The Epidermal Growth Factor Receptor Engages Receptor Interacting Protein and Nuclear Factor-κB (NF-κB)-inducing Kinase to Activate NF-κB

The transcription factor nuclear factor-κB (NF-κB) is activated by a diverse number of stimuli including tumor necrosis factor-α, interleukin-1, UV irradiation, viruses, as well as receptor tyrosine kinases such as the epidermal growth factor receptor (EGFR). NF-κB activation by the tumor necrosis factor receptor (TNFR) involves the formation of a multiprotein complex termed a signalosome. Although previous studies have shown that the activated EGFR can induce NF-κB, the mechanism of this activation remains unknown. In this study, we identify components of the signalosome formed by the activated EGFR required to activate NF-κB and show that, although the activated EGFR uses mechanisms similar to the TNFR, it recruits a distinct signalosome. We show the EGFR forms a complex with a TNFR-interacting protein (RIP), which plays a key role in TNFR-induced NF-κB activation, but not with TRADD, an adaptor protein which serves to recruit RIP to the TNFR. Furthermore, we show that the EGFR associates with NF-κB-inducing kinase (NIK) and provide evidence suggesting multiprotein complex formation between the EGFR, RIP, and NIK. Using a dominant negative NIK mutant, we show that NIK activation is required for EGFR-mediated NF-κB induction. We also show that a S32/36 IκBα mutant blocks EGFR-induced NF-κB activation. Our studies also suggest that a high level of EGFR expression, a frequent occurrence in human tumors, is optimal for epidermal growth factor-induced NF-κB activation. Finally, although protein kinase B/Akt has been implicated in tumor necrosis factor and PDGF-induced NF-κB activation, our studies do not support a role for this protein in EGFR-induced NF-κB activation.

The engagement of the epidermal growth factor receptor by its cognate ligand results in the generation of a number of intracellular signals (1). The initial changes induced by ligand binding are receptor dimerization, activation of the kinase ac-tivity of the receptor, and autophosphorylation of the receptor on tyrosine residues (2,3). Autophosphorylation of the receptor results in the creation of docking sites for a number of secondary signaling proteins bearing specific protein interaction domains such as the Src homology 2 domain, which interact specifically with phosphorylated tyrosine residues (4). As a consequence of this interaction, these secondary signaling proteins may themselves become activated and trigger a number of downstream signals. These signaling cascades result in the activation of a number of transcription factors such as AP-1 and STATS (5).
The NF-B 1 family of transcription factors plays an important role in inflammatory responses (6). A diverse number of stimuli including cytokines such as TNF␣ and IL-1, UV irradiation, and lipopolysaccharide are known to activate NF-B. In unstimulated cells NF-B is sequestered in the cytoplasm by the IB family of proteins (7). Binding of IB to NF-B masks nuclear localization signals on NF-B and prevents its translocation to the nucleus (8). Stimulation of cells with a diverse array of stimuli results in phosphorylation of IB␣ on both serines 32 and 36. This results in the ubiquitination and degradation of IB␣, allowing NF-B to translocate to the nucleus and activate transcription (9 -12). Considerable progress has been made in our understanding of how the TNF receptor activates NF-B. The 55-kDa TNFR1 is thought to be the more important receptor type in the activation of NF-B. Binding of TNF to the TNFR1 results in trimerization of the receptor and the recruitment of adaptor proteins such as TRADD to the receptor (13). TRADD in turn recruits receptor interacting protein (RIP) and TRAF2 to the receptor (14,15). TRAF2 binds to the NF-B-inducing kinase (NIK) (16). This results in the activation of NIK, although the mechanism of this activation remains unclear. NIK phosphorylates and activates the IKKs (17), which appear to exist in a large multiprotein complex termed the signalosome (18,19). The IKKs in turn phosphorylate IB␣ on serines 32 and 36. NIK appears to selectively target IKK␣ (20,21), whereas MEKK1 and atypical protein kinase Cs may activate both IKK␣ and IKK␤ (19,22).
Protein kinase B/Akt also appears to be involved in NF-B activation induced by PDGF and TNF␣. Activation of Akt requires the phosphorylation of Akt on residues Ser-473 and Thr-308 (23). Akt is activated by both TNF␣ and PDGF. Recent studies have indicated NF-B activation induced by these cytokines is inhibited by both wortmannin (a PI 3-kinase inhibitor) and by a kinase inactive Akt mutant, while expression of a constitutively activated myristylated Akt mutant was sufficient for NF-B activation (24,25). Akt has been shown to enhance the degradation of IBs (26). Furthermore, activated Akt can associate with the IKK complex and may activate IKK␣ by phosphorylating it at Thr-23 (24,25).
The epidermal growth factor receptor has also been shown to activate NF-B. EGF has been shown to activate NF-B in smooth muscle cells, in A431 cells, in fibroblasts, and in several estrogen receptor-negative EGF-overexpressing breast cancer cell lines (27)(28)(29). A previous study has shown that EGF stimulation of A431 cells leads to degradation of IB␣ (28). This study also showed that wild type IB␣ transfected into COS cells undergoes degradation upon treatment with EGF, whereas an IB␣ S32/36 mutant fails to undergo degradation, suggesting that activation of the EGFR leads to phosphorylation of IB␣ on serines 32 and 36 and subsequent activation of NF-B in a manner similar to the TNF receptor. A requirement for intracellular free calcium was also noted in this study. However, the proximal signals generated by the EGFR that lead to the activation of NF-B remain unknown.
In this study we present evidence showing that the EGFR activates NF-B by mechanisms similar to but not identical with the TNF receptor. We identify components of the signalosome involved in EGFR-induced NF-B activation and suggest a mechanism of NF-B activation at the level of protein-protein interaction. We show that the EGFR-induced NF-B activation involves some of the proteins previously identified as key proximal elements in the TNFR pathway of NF-B activation, but unlike for the TNFR we find that the protein kinase Akt is not involved in EGFR-mediated activation of NF-B in the cell types we have studied.
Transfections were done using the calcium phosphate technique using standard protocols, and expression of transfected genes was confirmed by Western blotting. For transient transfection experiments, cells were harvested 24 -48 h after transfection. For stable transfection, an HA-tagged IB␣ S32/36 mutant cloned into pcDNA 3.1 or the empty vector were transfected into MDA-MB-468 cells and mass populations of zeocin-resistant cells were screened by Western blotting.
Western Blotting and Immunoprecipitation-Standard protocols were used for immunoprecipitation and Western blotting (30,31). Quantitation of proteins was performed by using a Bio-Rad detergentcompatible protein assay kit. For immunoprecipitation, cells were lysed in a modified radioimmune precipitation buffer (150 mM NaCl, 1% Nonidet P-40, 0.25% deoxycholate, 1 mM EGTA, 1 mM NaF, 50 mM Tris-HCl, 1 mM phenylmethylsulfonyl fluoride, and 2 mM orthovanadate), and equal amounts of protein were incubated with the primary antibody for 90 min. Protein A-agarose or Protein G-agarose beads were subsequently added to the lysates and incubated overnight at 4°C. The beads were subsequently washed and solubilized in SDS sample buffer and then boiled and analyzed by SDS-polyacrylamide gel electrophoresis, followed by transfer to nitrocellulose. Western blots were developed with ECL reagents (Amersham Pharmacia Biotech). In experiments where cell lysates were examined directly, cells were lysed in SDS sample buffer, boiled, and subjected to SDS-polyacrylamide gel electrophoresis followed by Western blotting.
Electrophoretic Mobility Shift Assay-Nuclear extracts were prepared as previously described (32). Double-stranded oligonucleotide containing the consensus sequence for binding of NF-B was filled by the Klenow fragment with [␣-32 P]dCTP and the three other nonradiolabeled dNTPs. The sequence of the probe is as below.

SEQUENCE 1
For binding reactions, nuclear extracts (5-10 g of protein) were incubated with 1 g of poly(dI-dC) in 10 mM Tris-HCl buffer (pH 7.5) containing 0.1 mM EDTA, 3 mM dithiothreitol, 100 mM KCl, and 1% glycerol for 15 min at room temperature. The 32 P-labeled oligonucleotide probe (20,000 cpm) was then added and incubated for 30 min at room temperature. In competition assays, excess oligonucleotide competitor was preincubated with nuclear extracts for 15 min at room temperature. In supershift assays, antibodies were added 30 min before 32 P-labeled oligonucleotide. At the end of the incubation, 10ϫ DNA loading buffer was added and the sample was electrophoresed in a native 4.5% polyacrylamide gel at a constant voltage (10 V/cm). Gels were dried and exposed to x-ray film with intensifying screens overnight at Ϫ80°C.
Luciferase Assays-1 ϫ 10 5 R1hER cells were transfected using the calcium phosphate method. A dual-luciferase reporter assay system was used according to the instructions of the manufacturer (Promega). 30 ng of a pRL vector was cotransfected along with 300 ng of a SV40Bluc reporter plasmid (42) and 4 g of either empty vector or a dominant negative NIK mutant. Firefly luciferase activity was measured in a luminometer and normalized on the basis of Renilla luciferase activity.

A High Level of EGF Receptor Expression Is Optimal for EGF-induced NF-B
Activation-It has been reported that EGF induces NF-B in aortic smooth muscle cells (27), in A431 cells, transiently in mouse embryo fibroblasts (28), and in several breast cancer cell lines (29), but not in human omental microvascular endothelial cells (33). The activation of NF-B in the various cell types may be a consequence of high level of EGF receptor expression. To determine whether the level of EGF receptor expression influences the activation of NF-B, we used Rat-1 fibroblasts and compared them to R1hER cells, which are Rat-1 fibroblasts expressing a high level of the EGFR. Electrophoretic mobility shift assays showed that EGF stimulation failed to activate NF-B in Rat-1 cells (Fig. 1A,  lanes 1 and 2). However, exposure of R1hER cells to EGF resulted in a robust activation of NF-B (Fig. 1A, lanes 3 and  4). Furthermore, in R1hER cells we also detected constitutive activation of NF-B compared with untransfected Rat-1 cells. R1hER cells express about 7.5 ϫ 10 5 EGF receptors, which is a nearly 7-fold increase over untransfected cells (34). There is constitutive phosphorylation of the EGFR in R1hER cells, suggesting that the receptor is active in these cells even in the absence of ligand. Interestingly, the increase in the level of EGF receptor expression leads to a selective rather than a general amplification of intracellular signaling pathways. For example, ERK activation in response to EGF (100 ng/ml) was not increased in R1hER cells compared with Rat-1 cells (Fig.  1D). This observation also holds true at lower EGF concentrations (ranging from 0.1 to 1 ng/ml), which result in submaximal levels of ERK activation in both cell lines (data not shown). We did not detect an increase in ERK activation in R1hER cells compared with Rat-1 cells at any EGF concentration tested.
To determine whether the relationship between EGFR levels and NF-B activation was a consistent finding between cell types, we studied EGF-mediated NF-B activation in two breast cancer cell lines expressing different levels of the EGFR. The MDA-MB-468 cell lines expresses a high level of EGF receptors (1-3 ϫ 10 6 ), while the MCF7 cell line expresses a lower level of EGFR, about 2.4 ϫ 10 3 (35). EGF stimulation failed to induce a significant activation of NF-B in MCF7 cells, whereas a robust activation of NF-B was observed in MDA-MB-468 cells (Fig. 1B). On the other hand, activation of ERKs in response to EGF was similar in MCF7 and MDA-MB-468 cells (data not shown). Thus, our data suggest that, in the cell types used, a high level of EGFR expression is optimal for the activation of NF-B.
The NF-B family is composed of several subunits including, p50, p65, c-Rel, and Rel B. Supershift experiments were performed on MDA-MB-468 cells to identify which NF-B subunits were induced by EGF. In agreement with previous studies (28), we found that NF-B in MDA-MB-468 cells stimulated with EGF consists of p65 and p50 (Fig. 1C).

Expression of a Dominant Negative IB␣ S32/36 Mutant Blocks EGFR-induced NF-B Activation-NF-B
is regulated by its interaction with a group of cytoplasmic inhibitory proteins termed IB. The major species of this family of proteins is designated IB␣. Phosphorylation of IB␣ on serines 32 and 36 targets it for degradation via the ubiquitin-proteasome pathway. This releases NF-B, which translocates to the nucleus and activates transcription. It has been shown previously that EGF stimulation of cells results in the degradation of IB␣. To test whether an IB␣ mutant that cannot be phosphorylated at Ser-32/36 would block EGFR-induced NF-B activation, we stably transfected MDA-MB-468 cells with this mutant in a plasmid expression vector. Mass populations of zeocin-resist- ant cells were pooled and tested for expression of the HAtagged IB␣ mutant by Western blot analysis ( Fig. 2A). Having established that these cells expressed the transfected mutant, electrophoretic mobility shift assays were performed and compared with MDA-MB-468 cells expressing the empty vector. As can be seen in Fig. 2B, expression of a dominant negative IB␣ mutant blocks EGFR-induced NF-B activation, demonstrating that the EGFR utilizes elements of the canonical pathway of NF-B activation.
The EGFR Associates with RIP but Not TRADD-The activation of NF-B by TNF is mediated by a specific signalosome composed of distinct proteins. TRADD is an adaptor protein, which associates with the TNFR when the receptor is activated (13). It has a domain of about 80 amino acids in the carboxyl terminus, which is termed the death domain. The death domain of TRADD binds to a similar death domain on the TNFR. TRADD then recruits RIP and also TRAF2 to the TNFR (14,15). RIP may also bind directly to the TNFR, but this has been reported to be a weak interaction. To identify the proteins constituting the receptor tyrosine kinase signalosome at the EGF receptor, we transfected EGFR cDNA in a plasmid expression vector along with Myc-tagged TRADD into 293 EBNA cells to determine whether the two proteins would associate into a physical complex. Cells transfected with the two plasmids were subjected to immunoprecipitation with an EGFR antibody followed by immunoblotting with a Myc antibody. We failed to detect Myc-TRADD in EGFR immunoprecipitates in repeated experiments (Fig. 3A). We also failed to detect an interaction between the EGFR and Fas-associated death domain protein (FADD), another protein that interacts with the TNFR and Fas receptors and is involved in apoptosis (data not shown). However, when we cotransfected the EGFR with Myc-tagged RIP, which is a key protein in TNFR-induced NF-B activation, we detected a physical association between the two proteins. This interaction was detected by immunoprecipitating with an EGFR antibody and immunoblotting with Myc antibodies (Fig.  3B, middle panel). We estimate that 5-10% of the transfected RIP becomes associated with the EGFR in these cells as shown in Fig. 3C. The association can also be detected by immunoprecipitating with a Myc antibody and immunoblotting with the EGFR antibody (Fig. 3D). The kinase activity of the EGFR is required for association with RIP since a kinase-inactive EGFR mutant fails to bind RIP (Fig. 3D). Also, addition of tyrphostin AG 1478, a specific inhibitor of the EGFR kinase, blocks the association between wild type EGFR and RIP (Fig. 3D). 293 EBNA cells express very low levels of endogenous EGFR, and we did not detect EGF-induced ERK activation in these cells (data not shown). We found that the endogenous level of EGFR expression in 293 EBNA cells is not sufficient to mediate a detectable association with RIP. However, when we transfected increasing amounts of EGFR DNA into these cells along with a constant amount of RIP, an interaction between the two proteins can be detected with increasing levels of EGFR expression (Fig. 3B). This suggests that RIP may play a role in EGFmediated NF-kB activation since RIP is known to be a key component of TNFR-NF-B signaling, and also correlates with our observation that in certain cell types a high level of EGFR expression is required to activate NF-B. It should be noted that addition of EGF to cells did not increase the EGFR-RIP interaction in these cells. This is because ectopic expression of the EGFR leads to a substantial constitutive increase in the level of EGF receptor tyrosine phosphorylation in 293 EBNA cells as shown in Fig. 3F, where addition of EGF did not result in any further increase in tyrosine phosphorylation of the EGFR. This suggests maximal activation of the receptor in these cells even in the absence of ligand. This phenomenon has also been noted previously in 293T cells (36).
A physical association between the EGFR and endogenous RIP in untransfected MDA-MB-468 cells can also be detected. Cell lysates were immunoprecipitated with a EGFR antibody and immunoblotted with a RIP antibody. As can be seen in Fig.  4A, RIP coimmunoprecipitates with the EGFR in these cells. It should be noted that MDA-MB-468 cells express high levels of the EGFR and that there is considerable basal activation of the receptor in these cells even in the absence of exogenous EGF. This is shown in Fig. 4B, which demonstrates a significant autophosphorylation of the receptor in the absence of EGF. Although we do not see an increase in EGFR-RIP association when these cells are exposed to EGF (data not shown), consistent with our earlier observation in 293 cells, treating the cells with tyrphostin AG 1478 reverses the association between the two proteins (Fig. 4A). Taken together, this indicates that the protein complex responsible for signaling by the two receptors, namely, TNFR and the EGFR, are distinct.
The EGFR Associates with NIK-We next addressed the issue of whether NIK coimmunoprecipitates with the EGFR in vivo since NIK appears to be downstream of RIP in the TNFR NF-B pathway and a dominant negative NIK mutant blocks RIP-induced NF-B activation (16). 293 EBNA cells were cotransfected with FLAG-tagged NIK cDNA along with the EGFR. Concurrent with our observation with RIP, NIK associated with the EGFR with increasing levels of EGFR expression (Fig. 5A). This interaction was detected by immunoprecipitating cell lysates with an EGFR antibody and immunoblotting with anti-FLAG antibodies (Fig. 5A, middle panel), and also by immunoprecipitating with anti-FLAG antibodies and immunoblotting with an EGFR antibody (Fig. 5B). The relationship between increasing level of EGFR expression and association with NF-B-inducing protein seems to be more linear with Nik than with RIP. Similar to RIP, addition of EGF did not increase the EGFR-NIK association in these cells. In the absence of exogenous EGFR, we did not detect coprecipitation of the two proteins suggesting that the endogenous level of EGFR expression is too low in 293 EBNA cells to allow a significant interaction between the two proteins (Fig. 5, A and B).
The EGFR May Form Multiprotein Complexes with RIP and NIK-It is increasingly apparent that signal transduction involves the physical assembly of individual components of a pathway into multiprotein complexes. In the case of NF-B, a number of proteins such as NIK, IKKs, and IB␣ exist in a complex termed the signalosome. To investigate the formation of complex formation between the EGFR, RIP, and NIK, we cotransfected Myc-tagged RIP, Flag-tagged NIK, and the EGFR along with appropriate controls into 293 EBNA cells. The data shown in Fig. 6 suggest that multiprotein complexes may form among the three proteins. This was demonstrated by immunoprecipitating with an EGFR antibody and immunoblotting with a Myc antibody to detect the presence of RIP ( 6A). The same blot was stripped and reprobed with anti-FLAG antibodies to detect the presence of NIK (Fig. 6C). Similarly, in another independent experiment, anti-FLAG antibodies were used to immunoprecipitate the complex which was analyzed by immunoblotting with an EGFR antibody (Fig. 6B). This blot was later stripped and reprobed with a Myc antibody (Fig. 6D) to identify the presence of RIP. Thus, we detect both Rip and Nik in EGFR immunoprecipitates and also find both EGFR and Rip in Nik immunoprecipitates. This experiment suggests that the signaling complex at the EGFR may involve all three proteins. However, we have found that RIP can bind to NIK in the absence of exogenous EGFR in 293 EBNA cells, which express a low level of endogenous EGFR. Thus, finding both RIP and NIK in EGFR immunoprecipitates could also be a consequence of pairwise interactions between the three proteins. Our coimmunoprecipitation studies are suggestive of a direct physical A Dominant Negative NIK Mutant Blocks EGFR-mediated NF-B Activation-NF-B activation by cytokines involves signal transduction cascades, which involve a number of key intermediate signaling proteins. NIK was originally identified as a TRAF2-interacting protein (16). When overexpressed, NIK activates NF-B and kinase-inactive mutants of NIK behave as dominant negative inhibitors that block NF-B activation by TNF and RIP.
The observation that NIK forms a complex with the EGFR suggests that NIK may play a role in EGFR-induced NF-B activation. We tested this hypothesis in R1hER cells. As noted previously, EGF induces a substantial activation of NF-B in these cells. There is also a significant constitutive activation of NF-B in these cells compared with untransfected Rat-1 fibroblasts. A NIK kinase domain mutant (NIK-K429A/K430A) was transfected into R1hER cells along with an NF-B promoter linked to a luciferase construct along with appropriate controls. As can be seen in Fig. 7A, expression of a dominant negative NIK mutant significantly blocks the ability of the EGFR to activate NF-B-dependent transcription. A dominant negative NIK mutant blocks both the constitutive as well as the EGFdependent increase in NF-B activity in R1hER cells. This suggests that NIK is required for EGFR-induced NF-B activation. Fig. 7 (B and C) shows EGFR tyrosine phosphorylation and protein levels in transfected cells shown in Fig. 7A. Although expression of a dominant negative NIK blocks EGFRinduced NF-B activation, it does not influence EGFR expression or tyrosine phosphorylation.
Akt Does Not Play a Role in EGF-induced NF-B Activation-Previous studies have demonstrated that protein kinase B/Akt is involved in NF-B activation by cytokines such as PDGF and TNF (24 -26). The epidermal growth factor is also known to activate Akt. To investigate the role of Akt in EGFRinduced NF-B activation, the following experiments were undertaken. First, we confirmed that EGF stimulation leads to Akt activation in R1hER cells using a phosphospecific (Ser-473) antibody (Fig. 8A). However, EGF-induced Akt activation can also be detected in Rat-1 cells and since these cells do not appear to activate NF-B significantly in response to EGF, activation of Akt may not be an intermediary step in the activation of NF-B.
To address this definitively, we examined the effects of the PI 3-kinase inhibitor wortmannin since activation of Akt is mediated primarily through the PI 3-kinase pathway. The data presented in Fig. 8B clearly show that incubation of cells with wortmannin had no effect on EGF-induced NF-B activation in R1hER cells. Wortmannin also failed to block EGFR-induced NF-B activation in MDA-MB-468 cells (data not shown). Although wortmannin had no effect on EGF-induced NF-B activation, it completely blocked EGF-induced Akt activation in R1hER cells (Fig. 8C) as determined by immunoblotting EGFtreated cell lysates with phosphospecific Akt antibodies. A similar result was evident in MDA-MB-468 cells using both a phospho-Ser-473 antibody as well as a phospho-Thr-308 antibody (Fig. 8D). The observation that blocking EGF-induced Akt activation does not hinder EGF-induced NF-B activation argues strongly against a role of Akt in EGF-induced NF-B activation. However, a recent report has shown that EGFinduced NF-B activation can be blocked by inhibition of PI 3-kinase (29). The apparent discrepancy between our work and this study may be explained by the different time points of activation investigated by the two groups. The previous study has observed the effect at much later time points than we have investigated.

DISCUSSION
The response of mammalian cells to extrinsic stimuli involves the activation of signaling cascades with discrete specificity. There is increasing evidence that generation of a specific signal involves the formation of specific multiprotein complexes (37). The formation of such complexes involves selective recruitment of subcomplexes of signaling proteins to an activated receptor. This is followed by activation of signaling proteins through mechanisms such as phosphorylation and/or other protein-protein interaction. Identification of the composition of signal-specific multiprotein complexes is the initial step in understanding the mechanisms that underlie the generation of a specific signal. The EGF receptor is a prototypical tyrosine kinase that assembles a multiprotein complex to induce mitogenic signaling. In this work we have identified a novel receptor tyrosine kinase (EGF receptor)-bound complex, which mediates transactivation of NF-B.
Activation of the EGF receptor results in activation of a number of transcription factors including AP-1, STATs, and NF-B (28,38,39). Although EGF stimulation of cells has previously been shown to lead to degradation of IB␣ and NF-B activation, the proximal signaling events that lead to EGFR-induced activation of NF-B have not been elucidated. In this study we demonstrate an interaction between RIP and NIK, which are two key components of TNF-induced NF-B signal, and the EGFR. Our studies suggest a role for these proteins in NF-B activation by the EGFR and indicate that mechanisms of NF-B activation by the EGFR and TNF␣ are similar, but not identical.
RIP was initially identified as a CD95-interacting protein (40). Overexpression of RIP activates NF-B in cells (41,42). Jurkat cell lines, which lack RIP, are unable to activate NF-B in response to TNF␣. Introduction of RIP into these cells restores the ability of TNF to activate NF-B (42). Finally, cells from mice lacking RIP are unable to activate NF-B in response to TNF, establishing RIP as a key component of the TNF-induced NF-B signal (43). The mechanism of action of RIP remains obscure, but a recent study suggests that RIP binds to NEMO, another component of the TNFR NF-B pathway (44). We show that RIP physically associates with the EGFR when the two proteins are coexpressed in cells, and that increasing the level of EGF receptor expression in cells promotes this association. The kinase activity of the EGFR appears to be required for EGFR-RIP association since a kinaseinactive EGFR fails to bind RIP. We also show an association between the EGFR and RIP in untransfected cells, and show that this association is reversed by tyrphostin AG 1478, a specific inhibitor of the EGFR kinase. It is intriguing that the kinase activity of the receptor is required for its association with RIP. We have not observed RIP to be tyrosine-phosphorylated, and it is possible that other protein(s) may be required for this interaction in a phosphorylation-dependent manner. The finding that the EGFR-RIP association requires the kinase activity of the receptor also supports our hypothesis that the EGFR-RIP association is an important step in NF-B activation since the EGFR activates NF-B in a ligand-dependent manner.
RIP is recruited to the activated TNFR via an adaptor protein, TRADD (14). TRADD and RIP interact with each other via their death domain, which is a conserved sequence of amino acids also present in the TNF receptor. Although the EGFR associates with RIP, it does not bind to TRADD (or to another death domain-bearing protein, FADD). This suggests RIP may associate directly with the EGFR through a heterologous interaction or via some other protein(s) of the complex. Although the mechanism of action of RIP is not well understood, recruitment of RIP to the TNFR is likely to be a key step in NF-B signaling. Similarly, RIP binding to the EGFR may be a key step in the activation of NF-B by EGF.
NIK was originally identified as a TRAF2-binding protein.

Ectopic expression of NIK in cells results in NF-B activation.
A dominant negative NIK mutant blocks NF-B activation by a number of stimuli including TNF␣, IL-1, and RIP, suggesting that NIK functions downstream of RIP in the signaling cascade. Although the mechanism of activation of NIK itself remains unclear, NIK appears to activate the IKK enzymes leading to the phosphorylation of IB␣ on both of its regulatory serines at residues 32 and 36, which leads to IB degradation. We find that NIK associates with the EGFR when the two proteins are coexpressed and, as in the case of RIP, increasing the level of EGFR in cells promotes the association between NIK and the EGFR. We were unable to demonstrate an interaction between the EGFR and NIK in untransfected cells because of lack of a good NIK antibody. We show that a dominant negative NIK mutant blocks EGFR-induced NF-B activation, demonstrating that the interaction between the two proteins has functional significance.
Our data also suggest that EGFR, NIK, and RIP may form multiprotein complexes, indicating that EGFR-induced NF-B activation also involves the formation of multiprotein complexes like the signalosome described in TNFR-mediated NF-B signaling (18,44). We also show that transfection of a dominant negative IB␣ mutant that cannot be phosphorylated on serine at residues 32 and 36 blocks EGFR-induced NF-B activation, demonstrating again that the EGFR uses signals that are similar but not identical to the TNF receptor to activate NF-B.
Previous studies have invoked a role for the protein kinase Akt in NF-B activation by both the PDGF receptor and TNF␣. Both cytokines activate Akt primarily through activation of PI 3-kinase. A constitutively active Akt mutant activates NF-B. Wortmannin, a PI 3-kinase inhibitor, blocks NF-B activation by both cytokines as does introduction of a dominant negative Akt mutant. Our study, however, does not support a role for Akt in NF-B activation by the EGFR. We find that, although wortmannin blocks EGF-induced phosphorylation of Akt in MDA-MB-468 and in R1hER cells, it has no effect on EGFinduced NF-kB activation. This suggests that, unlike the TNFR and the PDGF receptor, Akt activation is not required for EGFR-mediated NF-B activation, at least in the cell types we have studied. Our results are in agreement with another recent study, which shows that inhibiting Akt activation in human endothelial cells does not block NF-B activation in response to TNF or IL-1 (45).
Previous studies have indicated that EGF induces a slow activation of NF-B in primary smooth muscle cells, a transient activation of NF-B in primary embryo fibroblasts, and a more robust activation in A431 cells and in several endoplasmic reticulum-negative breast cancer cell lines that express high levels of the EGFR. Our experiments also suggest that a high level of EGFR expression is optimal for EGF-mediated NF-B activation. This hypothesis is supported by our finding that there is a correlation between the level of EGFR expression and the recruitment of key NF-B signaling proteins (RIP and NIK) to the EGFR. It is also important to note that EGF stimulation of cells with high levels of EGFR expression (which are common in human tumors) leads to a selective rather than a general amplification of signals. For example, EGF stimulation of both Rat-1 fibroblasts and R1hER cells leads to increases in ERK activation, whereas Rat-1 fibroblasts fail to activate NF-B. Similar results are seen when comparing MCF7 cells to MDA-MB-468 cells.
The EGF receptor generates mitogenic signals in cells and a large number of human tumors express high levels of the EGFR (46). In addition, increased expression of the EGFR correlates with the malignant phenotype. NF-B, in addition to its role in inflammatory responses, is also involved in the control of cell growth and transformation. EGF is known to induce expression of genes involved in cell growth and some of these genes, such as c-myc, contain NF-B promoter binding sites (47,48). It has also been shown that activated forms of Ras and Raf activate reporter genes controlled by NF-B sites (49,50). EGF is known to activate both Ras and Raf (3). A dominant negative IB␣ mutant has been shown to block Ras transformation and decreased levels of IB␣ promote transformation (51,52). EGF-induced NF-B activation has also been implicated in cyclin D1-dependent cell cycle progression (29).
Taken together, this suggests that EGFR-mediated NF-B activation is likely to influence cell growth and proliferation. We have elucidated the proximal steps in EGF-mediated activation of NF-B. Our study indicates that the activated EGFR recruits RIP and NIK to initiate the signaling cascade via IB␣ to localize NF-B to the nucleus for transactivation. Recent studies have shown that NF-B also plays a key role in regulating apoptosis (53)(54)(55)(56). This is of particular interest since cells such as A431 and MDA-MB-468, which express high levels of the EGFR and show the most robust NF-B activation are known to undergo apoptosis in response to EGF (57,58). Fur-ther studies are needed to elucidate the role of NF-B in EGFinduced apoptosis and growth control and are likely to be relevant to a better understanding of the biology of human tumors as well as EGF-induced normal growth. Our identification of a novel complex at the EGF receptor distinct from the complex required for mitogenic signaling and from the complex assembled by the TNFR opens up new lines of investigation pertaining to EGF receptor-mediated signaling.