Tumor necrosis factor-induced nuclear factor kappaB activation is impaired in focal adhesion kinase-deficient fibroblasts.

Focal adhesion kinase (FAK) is widely involved in important cellular functions such as proliferation, migration, and survival, although its roles in immune and inflammatory responses have yet to be explored. We demonstrate a critical role for FAK in the tumor necrosis factor (TNF)-induced activation of nuclear factor (NF)-kappaB, using FAK-deficient (FAK-/-) embryonic fibroblasts. Interestingly, TNF-induced interleukin (IL)-6 production was nearly abolished in FAK-/- fibroblasts, whereas a normal level of production was obtained in FAK+/- or FAK+/+ fibroblasts. FAK deficiency did not affect the three types of mitogen-activated protein kinases, ERK, JNK, and p38. Similarly, TNF-induced activation of activator protein 1 or NF-IL-6 was not impaired in FAK-/- cells. Of note, TNF-induced NF-kappaB DNA binding activity and activation of IkappaB kinases (IKKs) were markedly impaired in FAK-/- cells, whereas the expression of TNF receptor I or other signaling molecules such as receptor-interacting protein (RIP), tumor necrosis factor receptor-associated factor 2 (TRAF2), IKKalpha, IKKbeta, and IKKgamma was unchanged. Also, TNF-induced association of FAK with RIP and subsequent association of RIP with TRAF2 were not observed, resulting in a failure of RIP to recruit the IKK complex in FAK-/- cells. The reintroduction of wild type FAK into FAK-/- cells restored the interaction of RIP with TRAF2 and the IKK complex and allowed recovery of NF-kappaB activation and subsequent IL-6 production. Thus, we propose a novel role for FAK in the NF-kappaB activation pathway leading to the production of cytokines.


Focal adhesion kinase (FAK) is widely involved in important cellular functions such as proliferation, migration, and survival, although its roles in immune and inflammatory responses have yet to be explored. We demonstrate a critical role for FAK in the tumor necrosis factor (TNF)-induced activation of nuclear factor (NF)-B, using FAK-deficient (FAK؊/؊) embryonic fibroblasts. Interestingly, TNF-induced interleukin (IL)-6 production was nearly abolished in FAK؊/؊ fibroblasts, whereas a normal level of production was obtained in FAK؉/؊ or FAK؉/؉ fibroblasts. FAK deficiency did not affect the three types of mitogen-activated protein kinases, ERK, JNK, and p38. Similarly, TNF-induced activation of activator protein 1 or NF-IL-6 was not impaired in FAK؊/؊ cells. Of note, TNF-induced NF-B DNA binding activity and activation of IB kinases (IKKs) were markedly impaired in FAK؊/؊ cells, whereas the expression of TNF receptor I or other sig-
naling molecules such as receptor-interacting protein (RIP), tumor necrosis factor receptor-associated factor 2 (TRAF2), IKK␣, IKK␤, and IKK␥ was unchanged. Also, TNF-induced association of FAK with RIP and subsequent association of RIP with TRAF2 were not observed, resulting in a failure of RIP to recruit the IKK complex in FAK؊/؊ cells. The reintroduction of wild type FAK into FAK؊/؊ cells restored the interaction of RIP with TRAF2 and the IKK complex and allowed recovery of NF-B activation and subsequent IL-6 production. Thus, we propose a novel role for FAK in the NF-B activation pathway leading to the production of cytokines.
Focal adhesion kinase (FAK) 1 is a non-receptor protein-tyrosine kinase implicated in controlling cellular responses to cell surface integrin with cell spreading and migration. Cellular focal adhesion contains structural and signaling proteins that in many cases are modified by tyrosine phosphorylation (re-viewed in Refs. [1][2][3]. FAK is phosphorylated not only by integrin signaling (4,5) but also by a variety of soluble growth factors including platelet-derived growth factor and vascular endothelial growth factor as well as growth hormone (see Refs. [6][7][8]. FAK is thus implicated to play important roles in signaling pathways initiated by integrins in cell migration, survival, and cell cycle regulation. In addition, FAK has also been shown to play an essential role in the survival of anchorage-dependent cells (10) or in antiapoptotic action during growth factor deprivation-induced apoptosis in human umbilical vein endothelial cells (11). Overexpression of FAK in many cells induces the constitutive activation of NF-B, which leads to the activation of survival genes as shown in human leukemic HL-60 cells, in which FAK protected cells from apoptosis caused by oxidative stress, etoposide, and ionizing radiation (12)(13)(14) or UV-induced apoptosis in Madin-Darby canine kidney cells (15). These findings imply that FAK generally has an antiapoptotic role in various cells. Furthermore, FAK triggers rapid cell cycle progression via activation of protein kinase C isoforms and cyclins (16). FAK is thus phosphorylated by various stimuli and is involved in cytoplasmic signaling downstream of a variety of cell surface receptors. However, to date, no reports have described the role of FAK in inflammatory and immune responses, whereas proline-rich tyrosine kinase 2, a related adhesion focal tyrosine kinase, was found to be activated by TNF or by ultraviolet irradiation (17).
Mechanisms of FAK activation are multiple, and other tyrosine-phosphorylated proteins such as proline-rich tyrosine kinase 2 are present in the cells (3,5,18). Thus, we investigated embryonic fibroblasts from FAK-deficient mice to study the role of FAK because FAK deficiency in mice was embryonic lethal as a result of defective developmental gastrulation events with deficits in cell migration (18 -20). Our focus was the responses of FAKϪ/Ϫ fibroblasts to TNF-␣. TNF-␣ is a potent inducer of IL-6 expression in fibroblasts, and we examined whether FAKϪ/Ϫ fibroblasts respond normally to TNF-␣ and produce cytokines. We found that FAKϪ/Ϫ fibroblasts responded poorly to stimulation with TNF-␣, and therefore we examined the underlying mechanism of their unresponsiveness to TNF-␣. We present evidence for the first time that a functional FAK molecule is required for IL-6 production by TNF-␣. More importantly, we have uncovered the role of FAK in TNF-␣-mediated NF-B activation through its association with receptor-interacting protein (RIP), a serine/threonine kinase.
Cell Culture and IL-6 Assay-Heterozygous FAKϩ/Ϫ and homozygous FAKϪ/Ϫ embryonic fibroblasts were originally established by Ilic and co-workers (19,20) and were provided to us through Dr. T. Mimura (Department of Allergies, Tokyo University School of Medicine; Ref. 21). These cells were maintained in Dulbecco's modified Eagle's medium (Nissui Seiyaku, Tokyo, Japan) supplemented with 10% heat-inactivated fetal bovine serum, 4 mM glutamine, 100 units/ml penicillin G, and 100 g/ml streptomycin. The absence of FAK in the FAKϪ/Ϫ fibroblasts was confirmed by immunoblot analysis using anti-FAK monoclonal antibody (Transduction Laboratories, Lexington, KY; see Fig. 6). Normal FAKϩ/ϩ mouse fibroblasts were obtained from the skin of newborn BALB/c mice by trypsinization. For the measurement of secreted IL-6, cells were incubated with or without TNF-␣ (10 ng/ml) in Dulbecco's modified Eagle's medium containing 1% fetal bovine serum for 12-24 h. IL-6 levels in the culture supernatants were determined using a commercial ELISA kit (BIOSOURCE). All samples were assayed at least in duplicate.
Transient Transfection and Luciferase Assay-HA-tagged FAK cDNA and mutant FAK cDNA subcloned into pRcCMV were originally constructed by Dr. Steven K. Hanks (22) and kindly provided to us (12). Plasmid DNAs were transfected into FAKϩ/Ϫ and FAKϪ/Ϫ fibroblasts using LipofectAMINE TM2000 (Invitrogen). Final amounts of transfected DNA for 24-well plates were adjusted to 1 g with empty vector, pRcCMV. 0.5 g of pRcCMV-HA-FAK and pRcCMV-HA-FAK (K454R) was cotransfected with 0.01 g of pRL-TK (Promega Co. Japan, Tokyo, Japan) and 0.1 g of pNF-B-Luc (Invitrogen). At 48 h after transfection, cells were harvested, and the luciferase activities were measured with a Lumat LB9501 (Bertold Japan, Tokyo, Japan). The efficiency of transfection was normalized with sea pansy luciferase activity as described elsewhere (23).
Immunoprecipitation and Immunoblot Analysis-Immunoprecipitation and immunoblot analysis were performed as described previously (23,24). In brief, harvested cells were lysed in lysis buffer (10 mM Tris-HCl, pH 7.4, 158 mM NaCl, 1% Triton X-100, 1% sodium deoxycholate, 1 mM EGTA, 1 mM Na 3 VO 4 , 2 g/ml aprotinin, and 2 g/ml leupeptin) on ice and cleaned by centrifugation to obtain whole cell extracts. Aliquots (250 g) of cell lysate were mixed with protein G-Sepharose (Amersham Biosciences) and various antibodies overnight at 4°C. The immune complexes were precipitated by centrifugation, washed five times with lysis buffer, and boiled in Laemmli sample buffer. Boiled samples were separated by SDS-PAGE, and the proteins were transfected to nitrocellulose membranes. Immunoblotting was performed with various antibodies and horseradish peroxidase-conjugated anti-rabbit or anti-mouse IgG antibody and visualized using the enhanced chemiluminescence Western blotting detection system (Amersham Biosciences).
Flow Cytometry-Flow cytometry using a fluorescence-activated cell sorter scan was done as follows. Cells were incubated with PE-conjugated rat anti-mouse TNFRI antibody (CD120a; IgG2a isotype; Immunotech, Beckman-Coulter, Marseilles, France) and the isotype-matched mouse control IgG, followed by PE-conjugated anti-mouse antibody. Stained cells were analyzed using FACSCalibur (Becton Dickinson).
In Vitro Kinase Assay-Immunoprecipitates obtained with anti-IKK␥ antibody were washed twice with lysis buffer and three times with kinase buffer (25 mM Hepes-NaOH, pH 7.5, 20 mM MgCl 2 , 20 mM ␤-glycerophosphate, 0.1 mM Na 3 VO 4 , 2 mM dithiothreitol, and 20 mM p-nitrophenylphosphate). The kinase reaction in 20 l of kinase buffer including 10 M ATP and [␥-32 P]ATP was carried out with 1 g of GST-IB␣ (amino acids 1-60) as a substrate for 20 min at 30°C. In some experiments, mutant GST-IB␣⌬C (S32A, S36R) was used instead of GST-IB␣. Samples were resolved by 15% polyacrylamide gel electrophoresis, and phosphorylated GST-IB␣ was visualized by autoradiography.
Reconstitution of FAK by Adenoviral Infection-Cells were plated at 1 ϫ 10 6 cells/ml in 6-cm culture plates and infected with adenovirus encoding ␤-galactosidase (Adv-LacZ) or HA-tagged wild type FAK (Adv-FAK) for 24 h at an optimal concentration of virus (25) with 1 ϫ 10 2 virions/cell (i.e. a multiplicity of infection of 100).
Electrophoretic Mobility Shift Assay (EMSA)-EMSA was carried out as described previously (23,24). The consensus double-strand oligodeoxynucleotide probes for NF-B, AP-1, and NF-IL-6 (Santa Cruz Biotechnology) were radioactively labeled using [␥-32 P]ATP and T4 polynucleotide kinase with standard procedures. Then, 10 g of nuclear protein prepared from cells was incubated with ␥-32 P-labeled doublestranded oligonucleotide probe. The binding reaction was carried out at room temperature for 30 min in a total volume of 25 l. Bound complexes were separated on 5% gel electrophoresis in TGE (tris-glycine-EDTA) buffer, dried, and visualized by autoradiography.

Defective TNF-␣-induced IL-6 Production in the FAKϪ/Ϫ
Fibroblasts-It is well documented that FAK plays important roles in the integrin signaling in cell migration, cell survival, or cell cycle regulation. We speculated that FAK might also play a role in cytokine signaling because in our preliminary studies, cytokine production was modulated in the FAK-transfected cells. In order to study the role of FAK, we obtained FAKϩ/Ϫ and FAKϪ/Ϫ embryonic fibroblasts from FAK-deficient mice, which were originally established by Ilic and co-workers (19,20). First, we examined TNF-␣-induced IL-6 production by the FAKϩ/Ϫ and FAKϪ/Ϫ fibroblasts as well as normal mouse embryonic skin fibroblasts (FAKϩ/ϩ cells). TNF-␣ induced marked IL-6 production in both FAKϩ/ϩ and FAKϩ/Ϫ fibroblasts, whereas only marginal IL-6 production (one-fifth or one-sixth that of FAKϩ/ϩ fibroblasts at 24 h) was observed by the FAKϪ/Ϫ fibroblasts, as shown in Fig. 1A. RT-PCR analysis also indicated that virtually no IL-6 mRNA was expressed in FAKϪ/Ϫ cells, whereas a substantial level of IL-6 mRNA was expressed in the FAKϩ/Ϫ and FAKϩ/ϩ cells, suggesting some critical role for FAK (Fig. 1B). This observation prompted us to further explore the role of FAK in the TNF-␣-induced signaling pathway. Because significant expression of IL-6 in response to TNF-␣ was seen in FAKϩ/Ϫ as well as wild type normal FAKϩ/ϩ fibroblasts, we used FAKϩ/Ϫ cells and FAKϪ/Ϫ cells in the subsequent studies.
FAK Has No Effect on the TNF-␣-induced MAP Kinase Pathway or Activation of AP-1 and NF-IL-6 -TNF-␣ has been implicated in the activation of one or more MAP kinases in different cell types for the induction of IL-6 production (26). To explore the possible involvement of the MAP kinase family in FAK function, we determined whether three types of MAP kinases (ERK, c-Jun NH 2 -terminal kinase (JNK), and p38) are activated (i.e. phosphorylated or not) using antibodies against each phosphorylated form. Activated forms of all types of MAP kinases were detected equally in the FAKϪ/Ϫ cells as well as in the FAKϩ/Ϫ cells in response to TNF-␣ (data not shown), indicating that the MAP kinase activation pathway is not involved in the critical role of FAK.
AP-1, NF-IL-6, and NF-B binding elements are known to be involved in activating the IL-6 gene (27,28). AP-1 can be activated directly through phosphorylation by JNK, and the expression of AP-1 components is induced through JNK-and p38-dependent pathways (29). To assess the effect of FAK on AP-1 activation, we measured AP-1 activity with an EMSA, using oligonucleotide probes for AP-1. As shown in Fig. 2A, TNF-␣ induced marked AP-1 DNA binding activity in both FAKϩ/Ϫ and FAKϪ/Ϫ cells, which is consistent with the results on the activation of MAP kinase. Binding specificity for AP-1 was confirmed by the complete disappearance with nonradiolabeled DNA (Fig. 2B). Similarly, NF-IL-6 activation was equally observed in FAKϩ/Ϫ and FAKϪ/Ϫ cells (Fig. 2, C and  D), indicating that FAK is not directly involved in the activation of NF-IL-6.
Impairment of TNF-␣-induced NF-B Activation in FAKϪ/Ϫ Cells-Because TNF-␣ is a potent activator of NF-B, which is one of the transcription factors necessary for activating the IL-6 gene, we tested the activation of NF-B using an EMSA as well as the reporter gene assay. The EMSA indicated that TNF-␣ induced rapid and marked NF-B DNA binding activity within 15 min in the FAKϩ/Ϫ cells, whereas only low level induction was observed in FAKϪ/Ϫ cells (Fig. 3A). A supershift assay revealed that the NF-B DNA binding proteins are mostly p65 and less abundant p50 or c-Rel, as shown in Fig. 3B. This observation was also confirmed by the translocation of activated NF-B components to the nucleus, as shown by immunoblotting. That is, whereas p65, p50, and c-Rel proteins were detected at 15 min in nuclear extracts from the TNF-␣-stimulated FAKϩ/Ϫ cells, these proteins were not detected in FAKϪ/Ϫ cells (Fig. 3C). In addition, significant NF-B luciferase activity was detected in the FAKϩ/Ϫ cells, whereas only minimal NF-B activation was observed in FAKϪ/Ϫ cells (Fig. 3D). Thus, FAK appears to facilitate TNF-␣ signaling by modulating NF-B activation, which is not dependent on the MAP kinase pathway.
IKK Activity Is Severely Impaired in the FAKϪ/Ϫ Fibroblasts-Because the activation of IKK is a key step in the activation of NF-B, we examined whether the activation of IKK by TNF-␣ differs between FAKϩ/Ϫ and FAKϪ/Ϫ cells. It was found that activation by TNF-␣ was severely impaired in FAKϪ/Ϫ cells when the IKK complex was immunoprecipitated with anti-IKK␥ (NEMO) antibody, and IKK activity was measured as IB␣ phosphorylation using GST-IB␣⌬C as a substrate (Fig. 4A, top lane). No phosphorylation was observed when the GST-IB␣⌬C (S32A, S36R) mutant was used as a substrate (Fig. 4A, bottom lane). Furthermore, when the degradation of IB␣ after stimulation with TNF-␣ was examined by Western blotting, it was found to be reduced significantly in FAKϪ/Ϫ cells (Fig. 4B), confirming the impairment of IKK activation in FAKϪ/Ϫ cells. Because these results suggested that FAK plays essential roles in the activation of NF-B, we further evaluated the expression levels of several upstream molecules involved in TNF-␣-induced NF-B activation including TNFRI, RIP, TRAF2, IKK␣, IKK␤, and IKK␥. As shown in Fig. 4C, no significant differences between FAKϩ/Ϫ and with TNF-␣ (10 ng/ml) for the periods indicated, and nuclear extracts were prepared for an EMSA. Radiolabeled probe for the consensus AP-1 binding sequence (5Ј-CGCTTGATGACTCAGCCGGAA-3Ј) was obtained from Santa Cruz Biotechnology. Arrow indicates specific AP-1⅐DNA complex. B, for competition analysis, nuclear extracts were incubated with a 50-fold excess of unlabeled DNA probe. C, FAKϩ/Ϫ cells and FAKϪ/Ϫ cells were treated with TNF-␣ (10 ng/ml) for the periods indicated, and nuclear extracts were prepared for an EMSA. Radiolabeled probe for the consensus NF-IL-6 binding sequence (5Ј-TGCAGATTGCGCAATCTGCA-3Ј) was obtained from Santa Cruz Biotechnology. Arrow indicates specific NF-IL-6⅐DNA complex. D, for competition analysis, nuclear extracts were incubated with a 50-fold excess of unlabeled DNA probe. FAKϪ/Ϫ cells were observed. Therefore, the defective NF-B activation in FAKϪ/Ϫ cells was not due to altered expression of signaling molecules in the TNF-␣ signaling pathway. Because the TNFRI detected by Western blotting might be a non-glycosylated form of the TNFRI precursor, we also analyzed the expression of the mature form of TNFRI on the cell surface. As shown in Fig. 4D, similar levels of cell surface TNFRI were observed in both FAKϩ/Ϫ and FAKϪ/Ϫ cells, confirming that there were no significant differences between these two cell lines in terms of TNFRI expression.
Critical Interaction of FAK with RIP-Because IKK activity was severely impaired in FAKϪ/Ϫ cells, we examined whether FAK physically interacts with RIP, which is essential for the TNFRI-mediated activation of IKKs (31)(32)(33). Coimmunoprecipitation assay revealed that FAK associated with RIP in a TNF-␣-dependent manner in FAKϩ/Ϫ cells, but not in FAKϪ/Ϫ cells (Fig. 5A). Inversely, in a pull-down assay using anti-RIP antibody, RIP also coimmunoprecipitated with FAK, confirming the physical association of FAK and RIP (Fig. 5B). The TNF-␣-dependent association of TNFRI with FAK and with RIP was demonstrated clearly in FAKϩ/Ϫ cells but not in FAKϪ/Ϫ cells (Fig. 5C). In addition, TRAF2 antibody coimmunoprecipitated with RIP (Fig. 5D), which has been demonstrated by Hsu et al. (30), only in FAKϩ/Ϫ cells. Furthermore, we found that IKK␥, which is an essential component of the IB kinase complex (31), recruited RIP in response to TNF-␣ (Fig.  5E), which appears to be an important step in the activation of TRAF2 (33,34). It should be noted that a significant association of RIP with TRAF2 and IKK␥ was observed in FAKϩ/Ϫ cells but not in FAKϪ/Ϫ cells, thus demonstrating the critical role of FAK in the physical association of RIP, TRAF2, and IKK␥.

Rescue of the Deficient Phenotype in FAKϪ/Ϫ Fibroblasts by the Reintroduction of Wild
Type FAK cDNA-To confirm that the defect in the interaction of RIP with TNFRI, TRAF2, and the IKK complex in FAKϪ/Ϫ cells was due to the disruption of FAK function, we examined whether the above interactions could be rescued in FAKϪ/Ϫ cells by the forced expression of wild type FAK. We thus reintroduced wild type FAK into FAKϪ/Ϫ cells using an adenovirus vector harboring FAK cDNA. As shown in Fig. 6A, FAK expression, in contrast to the Lac Z control, restored TNF-␣-induced interaction between RIP and TNFRI as determined by coimmunoprecipitation assay. In addition, FAK expression also recovered the recruitment of TRAF2 and IKK␥ to RIP (Fig. 6, B and C).
To further ascertain the effect of FAK on the activation of NF-B, we determined the NF-B DNA binding activity with an EMSA. As shown in Fig. 7A, reintroduction of wild type FAK into the FAKϪ/Ϫ cells restored TNF-␣-induced NF-B DNA binding activity to the levels seen in FAKϩ/Ϫ cells (Fig. 3A). The restoration of NF-B activation by the reintroduction of FAK was confirmed using the NF-B reporter gene assay (Fig.  7B), as compared with that in Fig. 3D. Furthermore, introduction of FAK in FAKϪ/Ϫ cells resulted in a significant increase of IL-6 secretion and IL-6 mRNA expression (Fig. 7, C and D). It should be noted that mutated FAK (K454R), a kinase-defective mutant, did not restore TNF-␣-induced NF-B activation (Fig. 7B). Therefore, wild type FAK with intact kinase activity is necessary for the TNF-␣-triggered FAK-associated signal pathways.

DISCUSSION
In this study, we demonstrated that TNF-␣-induced NF-B activation and subsequent IL-6 gene activation were severely impaired in FAKϪ/Ϫ cells. These observations indicated that FAK plays a critical role in the TNF-␣-mediated signal transduction pathway. Because our results demonstrated that FAK is indispensable for the TNF-␣-induced IL-6 gene activation pathway, we focused on the signaling molecules with which FAK may interact. Whereas FAK is at a crossroad for multiple signaling pathways (32), neither its role in the production of cytokines nor its intervention in the signals leading to the activation of cytokine genes has been studied thus far. Sieg et al. (7) demonstrated that FAK integrates growth factors (platelet-derived growth factor and epidermal growth factor) and integrin signals to promote cell migration and suggested a role as a receptor-proximal link between the growth factor receptor and integrin signaling pathway. However, the role of FAK in the activation of cytokine genes had not been studied. Thus, this is the first paper describing significant involvement of FAK in cytokine production.
The biological effects of TNF-␣ are regulated through interaction with two distinct TNFRs, TNFRI (p55) and TNFRII (p75). Upon cell stimulation with TNF-␣, TNFRI recruits tumor necrosis factor receptor 1-associated death domain, an adapter protein that binds to TRAF2 and RIP, a serine/threonine kinase (33)(34)(35)(36). Interaction of tumor necrosis factor receptor 1-associated death domain with Fas-associated death domain leads to apoptosis through the activation of caspase cascades, whereas the interaction of tumor necrosis factor receptor 1-associated death domain with RIP and TRAF2 seems to be involved in both NF-B and JNK activation (33,34). In contrast, Lee et al. (37) indicated that TRAF2 is essential for the activation of JNK but not NF-B, suggesting that TRAF2 is at the bifurcation point of two kinase cascades leading to the activation of NF-B and JNK (38). Particular attention has been paid to activation of the downstream IKK complex, which FIG. 3. Impaired NF-B activation in FAK؊/؊ fibroblasts in response to TNF-␣. A, nuclear extracts were prepared from FAKϩ/Ϫ and FAKϪ/Ϫ cells treated with TNF-␣ (10 ng/ml), and NF-B DNA binding activity was measured by an EMSA with radiolabeled probe containing a consensus NF-B binding site (5Ј-AGTTGAGGGGACTT-TCCCAGG-3Ј, obtained from Santa Cruz Biotechnology). Arrow denotes specific NF-B⅐DNA complexes. B, for supershift assay, nuclear extracts were incubated in the presence of 1 g of specific antibody against each NF-B component, p65, p50, c-Rel, and RelB. C, equal amounts of nuclear extracts (50 g of protein) were subjected to immunoblot analysis with specific antibody against p65, p50, and c-Rel. D, NF-B-dependent reporter assay was performed for the FAKϩ/Ϫ and FAKϪ/Ϫ cells stimulated with TNF-␣. Relative luciferase activity was determined as described under "Materials and Methods." Results are expressed as the mean Ϯ S.D. from seven independent experiments. is composed of two catalytic subunits, IKK␣ and IKK␤, and a regulatory subunit, IKK␥/NEMO (30, 39 -42). Activated IKKs thus phosphorylate IBs at serine residues 32 and 36, leading to their degradation and the subsequent activation of NF-B (43). Recently, it has been suggested that RIP interacts directly with IKK␥ and recruits IKK to the TNFRI complex. Interestingly, whereas both TRAF2 and RIP are required for the activation of IKK, neither of them alone is sufficient to activate IKK. This is interpreted to mean that TRAF2 is necessary for the recruitment of IKK, whereas RIP mediates the activation of

FIG. 4. Diminished IKK activation and IB␣ degradation in FAK؊/؊ fibroblasts.
A, in vitro kinase assay for IKKs. IKK␥ was immunoprecipitated with anti-IKK␥ antibody from FAKϩ/Ϫ and FAKϪ/Ϫ cells treated with TNF-␣ (10 ng/ml) for the periods indicated. In the top panel, IKK␥ immunoprecipitates were assayed for kinase activity using purified GST-IB␣ ⌬C (amino acids 1-60) as a substrate. In the bottom panel, mutant GST-IB␣⌬C (S32A, S36R) was used as a substrate. B, FAKϩ/Ϫ and FAKϪ/Ϫ cells were treated with TNF-␣ for the periods indicated, and the cell lysates were prepared for immunoblot analysis with anti-IB␣ antibody. C, whole cell lysates (40 g of protein) were used for immunoblotting using each specific antibody (anti-TNFRI, RIP, TRAF2, IKK␣, IKK␤, and IKK␥ antibody, respectively). D, FAKϩ/Ϫ and FAKϪ/Ϫ cells were stained with anti-PE-TNFRI and analyzed using FACSCalibur. Staining by isotype-matched control antibody is also indicated. IKK (41,42,44). Because RIP-deficient cells, but not TRAF2deficient cells, have an impaired NF-B activation in response to TNF-␣, RIP appears to be necessary and sufficient for NF-B signaling by TNF-␣. In contrast, an absence of TRAF2 in targeted mutant fibroblasts leads to defective activation of JNK by TNF-␣, indicating that TRAF2 is essential for TNF-␣-mediated JNK signaling (37,45).
In this study, we demonstrated that in the presence of FAK, RIP interacts with TRAF2 in a TNF-␣-dependent manner, which is required for the subsequent recruitment of IKK␥ to RIP. In contrast, no such interaction of RIP with TRAF2 was observed in the FAKϪ/Ϫ cells, suggesting that RIP is able to interact with TRAF2 and IKK␥ only in the presence of FAK. A critical unanswered question is how FAK interacts with the signaling molecules triggered by the binding of TNF-␣ to TN-FRI. A TNF-␣-induced physical association of TNFRI with FAK was clearly demonstrated in the coimmunoprecipitation assay, as shown in Fig. 5. In addition, a potential association of FAK with RIP dependent on TNF-␣ was evident. In the absence of FAK, the interaction of RIP with TRAF2, which appears to be required for the subsequent recruitment of IKK␥ to RIP, did not occur. This suggests that upon stimulation with TNF-␣, RIP is able to interact with FAK, allowing TRAF2 to participate in the recruitment of IKK␥ to RIP. We thus assumed that FAK acts as a bridge linking TRAF2 to RIP, which might serve as a platform for the interaction of these molecules as a critical step in the TNFRI signaling cascade.
FAK is a non-receptor protein-tyrosine kinase with several tyrosine phosphorylation sites, and it interacts with various intracellular signaling proteins including c-Src, phosphatidylinositol 3-kinase, and Grb2 (1-5, 12, 13). A kinase-defective mutant, K454R, failed to rescue TNF-␣-mediated NF-B activation in FAKϪ/Ϫ cells (Fig. 7B), indicating the significant role of FAK activity. In contrast, the mutant Y397F rescued TNF-␣-induced NF-B activation as well as did the wild type of FAK in FAKϪ/Ϫ cells (data not shown), suggesting that this autophosphorylation site and the subsequent binding of the p85 subunit of phosphatidylinositol 3-kinase or Shc adaptor proteins (3,46) are dispensable. Salazar et al. (47) similarly demonstrated that Src family kinases are required for integrinmediated signaling, but the autophosphorylation of FAK at Tyr-397 is not required for the stimulation of the G proteincoupled receptor. Thus, although FAK phosphorylated at Tyr-397 appears to be critical for adhesion-dependent signaling, mitogenic G protein-coupled receptor signaling by bombesin,  Fig. 6. NF-B DNA binding activity was measured with an EMSA. B, failure of mutated FAK (K454R) to restore NF-B activation. Wild type or K454R-mutated FAK constructs were transfected in the adenovectors, and NF-B luciferase activity was measured. C, restoration of TNF-␣-induced IL-6 production. FAKϪ/Ϫ cells were transfected with Lac Z or HA-tagged wild type FAK. Cells were stimulated with 10 ng/ml TNF-␣ for 24 h, and the IL-6 level in the culture supernatant was measured using an ELISA kit (BioSource). Data shown are the mean Ϯ S.D. from seven independent experiments. D, total RNA was extracted, and RT-PCR for IL-6 and glyceraldehyde-3-phosphate dehydrogenase was performed. bradykinin, endothelin, and lysophosphatidic acid does not appear to require this phosphorylation site. It should be noted that no common phenotypic features have yet been found among TNFRI, RIP, TRAF2, or IL-6-depleted mice and FAKϪ/Ϫ mice, presumably because little attention has been drawn to the participation of FAK in cytokine signaling. One result demonstrating an association of FAK with TNF-␣ signaling might be relevant to the observation that intestinal cells from TNFRIIϪ/Ϫ mice did not migrate in response to TNF-␣, and this migration required Src kinase-mediated FAK tyrosine phosphorylation (48).
TNF-␣ induces not only the activation of NF-B but also the rapid activation of three classes of MAP kinases, ERK, JNK, and p38 MAP kinase, in a variety of cell types (26,38,49,50). MAP kinase activities are unaffected in FAKϪ/Ϫ fibroblasts, ruling out the possible involvement of FAK with these MAP kinase cascades. However, it is widely accepted that FAK links integrins and downstream components of the integrin-dependent signaling pathway, such as Src or Fyn and Ras to the MAP kinase pathway (2-4, 51, 52). In addition, a related adhesion tyrosine kinase, identical to proline-rich tyrosine kinase 2, is involved in upstream ERK and JNK signaling in response to certain stresses (53). Because the level of FAK-related proteintyrosine kinase proline-rich tyrosine kinase 2 is elevated in FAKϪ/Ϫ fibroblasts (3,17), we assume that proline-rich tyrosine kinase 2 is involved in the activation of the MAP kinase family in response to TNF-␣ in FAKϪ/Ϫ cells.
Whereas we have shown that FAK is indispensable for TNF-␣-induced NF-B activation, marginal activation could be detected in FAKϪ/Ϫ cells by gel shift assay and reporter gene assay with minimal IL-6 production. Residual NF-B activity, although minimal, suggests the existence of an apparent FAKindependent pathway for TNF-␣-induced NF-B activation. Likely candidates are two members of the atypical protein kinase C subfamily of isozymes ( protein kinase C and / protein kinase C), because protein kinase C acts as a potent activator of NF-B. It has been described previously that atypical protein kinase C-binding protein, p62, bridges the atypical protein kinase Cs and RIP, leading to NF-B activation (54).
In conclusion, our results show the novel function of FAK as an important signal transducer of proinflammatory cytokines. These results revealed a new and critical function of FAK in mediating activation of NF-B by TNF-␣ that may regulate various inflammatory and immune responses.