Induction of TNF receptor I-mediated apoptosis via two sequential signaling complexes

Read Also

Although the molecular mechanisms of TNF-induced activation of pro-survival pathways (NF-kB, JNK) have been reasonably well elucidated (Baud and Karin, 2001; Devin et al., 2001), the principle deciding on whether TNF signals cell survival or cell death remains largely unknown.  Our data now provide evidence that the decision is not made at the level of the rapidly formed complex assembling around the ligand-bound TNFR1 at the plasma membrane.  Commitment to cell death is slow and is dependent on a complex that dissociates from TNFR1 (complex II) and which is found mostly in the cytoplasm. 

The results presented in this paper are compatible with the model outlined in Fig. 9.  TNFR1 stimulation leads to the rapid assembly of a complex (complex I) comprising the receptor itself, TRADD, RIP1, TRAF2, c-IAP1 and possibly other known (c-IAP2, FAN etc.) or yet unidentified proteins. Complex I is, however, devoid of FADD and caspase-8.  Complex I triggers the NF-kB signaling pathway via recruitment of the IKK complex (Zhang et al., 2000) whereas JNK is activated via TRAF2-mediated activation of MAP3-kinases (Chen and Goeddel, 2002).  Assembly of complex I occurs in lipid rafts (Legler et al., 2002) where posttranslational modifications of several complex-associated proteins are likely to occur.  For example, complexed TNFR1, which in its non-stimulated state exhibits an apparent molecular mass of 48 kDa, forms molecular species with apparent mw ranging from 48 kDa to up to 150 kDa.  Also, up to 50% of TRADD present in complex I undergoes modifications that increase its molecular mass from 35 kDa to approximately 44 kDa and 55 kDa.  Finally, complex I-associated RIP1 migrates as a smear with apparent mw ranging from 78 kDa to 120 kDa. 
Formation of complex I is transient since a large portion of TRADD, RIP1 and TRAF2 dissociate from TNFR1 within an hour, at a time when TNFR1 starts to undergo endocytosis.  Dissociation of TRADD was suggested to be dependent on TNFR1 endocytosis (Jones et al., 1999), although based on our data, endocytosis and dissociation do not strictly correlate.  Our data also do not reveal whether or not the extensive modifications seen cause dissociation.  In any case, after dissociation from TNFR1, the DD of TRADD (and RIP1) previously engaged in the interaction with the DD of TNFR1 becomes available for interaction with other DD-containing proteins. FADD is a likely interaction partner for TRADD, since TRADD and FADD were previously shown to interact via their respective DD (Hsu et al., 1996; Thomas et al., 2002; Varfolomeev et al., 1996).  Although a RIP1-FADD- interaction was also described (Varfolomeev et al., 1996), it is less likely to be of importance for complex II formation since TNFR1-induced apoptosis still proceeds in RIP1-deficient Jurkat cells (Holler et al., 2000).  Thus, similar to the DD of Fas, the DD of modified TRADD may act as a central platform for the recruitment and activation of FADD, leading to the subsequent binding of caspase-8.

After recruitment of FADD and caspase-8, the decision as to whether TNF acts to promote gene transcription or apoptosis has to be made.  Indeed, in contrast to complex I, the composition of complex II in apoptosis-resistant and sensitive cells differs.  In resistant cells, complex II comprises increased amounts of the two anti-apoptotic proteins c-IAP1 and FLIP_L and the expression of which is regulated by the transcriptional activity of NF-kB (Micheau et al., 2001; Wang et al., 1998).  Inhibition of the pro-apoptotic activity of caspase-8 is more likely to occur through FLIP_L, since enforced expression of FLIP but not c-IAP1 potently blocks TNF-mediated cell death (Micheau et al., 2001).  Moreover, FLIP-/- embryonic fibroblasts are highly sensitive to TNF-induced apoptosis and show rapid induction of caspase activities (Yeh et al., 2000).  In keeping with this observation, sixteen hrs after TNFR1 stimulation, complex II is devoid of FLIP_L in sensitive cells, while it contains increased quantities of caspase-10.  Caspase-8 is known to interact with itself, caspase –10 and with FLIPs, although the preferred interaction partner is FLIP_L (Irmler et al., 1997; Krueger et al., 2001; Wang et al., 2001).  Thus, in cells with high FLIP_L content, caspase-10 has limited access to caspase-8 within complex II, while in cells expressing low quantities of FLIP_L, high amounts of caspase-10 are found associated with caspase-8.  Whether FLIP_L and caspase-10 compete for the same site on caspase-8 or whether FLIP_L indirectly competes with caspase-10 remains to be determined.  Moreover, it is not known whether caspase-10 is an essential component in the pro-apoptotic complex II, since the role of caspase-10 in TNF-mediated or in Fas-and TRAIL–mediated apoptosis is uncertain (Kischkel et al., 2001; Sprick et al., 2002).

FLIP_L availability at the moment complex II is formed is dependent on a signal previously triggered by complex I (Kreuz et al., 2001; Micheau et al., 2001).  If NF-kB-activation promotes the expression of FLIP_L, the pro-apoptotic activity of caspase-8 is inhibited.  In contrast, if complex I-triggered NF-kB activation is not productive, the amount of available FLIP_L will rapidly diminish and the proapoptotic activity of caspase-8 will not be stopped. Such a model predicts that FLIP_L plays two important roles; on the one hand it regulates whether or not TNF triggers apoptosis, and on the other hand it is also able to act as a sensor for the fidelity of the signal emanating from complex I.

This model has interesting, more general implications as it predicts that the transcriptional activity of the NF-kB signaling pathway is controlled by (a) checkpoint(s), similar to checkpoints controlling the integrity of cell cycle progression. This control mechanism is triggered immediately after TNFR1 engagement but is operational only a few hours later, at a time when the success of the transcriptional activity of NF-kB can be assessed.  Cells with defective NF-kB signals (and thus having low quantities of FLIP and other anti-apoptotic proteins) will be eliminated through TNF-induced apoptosis.

The formation of complex II may also explain the different kinetics of apoptosis induced by TNFR1 and Fas.  Fas recruits FADD directly to the plasma membrane, and subsequent activation of the two DED-containing upstream caspases is rapid and can occur within minutes.  In contrast, TNFR1 is unable to recruit FADD directly but instead recruits adaptor proteins which upon dissociation can bind FADD in a second step.  Complex II formation is clearly FADD-dependent, as demonstrated using FADD-DN or FADD-deficient cells.  Interestingly, point mutations in FADD, inhibiting the association with Fas but not with TRADD nor caspase-8, have been identified (Thomas et al., 2002).  Reconstitution of Jurkat FADD-deficient cells with FADD constructs carrying these mutations severely impair Fas-induced apoptosis, but restore TNF-induced apoptosis (Thomas et al., 2002).  Moreover, overexpression of TRADD leads to FADD-dependent cell death (Yeh et al., 1998) placing FADD downstream of  TRADD.  Recent results even suggest that in promyeolytic cells, TRADD is able to trigger cell death from within the nucleus (Morgan et al., 2002). 

Upstream caspases have to be brought in close proximity for their activation (Boatright et al., 2003).  Assembly of death receptors upon ligand binding as well as Apaf-1 complexes upon cytochrome c leads to the formation of ideal platforms for caspase activation.  It is likely that TRADD remains oligomerized upon dissociation from TNFR1 and thus brings caspase-8/10 into close proximity after recruitment of FADD.  The TRADD-induced type II complex is the first example of a soluble, cytoplasmic complex that leads to caspase-8/10 activation.  Cells deficient in TRADD however, need to be studied to conclusively draw this conclusion.