Research Article 647

IntroductionTwo different NF-B pathways have been identified that lead totranscription from B sites, and are referred to as the canonical andnoncanonical (or alternative) pathways (Beinke and Ley, 2004;Bonizzi and Karin, 2004; Hayden and Ghosh, 2008; Vallabhapurapuand Karin, 2009). The canonical NF-B pathway is activatedfollowing stimulation with a broad range of stimuli such as TNF,IL-1, lipopolysaccharide (LPS) and genotoxic agents, and has amajor role in the control of innate immunity and inflammation(Beinke and Ley, 2004; Bonizzi and Karin, 2004; Hayden andGhosh, 2008; Vallabhapurapu and Karin, 2009). The canonicalpathway involves phosphorylation of IB by IB kinase (IKK)components (primarily IKK) and subsequent degradation of IB,which typically holds the RelA–p50 heterodimer in the cytoplasm.IB degradation allows NF-B to translocate to the nucleuswhere it binds to B sites, leading to gene transcription (Beinkeand Ley, 2004; Bonizzi and Karin, 2004; Hayden and Ghosh,2008; Vallabhapurapu and Karin, 2009).

The noncanonical NF-B pathway is activated by particularTNF receptor family members that bind to the TNF receptor-associated factors TRAF2 and/or TRAF3, such as LTR, CD40,CD27, CD30, BAFF-R and RANK, which are mainly involved insecondary lymphoid organ development, B cell survival andmaturation and osteoclastogenesis (Beinke and Ley, 2004; Bonizziand Karin, 2004; Hayden and Ghosh, 2008; Vallabhapurapu andKarin, 2009). The noncanonical pathway requires IKK-dependentNF-B2 (p100) processing in the proteasome to generate p52. p52

is associated with RelB and enters the nucleus and binds to Bsites in certain NF-B-responsive genes to activate transcription(Dejardin et al., 2002; Senftleben et al., 2001; Xiao et al., 2001).Processing of p100 requires its phosphorylation by IKK, whichin turn requires phosphorylation by NF-B-inducing kinase (NIK)(Senftleben et al., 2001; Xiao et al., 2004; Xiao et al., 2001). Rapidturnover of NIK prevents constitutive activation of the noncanonicalNF-B pathway in resting cells (Liao et al., 2004). Most currentdata in the literature with regard to stimuli-dependent stabilizationof NIK are consistent with a simple model (Zarnegar et al., 2008;Vallabhapurapu et al., 2008), in which TRAF3 recruits NIK to acomplex containing TRAF2 and inhibitor of apoptosis proteinscIAP1 and/or cIAP2 (cIAP1/2), where NIK ubiquitylation bycIAP1 or cIAP2 promotes its proteasomal degradation. The role ofTRAF2 in this complex is to recruit TRAF3 and the cIAP proteinsthrough direct interactions. This model predicts that the loss ofTRAF2, TRAF3 or both cIAP1 and cIAP2 would lead tononcanonical NF-B activation by the resulting stabilization of theNIK protein. Thus, cells deficient in TRAF2 or TRAF3 have highbasal levels of NIK and p52 (Gardam et al., 2008; Grech et al.,2004; He et al., 2007; He et al., 2006; Vallabhapurapu et al., 2008;Xie et al., 2007), whereas inhibition of cIAP proteins by small-molecule antagonists results in activation of the noncanonical NF-B pathway in a manner that is dependent on NIK stabilization,thereby enhancing B lymphocyte survival and proliferation(Varfolomeev et al., 2007; Vince et al., 2007). High levels of NIKand p52 in multiple myelomas might also be explained by several

SummaryThe current paradigm of noncanonical NF-B signaling suggests that the loss of TRAF2, TRAF3 or cIAP1 and cIAP2 leads tostabilization of NF-B-inducing kinase (NIK) to activate the noncanonical pathway. Although a crucial role of RIP1 in the TNF-induced canonical NF-B pathway has been well established, its involvement in noncanonical activation of NF-B through the TNFR1receptor, is unknown. Here we show that TNF is capable of activating the noncanonical NF-B pathway, but that activation of thispathway is negatively regulated by RIP1. In the absence of RIP1, TNFR1 stimulation leads to activation of the noncanonical NF-Bpathway through TRAF2 degradation, leading to NIK stabilization, IKK phosphorylation and the processing of p100 to generate p52.Thus although RIP1–/– mouse embryonic fibroblasts are sensitive at early time points to cell death induced by TNF, probably as aresult of lack of canonical NF-B activation, the late activation of the noncanonical NF-B pathway protects the remaining cells fromfurther cell death. The TNFR1-dependent noncanonical NF-B activation in RIP1–/– cells suggests that there is functional interplaybetween the two NF-B pathways during TNFR1 signaling, which might regulate the number and kinds of NF-B transcription factorsand thus finely control NF-B-dependent gene transcription.

Key words: TNF, TNFR1, RIP1, TRAF2, Noncanonical NF-B

Accepted 14 October 2010Journal of Cell Science 124, 647-656 © 2011. Published by The Company of Biologists Ltddoi:10.1242/jcs.075770

TNF-induced noncanonical NF-B activation isattenuated by RIP1 through stabilization of TRAF2Joo-Young Kim1, Michael Morgan2, Dong-Gun Kim1, Ju-Yeon Lee1, Lang Bai3, Yong Lin3, Zheng-gang Liu2

and You-Sun Kim1*1Institute for Medical Sciences, Ajou University School of Medicine, Suwon, 443-749, Korea2Cell and Cancer Biology Branch, Center for Cancer Research, National Cancer Institute, National Institutes of Health, 37 Convent Drive,Bethesda, MD 20892, USA3Molecular Biology and Lung Cancer Program, Lovelace Respiratory Research Institute, 2425 Ridgecrest Drive, Southeast, Albuquerque,NM 87108, USA*Author for correspondence (yousunkim@ajou.ac.kr)

Jour

nal o

f Cel

l Sci

ence

different genetic alterations, including deletion of TRAF2 orTRAF3 or bi-allelic deletions of both cIAP1 and cIAP2 (Annunziataet al., 2007; Keats et al., 2007). However, very few receptor-mediated stimuli have been described that cause noncanonicalactivation dependent solely upon degradation of TRAF2 in theabsence of TRAF3 degradation.

Receptor-interacting protein 1 (RIP1) has emerged as an essentialmolecule that is involved in death receptor signaling and cellularstress (Festjens et al., 2007). The C-terminal death domain of RIP1is capable of binding to death receptors with a similar death domainmotif in death receptors, such as TNFR1 and Fas, and the TRAIL(TNF-related apoptosis-inducing ligand) receptors (Chaudhary etal., 1997), and also interacts with other death domain-containingadaptor proteins, such as TRADD (TNFR1-associated DEATHdomain protein) and FADD (FAS-associated death domain protein)(Varfolomeev et al., 1996). As a TRADD- and TNFR1-bindingmolecule, RIP1 is a key effector molecule in the TNF-inducedactivation of the transcription factor NF-B (Kelliher et al., 1998).This is largely because the Lys63-linked polyubiquitylation ofRIP1 by cIAP1/2 (Lin et al., 2010) leads to its interaction withregulatory subunit of the IKK complex, IKK–NEMO, and thusRIP1 is essential for robust TNF activation of IKK and of thecanonical NF-B pathway (Ea et al., 2006; Poyet et al., 2000; Wuet al., 2006). Although RIP1 has a role in death receptor apoptosis(Morgan et al., 2009; O’Donnell et al., 2007; Wang et al., 2008),it is essential for death receptor-mediated necrotic cell death byFas and TNF (Holler et al., 2000; Lin et al., 2004), where it isrequired for the generation of reactive oxygen species (ROS) (Linet al., 2004). We have recently reported that RIP1 and TRADDinteract with NOXO1 to recruit and activate the NADPH oxidaseNox1, resulting in superoxide production during TNF-inducednecrotic cell death (Kim et al., 2007). Therefore, RIP1 is a crucialcomponent of TNF signaling that controls the divergence ofdistinct pathways and cellular outcomes in TNF-exposed cells.However, the involvement of RIP1 in the noncanonical NF-Bpathway has not been examined.

Our previous studies demonstrated that TRAF2, which is animportant effector molecule of TNF signaling, is also a crucial,non-redundant component in LTR signaling (Kim et al., 2005).However, we found that RIP1 is not involved in LTR-mediatedactivation of NF-B and JNK (Kim et al., 2005). In the currentstudy, we found that RIP1 is a negative regulator of thenoncanonical NF-B pathway in TNFR1 signaling. The absence ofRIP1 in cells treated with TNFleads to TRAF2 degradation, NIKstability, phosphorylation of IKK and p100 processing, resultingin noncanonical NF-B activation.

ResultsTNF activates the noncanonical NF-B pathway in RIP1–/–

MEFsAlthough a crucial role of RIP1 in the canonical TNF-inducedNF-B pathway has been well established (Devin et al., 2000; Eaet al., 2006; Kelliher et al., 1998; Poyet et al., 2000; Wu et al.,2006), its involvement in noncanonical NF-B activation isunknown. We carefully compared TNF-induced p100 processingin MEF cells and found that treatment with murine TNF led to asignificant increase in processing of p100 and the formation of p52in RIP1–/– mouse embryonic fibroblasts (MEFs), but not in wild-type (WT) MEFs (Fig. 1A). Unlike canonical NF-B activation,the TNF-induced processing of p100 in these cells appears slowlyand only becomes evident after 4 hours. We treated the RIP1–/–

MEFs with human recombinant TNF, which only binds to TNFR1in murine cells (Lewis et al., 1991). A similar pattern of p100processing was observed when treated with human TNF(supplementary material Fig. S1A), suggesting that p52 generationis mediated by TNFR1, but not TNFR2. Similar results wereobtained in RIP1–/– MEFs treated with an agonistic TNFR1-specificantibody (supplementary material Fig. S1B), and the p52 to p100ratio decreased substantially at the 4-hour time point when aTNFR1-specific blocking antibody was added (Michael Morgan,unpublished results). Next, we compared the role of RIP1 in early

648 Journal of Cell Science 124 (4)

Fig. 1. TNF induces activation of the noncanonical but not canonical NF-B pathway in RIP1–/– MEFs. (A)Western blots showing p100 processing inTNF-treated wild-type and RIP1–/– MEFs. Cells were treated with TNF (30ng/ml) for indicated times and cell lysates were applied for western blot withindicated antibodies. (B)(Top) Electrophoretic mobility shift assay (EMSA) innuclear cell lysates from wild-type, TRAF2–/– and RIP1–/– MEFs treated withTNF showing binding to NF-B probe. (Bottom) As a control, an anti-Sp-1antibody was used in western blot of the same nuclear lysates. (C)Westernblots of lysates from TNF-treated wild-type and RIP1–/– MEFs showingreduced IB degradation in RIP1–/– MEFs. (D)Western blots from RIP1–/–

MEFs and stably transfected RIP1–/–(FLAG–RIP1) MEFs showing recovery ofIB degradation during TNF (30 ng/ml) treatment. (E)Western blots oflysates from wild-type MEFs transfected with non-target control siRNA (NC)or RIP1 siRNA and treated with TNF for indicated times, showing increasedIB stability in RIP1-deficient cells.

Jour

nal o

f Cel

l Sci

ence

TNF-induced NF-B activation, which is proposed to be solelymediated by the canonical pathway (Rauert et al., 2009) and LTR-induced NF-B activation, which substantially activates thenoncanonical pathway (Dejardin et al., 2002). TNF treatment ledto potent NF-B activation in WT MEFs, with a partial reductionin TRAF2–/– MEFs and little activation in RIP1–/– MEFs, asdetermined by electrophoretic mobility shift assay (EMSA) at earlytime points (Fig. 1B). Some activation by EMSA was observed inRIP1–/– MEFs at high exposure. Nevertheless, this activation wassubstantially less than in WT cells. Since a recent report hasquestioned the role of RIP1 in early canonical NF-B activationusing different methodology (Wong et al., 2010), we also lookedat IB degradation in the same cells. IB degradation wassubstantially decreased in RIP1–/– cells when compared with WTcells (Fig. 1C). Because our MEFs were immortalized after goingthrough crisis, we wanted to verify that the effect was due to lossof RIP1 and not other molecules. We therefore stably transfectedFLAG–RIP1 into the RIP1–/– MEFs and observed that IBdegradation was restored when RIP1 expression was reconstituted(Fig. 1D). In addition, siRNA knockdown of RIP1 in WT MEFsalso led to a decrease in IB degradation (Fig. 1E). This datasuggests that although RIP1 might not be absolutely required forany TNF-induced canonical NF-B activation per se, it has asignificant and positive role in this process. By contrast, andconsistent with our previous finding (Kim et al., 2005), LTR-mediated noncanonical NF-B activation was similar in WT andRIP1–/– MEFs (supplementary material Fig. S1C), suggesting thatRIP1 is not important in the noncanonical NF-B activationinitiated through the LT receptor. Taken together, these resultssuggest that RIP1 is required for activation of the canonical pathwayby TNFR1 and it might negatively affect activation of thenoncanonical pathway specifically through TNFR1 when the cellsare treated with TNF.

TNF induces TRAF2 degradation in RIP1–/– MEFsThe noncanonical NF-B pathway is constitutively active inTRAF2–/– B cells and MEFs (Grech et al., 2004; Vince et al.,2007), and is degraded upon CD40 engagement in B cells (Brownet al., 2001; Liao et al., 2004), suggesting that TRAF2 functions asa negative regulator of p100 processing. We therefore carefullyexamined TRAF2 protein levels in TNF-treated RIP1–/– MEFs.Upon TNF treatment, there was marked reduction of TRAF2 inRIP1–/– MEFs, but not in WT MEFs (Fig. 2A and supplementarymaterial Fig. S2). Notably, reduction of TRAF2 protein levels inRIP1–/– MEFs correlated with p100 processing, whereas p52 proteinlevels were constitutively higher in TRAF2–/– MEFs, irrespectiveof TNF treatment. A closer kinetic study revealed that TRAF2degradation began 30 minutes after TNF treatment and TRAF2levels continued to decrease until 4 hours after treatment, which iswhen we observed processing of p100 to p52 (Fig. 2B). Treatmentwith both murine and human TNF led to a reduction in TRAF2protein levels to the same extent, suggesting that TNF-inducedreduction of TRAF2 is mediated by TNFR1 (supplementarymaterial Fig. S3A). TRAF2 was not degraded in WT or RIP1–/–

MEFs in response to treatment with TRAIL and no increase inp52 was observed in response to this ligand (supplementarymaterial Fig. S3B,C). However, treatment of either WT or RIP1–/–

MEFs with LIGHT led to generation of p52 in the absence ofTRAF2 degradation, suggesting that another mechanism for p52generation is present in this pathway (supplementary material Fig.S3C).

TRAF3 is degraded upon engagement of BAFF-R or CD40, andits degradation is therefore thought to be important in activation ofthe noncanonical NF-B pathway (Brown et al., 2001; Liao et al.,2004). TRAF3-deficient cells show constitutive processing of p100to p52 (Gardam et al., 2008). However, we failed to observe muchchange in TRAF3 protein levels in TNF-treated RIP1–/– MEFs(Fig. 2C). We previously saw little change in TRAF3 protein levelsduring noncanonical NF-B activation induced by anti-LTRantibody (Kim et al., 2005). Therefore, although TRAF3degradation might be involved in noncanonical NF-B activationdownstream of some TNF receptor family members in certain celltypes, it appears to not be required for activation of the noncanonical

649RIP1 attenuates noncanonical NF-B

Fig. 2. Deficiency of RIP1 leads to p100 processing and TRAF2degradation in response to TNF. (A)Western blots of lysates from TNF-treated wild-type, RIP1–/– and TRAF2–/– MEFs showing TRAF2 degradationand p52 generation in RIP1–/– MEFs. (B)Western blots of lysates from shorttime course of TNF-treated RIP1–/– MEFs showing a correlation betweenTRAF2 degradation and p52 generation. (C)Western blots of lysates fromwild-type and RIP1–/– MEFs showing TRAF2 degradation in TNF-treatedRIP1–/– MEFs cells, but lack of TRAF3 degradation. (D)Western blots oflysates from wild-type MEFs transfected with non-target control siRNA (NC)or RIP1 siRNA and treated with TNF for indicated times, showingdegradation of TRAF2 in RIP1-knockdown wild-type MEFs. (E)Western blotshowing protein expression levels of RIP1 in wild-type, RIP1–/– and RIP1–/–

(Flag-RIP1) MEFs. (F)Western blots of lysates from RIP1–/– MEFstransfected with FLAG or FLAG–RIP1 plasmid, respectively, showing noTRAF2 degradation and p52 generation in RIP1-reconstituted RIP1–/– MEFs.

Jour

nal o

f Cel

l Sci

ence

pathway downstream of TNFR1 or LTR in MEFs, and it isTRAF2 degradation that is associated with the TNF-inducedactivation of the noncanonical NF-B pathway in the RIP1–/–

MEFs.To verify that TRAF2 degradation and noncanonical NF-B

activation in the RIP1–/– MEFs is the authentic effect of RIP1deficiency, we first used siRNA to knock down RIP1 in wild-typeMEFs, and verified that TRAF2 degradation accompanied RIP1deficiency in TNF-treated cells (Fig. 2D). We next examined theeffect of RIP1 reconstitution in RIP1–/– MEFs in response to TNFtreatment. Transfection of FLAG-tagged RIP1 into RIP1–/– MEFsrestored the RIP1 protein level to endogenous levels (Fig. 2E).Reconstitution of RIP1 blocked TRAF2 degradation andsignificantly reduced processing of p100 that was induced byTNF (Fig. 2F), supporting the notion that RIP1 negativelyregulates noncanonical NF-B activation in response to TNFtreatment by preventing TRAF2 degradation.

To further substantiate the role of RIP1 in prevention ofnoncanonical NF-B activation, we infected A549 lung cancercells with a lentivirus expressing RIP1 shRNA. When RIP1 wasknocked down, TNF-induced p100 processing to p52 was clearlydetected (Fig. 3A), confirming that TNF-stimulated activation ofthe noncanonical pathway is not restricted to MEFs, and that RIP1negatively regulates this pathway. TNF-stimulated NF-B activityas measured by a NF-B luciferase reporter was significantlyreduced in the RIP1-knockdown cells (Fig. 3B); however, somethe remaining NF-B activity was additionally reduced bytransfection of IKK siRNA to block noncanonical NF-Bactivation (Fig. 3B). Since IKK is required upstream of thenoncanonical pathway, this probably indicates that at least some ofthe remaining NF-B activity in the RIP1-deficient cells is

attributable to activation of the noncanonical pathway. No reductionin NF-B-driven activity was seen in control cells transfected withIKK siRNA, suggesting that the noncanonical pathway is activatedonly in the absence of RIP1.

Degradation of TRAF2 leads to accumulation of NIKWe next examined the signaling events downstream of TRAF2degradation. Activation of the noncanonical NF-B pathwayrequires NF-B-inducing kinase (NIK) activity to induce p100processing (Xiao et al., 2004). TRAF3 recruits NIK to a complexthat promotes its proteasomal-mediated degradation, whereasTRAF2 recruits cIAP1 or cIAP2, which are necessary for thedegradation to take place (Vallabhapurapu et al., 2008; Zarnegar etal., 2008). Therefore, we hypothesized that TRAF2 degradationshould lead to the stabilization of NIK and the subsequent activationof the noncanonical NF-B pathway. TNF did induce NIKaccumulation in RIP1–/– MEFs, whereas no increase occurred inWT MEFs (Fig. 4A). As expected, NIK protein levels were highin TRAF2–/– MEFs, even in the absence of TNF (Fig. 4A). Toverify that this was a TNFR1-specific effect, we treated RIP1–/–

MEFs with an agonistic TNFR1-specific antibody (supplementarymaterial Fig. S4A) and NIK accumulated similarly as with TNFtreatment. To further eliminate the effect of TNFR2, we transfectedTNFR2–/– MEFs with siRNA to knock down RIP1 and treatedthese cells with TNF. We observed NIK accumulation in thesecells at 1 and 4 hours after TNF treatment (supplementary materialFig. S4B). Because NIK activates IKK by phosphorylation totrigger p100 processing (Senftleben et al., 2001; Xiao et al., 2004;Xiao et al., 2001), we looked at whether IKK is phosphorylatedwhen NIK levels increase in the absence of RIP1. There was atransient increase in IKK phosphorylation in TNF-treatedRIP1–/– MEFs that followed the increase of NIK (Fig. 4B). AlthoughNIK stability usually peaked between 1and 4 hours, after that theNIK stability and the IKK phosphorylation decreased, and thelevels of TRAF2 began to rise again. By 8 hours, NIK levels had

650 Journal of Cell Science 124 (4)

Fig. 3. NF-B activation in stable RIP-knockdown cells is dependent onIKK. (A)Western blots of lysates from stable A549-NC and A549-RIP1 KDcells treated with TNF (10 ng/ml) for the indicated times showing p52generation in response to TNF in the RIP-deficient A549 cells. (B)(Top) NF-B luciferase assay from A549-NC and A549-RIP1 KD cells mock transfectedor transfected with 5 nM IKK siRNA or negative control siRNA (NC)showing decrease in NF-B activity in response to IKK knockdown 48 hoursafter transfection (6 hours after TNF treatment). Data are means ± s.e.m.;*P<0.05. (Bottom) Knockdown of IKK was confirmed by western blot.

Fig. 4. Degradation of TRAF2 induces NIK accumulation andphosphorylation of IKK. (A)Western blots of NIK and TRAF2 in TNF(30 ng/ml)-treated wild-type, RIP1–/– and TRAF2–/– MEFs for the indicatedtimes. (B)Western blots of longer time course showing NIK accumulation andIKK phosphorylation in response to TNF in RIP1–/– MEFs.

Jour

nal o

f Cel

l Sci

ence

returned to that of the untreated cells despite the fact that theTRAF2 levels did not completely recover. This might suggest asecondary feedback inhibition on the stability of NIK in additionto TRAF2. Nevertheless, these results indicate that the degradationof TRAF2 in the absence of RIP1 precedes the appearance of NIKstabilization and phosphorylation of IKK, which precedes thegeneration of p52. Thus TRAF2 degradation is probably the crucialevent that leads to the activation of the noncanonical NF-Bpathway by TNF.

TRAF2 degradation in TNF-stimulated RIP1–/– MEFs isblocked by proteasomal inhibitionBecause our data indicate that TNF-induced degradation ofTRAF2 is a crucial step in p100 processing in these cells, weexamined the mechanism of TRAF2 degradation. Neither NIKstabilization and p52 generation, nor TRAF2 degradation wasinhibited in the presence of cycloheximide (supplementarymaterial Fig. S5A), indicating that de novo protein synthesis isnot required for induction of the upstream noncanonical pathway.TRAF2 degradation and p100 processing was likewise notaffected by the pan-caspase inhibitor zVAD (supplementarymaterial Fig. S5B), indicating that caspases are not the sourceof cleavage or degradation. Although TRAF2 degradation inRIP1–/– MEFs was not inhibited by zVAD, it was completelyinhibited by MG132 (supplementary material Fig. S6A),suggesting that the reduction of TRAF2 protein is due toproteasome-dependent degradation, rather than caspase-mediatedcleavage. Stabilization of TRAF2 by MG132 treatment resultedin the suppression of p100 processing in response to TNF(supplementary material Fig. S6B), which might further suggestthat TRAF2 negatively regulates the TNF-induced noncanonicalNF-B pathway in RIP1–/– MEFs. However, p100 processing isthought to also be dependent on the proteasome, so we examinedthe effect of short-term MG132 treatment on the basal levels ofp52 in TRAF2–/–, TRAF2–/–TRAF5–/– and TRAF3–/– MEFs, whichhave constitutive p52 processing (supplementary material Fig.S6C,D,E). Short-term treatment of MG132 did not substantiallyreduce p52 levels in these cell types. Interestingly, although thesecells have stable NIK levels because of a lack of TRAFs, the NIKprotein was more highly upregulated in all of these cell typesupon MG132 treatment (supplementary material Fig. S6C,D,E),suggesting that NIK expression is tightly negatively controlledby other mechanisms in addition to the TRAF-dependentubiquitylation that targets NIK to the proteasome. The increasedNIK in the MG132-treated cells would be predicted lead to higherp52 generation. However, we saw no such generation, suggestingeither that MG132 blocked additional p52 generation, or thatmaximal p52 generation was already occurring. These data wouldbe consistent with the hypothesis of Demchenko and colleagues,who, based on results from multiple myeloma cell lines withmutant TRAF proteins, postulated that an initial NIK stabilizationevent substantially activates noncanonical NF-B, whereas furtherstabilization of NIK has little cumulative effect on NF-Bactivation (Demchenko et al., 2010). These data indicate thatalthough it is clear that TRAF2 is being degraded in aproteasomal-dependent manner in TNF-stimulated RIP1–/–

MEFs, because proteasomal inhibition can block both TRAF2degradation and p100 processing, we could not conclude, basedon these experiments, that p52 generation was dependent onTRAF2 degradation in this case. We therefore sought other datathat would be consistent with this hypothesis.

RIP1 binding to the receptor complex limits therecruitment of TRAF2 to TNFR1When stimulated, TNFR1 recruits adaptor proteins, such as TRADD,RIP1 and TRAF2 to the receptor complex at the cell membrane,which is then internalized as a whole to form secondary signalingcomplexes (Schneider-Brachert et al., 2004). Although studies byour group and others have found that TRAF2 recruitment to TNFR1is TRADD dependent, the binding of RIP1 within the TNFR1complex does not absolutely require TRADD, but TRADDsubstantially enhances the RIP1 recruitment and is required forubiquitylation of RIP1 in the complex (Ermolaeva et al., 2008;Pobezinskaya et al., 2008). We performed co-immunoprecipitationexperiments in WT and RIP1–/– MEFs with an anti-TNFR1 antibodyand examined the recruitment of TRAF2. Both TRAF2 and RIP1were quickly recruited to TNFR1 following TNF treatment in WTMEFs (Fig. 5A). In RIP1–/– MEFs, TNF-induced TRAF2recruitment was dramatically increased by nearly 100-fold, whereasthe levels of TNFR1 protein precipitated were similar in both celltypes (Fig. 5A). The TRAF2 interaction was rapid and transient,reaching its peak at 5 minutes, whereas only a small fraction ofTRAF2 was found interacting with TNFR1 at 30 minutes (Fig. 5A).Also consistent with previous data (Devin et al., 2000), there was anenhanced level of TRADD pulled down by TNFR1 in RIP1–/– MEFscompared with levels in WT MEFs (Fig. 5A), indicating that RIP1might compete somewhat with TRADD in its interaction withTNFR1 in response to TNF. When co-immunoprecipitationexperiments were performed with an anti-TRADD antibody,significantly more TRAF2 interacted with TRADD in RIP1–/– MEFsthan in WT MEFs (Fig. 5B), suggesting that that RIP1 also limits

651RIP1 attenuates noncanonical NF-B

Fig. 5. RIP1 reduces the recruitment of TRAF2 to TNFR1 and TRADD.(A,B) Western blots of immunoprecipitates in TNF (30 ng/ml)-treated wild-type or RIP1–/– MEFs immunoprecipitated with (A) anti-TNFR1 antibody or(B) anti-TRADD antibody. Input: 1% of total lysates.

Jour

nal o

f Cel

l Sci

ence

TRAF2 recruitment to TRADD in response to TNF. Importantly,the level of ubiquitylation in the receptor complex increasedconsiderably (supplementary material Fig. S7), and this correspondedwith substantially higher levels of high molecular weight TRAF2 inthe complex (Fig. 5 and supplementary material Fig. S7), indicatingthat TRAF2 is more highly ubiquitylated in the receptor complex inthe RIP1–/– MEFs, which is consistent with data in supplementarymaterial Fig. S6 showing that the proteasome inhibition blocksTRAF2 degradation. These data thus suggest that RIP1 has apreviously unidentified function: to limit the recruitment of TRAF2to TNFR1, thus stabilizing TRAF2 levels and subsequentlypreventing the activation of the noncanonical NF-B pathway. Todetermine whether the kinase activity of RIP1 was important inpreventing TRAF2, we transfected RIP1–/– MEFs with kinase-deficient Xpress-RIP1 (K45A). Transfection of this plasmiddecreased TRAF2 degradation, NIK stability and generation of p52in response to TNF (supplementary material Fig. S8), althoughthese decreases were not as substantial as in our previous experiment

using nonmutant FLAG–RIP1 (Fig. 2F). It is not clear whether thiswas due to less RIP1 transfection efficiency or expression, use of adifferent epitope tag or lack of kinase activity. Nevertheless, thepartial protection seen in this experiment might indicate that thekinase activity of RIP1 is not required for its protection from TRAF2degradation.

Noncanonical NF-B activation protects against TNF-induced death in RIP1–/– MEFsRIP1 has important roles in death-receptor-induced cell death inboth apoptosis and necrosis (Holler et al., 2000; Kim et al., 2007;Lin et al., 2004; Morgan et al., 2009; O’Donnell et al., 2007; Wanget al., 2008). Owing to activation of the NF-B pathway, whichresults in production of pro-survival proteins, most cells are resistantto TNF in the absence of protein synthesis inhibitors. Weexamined TNF-induced cell death in the RIP1–/– MEFs, whichlack the strong canonical NF-B activation induced in WT MEFs(Kelliher et al., 1998). As shown in Fig. 6A, TNF alone was

652 Journal of Cell Science 124 (4)

Fig. 6. TNF alone induces apoptotic cell death inRIP1–/– MEFs at earlier time points, but death isinhibited at later time points. (A)Viability of TNF(30 ng/ml)-treated wild-type or RIP1–/– MEFs at 8hours visualized by light microscopy (left) or by MTTassay (right) (*P<0.05 compared with untreated groupin RIP1–/– cells; mean ± s.d.). (B)Caspase-3 westernblot in lysates from TNF-treated wild-type orRIP1–/– MEFs. (C)Time-course LDH leakage assay ofTNF-treated wild-type or RIP1–/– MEFs (mean ±s.d.). (D,E)Western blot analysis over long timecourse in TNF-treated RIP1–/– MEFs showingcleavage of caspase-3 (D) or PARP1 (E).(F)Apoptotic cell death measured by using a FITCAnnexin-V Apoptosis Detection kit in TNF-treatedwild-type or RIP1–/– MEFs.

Jour

nal o

f Cel

l Sci

ence

sufficient to induce apoptotic cell death in RIP1–/– MEFs at earliertime points (within 8 hours), but not in WT MEFs. Knockdown ofRIP1 in wild-type cells also led to cell death from TNF alone(supplementary material Fig. S9A), with corresponding cleavageof caspase-3 (supplementary material Fig. S9B). Conversely,ectopic expression of FLAG–RIP1 in these cells decreased theirsensitivity to TNF (supplementary material Fig. S9C), indicatingthat RIP1 deficiency is the likely reason for cell death. In agreementwith the cell cytotoxicity assay data, we detected caspase-3 cleavagein RIP1–/– MEFs upon TNF treatment, but not in WT MEFs (Fig.6B). Surprisingly, when we treated these cells for 24 hours, wefound that cell death did not increase after the 8-hour time point,as measured by LDH release (Fig. 6C) or morphological detectionunder a microscope (supplementary material Fig. S10). To confirmthis phenomenon, we examined caspase-3 activation and PARP1cleavage over a long time course. Caspase-3 activation increaseduntil 4 hours after TNF treatment, and then decreased rapidly(Fig. 6D). Consistently, PARP1 cleavage started at 1 hour andpeaked at 4 hours after TNF treatment and then diminished,accompanying the reduction of caspsase-3 activity (Fig. 6E). Thekinetics of cell death and its subsequent reduction was verified byAnnexin-V and propidium iodide staining (Fig. 6F). This analysisshowed that cell death peaked at about 4 hours and then wassubsequently reduced. TNF-induced cell death was sensitized inthe presence of CHX, whereas the pan-caspase inhibitor, z-VAD-fmk, completely blocked TNF-induced cell death (supplementarymaterial Fig. S11). The necrosis inhibitor necrostatin-1 had noeffect on cell death, indicating that only apoptotic cell deathoccurred, and consistent with the requirement of RIP1 in necroticcell death and the fact that this drug acts as an inhibitor of RIP1kinase activity (supplementary material Fig. S11).

The correlation of the decrease in caspase activation with theonset of strong p52 generation 4 hours after TNF treatment inRIP1–/– cells suggests that the noncanonical NF-B pathwayprotects these cells from further apoptotic cell death, probablythrough induction of anti-apoptotic genes. To see whether this wasthe case, we used siRNA to prevent NIK expression in RIP1–/–

MEFs. NIK protein expression was effectively eliminated fromoccurring in cells transfected with NIK siRNA, although TRAF2degradation continued to take place (Fig. 7A). The prevention ofNIK expression led to an increased sensitivity of the RIP1–/– MEFsto TNF, especially at the 8 hour time point, as observed by lightmicroscopy (Fig. 7B) or by LDH assay (Fig. 7C), thus showingthat the activation of the noncanonical NF-B pathway protectedRIP1–/– MEFs from TNF-induced cell death. As further evidencethat this was the case, the pretreatment of these cells with LIGHT,a well-known inducer of noncanonical NF-B activation preventedRIP1–/– MEFs from dying in response to TNF (Fig. 7D). Thus,although RIP1–/– MEFs undergo cell death early, as a result ofRIP1 deficiency, the late activation of the noncanonical NF-Bprotects surviving cells from long-term TNF toxicity.

DiscussionWe report here that TNFR1 is capable of activating thenoncanonical NF-B signaling pathway in addition to the canonicalpathway. Consistent with previous reports (Dejardin et al., 2002;Rauert et al., 2009), our data suggest that under normalcircumstances in MEF cells, TNF does not substantially activatethis pathway through TNFR1. During TNFR1 signaling, thepresence of RIP1 in the receptor complex not only serves topotentiate the canonical NF-B pathway, but also prevents

activation of the noncanonical NF-B pathway. RIP1 limits theamount of TRADD and TRAF2 proteins recruited to the TNFR1signaling complex, probably limiting extensive ubiquitylation ofTRAF2 in the receptor complex. When RIP1 is removed,ubiquitylation-dependent and proteasomal-mediated TRAF2degradation leads to NIK stabilization and subsequently theactivation of the noncanonical pathway. From the data in our study,we propose that RIP1 deficiency leads to a lack of canonical NF-B upon TNF treatment. This sensitizes the cells to cell death atearlier time points. However, as a result of the lack of RIP1,TRAF2 is degraded and NIK is stabilized, leading to noncanonicalNF-B activation, and thus protecting the cells from further death.

RIP1 has a weak binding affinity to TRAF2 through its kinaseand intermediate domains (Hsu et al., 1996), suggesting a possibledirect interaction between RIP1 and TRAF2 while in the complex.Because it has a strong affinity for RIP1, TRADD enables theefficient recruitment RIP1 to the TNFR1 receptor complex(Ermolaeva et al., 2008; Pobezinskaya et al., 2008). However, inthe absence of TRADD, RIP1 maintains a weak interaction withTNFR1, suggesting that there is also a direct interaction between

653RIP1 attenuates noncanonical NF-B

Fig. 7. Inhibition of noncanonical NF-kB pathway causes TNF-inducedcell death in RIP1–/– MEFs. (A)Western blots of lysates from RIP1–/– MEFstransfected with non-target control siRNA or NIK siRNA and treated withTNF (2 hours) showing that TRAF2 continues to be degraded in NIK-knockdown RIP1–/– MEFs. (B,C)Viability assays of RIP1–/– MEFs transfectedwith NIK siRNA or non-targeted control (NC) siRNA and treated with TNFshowing the effect of NIK knockdown on TNF-induced apoptosis in RIP1–/–

MEFs. (B)Representative phase-contrast microscopy images and (C) LDHleakage assay data (mean ± s.d.) are shown at indicated times. (D)MTT cellviability assay of RIP1–/– MEFs untreated, or pretreated for 5 hours withLIGHT then treated with TNF for 8 hours (*P<0.05 compared with untreatedgroup using Student’s t-test analysis; N.S., not significant) showing thatLIGHT pretreatment protects RIP1–/– MEFs from TNF-induced cell death.

Jour

nal o

f Cel

l Sci

ence

RIP1 and TNFR1 (Chen et al., 2008; Ermolaeva et al., 2008;Pobezinskaya et al., 2008). Our data here show that loss of thisinteraction (through a RIP1 deficiency) permits greater TRADDrecruitment to TNFR1. The comparative amount of TRAF2 increasein the receptor complex is even proportionally greater than theincrease in TRADD when compared with levels in WT cells.Therefore, it seems likely that RIP1 is not only competing to someextent with TRADD for TNFR1 binding, but also slows or obstructsTRAF2 binding to TRADD, perhaps by a kinetic constraint that isimposed by forming its own associations with TRADD and TRAF2.

In the absence of RIP1, not only was more TRAF2 recruited toTNFR1 complex, but also a greater abundance of higher molecularweight TRAF2 was found in the receptor complex, suggesting thatTRAF2 is further ubiquitylated, thus increasing TRAF2 proteindegradation. Since cIAP proteins are recruited by TRAF2 to theTNFR1 complex and TRAF2 is degraded by cIAP1 during TNFR2signaling (Li et al., 2002; Rothe et al., 1995; Shu et al., 1996), wehypothesize that the loss of RIP1 results in less competition forcIAP activity, and therefore more TRAF2 ubiquitylation. WithoutRIP1, cIAP proteins might also be in a less spatially constrainedenvironment, and therefore the enzymatic sites might be able tobetter interact structurally and therefore more substantiallyubiquitylate TRAF2.

The absence of RIP1 in the complex probably prevents theformation of a full IKK complex consisting of all the subunits (i.e.IKK, IKK, IKK), which is necessary for the full activation ofthe canonical pathway. We have previously shown that IKK andIKK can be recruited to an activated TNFR1 complex in theabsence of IKK through interaction with TRAF2 (Devin et al.,2000; Devin et al., 2001). However, others have shown that thepolyubiquitylated RIP1-recruited IKK subunit strongly stabilizescomplex formation (Ea et al., 2006; Poyet et al., 2000; Wu et al.,2006). Although an absolute primary role for RIP1 in activation ofthe canonical pathway has been recently questioned (Wong et al.,2010) based on observation of IB degradation and analysis ofgene expression in response to TNF, in our hands, the deficiencyof RIP1 in RIP1–/– MEFs or by RIP1 knockdown leads to asignificant decrease in TNF-induced IB degradation and NF-B binding as measured by EMSA, which is in agreement withnumerous previous studies. Because less IKK is tied up in thecanonical pathway, more IKK might be free to form IKKhomodimers to activate the noncanonical pathway downstream ofNIK (Senftleben et al., 2001). Inhibition of the regulatory IKKsubunit binding using peptide inhibitors has previously been shownto increase the basal level of IKK activity (May et al., 2000). Thus,RIP1 deficiency could conceivably contribute to noncanonicalactivation in two different ways.

Consistent with our current data regarding TRAF3 expression inTNF-treated RIP1–/– cells, we previously reported little change inTRAF3 protein levels in response to LIGHT-induced activation ofthe noncanonical NF-B pathway (Kim et al., 2005). However, weobserved extensive TRAF2 degradation during TNFR1 activation,but not during LIGHT signaling (Kim et al., 2005). These twopathways are therefore distinctly different from those of BAFF andCD40, where TRAF3 is degraded (Brown et al., 2001; Liao et al.,2004). BAFF does not cause TRAF2 degradation, whereas CD40stimulation results in loss of both TRAF2 and TRAF3 (Brown etal., 2001; Liao et al., 2004). However, TWEAK signaling throughits receptor, FN14, activates a lysosome-mediated degradation ofTRAF2 and cIAP1 in a cIAP1-dependent manner, and thus activatesthe noncanonical pathway (Vince et al., 2008). Consistent with

this, our data also indicate that TRAF2 degradation, as induced bya receptor stimulus, is sufficient for activation of the noncanonicalpathway. Based on our data, and the current model, one wouldpredict that TNFR2 stimulation would result in the activation ofthe noncanonical pathway, because TRAF2 is degraded by cIAP1during TNFR2 signaling (Li et al., 2002). Consistent with ourmodel, a new report has been published during the preparation ofthis manuscript showing that membrane-bound TNF can activatethe noncanonical pathway in primary T-cells and cancer cell linesthrough its binding to TNFR2 (Rauert et al., 2009).

Both TRAF2- and TRAF3-knockout mice die perinatally ofuncontrolled noncanonical NF-B activation (He et al., 2006;Vallabhapurapu et al., 2008; Yeh et al., 1997). In light of our data,and because RIP1–/– mice have similar prenatal lethality (Kelliheret al., 1998) that is partially rescued by TNFR1 deficiency (Cussonet al., 2002), it is possible that some of the defects in RIP1–/– miceare also due, in part, to hyperactive noncanonical NF-B signalingdownstream of TNF.

Previous reports have suggested that activation of thenoncanonical pathway serves to downregulate the canonicalpathway in different ways (Basak et al., 2007; Lawrence et al.,2005). Our studies show that TNFR1 is capable of inducing bothcanonical and noncanonical pathways, but that both pathways aredeliberatively regulated in opposite ways by RIP1. Thus, there isa functional interplay between the two pathways, which mightallow regulation of the number and kinds of NF-B transcriptionfactors and thus fine-tuning the regulation of NF-B-dependentgenes. The canonical pathway is clearly the dominant NF-Bpathway that ensures an early response of cells to TNFstimulation, whereas the noncanonical pathway might serve as acompensatory redundant mechanism. It is possible that when thecanonical pathway is suppressed, the non-canonical pathway isactivated to maintain a certain level of NF-B activity. For example,in some embryonic developmental stages, the canonical pathwayis inactivated in certain organs when RIP1 expression is lost (Wonget al., 2010). Although our results demonstrate that the noncanonicalpathway is a pro-survival signal that protects cells from TNF-induced cytotoxicity, it would be interesting to determine whetherthis pathway has other biological roles during the TNF response.

Materials and MethodsReagents.Recombinant murine and human TNF, mouse recombinant TRAIL, humanrecombinant LIGHT, agonistic (AF-425-PB) and blocking (MAB430) TNFR1antibodies, TRAF2 antibody and zVAD were purchased from R&D Systems. TRAF3,Sp-1, TRADD, ubiquitin and IB antibodies were purchased from Santa Cruz.RIP1 antibody was purchased from Transduction Laboratories. Caspase3 and PARP1antibodies were purchased from Pharmingen. NIK, phospho-IKK, p52 and IKKantibodies were purchased from Cell Signaling. MG132 and CHX were purchasedfrom Calbiochem. Agonistic monoclonal anti-murine lymphotoxin receptor (LTR)antibody was kindly provided by J. Browning (Biogene). Actin and FLAG antibodiesand Necrostatin-1 were purchased from Sigma.

Cell cultureWild-type (WT), RIP1–/– and TRAF2–/– mouse embryonic fibroblast (MEF) cellswere previously described (Kelliher et al., 1998; Yeh et al., 1997). MEFs werecultured in DMEM supplemented with 10% fetal bovine serum, 2 mM glutamine,100 U/ml penicillin and 100 g/ml streptomycin. RIP1–/– (FLAG–RIP1) stable celllines were established by transfecting cells with the designated vector and thenmaintaining in medium containing 10 g/ml puromycin. A549/NF-B-luc cells werepurchased from Panomics (Fremont, CA), maintained in RPMI 1640 mediumsupplemented with 100 g/ml hygromycin, and were infected with GIPZ-RIP1shRNA or negative control shRNA lentivirus (Open Biosystems, Huntville, AL). Thepuromycin-resistant cell clones were selected with puromycin (2 g/ml) medium andverified by RIP1 antibody by western blot, and were designated as A549/RIP1 KD(for EIP knockdown) and A459/NC (for negative control). They were maintained inRPMI 1640 with hygromycin (100 g/ml) and puromycin (2 g/ml).

654 Journal of Cell Science 124 (4)

Jour

nal o

f Cel

l Sci

ence

Western blot analysis and co-immunoprecipitationUpon treatment, cells were lysed in 0.5% NP-40 Tris M2 buffer (Kim et al., 2007).Equal amounts of cell extracts were resolved by 12% SDS-PAGE and analyzed bywestern blot. For immunoprecipitation assays, lysates were mixed and precipitatedwith antibody and protein-A–Sepharose or protein-G–Agarose beads by incubationat 4°C. The bound proteins were removed by boiling in SDS buffer and resolved in12% SDS-polyacrylamide gels for western blot analysis and visualized by enhancedchemiluminescence (ECL, Amersham).

Electrophoretic mobility shift assayNuclear extracts were prepared and analyzed as previously described (Kim et al.,2005). For the binding reaction, 5 g of nuclear extract was incubated at roomtemperature for 20 minutes with reaction buffer containing 20 mM HEPES (pH 7.9),50 mM KCl, 0.1 mM EDTA, 1 mM DTT, 5% glycerol, 200 g/ml BSA, and 2 gpoly(dI-dC)·poly(dI-dC). Then the 32P-labeled double-stranded oligonucleotide (1 ng,≥1�105 c.p.m.) containing the NF-B binding consensus sequence (5�-GGCAA -CTGGGGACTCTCCCTTT-3�) was added to the reaction mixture for an additional10 minutes at room temperature. The reaction products were fractionated on anondenaturating 5% polyacrylamide gel, which was then dried and subjected toautoradiography. As controls of nuclear protein content, an anti-Sp1 antibody wasused in EMSA and western blotting, respectively.

TransfectionCells were transfected with FLAG–RIP1 plasmid (Kim et al., 2007), Xpress-RIP1K45A mutant plasmid, FLAG–TRAF2(87–501) or FLAG empty vector withLipofectamine PLUS reagent according to the manufacturer’s protocol(GIBCO/BRL).

siRNA knockdownMEF cells were plated in six-well plates, and cells were transfected with 100 pmolesof NIK, RIP1 or non-targeting control (NC) RNAi oligo (Dharmacon) usingLipofectamine 2000 reagent (Invitrogen). After 48 hrs, cells were treated with TNF- for indicated times.

Luciferase assayIKK siRNA was transfected with INTERFERinTM (PolyPlus Transfection, SanMarcos, CA). After 48 hours, cells were treated with TNF as described in figurelegends, lysed, and luciferase activity was measured using a luciferase assay kit(Promega Corporation, Madison, WI) and normalized to protein concentration.

Cytotoxicity assaysCell death was determined using tetrazolium dye colorimetric test (MTT test) andabsorbance read at 570 nm. LDH leakage was quantified using a cytotoxicitydetection kit (Promega). Cell death was also measured by using a FITC-Annexin VApoptosis Detection kit (BD Pharmingen) according to the manufacturer’s manual.Representative images were also taken by a phase-contrast microscope.

We thank Wen-Chen Yeh and Tak Wah Mak for TRAF2–/– MEFs;Michelle Kelliher for RIP1–/– MEFs. This work was supported by aNational Research Foundation of Korea Grant funded by the KoreanGovernment (2009-0067084). This work was also supported byResearch Grant funded by the Ajou University School of Medicine (3-2008025-0) and by the grant of ‘The Ajou University ExcellenceResearch Program in 2010’.

Supplementary material available online athttp://jcs.biologists.org/cgi/content/full/124/4/647/DC1

ReferencesAnnunziata, C. M., Davis, R. E., Demchenko, Y., Bellamy, W., Gabrea, A., Zhan, F.,

Lenz, G., Hanamura, I., Wright, G., Xiao, W. et al. (2007). Frequent engagement ofthe classical and alternative NF-kappaB pathways by diverse genetic abnormalities inmultiple myeloma. Cancer Cell 12, 115-130.

Basak, S., Kim, H., Kearns, J. D., Tergaonkar, V., O’Dea, E., Werner, S. L., Benedict,C. A., Ware, C. F., Ghosh, G., Verma, I. M. et al. (2007). A fourth IkappaB proteinwithin the NF-kappaB signaling module. Cell 128, 369-381.

Beinke, S. and Ley, S. C. (2004). Functions of NF-kappaB1 and NF-kappaB2 in immunecell biology. Biochem. J. 382, 393-409.

Bonizzi, G. and Karin, M. (2004). The two NF-kappaB activation pathways and their rolein innate and adaptive immunity. Trends Immunol. 25, 280-288.

Brown, K. D., Hostager, B. S. and Bishop, G. A. (2001). Differential signaling and tumornecrosis factor receptor-associated factor (TRAF) degradation mediated by CD40 andthe Epstein-Barr virus oncoprotein latent membrane protein 1 (LMP1). J. Exp. Med.193, 943-954.

Chaudhary, P. M., Eby, M., Jasmin, A., Bookwalter, A., Murray, J. and Hood, L.(1997). Death receptor 5, a new member of the TNFR family, and DR4 induce FADD-dependent apoptosis and activate the NF-kappaB pathway. Immunity 7, 821-830.

Chen, N. J., Chio, II, Lin, W. J., Duncan, G., Chau, H., Katz, D., Huang, H. L., Pike,K. A., Hao, Z., Su, Y. W. et al. (2008). Beyond tumor necrosis factor receptor:

TRADD signaling in toll-like receptors. Proc. Natl. Acad. Sci. USA 105, 12429-12434.

Cusson, N., Oikemus, S., Kilpatrick, E. D., Cunningham, L. and Kelliher, M. (2002).The death domain kinase RIP protects thymocytes from tumor necrosis factor receptortype 2-induced cell death. J. Exp. Med. 196, 15-26.

Dejardin, E., Droin, N. M., Delhase, M., Haas, E., Cao, Y., Makris, C., Li, Z. W.,Karin, M., Ware, C. F. and Green, D. R. (2002). The lymphotoxin-beta receptorinduces different patterns of gene expression via two NF-kappaB pathways. Immunity17, 525-535.

Demchenko, Y. N., Glebov, O. K., Zingone, A., Keats, J. J., Bergsagel, P. L. andKuehl, W. M. (2010). Classical and/or alternative NF-kappaB pathway activation inmultiple myeloma. Blood 115, 3541-3552.

Devin, A., Cook, A., Lin, Y., Rodriguez, Y., Kelliher, M. and Liu, Z. (2000). The distinctroles of TRAF2 and RIP in IKK activation by TNF-R1: TRAF2 recruits IKK to TNF-R1 while RIP mediates IKK activation. Immunity 12, 419-429.

Devin, A., Lin, Y., Yamaoka, S., Li, Z., Karin, M. and Liu, Z. (2001). The alpha andbeta subunits of IkappaB kinase (IKK) mediate TRAF2-dependent IKK recruitment totumor necrosis factor (TNF) receptor 1 in response to TNF. Mol. Cell. Biol. 21, 3986-3994.

Ea, C. K., Deng, L., Xia, Z. P., Pineda, G. and Chen, Z. J. (2006). Activation of IKKby TNFalpha requires site-specific ubiquitination of RIP1 and polyubiquitin binding byNEMO. Mol. Cell 22, 245-257.

Ermolaeva, M. A., Michallet, M. C., Papadopoulou, N., Utermohlen, O., Kranidioti,K., Kollias, G., Tschopp, J. and Pasparakis, M. (2008). Function of TRADD in tumornecrosis factor receptor 1 signaling and in TRIF-dependent inflammatory responses.Nat. Immunol. 9, 1037-1046.

Festjens, N., Vanden Berghe, T., Cornelis, S. and Vandenabeele, P. (2007). RIP1, akinase on the crossroads of a cell’s decision to live or die. Cell Death Differ. 14, 400-410.

Gardam, S., Sierro, F., Basten, A., Mackay, F. and Brink, R. (2008). TRAF2 andTRAF3 signal adapters act cooperatively to control the maturation and survival signalsdelivered to B cells by the BAFF receptor. Immunity 28, 391-401.

Grech, A. P., Amesbury, M., Chan, T., Gardam, S., Basten, A. and Brink, R. (2004).TRAF2 differentially regulates the canonical and noncanonical pathways of NF-kappaBactivation in mature B cells. Immunity 21, 629-642.

Hayden, M. S. and Ghosh, S. (2008). Shared principles in NF-kappaB signaling. Cell132, 344-362.

He, J. Q., Zarnegar, B., Oganesyan, G., Saha, S. K., Yamazaki, S., Doyle, S. E.,Dempsey, P. W. and Cheng, G. (2006). Rescue of TRAF3-null mice by p100 NF-kappa B deficiency. J. Exp. Med. 203, 2413-2418.

He, J. Q., Saha, S. K., Kang, J. R., Zarnegar, B. and Cheng, G. (2007). Specificity ofTRAF3 in its negative regulation of the noncanonical NF-kappa B pathway. J. Biol.Chem. 282, 3688-3694.

Holler, N., Zaru, R., Micheau, O., Thome, M., Attinger, A., Valitutti, S., Bodmer, J.L., Schneider, P., Seed, B. and Tschopp, J. (2000). Fas triggers an alternative, caspase-8-independent cell death pathway using the kinase RIP as effector molecule. Nat.Immunol. 1, 489-495.

Hsu, H., Huang, J., Shu, H. B., Baichwal, V. and Goeddel, D. V. (1996). TNF-dependentrecruitment of the protein kinase RIP to the TNF receptor-1 signaling complex. Immunity4, 387-396.

Keats, J. J., Fonseca, R., Chesi, M., Schop, R., Baker, A., Chng, W. J., Van Wier, S.,Tiedemann, R., Shi, C. X., Sebag, M. et al. (2007). Promiscuous mutations activatethe noncanonical NF-kappaB pathway in multiple myeloma. Cancer Cell 12, 131-144.

Kelliher, M. A., Grimm, S., Ishida, Y., Kuo, F., Stanger, B. Z. and Leder, P. (1998).The death domain kinase RIP mediates the TNF-induced NF-kappaB signal. Immunity8, 297-303.

Kim, Y. S., Nedospasov, S. A. and Liu, Z. G. (2005). TRAF2 plays a key, nonredundantrole in LIGHT-lymphotoxin beta receptor signaling. Mol. Cell. Biol. 25, 2130-2137.

Kim, Y. S., Morgan, M. J., Choksi, S. and Liu, Z. G. (2007). TNF-induced activationof the Nox1 NADPH oxidase and its role in the induction of necrotic cell death. Mol.Cell 26, 675-687.

Lawrence, T., Bebien, M., Liu, G. Y., Nizet, V. and Karin, M. (2005). IKKalpha limitsmacrophage NF-kappaB activation and contributes to the resolution of inflammation.Nature 434, 1138-1143.

Lewis, M., Tartaglia, L. A., Lee, A., Bennett, G. L., Rice, G. C., Wong, G. H., Chen,E. Y. and Goeddel, D. V. (1991). Cloning and expression of cDNAs for two distinctmurine tumor necrosis factor receptors demonstrate one receptor is species specific.Proc. Natl. Acad. Sci. USA 88, 2830-2834.

Li, X., Yang, Y. and Ashwell, J. D. (2002). TNF-RII and c-IAP1 mediate ubiquitinationand degradation of TRAF2. Nature 416, 345-347.

Liao, G., Zhang, M., Harhaj, E. W. and Sun, S. C. (2004). Regulation of the NF-kappaB-inducing kinase by tumor necrosis factor receptor-associated factor 3-induceddegradation. J. Biol. Chem. 279, 26243-26250.

Lin, Y., Choksi, S., Shen, H. M., Yang, Q. F., Hur, G. M., Kim, Y. S., Tran, J. H.,Nedospasov, S. A. and Liu, Z. G. (2004). Tumor necrosis factor-induced nonapoptoticcell death requires receptor-interacting protein-mediated cellular reactive oxygen speciesaccumulation. J. Biol. Chem. 279, 10822-10828.

Lin, Y., Bai, L., Chen, W. and Xu, S. (2010). The NF-kappaB activation pathways,emerging molecular targets for cancer prevention and therapy. Expert Opin. Ther.Targets 14, 45-55.

May, M. J., D’Acquisto, F., Madge, L. A., Glockner, J., Pober, J. S. and Ghosh, S.(2000). Selective inhibition of NF-kappaB activation by a peptide that blocks theinteraction of NEMO with the IkappaB kinase complex. Science 289, 1550-1554.

655RIP1 attenuates noncanonical NF-B

Jour

nal o

f Cel

l Sci

ence

Morgan, M. J., Kim, Y. S. and Liu, Z. G. (2009). Membrane-bound Fas ligand requiresRIP1 for efficient activation of caspase-8 within the death-inducing signaling complex.J. Immunol. 183, 3278-3284.

O’Donnell, M. A., Legarda-Addison, D., Skountzos, P., Yeh, W. C. and Ting, A. T.(2007). Ubiquitination of RIP1 regulates an NF-kappaB-independent cell-death switchin TNF signaling. Curr. Biol. 17, 418-424.

Pobezinskaya, Y. L., Kim, Y. S., Choksi, S., Morgan, M. J., Li, T., Liu, C. and Liu, Z.(2008). The function of TRADD in signaling through tumor necrosis factor receptor 1and TRIF-dependent Toll-like receptors. Nat. Immunol. 9, 1047-1054.

Poyet, J. L., Srinivasula, S. M., Lin, J. H., Fernandes-Alnemri, T., Yamaoka,S., Tsichlis, P. N. and Alnemri, E. S. (2000). Activation of the Ikappa B kinases byRIP via IKKgamma/NEMO-mediated oligomerization. J. Biol. Chem. 275, 37966-37977.

Rauert, H., Wicovsky, A., Mueller, N., Siegmund, D., Spindler, V., Waschke, J., Kneitz,C. and Wajant, H. (2009). Membrane tumor necrosis factor (TNF) induces p100processing via TNF receptor-2 (TNFR2). J. Biol. Chem. 285, 7394-7404.

Rothe, M., Pan, M. G., Henzel, W. J., Ayres, T. M. and Goeddel, D. V. (1995). TheTNFR2-TRAF signaling complex contains two novel proteins related to baculoviralinhibitor of apoptosis proteins. Cell 83, 1243-1252.

Schneider-Brachert, W., Tchikov, V., Neumeyer, J., Jakob, M., Winoto-Morbach, S.,Held-Feindt, J., Heinrich, M., Merkel, O., Ehrenschwender, M., Adam, D. et al.(2004). Compartmentalization of TNF receptor 1 signaling: internalized TNFreceptosomes as death signaling vesicles. Immunity 21, 415-428.

Senftleben, U., Cao, Y., Xiao, G., Greten, F. R., Krahn, G., Bonizzi, G., Chen, Y., Hu,Y., Fong, A., Sun, S. C. et al. (2001). Activation by IKKalpha of a second, evolutionaryconserved, NF-kappa B signaling pathway. Science 293, 1495-1499.

Shu, H. B., Takeuchi, M. and Goeddel, D. V. (1996). The tumor necrosis factor receptor2 signal transducers TRAF2 and c-IAP1 are components of the tumor necrosis factorreceptor 1 signaling complex. Proc. Natl. Acad. Sci. USA 93, 13973-13978.

Vallabhapurapu, S. and Karin, M. (2009). Regulation and function of NF-kappaBtranscription factors in the immune system. Annu. Rev. Immunol. 27, 693-733.

Vallabhapurapu, S., Matsuzawa, A., Zhang, W., Tseng, P. H., Keats, J. J., Wang, H.,Vignali, D. A., Bergsagel, P. L. and Karin, M. (2008). Nonredundant and complementaryfunctions of TRAF2 and TRAF3 in a ubiquitination cascade that activates NIK-dependentalternative NF-kappaB signaling. Nat. Immunol. 9, 1364-1370.

Varfolomeev, E., Blankenship, J. W., Wayson, S. M., Fedorova, A. V., Kayagaki, N.,Garg, P., Zobel, K., Dynek, J. N., Elliott, L. O., Wallweber, H. J. et al. (2007). IAP

antagonists induce autoubiquitination of c-IAPs, NF-kappaB activation, and TNFalpha-dependent apoptosis. Cell 131, 669-681.

Varfolomeev, E. E., Boldin, M. P., Goncharov, T. M. and Wallach, D. (1996). Apotential mechanism of “cross-talk” between the p55 tumor necrosis factor receptor andFas/APO1: proteins binding to the death domains of the two receptors also bind to eachother. J. Exp. Med. 183, 1271-1275.

Vince, J. E., Wong, W. W., Khan, N., Feltham, R., Chau, D., Ahmed, A. U., Benetatos,C. A., Chunduru, S. K., Condon, S. M., McKinlay, M. et al. (2007). IAP antagoniststarget cIAP1 to induce TNFalpha-dependent apoptosis. Cell 131, 682-693.

Vince, J. E., Chau, D., Callus, B., Wong, W. W., Hawkins, C. J., Schneider, P.,McKinlay, M., Benetatos, C. A., Condon, S. M., Chunduru, S. K. et al. (2008).TWEAK-FN14 signaling induces lysosomal degradation of a cIAP1-TRAF2 complexto sensitize tumor cells to TNFalpha. J. Cell Biol. 182, 171-184.

Wang, L., Du, F. and Wang, X. (2008). TNF-alpha induces two distinct caspase-8activation pathways. Cell 133, 693-703.

Wong, W. W., Gentle, I. E., Nachbur, U., Anderton, H., Vaux, D. L. and Silke, J.(2010). RIPK1 is not essential for TNFR1-induced activation of NF-kappaB. CellDeath Differ. 17, 482-487.

Wu, C. J., Conze, D. B., Li, T., Srinivasula, S. M. and Ashwell, J. D. (2006). Sensingof Lys 63-linked polyubiquitination by NEMO is a key event in NF-kappaB activation[corrected]. Nat. Cell Biol. 8, 398-406.

Xiao, G., Harhaj, E. W. and Sun, S. C. (2001). NF-kappaB-inducing kinase regulates theprocessing of NF-kappaB2 p100. Mol. Cell 7, 401-409.

Xiao, G., Fong, A. and Sun, S. C. (2004). Induction of p100 processing by NF-kappaB-inducing kinase involves docking IkappaB kinase alpha (IKKalpha) to p100 andIKKalpha-mediated phosphorylation. J. Biol. Chem. 279, 30099-30105.

Xie, P., Stunz, L. L., Larison, K. D., Yang, B. and Bishop, G. A. (2007). Tumor necrosisfactor receptor-associated factor 3 is a critical regulator of B cell homeostasis insecondary lymphoid organs. Immunity 27, 253-267.

Yeh, W. C., Shahinian, A., Speiser, D., Kraunus, J., Billia, F., Wakeham, A., de laPompa, J. L., Ferrick, D., Hum, B., Iscove, N. et al. (1997). Early lethality, functionalNF-kappaB activation, and increased sensitivity to TNF-induced cell death in TRAF2-deficient mice. Immunity 7, 715-725.

Zarnegar, B. J., Wang, Y., Mahoney, D. J., Dempsey, P. W., Cheung, H. H., He, J.,Shiba, T., Yang, X., Yeh, W. C., Mak, T. W. et al. (2008). Noncanonical NF-kappaBactivation requires coordinated assembly of a regulatory complex of the adaptors cIAP1,cIAP2, TRAF2 and TRAF3 and the kinase NIK. Nat. Immunol. 9, 1371-1378.

656 Journal of Cell Science 124 (4)

Jour

nal o

f Cel

l Sci

ence