In the present study, we have demonstrated that (a) inhibition of either cathepsin B or caspases attenuates GCDC-induced apoptosis; (b) inhibition of caspases prevents the increase in cathepsin B activity, but cathepsin B inhibition does not prevent caspase activation; (c) increased caspase 8 hydrolytic activity accompanies treatment of McNtcp.24 cells with GCDC; (d) hepatocytes from Fas-deficient lpr mice are resistant to GCDC-induced apoptosis; and (e) GCDC induces Fas oligomerization, FADD aggregation, and apoptosis in the apparent absence of FasL. These novel observations not only place cathepsin B downstream of caspase 8 in bile salt–induced apoptosis but also suggest that caspase 8 activation occurs by a Fas-dependent, FasL-independent pathway. Each of these observations is discussed in greater detail below.
Cleavage of PARP and lamin B1
to signature fragments suggested activation of caspases 3 and 6 after GCDC treatment. These caspases are currently thought to be activated by upstream signaling caspases (13
). Therefore, we sought to identify the upstream protease responsible for initiating the GCDC-induced apoptotic cascade. Our results suggest that caspase 8 might play this role. This conclusion is based on the observed increase in IETD-AFC cleavage activity and on inhibition of GCDC-induced apoptosis by CrmA, a cowpox viral serpin that potently inhibits caspases 1, 4, and 8, but not caspases 3, 6, 7, and 10 (23
). In previous experiments, we were unable to find an increase in caspase 1 activity during GCDC-mediated apoptosis using the fluorogenic substrate YVAD-AMC (5
), thus eliminating caspases 1 and 4 as the CrmA-inhibitable protease causing apoptosis in our system. Caspase 10, another signaling caspase with NH2
-terminal death effector domains, is not inhibited by CrmA (39
). Thus, caspase 8 is likely to be the key target of CrmA during bile salt–induced apoptosis.
In addition to caspase 8, cathepsin B also appeared to be activated during GCDC-induced hepatocyte apoptosis. When we attempted to perform experiments to determine whether cathepsin B was downstream of caspase 8, we observed unexpected inhibition of cathepsin B with tetrapeptide fluoromethylketones and aldehydes designed to selectively inhibit caspases. Although the Ki for these inhibitors is over 1,000-fold lower for caspase 3 than for cathepsin B (the Ki for DEVD-CHO is 0.35 nM vs. 1.3 μM for cathepsin B), these inhibitors are frequently used in the 50–100 μM range in experiments using intact cells. Until the intracellular concentrations of the inhibitors are known, our results suggest that data relying solely on the use of high concentrations of these inhibitors need to be interpreted with caution, because the inhibitors might not be as selective as presumed.
As a result of these observations, we used CrmA to examine the relationship between caspases and cathepsin B. Inhibition of caspases with CrmA blocked the increase of cathepsin B activity, whereas inhibition of cathepsin B with FA-fmk or CA-074-Me did not block caspase activation. These data suggest that cathepsin B is downstream of and perhaps dependent on caspase 8 for activation. Consistent with this model, additional experiments using hepatocytes from a cathepsin B knockout mouse have confirmed that caspase 8 activation does not require cathepsin B, but cathepsin B does contribute to bile salt–induced apoptosis (Gores, G.J., unpublished observations).
The mechanism by which caspase 8 results in enhanced cathepsin B activity is currently unknown. Because cathepsin B is predominantly within acidic vesicles, a direct pathway might involve internalization of the Fas receptor/FADD/caspase 8 complex and fusion with cathepsin B–containing vesicles. Monney and coworkers (40
) have previously suggested this pathway for death-receptor signaling to explain their observation that alkalinization of acidic vesicles reduces apoptosis. The internalization of the Fas receptor signaling complex may be facilitated by activation of the Fas receptor in a ligand-independent manner as observed in our studies. Internalization of the DISC and fusion with acidic vesicles may also explain how ceramide generation by acidic sphingomyelinase (also present in vesicles) occurs in apoptosis (40
). Alternatively, caspase 8 could modulate vesicular function via an indirect, currently unidentified, cytosolic signaling pathway. Although further data are needed to elucidate the potential interactions between caspase 8 and acidic vesicles, the published data, along with the present observations on cathepsin B, suggest that acidic vesicles and their constituents may contribute to the execution of the apoptosis program.
Three additional observations suggest that bile salt–induced caspase 8 activation involves the Fas receptor and the adapter protein FADD. First, hepatocytes from Fas-deficient lpr mice are resistant to GCDC-induced apoptosis. Second, overexpression of the viral death-effector domain–containing protein MC159, which competitively inhibits the binding of caspase 8 to FADD, attenuates bile salt–induced apoptosis. Third, overexpression of DN-FADD inhibits GCDC-mediated apoptosis.
Despite the data implicating Fas and FADD in GCDC-induced apoptosis, FasL expression could not be detected by PCR even in GCDC-treated cells. Induction of Fas receptor–mediated cell death in the absence of FasL has been observed previously in Fas receptor overexpression systems (41
) and UV-treated cells (30
). Using a cross-linking approach described in the latter study, we have demonstrated GCDC-induced Fas oligomerization despite the apparent absence of FasL. Studies with GFP–DN FADD likewise demonstrate Fas aggregation in cells treated with toxic bile salts. The potential mechanisms by which toxic bile salts directly induce these interactions are unknown but likely relate to the hydrophobicity of the bile salts and their ability to intercalate in membranes or bind to proteins (42
). A better understanding of the precise mechanisms by which bile activates the Fas/FADD system may provide insight into Fas receptor dynamics as well as potential therapeutic strategies for cholestatic liver disease. These data do not, of course, rule out the possibility that bile salts might also promote oligomerization of other death-domain–containing receptors, such as the TRAIL receptor, which could also contribute to bile salt–mediated apoptosis in vivo.
In summary, data in the current study suggest that toxic bile salts cause cell death, in part by activating a protease cascade. The proximal signaling protease caspase 8 appears to be activated by toxic bile salts in a Fas receptor–dependent but FasL-independent manner. After caspase 8 activation, cathepsin B activity also increases. Inhibition of either protease attenuates apoptosis in vitro, suggesting that they both play a critical role in bile salt–induced apoptosis. The implications of these results for potential therapy of cholestatic liver disease are currently under investigation.