Microaerophilic conditions permit to mimic in vitro events occurring during in vivo Helicobacter pylori infection and to identify Rho/Ras-associated proteins in cellular signaling.

Molecular dissection of the mechanisms underlying Helicobacter pylori infection suffers from the lack of in vitro systems mimicking in vivo observations. A system was developed whereby human epithelial cells (Caco-2) grown as polarized monolayers and bacteria can communicate with each other under culture conditions optimal for each partner. Caco-2 cells grown on filter supports were inserted in a vertical position into diffusion chambers equilibrated with air and 5% CO(2) at their basolateral surface (aerophilic conditions) and 5% CO(2), 5% O(2), 90% N(2) (microaerophilic conditions) in the apical compartment. Remarkably, the epithelial polarized layer was stable under these asymmetric culture conditions for at least 24 h, and the presence of Caco-2 cells was necessary to maintain H. pylori growth. In contrast to previous studies conducted with non-polarized Caco-2 cells and other cell lines kept under aerophilic conditions, we found H. pylori-dependent stimulation of cytokine secretion (MCP-1 (monocyte chemoattractant protein-1), GRO-alpha (growth-regulated oncogene-alpha), RANTES (regulated on activation normal T cell expressed and secreted)). This correlated with nuclear translocation of NF-kappaB p50 and p65 subunits. Tyrosine phosphorylation of nine cellular proteins was induced or enhanced; we identified p120(RasGAP), p190(RhoGAP), p62dok (downstream of tyrosine kinases), and cortactin as H. pylori-inducible targets. Moreover, reduction of H. pylori urease expression was observed in adherent bacteria as compared with bacteria in suspension. In addition to mimicking several observations seen in the inflamed gastric mucosa, the novel in vitro system was allowed to underscore complex cellular events not seen in classical in vitro analyses of microaerophilic bacteria-epithelial cell cross-talk.

Upon infection of the host, enteric pathogens first encounter the host's mucosal surfaces lined by epithelial cells (1). The function of these cells in host defense goes far beyond the mere mechanical barrier separating the external environment from the internal milieu of the host. Epithelial cells have to be considered as an integral component of the mucosal immune system, as they are capable of providing the underlying mucosa with the message that an infection occurs (2). This is accomplished by the release of molecular messengers including cytokines and chemokines by the epithelial cells that orchestrate initial phases of the immune response. Given the molecular complexity of these events, their dissection requires the availability of in vitro systems, which mimics the in vivo situation. To fulfill this requirement, viability of all cell partners has to be guaranteed, and biologic read-outs should resemble those established in vivo.
Although in vivo studies have contributed to important breakthroughs regarding the morphological and physiological events resulting from the infection by Helicobacter pylori, the major causative agent of chronic gastritis, peptic ulcer, gastric adenocarcinoma, and mucosa-associated lymphoreticular tissue lymphoma (3,4), in vitro approaches are important to refine the study at the molecular and cellular levels. So far, the design of such a robust in vitro system has not been reported for the study of H. pylori-epithelial cell interactions. Two major difficulties have to be overcome: 1) gastric cell lines cannot be grown as polarized epithelia mimicking the gastric mucosa (5)(6)(7)(8), 2) or if so (9) do not stand the microaerophilic conditions needed for the particular metabolism of H. pylori. The low oxygen conditions prevailing in the gastric environment are important in the activation of H. pylori virulence genes (10). Among them, urease is required for initial gastric colonization (11), allowing the bacterium to maintain its periplasmic pH near neutrality (12,13). The expression of H. pylori CagA protein, strongly associated with duodenal ulceration, is influenced through the culture medium pH in vitro (14). These examples indicate that neglecting crucial bacterial growth parameters could preclude the refined study of H. pylori-epithelial cell interactions. Moreover, all studies have been performed under aerophilic conditions whereas a restricted oxygen atmosphere is required for proper H. pylori metabolic and growth functions (15).
For these important reasons, we designed a novel in vitro system maintaining optimal culture conditions for H. pylori added to polarized epithelial cells serving as an interface between the apical and basolateral compartments of diffusion chambers. Bacteria in BHI, 1 pH 5.5, 5 mM urea, 5% O 2 , 5% CO 2 , 90% N 2 kept dividing up only in the presence of the polarized Caco-2 monolayer. Bacterium-cell contact resulted in pedestal formation and brush border disruption as observed in vivo. Tyrosine phosphorylation of proteins including p120 RasGAP , p190 RhoGAP , p62dok, and cortactin was observed for the first time. In addition to NF-B nuclear translocation and IL-8 secretion, MCP-1, GRO-␣, and RANTES induction detected in biopsies could be reproduced under microaerophilic conditions only. In comparison with H. pylori floating in suspension, bacteria associated with Caco-2 cells exhibited decreased urease A and B subunits expression. The in vitro model developed in this study will represent a valuable mimic of host-pathogen interaction to examine complex molecular aspects of H. pylori infection, such as the bacterial virulence program, the mechanism of adherence and the contribution of epithelial cell to bacterial colonization, and the cross-talk with inflammatory cells.
Helicobacter pylori Culture Conditions-The H. pylori CagA ϩ strain ATCC 43504 was grown on agar plates made of 36 g/liter GC-agar base (Oxoid AG) containing 12.5% heat-inactivated horse serum (Invitrogen) and 1% IsoVitale X (Baltimore Biological Laboratories) in a microaerophilic atmosphere (90% N 2 , 5% CO 2 , 5% O 2 ) for 2 days before harvest into either plain BHI (BioMérieux) or into BHI supplemented with 0.25% yeast extract (BHI-C; Difco Laboratories) and 10% fetal calf serum (Seromed). The total number of bacteria was determined by measuring the A 600 of the bacterial suspension, 1.0 optical density unit corresponding to 10 8 bacteria.
The Asymmetrical Culture System-The diffusion chamber device developed by Grass and Sweetana (16) was modified to maintain both H. pylori and the polarized Caco-2 cells under optimal growth conditions. Caco-2 cells were seeded onto Snapwell filter (Costar) and allowed to form a tight, polarized monolayer (17). The filters carrying the cell monolayers were inserted between the two chambers, thus resulting in the physical separation into apical and basolateral compartments containing different culture media (Fig. 1). Up to six devices could be mounted serially on a support. Caco-2 cells were fed through the basolateral compartments containing C-DMEM perfused with air and 5% CO 2 . The apical compartments containing BHI-C, BHI, or BHI-U (BHI medium containing 5 mM urea, pH 5.5) were perfused continuously under microaerophilic gas conditions (90% N 2 , 5% CO 2 , 5% O 2 ) necessary for the optimal proliferation of H. pylori. The gas mixture flow rate was set at 50 ml/min, and the inlet pressure was fixed to 1.1 bar to give a steady delivery of the gas mixtures to each compartment. Constant administration of the gas mixtures into the airtight device for up to 24 h resulted in dissolved pO 2 and pCO 2 ranging around (mm of Hg measured in a Chiron Diagnostic blood gas analyzer) 110 and 34 in the apical BHI and 140 and 35 in basolateral C-DMEM, respectively. Oxygenbinding proteins in culture medium accounts for the 110 mm of Hg observed under microaerophilic conditions (18); we confirmed this assumption by measuring similar dissolved pO 2 in BHI kept in sealed flasks (Oxoid AG) equilibrated with 90% N 2 , 5% CO 2 , 5% O 2 .
The integrity of the epithelial Caco-2 cells as a function of the experimental conditions was reflected by the measurement of the TER, which was expressed as the percentage of the TER obtained at the beginning of the incubation period. In experiments dealing with exposure of Caco-2 cells to H. pylori, the bacterial inoculum was added to the apical compartments at a concentration ranging between 5 ϫ 10 5 and 9 ϫ 10 6 bacteria per ml.
Quantification of Adherent and Suspension H. pylori-Snapwell support carrying Caco-2 polarized monolayers were removed, transferred into 6-well dishes, and washed five times with 0.9% NaCl. The Caco-2 cells were then detached by a 5-min incubation at 37°C in the presence of trypsin/EDTA (Invitrogen) added to the apical (0.5 ml) and the basolateral (1.5 ml) surfaces. The adherent bacteria were dispersed by vigorous pipetting, serial dilutions (10 Ϫ2 to 10 Ϫ4 ) were applied onto triplicate agar plates left for 4 -5 days at 37°C under microaerophilic conditions (90% N 2 , 5% CO 2 , 5% O 2 ), and c.f.u. were determined. Suspension H. pylori recovered directly from the apical medium were assessed using serial dilutions ranging from 10 Ϫ3 to 10 Ϫ6 . For the experiment reported in Fig. 7, the equivalent of 5 ϫ 10 5 H. pylori (counted as above) recovered from either the supernatant or attached to Caco-2 cells were lysed in 15% trichloroacetic acid (final concentration). The whole lysate was resuspended in 50 mM Tris-HCl, pH 6.8, 2% SDS, 100 mM dithiothreitol, and resolved by SDS-PAGE (19).
Electron Microscopy-Polarized Caco-2 cells were washed five times with 0.9% NaCl, fixed overnight with 2.5% glutaraldehyde in 0.1 M sodium cacodylate, pH 7.4, post-fixed for 1 h with 2% osmium tetroxide in 0.1 M sodium cacodylate, dehydrated, and embedded in Epon 812 (Polysciences Inc.). Sections were stained with uranyl acetate and lead citrate and then examined using a Philips CM 10 transmission electron microscope.
Chemokine Enzyme-linked Immunosorbent Assay-Chemokines released by polarized Caco-2 cells in the basolateral cell culture medium were measured using sandwich enzyme-linked immunosorbent assay kits for human IL-8, MCP-1, GRO-␣, and RANTES (Quantikine, R & D Systems) and expressed as picograms of specific protein per milliliter of culture medium. Data are duplicates of two-three independent experiments.
EMSA-Caco-2 cells from three Snapwell filters were recovered by trypsinization and incubated for 15 min in 1 ml of hypotonic buffer (10 mM HEPES, pH 7.6, 15 mM KCl, 2 mM MgCl 2 , 0.1 mM EDTA, and protease inhibitors (Complete TM ; Roche Molecular Biochemicals)). Cells were lysed in 500 l of hypotonic buffer containing 0.2% Nonidet P-40 (Pierce). The nuclei were pelleted by centrifugation, and the supernatant corresponding to the cytoplasmic extracts was frozen immediately. The nuclei were resuspended in 270 l of a high salt buffer (25 mM HEPES, pH 7.6, 50 mM KCl, 0.1 mM EDTA, 10% glycerol, and protease inhibitors). DNA-binding proteins were extracted by the addition of 30 l of 3 M (NH 4 ) 2 SO 4 and gentle shaking for 30 min at 4°C. The extract was spun in a Beckman TLA-100.2 rotor for 15 min at 200,000 ϫ g. The supernatant was concentrated by 22% (NH 4 ) 2 SO 4 precipitation and spun in a Beckman TLA-100.2 rotor for 10 min at 100,000 ϫ g. The protein pellet was resuspended in 20 l of high salt buffer and stored at Ϫ70°C. Protein concentration was determined by the bicinchoninic acid assay (Pierce). Binding reactions were performed at room temperature for 30 min using 5 g of nuclear proteins and 0.5 ng (25,000 counts per min) of [␣-32 P]dCTP-radiolabeled oligonucleotide in 15 l of binding buffer containing (in mM) 10 Tris-HCl, pH 7.5, 50 NaCl, 50 KCl, 1 MgCl 2 , 1 EDTA, 5 dithiothreitol, 5% glycerol, and 0.75 g poly(dI-dC)poly(dI-dC) (Roche Molecular Biochemicals). The sequences of the consensus NF-B oligonucleotide used for EMSA were: coding strand, 5Ј-AGTTGAGGGGACTTTCCCAG-3Ј; non-coding strand, 5Ј-GCCTGG-GAAAGTCCCCTCAA-3Ј. DNA⅐protein complexes were resolved by electrophoresis on 5% non-denaturing polyacrylamide gels in 0.5ϫ TBE buffer (20) and visualized via autoradiography. In competition experiments, increasing molar excess (0.045 M Tris-borate, 1 mM EDTA) of the unlabeled consensus NF-B oligonucleotide was used and compared with mutated NF-B oligonucleotide consisting of the following: coding strand, 5Ј-AGTTGAGGCGACTTTCCCAG-3Ј; non-coding strand, 5Ј-GCCTGGGAAAGTCGCCTCAA-3Ј. To confirm that members of the NF-B family extracted from Caco-2 cells were contributing to the retardation of the DNA probe, rabbit Ab (1 g/reaction) against the NF-B subunits p50 or p65 (Santa Cruz Biotechnology) was added to EMSA mixtures during the binding reaction period.

Culture Conditions Optimized for H. pylori Growth Do Not
Alter the Polarized Caco-2 Monolayer-Increased oxygen tension (21), alkaline or neutral pH (22), and prolonged incubation (23) prompt formation of coccoid forms of H. pylori no longer able to adhere and signal like spiral forms (24). We thus sought to establish conditions optimal for H. pylori growth and not deleterious to epithelial Caco-2 cells grown as polarized monolayers. A microaerophilic gas mixture (5% O 2 , 5% CO 2 , 90% N 2 ) was applied to the apical compartment, whereas the basolateral surface of cells was exposed to air and 5% CO 2 . TER (a measure of the integrity of the Caco-2 cell polarized monolayer) resulting from polarized Caco-2 cells was stable either when BHI replaced C-DMEM or when the microaerophilic conditions were applied to the apical compartment ( Fig. 2A). In microaerophilic apical BHI, a multiplicity of infection of 10, respectively, 100 did not affect TER for up to 48 h (Fig. 2B). TER was preserved down to pH 5.0 (a pH value resembling that reported in the stomach antrum) when aerophilic or microaerophilic BHI was present in the apical compartment. 3 We found no drop in TER value when 5 mM urea, a substrate favoring H. pylori viability at low pH, was added (Fig. 2C). This defined that culture conditions ensuring optimal growth and adhesion (see below) of H. pylori, namely plain BHI, pH 5.5, 5 mM urea, and a microaerophilic gas mixture, are well tolerated by polarized Caco-2 monolayers.
The Presence of Caco-2 Cells Ensures Not Only Survival but Growth of H. pylori in the Asymmetrical Culture System-The definition of optimal culture conditions for H. pylori in the apical compartment is summarized in Table I times revealed that growth resumed between 8 and 12 h. Remarkably, the effect was dependent on the presence of polarized Caco-2 cells at the interface between the apical and basolateral compartment (compare columns 3 and 4). The data demonstrate that maintenance of live H. pylori under nonphysiological conditions depends on parameters, of which ignorance is likely to seriously affect its growth, and thus adhesion and resulting signaling properties (see below).
Because the microaerophilic gas conditions led to a much better bacterial growth capacity in the apical compartment, we then examined, using c.f.u. counting, whether this consistently improved adhesion of H. pylori to Caco-2 cells. At 4 h, Ͼ20 times more adhesive bacteria were recovered when microaerophilic BHI and BHI-C media were used as compared with aerophilic conditions (Fig. 3A). In BHI-U, this factor raised up to 100-fold in favor of the microaerophilic environment (Fig.  3A). Under microaerophilic conditions at 24 h, the capacity of H. pylori to adhere to Caco-2 cells was similar in all three media examined, whereas c.f.u. counts remained very low under aerophilic conditions (Fig. 3A). Adhesion is therefore linked closely with the preserved ability of H. pylori to multiply (Table I, column 3). H. pylori laid down over Caco-2 or AGS cells grown as non-polarized monolayers on plastic in C-DMEM yielded c.f.u. counts as low as in aerophilic BHI media. 4 We observed the average binding of seven bacteria per Caco-2 cell (5 ϫ 10 6 H. pylori per filter carrying 7 ϫ 10 5 Caco-2), and found that 10-fold differences in the initial bacterial load did not modify this ratio, thus arguing for specific association (Fig. 3B). Together, this indicates that optimized culture conditions for H. pylori and Caco-2 cells avoid possible nonspecific effects including epithelial cell apoptosis resulting from the addition of dead bacteria, bacterial debris (25,26), or excessive amounts of bacteria.
The Interaction of H. pylori with Polarized Caco-2 Resembles That Seen in Vivo-Caco-2 cells simply grown on plastic miss both brush border and tight junctions (27). In contrast, Caco-2 cells seeded on Snapwell filters exhibit such features, with expression of brush border sucrose isomaltase and basolateral polymeric Ig receptor (28). Given that Caco-2 cells maintained high TER values, and H. pylori showed much improved viability (see Figs. 2 and 3), we thus examined by electron microscopy what Caco-2 and H. pylori looked like in the novel system described herein. Similar to gastric biopsies, the association between Caco-2 epithelial cells and H. pylori triggers morphological changes including brush border effacement and pedestal formation (Fig. 4, A and B). The pictures obtained after 24 h of incubation showed the expected rod shape of H. pylori, as well as the presence of non-aggregated bacteria sitting at the Caco-2 surface (Fig. 4A). Under these conditions, no coccoid H. pylori could be observed at 24 h, yet this has been observed upon exposure to non-polarized epithelial cells maintained under 4 S. Cottet, unpublished observations.   (8). Actin rearrangement within the host cell occurred directly beneath the site of attachment of H. pylori, forming a very fine condensed structure concentric to the bacterium (Fig. 4B). No such alterations were seen in the absence of H. pylori or under aerophilic conditions not ensuring bacterial viability (Table I). In contrast to previous data showing no pedestal formation and actin filament rearrangement using non-gastric cells (29), the system as established satisfies this requirement, thus suggesting optimized cross-talk between H. pylori and Caco-2 cells.
H. pylori-induced Activation of NF-B and Phosphorylation of p120 RasGAP , p190 RhoGAP , p62dok, and Cortactin in Polarized Caco-2 Cells-Epithelial cells grown on plastic and exposed to H. pylori for 1-6 h have been shown to trigger NF-B activation (30,31). Using polarized Caco-2 cells, and thanks to the potential of the culture system to preserve H. pylori viability (Table I), this enabled us to extend the analysis to 24 h. We first compared the translocation properties of NF-B p50 and p65 subunits induced by H. pylori in aerophilic and microaerophilic BHI-U (Fig. 5A). The latter gave rise to a much more pronounced nuclear accumulation of both subunits (p50, 5.5-fold; p65, 3-fold). Nuclear translocation was strictly dependent on the incubation of Caco-2 cells with H. pylori and was sustained for up to 24 h. The level of p65 kept increasing between 2 and 24 h, whereas that of p50 reached steady-state after 2 h. Consistent with this, the higher amount of translocated p50 and p65 resulted in the formation of more DNA⅐NF-B complexes in EMSA (Fig. 5B, compare lanes 1 and 2 with lanes 5 and 6). Antisera to either subunit abolished (lanes 3 and 4) or reduced (lanes 7 and 8) the formation of the DNA⅐NF-B complex, indicating that the p50-p65 heterodimer represents the DNA binding form of NF-B activated after exposure to H. pylori. Competition experiments confirmed the specificity of binding to the consensus NF-B DNA probe (lanes 9 -12). Under identical EMSA conditions, barely detectable DNA⅐NF-B complexes were obtained with nuclear extracts from AGS cells grown on plastic 5 ; in agreement with Table I, dead or coccoid bacteria can trigger NF-B activation only to low levels.
Although H. pylori triggers protein phosphorylation in AGS cells (8), only scarce information exists as to the nature of these activities. We thus examined the pattern of phosphorylation of Caco-2 cytoplasmic proteins after incubation with H. pylori (Fig. 5C). In microaerophilic BHI-U, we found that ten protein bands were tyrosine-phosphorylated de novo, with apparent molecular weights (M r ) ranging from 47,000 to 184,000. Under aerophilic conditions, only doublets at 136 kDa yielded a signal. 6 The pattern was the same at 4 and 24 h, indicating that the interaction between the epithelial cells and the bacterium led to prolonged stimulation of pathways supposed to turn off rapidly. To shed light on the identity of protein triggered by H. pylori, we performed immunoprecipitation on Caco-2 lysates using anti-phosphotyrosine mAb, followed by Western blot analysis. Specific signals were obtained with antibodies against p120 RasGAP , p190 RhoGAP , p62dok, and cortactin. No equivalent signals were obtained in the absence of H. pylori or in aerophilic BHI-U (lanes 6 and 7). Likewise, no detection occurred after exposure of immunoblots with antibodies against ezrin and p130 cas . 7 p120 RasGAP , p190 RhoGAP , and p62dok are substrates for cellular tyrosine kinases and modulate Ras activity, a GTP-binding protein conveying signals from the membrane to the nucleus through the serine-threonine kinase cascade (32). Cortactin is involved in actin rearrangement and cell adhesion (33). The data indicate that microaero-philic conditions allow unraveling at the molecular level of sustained cellular signals possibly involved in cell transformation and shaping.
Chemokines Mapped in the Gastric Mucosa of Infected Patients Are Secreted by Caco-2 Cells upon H. pylori Binding-Activation of NF-B and protein phosphorylation are events required to regulate positively the transcription of genes coding for cytokines and chemokines (34). Previous reports using human stomach biopsies have shown that H. pylori infection resulted in the production in the stomach antrum of IL-8, MCP-1, GRO-␣, and RANTES (35). However, with the exception of IL-8, no induction of other chemokines has been reported using classical in vitro systems. Because RT-PCR analysis of cellular transcripts does not reflect strictly the protein production in the supernatant (36), we examined expression at the protein level using sandwich enzyme-linked immunosor-5 B. Corthésy, unpublished observations. 6 S. Cottet, unpublished observations. 7 S. Cottet and B. Corthésy, unpublished observations. bent assay. Major differences in chemokine secretion were observed as a function of gaseous and medium conditions (Fig. 6). Rise in IL-8 level occurred under aerophilic conditions in BHI-U and C-DMEM, with a lag time of 8 h in BHI-U; this resembles the situation seen with non-polarized cells incubated with H. pylori (37). Under microaerophilic conditions, detection of IL-8 was significant at 24 h in BHI-U (p Ͻ 0.002), whereas incubation in C-DMEM or without bacteria led to reduced chemokine release by Caco-2 cells (Fig. 6A). No specific changes because of exposure to H. pylori was observed for MCP-1 under aerophilic conditions; similar to IL-8, only at 24 h was the level of MCP-1 enhanced 3-fold (p Ͻ 0.008) in microaerophilic BHI-U as compared with controls (Fig. 6B). The RANTES profile exhibited the same kinetics and aspects, with a 2.5-fold up-regulation (p Ͻ 0.001) occurring at 24 h in microaerophilic BHI-U (Fig. 6C). Finally, to underscore weak activation of GRO-␣ (p Ͻ 0.02), microaerophilic BHI-U was required (Fig. 6D). H. pylori killed with gentamicin or 4% formalin did not induce production of chemokines above levels seen in the absence of bacteria under any gas conditions. 8 Extrapolation of these data to Table  I suggests that H. pylori in growing phase provide optimal cross-talk to induce expression and secretion of proinflammatory mediators by epithelial cells.
H. pylori Urease Expression Is Down-regulated upon H. pylori Attachment to Caco-2 Cells-Given the role of Caco-2 cells in favoring H. pylori growth, we then examined what happened to the bacterium during the cross-talk with the epithelial cells. We focused our analysis on the expression of urease, a virulence factor required at the time the bacterium establishes in the stomach (38). To avoid the possible loss of surface-bound urease, we compared whole lysates corresponding to 5 ϫ 10 5 H. pylori. We observed that the bacteria in the medium produce three to four times more urease subunits A and B than bacteria that had adhered to Caco-2 monolayers (Fig. 7). This held true at 4 and 24 h, suggesting that enzyme expression is somehow regulated by the less acidic environment in the vicinity of the epithelial cell, in agreement with the observations of Akada et al. (39). We could not detect such differences under aerophilic conditions or after incubation of H. pylori with AGS cells grown on plastic. DISCUSSION The lack of an in vitro model has considerably impaired the study of H. pylori-host cell interaction at the cellular and molecular levels. Numerous data that have been obtained to date suffer from the limitation that polarized monolayers were not used. Furthermore, incubation of H. pylori and epithelial cells was performed under aerophilic conditions, which are not appropriate to ensure sustained bacterial growth for more than a few hours. Moreover, no data were obtained on the role of epithelial cells in H. pylori viability/growth or on modulation of H. pylori virulence factors. This can undoubtedly interfere with the underscoring of complex signal transduction pathways that require optimal bacteria-host cell interaction.
We have therefore designed an in vitro system based on the use of a H. pylori strain encoding the cag pathogenicity island and polarized intestinal Caco-2 cells, which approached the optimal culture conditions for both the bacteria and epithelial cells. The choice of the bacterial strain was based on the prevalent role of the cag pathogenicity island in gastric diseases and induction of initial events necessary for interaction with epithelial cells (40) and subsequent gene activation (41). Although not a strict equivalent to gastric cell lines, Hep-2, HT-29, T-84, Madin-Darby canine kidney cells, or Caco-2 cells have been used consistently to study IL-8 release, H. pylori-host interaction, permeability increase, and polymorpho-nuclear leukocyte migration (42)(43)(44)(45)(46). Moreover, the finding that H. pylori can associate with the duodenal and colonic mucosa (47)(48)(49) makes the approach of using an intestinal cell line valid. Several new molecular data could be obtained that remained elusive to date in other systems used previously to study H. pylori-host cell interaction. Experimental read-outs 8  identical or close to physiological observations make it relevant to the refined dissection of the complex consequences of H. pylori infection.
The diffusion chamber we used offers the flexibility to select for different culture media, pH settings, and gas mixtures in either the apical or the basolateral compartment. The integrity of the Caco-2 monolayer was maintained using BHI at pH values between 5 and 7.3, not affected by the presence of up to 10 8 bacteria for 48 h, not sensitive to the addition of 5 mM urea for 24 h, and preserved using microaerophilic gas conditions in the apical medium. Together, the data indicate that the unusual conditions Caco-2 monolayers can stand at the apical membrane permitted exposure to H. pylori kept under ideal culture conditions without deleterious morphological consequences that might affect the interaction (50) and resulting signal transduction pathways.
Following the binding of H. pylori to polarized Caco-2 cells, we observed inducible tyrosine phosphorylation of ten cellular proteins, with apparent M r ranging from 47,000 to 184,000. In the absence of H. pylori genes homologous to eukaryotic or bacterial tyrosine kinases (51,52), phosphorylation has to be because of Caco-2 cell activities (53). The 151-kDa band might correspond to the 145-kDa band observed in gastric AGS cells (54). This protein was identified as H. pylori CagA antigen translocated into epithelial cells upon H. pylori attachment (55)(56)(57). In AGS cells again, the moderately increasing tyrosine phosphorylation of a protein with a M r of 105,000 was reported (8), which might correspond to the 104 -107-kDa doublet observed similarly in Caco-2 cell lysate.
Reactivity of tyrosine-phosphorylated proteins with specific antiserum and mAb to Ras/Rho-associated proteins indicates that a novel pathway of cellular signaling by H. pylori has been FIG. 6. Release of chemokines by the epithelial polarized Caco-2 cells exposed to H. pylori. 3 ϫ 10 6 bacteria/ml in either BHI-U (black bars) or C-DMEM (gray bars) were added to the apical compartments, and these latter were incubated under microaerophilic (Micro) and aerophilic (Aero) conditions. Controls included measuring chemokine release at 24 h in BHI-U in the absence of H. pylori (white bars). Results are means Ϯ S.D. of two to three experiments (n ϭ 3 filters/group). H.p., H. pylori.
identified. Phosphorylated p62dok binds to p120 RasGAP and down-regulates its Ras GTPase activity. This observation contributes to link proteins of the Ras superfamily involved in the control of normal and neoplastic proliferation and the role played by H. pylori in gastric adenocarcinoma. Further, p120 RasGAP associates with p190 RhoGAP to regulate actin dynamics and cytoskeleton rearrangement (58). This suggests the notion that p120 RasGAP connects the Rho and Ras pathways (32) through mechanisms involving H. pylori-regulated tyrosine-phosphorylated proteins. Detection of H. pylori-triggered phosphorylation of cortactin provides a clue for cytoskeletal reorganization seen upon bacterial binding. The perversion of cellular proteins appears as a paradigm of host-pathogen interactions. For example, enteropathogenic Escherichia coli induces tyrosine phosphorylation of three eukaryotic proteins, all apparently cytoskeletal-associated (59). Along the same line, Listeria monocytogenes induces the tyrosine phosphorylation of two isoforms (42 and 44 kDa) of the mitogen-activated protein kinase (60) found downstream of the ras-raf-mitogen-activated protein kinase/extracellular signal-regulated kinase kinase pathway.
Intracellular signaling is required for short term activation of chemokine transcription by epithelial cells. We have shown in our system that inflammatory chemokines are produced in response to H. pylori. How does it compare with in vitro and in vivo gastric profiles? In KATO-III and MKN 45 gastric cells, IL-8 transcription induced by H. pylori was dependent on protein-tyrosine kinases (61)(62)(63). Effect of H. pylori on NF-B activation in gastric (KATO III, MKN 45) and colonic (HT-29) cell lines is well documented (50). While confirming these observations, our system brought up information on activation of chemokines such as MCP-1, GRO-␣, and RANTES known to be involved in gastric inflammation in vivo (35). Although produced at relatively low levels, the chemokines were expressed with delayed kinetics by Caco-2 cells. Differences in the kinetics of production, coupled with quantitative differences in their production, suggest that epithelial cells may play a regulatory role in mucosal inflammation by influencing temporal and spatial recruitment of leukocytes within the mucosa. Features like late secretion of chemokines and concomitant sustained NF-B activation most likely reflect that only H. pylori maintained under microaerophilic conditions "communicates" adequately with Caco-2 cells to which it is attached. It remains to be appreciated whether the sustained cellular signals observed in this study with a 1:5 ratio of Caco-2 cells to H. pylori might somehow reflect the chronicity of the infection as occurring in the gastric mucosa.
Urease activity is essential for initiating colonization of the stomach of animal models by Helicobacter sp. (64). Our data support the notion that urease per se does not function as an adhesin (4), because adherent H. pylori express three to four times less urease subunits than their suspension counterpart. It might well be that the urease signal contributed by adherent H. pylori corresponds to the cytoplasmic form, which appears essential to recurrence of infection after treatment with urease inhibitors (65). Thus, the system described herein could serve to screen for such novel inhibitors (66) at reduced costs. In addition, the advantage of using polarized cells responding in vitro like gastric cells should permit the future study of transmigration of polymorpho-nuclear leukocytes as a function of the H. pylori strain and possibly assay modulators of inflammation including IgA under well controlled experimental conditions.