p21-activated Kinase 1 Activates the Nuclear Factor κB (NF-κB)-inducing Kinase-IκB Kinases NF-κB Pathway and Proinflammatory Cytokines in Helicobacter pyloriInfection*

Helicobacter pylori, the causative agent of several human gastric diseases, induces activation of the immediate early response transcription factor nuclear factor κB (NF-κB), which subsequently triggers release of proinflammatory cytokines in colonized epithelial cells. Here we report that inH. pylori infection p21-activated kinase 1 (PAK1) activates NF-κB. Activated PAK1 associates with NF-κB-inducing kinase, which upon activation directs the activity of IκB kinases to IκBα. Our results indicate that in epithelial cells PAK1 participates in a unique pathway that links H. pylori-dependent effector molecules to the activation of NF-κB and the induction of the innate immune response.

Exposure of cells to various stimuli results in phosphorylation, ubiquitination, and subsequent degradation of IB molecules. The liberated nuclear factor B (NF-B) 1 dimers are translocated to the nucleus, where they activate transcription of target genes (1). Key components of the intracellular signal transduction pathways regulating NF-B activation are represented by NIK (2) and the IB kinases (IKKs) (3). IKK␣ and IKK␤ are the catalytic subunits of a protein kinase complex that phosphorylates IB molecules (4). IKK␥ (or NF-B essential modulator (NEMO)) represents the regulatory subunit (5,6).
We have studied the mechanism of NF-B activation in response to human pathogenic Helicobacter pylori in epithelial cells where NF-B, one of the main activators of the inflammatory response, triggers the induction of immune function including cytokine/chemokine production, growth control, and apoptosis (1). The epithelial cytokine/chemokine response is particularly important in the early stages of H. pylori-induced inflammation and is often followed by diseases like gastritis, peptic ulcer (7), gastric cancer (rarely), and low grade B-cell mucosa-associated lymphoid tissue gastric lymphoma (8). The major disease-associated, genetic difference in H. pylori strains is the presence or absence of a pathogenicity island (PAI) containing 31 genes, which code for proteins involved in a specialized type IV secretion machinery (9). H. pylori infection modulates the host cells by bacterial protein translocation. Several groups have recently shown for the first time that H. pylori translocates the CagA protein by a type IV secretion system. Tyrosine phosphorylation of CagA induces changes in the tyrosine phosphorylation state of proteins in the epithelial cell (10 -14). Knockouts of certain PAI genes (CagA, CagF, and CagN), which are not necessary for the functional integrity of the type IV secretion apparatus, suppress or reduce the activation of NF-B (15) and affect the secretion of cytokines/ chemokines (16). Thus the integrity of the type IV secretion machinery is an absolute requirement for NF-B activation in H. pylori infection. The identity of the effector molecule(s) of H. pylori and the mechanism of integration of the signals that induce NF-B are not known so far. Therefore, we examined the mechanism of the H. pylori-induced activation of proinflammatory cytokine genes in the course of NF-B activation in detail.

EXPERIMENTAL PROCEDURES
Cell Culture and H. pylori Infection-Gastric epithelial cells (AGS) and 293 or HeLa cells were grown in RPMI 1640 containing 4 mM glutamine (Life Technologies, Inc.), 100 units/ml penicillin, 100 g/ml streptomycin, and 10% fetal calf serum (Life Technologies, Inc.) in a humidified 5% CO 2 atmosphere. 16 h before infection, the medium was replaced by fresh RPMI 1640 medium supplemented with 0.1% fetal calf serum. The H. pylori P12 strain or the PAI strain (17) was cultured for 48 -72 h on agar plates containing 10% horse serum in a microaerophilic atmosphere (generated by Campy Gen, Oxoid) at 37°C. For the infection, the cell monolayer (50 -70% confluence) was incubated with the bacteria to the multiplicities of infection (MOI) of 50 for different periods of time. Infection with H. pylori was routinely monitored by light microscopy.
Electrophoretic Mobility Shift Assay-Nuclear extracts were prepared by using a nonionic detergent method as described previously (18). Detection of NF-B was performed with a [␣-32 P]dATP-labeled Ig oligo probe containing the NF-B recognition site. The DNA binding reactions were performed with 20 l of binding buffer (2 g of poly(dI-dC), 1 g of bovine serum albumin, 5 mM dithiothreitol, 20 mM HEPES, pH 8.4, 60 mM KCl, and 10% glycerol) for 20 min at 30°C. For competition experiments, the cold oligonucleotide probe was used, and supershift analysis was performed using antibodies against p65, p50, or c-Rel (18). The reaction products were analyzed via 5% polyacrylamide gel electrophoresis using 12.5 mM Tris, 12.5 mM boric acid, and 0.25 mM EDTA, pH 8.3. The gels were dried and exposed to Amersham TM film (Amersham Pharmacia Biotech) at Ϫ70°C using an intensifying screen.
RNA Isolation and Reverse Transcriptase-Polymerase Chain Reaction-Total RNA was isolated using Trizol reagent (Life Technologies, Inc.) as recommended by the manufacturer's instructions. Total RNA (1 g) was reverse transcribed into single-stranded cDNA with Superscript IIRT (Life Technologies, Inc.) and oligo(dT) primers. Amplification of cytokine cDNAs and ␤-actin cDNA as an internal control in each reaction was carried out by polymerase chain reaction with the primers as described previously (19). A subsaturating number of cycles allowed a semi-quantitative analysis within the infection kinetics. For inhibition of IB␣ degradation by the 26 S proteasome, the cells were preincubated for 60 min using 10 M lactacystin (Affiniti) before the bacteria were added. Polymerase chain reaction products were visualized by ethidium bromide staining after agarose gel electrophoresis.
Transient Transfections and Reporter Assays-Transactivating activity of NF-B was analyzed in 293 or HeLa cells by cotransfection of a luciferase expression plasmid (400 ng) containing three repeats of the NF-B human immunodeficiency virus-binding site and expression constructs using cationic liposomes (DAC-30, Eurogentec). 16 h after transfection cells were infected with H. pylori, treated with 10 ng ml Ϫ1 tumor necrosis factor ␣ (TNF␣) (Promega), or left untreated. Luciferase assays were performed 3-4 h after treatment as recommended by the manufacturer's instructions (Promega). The data presented are representative of more than three independent experiments. A pSV-␤-galactosidase vector (Promega) was used for normalization of transfection efficiency. The results were recorded on a Wallac 409 ␤-counter (Berthold-Wallac) and given as fold induction or as percent induction compared with the control.

NF-B Activation in H. pylori Infection
Involves IKK Activity-AGS cells (neuroendocrine differentiated gastric carcinoma cells) infected with H. pylori or treated with TNF␣ for various periods of times, fractionated into cytoplasmic and nuclear components, and subjected to electrophoretic mobility shift assay show an equivalent increase of NF-B DNA binding activity within 90 and 10 min, respectively (Fig. 1a, lanes 1-9). The H. pylori-induced binding activity was strongly reduced and partially super-shifted when anti-p50 or anti-p65 antibodies were used, whereas an anti-c-Rel antiserum or preimmune serum did not show a reduction (Fig. 1a, lanes 10 -13). The specificity of the DNA binding activity was examined by adding increasing amounts of the non-labeled double-stranded Ig oligonucleotide for competition (Fig. 1a, lanes 14 -16). In contrast to the wild type H. pylori strain, an isogenic mutant strain (PAI, no expression of the type IV secretion machinery) did not show NF-B activation (Fig. 1a, lanes [17][18][19]. This experiment excludes a role for lipopolysaccharide in H. pylori-induced NF-B activation in epithelial cells. To determine whether NF-B activation is sustained at later stages of the infection, we studied H. pylori infection up to 6 h after infection. Maximal DNA binding of NF-B was observed between 90 and 180 min after infection and was not detectable later than 240 min after infection (Fig. 1a, lanes 20 -22). De novo protein synthesis was not required for H. pylori-induced NF-B activation (data not shown). Furthermore, H. pylori infection or TNF␣ stimulation led to a rapid loss of IB␣ as analyzed in an immunoblot (Fig.  1b). To assess changes in cytokine gene expression, AGS cells were challenged to infection with H. pylori. Cytokine mRNA expression was determined at different time points after infection by reverse transcriptase-polymerase chain reaction of total RNA prepared from infected cells. The cytokine mRNA levels were compared with the constitutive ␤-actin mRNA in the same polymerase chain reactions. Infection with H. pylori led to an increased synthesis of granulocyte-macrophage colonystimulating factor, TNF␣, and interleukin-8 as soon as 45 min after infection (Fig. 1c, lanes 1-4). Treatment of the cells with the proteasome inhibitor lactacystin prior to H. pylori infection suppressed the cytokine mRNA expression (Fig. 1c, lanes 5-8).
The correlation of NF-B activation and IB␣ degradation, which is blocked by a proteasome inhibitor, demonstrates that cytokine/chemokine genes are indeed subject to a coordinate cellular regulation in response to H. pylori infection.
To address whether H. pylori-induced NF-B activation and IB␣ degradation involves IKK activity, we tested the ability of H. pylori to induce phosphorylation of endogenous IKKs. AGS cells were infected for the indicated periods of time. IKK␣ or IKK␤ (Fig. 2, a and b) were immunoprecipitated from cell extracts with appropriate antibodies and analyzed subse-  (Fig. 2, a and b, lower  panels).
To analyze the ability of H. pylori to induce NF-B transactivation activity, we tested the effects of activation of H. pylori on the expression of an NF-B-dependent reporter gene in transiently transfected 293 cells. H. pylori infection (MOI 50) or TNF␣ treatment of cells increased transcription of the reporter gene (Fig. 2c). When we tested the effect of a kinaseinactive IKK␤(K44A) construct on the expression of an NF-Bdependent reporter gene in transiently transfected and H. pylori-infected or TNF␣-treated 293 cells, we observed suppression of NF-B activation. This indicates that H. pylori-induced NF-B activation involves IKK␤ (Fig. 2d). For a control, we analyzed the effect of kinase-inactive MAP kinase kinase (MKK4(K116R)) on the H. pylori-induced NF-B activation, which was not affected.
PAK1 Activates NIK and the IKK Complex-Several kinases (e.g. NIK, mitogen-activated protein kinase (MAPK)/extracellular regulated kinase (ERK) kinase kinase 1 (MEKK1), IKK-i, TANK-binding kinase 1/NF-B-activating kinase, mixed-lineage kinase 3) have been shown to be signaling intermediates that act as direct activators of the IKK complex (2, 20 -24). The cellular selection of the kinases might be dependent on cell type specificity and distinct extracellular stimuli. To study the H. pylori-induced upstream kinase involved in activation of the IKK complex, we analyzed the role of NIK and the MEKK1 in H. pylori-induced expression of an NF-B-dependent reporter gene in transiently transfected 293 cells. H. pylori-or TNF␣stimulated NF-B-dependent reporter gene activity was suppressed when kinase-inactive mutants of NIK (NIK(624 -947); NIK(K429A,K430A)) were expressed (Fig. 3a). In contrast to kinase-inactive NIK, the kinase-inactive MEKK1(K432M) did not significantly block H. pylori-induced NF-B activation (Fig.  3a), whereas phorbol 12-myristate 13-acetate-induced NF-B activation was affected by MEKK1(K432M) (data not shown). These results suggest that H. pylori-induced NF-B activation involves NIK but not MEKK1 in this kinase cascade and exclude a functional cooperation between those kinases in the activation of IKK activity. Compelling evidence that NIK is involved in the activation of NF-B has been shown in response to CD40 and CD3/CD28 induction (25,26).
Our experimental evidence that the MAPK kinase kinase NIK represents a crucial factor in H. pylori-directed activation of the IKK complex led us to search for putative upstream regulators of NIK. Colinear amino acid sequence alignment of the kinase domain in NIK revealed homology to MEKK1 and Ste11 apart from other kinases, and Ste11 represents a downstream kinase of Ste20, the yeast homologue of mammalian PAK1 (27). Therefore, we examined whether PAK1, which is a stress response kinase of the MAPK kinase kinase kinase level, represents an upstream kinase for NIK. In AGS cells the activity of PAK1 was induced severalfold in response to pathophysiological stress raised by H. pylori infection. The kinase activity of immunoprecipitated PAK1 was measured with myelin basic protein as a substrate (Fig. 3b, lanes 1-4, upper  panel). TNF␣ weakly induced PAK1 (Fig. 3b, lanes 5-7, upper  panel). Immunoblot analysis confirmed the presence of similar quantities of PAK1 in each of the extracts used for immunoprecipitation (Fig. 3b, lower panel).
To study the mechanism of NF-B activation, we analyzed whether PAK1 affects the H. pylori-induced NF-B-dependent transcriptional activity. Kinase-inactive PAK1(K299R) blocked the expression of an NF-B-dependent reporter gene in transiently transfected and H. pylori-infected 293 cells in a dosedependent manner (Fig. 3c). In contrast to H. pylori infection, kinase-inactive PAK1(K299R) only slightly affected the NF-B transactivation activity in TNF␣-treated cells (Fig. 3c). To exclude possible effects on the putative upstream components of PAK1 due to titration of Rho-GTPases, we performed experiments with a PAK1(K299R) construct mutated in histidine 83 and 86, which prevents the binding of Rho-GTPases. This construct clearly blocked H. pylori-induced NF-B activation. Furthermore, we used a PAK1 construct containing the residues 83-149. PAK1(83-149) has been demonstrated to inhibit the autophosphorylation of PAK1 by blocking a critical phosphoacceptor site that is required for its kinase activity and does not bind to Rac1 (28). Overexpression of this construct blocked H. pylori-induced activation of NF-B in a dose-dependent manner (Fig. 3c). For a control, we transiently transfected a PAK1-(83-149) construct with the leucine 107 to phenylalanine mutation (PAK1 (83-149,L107F)), which inactivates the autoinhibitory domain of PAK1. This construct did not block H. pylori-induced NF-B activation. To determine the signaling pathways that couple PAK1 to NF-B activation, we tested To identify NIK as a potential downstream component of PAK1 in H. pylori infection, we studied the interaction of these kinases. AGS or HeLa cells were transfected with Ha-tagged PAK1 constructs or Flag-tagged NIK constructs. The kinases were then immunoprecipitated. The immunoprecipitates were subjected to SDS-PAGE and analyzed in an immunoblot with the indicated antibodies. We observed interaction of Ha-PAK1(K299R) with Flag-NIK (Fig. 4a, lane 1), high affinity interaction between Ha-PAK1(K299R) and Flag-NIK(K429A, K430A) (lane 2), and interaction of Ha-PAK1 with Flag-NIK(K429A,K430A) (lane 3). We also performed these experiments with constitutive active Ha-PAK1(L107F), which resulted in a detectable interaction with Flag-NIK(K429A, K430A) (lane 4). These results indicate that kinase-active PAK1 could target the MAPK kinase kinase NIK. In contrast to the PAK1/NIK interaction, we observed no direct interaction between PAK1 and the IKK complex components IKK␣, IKK␤, or NEMO/IKK␥ (Fig. 4, b and c). Appropriate controls show that NIK interacted with IKK␣ (Fig. 4b) and that IKK␤ interacted with NEMO/IKK␥ (Fig. 4c). Similar to the kinase-inactive PAK1(K299R), the constitutively active PAK1(L107F) did not interact with an IKK subunit (data not shown).
To analyze endogenous PAK1/NIK interaction, we tried to  1, 5,  and 6, upper panel). Immunoblot analysis confirmed the presence of similar quantities of Flag-NIK and endogenous PAK1 in each of the extracts used for immunoprecipitation (Fig. 4d,  middle and lower panels). To study whether this interaction is direct, we carried out an in vitro analysis using 35 S-labeled proteins translated in wheat germ extracts. After cotranslation of Flag-NIK and HaPAK1 (Fig. 4e, lane 1) followed by anti-Flag immunoprecipitation (lane 4), we detected PAK1 in the immunoprecipitate. The converse experiment of immunoprecipitation with an anti-Ha antibody allowed the detection of Flag-NIK in the immunoprecipitate (data not shown).
Our results raised the possibility that PAK1 could function as the kinase that activates NIK in H. pylori infection. The analysis of H. pylori-infected cells, which were transfected with kinase-inactive NIK(K429A,K430A), resulted in the occurrence of an additional NIK form with a slightly decreased mobility in SDS-PAGE as detected in an immunoblot (Fig. 5a, lane 2). When NIK(K429A,K430A) was expressed without H. pylori infection, this slower migrating form of NIK was weakly visible (Fig. 5a, lane 1). Cells cotransfected with constitutive active PAK1(L107F) caused an accumulation of the more slowly mi- grating band of NIK(K429A,K430A) (Fig. 5b, lane 4), whereas cotransfection with kinase-inactive PAK1(K299R) suppressed this slower band (lane 3). The upshifted band of NIK in H. pylori-infected cells was eliminated by phosphatase treatment (Fig. 5a, lane 5), suggesting that it represented phosphorylated NIK. These data suggest that, once activated, PAK1 binds and phosphorylates NIK.
To show that PAK1 is critical for H. pylori-induced activation of the IKK complex, IKK␣ was cotransfected with kinase-inactive PAK1(K299R), and the cells were either infected with H. pylori or stimulated with TNF␣. Kinase assays of immunoprecipitated IKK␣ were performed, and the effect of kinase-inactive PAK1 on IKK␣ phosphorylation was analyzed (Fig. 5b). Kinase-inactive PAK1(K299R) (Fig. 5b, lane 5) as well as NIK(K429A,K430A) (lane 6) could efficiently block the H. pylori-induced IKK␣ phosphorylation (lane 4), whereas TNF␣induced IKK␣ phosphorylation (lane 2) was not blocked by PAK1 (K299R) (lane 3). These findings strongly suggest that PAK1 and NIK represent crucial components in H. pyloriinduced NF-B activation. Thus, like Cot/TPL2, TAK1(TAB1/ TAB2), Raf, and PKC (26, 29 -34), the activation of PAK1 results in the activation of NF-B by certain stimuli, but distinct from pathways used by TNF␣. The activation of the IKK complex by these kinases is indirect and involves other kinases or adaptor molecules, and not all signals proceed through IKK␣. It has been shown that PKC-induced signaling targets IKK␤ (35) and that the adaptor protein receptor-interacting protein targets NEMO/IKK␥ to recruit the IKK complex to the TNF receptor (36), which could involve TRAF2 (37). DISCUSSION In summary, our results identify PAK1 as an integral component of the H. pylori-induced activation of NF-B in epithe-lial cells. Activated PAK1 phosphorylates NIK and directs the activity of IKKs to IB␣. This signaling causes NF-B activation and induction of proinflammatory cytokines in H. pyloriinfected gastric epithelial cells (Fig. 6). Because PAK1 does not play an important role in NF-B activation by TNF␣ or phorbol 12-myristate 13-acetate, this study corroborates the concept that different NF-B-activating stimuli use different signaling components.
Because the integrity of the type IV secretion machinery is an absolute requirement for NF-B activation in H. pylori infection (38), it is reasonable to speculate that the component(s) that induce(s) this transcription factor must either be injected into the eucaryotic target cell or activate(s) contact-dependent eucaryotic cell-surface receptors. This hypothesis is supported by the notion that only those PAI factors (CagA, CagF, and CagN) that are not necessary for the functional integrity of the type IV secretion apparatus do not induce activation of NF-B (15). Thus PAK1 functions in gastric epithelial cells in a unique pathway that links H. pylori-dependent effector molecules to the activation of NF-B and the induction of the innate immune response. The central role of NF-B in the innate immune response suggests that many pathogenic microorganisms have evolved mechanisms to interfere with the function of NF-B or modulate the NF-B signal transduction. In contrast to H. pylori, the microbial pathogen Yersinia pseudotuberculosis inhibits the activation of NF-B by active translocation of the virulence factor yersinia outer protein J via a type III secretion system. The yersinia outer protein J protein was shown to bind to IKK␤ and the superfamily of mitogenactivated protein mitogen-activated protein kinase kinases and inhibits phosphorylation and subsequent activation of these kinases (39). Thus, by blocking IKK␤ and MKKs Yersinia in- hibits the release of cytokines and other immunomodulatory factors during innate immune response. The identity of the effector molecule(s) of H. pylori that induce(s) NF-B is not known so far. We speculate that PAK1 becomes targeted directly or indirectly by H. pylori factor(s). The identification of the H. pylori factor(s) in this process of NF-B activation will unravel the mechanism of the H. pylori-induced signaling in the future. Interestingly, the viral protein negative regulatory factor from human simian immunodeficiency virus targets PAK1 and induces nuclear responses. The negative regulatory factor-induced mechanism of PAK1 activation involves the si-multaneous interaction of negative regulatory factor with the SH3 domain of Vav, which induces its guanine nucleotide exchange factor activity for Cdc42 and Rac1. The activation of Rac1 and Cdc42 leads to their subsequent dissociation from Vav and strongly increases their affinity for PAK1 (40).
Based on their structure the PAK kinases in mammals can be divided into two subfamilies: the PAK subfamily, containing an NH 2 -terminal catalytic p21-binding domain (also known as cdc42/Rac1-interactive binding domain) and a COOH-terminal kinase domain, and the germinal center kinase-like subfamily, which contains an NH 2 -terminal catalytic domain and lacks the p21-binding domain. In the first description of PAK it was shown that the serine/threonine protein kinase activity of PAK could be stimulated by the binding of activated GTP-bound Rac1 and Cdc42. More recently, a variety of studies have suggested that PAKs can participate in a broad range of cellular events that include cytoskeletal responses as well as certain signaling events (41). For example, PAKs can catalyze the phosphorylation of the heavy chain in myosin 1 (42), in p47phox (43), and in LIM kinase (44). In our study we have shown for the first time that PAK1 interacts with and phosphorylates NIK, which represents a specific activating mechanism in NF-B regulation in H. pylori-infected epithelial cells. The recent observation that PAK1-mediated NF-B activation does not involve activation of the IKKs could be explained by a different experimental approach (45). In our future work we plan to analyze the whole process of H. pylori-induced NF-B activation including the identification of the microbial factors. Regarding the phosphorylation of NIK by PAK1, further experiments will determine the phosphorylation sites. NIK putatively contains a consensus site for phosphorylation by PAK (46). Therefore, NIK phosphorylation could take place in the NH 2 -terminal region, possibly activating NIK to autophosphorylate and stimulate catalytic activity, as proposed for the activation of NIK by TAK1 (29). Because PAK1 is known to activate the JNK pathway in H. pylori infection (47), this finding contributes to our understanding of how a given stimulus can simultaneously activate JNK as well as NF-B. Other kinases could also act as dual activators of JNK and NF-B, e.g. MEKK1 or mixed-lineage kinase 3 activate JNK via phosphorylation of MKK4 (48,49) and trigger NF-B activation by direct phosphorylation of IKKs (24,35). Furthermore, dual activation of JNK and NF-B activities has been shown by ectopic expression of plenty of SH3s, a Rac1-regulated protein, which consists of four SH3 domains (50).
Gastric inflammation is a hallmark of H. pylori infection, and our work contributes to the elucidation of the steps involved in H. pylori-induced NF-B signal transduction and activation of an innate host immune response. The understanding of the host cell response mechanism may reveal potential targets for drug intervention in the case of pathological activation of this transcription factor in inflammatory diseases.