Apoptosis (programmed cell death) has been shown to play a major role in development and in the pathogenesis of numerous diseases. A principal mechanism of apoptosis is molecular interaction between surface molecules known as the “death receptors” and their ligands. Perhaps the best-studied death receptor and ligand system is the Fas/Fas ligand (FasL) system, in which FasL, a member of the tumor necrosis factor (TNF) family of death-inducing ligands, signals death through the death receptor Fas, thereby resulting in the apoptotic death of the cell. Numerous cells in the liver and gastrointestinal tract have been shown to express Fas/FasL, and there is a growing body of evidence that the Fas/FasL system plays a major role in the pathogenesis of many liver and gastrointestinal diseases, such as inflammatory bowel disease, graft vs. host disease, and hepatitis. Here we review the Fas/FasL system and the evidence that it is involved in the pathogenesis of liver and gastrointestinal diseases.
kerr, wyllie, and currie published in 1972 (61) what has become recognized as a landmark report, in which they described a controlled, apoptotic cell death that occurs in a wide range of physiological situations and proposed a role in tissue kinetics involving dynamic cell turnover and homeostasis. Remarkable about the cell death they described was not only the diversity of the tissue in which it was observed but also the consistency of the morphological description of the dying cells. Apoptotic cells were observed to undergo shrinkage involving nuclear margination and chromatin condensation as well as a characteristic membrane blebbing followed by engulfment by neighboring cells. Integrity of the plasma membrane was maintained in the dying cell, and cellular contents were not released into the extracellular milieu. Wyllie contributed another major advance in 1980 (143) when he demonstrated that nuclear apoptotic events included activation of an endogenous nuclease that resulted in degradation of chromatin into nucleosome-sized DNA ladders that have become a biochemical hallmark of apoptosis.
One of the major functions of apoptosis is maintaining the appropriate number of cells (i.e., “homeostasis”). Much of our knowledge on this subject is from studies on the nematode Caenorhabditis elegans, which has exactly 1,090 cells of which 131 must die during development to ensure proper maturation (30). The genes that regulate the precise number of cells that C. elegans loses have been shown to have mammalian analogs, and these genes have subsequently been shown to be important in regulation of apoptosis of mammalian cells (50, 83). Similar to the nematode, and unlike most tissues, the size of the liver and the number of intestinal epithelial cells are tightly regulated, and it is reasonable to hypothesize that apoptosis plays a major role in this regulation. The liver has regenerative capability such that after injury or resection the liver normally returns to its original size and function. Similarly, the number of intestinal epithelial cells (IEC) found on the villus and crypt appears to be tightly controlled. IEC develop from stem cells in the crypt where they migrate up to the villus tip in 2–3 days and “slough” off the villus tip (49, 51). With each epithelial cell that sloughs off, a new epithelial cell is generated in the crypt such that the villus height remains constant. Although apoptosis most likely ensures that the liver does not grow too large or that villus height remains constant, it remains to be formally demonstrated that these processes are regulated by apoptosis. Although genetically engineered “knockout” of mammalian analogs of C. elegans genes, which have been shown to regulate apoptosis, resulted in significant kidney, brain, extremity, and craniofacial abnormalities, there have been to date no reported abnormalities of the liver and gastrointestinal tract in these knockout mice (76, 108).
FAS-MEDIATED DEATH SIGNALING: MULTIPLE ROADS TO DEATH
In recent years we have realized that apoptosis is mediated through a series of biochemical steps. Perhaps the best-characterized apoptotic pathway is that induced through the surface molecule Fas (Refs. 6, 93; see Fig. 1). Engagement of Fas by FasL or anti-Fas antibodies results in trimerization of Fas, followed by recruitment of a number of proteins to form a complex known as the death-inducing signaling complex (DISC). After formation of the DISC with Fas, FADD (20), and pro-caspase-8 (12, 91) as the primary components, pro-caspase-8 is cleaved and activated. Active caspase-8 then cleaves bid (77), a bcl-2 family member that causes release of mitochondrial cytochrome c (62, 63, 74). On release from the mitochondria, cytochrome c interacts with APAF-1 (148) and caspase-9 (69) to form a complex known as the apoptosome (103). Formation of the apoptosome results in cleavage and activation of caspase-9, which in turn leads to the cleavage and activation of caspase-3 (97), a central executioner caspase that cleaves and activates downstream caspases that are largely responsible, directly and indirectly, for systematically dismantling the apoptotic cell from within. For a comprehensive review of caspases, see Refs. 135 and 142. Structural breakdown of the cell includes cleavage of gelsolin (65), which results in cleavage of actin filaments, as well as direct cleavage of fodrin (nonerythroid spectrin) (81). Together these two events can be held responsible for the gross morphological changes observed in programmed cell death. Although caspases have not been directly linked with membrane events, the asymmetrical distribution of phosphatidylserine (PS) moieties is lost (80, 82) and the resultant presentation of PS on the external surface of a dying cell facilitates its engulfment (113) by professional or nonprofessional phagocytes, thereby minimizing cellular debris and the possibility of an inflammatory response.
Active caspase-3 is also responsible for cleavage of the inhibitor ICAD [Ref. 111; also identified as DFF45 (75)], which releases the caspase-dependent DNase [CAD (31); also DFF for DNA fragmentation factor (75)], the nuclease that gives rise to the DNA laddering that is a signature of apoptosis. These nuclear events in apoptosis have been the subject of some debate since the nucleus has been shown to be dispensable for many apoptotic deaths in a number of cell types by various stimuli (56, 95, 137), but the fact remains that most cells in a physiological setting have functional nuclei that must be properly disposed with the remainder of the cell and nuclear breakdown and DNA laddering have long been hallmarks of apoptosis.
Although the pathway described above is probably valid, recent evidence suggests that the Fas signaling pathway that leads to cell death is significantly more complex and involves at least two major pathways, termed type I and type II. The molecular decision to signal via one pathway or the other occurs at the level of the DISC. In type I cells, caspase-8 is recruited in sufficient amounts at the level of the DISC to directly activate caspase-3, thereby bypassing the need for mitochondrial cytochrome c release (114), whereas if caspase-8 in the DISC is present in lower levels, the pathway described initially occurs and this pathway is deemed type II. That is, caspase-8 cleaves bid, which then induces the release of mitochondrial cytochromec. Cytochrome c release subsequently induces the formation of the apoptosome, which results in the feed-forward activation of the caspase-protease cascade.
The existence of type I and type II pathways helps explain initially conflicting findings that, in some circumstances, bcl-2 is capable of preventing Fas-mediated death, i.e., in type II cells but not type I cells (114). Furthermore, recent studies in genetically targeted mice support the possibility that the usage of type I and type II pathways is determined by tissue specificity. Hepatocytes from bid−/− mice have been shown to be relatively resistant to Fas-mediated death, yet thymocytes from bid−/− mice were shown in the same study to be sensitive to Fas-mediated cell death (145), suggesting that thymocytes use the type I pathway whereas hepatocytes utilize the type II pathway. In agreement, Yoshida et al. (147) demonstrated that thymocytes lacking APAF-1 (a molecule that is predicted from Fig. 1 to be involved in the type II pathway) are extremely resistant to numerous apoptotic stimuli yet sensitive to Fas-mediated death.
Finally, two other pathways have been implicated in Fas-mediated death. Although the relative importance of these two pathways is unclear and controversial, it should be stressed that more than one pathway for Fas-mediated cell death exists and this is important because, as we have observed with type I and type II cells, the pathway(s) leading to Fas-mediated death in lymphocytes (which is the most commonly described) may not be the same as that in cells found in the liver and gastrointestinal tract. In the first pathway, engagement of Fas has also been shown in some cells to lead to cell death by activation of the JNK pathway via DAXX (144) by a process that does not appear to involve FADD or caspase-8 (20). However, the relevance of Fas-DAXX interactions has been brought into question in light of the phenotype of the DAXX knockout mice (84). In a second pathway, engagement of Fas has been shown to induce rapid increase in ceramide (27, 133), which has been shown to be a potent intracellular mediator of cell death (99). However, if and how ceramide is linked to the presently known Fas-mediated death pathway is unclear. Ceramide levels have been shown to increase under numerous apoptotic stimuli (41, 46), and it is not yet clear whether the increase in ceramide observed after engagement of the Fas receptor is an active participant in the pathway leading to apoptotic cell death or merely a byproduct in dying cells.
It is important to note that not all Fas-bearing cells are susceptible to Fas-mediated killing. One mechanism of protection is by expression of inhibitors such as the inhibitor of apoptosis proteins (IAP) family of protease inhibitors that have been found to block the action of key caspases such as caspase-3, -7, and -9 (28). For example, HepG2 hepatocarcinoma cells are relatively resistant to Fas-mediated killing unless treated in the presence of actinomycin D, during which treatment there is a marked reduction in ILP, a member of the IAP family (128). Another inhibitor of Fas signaling is through FLICE-inhibitory protein (FLIP), which acts by binding to FADD in a nonsignaling complex, thus acting as a dominant negative caspase-8 (FLICE) (54). Elevated levels of c-FLIP have been proposed to impart Fas resistance in resting lymphocytes and may be coordinately downregulated after lymphocyte activation and during S phase of the cell cycle (2, 115). c-FLIP levels are elevated in lymphocytes in G1 phase, which corresponds with reduced sensitivity to death receptor signaling in these cells at this point in the cell cycle. A viral FLIP (v-FLIP) has been identified that is believed to contribute to a viral strategy to prevent cell death of infected cells, thus avoiding immune surveillance (134).
It is also worth mentioning that engagement of Fas on Fas-positive cells does not lead necessarily to apoptosis or only to apoptosis. Using Fas+ HT-29 colon cancer cell lines, Abreu-Martin et al. (1) demonstrated that engagement of Fas by anti-Fas did not result in apoptosis but instead led to interleukin (IL)-8 secretion. Interestingly, the addition of interferon-γ made these cells susceptible to Fas-mediated death, suggesting that Fas-positive cells can be resistant to Fas-mediated cell death under normal conditions but in certain inflammatory conditions can become susceptible. In addition, Gao et al. (40) demonstrated that engagement of Fas on splenocytes led to IL-10 secretion. In this later study, it is unclear whether engagement of Fas leads to apoptosis of splenocytes because resting splenocytes normally express very low levels of Fas and are resistant to Fas-mediated cell death.
Another mechanism to downregulate tissue effects of Fas-mediated killing is the expression of decoy receptors. DcR1 and DcR2 are membrane-bound receptors that compete for death ligand binding by DR4 and DR5 but do not transmit a death signal. In this way, a death ligand is rendered unable to engage the apoptotic pathway through a functional death receptor (5). Another decoy receptor, DcR3, is produced in a secreted form and binds competitively to FasL. The DcR3 gene is amplified in a number of colorectal carcinomas, and an increase in DcR3 by these tumors has been proposed to be responsible for immune evasion, thus giving the tumor an additional survival advantage (106).
In some physiological conditions hepatocytes have been found that produce a truncated variant of Fas lacking the transmembrane domain. mRNA encoding a secreted soluble Fas (sFas) has been detected in a number of hepatitis and hepatocellular carcinoma studies (92, 120). It is believed that, like a secreted decoy receptor, sFas binds FasL, thereby preventing engagement of surface Fas, which is required to transmit the death signal. However, without coordinate downregulation of membrane-bound Fas, the effects of sFas would be transient and act only in a limited microenvironment.
IMPACT OF FAS AND FASL: LOCATION, LOCATION, AND LOCATION
Fas and FasL in lymphoid cells.
Perhaps the best-characterized physiological system involving Fas/FasL-mediated apoptosis is that observed in lymphoid cells. A connection implicating Fas/FasL in lymphoid apoptosis was made when mice bearing the lpr (lymphoproliferative) and gld(generalized lymphoproliferative disorder) phenotypes (25, 94) were discovered to be associated with mutations in the Fas and FasL genes, respectively. Both lpr and gld mice develop lymphadenopathy and splenomegaly, and in some strains we see the penetrance of autoimmune disorders (94). Similarly, mutations of Fas and FasL in humans are associated with autoimmune lymphoproliferative syndrome (ALPS) types Ia and Ib, respectively (9, 67, 68, 79, 125).
A major role of Fas and FasL on lymphoid cells appears to be controlling immune responses by regulating the number of lymphocytes through a process known as activation-induced cell death (AICD) (16,29, 57). After antigen-driven expansion of lymphocytes, there is an upregulation of FasL on the activated T cells that is required for subsequent death and removal of activated lymphocytes (139). Persistence of a population of activated T cells after the removal of antigen could lead to potential deleterious effects such as autoreactivity. Clearance of activated T cells through FasL has been demonstrated to occur in an autonomous fashion, suggesting a self-limiting feedback inherent in some activated lymphocytes (16). In addition to AICD, activated T cells can also be deleted by expression of FasL in peripheral, nonlymphoid tissues. This is discussed in detail below.
It is important to remember that studies on Fas/FasL in controlling an immune response have been done largely on lymphocytes in the periphery or in vitro and not specifically in organs such as the liver and gastrointestinal tract. Although an uncontrolled lymphocyte response may have dire consequences indirectly on the liver and gastrointestinal system, the most potentially harmful effect from FasL present on lymphoid cells is its indiscriminant cytolytic activity toward Fas-positive cells. In many immune reactions lymphocytes (which are presumably activated and express FasL) infiltrate the intestine and liver, and it is in this manner that lymphoid FasL can cause injury in target cells in these organs.
Like most members of the TNF family of apoptosis-inducing ligands (with the exception of LT-α), FasL is expressed as a type II transmembrane molecule (127), associated either with intracellular membranes of secretory vesicles or with the plasma membrane, where it may mediate apoptosis through the interaction with its receptor. It was first noted that transmembrane TNF-α can be proteolytically cleaved by a metalloprotease to its soluble form. Recently, the TNF-α-converting enzyme (TACE) protease has been cloned (19), and specific inhibitors of TACE are therapeutically used to block the generation of soluble TNF-α in different diseases. Similar to TNF-α, most other members of the family, including FasL, can be processed by metalloproteases into their soluble form. Although soluble TNF-α has a well-characterized inflammatory activity, the physiological role of soluble FasL (sFasL) is less clear. sFasL still trimerizes efficiently and thus still binds to its membrane-bound receptor. The cleavage site of FasL has been recently identified (117). In 293 cells, FasL is cleaved between amino acids 126 and 127 and the proteolytic processing does not affect the domain responsible for trimerization. However, sFasL has been found to be only a poor inducer of apoptosis and very high concentrations are required to trigger apoptosis in otherwise sensitive cells. Thus membrane anchoring of the molecule appears to be important for the apoptosis-inducing activity of FasL. Although both forms of FasL bind efficiently to their receptors, receptor-bound sFasL appears to be rapidly internalized and leads to the downregulation of cell surface Fas receptor. Similar observations have been made with other immunologically important receptors, including T-cell receptor and chemokine receptors. sFasL may therefore have an antagonistic rather than agonistic activity and may even compete for the apoptosis-inducing activity of membrane-bound FasL (126, 131). In contrast, if sFasL is further polymerized, e.g., through cross-linking antibodies, it regains its apoptosis-inducing activity (117), presumably through stabilization of the ligand-receptor complex on the cell surface.
The generation of sFasL through metalloprotease activity is also observed in vivo in a variety of inflammatory diseases. Serum levels of sFasL are significantly elevated in patients suffering from acute graft vs. host disease (GVHD) after bone marrow transplantation (26, 58). Similarly, elevated sFasL levels positively correlate with liver damage during acute self-limited and fulminant hepatitis (120). Thus tissue damage appears to be sensed by tissue cells, leading to enhanced metalloprotease activity and sFasL generation. It is tempting to speculate that under these disease conditions sFasL may have a protecting function and may reduce tissue damage by competing for membrane-bound FasL on infiltrating cytotoxic T cells and down-modulating tissue Fas receptor. Whereas the conversion of membrane-bound cytotoxic FasL to its protective soluble form may represent an important regulatory mechanism by which sensitive tissues, such as hepatocytes and IEC, protect themselves from excessive immune responses, it may also have been adapted by tumor cells to inhibit antitumor immune responses. A variety of tumor cells, originating from almost all different types of tissues, have been shown to express FasL and may thus escape immune surveillance by inducing apoptotic cell death in tumor-infiltrating cytotoxic T cells (141). In addition to this cytotoxic immune escape mechanism, tumor cells may protect themselves from FasL-mediated suicide and Fas-mediated killing by T lymphocytes by down-modulation of its receptor (73, 86). Therefore, sFasL may represent a disease-promoting risk factor, a role that has been proposed for sFas in patients suffering from autoimmune disease (24).
FAS LIGAND IN NONLYMPHOID TISSUES
Although the immune system was the first and is still the best-characterized system in which Fas and FasL are principal mediators of apoptosis, a role for Fas/FasL in nonlymphoid tissues is becoming increasingly evident (Fig. 2). French et al. (37) observed expression of FasL in nonlymphoid sites in the developing mouse embryo and in adult mice. In addition to lymphoid tissues (spleen and thymus), these researchers also detected significant levels of FasL mRNA in nonlymphoid tissues at sites of immune privilege (testes, uterus, and ovaries) and at sites that accommodate frequent lymphocytic infiltration (small intestine, large intestine, lung, and liver), suggesting, in addition to a role in AICD as discussed above, a role for FasL in physiological cell turnover and protection against potentially harmful lymphocytes. Two of the most closely studied systems involving the expression of FasL in nonlymphoid tissue are the phenomena of immune privilege and peripheral deletion.
Immune privilege is the term given to sites that are receptive to transplant because of the lack of resident or infiltrating lymphocytes. It is believed that immune privilege is a mechanism to prevent the slightest inflammatory response that can accompany an immune response (42). For example, when an inflammatory response is generated in the eye, a principal site of immune privilege, there is a dramatic reduction in vision (34). Constitutive expression of FasL in immune privilege sites ensures that any Fas+ lymphocytes that infiltrate these tissues are removed rapidly and efficiently. FasL was shown to be important in immune privilege when Griffith et al. (43, 44) demonstrated that virus-infected cells injected into the anterior chamber of the eye were cleared rapidly and without resulting in recruitment of lymphocytes or granulocytes to the site of infection. Similar infection in the eyes of FasL-defective gld mice resulted in recruitment and lymphocytic infiltration followed by inflammation at the site of infection.
The FasL model of immune privilege has been employed as a technique to avoid graft rejection. Bellgrau et al. (8) observed that FasL+ Sertoli cells injected under the kidney capsule of recipient mice were not rejected, whereas Sertoli cells fromgld mice did not survive transplantation. By virtue of the presence of surface FasL, the graft is believed to defend itself against host attack. This strategy, however, does not appear to be universal, because FasL+ islet β-cells transplanted under the kidney capsule of allogeneic recipient mice resulted in a massive granulocytic infiltration and efficient clearance of the graft (3). In this experimental system, expression of FasL on β-cells did not confer any resistance to rejection compared with that of FasL-negative β-cells.
Expression of FasL does not automatically confer immune privilege on a tissue. In fact, under some circumstances, the presence of surface FasL has been shown to evoke an inflammatory response. When Neuro-2a neuroblastoma cells transfected to express FasL were injected into syngeneic recipient mice, there was a major infiltration by neutrophils resulting in inflammation at the site of injection (119). There was no such inflammatory response if the donor cells were not expressing FasL, if the animals were treated with neutralizing antibody to FasL, or if the FasL-transfected Neuro-2a cells were introduced into lprmice. In similar model systems, it has been demonstrated that rejection of FasL-bearing transplanted tumor cells in syngeneic hosts was carried out by infiltrating CD8+ cytotoxic lymphocytes (CTL) and natural killer (NK) cells (4, 118). The inflammatory response observed in these situations appears to be due, at least in part, to the effects of transforming growth factor (TGF)-β in the microenvironment of the donor cells. Chen et al. (22) showed that the strong inflammatory response observed after injection of CT26-CD95L into syngeneic hosts was suppressed if TGF-β was present at the site of injection. The suppressive effects of TGF-β on the inflammatory response in this experimental system are believed to be through inhibition of neutrophil activation.
After an antigenic stimulus is removed, the immune system is faced with the task of removing activated lymphocytes from circulation (Fig.3). Failure to do so often leads to autoimmune and lymphoproliferative disorders. In fact, the characterization of the lymphoproliferative (lpr) and generalized lymphoproliferative disorder (gld) phenotypes arise from mutations in the Fas and FasL genes, respectively (25, 94). It is now clear that peripheral deletion of activated lymphocytes is mediated by Fas/FasL interactions by one of two mechanisms. AICD is a process by which lymphocytes act as both effector and target cells in apoptosis, and it has been demonstrated that lymphocytes, when properly activated, can induce apoptosis in an autonomous fashion. A second mechanism of peripheral deletion has been described recently (13) in which activated Fas-bearing lymphocytes are deleted after an upregulation of FasL expression by nonlymphoid tissues at common sites of major lymphocytic infiltration, notably, hepatocytes and intestinal epithelial cells. This model explains, at least in part, the observation of splenomegaly in lpr and gld mice and in humans with type I ALPS, which is attributed to genetic defects in Fas or FasL.
ROLE OF FAS/FASL IN DISEASES OF LIVER AND GASTROINTESTINAL SYSTEMS
Studying Fas/FasL in gut and liver disease: words of caution.
When studying the role of Fas/FasL in gut and liver disease it is important to remember that demonstrating that cells found in diseased tissue express Fas and FasL is not sufficient to demonstrate that Fas-mediated cell death is involved. In fact, as discussed above, not all Fas-expressing cells are susceptible to Fas-mediated cell death (1), and thus it is preferable to demonstrate that Fas+cells are susceptible to death through the use of either an anti-Fas monoclonal antibody (MAb) (1, 100, 101) or cross-linked sFasL (117).
Similarly, the presence of FasL on the cell membrane does not necessarily imply that the FasL expressed is functional (i.e., capable of mediating death of Fas-sensitive target cells). Depending on the type of tissue, membrane FasL in some tissues can be rapidly cleaved by metalloproteases (60) into a soluble and potentially less cytotoxic form. Furthermore, in an issue perhaps more relevant to tumor cells, mutations in FasL can result in normal levels of FasL expression on the surface, but the FasL that is expressed has lost its ability to induce Fas-mediated cell death. This appears to be the case with thegld mutation, where a point mutation results in a normal level of FasL expression but the FasL that is expressed is nonfunctional (130). Thus, if possible, it is preferable to demonstrate that the FasL that is expressed is capable of inducing death in Fas-sensitive target cells (72, 109). The specificity of cell death in these systems can readily be examined with the use of FasFc (17, 72) or anti-Fas blocking MAbs (47, 60) or by comparing the susceptibility of Fas-transfected and nontransfected target cells (109, 112).
If Fas and FasL have a major role in diseases of the liver and gastrointestinal tract, it is probably more so in pathological apoptosis rather than homeostatic apoptosis. Although apoptosis plays a major role in the homeostasis of the liver and gastrointestinal tract, there have been to date no reports of increased liver and intestinal tumors in FasL- and Fas-mutated gld and lpr mice. There is, however, a possibility that Fas and FasL play a role in the homeostasis of the liver. Fas-null mice have been reported to have larger livers composed of hepatocytes with enlarged nuclei typical of senescent cells, suggesting that Fas/FasL may regulate the removal of unwanted old hepatocytes. Interestingly, these findings have not been reported in lpr mice, but it has been suggested that this is the result of leakiness in the lpr mutation, allowing some residual Fas signaling.
There are several reasons to believe that the Fas/FasL system is of special importance in pathological disease of the liver and gastrointestinal system. First, many cells in the liver and gastrointestinal system have been shown to express Fas and be highly susceptible to Fas-mediated death. The hepatocytes in the liver appear to be especially susceptible to Fas-mediated death, because injection of anti-Fas induces massive injury to the liver and not elsewhere. In the gastrointestinal tract, T cells in human lamina propria of the colon have been shown to be more susceptible than T cells in the periphery to Fas-mediated apoptosis from either anti-Fas or anti-CD2 activation (11, 27) and IEC have been shown to express Fas and under certain conditions be sensitive to Fas-mediated cell death (45, 72,112). Finally, in addition to being more susceptible to Fas-mediated death, intestinal lymphocytes, and under certain situations IEC, have been shown in several studies to be able to readily induce Fas-mediated death. Overall, these results suggest that the liver and gastrointestinal tract are organs that are primed for Fas-mediated injury (13, 27, 72).
Fas/FasL in GVHD.
GVHD is an immunologically mediated disease in which donor (graft) T cells recognize recipient (host) cells as foreign, thereby resulting in the death of host cells (reviewed in Ref. 35). The disease is manifested by systemic injury that results in depletion of host immune cells, wasting, and eventually death. GVHD is important in gastrointestinal and liver disease because the intestinal epithelium and liver hepatocytes are major targets of injury in this disease. Ponec et al. (107) recently reported the observation that intestinal lesions in human GVHD are characterized by apoptosis of intestinal crypt epithelial cells, resulting in crypt destruction. Injury to these two organs can potentially contribute to systemic symptoms because the intestinal epithelium and liver both serve as an important barrier in preventing the translocation of harmful toxins and bacteria found normally in the intestinal lumen. Furthermore, the intestinal epithelium serves as an absorptive organ for vital nutrients, and dysfunction can lead to diarrhea, malabsorption, and malnutrition.
Because GVHD is mediated primarily by donor T cells and T cells mediate cytotoxicity of target cells through two major mechanism of apoptosis, one dependent on perforin and the other dependent on FasL (15), it is reasonable to expect that a major mechanism of injury in GVHD is from FasL present on donor cells inducing death of Fas-positive host cells. Several studies have shown that FasL present on donor cells is involved in mediating some of the systemic symptoms seen in GVHD. Via et al. (140) demonstrated that depletion of host lymphocytes is in part dependent on donor FasL. In addition, two groups have shown that wasting and lethality are significantly attenuated in GVHD when donor cells lack functional FasL (7, 15). However, because most studies using murine models focus on the systemic complication (this is probably because it is difficult to accurately quantify organ injury), it is not evident from these studies whether Fas and FasL play a role in liver and gastrointestinal complications seen in GVHD.
In perhaps the only study that addressed the involvement of Fas/FasL in inducing liver injury during GVHD, Baker et al. (7) demonstrated that skin and liver injury in GVHD is almost completely dependent on functional donor FasL, but they stated that they did not observe any differences in intestinal injury. However, how closely the intestine was examined is not clear. Using neutralizing MAbs that block Fas- and TNF-mediated death, Hattori et al. (47) demonstrated that a severe form of intestinal injury seen in GVHD was more likely to be the result of TNF than of FasL. Although we can conclude that Fas/FasL probably does not play a major role in this model of severe intestinal GVHD, it is probably premature to conclude that Fas and FasL have no role in inducing intestinal epithelial injury in GVHD because it is unclear whether the neutralizing anti-Fas ligand antibody used by Hattori et al. might have been more effective in preventing intestinal epithelial injury if used at a larger dose or in a milder form of intestinal GVHD. Using a different model of murine GVHD in which intestinal epithelial injury is milder and quantitated by IEC apoptosis, two groups have demonstrated that donor T cells infiltrating the intestinal epithelium during GVHD are very capable of mediating IEC apoptosis through largely a Fas-mediated process (72, 112). It is likely that both TNF and FasL play a role in inducing intestinal epithelial injury during GVHD, but the importance of each may depend on which model is examined and on how epithelial injury is quantified.
Role of Fas/FasL in inflammatory bowel disease.
Most studies implicating Fas/FasL in human inflammatory bowel disease have been performed on ulcerative colitis rather than Crohn's disease. The role of Fas/FasL in ulcerative colitis is centered on the hypothesis that Fas+ IEC are targeted by FasL+lymphocytes resulting in IEC apoptosis. A corollary to this hypothesis is that injury to the IEC subsequently leads to intestinal lymphocytes being exposed to luminal antigens resulting in an uncontrolled inflammatory response. In agreement with this hypothesis, apoptosis of IEC in the crypts has been reported in ulcerative colitis (55, 124) and Fas and FasL have been shown to be upregulated on IEC and intestinal lymphocytes, respectively, in ulcerative colitis (15, 89, 138).
In a different proposed role for Fas/FasL in Crohn's disease, two groups made the interesting observation that lymphocytes found in the inflamed tissue of Crohn's patients expressed Fas yet were relatively resistant to Fas-mediated cell death compared with lymphocytes from noninflamed controls (10, 53). Thus, if Fas/FasL is involved in the eradication of potentially harmful Fas+ T cells in the intestine, then the resistance of intestinal lymphocytes to Fas-mediated death in Crohn's disease may allow potentially harmful lymphocytes to survive longer and induce more damage of the intestine.
Interestingly, several studies have shown that nonlymphoid cells found in the intestine are capable of expressing FasL. Paneth cells appear to constitutively express FasL, and in ulcerative colitis IEC express FasL (89). Although the significance of these nonlymphoid cells expressing FasL is speculative, it is possible that FasL expression here is a protective response by these nonlymphoid cells against harmful neighboring Fas+ T cells (analogous to the role of FasL in immune privilege discussed above). Conversely, FasL expression can induce apoptosis of surrounding cells, thereby contributing to the pathogenesis of ulcerative colitis.
Despite the availability of numerous models of murine inflammatory bowel disease, very few studies have evaluated the role of Fas/FasL in these models. In a murine model of colitis in which CD4+cells were transferred into immune-compromised SCID mice, which essentially have no B or T cells, Bonhagen et al. (14) demonstrated that CD4+ cells that infiltrate the colon express FasL and are capable of killing target cells in vitro through a Fas-mediated pathway. In a similar model of murine colitis, in which nonallogeneic bone marrow cells are transferred into T cell- and NK cell-deficient Tg26 mice (BM→Tg26), Simpson et al. (122) demonstrated that lymphocytes found in the inflamed colon were capable of inducing Fas-mediated apoptosis in vitro. Although both studies clearly demonstrated that lymphocytes found in the diseased intestine are readily capable of inducing Fas-mediated death, it is unclear whether the Fas/FasL pathway is involved in the pathogenesis of the colitis. In fact, when gld bone marrow cells were injected into Tg26 mice there was no clear reduction in the level of colitis, suggesting that the Fas/FasL system plays a minor role in the pathogenesis of this model of colitis. Clearly, studies in which the host mice are in alpr background and studies investigating the role of Fas/FasL in other murine models of inflammatory bowel disease are needed.
One of the early indications that Fas/FasL may be important in the maintenance of hepatic homeostasis and therefore in the development of hepatic disorders was the observation that anti-Fas antibody, when administered to mice in vivo, caused massive hepatic apoptosis resulting in hepatic failure and death soon thereafter (101). It was postulated that the effects of anti-Fas antibody were caused by the induction of apoptosis of hepatic lymphocytes; however, it soon became evident that the issue was more complex. Given the high susceptibility of the liver to injury from anti-Fas antibody, it is not surprising that the role of Fas/FasL has been the intense topic of numerous studies of diseases of the liver. It is unlikely, however, that Fas/FasL plays a major role in normal hepatocellular turnover becauselpr and gld mice do not display an increased incidence of liver neoplasia despite nonfunctional Fas or FasL (25).
Liver pathologies that arise from dysfunctional Fas or FasL are more likely the result of infiltrating or resident lymphocytes. There are two views on the role of the liver in the deletion of peripheral lymphocytes. It is either a site where dying lymphocytes home and are subsequently removed from action or a site where live, activated lymphocytes are induced to die and are deleted. Regardless of the contribution of each of the scenarios, the liver has been shown to be a site for clearance of lymphocytes (52). Also, if there is a reduction in peripheral deletion such as in the lpr and gld mouse models, as well as in GVHD, the liver is one of the target organs that suffer from distress inflicted by autoreactive or inappropriately active lymphocytes.
Fas/FasL in liver transplants.
Because of the high sensitivity of the liver to Fas-mediated injury and a major role of CTL in graft rejection, it is not surprising that several studies have focused on the role of Fas/FasL liver transplant rejection (32, 59, 70, 71). In a rodent model of liver transplantation in which donor hepatocytes are injected into host spleen, Kawahara et al. (59) demonstrated that host FasL in the spleen plays a role in rejection of donor hepatocytes. However, from their studies as well as others (70) it appears that the Fas/FasL system is not the only mechanism involved.
Interestingly, despite the high susceptibility of hepatocytes to undergo Fas-mediated death, the issue of whether FasL can be present on donor hepatocytes during rejection or whether FasL+hepatocytes can help prevent liver graft rejection is still unresolved. FasL mRNA levels have been shown to increase in a transplanted liver undergoing rejection, but it is unclear from these studies whether the increase in FasL mRNA is from infiltrating donor lymphocytes of host hepatocytes (32). At first thought, the idea of hepatocytes upregulating FasL expression to prevent rejection appears unlikely because expression of FasL by adenoviral transfection has been shown to result in massive hepatitis and death (90). However, using liposome vesicles to transfect various levels of FasL into the liver, Li et al. (71) demonstrated that relatively low levels of FasL on hepatocytes did not induce hepatitis and, perhaps more interesting, demonstrated that livers expressing low levels of FasL survived longer when transplanted into allogeneic-mismatch recipients.
Finally, Krams et al. (66) made the observation that the liver is capable of producing sFas, which is capable of blocking Fas-FasL interactions and hence Fas-mediated death. Perhaps more intriguing, sFas levels were found to be lower in patients undergoing liver graft rejection compared with patients with stable grafts. This suggests that sFas may provide protection from Fas-mediated injury during rejection and that lower levels of sFas produced by the liver may predispose a liver graft to rejection.
Hepatitis B and C.
Numerous studies have implicated Fas and FasL in both hepatitis B and C. Although human hepatocytes appear by histochemical studies to express relatively low levels of Fas, they also appear to be extremely sensitive to Fas-mediated cell death (38). Fas expression has been shown to be upregulated on hepatocytes from both hepatitis B and C (39,48, 88, 102), and the degree of Fas expression correlates with the severity and location of liver inflammation. However, it remains unclear whether upregulation of Fas expression contributes to the pathogenesis of viral liver disease. One study showed that infection of a hepatoma cell line with the hepatitis C core protein increased susceptibility to Fas-mediated death, yet this did not appear to be the result of increased Fas expression (110).
Lymphocytes found in the inflamed region of the liver in viral hepatitis have been shown to express FasL (48, 85). Presumably, it is FasL on these lymphocytes mediating the death of Fas+hepatocytes that contributes to liver injury, resulting in end-stage liver disease. In agreement with this hypothesis, Kondo et al. (64) demonstrated that CTL specific for the viral hepatitis B surface antigen induced acute liver disease through a Fas-mediated mechanism in transgenic mice expressing the hepatitis B surface antigen on hepatocytes.
Alcohol liver disease.
Interestingly, there have been very few studies on the role of Fas/FasL in alcohol liver disease. One of these reports demonstrated that Fas is minimally upregulated in alcohol liver disease (38); however, most of the liver examined appeared to have been at the end stage of alcohol liver disease (cirrhosis), where there is normally very little inflammation. Thus it is plausible that Fas expression can be upregulated in acute alcohol hepatitis, a stage in which inflammation is more prominent. Interestingly, the authors reported that hepatocytes from alcohol liver cirrhosis expressed high levels of FasL; however, any contribution to the disease is speculative at this time.
Murine models of hepatitis.
In several murine models of hepatitis, several groups have demonstrated that Fas/FasL plays an important role in most inflammatory hepatitis. Using mice that lack functional Fas (lpr) and FasL (gld), two groups have demonstrated that hepatitis induced by injection of the potent T cell stimulator concanavalin A is mediated by Fas (118, 129). In support of the involvement of Fas in murine hepatitis, Kondo et al. (64) demonstrated that Fas-null mice were resistant to the hepatitis induced by Propionibacteriuminfection and subsequent lipopolysaccharide challenge. Finally, Fleck et al. (36) demonstrated that the hepatitis seen in murine cytomegalovirus is less severe in lpr mice.
Copper is usually present as a trace element required for optimal function of such enzymes as cytochrome oxidase and superoxide dismutase, among others, and when present in excess can lead to protein and lipid oxidation that, in turn, leads to production of reactive oxygen species. Wilson's disease (WD) is an autosomal recessive disorder that results from copper overload (116), and in 1993 several groups identified the underlying cause of WD as mutations in a gene encoding a copper-transporting P-type ATPase (18, 21, 105, 132). Clinically, WD manifests with hepatolenticular degeneration in early stages followed by decreased motor control as copper excess accumulates in the brain. A link between copper metabolism deficiencies in WD and apoptosis was drawn by Strand et al. (123), who noted that free radicals observed in cells with excess copper (II) ions could be functioning in a biochemical manner similar to free radicals frequently generated during apoptosis. Using HepG2 hepatocarcinoma cells treated with exogenous copper, these researchers demonstrated that Cu2+-induced apoptosis acts through Fas. Treatment of HepG2 cells with 100 μM Cu2+ resulted in an induced upregulation of FasL, and death in this system was inhibited by the broad-spectrum caspase inhibitor z-VAD-fmk.
There is also a growing body of evidence that Fas/FasL plays a role in biliary disease in mice. Apoptosis of rodent hepatocytes by bile salt or from hepatic ligation appears to be Fas mediated (33, 87). Finally, Chen et al. (23) recently reported an interesting observation that apoptosis of a biliary epithelial cell line with Cryptosporidiais partially Fas mediated and that the mechanism is in part due to upregulation of FasL on the biliary cells.
Fas/FasL in tumors of liver and gastrointestinal tract.
In the absence of T cells, Fas-deficient mice have been shown to develop malignant and lethal B cell lymphoma, suggesting that Fas can function as a tumor suppressor (104). As of yet, there have been no reports of liver and gastrointestinal tumors in these mice, which raises doubt as to whether Fas can function as a tumor suppressor in these organs. Nevertheless, the notion that hepatoma and colon cancer can escape immune surveillance by FasL-bearing tumor-infiltrating lymphocytes by becoming resistant to Fas-mediated death is an attractive hypothesis. Several hepatoma and colon cancer cell lines have been shown to be resistant to Fas-mediated death (96,136). One interesting mechanism by which colon cancers become resistant to Fas-induced cell death is through the expression of “decoy receptors,” which bind and block FasL (106).
Another intriguing role for Fas/FasL in cancer has been proposed and termed “Fas counterattack.” According to this theory, tumor cells express FasL, enabling them to evade immune destruction by inducing apoptosis of activated Fas+ T cells. Several colon cancer cell lines have been shown to express functional FasL (98). In addition, expression of functional FasL has been reported to be found more commonly on hepatic metastatic colon cancer than primary colon cancer, raising the possibility that the presence of FasL enables some colon cancer to more easily infiltrate the Fas-sensitive liver (78, 121, 146). Although these studies suggest that FasL expression in colon cancer is advantageous to the tumor, it is not entirely clear that this occurs in vivo. Chen et al. (22) demonstrated that overexpression of FasL in the CT-26 colon cancer line resulted in it being more readily rejected when transferred into immune-deficient mice.
There is significant evidence that Fas and FasL play a significant role in the pathogenesis of a wide spectrum of liver and gastrointestinal disease. Although a major mechanism of tissue destruction involving Fas/FasL is through the death of Fas-positive target cells by FasL+ lymphoid cells, there are several important issues that require more clarification. First, there is growing evidence that nonlymphoid cells found in the liver and gastrointestinal organs express FasL. Whether the expression of nonlymphoid FasL in these organs will exacerbate or ameliorate liver and gastrointestinal diseases remains to be shown. Second, sFasL and sFas have both been shown to be increased in several liver and gastrointestinal diseases. However, whether sFas and sFasL play an important role in these diseases in vivo remains to be shown. Finally, it is important to remember that Fas is only one member of a growing list of the TNF receptor family of proteins that are capable of inducing apoptosis in the presence of their respective ligand. Thus future studies addressing the role of other members of the TNF receptor apoptosis family (TNFR1, DR3/TRAMP/WSL, and DR4, DR5) with their respective TNF-like ligands (TNF and LT-α, TWEAK/APO-3L, and TRAIL/APO-2L) in the gastrointestinal and liver diseases discussed here will be of extreme interest.
The excellent secretarial skill of L. Gentry and T. Smith and words of encouragement from Dr. W. A. Olsen are greatly appreciated.
Address for reprint requests and other correspondence: Dr. Tesu Lin, Searle #10–541, GI Division, Dept. of Medicine, Northwestern Univ., 303E Chicago, Chicago, IL 60611 (E-mail:).
M. J. Pinkoski is a postdoctoral fellow of the Medical Research Council of Canada. This is manuscript no. 327 of the La Jolla Institute for Allergy and Immunology.
- Copyright © 2000 the American Physiological Society