NF-κB meets ROS: an ‘iron-ic’ encounter

Since becoming abundant in the atmosphere approximately 2.3 billion years ago, oxygen has been a defining element for life on our planet. One needs not be a biologist to know the importance of oxygen for sustaining life – termed in fact ‘aerobic’ life. Our body is built around the need to maximize exploitation of oxygen for the production of energy. The respiratory and cardiovascular systems are exemplary illustrations of this need. Like all good things, however, oxygen can also be extremely harmful. Its original accumulation on Earth caused the extinction of most existing life forms, defenseless against oxidative damage. Even a layperson is aware of this potential toxicity of oxygen. Indeed, nowadays antioxidants are among the most heavily advertised dietary supplements on the market. Yet, it would surprise most to know that the potent reactivity of oxygen and its products – so-called reactive oxygen species (ROS) – is purposefully used by nature to transduce signals that actively trigger cell suicide or programmed cell death (PCD), as well as other biological responses.

One pathway that seems to fully exploit this reactivity of ROS for inflicting cell death is that initiated by TNFα engagement of TNF-R1, a pathway that plays a central role in immunity, inflammation, cell growth, cell death and differentiation.^{1, 2, 3} This pathway is also crucial for pathogenesis of human diseases such as cancer and chronic inflammatory conditions, including rheumatoid arthritis (RA) and inflammatory bowel disease (IBD).^{1, 3, 4} Not surprisingly, it has been the subject of intense investigation for over one century.¹ TNFα-induced killing is antagonized by activation of NF-κB-family transcription factors^{3, 4} – which act as master coordinators of immune and inflammatory responses.⁴ The prosurvival activity of NF-κB is also crucial for lymphocyte development, tumorigenesis and cancer chemoresistance.^{4, 5} In recent years, remarkable progress has been made in our understanding of the mechanisms governing TNFα-induced death and NF-κB-mediated survival.^{3, 4} As it turns out, ROS have now taken center stage in the intricate multitude of players that control cell fate downstream of TNF-R1, as they appear to be at an obligatory crossroads of the opposing pathways for life and death elicited by stimulation of this receptor. Indeed, now there is hope that this new understanding of TNF-R-induced pathways may lead to the development of new approaches for treatment of widespread human diseases.

Death-inducing JNK signaling downstream of TNF-Rs

The ability to trigger PCD is a signature characteristic of TNF-R1, a prototypic example of the so-called ‘death receptors.’^{1, 2} We and others have shown that induction of PCD by TNFα depends on activation of the c-Jun-N-terminal kinase (JNK) cascade,^{6, 7} a major mitogen-activated protein kinase (MAPK) cascade.⁸ Although some doubts were cast initially, it is now widely accepted that JNK plays an obligatory role in TNF-R-induced death signaling.³ Paradoxically, however, despite the well-documented ability of TNF-Rs to induce PCD,^{1, 2} cells do not normally die following stimulation of this receptor – owing to the prosurvival effects of activation of NF-κB.^{3, 4} Accordingly, TNFα-mediated cytotoxicity can be promptly unveiled by the inhibition of NF-κB.^{3, 4} Notably, suppression of the JNK cascade by various means effectively rescues NF-κB-deficient cells from this cytotoxicity.^{6, 7} It is only prolonged activation of JNK, however, that appears to be involved in induction of cell death.³ In the presence of NF-κB, TNFα only causes transient elevation of JNK activity,^{6, 7} which is why this elevation normally occurs without significant PCD. Conversely, preventing nuclear translocation of NF-κB impairs the shutdown of JNK signaling,^{6, 7} thereby revealing a subsequent, sustained phase of JNK activation by TNF-Rs, one that is involved in PCD signaling. This NF-κB-mediated control of sustained activation of the JNK cascade is in fact a pivotal protective mechanism against TNFα-induced cytotoxicity.^{6, 7} Of note, some reports have recently suggested that activation of JNK by TNFα might actually exert a protective effect against TNF-R-mediated killing.^{9, 10} Another study using JNK null fibroblasts indicates that, while promoting necrosis-like PCD, JNK activation downstream of TNF-Rs suppresses apoptosis.¹¹ The bases for these apparent discrepancies are not clear, but seem consistent with the notion that the actual outcome of activation of JNK ultimately depends upon the biological context and tissue in which this activation occurs.⁸

The exact mechanism(s) by which JNK promotes PCD remains unclear. In some systems, JNK-induced death signaling involves modulation of gene expression, achieved in part through phosphorylation of the c-Jun transcription factor.⁸ During challenge with TNFα or stress stimuli, however, the main targets of ROS-triggered JNK-mediated cytotoxicity appear to be constitutively present in the cell.^{8, 12} A recent report offers some mechanistic insights into the bases for the role of JNK in TNFα-induced PCD.¹² This report suggests that activation of JNK by TNF-Rs triggers death by causing proteolytic processing of the ‘BH-3-only’ protein Bid,² into jBid¹² – a product distinct from caspase-8-generated, tBid.^{2, 12} jBid then translocates to mitochondria to trigger selective release into cytosol of the apoptogenic factor Smac/Diablo,^{2, 12} which in turn binds to and inactivates the caspase inhibitor c-IAP1,^{2, 12} thereby activating caspase-8 and ultimately PCD¹² (Figure 1). Exactly how JNK promotes the formation of jBid, however, is not known. Moreover, it is not clear whether Bid is the only target of proapoptotic JNK signaling. In fact, killing caused by overexpressed MKK7-JNK proteins, which mimic constitutively active JNK, was shown to require Bax-like factors of the Bcl-2 group, but to occur normally in Bid^−/− cells,¹³ indicating that proteins other than Bid participate in JNK-mediated killing in this system. Additionally, ROS-induced JNK activity can trigger caspase-independent, necrosis-like PCD;¹¹ but how JNK mediates TNFα-induced necrosis remains unknown.

Figure 1

Life and death pathways downstream of TNF-R1. TNF-α engagement of TNF-R1 initiates accumulation of ROS, which promotes cell death through activation of the JNK MAPK signaling. This occurs via at least two mechanisms: inactivation of MKPs and activation of MAPKKKs. Death signaling triggered by ROS-mediated activation of JNK is antagonized by NF-κB, which upregulated the expression of protective genes such as FHC and Mn-SOD

Involvement of ROS in the induction of JNK and PCD downstream of TNF-Rs

New studies have shown that TNF-R-induced killing also relies upon an accumulation of ROS.^{14, 15, 16} Indeed, antioxidant treatment can afford near-complete protection against killing caused by stimulation of TNF-Rs in various cell types.^{14, 15} Interestingly, ROS-induced cytotoxicity is mediated through activation of the JNK pathway,^{14, 16} and sustained activation of this pathway by TNFα depends in fact on ROS .^{14, 15, 16} Thus, the activities of ROS and JNK seem to participate in the same death-inducing mechanism triggered by TNF-Rs. Recent studies now provide important clues to how ROS promote activation of the JNK pathway. A study by Karin and colleagues has identified JNK phosphatases of the MKP group,⁸ such as MKP-1, -3, -5 and -7, as critical molecular targets of ROS in PCD signaling downstream of TNF-Rs.¹⁶ ROS-mediated inactivation of MKPs – key factors for controlling activity of MAPKs⁸ – appears to involve oxidation of a critical cysteine residue in the catalytic domain of these enzymes to sulfenic acid.¹⁶ Underscoring the physiological relevance of this inhibitory mechanism, this catalytic cysteine exhibits a lower pK_a than other cysteines of MKPs, and so is naturally more prone to oxidation than these other residues.¹⁶ ROS-dependent loss of MKP function causes persistent activation of JNK by TNFα and, ultimately, PCD via both the necrotic and apoptotic pathways.¹⁶ Accordingly, antioxidant agents effectively block sustained activation of JNK and both forms of PCD downstream of TNF-Rs.^{14, 15, 16} It should be cautioned, though, that the conclusions of this study were reached mainly using pharmacological and dominant-negative inhibitors,¹⁶ and so still await validation through genetic means.

Additionally, it appears unlikely that inactivation of MKP phosphatases is the only mechanism by which ROS promote JNK signaling. Previous studies by Ichijo and colleagues indicate, in fact, that ROS also trigger activation of ASK1/MEKK5,¹⁷ a TRAF2-binding MAPK kinase kinase (MAPKKK) required for sustained induction of JNK and PCD downstream of TNF-R1.^{8, 17} Notably, this model is supported by knockout data, because fibroblasts lacking ASK1 exhibit a severe defect in activation of JNK and PCD in response to TNFα.¹⁷ As with the inactivation of MKPs,¹⁶ ROS-mediated induction of ASK1 seems to involve inhibition of the redox-sensing protein, thioredoxin.¹⁷ Thus, the magnitude and duration of TNFα-induced JNK activity is likely controlled by a balance between the actions of inducing kinases and inhibiting phosphatases. It is plausible that the relative importance of these mechanisms in induction of the JNK cascade by TNF-Rs is dictated by biological context. Of note, the precise composition of MAPKKKs involved in acute and persistent activation of JNK by TNF-Rs is still unknown.⁸ Identifying these MAPKKKs will most certainly help clarify the bases for the ROS-mediated regulation of the JNK cascade.

Key questions in TNF-R1-induced ROS signaling

An important open issue concerns the causal relationship between ROS and JNK. Whereas most studies seem to agree that ROS are upstream of JNK in the TNF-R-induced signaling pathway,^{14, 15, 16, 17} some evidence suggests that there might be a more complex, reciprocal interplay between ROS and JNK in this pathway.¹¹ It was in fact proposed that in the TNF-R1-triggered pathway for necrosis, ROS might actually lie downstream (rather than upstream) of activation of JNK.¹¹ In NF-κB-deficient fibroblasts, the dramatic accumulation of ROS observed following the triggering of TNF-Rs was abrogated by the introduction of the JNK1^−/− and JNK2^−/− mutations.¹¹ Thus, ROS and JNK may participate in a mutual amplification mechanism that is ultimately responsible for inflicting cell death. It should be cautioned, though, that while a dynamic relationship between ROS and JNK is likely to exist, the antioxidant effects of JNK deletion in this system might be owed, at least in part, to an inhibition of cell death (see below). Interestingly, however, the study suggests that the relative positioning of ROS with respect to JNK signaling might dictate which form of PCD is ultimately elicited in response to treatment with TNFα (i.e. necrotic versus apoptotic)¹¹ (see also Pham et al.,¹⁴ Sakon et al.,¹⁵ Kamata et al.,¹⁶). Moreover, irrespective of the nature of the PCD mechanism that is triggered by TNFα, the study shows that activation of NF-κB is capable of blocking TNFα-induced cytotoxicity.¹¹

Another major future challenge will be identifying the mechanisms by which TNF-Rs induce formation of ROS. It is usually assumed that TNFα-induced ROS originate in mitochondria – the main source of oxygen radicals in eukaryotes.¹⁸ However, most prior studies dealing with this issue have measured TNF-R-triggered ROS production at relatively late times (i.e. several hours),^{14, 15, 18} and so these measurements were likely affected by the oxidative burst that invariably follows mitochondrial outer membrane depolarization,² an event often signifying that cells have already committed to die.² Thus, whether mitochondrial generation of ROS is a cause or an effect of cell death remains unclear. Indeed, previous observations that ROS formation is not triggered by stimulation of IL-1β-R^{15, 16} and that induction of these species by TNF-R1 is blocked by combined deletion of JNK1 and JNK2¹¹ might be a mere reflection of a lack of cell death in these systems. The notion that mitochondria are the primary source of ROS produced downstream of TNF-R1 is also at odds with the weak protective activity of the overexpression of the mitochondrial ROS scavenger, Mn²⁺ superoxide dismutase (Mn-SOD)^{14, 15} (see below). Extra-mitochondrial sources of ROS have in fact been identified and proposed to play a role in induction of PCD by TNF-R1.¹⁹ Ultimately, the issue of the origin of TNFα-induced ROS will need to be addressed by the use of genetic tools and highly specific and sensitive methods for detection of early formation of ROS and discrimination between individual oxygen species.

The targeting of ROS by NF-κB

The importance of ROS formation in TNF-R signaling to PCD is underscored by the recent discovery that preventing this formation is a key protective function ascribed to NF-κB.^{14, 15} We and others have shown that blocking NF-κB activation leads to exaggerated accumulation of ROS following stimulation of TNF-Rs and that suppression of this accumulation virtually abrogates TNF-R-induced killing in NF-κB-deficient cells.^{14, 15} This antioxidant action represents in fact a pivotal mechanism by which NF-κB restrains sustained activation of the JNK pathway by TNFα.^{14, 15} The bases for the NF-κB-mediated control of ROS production are now beginning to be elucidated (Figure 1). Using a microarray-based screen of cDNA libraries, we have identified ferritin heavy chain (FHC) as a key mediator of the antioxidant and antiapoptotic activities of NF-κB downstream of TNF-Rs.¹⁴ Together with light chains (FLC), FHC assembles into Ferritin multiprotein complexes²⁰ – the primary iron storage devices in cells.²⁰ The FHC-mediated inhibition of PCD induced by TNFα involves suppressing the induction of ROS and thereby sustained activation of JNK signaling.¹⁴ This action of FHC depends on sequestration of free iron¹⁴ – a metal that catalyzes generation of ROS in mitochondria and through Fenton and Haber-Weiss reactions,^{18, 20} resulting in formation of highly reactive hydroxyl (•OH) radicals.^{18, 20} Interestingly, knockdown experiments suggest that, in some tissues, FHC might represent a dominant component of the protective mechanism activated by NF-κB for inhibiting TNF-R-induced PCD.¹⁴ Notably, FHC is one of several acute-phase proteins, upregulated in the liver during the organismal response to stress, injury and infection,²⁰ and so, might mediate a systemic cytoprotective mechanism activated by NF-κB during chronic inflammation, when the potential exists for extensive ROS-mediated, tissue damage.^{1, 2, 3} Indeed, induction of FHC might also contribute to some of the important functions of NF-κB in oncogenesis, tumor progression and cancer resistance to anticancer therapy.^{3, 4, 5}

Studies by other laboratories had previously shown that the protective activity of NF-κB against ROS-inflicted damage is also mediated by upregulation of the antioxidant enzyme, Mn-SOD⁴ – which catalyzes dismutation of superoxide anion (•O₂−) into hydrogen peroxide (H₂O₂).^{4, 18} Interestingly, whereas in some contexts Mn-SOD is likely to be an important element for NF-κB-mediated suppression of PCD,⁴ its ectopic expression seems to afford little or no protection against TNFα-induced cytotoxicity in NF-κB-deficient cells.^{14, 15} Hence, while required for controlling TNFα-prompted changes in redox status, Mn-SOD alone does not appear to be sufficient to mediate the ROS-inhibiting and protective functions of NF-κB. For this control to be effective, in fact Mn-SOD might have to cooperate with FHC. It is plausible that whereas induction of Mn-SOD promotes dismutation of •O₂− into H₂O₂, FHC-mediated iron depletion might be required to enable disposal of H₂O₂ by catalases and peroxidases.^{14, 18}

Bases for the NF-κB-mediated control of the JNK cascade

Combined upregulation of FHC and Mn-SOD provides a basis for the antioxidant activity of NF-κB. It also provides an indirect link between the NF-κB and JNK pathways (Figure 1). There are, however, also direct means by which NF-κB controls proapoptotic JNK signaling. A handful of downstream targets of NF-κB was in fact shown to encode bona fide inhibitors of the JNK cascade.³ These include: the c-IAP-like caspase inhibitor, XIAP,⁷ the zinc-finger protein A20³, and the Gadd45-family member, Gadd45β/Myd118.⁶ Whereas this latter factor was shown to associate physically with and inhibit MKK7/JNKK2,³ the dominant JNK kinase induced by TNFα,^{3, 8} the mechanisms for XIAP- and A20-mediated inhibition of the JNK cascade remain unknown. Moreover, genetic evidence for a role in the NF-κB-mediated containment of TNFα-induced JNK signaling has so far been obtained for A20 and Gadd45β.³ XIAP^−/− mice were instead reported to have no apparent phenotype in control of this signaling and PCD in response to TNFα.⁴ Hence, NF-κB appears to halt the JNK cascade through two distinct mechanisms, directly, via induction of Gadd45β, A20 and XIAP, indirectly, via induction of FHC and Mn-SOD, which blunt production of ROS (Figure 1). Possibly, this redundancy reflected in the use of multiple effectors may serve to ensure the effective shutdown of JNK signaling. Additionally, the biological relevance of these downstream effectors to the prosurvival function of NF-κB is likely to differ depending upon tissue and biological context,³ and so, the flexibility of this program might enable an organism to tailor the prosurvival response according to specific contexts and needs.

Relevance of the NF-κB-mediated control of ROS and JNK signaling to human diseases and future prospective

Disturbances of the NF-κB-mediated suppression of TNF-R-induced PCD play a critical role in human diseases.^{3, 4, 5} The mutual positive regulation of TNFα and NF-κB is critical for perpetuating chronic inflammation in conditions such as RA and IBD,^{3, 4} and indeed, inhibitors of NF-κB and neutralizing anti-TNFα antibodies are routinely used for treatment of these conditions.^{3, 4} NF-κB-targeting agents such as proteasome inhibitors and glucocorticoids are also used to treat human malignancies.^{3, 4} The clinical utility of NF-κB inhibitors, however, is greatly limited by their severe side effects, including immunosuppressive effects. A radically new approach to therapy is therefore needed. An ideal drug would be one that enables targeting specific downstream effectors of NF-κB, rather than NF-κB itself. Thus, a capital future challenge includes determining the exact mechanisms by which NF-κB blunts ROS accumulation and JNK signaling in specific patho-physiological contexts. This might lead to the development of new drugs that selectively interrupt the crosstalk between NF-κB and the ROS/JNK pathway in these contexts, and thereby promote PCD selectively in diseased tissues, without harming other important functions of NF-κB such as immune functions.⁴ Indeed, an augmentation of ROS or JNK signaling, achieved through blockade of select NF-κB targets might trigger PCD in self-reactive and proinflammatory cells within sites of inflammation – where there are high concentrations of TNFα. This represents a major therapeutic goal. A recent study reports an important step in this direction.¹⁶ This study shows that, in the liver, ROS-mediated activation of JNK signaling selectively regulates TNF-R-mediated injury induced by concanavalin-A, but not regeneration post-partial hepatectomy,¹⁶ albeit both processes are governed by an integration of activities of TNF-Rs, JNK and NF-κB.^{3, 16} In addition to providing an effective new tool for anti-inflammatory therapy, drugs that target the NF-κB/JNK crosstalk might have beneficial effects for the treatment of cancer. Indeed, there is now evidence that whereas NF-κB promotes survival of certain late-stage tumors and cancer chemo- and radio-resistance,^{4, 5} the activities of JNK and ROS mediate the cytotoxic effects of radiation and certain chemotherapeutic drugs (e.g. topoisomerase inhibitors), and that activators of JNK can act as tumor suppressors.^{3, 8}