Pathways of cell death.
PCD can be either apoptotic or necrotic. Apoptosis is characterized by membrane blebbing, shrinking, and condensation of the cell and its organelles (13
). Two well-established pathways lead to apoptosis: the death receptor (DR) (extrinsic) pathway and the mitochondrial (intrinsic) pathway (15
). Both pathways depend on cysteine proteases called caspases (15
). However, apoptosis-like PCD can sometimes proceed without caspase activation (17
). Furthermore, caspase activation does not always lead to cell death (19
), and caspase-8 also has pro-survival functions (20
). Necrosis is characterized by swelling of the cell and its organelles, culminating in membrane disruption and cell lysis, often accompanied by inflammation. Failure of energy metabolism and massive generation of ROS are each thought to cause necrosis (22
NF-κB suppresses both PCD types, although initially it was thought to antagonize only apoptosis. The first clear evidence for NF-κB as a PCD inhibitor was provided by RelA knockout mice that die mid-gestation by massive liver apoptosis (23
). The role of NF-κB in embryonic liver survival, brought about by inhibition of TNFR1-mediated apoptosis (24
), is underscored by the very similar phenotypes of mice lacking IKKβ (4
) or IKKγ (26
). A protective role for NF-κB in adult liver was confirmed in mouse models of liver damage (10
) and involves inhibition of both apoptosis and necrosis (9
). We will discuss the various mechanisms by which NF-κB suppresses PCD (Figure ).
Figure 2 Control of cell survival and death through NF-κB–JNK cross-talk. Positive feedback loops exist between ROS and caspases, caspases and JNK, and JNK and ROS. Negative feedback loops exist between NF-κB and caspases, and NF-κB (more ...) NF-κB and caspases.
There are 2 groups of DRs, based on their signaling complexes. The first group comprises Fas, DR4, and DR5, which directly recruit the death domain–containing (DD-containing) adaptor Fas-associated death domain (FADD), procaspase-8, procaspase-10, and the cellular FLICE-inhibitory protein (FLIP) to form death-inducing signaling complexes (DISCs) (29
). The second group comprises TNFR1, DR3, DR6, and ectodysplasin A receptor (EDAR). TNFR1 forms a signaling complex (complex I) at the plasma membrane by recruiting the adaptor TNFR1-associated DD protein (TRADD) and the signaling proteins TNFR-associated factor 2 (TRAF2), TRAF5, and receptor-interacting protein 1 (RIP1). After assembly, complex I dissociates from TNFR1, which can then recruit FADD and caspase-8 and trigger an apoptotic response (30
NF-κB as a transcription factor induces genes whose products prevent PCD. An elicitor of NF-κB activation is TNF-α, which is a rather poor inducer of PCD. TNF-α triggers PCD only when new protein or RNA synthesis is inhibited or in NF-κB–deficient cells. NF-κB exerts its pro-survival activity through several antiapoptotic proteins, including FLIP, Bcl-XL
, A1/Bfl-1, cellular inhibitor of apoptosis (c-IAP), X chromosome–linked inhibitor of apoptosis (XIAP), TRAF1, and TRAF2 (2
). FLIP inhibits apoptosis by interfering with caspase-8 activation (30
). c-IAP and XIAP directly bind and inhibit effector caspases, acting downstream of initiator caspases.
ROS and the NF-κB–JNK cross-talk.
The role of JNK in PCD has been controversial, because it has both survival and death-enhancing effects. The clearest evidence for JNK as regulator of PCD comes from analysis of knockout mice: JNK1- or JNK2-deficient mice are relatively resistant to induction of fulminant hepatitis in response to concanavalin A, a pathology that depends on activation of TNFR1 and other DRs (10
The ratio between JNK and NF-κB activities controls cell survival or death, not only in response to TNFR1 but also in response to other death stimuli (35
). Whereas TNF-α leads to transient JNK activation in WT cells, it leads to prolonged JNK activation in cells that cannot activate NF-κB (9
). The pro-survival activity of NF-κB depends on this ability to prevent prolonged JNK activation (9
). Prolonged JNK activation following concanavalin A administration was also seen in mice lacking IKKβ in liver cells, resulting in massive TNFR1-dependent hepatocyte death (10
). In the liver, however, TNFR1 and JNK signaling is also required for regeneration or compensatory hepatocyte proliferation following partial hepatectomy or chemically induced injury (41
). Thus, NF-κB may be a critical regulator of cell survival and death through its ability to control the duration of JNK activation (Figure ).
Prolonged JNK activation in NF-κB–deficient cells implies that NF-κB induces expression of JNK inhibitors. Such a function was proposed for GADD45β (43
) and XIAP (39
). However, analysis of GADD45β- or XIAP-deficient fibroblasts failed to reveal changes in the kinetics of JNK activation (31
), suggesting that NF-κB regulates JNK activation through a different mediator.
ROS are likely the mediators (40
). ROS, including H2
, and HO•
radicals, are generated through many enzymatic pathways, but their major source is leakage of electrons from the mitochondrial respiratory chain (45
). ROS activate kinases through oxidation of kinase-interacting molecules. For instance, ROS activate tyrosine kinases by inactivating protein tyrosine phosphatases through oxidation of a highly reactive cysteine residue at their catalytic site (46
). Similarly, ROS mediate the NF-κB–JNK cross-talk through their ability to inactivate various MAPK phosphatases (MKPs) involved in JNK inactivation (45
TNF-α induces ROS accumulation in many cell types, and these ROS are important mediators of PCD (22
). TNF-α–induced ROS accumulation is seen in NF-κB–deficient cells, but not in NF-κB–competent cells (9
). Treatment of cells with the antioxidant butylated hydroxyanisole (BHA) has no effect on transient JNK activation triggered by TNF-α, but it suppresses prolonged JNK activation and PCD in TNF-α–treated NF-κB–deficient cells (9
). This protective effect is due to BHA’s ability to prevent oxidation of MKPs, ensuring transient JNK activation (9
). Expression of dominant-negative mutants of MKPs leads to prolonged JNK activation and allows killing by TNF-α of NF-κB–competent cells, which otherwise are TNF-α–resistant (9
The loss of NF-κB activity results in ROS accumulation because NF-κB induces expression of several antioxidant genes such as manganese superoxide dismutase (MnSOD), ferritin heavy chain (FHC), glutathione-S
-transferase, and metallothionein (50
). Overexpression of mitochondrial MnSOD protects cells from TNF-α–induced cytotoxicity (9
). Overexpression of FHC also suppresses TNF-α–induced PCD along with attenuation of prolonged JNK activation (52
). Another interesting observation is that TNF-α induces expression of a number of cytochrome p450 family members, such as CYP1B1, that enhance ROS production in NF-κB–deficient fibroblasts (37
). Taken together, these findings show that NF-κB protects cells from oxidative stress by activating expression of various antioxidant systems, whose failure enhances TNF-α–induced PCD.
The mechanism of TNF-α–induced ROS production is unclear. One possible source of ROS is the cytosolic phospholipase A2 (53
). However, several lines of evidence suggest that mitochondria are the main source of ROS during TNF-α–induced PCD (22
). Compared with our understanding of DR-induced caspase activation, the mechanism of DR-induced ROS production is obscure. TNF-α does not induce ROS accumulation and programmed necrosis in FADD- or RIP1-deficient cells, indicating essential roles for FADD and RIP1 (54
). In contrast to the established function of RIP1 as an adaptor molecule in NF-κB activation, its kinase activity is necessary for Fas-induced necrosis, which mostly occurs in caspase-8–deficient cells (55
). Interestingly, inhibition of caspases potentiates ROS accumulation and cell death (53
). Although DRs can induce ROS accumulation without caspase activation in certain cell types, caspase activation can also lead to mitochondrial damage and ROS accumulation (33
). Thus, caspase-dependent and -independent mechanisms might be involved in ROS accumulation.
JNK activation may also enhance ROS accumulation, potentiating TNF-α–stimulated necrosis (60
). Although the mechanism by which JNK potentiates ROS accumulation is unclear, a positive feedback loop between ROS accumulation and JNK activation may exist (Figure ). Such a loop may also involve caspase activation. Although caspases are not involved in TNF-α–induced prolonged JNK activation in NF-κB–deficient cells (40
), caspase-mediated cleavage of upstream MAP3Ks may cause constitutive JNK activation (61
). JNK activation also contributes to caspase activation, an effect mediated through enhanced cytochrome c
release, during UV-induced apoptosis (62
). Alternatively, JNK causes caspase activation through jBid formation during TNFR1 signaling (63
). Importantly, NF-κB suppresses all of these amplification loops by inducing expression of caspase inhibitors, Bcl-2 family members, and antioxidants (Figure ). Interestingly, negative feedback loops exist between NF-κB and various death-promoting proteins. Caspase-mediated cleavage of RelA and IKKβ can prevent NF-κB activation (64
). Caspases can also cleave IκB to generate a degradation-resistant NF-κB inhibitor (66
). Oxidation of a cysteine residue in the RHD of RelA prevents its binding to DNA (67
), whereas oxidation of another cysteine within the activation loop of IKKβ interferes with its activation (68
). It is unlikely, however, that all of these regulatory loops and modifications take place simultaneously, and a major challenge for the future is to sort out the events that do take place during different physiological and pathophysiological conditions. It is possible to use some of these regulatory loops in designing drugs and therapeutic strategies to kill cancer cells. Fas and TNF-α can induce both apoptosis and necrosis, and so do anticancer drugs. In L929 cells, for instance, TNF-α triggers mostly necrosis, whereas Fas can induce necrosis only when the apoptotic pathway is suppressed (69
). FADD and RIP play central roles in controlling the choice between the 2 death pathways (70
). NF-κB activation also inhibits programmed necrosis, in addition to its role in prevention of apoptosis.