|Home | About | Journals | Submit | Contact Us | Français|
The IκB kinase (IKK)-related kinase NAK (also known as TBK or T2K) contributes to the activation of NF-κB-dependent gene expression. Here we identify NAP1 (for NAK-associated protein 1), a protein that interacts with NAK and its relative IKK (also known as IKKi). NAP1 activates NAK and facilitates its oligomerization. Interestingly, the NAK-NAP1 complex itself effectively phosphorylated serine 536 of the p65/RelA subunit of NF-κB, and this activity was stimulated by tumor necrosis factor alpha (TNF-α). Overexpression of NAP1 specifically enhanced cytokine induction of an NF-κB-dependent, but not an AP-1-dependent, reporter. Depletion of NAP1 reduced NF-κB-dependent reporter gene expression and sensitized cells to TNF-α-induced apoptosis. These results define NAP1 as an activator of IKK-related kinases and suggest that the NAK-NAP1 complex may protect cells from TNF-α-induced apoptosis by promoting NF-κB activation.
Transcription factor NF-κB plays a central role in inducing the expression of many genes that contribute to diverse biological functions, including cell proliferation, cell survival, oncogenesis, and inflammatory and immune responses (1, 13, 41, 45). In nonstimulated cells, NF-κB exists as hetero- and homodimers that are sequestered in the cytoplasm in a complex with members of the IκB family of inhibitor proteins (14, 24, 42, 46). Various extracellular signals, including proinflammatory cytokines such as tumor necrosis factor alpha (TNF-α) and interleukin-1β (IL-1β), induce the phosphorylation of IκBα on two conserved serine residues (Ser32 and Ser36) near its NH2 terminus (5, 9, 44). This phosphorylation results in recognition of IκBα by a specific multicomponent ubiquitin ligase (β-TrCP-E3RSIκB-FWD1), leading to its polyubiquitination and eventual degradation by the 26S proteasome (15, 52). The liberated NF-κB dimers translocate to the nucleus, where they are subjected to additional regulatory events that enhance their ability to activate transcription of genes that contain κB sites in their promoters/enhancers (12).
The signal-induced phosphorylation of IκBα and other IκBs is catalyzed by an IκB kinase (IKK) that was originally identified as a high-molecular-mass (~700-kDa) complex (10). The IKK complex contains two catalytic subunits, IKKα and IKKβ, and a regulatory subunit known as NEMO or IKKγ (28, 35, 49, 53). Both Cdc37 and Hsp90 have recently been proposed as additional components of the IKK complex (6). Activation of the IKK complex requires the phosphorylation of two serine residues located in the activation loops of both IKKα and IKKβ (8). Although NF-κB-inducing kinase and other members of the mitogen-activated protein kinase (MAPK) kinase kinase family, including MEKK1, -2, and -3, were proposed to function as IKK kinases (21, 31, 47, 51, 54), genetic evidence does not support their physiological participation in IKK activation (12). NF-κB-inducing kinase, for instance, was recently shown to act in a pathway that leads to activation of IKKα but does not depend on IKKβ or IKKγ (38).
Two IKK-related kinases, NAK (also known as TBK1 or T2K) and IKK (also known as IKKi), were identified and shown to be present in complexes distinct from the classical IKK complex (4, 32, 33, 40, 43). We had suggested that NAK functions upstream of IKK and activates IKKβ by direct phosphorylation of serine residues in its activation loop in response to cell stimulation with phorbol 12-myristate 13-acetate (PMA) or growth factors (43). TBK1 was identified as a kinase that binds to TANK (I-TRAF) and was proposed to function in a signaling pathway that links TRAF-TANK to the IKK complex (33). IKK (IKKi) was suggested to associate with unidentified kinases to form a PMA-inducible IκB kinase complex that phosphorylates IκBα on both Ser32 and Ser36 (32). However, the complete lack of IκBα phosphorylation and degradation in IKKα-IKKβ double-knockout cells exposed to various extracellular stimuli (22) excluded the possibility that IKK-related kinases function as direct IκB kinases.
The phenotype of NAK knockout mice is highly similar to those of mice deficient in IKKβ, IKKγ, or the p65/RelA subunit of NF-κB (2, 4, 23, 27, 36). The observation that the embryonic lethality and liver cell apoptosis apparent in NAK (T2K) knockout mice are prevented by inactivation of the gene for the type1 TNF-α receptor (4) suggests that NAK plays an important role in TNF-α-mediated NF-κB activation. However, the analysis of NAK-deficient cells revealed that inducible IκB degradation remained intact, despite a decrease in NF-κB transcriptional activity (4). Thus, it appears that NAK and probably IKK function in a yet-to-be defined step in the NF-κB activation pathway. Recently, IKK and possibly NAK were suggested to function as virus- or double-stranded-RNA-activated kinases that phophorylated two transcription factors involved in interferon gene induction (11, 39). However, the analysis of IKK-deficient cells failed to reveal defective activation of these transcription factors (20). Thus, the function of NAK and IKK remains enigmatic.
IKK-related kinases possess one serine residue in their canonical activation loop, and the phosphorylation of this serine is required for their activity (32). Gel filtration revealed that IKK is a component of a high-molecular-weight complex, suggesting that some component of this complex may function as a regulatory subunit required for IKK or NAK activation. Therefore, the identification of IKK- or NAK-associated molecules may provide new insight into their mechanism of activation and function in NF-κB signaling.
We now describe the identification and characterization of NAP1 (for NAK-associated protein 1), a protein that binds to and activates IKK-related kinases both in vitro and in vivo. NAP1 facilitates oligomerization of NAK molecules, and NAK-NAP1 complexes are components of a high-molecular-weight complex that is activated on exposure of cells to TNF-α. We also found that NAK phosphorylates serine 536 on the COOH-terminal region of p65/RelA, and NAP1 enhances this phosphorylation. With the use of RNA interference technology, we also demonstrate that the NAK-NAP1 complex is important for NF-κB-dependent gene expression and prevention of apoptosis in response to TNF-α.
The pLexA-NAK plasmid was generated by insertion of the full-length human NAK-coding sequence in frame into pBTM116. pBTM116-NAK was transformed into the yeast strain L40 (LYS::lexA-HIS3), and this strain was transformed with a HeLa cell cDNA library (Clontech). Positive colonies were selected on yeast synthetic medium lacking histidine, leucine, and tryptophan and were then tested for β-galactosidase activity. The cDNA for human NAP1 was confirmed by conventional screening of a λ phage library of human testis cDNA (Stratagene) with a probe derived from the two-hybrid clones. The human NAP1 cDNA contains an in-frame stop codon upstream of the first ATG in the predicted open reading frame.
Polyadenylated RNAs isolated from various human tissues (Clontech) were subjected to Northern blot hybridization as described previously (43).
Monoclonal antibodies to NAP1 were generated with recombinant human NAP1 expressed in Sf9 cells as the immunogen by a modified version of a method described previously (19). The fused hybridomas were distributed into 96-well culture plates (Sumitomo Bakelite) at a density of 1.4 × 105 cells per well and were maintained in RPMI 1640 (Gibco) supplemented with 10% fetal bovine serum, 0.1 mM hypoxanthine, 0.4 μM aminopterin, and 16 μM thymidine. Positive hybridomas were subcloned twice by limiting dilution and then propagated as ascites in Pristane-treated nude mice (Balb/c nu/nu; CLEA). Antibodies were purified from ascites by precipitation with caprylic acid and ammonium sulfate fractionation.
A mouse monoclonal anti-Flag antibody (M5; Sigma), a rabbit anti-Myc tag serum (MBL, Nagoya, Japan), and an antibody specific to phospho-IκBα (Ser32) (92415; NEB) were used at a 1:1,000 dilution for immunoblotting. Antibodies specific to phospho-c-Jun (Ser63) and phospho-ATF-2 (Thr71) were purchased from Cell Signaling Co.
For generation of pcDNA3.1Myc/His6NAP1, we first performed PCR with the 5′ primer 5′-AAA GAA TTC GCC ACC ATG GAT GCA CTG GTA GAA GAT (S-1), the 3′ primer 5′-AAA CTC GAG ATT CTT ATA AAG GCA GTT CTG (AS-1), and human NAP1 cDNA as the template. The resulting PCR product was digested with EcoRI and XhoI and then ligated into pcDNA3.1/Myc-HisA (Invitrogen). The plasmid pcDNA3.1Myc/His6NAP1(158-270) was generated by PCR with the primers 5′-AAA GAA TTC GCC ACC ATG TGG GAG GTG GAA AAG TTG AGC and 5′-AAA CTC GAG GCT GTC TCT TCC AAG GTC TTC and with pcDNA3.1Myc/His6NAP1 as the template. The vector pcDNA3.1Myc/His6NAP1(Δ158-270) was generated by two-step PCR with S1 and AS-1, the inner primers 5′-GAA CTA CTA AGA AAA CTG AAA ACC ACA AAA CTG CAC TTG ATG AAT and 5′-AAA ATT CAT CAA GTG CAG TTT TGT GTT TGA TGA AGG TGG ATT TAA, and pcDNA3.1Myc/His6NAP1 as the template. Expression vectors for Flag-NAK, Flag-IKKβ, their kinase-inactive mutants, and Flag-NEMO have been described previously (43). The vector pcDNA3FlagIKK was kindly provided by T. Maniatis (Harvard University).
293T, MDAH041, and HeLa cells were maintained as described previously (30). For immunofluorescence analysis, MDAH041 cells were fixed with 4% paraformaldehyde, permeabilized with 0.2% Triton X-100, and incubated with either antibodies to NAP1 (1:200 dilution) or normal mouse serum (1:1,000 dilution). Immune complexes were detected with Cy3-conjugated goat antibodies to mouse immunoglobulin G (1:200 dilution; Jackson Immuno Research).
The plasmid pGEX-NAP1 was generated by ligation of the EcoRI-XhoI fragment of pcDNA3.1Myc/His6NAP1 into pGEX5X-1. The corresponding plasmids encoding NAP1 residues 102 to 392 (N101), 158 to 392 (N157), 271 to 392 (N270), 1 to 302 (C302), 1 to 270 (C270), or 1 to 157 (C157) were generated by PCR with the following sets of primers: 5′-AAA GAA TTC GAT AGA GAT AAT TTG AAG AGC and AS-1 for N101, 5′-AAA GAA TTC TGG GAG GTG GAA AAG TTG AGC and AS-1 for N157, 5′-AAA GAA TTC ACA AAA CTG CAC TTG ATG AAT and AS-1 for N270, S-1 and 5′-AAA CTC GAG AGG GGA AGA TGT GGT ATG ACA for C302, S-1 and 5′-AAA CTC GAG GCT GTC TCT TCC AAG GTC TTC for C270, and S-1 and 5′-AAA CTC GAG GTT TGA TGA AGG TGG ATT TAA for C157. The plasmids pGEX-NAP1(158-270) and pGEX-NAP1(Δ158-270) were generated by inserting the EcoRI-XhoI fragments of pcDNA3.1Myc/His6NAP1(158-270) and pcDNA3.1Myc/His6NAP1(Δ158-270), respectively, into pGEX5X-1. Glutathione S-transferase (GST)-fused wild-type p65 and its mutant p65 proteins were generated as described previously (37). All plasmids were introduced into Escherichia coli, the encoded GST fusion proteins were purified, and GST pull-down assays were performed as described previously (48).
Baculoviruses expressing Myc- and His6-tagged NAK were generated as described previously (43). Baculoviruses expressing Myc- and His6-tagged NAP1 were generated by removing the EcoRI/PmeI fragment from pcDNA3.1Myc/His6NAP1 into pVL1392, which was cut with EcoRI and SmaI. One microgram of pVL1392NAP1Myc/His6 was cotransfected into Sf9 cells with 2.5 μg of linearized baculovirus DNA (BaculoGold; PharMingen). Baculoviruses expressing p65, p50, and Myc- and His6-tagged IκBα were generated by cotransfection of linearized baculovirus DNA with pVL1392p65, pVL1392p50, and pVL1392IκBαMyc/His6, respectively.
HeLa cells were left unstimulated or were treated for 10 min with 20 ng of TNF-α per ml. Cells were washed with phosphate-buffered saline and collected. Cytosolic S100 extracts, prepared as previously described (10), were fractionated on a Superose 6 gel filtration column. Elution (600 μl per fraction) was performed in buffer A supplemented with 0.1% Brij-35 and 250 mM NaCl (10).
293T or HeLa cells (106) were plated in 10-cm-diameter dishes and transfected with the indicated amounts of mammalian expression vectors with the use of a Fugene 6 transfection kit (Boehringer). After 30 h, the cells were washed once with ice-cold phosphate-buffered saline and lysed in IP-kinase buffer as described previously (43). Cellular debris was removed by centrifugation at 10,000 × g for 10 min, and the resulting supernatant was subjected to immunoprecipitation with various antibodies and 30 μl of protein A- and protein G-Sepharose (50% suspension) (Boehringer).
Immunoprecipitation, immunoblotting, and kinase assay were performed as described previously (43). c-Jun NH2-terminal kinase (JNK) and p38 MAPK activities were examined by immune-complex kinase assays as described previously (17).
For reporter gene assays, 293T cells grown on 35-mm-diameter plates were transfected with an NF-κB-Luc reporter plasmid and a wild-type or mutant pcDNA3.1Myc/His6NAP1 vector, in the absence or presence of pcDNA3FlagNAK. Luciferase activity was determined as described previously (43).
Double-stranded RNAs (21 bp) were synthesized by B-Bridge International (Saitama, Japan). The NAP1 small interfering RNA (siRNA) was targeted to the coding sequence 5′-TCTGTTGCTTCCCATTTTGCT, which corresponds to nucleotides 100 to 120 of the open reading frame of human NAP1 cDNA. Green fluorescent protein (GFP) siRNA, used as a control, was described previously (26). Cells were transfected with each siRNA with the use of the Oligofectamine 2000 reagent (Invitrogen). Cells were subjected to the terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling (TUNEL) technique for the detection of apoptotic cells with a kit (Promega).
The NAP1 cDNA sequence has been deposited with GenBank (accession number AY151386).
Although extracellular stimuli such as TNF-α and platelet-derived growth factor stimulate endogenous NAK activity (43), the precise mechanism of NAK activation remains unclear. To identify NAK-binding proteins that might regulate its activity, we performed a yeast two-hybrid screen with full-length NAK as a bait. Among a total of 2 × 106 transformants obtained with a HeLa cell cDNA library, 63 positive colonies were detected, and 35 of them were confirmed to be lacZ positive. Most of the confirmed positive colonies contained overlapping cDNAs derived from a single gene. Sequence analysis revealed that this cDNA clone is identical to previously identified cDNA with unknown function, named AZ2, whose expression is induced by a demethylating stimulus (29). In light of its function, we renamed the product of this gene NAP1 (for NAK-associated protein 1). The human NAP1 cDNA sequence predicts a translation product of 392 amino acids with a calculated molecular mass of 44,800 Da (Fig. (Fig.1A).1A). NAP1 exhibits low but significant similarity (28%) to TANK (I-TRAF). The smallest cDNA isolated by the yeast two-hybrid screen contained the COOH-terminal 202 residues of NAP1, indicating that this region is sufficient for interaction with NAK.
Northern blot analysis revealed two major NAP1 transcripts of 3.2 and 4.2 kb in all human tissues examined, whose abundance was highest in pancreas and testis (Fig. (Fig.1B).1B). Although NAK mRNA is barely detectable in thymus (43), NAP1 transcripts are present at relatively high levels in this tissue.
We next examined the subcellular localization of NAP1 protein by indirect immunofluorescence analysis. Whereas virtually no signal was detected with normal mouse serum as a control, a mouse monoclonal antibody to NAP1 yielded a pronounced cytoplasmic signal in human MDAH041 fibroblasts (Fig. (Fig.1C).1C). The predominant cytoplasmic localization of NAP1 is consistent with the subcellular distribution of NAK (43).
To map the minimal region of NAP1 required for binding to NAK, we performed in vitro binding assays with GST fusion proteins containing full-length NAP1 or various deletion mutants thereof. The NH2-terminal truncation mutants comprised the COOH-terminal regions of the protein beginning at amino acid 102, 158, or 271. The COOH-terminal truncation mutants comprised the NH2-terminal regions of the protein up to amino acid 157, 270, or 302. The GST fusion proteins were tested for the ability to bind Myc epitope-tagged NAK overexpressed in Sf9 insect cells (Fig. (Fig.2A).2A). Truncation of the NH2-terminal region of NAP1 by up to 157 amino acids (N101 and N157) did not affect NAK binding, whereas deletion of an additional 113 NH2-terminal residues (N270) completely abolished binding to NAK. Deletion of the COOH-terminal region comprising amino acids 271 to 392 (C270) resulted in a small decrease in the extent of NAK binding, whereas deletion of residues 158 to 392 (C157) abolished NAK binding activity. Consistent with these results, a mutant construct comprising only amino acids 158 to 270 bound to NAK as efficiently as the full-length protein, whereas a mutant lacking residues 158 to 270 (Δ158-270) did not bind NAK. These observations thus indicate that the region of NAP1 responsible for binding to NAK is located between amino acids 158 and 270. Curiously, the similarity between this region of NAP1 and the equivalent region of TANK is only 34%.
We next performed coimmunoprecipitation experiments to determine whether NAP1 interacts with NAK in transfected 293T cells. Flag-NAK coprecipitated both with full-length NAP1 and with NAP1(158-270) but not with NAP1(Δ158-270) (Fig. (Fig.2B).2B). These results were thus consistent with those of the in vitro binding assay and confirmed that the NAK-binding domain of NAP1 lies between residues 158 and 270.
The IKK-related kinases NAK and IKK are not present within the classical IKK complex (32, 43). We investigated whether NAP1 interacts with IKK, IKKβ, or IKKγ (NEMO). 293T cells were transfected with expression vector for Flag-tagged NAK, IKK, IKKβ, or IKKγ (NEMO) in the absence or presence of a vector for Myc epitope-tagged NAP1, and cell lysates were subjected to immunoprecipitation with antibodies to Myc. The resulting immunoprecipitates were immunoblotted with antibodies to Flag. Both Flag-NAK and Flag-IKK, but neither Flag-IKKβ nor Flag-IKKγ, were detected in the immunoprecipitates prepared from cells expressing Myc-NAP1 (Fig. (Fig.2C).2C). In similar experiments, Flag-IKKα was found not to coprecipitate with Myc-NAP1 (data not shown). These results thus indicate that NAP1 binds specifically to IKK-related kinases and does not interact with components of the classical IKK complex.
Immunoblot analysis revealed that NAK and NAP1 were efficiently coprecipitated from lysates of nonstimulated HeLa cells with antibodies to NAP1 and NAK, respectively (Fig. (Fig.3A).3A). To examine further the interaction of endogenous NAK and NAP1, we incubated HeLa cells in the absence or presence of TNF-α and fractionated cell lysates by gel filtration on a Superose 6 column. The resulting column fractions were immunoblotted with antibodies to IKKα, IKKγ, NAK, or NAP1 (Fig. (Fig.3B).3B). Endogenous NAP1 was detected as two distinct polypeptide species, one of 59 kDa coeluting with a ~600-kDa complex that contained the majority of NAK and the other of 47 kDa coeluting with a ~250-kDa complex that contained smaller amounts of NAK. The elution profiles of IKKα and IKKγ differed from those of NAP1 and NAK. These results further confirmed that endogenous NAK did not interact with the classical IKK complex but did coelute with NAP1. The identities of the two NAP1 polypeptides are not fully clear at this point, but they may be related to the two different-sized transcripts seen in Fig. Fig.1B.1B. No effect of TNF-α treatment on the elution profiles was detected.
The column fractions were also immunoprecipitated with antibody to NAK, and the resulting immunoprecipitates were assayed for their ability to phosphorylate GST-p65C (see below). NAK-associated p65 kinase activity was detected exclusively in the higher (600-kDa)-molecular-mass fraction but not in the lower (250-kDa)-molecular-mass fraction. Although TNF-α treatment significantly increased NAK-associated p65 kinase activity, it did not affect the hydrodynamic properties of either of the two NAK-NAP1 complexes.
We next examined the effect of NAP1 binding on NAK kinase activity assayed with GST-IKKβ(132-206) as a substrate. NAK phosphorylated GST-IKKβ(132-206), GST-NAP1, and itself (Fig. (Fig.4A).4A). GST-NAP1, but not GST alone, increased the kinase activity of NAK towards IKKβ in a dose-dependent manner. No increase in enzyme activity was apparent with a kinase-dead mutant of NAK (data not shown). While GST-IKKβ(132-206) phosphorylation was enhanced up to 5.5- to 6-fold by NAP1, the autophosphorylation activity by NAK was enhanced by no more than 1.4-fold.
To determine whether NAP1 also activates NAK in vivo, we subjected 293T cells expressing Flag-NAK in the absence or presence of Myc-NAP1 to immunoprecipitation with antibodies to Flag or to Myc and then assayed the resulting precipitates for phosphorylation of GST-IKKβ(132-206). Whereas almost equal amounts of Flag-NAK were immunoprecipitated in the absence or presence of Myc-NAP1, the kinase activity of the NAK immunoprecipitate was much higher when it was isolated from Myc-NAP1-expressing cells (Fig. (Fig.4B,4B, lanes 5 and 6). The NAK-associated NAP1 was also phosphorylated.
Given that all of the Flag-NAK molecules immunoprecipitated by the antibodies to Flag might have not been associated with NAP1, we might have underestimated the extent of NAK activation by NAP1. We therefore also immunoprecipitated NAK-NAP1 complexes with antibodies to the Myc epitope fused to NAP1. In this instance, all of the NAK molecules in the immunoprecipitates should be associated with NAP1. The amount of cell lysate used for immunoprecipitation with antibodies to Myc was 1.5 times that used for immunoprecipitation with antibodies to Flag in an attempt to equalize the amount of NAK precipitated. Although the amount of NAK in the Myc-NAP1 immunoprecipitates was slightly lower than that in the Flag-NAK immunoprecipitates, the kinase activity of the former was 5.5 times higher than that of the latter (Fig. (Fig.4B,4B, compare lanes 5 and 9). Immunoprecipitation of cells expressing Myc-NAP1 alone did not yield a kinase-active complex (data not shown), eliminating the possibility that NAP1 forms a complex with another unknown kinase. These results strongly suggest that NAP1 directly activates NAK both in vitro and in vivo.
We next asked whether NAP1 facilitates oligomerization of NAK molecules. An expression vector for Myc-NAP1 was transfected into 293T cells, together with expression vectors for either Flag-NAK or Myc-NAK or both. Transfected cell lysates were immunoprecipitated with anti-Flag antibody and then immunoblotted with anti-Myc antibody. Myc-NAK was efficiently coprecipitated with Flag-NAK when Myc-NAP1 was coexpressed (Fig. (Fig.4C,4C, lane 6). Similar results were obtained when the kinase-negative Flag-NAK and Myc-NAK were used (data not shown). These results clearly indicated that NAP1 facilitates oligomerization of NAK molecules in vivo.
We next examined the effect of NAP1 on activation of an NF-κB-dependent reporter gene by NAK (Fig. (Fig.5A).5A). Ectopic expression of NAP1 alone induced a small increase in reporter gene expression, presumably as a result of activation of endogenous NAK. Overexpression of NAK alone induced an approximately sixfold increase in luciferase activity, and coexpression of full-length NAP1 enhanced this effect in a dose-dependent manner. The NAP1(158-270) mutant also increased NAK-induced luciferase activity, whereas NAP1(Δ158-270) had no such effect. The NAP truncation mutants C157 and N270 (Fig. (Fig.2A)2A) did not affect NAK-induced luciferase activity (data not shown). Therefore, the minimal NAK-binding region of NAP1 is sufficient for the enhancement of NAK-induced NF-κB activation.
Given that our previous results (43) suggested that NAK functions downstream of protein kinase C, we investigated whether NAP1 affects the activation of NF-κB by TNF-α or PMA. Treatment of 293T cells with TNF-α or PMA increased the activity of an NF-κB-dependent reporter gene by factors of ~10 and 5, respectively (Fig. (Fig.5B).5B). Ectopic expression of NAP1 potentiated these effects in a dose-dependent manner. Importantly, NAP1 did not affect TNF-α- or PMA-induced activation of an AP-1-dependent reporter gene, suggesting the specific function of NAK-NAP1 complex in NF-κB signaling (Fig. (Fig.5C).5C). Although a dominant negative mutant of NAK was previously shown not to inhibit the activation of NF-κB by TNF-α (43), our present results suggest that NAP1 may contribute to TNF-α-induced NF-κB activation.
Given that NAP1 affected the autophosphorylation and IKKβ-phosphorylating activities of NAK differentially (Fig. (Fig.4A),4A), we investigated whether the NAK-NAP1 complex might affect the activation of JNK or p38 MAPK. HeLa cells were transfected with an expression vector for NAK and various amounts of an expression vector for NAP1, and the activation of JNK and p38 MAPK was evaluated by immune-complex kinase assays with c-Jun and ATF-2, respectively, as substrates. Neither JNK nor p38 MAPK was activated by overexpression of NAK in the absence or presence of NAP1 (data not shown).
Although the experiments using dominant negative NAK did not provide direct evidence for the participation of NAK in TNF signaling (43), the phenotype of NAK−/− mice is very similar to that of IKKβ- and p65-deficient mice (3, 23, 25). In addition, NAK−/− mouse embryonic fibroblasts exhibited decreased activation of some NF-κB-responsive genes in response to TNF-α despite normal induction of IKK activity, IκBα degradation, and NF-κB DNA binding activity (4). These observations suggest that NAK could directly modulate the transcriptional activity of the NF-κB but not its nuclear transport. Thus, we examined whether NAK can phosphorylate NF-κB subunits. A p65-p50-His6-IκBα ternary complex was generated by infection of Sf9 cells with baculovirus expression vectors and purified by using ProBond resin (Invitrogen). Phosphorylation experiments revealed that NAK readily phosphorylated p65 but not p50 or IκBα (Fig. (Fig.6A,6A, KA). In additional experiments, immunopurified Flag-NAK phosphorylated the GST-fused COOH-terminal portion of p65 (GST-p65C) but not its NH2-terminal portion (GST-p65N) or the S536A mutant of GST-p65C (Fig. (Fig.6B6B lanes 2, 3, and 4). Coexpression of NAP1 enhanced the NAK kinase activity towards GST-p65C (Fig. (Fig.6B,6B, lane 7). These results suggest that NAK-NAP1 phosphorylated serine 536 of the p65 subunit, whose phosphorylation was suggested to enhance its transcriptional activity (16, 50).
Finally, we investigated the physiological role of NAP1 in mammalian cells through the use of RNA interference. We synthesized an RNA duplex directed against nucleotides 100 to 120 of the NAP1-coding sequence as well as a control RNA targeted to the coding sequence of GFP. Transfection of cells with the NAP1 siRNA resulted in a 90% reduction in both forms of NAP1, with molecular masses of 59 and 47 kDa, indicating that both polypeptides are encoded by the same gene (Fig. (Fig.7A).7A). Although the level of NAK was unchanged, the amount of NAK associated with NAP1 was also reduced by 90%. The GFP siRNA had no effect on total NAK or NAP1 levels.
We next examined the effect of NAP1 depletion on the activation of NF-κB induced by TNF-α, PMA, or IL-1β. Whereas the GFP siRNA had no effect on activation of the NF-κB-dependent reporter construct, the NAP1 siRNA reduced the stimulatory effects of TNF-α and PMA, but not of IL-1β, on luciferase expression (Fig. (Fig.7B),7B), suggesting that NAP1 contributes to full NF-κB activation by TNF-α or PMA. Interestingly, IκBα phosphorylation upon TNF-α treatment was only slightly reduced in cells depleted of NAP1 (Fig. (Fig.7C).7C). These results suggest that NAP1 modulates NF-κB activation at a point distal to IκBα phosphorylation and degradation.
The activation of NF-κB contributes to inhibition of apoptosis, and loss of NF-κB function in mice results in TNF-α-dependent liver apoptosis (3). We examined whether depletion of NAP1 affects TNF-α-induced apoptosis. Whereas ~10% of the TNF-α-treated nontransfected cells (data not shown) or of TNF-α-treated cells transfected with the GFP siRNA were TUNEL positive, ~40% of the cells transfected with the NAP1 siRNA and exposed to TNF-α were TUNEL positive (Fig. (Fig.7D).7D). These results suggest that NAP1 protects cells against TNF-α-induced apoptosis by promoting the activation of NF-κB.
Previous studies have suggested that the IKK-related kinases NAK and IKK play an important role in the signaling pathways by which extracellular stimuli such as platelet-derived growth factor and TNF-α activate NF-κB. However, the precise functions and regulation of these kinases have been unclear. To identify regulatory proteins that might mediate activation of NAK in response to upstream stimuli, we performed a yeast two-hybrid screen with NAK as a bait. Here, we describe the cloning and characterization of one such NAK-binding protein, NAP1, which we have shown functions as an activator of IKK-related kinases. NAP1 can bind directly to NAK, stimulates its kinase activity, facilitates its oligomerization, and promotes NAK-induced as well as TNF-α- or PMA-induced activation of NF-κB.
NAK was previously shown to bind TANK, a TRAF-binding protein, and to form a ternary complex with TRAF2 and TANK in transfected cells (33). Although NAP1 exhibits limited sequence similarity to TANK, the abilities of these two proteins to bind TRAF family members differ markedly. Whereas TANK binds TRAF1, -2, and -3 (7), endogenous NAP1 did not stably interact with TRAF1, -2, -3, or -6 in HeLa cells incubated in the absence or presence of TNF-α or IL-1β (data not shown). The central region of TANK contains both the consensus TRAF-binding sequence PXQXS/T, which is present in CD40, CD30, CD27, and LMP1, and the DEED motif, a short acidic TRAF-binding sequence present in 4-1BB, Ox40, and CD30 (7). However, neither of these sequence motifs is conserved in NAP1. NAP1 is thus unlikely to be capable of direct interaction with TRAF proteins. TANK exerts a dose-dependent biphasic effect on TRAF-mediated NF-κB activation, enhancing it at lower doses and inhibiting it at higher doses (7, 33). TANK also competes with receptor tails for binding to TRAF proteins and can thereby prevent ligand-dependent recruitment of TRAFs to receptor complexes and inhibit the ability of the latter to mediate NF-κB activation (34). As NAP1 exerts only a stimulatory effect on NF-κB activation, it is unlikely to modulate TRAF function, as TANK does.
NAP1 mRNA was detected in all human tissues examined. Its tissue distribution thus differs from that of NAK mRNA in that it is relatively abundant in the thymus, a tissue in which NAK mRNA is virtually undetectable (43). This observation suggests that NAP1 may also interact with proteins other than NAK. One such protein may be the IKK-related kinase IKK, which interacted with NAP1 as efficiently as NAK in transfected cells. Furthermore, GST-NAP1 activated IKK in a concentration-dependent manner in vitro (data not shown). Given that IKK is expressed predominantly in immune cells, including those in the thymus (40), NAP1 may function as a common regulatory subunit of the IKK-related kinases.
Although the precise mechanism by which NAP1 activates NAK remains to be elucidated, the binding of NAP1 to NAK may induce NAK oligomerization. Such an oligomerization might affect both the kinase activity of NAK and its substrate specificity, because NAP1 binding stimulated the autophosphorylation of NAK to a lesser extent than its effect on NAK activity toward p65 and IKKβ. Given that NAK activation upon TNF-α may depend on its formation of a high-molecular-weight complex (Fig. (Fig.3B),3B), NAP1 is likely to function as a chaperone or a cofactor that stabilizes this complex or is involved in recruitment of additional subunits.
Although some of the molecular mechanisms of TNF-α signal transduction are relatively well understood, control of the balance between apoptosis and cell survival as the outcome of TNF-α exposure is not very clear. There are also hints from the phenotypes of knockout mice that TNF-α may potentiate NF-κB-mediated gene expression by more than one mechanism. In diverse cell types, NF-κB is an important inhibitor of TNF-α-induced apoptosis (18). Prevention of NF-κB function in mice results in massive liver degeneration mediated by TNF-α-dependent apoptosis, as first detected in p65-deficient mice (3). A similar phenotype was observed in IKKβ- and IKKγ-deficient mice (23, 25, 27, 36). Indeed, both IKKβ and IKKγ are required for NF-κB activation. More surprising was the finding that although NAK-deficient mice also exhibit TNF-α-dependent liver apoptosis, cells derived from these mice show normal induction of NF-κB DNA binding activity and normal IκB degradation. It was therefore suggested that NAK affects NF-κB activity at a step distal to IκB degradation (4). Our present results show that the NAK-interacting protein NAP1 protects cells from TNF-α-induced apoptosis, presumably by promoting the activation of NF-κB. Most importantly, we now found that, at least in vitro, NAP1-associated NAK activity can phosphorylate serine 536 of the p65 subunit. Phosphorylation of this site was suggested to enhance the transcriptional activity of p65 (16, 50). Therefore, the NAK-NAP1 complex could be an important component of the NF-κB activation pathway through the phosphorylation of p65.
After the submission of this paper, NAK and IKK were reported to play an essential role in the antiviral response through phosphorylation of IRF-3 and IRF-7 (11, 39). It should therefore be interesting to examine whether NAP1 is also involved in these processes.
We thank T. Maniatis for kindly providing plasmids; S. Futatsuka and S. Kawamoto for generating the antibodies to NAP1; and H. Niida, H. Murakami, and other members of the Nakanishi lab for helpful discussions and critical reading of the manuscript.
This work was supported in part by a Grant-in-Aid for Scientific Research on Priority Areas and for Scientific Research (B) from the Ministry of Education, Science, Sports, and Culture of Japan to M.N. Work in the lab of M.K. was supported by grants from the National Institute of Health. M.D. was supported by a Sontag Foundation fellowship from the National Arthritis Research Foundation, and M.K. is the Frank and Else Schilling-American Cancer Society Research Professor.