Gastric infection by Helicobacter pylori is invariably associated with chronic inflammation.1 The topographical extent, severity and clinical sequelae of chronic gastritis depend on a range of host, bacterial and environmental co-factors. Most infected individuals are symptom free, but H pylori associated chronic gastritis may lead to peptic ulceration. A small proportion of infected individuals progress along a multistage pathway to gastric adenocarcinoma, via stages of multifocal gastric atrophy, intestinal metaplasia and dysplasia.2 It is clear that H pylori associated diseases are associated with major architectural alterations in the organisation of the gastric epithelium but this tissue remodelling process remains poorly understood.

The composition of the extracellular matrix (ECM), and the balance between synthesis and degradation of its components, plays a fundamental role in determining tissue organisation. Major biological players in the regulation of the ECM are the matrix metalloproteinases (MMPs). Over 20 MMPs have been identified in man. This complex enzyme system influences both proteolytic degradation of structural ECM components and the activation/inactivation of a diverse array of cell-surface- and ECM-bound proteins, including growth factors and cytokines.3 Host-derived MMPs are necessary for the successful execution of the inflammatory response to infection,4 and play a key role in cell proliferation, migration, differentiation and survival.5

Elevated gastric mucosal abundance of a number of MMPs has been reported in H pylori infection,6^–8 as might be anticipated given the known role for these proteases in inflammation. However, the activity of MMPs is tightly regulated by the release of specific inhibitors, the tissue inhibitors of metalloproteinases (TIMPs).9 The balance between ECM catabolism and anabolism will depend on the relative activity of proteases (MMP) and anti-proteases (TIMP). We have little knowledge of the effect of H pylori infection on the regulation of TIMP activity in the gastric mucosa. The aim of this study was to investigate changes in TIMP expression in the gastric mucosa in chronic gastritis associated with Helicobacter infection in man and in a murine model.

METHODS

Patients and tissue samples

Endoscopic pinch biopsies of gastric mucosa were obtained from consecutive patients attending for the investigation of dyspepsia, avoiding macroscopic lesions (eg, ulceration). Subjects with a history of recent intake of antibiotics (preceding month) were excluded. Details of current medication and endoscopic findings were recorded. Immediately prior to endoscopy, a fasting venous blood sample was obtained and aliquots of serum and plasma were separated and stored at −20°C until assay. During endoscopy, gastric corpus biopsies were obtained from the greater curve for histopathology and for either RT-PCR or immunoblot analyses as detailed below. Additional antral biopsies were taken for histopathology and rapid urease testing. Histopathology samples were immediately fixed in formalin. Samples for RNA analyses were placed in RNAlater (Ambion, Austin, TX, USA) and stored at −80°C, whereas samples for immunoblotting were placed in RIPA buffer (Upstate, Lake Placid, New York, USA), stored −80°C, as previously described.

H pylori status was initially assessed by rapid urease test (Prontodry; Medical Instruments Corporation, Solothurn, Switzerland) and subsequently confirmed by histology and serology. Fifty-three patients were studied in total (27 H pylori negative and 26 H pylori positive). The study was approved by the local ethics committee and all participants gave written informed consent.

Histopathology

The presence and severity of gastritis was graded in paraffin sections of antral and corpus biopsies stained with haematoxylin and eosin according to the revised Sydney System. Acute inflammation (activity) and chronic inflammatory cell infiltrate (inflammation) were each scored on a scale 0, 1, 2 or 3 by the same pathologist (AK) who was unaware of subsequent assay results.

Blood assays

H pylori serology was performed using a commercial ELISA assay (Biohit, Torquay, UK). The concentration of amidated gastrins in plasma was determined by radioimmunoassay using antibody L2 specific for the COOH-terminal amide sequence of gastrin, as described previously.10 Serum pepsinogens have been employed as a marker of gastric corpus atrophy, with very low values associated with severe atrophic gastritis. Based on histopathology of greater curve biopsies, none of the patients studied had significant gastric corpus atrophy (atrophy grade 2 or 3). However, atrophy can be patchy and multifocal such that accurate grading of corpus atrophy requires biopsies from multiple sites. To verify that none of our subjects had established corpus atrophy, we assayed serum pepsinogen I levels by commercial ELISA (Biohit) and found that no patient had very low levels (<25 μg/l).

RNA extraction and real-time quantitative RT-PCR analysis

Total RNA was extracted from gastric corpus biopsies (RNeasy; Qiagen, Crawley, UK), reverse transcribed, and then cDNA was amplified with specific primers designed using RefSeq database (table 1). The sequences, usually within 200–300 bp from the polyA tail, were identified and 24–25-mer primer pairs were designed to amplify a 100–160 bp fragment. Primer specificity was tested by BLAST searches. The synthesised primers were tested for lack of primer-dimer formation and for the absence of non-specifically amplified, longer or shorter than expected PCR products. PCR was performed using Excite Core reagents (Biogene, Cambridge, UK) and a Rotorgene (Corbett Research, Cambridge, UK) for 45 cycles. The cycling conditions were an initial denaturation step of 95°C for 10 min followed by 45 cycles of 95°C for 30 s, 65°C for 60 s (55 s for all TIMPs, 60 s for MMP-3 and -8) and 72°C for 30 s. The expected product size was confirmed by running the PCR product on 1.5% agarose gels. The Lightcycler h-HPRT housekeeping gene set (Roche Diagnostics, Mannheim, Germany) was used to normalise samples. Threshold cycles (Ct) were defined as the fractional cycle number at which the fluorescence reached 10 times the standard deviation of the baseline. Gene expression ratios relative to h-HPRT were calculated using the delta-delta method.

Immunoblot analysis

Western blotting was performed as previously described.10 Briefly, protein extracts were prepared in lysis buffer buffer containing 20 mM Tris (pH 7.8), 150 mM NaCl, 2 mM EDTA, 0.5% Nonidet P-40, 10 mM NaF, 15 mg/ml benzamide, 8.8 mg/ml sodium orthovanadate, 0.5 mM dithiothreitol, 10 mg/ml phenylmethylsulfonyl fluoride, and one protease inhibitor mixture tablet per 10 ml of lysis buffer. A 40 μg quantity of each lysate was electrophoresed on a 12% SDS-polyacrylamide gel. After electrophoresis, the proteins were blotted on nitrocellulose membranes, and immunodetection of the proteins was performed. The primary anti-TIMP antibodies used were obtained from RDI (Flanders, NJ, USA). Enhanced chemiluminescence (SuperSignal West Pico chemiluminescent substrate; Pierce, Tattenhall, UK) was used to identify the proteins of interest.

Immunohistochemistry

Cellular localisation of TIMP-1, -3 and -4 peptides was studied by indirect immunohistochemistry using formalin-fixed paraffin sections using the Chemmate Envision (Dakocytomation, Ely, UK) horseradish-peroxidase-labelled polymer technique. Sections were cut at 3 μm thickness and mounted on aminopropyltriethoxysilane-coated slides. Sections were dewaxed and rehydrated in absolute industrial methylated spirits and running water. Antigen retrieval was carried out by immersing the sections in EDTA buffer solution, pH 7.0, and heating using a pressure cooker for 2 min at full pressure followed by washing in running tap water. An automated immunostainer (Lab Vision) was used for staining. Sections were treated with a solution of 3% H₂O₂ to block non-specific staining activity. The sections were then incubated with the primary antibody (table 2) for 40 min at room temperature followed by washing in 0.05% Tris buffered saline. The sections were then incubated with the secondary antibody (Chemmate Envision; Dakocytomation) for 10 min at room temperature followed by washing in 0.05% Tris buffered saline. Colour was developed using 3,3′-diaminobenzadine substrate solution. Sections were washed, counterstained with Harris’s haematoxylin, dehydrated in absolute industrial methylated spirits, cleared in xylene, and mounted with DPX medium.

Real-time PCR assay of gastric TIMP mRNA abundance in the stomach of H felis infected INS-GAS transgenic mice

We also examined the abundance of TIMP mRNA in the stomach of INS-GAS transgenic mice infected with Helicobacter felis. The conditions and methods employed have been described elsewhere.11 12 Briefly, male INS-GAS mice were inoculated with H felis (ATCC 49179) at age 2–3 months, and infection was confirmed at 12 and 24 weeks post-inoculation by antral urease test as previously described.13 The present study used total RNA extracted from animals at 3 months (control, n = 6; infected, n = 5) and 6 months (control, n = 5; infected, n = 5). Real-time quantitative RT-PCR was performed as previously described,14 15 using primers specific for TIMP-1, -3 and -4. Sequences for RT-PCR primers in the murine studies are shown in table 1.

Statistical analysis

Results are presented as means (SEM); comparisons were made using a Student t test (SPSS for Windows, version 11.0; SPSS, Chicago, Illinois, USA) and were considered significant at p<0.05. Significant differences between infected and uninfected subjects are denoted by an asterisk in the figures.

RESULTS

Quantification of TIMP and mRNA and protein abundance in gastric corpus of H pylori infected and uninfected patients

We used real-time RT-PCR to quantify the abundance of mRNAs for TIMP-1 to -4 in the gastric corpus. Compared with uninfected controls (n = 14 subjects), mRNAs for TIMP-1, -3 and -4 were significantly more abundant in gastric corpus of H pylori positive subjects (n = 13) (fig 1A). Western blot analyses confirmed these findings (fig 1B), demonstrating increased abundance of protein expression in biopsies from H pylori positive subjects (n = 12) versus uninfected controls (n = 13). In addition, we quantified gastric corpus mRNA expression for selected proteases (MMP-3 and -8) and found no significant difference between infected and uninfected subjects (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Abundance of TIMP mRNA and protein in biopsies taken from the corpus mucosa of H pylori positive and negative subjects attending for gastroscopy. Each bar represents mean (SEM); *p<0.05. (A) Quantitative real-time PCR for TIMP-1 to -4, with values expressed in arbitrary units relative to the house keeping gene HPRT. (H pylori positive, n = 13; H pylori negative, n = 14). For H pylori positives, data are presented for a normogastrinaemic sub-group (gastrin <35 pM) and for all infected subjects. (B) Quantification of western blots for TIMP-1, -3 and -4 with protein abundance expressed relative to GAPDH. (H pylori positive, n = 12; H pylori negative, n = 13). Images of representative western blots are shown. For H pylori positives, data are presented for a normogastrinaemic sub-group (gastrin <35 pM) and for all infected subjects.

Cellular distribution of TIMP protein expression in the human gastric corpus by immunohistochemistry

In the light of the above findings for TIMP-1, -3 and -4, we performed indirect immunohistochemistry to examine the cellular localisation of the upregulated proteins (fig 2A–F). This provided further support for the observed changes in overall expression of TIMPs according to H pylori status, with more intense immunopositivity seen in H pylori positive cases (fig 2A, C). The cellular localisation of individual TIMPs in the gastric mucosa showed specific patterns, with each protease inhibitor having a well-defined distribution. Surface gastric epithelial cells demonstrated strongest immunoreactivity for TIMP-4 (fig 2G) but weaker or focal immunolabelling for TIMP-1 and -3. Glandular epithelial cells exhibited strongly positive immunoreactivity for TIMP-1, -3 and -4, especially in chief cells in H pylori positive cases. All TIMPs showed focal positivity in stromal cells. Leucocytes were positive for TIMP-3 and -4, but only weakly positive for TIMP-1. Strong immunolabelling of cells with the morphology of myofibroblasts was seen for TIMP-3 (fig 2F).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2 Immunohistochemical staining for TIMP proteins in sections of formalin-fixed endoscopic biopsies from gastric corpus mucosa. Biopsies were stained for TIMP-1, -3 and -4 using specific antibodies. (A) TIMP-1 (H pylori positive; ×200 magnification) – immunopositivity of glandular epithelial cells (white arrow, g) and weak immunolabelling of stromal inflammatory cells (black arrow, i). (B) TIMP-1 (H pylori positive; ×400 magnification) – higher power view showing weak immunopositivity of stromal inflammatory cells. (C) TIMP-1 (H pylori negative; ×200 magnification) – absent immunoreactivity in an uninfected subject. (D) TIMP-3 (H pylori positive; ×200 magnification) – immunolabelling of surface (white arrow, s) and glandular epithelial cells (white arrow, g). (E) TIMP-3 (H pylori positive; ×400 magnification) – higher power view showing focal immunopositivity of surface epithelial cells (white arrow, s). (F) TIMP-3 (H pylori positive; ×200 magnification) – stromal cell staining, including immunoreactivity of cells with spindle-shaped morphology of myofibroblasts (black arrow, m) in addition to inflammatory cells (black arrow, i). (G) TIMP-4 (H pylori positive; ×200 magnification) – immunopositivity of surface (white arrow, s) and glandular (white arrow, g) epithelial cells. (H) TIMP-4 (H pylori positive; ×200 magnification) – immunopositivity of glandular epithelial cells (white arrow, g), particularly chief cells.

Variation in gastric TIMP abundance with grade of inflammation and circulating gastrin

TIMPs are expressed by a number of cell types, including infiltrating inflammatory cells (leucocytes). In order to assess whether gastric corpus expression of TIMPs was predominantly influenced by the degree of leucocytic infiltration, we compared TIMP mRNA abundance among H pylori positive subjects according to scores for activity and chronic inflammation. For the purpose of this analysis, patients were dichotomised on the basis of gastric corpus histology scores (mild: Sydney score zero or 1; severe: Sydney score 2 or 3). No significant difference was found for any of the TIMPs when comparing subjects with mild versus severe activity or mild versus severe inflammation (table 3). This suggests that levels of TIMP mRNA in gastric biopsy samples are not simply reflective of the density of inflammatory cell infiltrate and that other cellular sources (eg, epithelial cells and stromal cells) may be more important in determining total tissue expression levels.

Based on the results of gastrin assay, subjects were categorised as hypergastrinaemic (>35 pM) or normogastrinaemic. The abundance of gastric corpus mRNA and protein for TIMP-1 to -4 did not vary significantly according to gastrin levels, irrespective of H pylori status. Differences in gastric TIMP abundance between H pylori positive and negative patients remained significant when comparing normogastrinaemic subjects only (fig 1A, B), suggesting that gastrin is not a major mediator of increased TIMP expression in response to infection.

Quantification of TIMP mRNA abundance in the stomach of H felis infected INS-GAS transgenic mice

Consistent with our human data, we found increased abundance of mRNA for TIMP-1 and -3 in the stomach of Helicobacter-infected animals compared with controls at 6 months post-inoculation (fig 3). At this time point, mice have not developed significant hypergastrinaemia compared with controls. In this murine H felis model, we found gastric TIMP-4 levels were below the detection limit of the assay. No significant increases in abundance of any of the TIMPs were apparent after H felis infection at the earlier time point of 3 months (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3 Abundance of TIMP mRNA in gastric mucosa of H felis infected and uninfected (control) INS-GAS mice. Results of real-time quantitative PCR with mRNA abundance expressed relative to β-actin. Data for 6 months post-infection (control, n = 5; H felis, n = 5). *p<0.05. Data are presented as percentages of control.

DISCUSSION

In this study we show that there are statistically significant elevations in the level of mRNA and protein for TIMP-1, -3 and -4 in the glandular (corpus) mucosa of patients infected by H pylori in comparison with uninfected control subjects. Furthermore, we showed a similar pattern of altered TIMP expression in an animal model of Helicobacter infection, with elevated levels of both TIMP-1 and -3 mRNA in the stomach of INS-GAS transgenic mice infected with H felis.

Immunolocalisation of TIMP proteins using specific antibodies revealed that there are several cellular sources for these protease inhibitors in the human gastric corpus mucosa, including both epithelial and stromal cells. Our immunohistochemical observations suggest that individual TIMPs are upregulated in specific cell types and/or compartments. TIMP-1, -3 and -4 proteins were abundant in glandular epithelial cells, but only TIMP -3 and -4 exhibited strong staining of the surface epithelium. Strong immunoreactivity of myofibroblast-like cells was restricted to TIMP-3. These observations suggest different regulatory signals and roles for individual protease inhibitors, requiring further characterisation.

The existing literature relating to TIMP expression in benign gastric disease is limited. Upregulation of TIMP-1 mRNA has been reported in epithelial and stromal cells adjacent to experimental gastric ulcers in rodents.16 In a small study in man, Bergin et al were unable to demonstrate differences in expression of TIMP-1 and -2 protein in gastric biopsies from H pylori infected and uninfected subjects.7 Stimulation of an epithelial cell line (AGS cells) with H pylori resulted in increased secretion of TIMP-3 but not TIMP-1 or TIMP-2.17 The present report is the first to fully characterise the abundance of all four TIMPs at the mRNA and protein levels in the gastric corpus mucosa in H pylori infected subjects. Our mRNA and protein analyses produced consistent findings, with selective upregulation of TIMP-1, -3 and -4 in biopsies from infected subjects.

It is perhaps not surprising that tissue levels of specific TIMPs are raised in biopsies of inflamed gastric mucosa compared with normal gastric mucosa. These proteins have a number of well-defined roles in inflammation and repair, and their cellular sources include leucocytes.9 However, we found no strong correlation between the intensity of acute or chronic inflammatory cell infiltrate and the total level of TIMP mRNA, suggesting that other cellular sources are important in determining total protease inhibitor abundance. Immunohistochemistry suggested that both epithelial cells and stromal cells are key sources for TIMPs. In the present study, we focussed on studying changes in the specialised glandular compartment of the corpus mucosa where chronic infection may be associated with dramatic remodelling during the process of gastric atrophy.2

In the present study, we did not examine the mechanisms whereby H pylori associated gastritis leads to selective upregulation of protease inhibitors in specific cell types. We speculated that the hormone, gastrin, might induce TIMP expression. Hypergastrinaemia may occur in response to H pylori infection or intake of potent acid suppressing drugs. However, the differences we observed in overall abundance of TIMPs between H pylori infected and uninfected subjects remained significant when only normogastrinaemic subjects were compared. Furthermore, our animal data come from a model of Helicobacter infection in which gastrin levels in infected mice are not significantly different from uninfected controls at the time points studied.18 19 This suggests that it is the infection and associated inflammation, rather than hypergastrinaemia per se, that are responsible for upregulating the protease inhibitors.

TIMP-family members differ in their ability to inhibit individual MMPs. Whereas TIMP-4 inhibits most of the metalloproteinases, the other TIMPs appear to have a more restricted set of target proteases.9 The net effect of enhanced TIMP production on the activity of individual MMPs in the gastric mucosa in H pylori infection will depend not only on their protease specificity, but also on the relative magnitude and direction of any infection-induced changes in MMP production. Elevated gastric abundance or activity of MMP-2,7 MMP-76 8 and MMP-97 has been reported in H pylori infection, but we found no significant increase in mRNA abundance for MMP-3 or -8 in the gastric corpus. Raised tissue levels of protease inhibitors have been proposed as a pro-fibrogenic factor in conditions such as chronic liver disease20 and Crohn disease.21 Gastric atrophy is characterised by loss of specialised cells and expansion of ECM (fibrosis)22 and enhanced TIMP expression might favour gastric mucosal fibrosis.

The biological effects of MMPs and TIMPs are not restricted to regulating catabolism of ECM structural proteins. The protease system is involved both in the activation and inactivation of cytokines and growth factors.9 23 24 We have reported that H pylori induces epithelial-derived MMP-7 (25) which increases the bioavailability of insulin-like growth factor-II released from myofibroblasts, serving as a stimulus for gastric myofibroblast and epithelial cell proliferation.26 Furthermore, it is increasingly recognised that TIMP proteins have biological effects that are not related directly to their enzyme-inhibitory actions.9 Observed effects on cell growth and survival cannot always be clearly attributed to abrogation of MMP activity, and individual TIMPs exhibit apparently opposing phenotypic effects in different systems.9

The reasons why only a minority of H pylori infected subjects develop corpus atrophy and subsequent cancer are unclear, but host genetic factors, bacterial virulence factors and environmental co-factors are all implicated. Interestingly, recent work suggests that genetic variants in specific matrix metalloproteinase genes (MMP-7 and -9) may be associated with increased risk of gastric ulcer among H pylori infected subjects.27 Hence, the balance between the MMP and TIMP system in an individual may be a key factor in determining the fate of the gastric mucosa in response to infection. Examination of the influence of host genetic and bacterial virulence factors on levels of gastric TIMP production in man may give important insights into disease pathogenesis.

CONCLUSIONS

H pylori infection is associated with upregulation of TIMP-1, -3 and -4 in the gastric corpus mucosa. Based on immunohistochemical observations, cellular sources of TIMPs include gastric epithelial cells, infiltrating inflammatory cells and stromal myofibroblasts. These multifunctional proteins are likely to be key players in redefining the gastric microenvironment in response to H pylori infection and in determining the clinical outcome of infection.