The liver is an organ of immense complexity which carefully guards its secrets and continues to reveal and conceal its intricacies. This large organ regulates intermediary metabolism, detoxifies endobiotics and xenobiotics, manufactures critical circulating proteins, has a pivotal excretory function (i.e., bile formation), and can be viewed as a key participant of the innate immune system. In vertebrates, the liver is essential for survival because no other organ can compensate for this multiplicity of functions. A key liver cell participating in each of these processes is the hepatocyte. Disturbances in any of the aforementioned intrinsic hepatic functions and/or exuberant activation of the innate immune system stresses the organ, endangering hepatocyte survival. Indeed, biomarkers for hepatocyte injury—serum aminotransferases—are present in all humans. What constitutes elevation of these biomarkers separating an injured from a “healthy” liver is uncertain and is subject of intense medical interest.1 We posit here that the mere presence of aminotransferases in the circulation implies basal hepatocyte stress and turnover, and perhaps some level of stress and hepatocyte death is simply to be expected given the burden of the aforementioned hepatocyte functions.
Over the last several decades, science has provided considerable insight into how cells sense and succumb to stress. We now realize that cell death commonly occurs by apoptosis, a process characterized by activation of intracellular proteases termed caspases.2 Once these zymogens are activated, they demolish the cell resulting in nuclear fragmentation, cell shrinkage, and scission of the cell into membrane-defined remnants, termed apoptotic bodies. Indeed, human cell apoptosis has long been recognized in liver biopsies,3, 4 and apoptotic bodies were initially referred to by hepatopathologists as Councilman bodies.
Apoptosis occurs by an extrinsic death receptor pathway or by an intrinsic intracellular pathway culminating in mitochondrial dysfunction. In the hepatocyte, these pathways converge in that the death receptor pathway also requires mitochondrial dysfunction for efficient apoptosis.5 The mitochondrial pathway of cell death is controlled by interactions between members of the Bcl-2 family of proteins.6 These proteins can be categorized into subsets. The guardians or antiapoptotic members of this family include Bcl-2, Bfl-1/A1, Mcl-1, Bcl-xL and Bcl-w. The multidomain executioners or proapoptotic members of this family include Bcl-2-associated X protein (Bax) and Bak.6 Bax and Bak are essential for cell death, directly induce mitochondrial dysfunction and are often redundant in that expression of one compensates for the lack of the other. Although the multidomain proteins share four Bcl-2 homology domains, the messengers or biosensors of cell death display only the third Bcl-2 homology domain. These proteins, referred to as BH3-only proteins, include BH3-interacting domain death agonist (Bid), Bim, Bmf, p53 up-regulated modulator of apoptosis (Puma), Noxa, Hrk, Bad, and Bik. The mechanisms by which these messengers induce cell death remain controversial; however, Bid, Bim, and Puma are thought to directly activate the executioners Bax and/or Bak,7–9 whereas the others may antagonize the ability of the antiapoptotic proteins to prevent Bax and Bak activation.10 Thus, the antiapoptotic proteins of this family are essential to prevent apoptosis upon activation of BH3-only proteins following a death stimulus.
What Do We Know About these Antiapoptotic Proteins and the Hepatocyte?
Interestingly, the founding member of this family, Bcl-2, is not expressed in murine hepatocytes at the protein level,11 whereas Bcl-w and Bfl-1/A1 knockout mice have no liver phenotype.12–14 However, both the potent antiapoptotic proteins Bcl-xL and Mcl-1 are expressed by hepatocytes. These cytoprotective proteins appear to have nonredundant functions. For example, murine hepatocyte-specific genetic deletion of either Bcl-xL or Mcl-1 results in spontaneous hepatocyte apoptosis, elevated serum aminotransferase values, and an enhanced profibrogenic response of the liver to injury.15, 16
To further address the question of nonredundant and/or cooperative cytoprotective functions of these two Bcl-2 proteins in the hepatocyte, Hikita et al., in this issue of HEPATOLOGY,17 generated mice with combined hepatocyte deletion of Bcl-xL plus Mcl-1. The resultant double knockout mice had impaired liver development characterized by reduced liver volume and died in the early neonatal period due to severe hepatic failure. The severe liver failure was well-characterized with elevated serum bilirubin and ammonia levels. Apparently, mice survived in utero due to compensatory processes by the placenta and/or until additional proapoptotic stimuli developed after birth. A gene dose effect was documented in this study by using heterozygotes for Mcl-1 and/or Bcl-xL.17 Hepatic deletion of a single allele for bcl-x or mcl-1 gene, which is associated with a marked reduction of Bcl-xL or Mcl-1 protein levels, respectively, does not induce hepatocyte apoptosis under physiological conditions. However, mice lacking a single allele for each bcl-x and mcl-1 genes, with reduced protein levels of both Bcl-xL and Mcl-1, develop a spontaneous apoptotic liver phenotype similar to that observed in mice with hepatocyte-specific deletion of Bcl-xL or Mcl-1. Again, this latter observation is consistent with nonredundant function of these two proteins, as well as a chronic, low-level hepatocyte death stimulus.
Bad, Bcl-2 agonist of cell death; Bak, Bcl-2 antagonistic killer; Bax, Bcl-2 associated X protein; Bcl-2; B-cell lymphoma-2; Bcl-xL, Bcl-extra long; Bid, Bcl-2 interacting death domain agonist; Bik, Bcl-2 interacting killer; Bim, Bcl-2 interacting mediator of death; Bmf, Bcl-2 modifying factor; Hrk, hara-kiri; Mcl-1, myeloid cell leukemia factor-1; Noxa, named for damaged; Puma, p53 up-regulated modulator of apoptosis; TNF-R1, tumor necrosis factor–receptor 1; TRAIL-R1, TNF-related apoptosis inducing ligand receptor 1.
What Is the Mechanism Responsible for the “Physiologic” Proapoptotic Stress Afflicting the Hepatocyte?
The article by Hikita et al.17 indicates that the murine liver is continuously subjected to proapoptotic stress under physiological conditions, which is antagonized by the dynamic duo of the antiapoptotic proteins Mcl-1 and Bcl-xL. What is responsible for this stress? Hikita et al., also partially addressed this question by generating mice deficient for both the BH3-only protein Bid and Mcl-1. Bid deletion protected the Mcl-1 deficient liver from injury. Thus, Bid, in part, mediates “spontaneous or physiologic” proapoptotic stress in hepatocyte. Bid is thought to transduce apoptotic death in response to death receptor stimulation after its cleavage by caspase 8 into a truncated form of Bid (or tBid; Fig. 1A).18 Thus, the data implicate death receptors in the proapoptotic stress of the hepatocytes.
What Death Receptor Is Responsible for the Generation of tBid?
The current data do not completely answer this question, but of the four death receptors expressed by the hepatocyte—Fas, tumor necrosis factor–receptor 1 (TNF-R1), and TNF-related apoptosis inducing ligand receptors 1 and 2 (TRAIL-R1/TRAIL-R2)19—a rational candidate is Fas. For example, genetic deletion of Bid has previously been shown to protect the murine liver from Fas-mediated acute liver failure,20 but not against acute liver failure mediated by TNF-α plus galactosamine.21 Furthermore, TNF-α, which preferentially activates the cytoprotective transcription factor nuclear factor-κB rather than pro–cell death signals under physiological conditions,22 is unlikely to contribute to the hepatocyte apoptosis observed by Hikita and coworkers. The other death ligand, TRAIL, also does not induce apoptosis in healthy hepatocytes.23 Consistent with Fas-mediated Bid activation as a mechanism for apoptosis in hepatocytes deficient in Mcl-1 and perhaps Bcl-xL, overexpression of Bcl-xL or Mcl-1 rescues mice from acute liver failure by Fas,24, 25 and single-allele deletion of Mcl-1 or antisense targeted knockdown of Bcl-xL renders hepatocytes more susceptible to Fas-mediated hepatocellular damage.17, 26 Interestingly, another potent antiapoptotic BH3-only protein, Bim, also contributes to acute murine liver injury by Fas.21, 27 Because Fas is abundantly expressed in the liver, hepatocytes may be continuously exposed to constant, albeit low, levels of Fas-induced Bid and Bim activation. Sufficient Mcl-1 and Bcl-xL are then necessary to block the deleterious effects of Bid and Bim (Fig. 1B). Additional studies will be necessary to address this interpretation of the existing data, such as crossing the Bcl-xL and Mcl-1 conditional double knockout animals with Fas, Bim, Bid, and caspase 8 knockout mice.
Why Are Both Bcl-xL and Mcl-1 Required to Protect the Hepatocyte?
Current models of apoptosis signaling suggest that antiapoptotic proteins bind to and sequester BH3-only proteins or inhibit activation of Bax and Bak (Fig. 1B).6 Because the antiapoptotic proteins are structurally similar in sharing the four BH domains and all display a shallow groove capable of binding the BH3-only proteins, they have been assumed to exert similar functions. In addition, both Mcl-1 and Bcl-xL also bind Bak and inhibit Bax activation.28 However, the unexpected pivotal finding of this study is the nonredundant roles exerted by Bcl-xL and Mcl-1 in the hepatocyte.
Previous studies suggest that Bim and Bid bind to both Bcl-xL and Mcl-1.29, 30 Therefore, it is first tempting to speculate that, under physiological conditions, Bcl-xL and Mcl-1 are inundated with proapoptotic pressure from Bim and t-Bid. Both Bcl-xL and Mcl-1 equally bind the BH3-only proteins but the hepatocyte requires abundant expression of both proteins to protect itself from cellular demise (Fig. 2A). In this stoichiometric model, loss of one or the other antiapoptotic protein would diminish the capacity of the hepatocyte to protect itself from the BH3-only protein stress resulting in apoptosis. Loss of an adequate threshold level of protective Bcl-2 proteins could be further aggravated by the sensitizer BH3-only family members Bad and Noxa; Bad selectively counteracts the function of Bcl-xL, whereas Noxa exclusively counteracts Mcl-1,29 as a result releasing t-Bid and Bim from their sequestration by Bcl-xL and Mcl-1. Such a model has been identified in selected cancer cell lines, and these cells have been considered “primed for death”.31 This model predicts that Bcl-xL or Mcl-1 complexed to BH3-only proteins would be readily identified in normal hepatocytes—a testable hypothesis. However, we do not favor this model because any additional BH3-only protein stress should be hepatotoxic. The BH3 mimetic ABT-737, in development for cancer therapy, has not been reported to be hepatotoxic when administered exogenously to mice.32 In contrast, it does exert single-agent cytotoxicity in malignant cell lines primed for death.33
Our preferred model to explain the nonredundant function of Bcl-xL and Mcl-1 during physiologic hepatocyte apoptosis implies that these two protective proteins bind to and sequester different BH3-only proteins (Fig. 2B). Although previous studies have demonstrated binding of Bim and Bid to both Bcl-xL and Mcl-1,29 the experimental settings did not recapitulate the lipid milieu of the outer mitochondrial membrane, the precise conformation of the intact BH3-only protein, or the posttranslational modifications of the Bcl-2 proteins. Further, they employed Bcl-xL and Mcl-1 which lacked their transmembrane domains. Consistent with this criticism, several posttranslational modifications of Mcl-1 or Bcl-xL have been reported to render them more selective to bind to and sequester one BH3-only protein member over another or to disable their antiapoptotic function. For example, Mcl-1 can be phosphorylated at serine-64, a modification near its transmembrane domain which potentiates its binding to the BH3-only proteins Bim and Noxa.34 Also, deamination of Bcl-xL at asparagine-52 and asparagine-66 disrupts its antiapoptotic activity.35 Finally, it is also possible that posttranslational modifications alter the ability of Bcl-xL and/or Mcl-1 to prevent Bax or Bak activation. For example, c-Jun N-terminal kinase–dependent phosphorylation of Bcl-xL at serine-62 disables the ability of the protein to bind to Bax.36 This proposed model argues that Mcl-1–deficient mice that overexpress Bcl-xL, and vice versa, would still experience elevated serum aminotransferase levels and apoptosis. These concepts can be tested by studies examining posttranslational modifications of these antiapoptotic proteins in hepatocytes.
In summary, the current article by Hikita et al.17 suggests that the liver's dance with death receptors defines a precarious relationship defined by constant proapoptotic stress. The hepatocyte appears to have adapted by expressing two antiapoptotic proteins, each assigned a specific survival role. Without both guardians at the gate, the liver is overwhelmed by death cues and will fall into the abyss of organ failure. More information is needed to answer the questions raised above. Perhaps as this information emerges, it can be used to develop hepatoprotective strategies for the treatment of human liver disease.