Reactive Oxygen Species (ROS), Troublemakers between Nuclear Factor-κB (NF-κB) and c-Jun NH₂-terminal Kinase (JNK)

INTRODUCTION

There are several levels of cross-talk between nuclear factor κB (NF-κB) and c-Jun NH₂-terminal kinase (JNK; Ref. 1 ). Both NF-κB and JNK share several common upstream signaling molecules, such as interleukin 1 receptor-associated kinase (2 , 3) , mitogen-activated protein/extracellular signal-regulated kinase kinase kinase 1 (4) , protein kinase Cζ (5 , 6) , transforming growth factor-β activated kinase (7) , and Act1/CIKS (8 , 9) , for their activation. In addition, NF-κB-induced tumor necrosis factor (TNF) α, interleukin 1 (10) , Fas (11) , and/or Fas ligand (12) are considered to be critical for JNK activation under certain circumstances (13, 14, 15) , whereas JNK has been shown to be able to regulate NF-κB by inducing β-TrCP, an essential component for inhibitor of NF-κB (IκB) degradation (16) . Furthermore, a collaborative or synergistic action between NF-κB and JNK has been demonstrated on the transcriptional induction of a number of genes responsible for inflammation, apoptosis, and tumorigenesis (17 , 18) . Thus, it is not surprising to assume that NF-κB and JNK are accomplices with each other during a number of pathological processes.

However, several recent studies have revealed a novel relationship between NF-κB and JNK, i.e., NF-κB antagonizes JNK (19) . We have first shown that NF-κB inhibition by expression of a kinase-mutated IκB kinase (IKK) β enhanced and prolonged JNK activation by arsenic, a classic stress inducer, in human bronchial epithelial cells (19) . This enhancement of JNK activation appears to be responsible for the increased induction of growth arrest- and DNA damage-inducible protein 45α by arsenic. By a similar approach, Javelaud and Besancon (20) demonstrated later that inhibition of NF-κB activation as a consequence of the overexpression of a degradation-resistant form of IκBα(A32/36) prolonged TNFα-induced JNK activation in Ewing sarcoma cells. This prolonged JNK activation is correlated with the increased TNFα-induced cell apoptosis. Conversely, hyperactivation of NF-κB as a result of Par4 deficiency impaired JNK activation (21) .

A fascinating question is why inhibition of NF-κB, a transcription factor that is frequently accused as a co-conspirator with JNK under many circumstances, results in an amplification of JNK activation. Two intriguing answers to this question were provided by Tang et al. (22) and De Smaele et al. (23) , respectively, later in 2001. Using ikkβ^−/− or rela^−/− mouse embryo fibroblasts, results from Tang et al. (22) suggested that the enhanced TNFα-induced JNK activation is possibly due to a decrease of XIAP, of which the gene transcription is regulated by NF-κB. In other words, XIAP is an endogenous inhibitor for JNK activation. At about the same time, De Smaele et al. (23) suggested that deficiency of NF-κB in rela^−/− cells caused impairment of TNFα-induced GADD45β expression, leading to a sustained JNK activation. However, there are some reservations concerning these findings among many experts. First, earlier studies indicated that both GADD45β and XIAP are capable of activating, rather than inhibiting, JNK through up-regulation of upstream kinases (24 , 25) . Second, genetic disruption of gadd45β or xiap has no effect on signal-induced JNK activation (26 , 27) . Third, recent studies suggested that the antiapoptotic function of XIAP requires JNK activation (28) . Lastly, the T-cell receptor-mediated activation of JNK, p38, and extracellular signal-regulated kinase was substantially suppressed in gadd45β^−/− CD4⁺ T cells (29) . Therefore, it appears that there must be mechanisms other than the decrease of XIAP or GADD45β for the potentiated JNK activation in cells where NF-κB is inhibited.

One factor worth consideration is oxidative stress induced by increased generation of reactive oxygen species (ROS). ROS have been viewed previously as general messengers for signal-induced NF-κB activation (30, 31, 32) . However, there is no shortage of dissenters who believe that ROS are not activators, but rather, inhibitors for NF-κB (33, 34, 35) . Recent evidence supports the notion that ROS themselves are not activators for NF-κB (36) . At the activity level, ROS may oxidize NF-κB subunits and, thus, impair the DNA binding and transcriptional activities of NF-κB (37 , 38) . At the activation level, the ubiquitination and degradation of NF-κB inhibitor, IκBα, is dependent on the kinase activity of IKK complexes. Structural comparison of the IKK kinase domain with the corresponding domain of other related kinases, including JNK, p38, phosphoinositide-dependent protein kinase 1, Casein kinase II, and mitogen-activated protein/extracellular signal-regulated kinase kinase kinase 1, revealed that only IKKα and IKKβ contain a redox sensitive cysteine residue (cys179) in this domain (39 , 40) . Previous studies by Kapahi et al. (41) and Rossi et al. (39) indicated that the cys179 in the T-loop of IKKβ could be directly modified by arsenite and 15-deoxy-12.14-prostaglandin J2, respectively. Accordingly, this residue may also be sensitive to oxidative modification by ROS, which could explain the observed oxidative inactivation of IKKβ kinase activity in mouse alveolar epithelial cells exposed to ROS (42) . In contrast, ROS are potent activators for JNK by oxidative inactivation of the endogenous JNK inhibitors, such as JNK phosphatases and glutathione S-transferase π (43 , 44) . Thus, ROS may act as unfair brokers and troublemakers between NF-κB and JNK by inhibiting one but promoting another, creating a new form of cross-talk between these two important stress-responsive systems.

Evidence to support the view that ROS mediate unbalanced cross-talk between NF-κB and JNK was provided recently by Sakon et al. (45) and Chen et al. (46) . Both groups demonstrated independently that the increased accumulation of ROS, mainly H₂O₂, is responsible for the prolonged JNK activation in TNFα-stimulated rela^−/− cells and arsenic-challenged ikkβ^−/− cells, respectively. Strikingly, in addition to the remarkable enhancement of JNK activation, both groups also demonstrated a prolongation of extracellular signal-regulated kinase and p38 activation, although the extent of this prolongation was diminished compared with that of JNK in rela^−/− cells or ikkβ^−/− cells in response to stress. Compared with wild-type cells, significant levels of H₂O₂ were detected in ikkβ^−/− cells even before treatment with arsenic. Addition of arsenic increased H₂O₂ further (46) . In rela^−/− cells where the basal level of H₂O₂ was similar to that in wild-type cells, a dramatic accumulation of H₂O₂ was observed after TNFα administration (45) . Antioxidants, NAC and BHA substantially reduced JNK activation induced by arsenic and TNFα, respectively.

On the basis of the earlier findings by Tang et al. (22) and De Smaele et al. (23) , the question to be asked is does this ROS-dependent prolongation of JNK activation have some connections with the diminishment of GADD45β or XIAP in rela^−/− or ikkβ^−/− cells. The answer appears not, at least in the experiment of ectopic expression. Transfection of either GADD45β or XIAP showed no effect on both signal-induced ROS accumulation and prolonged JNK activation (45) . Furthermore, reverse transcription-PCR or DNA microarray revealed no appreciable difference of GADD45β or XIAP mRNA between wild-type and ikkβ^−/− cells (46) .

The implication of ROS in stress-induced JNK activation in a situation when NF-κB is inhibited is of particular interest in light of their previous known role in promoting cell apoptosis and carcinogenesis. Now the question is why NF-κB impairment, by either ikkβ gene knockout or rela gene disruption, causes ROS accumulation. A good starting point would be to determine whether NF-κB inhibition causes decreased expression of endogenous antioxidant enzymes, such as SODs, GPx, and catalase, which are capable of dismutating O₂⁻ or eliminating H₂O₂. It is interesting to note that there is no deficiency or decrease in the expression of these antioxidant enzymes in rela^−/− and ikkβ^−/− cells, respectively (45 , 46) . In contrast, DNA microarray and reverse transcription-PCR analyses suggested a marginal increase of SOD1 expression in ikkβ^−/− cells (46) . This indicates that at least on the expression levels, there should be enough intracellular antioxidant enzymes, no matter whether the NF-κB is inhibited or not. For the nonprotein antioxidant systems, Sakon et al. (45) found an appreciable depletion of reduced glutathione and NADPH, two nonenzymatic endogenous antioxidants, in rela^−/− cells and assumed that this is the cause of ROS accumulation.

Another possibility needs to be examined is the enhanced capacity of ROS formation in rela^−/− or ikkβ^−/− cells. A number of enzymes with peroxidase activity, such as cyclooxygenases, lipoxygenases, and cytochromes p450 family members, are pivotal for the generation of ROS in the cytosol (47, 48, 49) . Evidence available to support this possibility is the finding of increased expression of p450 family member cyp1b1 in ikkβ^−/− cells compared with that in wild-type cells (46) . Similarly, in human bronchial epithelial cells, transfection of a kinase-mutated IKKβ increases expression of several other p450 members. 3 These findings are not surprising, however, because a negative influence of NF-κB on aryl hydrocarbon receptor has been established previously (50 , 51) . The transcription of many p450 genes is largely dependent on the activation of aryl hydrocarbon receptor, a nuclear protein forming a heterodimer with a related helix-loop-helix protein, aryl hydrocarbon receptor nuclear translocator (52) . By recognizing and binding to the xenobiotic-responsive elements in the promoter or enhancer region of p450 family genes, this complex up-regulates the transcription of p450 genes. Both mouse and human cyp1b1 genes exhibit an exceptional sequence characteristic, containing a number of xenobiotic-responsive elements or xenobiotic-responsive element-like elements in their 5′-flanking region (53) . Therefore, gene deficiency of rela or ikkβ will ablate the antagonism of NF-κB on aryl hydrocarbon receptor and consequently amplify transcription of the p450 genes. Nevertheless, whereas the activity of xenobiotic metabolism of p450 family members has been well established, the contribution of these members to ROS generation remains largely unaddressed. A number of p450 isoforms are thought to be capable of catalyzing ROS generation during their metabolism, and oxidation of endobiotic and xenobiotic chemicals or hormones (48) .

Cell apoptosis appears to be an unavoidable topic when both NF-κB and JNK are discussed. It has been generally accepted that deficiency or inhibition of NF-κB promotes apoptosis in many types of cells (54 , 55) . The pro- or antiapoptotic effect of JNK activation, however, is an extensively debatable issue. The reports by Tang et al. (22) and De Smaele et al. (23) suggested a proapoptotic role of sustained JNK activation in the cells where NF-κB is inhibited. It was also noted that activation of JNK caused caspase 8-independent cleavage of Bid and subsequent release of Smac/DIABLO from mitochondria, leading to the disruption of TRAF2-cIAP1 complex and cell apoptosis (56) . Despite these findings, several recent studies suggest that activation of JNK is actually protective for the cells from TNFα-induced apoptosis (57 , 58) . The notion that JNK protects cell from apoptosis is largely based on the pioneering study by Reuther-Madrid et al. (57) who demonstrated that inhibition of JNK promoted TNFα-induced apoptosis in the cells that lack functional NF-κB. Additional evidence indicating antiapoptotic role of JNK was from the gene knockout studies, which suggested increased apoptosis in the forebrain and/or hindbrain in Jnk1^−/− Jnk2^−/− mouse embryos (59 , 60) . In support of this view, a recent study by Lamb et al. (58) indicates that the antiapoptotic function of JNK is mediated by JunD. JunD may be an important factor contributing to the up-regulation of cIAP2, an antiapoptotic protein capable of suppressing apoptotic proteases, caspases (55 , 58) . The latest evidence indicating antiapoptosis of JNK was provided by Zhang et al. (61) who demonstrated that the enhancement of JNK activation due to NF-κB inhibition in epidemis fails to trigger cell apoptosis but instead increases cell proliferation. It appears to be difficult currently to answer the question why JNK mediates two opposite effects in cell apoptosis.

The finding that oxidative stress is a major player for the prolonged JNK activation in ikkβ^−/− or rela^−/− cells adds an important piece to the puzzle of ROS on cellular signaling pathways. In normal cells, the generation of ROS is limited. Inhibition of NF-κB possibly resembles the opening of Pandora’s box, leading to the flowing out of ROS and inevitably, initiation of a stress response by oxidative damage of intracellular molecules. A self-amplification loop of ROS is formed, predominantly, due to the oxidative inactivation of NF-κB, which causes sustained JNK activation (Fig. 1) ⇓ . Nevertheless, this finding also poses several questions. First and foremost, blockage of NF-κB activation has been considered as a rational strategy in the control of inflammatory diseases. If accumulation of ROS occurs as a result of NF-κB inhibition, how can one avoid the adverse side effect of oxidative stress? Would this accumulation of ROS occur only in ikkβ or rela gene knockout cells where NF-κB activation is almost eliminated, but not in the cells where the inhibition of NF-κB is mild? A second question concerns the role of ROS in NF-κB inhibition-induced cell death. Apoptotic response of ikkβ^−/− or rela^−/− cells was attributed previously to the deficiency of antiapoptotic protein expression in these cells (55 , 62) . If ROS accumulation is the main reason for the sustained activation of JNK, any measure that reduces cellular oxidative stress should be effective in protection of the cells from either apoptotic or necrotic death. Such protection, however, is mild as shown in the report by Sakon et al. (45) . Lastly, CYP1B1 has been identified as a critical enzyme contributing to the development of breast cancer through the 4-hydroxylation of estrogen (63) . Then the question arises, is NF-κB activation beneficial, rather than harmful as assumed previously (62 , 64) , for the control of cancer development? More detailed and extensive study is required to answer these difficult questions.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Inhibition of nuclear factor-κB (NF-κB) initiates an amplification loop of reactive oxygen species (ROS). In the cells where ikkβ or rela gene is deficient, the activation or activity of NF-κB is inhibited. This inhibition blocks the antagonism of NF-κB on aryl hydrocarbon receptor, leading to enhancement of aryl hydrocarbon receptor function and increased expression of p450 family members. Elevated expression of p450 members, such as cyp1b1, has a positive influence on the generation of ROS, which inhibit NF-κB further by oxidative modification of inhibitor of NF-κB kinase and NF-κB p50 subunit. The accumulated ROS cause sustained c-Jun NH₂-terminal kinase (JNK) activation leading to either cell death (block arrow) or cell survival (dashed arrow). The inset shows a second ROS amplification loop, ROS-JNK-BAD-mitochondria-ROS. pBAD, phosphorylated BAD.