Stimulation of MMP-7 (matrilysin) by Helicobacter pylori in human gastric epithelial cells: role in epithelial cell migration

Epithelial cells in the gastrointestinal, urogenital and respiratory tracts are frequently exposed to bacteria. The maintenance of normal epithelial integrity and function therefore depends on the execution of appropriate adaptive responses, which may include interactions with immune cells (e.g. by release of cytokines, chemokines or antigen presentation) and control of epithelial cell proliferation, migration or apoptosis. Recent studies suggest that matrix metalloproteinase 7 (MMP-7), or matrilysin, plays a role in mediating responses to bacteria in several different systems, although the relevant cellular mechanisms are uncertain(Lopez-Boado et al., 2000; Lopez-Boado et al., 2001). Induction of MMP-7 also occurs in tumours in many organs including colon,stomach and pancreas (Crawford et al.,2002; McDonnell et al.,1991). In the stomach, it is well established that infection with Helicobacter pylori predisposes to gastric cancer in some subjects(Uemura et al., 2001); this appears to be one of the clearest examples of a growing number of instances in which bacteria facilitate tumourigenesis(Lax and Thomas, 2002; Peek and Blaser, 2002). The responses of epithelial cells to H. pylori infection that might influence the subsequent progression to cancer are still largely unknown, and the possible role of MMP-7 in this system is uncertain.

The MMPs comprise a family of over 25 members implicated in the proteolysis of extracellular matrix, growth factors, cytokines, adhesion molecules and protease inhibitors such as the serpins, resulting in both gain and loss of function (Nagase and Woessner,1999; Seiki,2002). It is now widely appreciated that induction of distinct profiles of MMPs, or of their inhibitors, the tissue inhibitors of metalloproteinases (TIMPs), is a feature of the tissue remodelling that occurs in many normal and disease processes (e.g. in development, wound healing,cancer and a wide variety of inflammatory conditions)(Murphy and Gavrilovic, 1999). In the gastrointestinal tract, epithelial cells are often not the source of the increased MMP production that occurs in response to disease. Instead, it is the sub-epithelial (i.e. stromal or mesenchymal) cells that produce the MMPs that play a role in mucosal responses, at least in part by influencing the environment at the epithelial basolateral membrane. However, an exception is MMP-7, which is characteristically produced by epithelial cells and plays a role in a variety of epithelial responses. Thus, for example, in murine small intestinal Paneth cells, MMP-7 may act as a processing enzyme to convert prodefensins to their active form, thereby playing a role in the innate defence system of the gut (Wilson et al.,1999).

The epithelial cells of the gastrointestinal tract are protected from ingested bacteria in part by the barrier presented by gastric acid secretion that generates an initial environment unfavourable to the survival of microoganisms. However, unusually, the gastric pathogen H. pylori(Tomb et al., 1997) colonizes the microenvironment generated between the surface of gastric epithelial cells and the overlying mucus glycoprotein gel. In many subjects, infection with H. pylori is clinically silent; but in some patients it is associated with duodenal ulcer, and in others it is associated with a progression through chronic atrophic gastritis to gastric adenocarcinoma(Fox and Wang, 2001; Uemura et al., 2001). The cellular and molecular mechanisms that determine the precise constellation of epithelial cell responses to infection are far from clear. However, this is a relatively attractive system for studies of the way that bacterial infection might control MMP-7 expression because: (1) it is relatively straightforward to obtain primary gastric epithelial cells from H. pylori positive and control subjects; (2) there is a clear link between H. pylori and gastric cancer; and (3) MMP-7 expression is known to be increased in gastric cancer (Honda et al., 1996). We report here increased expression of MMP-7 in response to H. pylori, the consequences for epithelial cell migration, and the relevant signalling pathways.

AGS cells were obtained from the American Type Culture Collection (ATCC,Manassas, VA), routinely cultured in HAMS/F12 medium supplemented with 10%fetal bovine serum (FBS) and 1% penicillin/streptomycin and subcultured with 1% trypsin/versene (Life Technologies, Paisley, UK) as described(Wroblewski et al., 2002). Expression vectors for constitutively active (CA)-RhoA (L63RhoA), Rac (L61Rac)and cdc42 (L61cdc42), and for dominant-negative (DN)-RhoA (N19RhoA) and DN-Rac(N17Rac) were gifts from Alan Hall (University College, London, UK). A DN-c-jun plasmid was provided by Tim Wang (UMAS, Worcester, MA) and the c-fos overexpressing vector was generated as described previously(Deavall et al., 2000). Lynn Matrisian (Vanderbilt University, Nashville, TN) generously donated a vector consisting of 2.3 kb of human MMP-7 promoter coupled to firefly luciferase(i.e. MMP-7-luc). BAY11-0782, PD98059, epidermal growth factor (EGF), human recombinant MMP-7, MMP-7 antisense (AS) oligonucleotides (both unlabelled and fluorescein labelled), and control oligonucleotide were obtained from Calbiochem (Nottingham, UK). Interleukin 8 (IL-8), IL-6 and tumour necrosis factor α (TNF-α) were purchased from R&D Systems (Minneapolis,MN). Clostridum botulinum C3 toxin was obtained from Sigma (Poole,Dorset, UK).

We routinely used H. pylori strain 60190 (ATCC). In a few experiments, we also used two gastric isolates, strain E5(cag⁺, vacA^-) and E6(cag⁺, vacA⁺). Bacteria were grown in a microaerophilic atmosphere at 37°C on fresh chocolatized Columbia blood agar for 3–9 days (Oxoid, Basingstoke, UK). Unless otherwise stated, H. pylori was added to human gastric glands or AGS cells at a multiplicity of infection (MOI) of 100. In some experiments,bacteria were added to 0.2 μm Anapore filter inserts (Life Technologies)that were suspended above AGS cells. Two such systems were studied: either bacteria was added directly to the filters, or to AGS cells cultured on filters that were then inserted over AGS cells transfected with MMP-7-luc (see below). In other experiments, H. pylori were sonicated for 20 minutes and added directly to AGS cells.

Endoscopic pinch biopsies of the gastric corpus and antrum were obtained from patients attending routine gastroscopy for investigation of dyspepsia. Most patients had no macroscopic gastroduodenal pathology; none of them had neoplastic disease, and a total of four H. pylori-positive patients had peptic ulcer (3 duodenal, 1 antral). H. pylori status was initially assessed by a rapid urease test (Prontodry, Medical Instruments Corporation, Solothurn, Switzerland), and subsequently confirmed by histology. On routine histology, H. pylori-positive biopsies typically exhibited active chronic gastritis as expected. The present report is based on studies of 41 patients (25 female, mean age 53.5±1.6 years) who were H. pylori positive, and 45 who were H. pylori negative (29 female,mean age 50.9±1.8 years) and were used as controls. Samples were assigned randomly within the two groups either to extraction for western blotting or to gastric gland culture. The study was approved by the Ethics Committee of Royal Liverpool and Broadgreen University Hospitals NHS Trust. All patients gave informed consent.

The concentration of amidated gastrins in plasma was determined by RIA using antibody L2 specific for the C-terminal amide sequence of gastrin, as described previously (Varro et al.,1997).

Protein extracts were prepared in RIPA or lysis buffer and western blotting was performed as previously described(Varro et al., 2002) using a rabbit anti-MMP-7 antibody (Chemicon International, Temecula, CA) or anti-NF-κB (p65, p50) antibodies (Santa Cruz Biotechnology, Santa Cruz,MA). Samples were reprobed with a goat anti-β-actin antibody (Santa Cruz Biotechnology).

For immunohistochemistry (IHC), a goat anti-MMP-7 antibody (Santa Cruz Biotechnology) was used with a fluorescein isothiocyanate (FITC)-conjugated donkey anti-goat IgG (Jackson IR, West Grove, PA) as previously described(Wroblewski et al., 2002). In colocalization studies, the following antibodies were used: mouse anti-gastric mucin (Sigma), rabbit anti-pepsinogen (gift from Mike Samloff, Center for Ulcer Research, Los Angeles, CA), rabbit anti-H⁺/K⁺ATP-ase (Calbiochem), rabbit anti-somatostatin(Santa Cruz Biotechnology), rabbit anti-chromogranin A and guinea-pig anti-gastrin (Hussain et al.,1999), mouse anti-vimentin and α-smooth muscle actin(Research Diagnostics, Flanders, NJ). Texas-Red-labelled donkey anti-mouse,anti-rabbit or anti-guinea-pig IgG (Jackson IR) were used as appropriate. For studies of the localization of MMP-7 in gastric biopsies, formalin-fixed paraffin sections were treated with 0.9% hydrogen peroxide in methanol to block endogenous peroxide, and antigen was retrieved by heating in 10 mM EDTA,pH 7.0 for 3 minutes at full pressure in a pressure cooker. Primary rabbit anti-MMP-7 (Chemicon) was employed with detection by the ChemMate EnVision HRP/DAB system (DakoCytomation, Ely, UK).

Human biopsies were sliced into 2 mm² segments using a razor blade and washed in three changes of Hank's Balanced Salt Solution (HBSS). Tissue was incubated in 5 ml 1 mM dithiothreitol for 15 minutes, with continuous gassing with 5% CO₂/95% O₂ at 37°C, and shaking at 100 cycles per minute. Tissue was then washed in HBSS (three times), incubated in 0.5 mg ml^-1 collagenase for 30 minutes (Roche,Lewes, UK), washed again in HBSS (three times) and incubated for a further 30 minutes in collagenase (0.5 mg ml^-1). The tissue was then triturated using a wide mouthed pipette and larger fragments allowed to settle under gravity for 45 seconds. The supernatant containing isolated glands was removed and transferred to a clean tube, and shaken vigorously to release additional glands that were allowed to sediment for 45 minutes. The supernatant (which contained individual cells and debris) was carefully removed and discarded. The isolated glands were routinely cultured up to 72 hours in Dulbecco's Modified Eagle's Medium (DMEM) supplemented with 10% FBS and 1% antibiotic-antimycotic solution (Sigma), at 37°C in 5%CO₂/95% O₂, and the medium was changed every 24 hours.

Human gastric glands were cultured in 6-well dishes. In experiments where AS oligonucleotides were used, either MMP-7 AS oligonucleotide (2 μM) or negative control oligonucleotides (2 μM) were added at the start of culture. Medium and oligonucleotides were replaced daily. Intracellular accumulation of oligonucleotides was confirmed using FITC-labelled MMP-7 AS oligonucleotide (2 μM) viewed live under a Leica DMIRE2 microscope 17 hours after addition. Routinely, gastric colonies were cultured for 17 hours and then mounted on a motorized stage fitted on a Leica DMIRE2 microscope in a heated, humidified chamber (Solent Scientific, Portsmouth, UK) and images captured with a Hamamatsu Orca ER camera (Hamamatsu Photonics, Hamamatsu City,Japan) controlled by Kinetic Imaging AQM-2001 software (Kinetic Imaging,Liverpool, UK). Serial images were analysed using Lucida 4.0:Analyse software(Kinetic Imaging), and gland cell spreading was calculated as the mean area of a gland normalized for cell number determined by Hoechst 33342 (Molecular Probes, Eugene, OR) nuclear staining.

Transwell migration and invasion assays were performed using AGS cells cultured in 24-well plates containing either 8 μm pore Biocoat® control inserts (migration assays) or Matrigel-coated inserts (invasion assays),according to the manufacturer's instructions (Becton Dickinson, Bedford, MA). Briefly, 2.5×10⁴ AGS cells were diluted in 0.5 ml serum free(SF) medium and placed in the insert immersed in 0.75 ml SF medium with or without H. pylori and a mouse monoclonal MMP-7 neutralizing antibody(Chemicon) for 15–22 hours. After incubation, non-invading cells were removed from the upper surface of the membrane by scrubbing, and invading cells on the lower surface of the membrane were stained with Diff-Quik reagent(Dade Behring, Dudingen, Switzerland). Membranes were then removed, mounted in immersion oil and invading cells counted using a Zeiss 25 Axiovert microscope(Carl Zeiss, Welwyn Garden City, UK) and Intellicam software (Matrox, Stoke Poges, UK).

Cells (2×10⁵) were plated in 6-well plates in full medium(FM). The following day, medium was removed and cells were transfected using TransFast (Promega, Madison, WI) in SF medium for 1 hour. Routinely, MMP-7-luc was used at 0.5–1.0 μg well^-1. After transfection, 2 ml FM was added and cells incubated for 20–24 hours. Media was then replaced with 2 ml SF medium and cells were incubated with H. pylori and other compounds as indicated for 6 hours. Luciferase activity was measured with Bright-Glo™ or Dual Glo™ (Promega) using a LumiCount Platereader(Packard BioScience, Pangbourne, UK) according the manufacturer's protocol. Results are presented as fold increase over unstimulated control, so 1.0 signifies no change in luciferase activity. Protein concentration was determined when appropriate using Lowry protein assay kit (Sigma) to monitor plating efficiency and cell death.

To identify the transcription factors that might bind cis-acting DNA elements in nuclear extracts in response to H. pylori, we used Mercury^TM Transfactor Profiling Kit-Inflammation 1 (BD Biosciences Clontech, Palo Alto, CA) according to the manufacturer's protocol. Nuclear extracts were prepared from control cells, and cells were incubated with H. pylori using NE-PER^TM (Pierce Biotechnology, Rockford,IL).

Results are presented as means±s.e.m.; comparisons were made using a t test, and were considered significant at P<0.05.

In initial studies, we examined the abundance of MMP-7 in western blots of gastric biopsies from control and H. pylori-infected subjects. In all subjects, a band corresponding to 28 kDa pro-MMP-7 was detected. However, the abundance was greater in both the corpus and antral region of the stomach from H. pylori-positive subjects compared with controls(Fig. 1). There is a tendency in H. pylori infection towards elevated plasma concentrations of the gastric hormone gastrin (Calam et al.,1997), and so we considered the possibility that gastrin might mediate the effect of H. pylori on MMP-7. However, when H. pylori-positive subjects with plasma gastrin concentrations in the normal range for this assay (<30 pM) were compared with control subjects, there remained a significant difference in the relative abundance of MMP-7 in western blots in the two groups (Fig. 1).

The localization of MMP-7 was determined in both paraffin sections of gastric biopsies and cultured gastric glands. In paraffin sections, MMP-7 immunoreactivity was identified in surface epithelial cells where there was cytoplasmic localization that tended to be subnuclear (i.e. on the basolateral side); there was also staining deep in the gastric gland compatible with localization to chief cells (Fig. 2). The staining in samples from H. pylori-positive subjects was more intense than in those from H. pylori-negative subjects, but the cellular distribution was similar. Cultured gastric glands from the corpus and antrum of normal and H. pylori-positive subjects contained the major cell types found in vivo. Thus, in corpus cultures,antibody to H⁺/K⁺ATPase identified approximately 10% of the total cells as parietal cells, and antibodies to pepsinogen and mucus glycoprotein identified chief and mucus cells (approximately 25 and 35%)respectively; enterochromaffin-like (ECL) cells, revealed by chromogranin A immunoreactivity, were typically <5% total. In antral cultures, parietal cells were absent, and there were abundant mucus (>50% total) and pepsinogen (approximately 40% total) cells, but G- and D-cells revealed by gastrin and somatostatin immunoreactivity, respectively, were typically <5%of total. MMP-7 immunoreactivity was found in many cells that were also stained with antibodies to mucus glycoprotein(Fig. 2), or to pepsinogen,compatible with localization to mucus and chief cells, respectively. In both cases, only a subset of the two cell types expressed MMP-7. By contrast, in the corpus, neither parietal cells nor ECL cells exhibited MMP-7 immunoreactivity (not shown). Moreover, in the gastric antrum, MMP-7 immunoreactivity was not identified in G- or D-cells (not shown). The pattern of localization in different cell types did not differ between H. pylori-positive and -negative subjects.

The majority of cultured glands did not contain cells staining with antibodies to α-smooth muscle actin, or to vimentin (not shown). In those glands in which positive staining was encountered with these antibodies,the stained cells never accounted for more than 3% of the total. Thus, the cultured gastric glands were largely epithelial with little or no contamination by myofibroblasts or fibroblasts. Moreover, the localization of MMP-7 in cultured epithelial cells was compatible with that seen in tissue fixed immediately after sampling.

In cultured glands, MMP-7 was typically sequestered in vesicles that were distinct from those reacting with either mucin or pepsinogen antibodies(Fig. 2). Interestingly, there was a clear localization of MMP-7 to the plasma membrane of cells at the periphery (i.e. the migrating edge) of cultured gastric glands. This pattern was seen in both H. pylori-positive and -negative subjects. When H. pylori was added to control gastric gland cultures, the peripheral cells exhibited a tendency to extend lamellipodia, and MMP-7 immunoreactivity was identified in these structures (Fig. 2).

We then examined the spreading of cultured gastric glands of both H. pylori-positive and -negative subjects. Over periods of up to 48 hours in culture, cells migrated from isolated gastric glands to form monolayer islands, or colonies, of cells. During this process, cell-cell contacts were maintained and, although time-lapse video-microscopy revealed cellular movements indicative of remodelling of these contacts, cells characteristically did not migrate away from the colony. Careful tracking of individual cells provided no evidence for mitosis of gland cells, and in separate studies we also found no evidence of proliferation indicated by nuclear PCNA in cultured glands (S. Kennedy and A. Varro, unpublished). The progressive increase in area covered by individual glands therefore represents both migration and spreading, but not proliferation. Importantly, the expansion of these islands of gastric gland cells from H. pylori-positive subjects was significantly greater than in controls(Fig. 3). In order to determine whether there was a role for MMP-7 in this response, we examined the effect of application of MMP-7 AS oligonucleotides. In validation experiments, the latter were shown to suppress MMP-7 induction by H. pylori in AGS cells (data not shown). We then showed that MMP-7 AS oligonucleotides inhibited migration of cultured glands from H. pylori-positive subjects (Fig. 3). By contrast,application of MMP-7 AS oligonucleotides had no effect on the spreading of gland cells from control subjects. In addition, scrambled oligonucleotides used as a control had no effect on the gland cell spreading.

We explored the induction of MMP-7 by H. pylori in a gastric epithelial cell line (AGS cells). Application of H. pylori to AGS cells for 16 hours increased the abundance of the active 19 kDa form of MMP-7 determined by western blot (Fig. 4). The functional significance of this induction was then examined using migration and invasion assays. In this model, H. pylori increased migration of AGS cells through 8 μm Transwell filters, and stimulated invasion through Matrigel-coated filters compared with control cells (Fig. 4). A role for MMP-7 in both migration and invasion was shown by the inhibition of H. pylori-stimulated migration and invasion by addition of neutralizing MMP-7 antibody (Fig. 4). Application of sonicates of H. pylori had no effect on AGS cell migration or invasion, suggesting that a soluble mediator was not involved.

In order to study the mechanism of H. pylori-induced expression of MMP-7, we first showed that addition of H. pylori to cultured AGS cells increased the expression of a construct consisting of 2.3 kb of the human MMP-7 promoter coupled to a luciferase (i.e. MMP-7-luc) reporter. The induction of MMP-7-luc in AGS cells was similar for two strains that were cag⁺ and vacA⁺ (E6 and 60190). A third strain that was vacA^- (E5) also produced similar responses. Thus, although the secreted cytotoxin VacA has been suggested to activate the small GTPase Rac (Hotchin et al., 2000), and studies described below implicate Rac in the induction of MMP-7, the data suggest this toxin was not required for the AGS cell responses studied here (Fig. 5). In addition, sonicates of H. pylori had no stimulatory effect on MMP-7-luc (not shown). Moreover, when H. pyloriwas applied to 0.2 μm filters cultured over AGS cells transfected with MMP-7-luc, there was again no increase in MMP-7-luc expression(Fig. 5). Together, these lines of evidence suggest that release of a soluble mediator by the bacterium is unlikely to be the main mechanism involved in MMP-7 induction.

It is known that H. pylori increases shedding of HB-EGF and stimulates production of proinflammatory cytokines (IL-6, TNF-α) and chemokines (IL-8) (Crabtree et al.,1994; Moss et al.,1994; Romano et al.,1998; Wallasch et al.,2002). In AGS cells, EGF-family members, IL-8, TNF-α and IL-6 were all found to increase MMP-7-luc expression(Fig. 5). We therefore examined the possibility that H. pylori increased MMP-7 expression in AGS cells via a paracrine mediator. AGS cells were transfected with MMP-7-luc, and then cocultured with untransfected AGS cells grown on 0.2 μm pore filters suspended above the transfected cells. Addition of H. pylori to AGS cells grown on 0.2 μm Transwell filters had no effect on MMP-7-luc expression in the cocultured cells. Putative paracrine signals may of course be diluted in this system, but in the same experimental system using luciferase expression driven by a promoter sequence of plasminogen activator inhibitor-2 (Varro et al.,2002) there was induction of luciferase by H. pylori up to 3–5 fold. Thus, the data support the view that a direct effect due to adherence of the bacteria was responsible for stimulating MMP-7-luc expression(Fig. 5).

Previous studies have implicated NF-κB and AP-1 in H. pylori-stimulated transcriptional responses(Meyer-ter-Vehn et al., 2000; Munzenmaier et al., 1997). We first confirmed that addition of H. pylori to AGS cells increased p50 and p65 NF-κB translocation to the nucleus by western blotting, and that binding of p50 and p65 NF-κB to a consensus NF-κB cis-element was increased in H. pylori-stimulated AGS cell nuclear extracts(Fig. 6). The functional significance of these findings for MMP-7 expression was then demonstrated by showing that an inhibitor of IκB degradation (BAY11-7082) reduced the MMP-7-luc response to H. pylori to control levels. In addition, an increase in c-fos binding to a consensus AP-1 cis-element was demonstrated in nuclear extracts, and cotransfection of AGS cells with MMP-7-luc and a vector encoding c-fos increased expression 2–3 fold(Fig. 6). It is known that H. pylori increases AP-1 activity via the Erk1/2 pathway(Meyer-ter-Vehn et al., 2000)and, consistent with the involvement of this pathway in the present studies,we found that PD98059 (which inhibits the Erk1/2 kinase, MEK) inhibited the response to H. pylori (Fig. 6).

H. pylori is reported to activate small GTPases of the Rho family(Palovuori et al., 2000). We therefore screened three representatives of this GTPase family (RhoA, Rac1,cdc42) for the capacity to stimulate MMP-7-luc expression. Cotransfection of AGS cells with a vector encoding a constitutively active form of RhoA(L63RhoA) strongly stimulated MMP-7-luc; there was a moderate response to a constitutively active form of Rac (L61Rac), whereas a constitutively active form of cdc42 (L61cdc42) had little effect(Fig. 7). We then showed that induction of MMP-7-luc by H. pylori was mediated by RhoA and Rac because cotransfection of MMP-7-luc and a DN-RhoA vector (N19RhoA), or DN-Rac(N17Rac), reduced the H. pylori-increased expression of MMP-7-luc(Fig. 7). In addition,application of the RhoA inhibitory toxin, C3-transferase, reduced MMP-7-luc responses to H. pylori by 76.9±4.4%. The MMP-7-luc responses to L63RhoA and L61Rac were inhibited by BAY11-7082, indicating that NF-κB was downstream of both RhoA and Rac(Fig. 7). Interestingly,cotransfection of MMP-7-luc and L63RhoA with a DN-jun vector resulted in significant inhibition of MMP-7-luc expression, compatible with activation of AP-1 downstream of RhoA. By contrast, the DN-jun vector had no effect on Rac activation of MMP-7-luc expression (Fig. 7).

The present study shows that the gastric oncogenic pathogen H. pylori is associated with increased expression of MMP-7 in a subset of gastric epithelial cells. Moreover, we found that induction of MMP-7 by H. pylori plays a role in stimulating gastric epithelial cell migration. The induction of MMP-7 by H. pylori appears to require adhesion of bacteria to epithelial cells and to be mediated by small GTPases of the Rho family through activation of NF-κB and AP-1 transcription factors.

Recent reports have noted that MMP-7 expression is induced by bacteria in several different epithelia, including colon, bladder and airways, but the cellular mechanisms responsible for this response are unclear(Lopez-Boado et al., 2000; Lopez-Boado et al., 2001; Parks et al., 2001). In mouse small intestine, MMP-7 is expressed in Paneth cells, where it is proposed to function as a processing enzyme for mediators of innate immunity, theα-defensins (Wilson et al.,1999). However, this function seems to be species specific since,in human small intestine, trypsin has been reported to function as a prodefensin-processing enzyme (Ghosh et al., 2002). Moreover, in normal human small and large intestine,MMP-7 was reported to be absent (Ghosh et al., 2002), although expression may increase in ulcerative colitis in association with dysplasia (Newell et al., 2002). By contrast, we found detectable expression of MMP-7 in epithelial cells in both antral and corpus regions of normal human stomach,and in both cases expression was increased in the presence of the gastric pathogen H. pylori.

In mucus and pepsinogen-secreting cells, which were the main MMP-7-containing cells, MMP-7 was found in vesicles distinct from those containing the primary exocrine secretory product. Interestingly, MMP-7 was found at the leading edge of migrating cells, including lamellipodia. By using time-lapse video-microscopy of isolated gastric gland fragments, we showed that cells progressively migrated to form islands, or monolayer colonies of cells, over 2–3 days through a combination of cell migration and cell spreading. This process was accelerated in H. pylori-positive biopsies, and the response was reversed in the presence of MMP-7 AS oligonucleotides. Moreover, in a cancer cell line (AGS cells) H. pylori stimulated both migration and invasion through artificial basement membrane, and in both cases this was inhibited by neutralization of MMP-7. Together, these data support the idea that H. pylori induction of MMP-7 is a functionally important host cell response mediating cell migration in the stomach.

The stimulation of gastric epithelial cell migration in response to H. pylori occurs in otherwise normal gastric gland cells. Related processes have also been described in airways(Dunsmore et al., 1998; Parks et al., 2001). These phenomena may well be protective and serve to maintain epithelial integrity in the face of epithelial damage and infection. However, in stomach, the induction of MMP-7 may also play a role in initiating events that predispose towards malignancy. Thus, there is increased expression of MMP-7 in gastric cancer, particularly at the migrating front of tumours(Adachi et al., 1998; Ajisaka et al., 2001; Honda et al., 1996), and AS inhibition of MMP-7 expression suppressed tumour invasion but not proliferation (Yonemura et al.,2001). In addition, MMP-7 degrades pro-apoptotic factors and proinflammatory cytokines (e.g. TNF-α and Fas ligand). Decreased availability of the latter would tend to suppress apoptosis and may therefore serve to preserve cells after DNA damage and so contribute to tumorigenesis(Mitsiades et al., 2001; Vargo-Gogola et al.,2002).

Previous work has shown that MMP-7 expression is regulated by the PEA3 Ets transcription factor, which appears to enhance transcriptional responses to both β-catenin and AP-1 transcription factors(Crawford et al., 1999; Crawford et al., 2001). Our data suggest that activation of AP-1 by H. pylori(Naumann et al., 1999) is one component of the MMP-7 transcriptional response. But, in addition, we found that H. pylori stimulation of NF-κB activity contributed to the induction of MMP-7. The activation of NF-κB by H. pylori has been described by several groups and has been linked to induction of IL-8(Gupta et al., 2001; Keates et al., 1997; Munzenmaier et al., 1997; Wada et al., 2001). The present data linking NF-κB activation to induction of MMP-7 suggest a wider range of responsive genes than previously supposed. Moreover, although IL-8 increased MMP-7 expression, the present data suggest that this is not an important component of the response to H. pylori. Similarly, the data suggest that other possible endocrine, paracrine, cytokine or bacterial mediators are unlikely to mediate the MMP-7 response to H. pylori.Instead, we suggest that cell adhesion of the bacterium and activation of small GTPases of the Rho family (Churin et al., 2001; Hotchin et al.,2000; Palovuori et al.,2000) are responsible for induction of MMP-7. In particular, our data suggest that induction of MMP-7 via NF-κB occurs as a consequence of activation of Rho and Rac (Montaner et al., 1999). Interestingly, there was differential activation of AP-1 and NF-κB, since both Rho and Rac activated the latter, but only Rho activated AP-1.

Infection with H. pylori affects approximately 50% of the western population; some patients exhibit a progression through chronic gastric atrophy to cancer, others develop peptic ulcer, but most do not exhibit either disease (Blaser and Berg, 2001; Peek and Blaser, 2002; Uemura et al., 2001). The different outcomes from infection are thought to reflect host, pathogen and environmental variables (Blaser and Berg,2001; Fox and Wang,2001). The ability of epithelial cells to migrate after damage is usually considered to be an important component of the host defence mechanism. In this context, we suggest that H. pylori induction of MMP-7 can be considered part of a protective host response. In addition, though, we also suggest that, over longer periods, the prolonged induction of MMP-7 accelerates an oncogenic progression via disruption of epithelial organization and increased invasion. The identification of increased MMP-7 expression as a consequence of H. pylori infection should provide a basis for detailed study of the links between bacterial infection and cancer.

We thank Cathy McLean for skilled technical assistance, Claire Phythian for help with some luciferase assays, Debbie Sales for help with H. pylori culture and Susie Kennedy for help with some immunocytochemistry on cultured glands. We also thank the following for making available plasmids for this study: Alan Hall, Tim Wang, Lynn Matrisian and John Neoptolemos. This work was supported by grants from the Medical Research Council and the Wellcome Trust.