cells for CD95-induced death.15 Certainly, a failure of the blood supply, as discussed above, would cause such stress, and it will be of interest to ascertain if this can convert Type II cells to Type I cells.
Finally, one might be enticed to consider the possibility that liver destruction via CD95 ligation may proceed not only by apoptosis but also by necrosis. Several molecular mechanisms whereby necrosis can be “programmed” are known.1 However, Hikita et al.3 showed that cyclophilin D, which is required for some forms of necrosis,16 does not play a role in CD95-induced liver damage in the absence of BAX and BAK. Furthermore, because it has long been known that caspase inhibitors (which preferentially go to the liver in vivo) block CD95 ligation-induced lethality,17 the authors also confirmed that the lethality in their mice was similarly blocked by caspase inhibitors. Tellingly, a recently uncovered pathway of necrosis is antagonized by caspase-8,18, 19 but based on these results, it does not appear to play a role in CD95-mediated liver destruction.
The liver is more than hepatocytes. Understanding how the ligation of death receptors, either in experimental models or disease states, causes liver injury and how this can be abated remains a priority.