The β₁ Integrin Activates JNK Independent of CagA, and JNK Activation Is Required for Helicobacter pylori CagA⁺-induced Motility of Gastric Cancer Cells^,

Abstract

The Helicobacter pylori CagA protein is translocated into gastric epithelial cells through a type IV secretion system (TFSS), and published studies suggest CagA is critical for H. pylori-associated carcinogenesis. CagA is thought to be necessary and sufficient to induce the motogenic response observed in response to CagA⁺ strains, as CagA interacts with proteins involved in adhesion and motility. We report that H. pylori strain 60190 stimulated AGS cell motility through a CagA- and TFSS-dependent mechanism, because strains 60190ΔcagA or 60190ΔcagE (TFSS-defective) did not increase motility. The JNK pathway is critical for H. pylori-dependent cell motility, as inhibition using SP600125 (JNK1/2/3 inhibitor) or a JNK2/3-specific inhibitor blocked motility. JNK mediates H. pylori-induced cell motility by activating paxillin, because JNK inhibition blocked paxillin^Tyr-118 phosphorylation, and paxillin expression knockdown completely abrogated bacteria-induced motility. Furthermore, JNK and paxillin^Tyr-118 were activated by 60190ΔcagA but not 60190ΔcagE, demonstrating CagA-independent signaling critical for cell motility. A β₁ integrin-blocking antibody significantly inhibited JNK and paxillin^Tyr-118 phosphorylation and cell scattering, demonstrating that CagA-independent signaling required for cell motility occurs through β_1. The requirement of both Src and focal adhesion kinase for signaling and motility further suggests the importance of integrin signaling in H. pylori-induced cell motility. Finally, we show that JNK activation occurs independent of known upstream kinases and signaling molecules, including Nod1, Cdc42, Rac1, MKK4, and MKK7, which demonstrates novel signaling leading to JNK activation. We report for the first time that H. pylori mediates CagA-independent signaling that promotes cell motility through the β₁ integrin pathway.

Helicobacter pylori infects one-half of the world's population, establishing a chronic infection in the gastric mucosa that persists for the lifetime of the host (1, 2). Although most infections only manifest as superficial gastritis, many progress to mucosal necrosis, ulceration, and atrophic gastritis, a precursor lesion of gastric adenocarcinoma (3). In animal studies, 37% of gerbils infected with virulent strains of H. pylori developed stomach tumors, demonstrating a direct link between H. pylori and gastric carcinogenesis (4). Additionally, epidemiological studies suggest that H. pylori infection increases the risk of developing gastric cancer 6-fold, emphasizing the importance of this bacterium in gastric carcinogenesis (5).

H. pylori pathogenesis varies based on the expression of virulence factors used for bacterial colonization and disease progression. The vacA gene is encoded by virtually all H. pylori strains, but the intense vacuolation caused by VacA varies based on genetic mosaicism (6). Peptic ulceration strongly correlates with strains encoding the most active forms of VacA (6).

The cag pathogenicity island (cag PAI)3 contains 32 genes, many of which encode components of a putative type IV secretion system (TFSS). The only known protein transported by the TFSS is CagA, which is also expressed from the cag locus (7). During infection, CagA translocates into gastric epithelial cells via the TFSS and is phosphorylated at multiple sites by Src family kinases and c-Abl (8–10). CagA then influences signal transduction pathways by docking with host signaling proteins (11–14). Patients infected with CagA-positive H. pylori strains exhibit higher grades of gastric inflammation, atrophic gastritis, and an increased risk of the development of gastric adenocarcinoma (15–17).

In vitro experiments show that epithelial cells cultured with CagA⁺ bacteria transition from the unstimulated “cobble-stone” morphology to the “hummingbird” phenotype indicative of motile cells (18–21). Additionally, H. pylori stimulates gastric cancer cell invasion through in vitro basement membranes, suggesting a role for H. pylori in cancer progression to metastasis (22–24). The mechanism of H. pylori-induced cell motility and invasion are unclear, although recent studies show that CagA associates with key regulators of cell adhesion, motility, and invasion, including c-Met, Grb2, SHP-2 phosphatase, and ZO-1 (11, 12, 14, 22, 25). The critical role of CagA in cancer cell motility was emphasized by Higashi et al. (11) who showed that CagA transfection of AGS cells was sufficient to induce the motile phenotype. These data suggest that CagA stimulates all signaling necessary to induce cell motility, although this hypothesis is not universally accepted (22).

In this study, we identified JNK as a key mediator of H. pylori-stimulated cell motility, and we found that JNK was activated through a CagA-independent but still TFSS-dependent mechanism. We then evaluated the molecular mechanism of CagA-independent JNK activation, and we determined that CagA-independent activation of JNK occurs through β₁ integrin and Src signaling. Furthermore, we identified paxillin as a downstream target of H. pylori-dependent JNK activation and that paxillin activity is required for H. pylori-stimulated gastric cancer cell motility. These data show for the first time a mechanism of CagA-independent signaling that promotes cell motility, and we conclude that a combination of CagA-dependent and CagA-independent cell signaling is required for H. pylori-induced gastric cancer cell motility.

EXPERIMENTAL PROCEDURES

Cell Culture and Reagents—AGS gastric adenocarcinoma cells (ATCC, Manassas, VA) were cultured in Ham's F-12 media (Mediatech, Herndon, VA) supplemented with 10% fetal calf serum (Gemini, West Sacramento, CA), 100 μg/ml streptomycin, 100 IU/ml penicillin (Mediatech, Herndon, VA), and 0.0375% (w/v) sodium bicarbonate (Mediatech, Herndon, VA) at 37 °C in 5% CO₂. Cells were cultured to 75% confluency and split either 1:5 or 1:10 using 0.025% EDTA to gently detach cells from plastic. To prevent spontaneous cell scattering associated with increased passage number, fresh stocks were thawed out monthly.

Pharmacological inhibitors LY294002, SP600125, JNK2/3 inhibitor (IX), and Bay11-7082 were obtained from EMD Biosciences (San Diego). Cycloheximide and PP2 were obtained from Sigma. The β₁ blocking antibody AIIB2 was obtained from Developmental Studies Hybridoma Bank (Iowa City, IA). Prior to experiments, subtoxic concentrations of each inhibitor were selected, and Western blots to detect phosphorylated proteins were performed to confirm the functionality of the inhibitors at the concentrations used. For experiments, cells were pretreated with inhibitors for 30 min (LY294002, 50 μm; SP600125, 40 μm; cycloheximide, 10 μg/ml), 2 h (inhibitor IX, 625 nm), 2 h (PP2, 25 μm), or 3 h (Bay11-7082, 0.5 μg/ml) prior to the addition of bacteria.

Bacterial Strains and Culture Conditions—H. pylori strains 60190 (ATCC 49503, cag PAI⁺, vacA s1/m1) and Tx30a (ATCC 51932, cag PAI^-, vacA s2/m2) were obtained from ATCC (Manassas, VA) and grown on trypticase soy agar plates supplemented with 5% adult defibrinated bovine blood (Gemini, West Sacramento, CA) at 37 °C in 5% CO₂ overnight prior to use in experiments. H. pylori mutant strains with disrupted cagA (60190ΔcagA), cagE (60190ΔcagE), and vacA (60190ΔvacA) genes were a kind gift from Dr. Richard Peek (Vanderbilt University, Nashville, TN). These strains were grown on the same plates as wild-type bacteria but under kanamycin selection (50 μg/ml). Bacteria were passaged daily, and fresh bacteria were thawed on a monthly basis.

The H. pylori strain G27 was used in this study, and isogenic mutants, cagA and cagM, were constructed by natural transformation of a kanamycin cassette flanked by regions homologous to the disrupted genes (26, 27). H. pylori G27 and mutant strains were routinely cultured on horse blood agar (blood base agar number 2, 8% (v/v) horse blood (Bio-Lab, Victoria, Australia)) supplemented with antibiotics (27). Bacteria were grown at 37 °C for 1–2 days under microaerobic conditions in an anaerobic jar containing a Campygen gas mix of 5% O₂, 10% CO₂, and 85% N₂ (Oxoid, Hampshire, UK). Liquid broth cultures were incubated in 25-cm³ tissue culture flasks (Iwaki, Japan) in a final volume of 10 ml of brain heart infusion broth containing 10% (v/v) fetal bovine serum (Thermo Electron, Melbourne, Australia) and with shaking at 125 rpm and 37 °C overnight, prior to use in experiments. H. pylori isogenic mutants with disrupted cagA (G27ΔcagA) and cagM (G27ΔcagM) were grown on the same plates as wild-type bacteria but under kanamycin selection (20 μg/ml).

Infection of AGS Cells—At least 1 day prior to infection, AGS cells were seeded at a density of 2.5–6 × 10⁴ cells/ml in antibiotic-free media. For each experiment, 1 day-old bacteria were suspended in warmed, CO₂-charged antibiotic-free media, and bacterial density was measured by spectrophotometer at 600 nm. Bacteria were then added to cells at a multiplicity of infection (m.o.i.) of 100. Bacterial contact with cells was synchronized by centrifugation at 600 × g for 4 min, after which cells were maintained at 37 °C and 5% CO₂ throughout each experiment. Control cells were prepared under identical conditions.

Scatter Assays—Cells were cultured alone or with bacteria ± inhibitors for 18–20 h, and monolayers were fixed with 4% paraformaldehyde. After three washes with PBS, F-actin was fluorescently labeled with Oregon Green 488 phalloidin (Molecular Probes, Eugene, OR) suspended in BSP (bovine serum albumin, saponin, PBS). Fluorescent and phase images were acquired by wide field fluorescent microscopy.

Western Blot Analysis—Lysates from bacteria alone or cells co-cultured with bacteria were collected in boiling protein loading buffer (0.125 m Tris-HCl, pH 6.8, 4% SDS, 0.13 mm bromphenol blue, 1 m sucrose). Proteins were separated by 7.5–12% acrylamide gel electrophoresis and transferred to nitrocellulose or polyvinylidene difluoride membranes (Pall, Pensacola, FL). Membranes were blocked with 5% milk or 0.1% gelatin in TBST (20 mm Tris, 137 mm NaCl, 0.1% Tween 20, pH 7.5) and incubated overnight with antibodies specific for tubulin (Neo-Markers, Fremont, CA), CagA (Austral Biologicals, San Ramon, CA), β-actin (Sigma), Rac1 (BD Biosciences), phospho-JNK, phospho-AKT, JNK, Cdc42, phospho-MKK4, MKK4, phospho-MKK7, MKK7, phospho-paxillin^Tyr-118, and paxillin (Cell Signaling Technology, Beverly, MA) and phospho-paxillin^Ser-178 (EMD Biosciences, San Diego) in 5% bovine serum albumin or 0.1% gelatin in TBST. Blots were followed with horseradish peroxidase-conjugated secondary antibodies (Amersham Biosciences), and proteins were detected by ECL (Amersham Biosciences).

Colloidal Gold Motility Assay—Assay was derived from colloidal gold phagokinetic assay as described previously (28). Briefly, 12-mm coverslips were immersed in a gelatin solution (Sigma, 300 bloom; 500 mg in 300 ml of water) and heated at 90 °C for 10 min. The gelatin was then removed, and the coverslips were dried at 70 °C for 45 min. After the coverslips cooled, they were aseptically transferred to 24-well plates. The colloidal gold suspension was prepared by mixing 11 ml of water, 6 ml of a 36.5 mm Na₂CO₃ solution, and 2 ml of a 14.5 mm AuHCl₄ solution (Fisher). The mixture was gently swirled high over a Bunsen burner until the first sign of boiling, after which the flask was immediately removed from heat, and 2 ml of a 0.1% formaldehyde solution was quickly added.

The hot solution was allowed to slightly cool on the bench top as the flask was swirled, during which time the solution turned moderately brown in color. After the gold solution changed color, 2 ml were added atop each coverslip in the 24-well plate, and the plates were incubated at 37 °C and 5% CO₂ for 3 h to allow the gold particles to settle onto the coverslips. Coverslips were then gently washed and stored in PBS at 4 °C until use.

For motility assays, 1 × 10⁴ cells were seeded onto prepared coverslips and spun at 600 × g to maximize cell attachment to the substratum.4 After 6–12 h of recovery time, bacteria were added as described above. In inhibition assays, the inhibitors were added 30 min prior to the addition of bacteria, except for SU11274, which was added overnight following a 6-h recovery period.

After 18–22 h, cells were fixed with 4% paraformaldehyde and permeabilized with BSP for 1 h at room temperature without agitation. Coverslips were rinsed three times with PBS and mounted onto slides with Slowfade Gold antifade reagent with 4′,6-diamidino-2-phenylindole (Molecular Probes, Eugene, OR). Phase and fluorescent images were taken of each field, and the area was cleared by single or small colonies of cells measured using ImageJ software (National Institutes of Health). The area was then divided by the number of nuclei in the corresponding fluorescent image to give the average area cleared per cell. Fifteen to 30 fields were visualized in this manner for an average of 100 cells per coverslip, and between one and three coverslips were used per experimental condition. Data for two to three separate experiments were compared and presented as the fold change over control or area cleared per cell. Standard deviation and error were included, and p values were calculated by either the paired two-sample t test for means (Microsoft Excel) or analysis of variance with Tukey's HSD test (GraphPad Prism).

Short Interfering RNA (siRNA) Transfection—Transfection protocol was modified from Chan et al. (29). Briefly, 1.25 × 10⁴ AGS cells were seeded into a 24-well plate, and the following day were transfected with Lipofectamine 2000 (Invitrogen) according to product protocol using 1.5 μl of stock Cdc42 or luciferase siRNA (20 μm; a kind gift from Dr. Marc Symons, Feinstein Institute for Medical Research, Manhasset, NY). After 2 days fresh medium was added, and cells were either infected with H. pylori for scatter assays or collected for Western blot analysis.

Lentiviral Delivery of Short Hairpin (sh) RNA—Stable shRNA cell lines were generated using MISSION shRNA lentiviruses (Sigma) according to manufacturer's protocol. Briefly, 5 × 10⁴ AGS cells were seeded in 6-well dishes, and each well was left untreated or infected with one of the five lentiviral clones provided in each target transcript kit (m.o.i. of 0.5) plus Polybrene (8 μg/ml). The following day, virus was aspirated and replaced with fresh media, and the next day the cells were washed, and media containing puromycin (0.6 μg/ml) was added for selection of stable transfectants. After 3 days of daily washes, stable clones were screened for maximal target protein expression knockdown. The clones selected were as follows: Rac1 (MISSION shRNA TRCN0000004873), MKK4 (MISSION shRNA TRCN0000039916), MKK7 (MISSION shRNA TRCN0000000587), paxillin (MISSION shRNA TRCN-0000123138), and non-target control (MISSION shRNA SHC002V). Stable GFP shRNA cell lines were generated by infecting AGS cells with the lentiviral shRNA vector GFP-FSIPPW (a kind gift from Dr. Andrew Kung, Dana-Farber Cancer Institute, Boston). Stable cells were cultured in puromycin-containing media.

Adenovirus Infection—Transient expression of GFP-tagged FAK-related non-kinase (Ad-GFP-FRNK) or GFP (Ad-GFP) was facilitated through adenoviral delivery. Ad-GFP-FRNK and Ad-GFP viruses were a kind gift of Dr. Joan Taylor (University of North Carolina, Chapel Hill). AGS cells were incubated overnight with virus at an m.o.i. of 10. The following day, cells were then infected with H. pylori strains for either 1 h (Western blot analysis) or overnight. After overnight incubation, cells were fixed and stained for F-actin using Alexa Fluor 546 phalloidin (Molecular Probes). GFP and Texas Red images were captured, and merge images were generated using ImageJ (National Institutes of Health).

Nod1KD Cell Analysis—AGS cells stably expressing siRNAs to either the caspase-activation recruitment domain of Nod1 or an irrelevant gene (EGFP) were generated by Dr. J. Viala (Institut Pasteur, Paris) (27). Briefly, AGS cells were transfected with plasmid constructs in which short hairpin RNA to the genes of interest had been cloned into psiRNA-hHIneo (InvivoGen, San Diego). To select for clones that had stably incorporated the respective plasmids into their genomic DNA, the cells were grown in RPMI 1640 medium containing 10% fetal calf serum and 400 μg/ml G418 (Invitrogen). G418-resistant cells were isolated and expanded so as to generate stable knockdown clones for NOD1 or EGFP genes. The characterization of these clones will be described in detail elsewhere.5 Successful knockdown of Nod1 expression in these cells was published previously (30).

Chemokine Analysis—IL-8 production by EGFP and Nod1KD cells induced by H. pylori co-culture was determined by collecting 24-h supernatants and using the OptEIA kit (BD Biosciences).

RESULTS

H. pylori-dependent AGS Cell Scattering and Motility Require CagA and the TFSS—Epithelial cell lines co-cultured with Cag PAI⁺ H. pylori strains exhibit the hummingbird phenotype associated with epithelial-mesenchymal transition (14, 18). To demonstrate that a similar phenotype occurred in response to exposure to H. pylori in our experimental system, AGS gastric cancer cells were co-cultured with H. pylori 60190 (Cag PAI⁺, vacuolating) or Tx30a (Cag PAI^-, nonvacuolating) for 18 h, and cells were fixed and fluorescently labeled for F-actin. As shown in Fig. 1A, panel ii, AGS cells cultured with H. pylori 60190 exhibited the hummingbird phenotype indicative of motile cells, whereas cells cultured with H. pylori Tx30a (Fig. 1A, panel iii) showed no morphological changes compared with untreated cells (Fig. 1A, panel i).

FIGURE 1.

H. pylori-stimulated AGS cell scattering and motility are dependent upon CagA and the TFSS. A, AGS cells were cultured alone (panel i) or with 60190 (panel ii), Tx30a (panel iii), 60190ΔcagA (panel iv), 60190ΔcagE (panel v), or 60190ΔvacA (panel vi) for 18 h, after which cells were fixed and stained for F-actin. B, Western blot analysis was performed on bacterial lysates using a CagA-specific antibody. C, graph represents the effect H. pylori strains have on cell motility. Data presented as a fold change compared with control cells in three separate experiments and include the standard error. UT, untreated. ^*, untreated versus 60190 or VacA (p < 0.01, using paired two-sample t test for means). All micrographs, blots, and motility data are representative of multiple experiments.

To determine the bacterial factors required for AGS cell scattering, isogenic mutants of H. pylori 60190 with gene disruptions in cagA (60190ΔcagA), cagE, encoding a gene product required for full TFSS functionality (60190ΔcagE) or vacA (60190ΔvacA) were used in co-culture assays. Western blot analysis was first performed on whole bacteria lysates to confirm that the mutant strains 60190ΔcagE and 60190ΔvacA, but not 60190ΔcagA, expressed CagA (Fig. 1B). AGS cells were then co-cultured with each of these strains, and the extent of cell scattering was determined by immunofluorescence microscopy. As shown in Fig. 1A, panels iv and v, 60190ΔcagA and 60190ΔcagE, respectively, did not induce wild-type cell scattering, demonstrating that the delivery of CagA into host cells is necessary to induce the hummingbird phenotype. The mutant strain 60190ΔvacA caused a scattering phenotype similar to wild-type 60190 (Fig. 1A, panels vi and ii, respectively), demonstrating that the vacuolating cytotoxin of H. pylori plays no role in the induction of this morphological response.

Previous studies have measured H. pylori-induced cell motility as the percent of cells per field that exhibit the hummingbird phenotype (21, 31–33). To quantitate the level of participation of bacterial and host proteins in cell motility, a modified colloidal gold phagokinetic assay was utilized to investigate the role of CagA, the TFSS itself (using the CagE mutant), and the vacuolating cytotoxin in cell motility. Briefly, cells and bacteria were seeded onto a colloidal gold substrate, and cell motility was measured as a function of the area that cells cleared as they moved during the assay. As shown in Fig. 1C, H. pylori 60190 stimulated a 2-fold increase in cell motility over untreated cells. Additionally, only the VacA mutant caused a comparable increase in motility over untreated cells, although no significant increase in motility was observed by Tx30a and the CagA or TFSS mutants. These data correlate with our scatter data from Fig. 1A and demonstrate that cell motility is a CagA- and TFSS-dependent but VacA-independent event.

H. pylori-induced Cell Scattering and Motility Require JNK Signaling—Recent evidence shows that the JNK pathway is a key mediator of cytoskeletal extensions and cell motility in a number of experimental systems (34, 35). To determine whether JNK plays a role in H. pylori-induced cell scattering and motility, AGS cells were pretreated with the pan-JNK inhibitor, SP600125 or DMSO for 30 min prior to the addition of H. pylori strains for 18 h (Fig. 2A). Cells pretreated with SP600125 did not scatter in response to H. pylori compared with DMSO control cells (Fig. 2A), demonstrating that JNK is required for cell scattering. Motility assays were then employed, and Fig. 2B shows that SP600125 significantly decreased AGS cell motility, demonstrating that JNK signaling is required for H. pylori-induced cell scattering and motility.

FIGURE 2.

H. pylori-induced cell motility requires JNK signaling. A, cells were co-cultured alone (panels i and iii) or with H. pylori strain 60190 for 18 h following pretreatment for 30 min with JNK inhibitor SP600125 (40 μm) or with DMSO as a carrier control. Cells were fixed and stained for F-actin. B, graph represents the effects of JNK inhibition on H. pylori-induced cell motility. Data are presented as fold change compared with control cells in three separate experiments and include standard error. C, AGS cells were co-cultured with the indicated H. pylori strains for 2 h, and lysates were collected. Western blot analysis was performed to determine the JNK activation profile using phospho-specific antibodies. Total protein was also probed as a load control. Arrowheads indicate the 54-kDa JNK2/3 isoforms and the 46-kDa JNK1 isoform. UT, untreated. ^*, untreated versus H. pylori (p < 0.01); ^**, H. pylori versus H. pylori + inhibitor (p < 0.01, using paired two-sample t test for means). All micrographs, blots, and motility data are representative of multiple experiments.

H. pylori Activates JNK in a CagA-independent, TFSS-dependent Manner—The report by Higashi et al. (11) suggests that CagA is sufficient to stimulate all pathways required for cell motility. If this hypothesis is true, then all signaling pathways required for H. pylori-induced cell motility would require CagA delivery for activation. To test this hypothesis, Western blot analysis was performed on cells co-cultured with the parental and mutant H. pylori strains, and JNK phosphorylation in response to different bacterial stimuli was analyzed (Fig. 2C). Surprisingly, JNK was phosphorylated in response to co-culture with 60190, 60190ΔcagA, or 60190ΔvacA, whereas strains lacking a functional TFSS (Tx30a and 60190ΔcagE) showed no induction. These data show that JNK signaling, which is required for H. pylori-stimulated cell motility, is activated in a CagA-independent but still TFSS-dependent manner.

H. pylori Stimulates JNK through a Nod1-independent Mechanism—Although activation of JNK occurs in a CagA-independent but still TFSS-dependent manner, the mechanism of TFSS-dependent JNK signaling is unknown. A recent report by Viala et al. (27) demonstrated that peptidoglycan, a component of the bacterial cell wall, is transported into the cytoplasm and recognized by the pathogen recognition molecule Nod1, causing NF-κB activation and IL-8 secretion. Nod1 is also reported to regulate JNK and p38 activity, and therefore peptidoglycan-mediated Nod1 induction may be the mechanism of CagA-independent JNK activation leading to cell motility (36, 37). To test this hypothesis, cells stably expressing siRNAs targeting Nod1 (Nod1KD) or an irrelevant gene (EGFP) were analyzed for JNK phosphorylation after co-culture with H. pylori strain G27 and isogenic mutant strains G27ΔcagA or G27 ΔcagM (TFSS-defective) (26). Nod1 expression was significantly abolished in the Nod1KD cells, as published previously (30). To confirm a knockdown of Nod1 expression, H. pylori-stimulated IL-8 production was analyzed by enzyme-linked immunosorbent assay, and supplemental Fig. 1A confirms a significant decrease in IL-8 production by Nod1KD cells in response to H. pylori G27. In supplemental Fig. 1B, G27 stimulated JNK phosphorylation in a CagA-independent manner in both EGFP and Nod1KD cell lines, which demonstrates two important points. First, CagA-independent JNK activation is not specific to H. pylori strain 60190. Second, the bacterium stimulates JNK phosphorylation independent of Nod1 signaling. Furthermore, supplemental Fig. 1C shows that Nod1KD cells exhibit the scattered phenotype in response to H. pylori similar to EGFP control cells. These observations demonstrate that H. pylori activates JNK-dependent cell scattering through a Nod1-independent mechanism.

H. pylori-induced JNK Phosphorylation and Cell Scattering Occur through Integrin Signaling—A recent publication by Kwok et al. (38) showed that the TFSS requires the interaction between the TFSS-associated CagL protein and the α₅β₁ integrin complex to initiate translocation of CagA into the cell. This interaction also activates FAK and Src, both of which play a key regulatory role in integrin signaling (38). This group pretreated AGS cells with a β₁-blocking antibody (AIIB2), which prevented CagA translocation and phosphorylation (38). Evidence also shows that integrin signaling can stimulate JNK activity (39). To determine whether TFSS-dependent β₁ signaling causes activation of JNK leading to H. pylori-induced cell motility, AGS cells were pretreated with AIIB2 for 1 h prior to co-culture with H. pylori 60190. Western blot analysis was then performed, and Fig. 3A shows that incubation with the β1 blocking antibody significantly, though not completely, blocked JNK phosphorylation. Scatter assays were also performed, and Fig. 3B shows that pretreatment with AIIB2 significantly blocked the robust scattering phenotype induced by H. pylori alone, although the cell colonies loosen up, indicative of incomplete inhibition of motility. These data demonstrate, however, that β₁ integrin signaling is required for CagA-independent JNK phosphorylation and cell scattering.

FIGURE 3.

The β₁ integrin is required for H. pylori-dependent JNK phosphorylation and cell scattering. A, AGS cells were pretreated for 1 h with the β₁ blocking antibody AIIB2 (5 μg/ml) or an equal aliquot of serum-free media (SF Media) prior to co-culture with H. pylori 60190, and cell lysates were analyzed for phospho-JNK activity by Western blot analysis using phospho-specific antibodies. Tubulin was also probed as a load control. B, AGS cells were pretreated for 1 h as indicated in A and then cultured alone or with H. pylori 60190 for 18 h. Cells were then fixed and stained for F-actin. All blots and micrographs are representative of multiple experiments.

Kwok et al. (38) also show that H. pylori-mediated β₁ integrin stimulation results in activation of both Src and FAK. These two kinases form a signaling complex that mediates downstream integrin signaling (40). Therefore, we tested if Src and FAK influenced JNK phosphorylation and cell scattering. In Fig. 4A, cells pretreated with the Src inhibitor, PP2, showed a significant decrease in H. pylori-stimulated JNK phosphorylation compared with DMSO alone. Additionally, PP2 blocked H. pylori-induced cell scattering, as shown in Fig. 4C. These data demonstrate that Src is required for H. pylori-induced JNK activation and cell scattering.

FIGURE 4.

Src, but not FAK, is required for H. pylori-induced JNK phosphorylation. A, AGS cells were pretreated with DMSO or PP2 for 2 h before culture alone or with H. pylori 60190 for 1 h. Cells were collected and analyzed for phospho-JNK activity by Western blot analysis using phospho-specific antibodies. B, AGS cells were incubated overnight with adenovirus encoding GFP-tagged FAK-related non-kinase (Ad-GFP-FRNK) or GFP alone (Ad-GFP) as a control (m.o.i. of 10). Cells were then co-cultured for 1 h with H. pylori 60190. Lysates were analyzed for JNK activity. Tubulin was also probed as a load control. Arrowheads indicate the 54-kDa JNK2/3 isoforms and the 46-kDa JNK1 isoform. C, cells pretreated with either DMSO or PP2 were incubated overnight alone or with H. pylori 60190, and cells were fixed and stained for F-actin.

To address the role of FAK in JNK activation and cell scattering, we employed a protein consisting of the carboxyl-terminal noncatalytic domain of FAK, termed FAK-related non-kinase (FRNK). FRNK is a separate protein endogenously expressed which, when overexpressed, inhibits FAK activity (41). Therefore, to inhibit FAK in our studies, GFP-labeled FRNK or GFP alone was expressed in AGS cells using a replication-defective adenovirus construct (Ad-GFP-FRNK and Ad-GFP, respectively) prior to co-culture with H. pylori 60190 for 1 h. As shown in Fig. 4B, FRNK expression did not inhibit H. pylori-induced JNK phosphorylation compared with the GFP control construct, demonstrating that FAK is not required for JNK activity. In Fig. 5, however, cells that expressed GFP-FRNK showed a striking inhibition of H. pylori-induced cell scattering. These cells showed high nuclear FRNK localization and loss of cortical actin and stress fibers, compared with GFP control cells, which still scattered in response to H. pylori. These data demonstrate that although JNK requires Src activity, JNK phosphorylation occurs independent of FAK. But inhibition of both Src and FAK blocks the H. pylori-induced morphogenic response.

FIGURE 5.

H. pylori-induced cell scattering requires FAK activity. AGS cells were incubated overnight with Ad-GFP-FRNK or Ad-GFP. Cells were cultured overnight alone or with H. pylori 60190. Cells were then fixed and stained for F-actin. Images of GFP-labeled and F-actin-labeled cells were then captured.

H. pylori Stimulates JNK through a PI3K, Cdc42-, and Rac1-independent Mechanism—A key pathway that regulates cancer cell survival is the PI3K pathway, which is known to play a role in integrin signaling and JNK activation (42, 43). To determine whether PI3K regulates JNK activity, AGS cells were pretreated with LY294002 prior to co-culture with H. pylori 60190 for 1 h. Western blot analysis was performed on these lysates, and supplemental Fig. 2A shows that H. pylori-induced JNK activity was not affected by LY294002, which demonstrates that JNK is not regulated by PI3K activity.

The Rho GTPases Cdc42 and Rac1 are well known regulators of actin cytoskeletal changes; Cdc42 mediates filopodial protrusions at the leading edge of motile cells, and Rac1 controls the lamellipodial sheets that extend the cell forward (44). Yamauchi et al. (45) used dominant-negative constructs to show that Cdc42 and Rac1 also regulate neuronal extension by stimulating JNK activity. Additionally, H. pylori is known to activate both Cdc42 and Rac1 (46). To determine whether these Rho GTPases play a role in H. pylori-dependent JNK activation and cell motility, siRNA technology was applied to knockdown expression of Cdc42 or Rac1 for Western blot analyses and scatter assays. AGS cells were transiently transfected with Cdc42 siRNAs or control siRNAs (luciferase), co-cultured with H. pylori 60190 for 1 h, and lysates were collected for Western blot analysis. The supplemental Fig. 2B shows that Cdc42 siRNA (siCdc42)-treated cells show considerable loss of Cdc42 compared with control cells (siLuc), but no difference in H. pylori-stimulated JNK phosphorylation was observed between the two transfection conditions. Furthermore, supplemental Fig. 2C shows that there was also no difference in H. pylori-stimulated cell scattering between siCdc42 (panel ii) and siLuc cells (panel i), demonstrating that Cdc42 is not required for JNK phosphorylation or cell scattering.

To determine whether JNK activity is regulated by H. pylori-dependent Rac1 activation, AGS cells stably expressing Rac1 shRNAs were generated by lentiviral delivery (see “Experimental Procedures”). The supplemental Fig. 2D shows that Rac1 shRNA-expressing cells (shRac1) exhibit significant Rac1 protein knockdown compared with GFP shRNA-expressing control cells (shGFP). Additionally, in supplemental Fig. 2, E and F, respectively, Rac1 knockdown did not affect H. pylori-induced JNK phosphorylation or cell scattering. These data demonstrate that H. pylori stimulates JNK through a Cdc42- and Rac1-independent mechanism.

H. pylori Stimulates MKK4 Phosphorylation but Activates JNK Independent of MKK4—The MAP kinase kinase 4 (MKK4) is one of two MAP kinase kinases identified as direct upstream JNK kinases (47, 48). To determine whether H. pylori activates MKK4 to stimulate JNK activation, AGS cells were co-cultured with the indicated H. pylori strains and collected for Western blot analysis. As shown in supplemental Fig. 3A, MKK4 was phosphorylated in response to H. pylori 60190, H. pylori 60190ΔcagA, and H. pylori 60190ΔvacA but not H. pylori 60190ΔcagE, demonstrating that MKK4 is activated in a CagA-independent, TFSS-dependent manner similar to JNK. To further determine whether MKK4 was required for JNK phosphorylation, AGS cells stably expressing shRNAs against MKK4 (shMKK4) were generated. The supplemental Fig. 3B shows successful and efficient MKK4 expression knockdown compared with non-target lentivirus-infected cells (shCtrl). Also, although H. pylori-stimulated phosphorylation of JNK isoforms 2 and 3 (JNK2/3) was unaffected in the shMKK4 cells (supplemental Fig. 3B, upper arrowhead), JNK isoform 1 (JNK1) phosphorylation was significantly reduced in the knockdown cells compared with shCtrl cells (lower arrowhead). This suggests that H. pylori-stimulated MKK4 demonstrates specificity for the JNK1 isoform. Cells were then co-cultured with H. pylori 60190 for scatter assays, and supplemental Fig. 3C shows that shMKK4 cells still scattered in response to H. pylori, suggesting that neither MKK4 nor JNK1 phosphorylation are required for the H. pylori-induced morphological response.

H. pylori Stimulates JNK Phosphorylation and Cell Scattering through an MKK7-independent Mechanism—Besides MKK4, only MKK7 is known to directly regulate JNK activity (48). Therefore, one would predict that if JNK is required for cell scattering and shMKK4 cells show a loss of JNK1 phosphorylation but the cells still scatter, then MKK7 would regulate JNK2/3 activity and be required for H. pylori-dependent cell scattering. Indeed, when JNK2/3 activity was pharmacologically inhibited using an inhibitor specific to isoforms 2 and 3 but not 1, H. pylori-stimulated cell scattering was significantly blocked, although the normal phenotype was not completely restored; this demonstrates a requirement for JNK2/3 activity in H. pylori-stimulated cell scattering (supplemental Fig. 4A). Stable AGS cells expressing shRNAs against MKK7 were then generated (shMKK7) to address the role of MKK7 in H. pylori-stimulated signaling and cell scattering. Surprisingly, supplemental Fig. 4B shows no difference between shCtrl and shMKK7 cells in the phosphorylation of any JNK isoforms in the presence of H. pylori 60190, although shMKK7 cells showed significant loss of MKK7 expression. Additionally, shMKK7 cells scattered to a similar extent to control cells in response to H. pylori 60190 (supplemental Fig. 4C), demonstrating that H. pylori induces JNK phosphorylation and cell scattering independent of MKK7.

JNK Phosphorylation and Cell Scattering Occur Independent of Both MKK4 and MKK7—Although the JNK2/3 inhibitor blocked H. pylori-dependent cell scattering, MKK7 expression knockdown failed to prevent JNK2/3 phosphorylation. To test the hypothesis that loss of one of these kinases is complemented by the presence of the other, AGS cells stably expressing both MKK4 and MKK7 shRNAs (shMKK4/7) cells were generated by lentiviral delivery and used in co-culture experiments. The supplemental Fig. 5A confirms that MKK4 and MKK7 expression is almost completely abrogated in shMKK4/7 cells, and the supplemental Fig. 5B indicates that H. pylori-induced JNK 2/3 phosphorylation was unaffected in these both control and shMKK4/7 cells. Additionally, shCtrl cells or shMKK4/7 cells were co-cultured with H. pylori 60190 for 18 h and fixed and stained for F-actin. The supplemental Fig. 5C shows that H. pylori-stimulated cell scattering was not blocked, demonstrating that H. pylori-stimulated JNK2/3 phosphorylation and cell scattering occur independent of both MKK4 and MKK7.

JNK Mediates H. pylori-dependent Cell Motility through Paxillin—A major function of JNK is to regulate activity of the AP-1 transcription factor, which in turn alters gene expression (48–50). Additionally, recent evidence shows that JNK can influence cell motility by activating downstream effectors that stabilize microtubules and regulate actin reorganization and cell adhesion (35, 51, 52). Cells pretreated with the protein synthesis inhibitor, cycloheximide, or the NF-κB inhibitor, Bay11-7082, were unable to block H. pylori-stimulated cell scattering in separate experiments, demonstrating that JNK mediates H. pylori-induced cell scattering through a gene expression-independent mechanism (supplemental Fig. 6).

Paxillin is a component of focal adhesions, which facilitate attachment of the actin cytoskeleton to the extracellular matrix (51). Paxillin is phosphorylated at multiple serine and tyrosine residues by different upstream activators to regulate focal adhesion turnover and cell migration (53). Phosphorylation at tyrosine residue 118 (paxillin^Tyr-118) occurs through Src and FAK in response to growth factors, and serine 178 (paxillin^Ser-178) is known to be phosphorylated by JNK (35, 51–53). Because of this link between JNK and paxillin, Western blot analyses were performed on AGS co-culture lysates to determine paxillin activation in response to H. pylori 60190. As shown in Fig. 6A, paxillin was phosphorylated at both Tyr-118 and Ser-178 in a CagA-independent but TFSS-dependent manner. AGS cells stably expressing paxillin shRNAs were generated (shPax) to address the role of paxillin in H. pylori-induced cell signaling and motility, and Fig. 6B shows loss of detectable paxillin in shPax cells compared with shGFP control cells. Cell scattering and motility assays show a striking decrease in shPax cell scattering and motility compared with shGFP cells in response to H. pylori co-culture (Fig. 6, C and D, respectively), which demonstrates that paxillin is required for H. pylori-induced cell scattering and motility.

FIGURE 6.

H. pylori requires paxillin to induce cell scattering and motility. A, AGS cells were co-cultured with indicated H. pylori strains for 1 h, and lysates were collected for Western blot analysis. Paxillin phosphorylation at Ser-178 or Tyr-118 was detected using phospho-paxillin (Ser-178)- or phospho-paxillin (Tyr-118)-specific antibodies. B, AGS cells stably expressing shRNAs targeting paxillin (shPax) or GFP (shGFP) were collected for Western blot analysis, and total paxillin was detected using anti-paxillin antibodies. Tubulin was also probed as a load control. C, shGFP (panel i) or shPax (panel ii) cells were co-cultured with H. pylori 60190 for 18 h, and fixed and stained for F-actin. D, graph represents the effects of paxillin expression knockdown on H. pylori-induced cell motility compared with control cells in three separate experiments and includes standard error. Data presented as average area cleared per cell from multiple experiments and includes standard error. ^*, shGFP+H. pylori versus shGFP alone (p < 0.001); ^**, shPax+H. pylori versus shPax alone (p > 0.05; determined by analysis of variance and Tukey's HSD test). UT, untreated cells.

To determine whether JNK regulates paxillin activity, AGS cells were pretreated with the pan-JNK inhibitor SP600125 (targets isoforms 1–3) prior to co-culture with H. pylori 60190 for 1 h. Western blot analysis was then performed on the lysates, and Fig. 7A shows that phosphorylation at paxillin^Ser-178 was significantly decreased in response to the JNK inhibitor. Next, shGFP or shMKK4 cells were co-cultured with H. pylori 60190 for 1 h and collected for Western blot analysis to determine whether paxillin^Ser-178 phosphorylation is induced by JNK1. As shown in Fig. 7B, shMKK4 cells showed no paxillin^Ser-178 phosphorylation compared with strong induction in shGFP cells. This suggests that paxillin^Ser-178 may be a substrate of JNK1, but phosphorylation at Ser-178 is not required for H. pylori-stimulated cell scattering.

FIGURE 7.

H. pylori stimulatesβ₁-mediated paxillin (Tyr-118) phosphorylation independent of MKK4 or MKK7. All Western blot analyses were derived from lysates of cells co-cultured with H. pylori 60190 for 1 h unless indicated otherwise. A, AGS wild-type cells were pretreated with the JNK1/2/3 inhibitor SP600125 (40 μm) or DMSO. B, shGFP or shMKK4 cells were cultured alone or with H. pylori, and paxillin^Ser-178 phosphorylation at was detected using phospho-paxillin^Ser-178-specific antibodies. C, AGS wild-type cells were pretreated with the DMSO, SP600125 (40 μm), AIIB2 antibody (5 μg/ml), an equal volume of serum-free media (SF media), or PP2 (25 μm), or transfected with Ad-GFP-FRNK (m.o.i. of 10) or Ad-GFP (m.o.i. of 10). D, untreated shCtrl, shMKK4, shMKK7, or shMKK4/7 cells were co-culture with H. pylori, and paxillin^Tyr-118 phosphorylation at was detected using phospho-paxillin^Tyr-118-specific antibodies. Tubulin was also probed as a load control. All blots are representative of multiple experiments.

Fig. 7C shows that paxillin^Tyr-118 phosphorylation was also blocked in the presence of SP600125, demonstrating that paxillin^Tyr-118 is regulated by JNK activity. Fig. 7C also shows that paxillin^Tyr-118 phosphorylation is reduced in response to pretreatment with AIIB (β₁ blocking antibody) or PP2 (Src inhibitor) or FRNK expression (endogenous FAK inhibitor) compared with control conditions, demonstrating that H. pylori stimulates paxillin^Tyr-118 through the integrin signaling pathway. Western blot analysis also showed no change in paxillin^Tyr-118 phosphorylation in response to H. pylori in shMKK4, shMKK7, and shMKK4/7 cells compared with shCtrl cells, which shows that paxillin^Tyr-118 is phosphorylated independent of known upstream kinases (Fig. 7D). These data also suggest that this paxillin^Tyr-118 may be important in H. pylori-dependent motility.

DISCUSSION

In recent years, CagA has been shown to be sufficient to induce cell scattering (hummingbird morphology) when transiently expressed in cells, and CagA is thought to be the major stimulus of cell morphological changes (11, 12, 21). These previous data conflict with our findings in that the JNK pathway, which is required for H. pylori-stimulated cell motility, is induced independent of CagA. When comparing experimental systems, we propose that live bacterial co-culture is more physiologically relevant than overexpression of CagA, because proper expression levels and intracellular location are maintained, and any protein modifications occurring inside the bacteria or during translocation that may influence CagA activity are more likely to occur (54, 55). The report by Al-Ghoul et al. (32) demonstrating that Cag PAI gene products independent of CagA were required for cell scattering further suggests that CagA alone is insufficient to stimulate cell scattering.

Our data are consistent with reports of TFSS-dependent, but CagA-independent host responses, such as cyclin D1 expression, NF-κB activation, and the production of cytokines (27, 56, 57). Recent studies also showed that H. pylori induced cell invasion through a combination of CagA-dependent and CagA-independent (but TFSS-dependent) signaling (22, 32).

Mechanisms of TFSS-dependent signaling independent of CagA have only recently been reported. The cytoplasmic pattern recognition receptor, Nod1, was shown to be activated in response to peptidoglycan that was transported through the H. pylori TFSS to stimulate NF-κB activity (27). Evidence also showed that Nod1 can regulate both JNK and p38 activity, although supplemental Fig. 1 shows that AGS cells lacking Nod1 expression still exhibited H. pylori-induced JNK activity and cell scattering, suggesting that the TFSS does not activate JNK signaling through Nod1 (36, 37). More recently, Kwok et al. (38) showed that the TFSS activates integrin signaling through interaction of the bacterial CagL protein with the α₅β₁ integrin heterodimer. Furthermore, this interaction resulted in the activation of Src and FAK, both known to participate in integrin-mediated signaling leading to JNK activation (58). Fig. 3 and Fig. 7C strongly suggest that the H. pylori CagL-β₁ integrin interaction stimulates CagA-independent JNK-mediated cell motility through paxillin. One may argue that the β₁ blocking antibody inhibits focal adhesions, but because the cells still attach to the substratum, we believe that focal adhesions remain intact.

Upon translocation via the TFSS, CagA is phosphorylated at multiple EPIYA residues by Src family kinases and c-Abl (8–10, 59). CagA then causes dephosphorylation of the activation domain of Src, leading to deregulation of multiple cytoskeletal regulatory pathways and cell motility (60, 61). The inactivation of Src by CagA does not conflict with our data showing Src-mediated JNK phosphorylation, because we detect JNK activation within 30 min of stimulation by H. pylori, and CagA-mediated Src inactivation occurs after 3–4 h.

We also show that FAK is not required for H. pylori-induced JNK phosphorylation, but it is required for paxillin^Tyr-118 phosphorylation and cell scattering, which suggests that FAK bypasses JNK to directly target paxillin. This was not surprising, because paxillin^Tyr-118 is a known target of FAK (62). Integrin-mediated stimulation of FAK occurs through an undefined mechanism at tyrosine 397, which causes a conformational change that creates a high affinity Src homology 2 domain for Src. Src then mediates further tyrosine phosphorylation of FAK, leading to downstream signaling (40). Therefore, as shown in Fig. 8, we propose that Src-mediated FAK activity leads to paxillin^Tyr-118 phosphorylation, which is also stimulated by Src-mediated JNK activation.

FIGURE 8.

H. pylori stimulates cancer cell motility through the combined CagA-dependent and -independent signaling. CagA is injected into the cell via the α₅β₁ integrin and associates with known regulators of cell adhesion and motility. CagA-independent activation of the β₁ integrin signaling pathway leads to Src-induced JNK stimulation, which in combination with FAK causes paxillin^Tyr-118 phosphorylation. Although both CagA-dependent and CagA-independent signaling pathways are necessary, neither pathway is sufficient by itself for H. pylori-induced cell motility.

Because the integrin-mediated signaling pathway leading to JNK activation is reported to occur independent of Rac1 and PI3K, our data showing that Cdc42, Rac1, and PI3K are not important for H. pylori-dependent JNK phosphorylation strengthen our model of JNK activation (39).

H. pylori-dependent JNK2/3 is required for AGS cell scattering (supplemental Fig. 4A), but JNK2/3 activation and cell scattering occur independent of MKK4 and MKK7, as shown in supplemental Figs. 3–5. We therefore propose that JNK2/3 phosphorylation and H. pylori-dependent cell scattering occur through activation of a third unidentified MAP kinase kinase that targets JNK. Yamauchi et al. (58) reported activation of JNK independent of either MKK4 or MKK7 using dominant-negative constructs and termed this unknown kinase MKK-X. This hypothesis is conceivable, as JNK isoforms interact with scaffolding proteins that form signaling complexes to promote JNK activation and downstream signaling (48, 63). Because many MAP kinase kinases and MAP kinase kinase kinases, such as MKK1, MKK3, and mixed lineage kinases, are known to be recruited to JNK scaffold proteins, JNK2/3 may be activated by one of these other kinases (48, 63). One of these signaling complexes was reported by Takino et al. (63) who showed that the α₅β₁ ligand, fibronectin, stimulated JNK activity through complex formation of FAK and the JSAP1 scaffold protein. This interaction was enhanced by Src and led to cell migration. These data strengthen the hypothesis of integrin-mediated activation of JNK and the role of scaffold proteins in cell motility.

JNK signaling is emerging as a key mediator in cell migration and invasion. Besides regulating gene expression through the AP-1 transcription factor complex, JNK can activate proteins that regulate microtubule stabilization and focal adhesion turnover (35, 48–50, 64). The supplemental Fig. 6 demonstrates that H. pylori-dependent cell scattering does not require de novo protein synthesis, which agrees with prior reports of JNK mediating cell migration through a gene expression-independent mechanism (34). We show that H. pylori-induced motility requires paxillin expression, and paxillin^Ser-178 phosphorylation is dependent upon MKK4 and JNK activity but is not required for cell motility (Figs. 6 and 7 and supplemental Fig. 3). Paxillin^Tyr-118 phosphorylation is dependent upon JNK activity, as shown using the pan-JNK inhibitor, SP600125 (Fig. 7C). Since the JNK2/3 inhibitor blocked H. pylori-induced cell scattering, the fact that paxillin^Tyr-118 phosphorylation requires JNK activation and occurs independent of MKK4/7 ablation suggests that JNK2/3 mediates cell motility through Y118 phosphorylation.

Fig. 8 summarizes our model of H. pylori-induced cell motility. The process is initiated by the interaction of CagL with the β₁ integrin, which promotes translocation of CagA via the TFSS into tumor cells. β₁ signaling causes Src-mediated CagA phosphorylation (along with c-Abl) to facilitate CagA-dependent signaling that promotes cell motility. Src also stimulates FAK activation, which leads to activation of paxillin. Src also mediates JNK2/3 activation through a mechanism independent of known upstream JNK kinases, and JNK2/3 also stimulates paxillin phosphorylation, which promotes focal adhesion turnover necessary for cell motility.

Although these CagA-independent events are necessary, they are insufficient to induce cell scattering and motility without CagA-dependent signaling. Thus, we propose that the combination of CagA-dependent and CagA-independent (but still TFSS-dependent) events is required to stimulate an epithelial-mesenchymal transition-like response in gastric cancer cells in a gene expression-independent manner.

These data demonstrate that the TFSS plays a greater role in host cell physiology than just to deliver CagA from the bacterium into the host cell cytoplasm, and further studies are needed to determine the greater scope of CagA-independent signaling. These studies will also identify additional host cell players that participate in cell motility that may contribute further insight into the mechanisms of gastric cancer progression.

Note Added in Proof—Since submission, we have generated data demonstrating that the β₁ blocking antibody, AIIB2, does not prevent focal adhesion assembly because we detected no change in the frequency of focal contacts between control and AIIB2-treated cells, as visualized by immunofluorescence microscopy.