Article Text

Download PDFPDF
Original article
Epidermal growth factor receptor inhibition downregulates Helicobacter pylori-induced epithelial inflammatory responses, DNA damage and gastric carcinogenesis
  1. Johanna C Sierra1,
  2. Mohammad Asim1,
  3. Thomas G Verriere1,
  4. M Blanca Piazuelo1,
  5. Giovanni Suarez1,
  6. Judith Romero-Gallo1,
  7. Alberto G Delgado1,
  8. Lydia E Wroblewski1,
  9. Daniel P Barry1,
  10. Richard M Peek Jr1,2,3,
  11. Alain P Gobert1,4,
  12. Keith T Wilson1,2,3,4,5
  1. 1Division of Gastroenterology, Hepatology, and Nutrition, Department of Medicine, Vanderbilt University Medical Center, Nashville, Tennessee, USA
  2. 2Department of Cancer Biology, Vanderbilt University Medical Center, Nashville, Tennessee, USA
  3. 3Department of Pathology, Microbiology, and Immunology, Vanderbilt University Medical Center, Nashville, Tennessee, USA
  4. 4Center for Mucosal Inflammation and Cancer, Vanderbilt University Medical Center, Nashville, Tennessee, USA
  5. 5Veterans Affairs Tennessee Valley Healthcare System, Nashville, Tennessee, USA
  1. Correspondence to Dr Keith T Wilson, Vanderbilt University Medical Center, 2215 Garland Ave, MRB IV 1030C, Nashville, TN 37232, USA; keith.wilson{at}


Objective Gastric cancer is the third leading cause of cancer death worldwide and infection by Helicobacter pylori is the strongest risk factor. We have reported increased epidermal growth factor receptor (EGFR) phosphorylation in the H. pylori-induced human carcinogenesis cascade, and association with DNA damage. Our goal was to determine the role of EGFR activation in gastric carcinogenesis.

Design We evaluated gefitinib, a specific EGFR inhibitor, in chemoprevention of H. pylori-induced gastric inflammation and cancer development. Mice with genetically targeted epithelial cell-specific deletion of Egfr (EfgrΔepi mice) were also used.

Results In C57BL/6 mice, gefitinib decreased Cxcl1 and Cxcl2 expression by gastric epithelial cells, myeloperoxidase-positive inflammatory cells in the mucosa and epithelial DNA damage induced by H. pylori infection. Similar reductions in chemokines, inflammatory cells and DNA damage occurred in infected EgfrΔepi versus Egfrfl/fl control mice. In H. pylori-infected transgenic insulin-gastrin (INS-GAS) mice and gerbils, gefitinib treatment markedly reduced dysplasia and carcinoma. Gefitinib blocked H. pylori-induced activation of mitogen-activated protein kinase 1/3 (MAPK1/3) and activator protein 1 in gastric epithelial cells, resulting in inhibition of chemokine synthesis. MAPK1/3 phosphorylation and JUN activation was reduced in gastric tissues from infected wild-type and INS-GAS mice treated with gefitinib and in primary epithelial cells from EfgrΔepi versus Egfrfl/fl mice. Epithelial EGFR activation persisted in humans and mice after H. pylori eradication, and gefitinib reduced gastric carcinoma in INS-GAS mice treated with antibiotics.

Conclusions These findings suggest that epithelial EGFR inhibition represents a potential strategy to prevent development of gastric carcinoma in H. pylori-infected individuals.


Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Significance of this study

What is already known on this subject?

  • Helicobacter pylori infection is the primary cause of gastric cancer, which remains a major public health problem as the third leading cause of cancer deaths.

  • Antibiotic-based eradication of H. pylori may not be of benefit once advanced histological lesions have occurred, indicating that other chemopreventive strategies are needed.

  • Epidermal growth factor receptor (EGFR) overexpression and transactivation (phosphorylation) is associated with gastric carcinogenesis in humans, but EGFR has not been investigated in animal models of H. pylori-induced gastric dysplasia and carcinoma.

What are the new findings?

  • Administration of a clinically available EGFR inhibitor, gefitinib, blocked H. pylori-stimulated phosphorylation of EGFR in gastric epithelial cells in animals, and in vitro.

  • Gefitinib administration reduced infiltration of inflammatory cells, chemokine and cytokine expression, DNA damage and development of gastric dysplasia and carcinoma in mouse and gerbil models of H. pylori-induced gastric cancer. C57BL/6 mice with genetically engineered epithelial-specific knockout of Egfr also exhibited attenuated inflammation and DNA damage.

  • Epithelial EGFR signalling occurred through MAPK1/3 (ERK1/2) and activator protein 1 pathways, which thus may contribute to the pro-inflammatory and protumourigenic responses.

  • Antibiotic eradication of H. pylori did not reduce EGFR phosphorylation in a human longitudinal cohort or in insulin-gastrin mice, and gefitinib significantly reduced gastric carcinoma in mice treated with antibiotics.

How might it impact on clinical practice in the foreseeable future?

  • These studies provide evidence that EGFR inhibition may be an effective strategy for the chemoprevention of H. pylori-associated gastric carcinogenesis, particularly in individuals at high-risk for neoplastic progression.


Infection with Helicobacter pylori is directly associated with development of gastric cancer in humans,1 the third leading cause of cancer death worldwide ( While systematic antibiotic-based eradication has been proposed,2 treatment failure is common, and high rates of recurrence and/or recrudescence occur in middle-income countries.3 ,4 Furthermore, the benefit of H. pylori eradication is uncertain once precancerous lesions develop.5 Thus, alternative strategies are needed to reduce gastric cancer incidence.

Epidermal growth factor receptor (EGFR) is a transmembrane glycoprotein belonging to the receptor tyrosine kinase group.6 EGFR activation enhances cell growth, proliferation, differentiation and wound healing, but has been implicated in development of malignancies, including gastric cancer.7 ,8 EGFR overexpression is observed in 27%–44% of gastric cancer cases and is associated with worse outcomes.7 ,9 Our group previously reported that EGFR transactivation (phosphorylation) is linked to DNA damage in gastric epithelial cells (GECs) after infection with H. pylori, both in cell lines and in tissues, and that epithelial levels of activated EGFR are increased in human gastritis and atrophic gastritis, but not in the later stages of the cancer cascade.10 Importantly, levels of phosphorylated EGFR (pEGFR) were upregulated in baseline gastric biopsies from subjects in a long-term cohort in Colombia that later progressed to intestinal metaplasia or dysplasia as compared with non-progressors.10 These data suggest that EGFR activation is involved in events leading to gastric cancer. Gefitinib, an inhibitor of EGFR, has been used in the treatment of a variety of cancers and has shown efficacy in human non-small cell lung cancer and proximal gastric cancer,11 ,12 and as a chemopreventive agent for lung and breast cancer in animal models.13 ,14

Herein, we show that gefitinib treatment of H. pylori-infected mice and gerbils attenuated EGFR phosphorylation, the pro-inflammatory response of GECs, DNA damage and development of inflammation, dysplasia and cancer. Epithelial-specific gene knockout of Egfr also decreased innate immune response of GECs, gastric inflammation and DNA damage. The protective effect of gefitinib was linked to inhibition of mitogen-activated protein kinase 1/3 (MAPK1/3)-activator protein 1 (AP-1) signalling. Our findings indicate a critical role for EGFR signalling in gastric inflammation and carcinogenesis, and that inhibition of EGFR may serve as a strategy for decreasing cancer risk.


Cells and bacteria

Mouse conditionally immortalised stomach epithelial (ImSt) were used as described.10 ,15 Human adenocarcinoma gastric cells (AGS) cells were from American Type Culture Collection (ATCC). Cells were pretreated with 10 μM gefitinib, PD98059 (2-(2′-amino-3′-methoxyphenyl)-oxanaphthalen-4-one) or SR11302, for 1 hour and infected with H. pylori PMSS1, 7.13 or 60190 and its isogenic cagA and vacA mutants, at a multiplicity of infection of 10–100.10

Generation of EgfrΔepi mice

Egfrtm1Dwt mice were crossed to a Tg(Foxa3-cre)1Khk transgene-expressing mouse strain,16 ,17 generating Egfrfl/f/;Foxa3cre/+ (EgfrΔepi) mice. Littermate Egfrfl/fl and EfgrΔepi mice were used for experiments.

Animals and infections

Mice and gerbils were fed a defined rodent diet (AIN-76A, BioServ), supplemented or not with 200 ppm gefitinib,18 1 week before infection and for the duration of experiments. C57BL/6, Egfrfl/fl, EgfrΔepi and transgenic insulin-gastrin (INS-GAS) mice were infected with 1×109 colony-forming unit (CFU) of H. pylori PMSS1, a vacA s2/m2 strain15; Mongolian gerbils were infected with 2×109 CFU of H. pylori 7.13,19 a VacA-negative strain.20 Colonisation of all included animals was verified by culture.19 ,21

Isolation of epithelial cells from gastric tissue

GECs were isolated by dissociation and dispersion.15

Western blotting

Proteins from cell lines and tissues were analysed as described.10 Antibodies are summarised in online supplementary methods.

Immunohistochemistry and imaging analysis

See online supplementary methods.

Flow cytometry

See online supplementary methods.

Chemokine and cytokine analysis

Gastric tissues were lysed and processed using a Mouse Milliplex kit (Millipore) and analysed on a Luminex FlexMap 3D machine.17 Chemokine (C-X-C motif) ligand 1 (CXCL1) was also measured using the DuoSet ELISA Mouse CXCL1 kit (R&D Systems).

mRNA analysis

Expression of Cxcl1, Cxcl2, Il1b and CXCL8 was assayed by real-time PCR.15 ,17

Culture of primary GEC monolayers

See online supplementary methods.

Statistical analysis

Results are expressed as mean±SEM. For multiple groups, comparisons were performed using the Kruskal-Wallis non-parametric multiple comparison test followed by paired comparisons using the Mann-Whitney U test. Contingency analysis was performed using the χ2 test.


Gefitinib effectively reduces EGFR activation and DNA damage in response to H. pylori in vivo

EGFR activation in gastric tissues has been demonstrated as early as 24 hours after H. pylori infection.22 Accordingly, EGFR was phosphorylated at tyrosine residue 1068 in mice infected for 1, 2 or 7 days with PMSS1 (see online supplementary figure S1 and figure 1A). EGFR activation in GECs was blocked when the diet was supplemented with gefitinib (figure 1A). Colonisation density was increased in gefitinib-treated versus control mice (see online supplementary figure S2A). Chronic inhibition was confirmed by immunohistochemistry showing a marked attenuation of pEGFR in gastric tissues from mice infected with H. pylori strain PMSS1 for 1 month and treated with gefitinib compared with animals receiving the control diet (figure 1B).

Figure 1

Levels of pEGFR and pH2AFX in C57BL/6 mice infected with Helicobacter pylori PMSS1 on control or gefitinib-supplemented diet. (A) Western blot analysis for pEGFR (Y1068) and total epidermal growth factor receptor (EGFR) in gastric epithelial cells isolated 7 days postinfection. (B) Immunohistochemistry for pEGFR in gastric tissues after 4 weeks infection. (C) Percentage of cells pH2AFX-positive by flow cytometry analysis. (D and E) Quantification of nuclei positive for pH2AFX, and representative pH2AFX immunohistochemistry. *p<0.05, ***p<0.001 vs uninfected mice; §p<0.05, §§p<0.01 vs PMSS1 infected on control diet. In B and E, scale bar=100 µm.

There was increased phosphorylation of H2AFX at serine 139, a marker of DNA damage,23 when assessed by flow cytometry in GECs isolated from H pylori-infected mice versus uninfected animals (figure 1C). Significant reduction of pH2AFX-positive GECs was observed in infected mice maintained on the gefitinib-supplemented diet (figure 1C). A similar reduction in GECs positive for pH2AFX was observed by immunohistochemistry, quantified by image analysis, in infected mice on the gefitinib diet (figure 1D, E).

EGFR inhibition leads to a reduction on H. pylori-induced inflammatory response

We directly evaluated levels of inflammatory cell infiltration in the gastric mucosa of H. pylori-infected mice by immunohistochemistry for myeloperoxidase (MPO), an enzyme abundantly expressed in polymorphonuclear cells (PMNs). The number of MPO-positive cells was increased in PMSS1-infected mice as compared with uninfected controls (figure 2A), and inhibition of EGFR activation by gefitinib significantly reduced the number of infiltrating cells as quantified by image analysis (figure 2A, B). We then assessed chemokines and cytokines produced in gastric tissues by Luminex-based multiplex protein analysis. In mice infected with H. pylori, there were increased concentrations of the chemokines CXCL1, CXCL2 and CCL5; the macrophage-derived cytokines CSF3 and interleukin (IL)-1β and the prototype T helper 17 cytokine, IL-17 (figure 2C). The induction of each of these pro-inflammatory parameters was decreased with gefitinib treatment in infected animals (figure 2C). Additional targets in the multiplex assay not significantly affected by gefitinib treatment are shown in online supplementary table S1.

Figure 2

MPO and chemokine/cytokine levels in C57BL/6 mice infected with PMSS1 for 4 weeks and maintained on control or gefitinib-supplemented diet. (A) Quantification of MPO-positive cells assessed by immunohistochemistry. (B) Representative MPO staining in gastric tissues. Scale bar=100 µm in upper panel. Scale bar=50 µm in lower panel. (C) Chemokine and cytokine levels in gastric tissue lysates assessed by Luminex assay. *p<0.05, ***p<0.001 vs uninfected mice; §p<0.05, §§p<0.01, §§§p<0.001 vs PMSS1 infected on control diet.

To confirm the role of epithelial EGFR as a critical mediator of H. pylori-induced gastric inflammation, we used EgfrΔepi mice. Western blot analysis demonstrated loss of EGFR in GECs isolated from EgfrΔepi mice, when compared with Egfrfl/fl mice (figure 3A). This was confirmed by immunohistochemistry for pEGFR, showing epithelial staining in the tissues from PMSS1-infected Egfrfl/fl mice and a marked reduction in infected EgfrΔepi mice (figure 3B). There was no difference in colonisation between Egfrfl/fl and EgfrΔepi mice (see online supplementary figure S2B). Egfrfl/fl mice infected with H. pylori showed more MPO-positive cells than uninfected animals (figure 3C) and a significant reduction of MPO staining was observed in H. pylori-infected EgfrΔepi mice (figure 3C). Concomitantly, levels of Cxcl1 and Cxcl2 mRNA in isolated GECs from infected animals were reduced in EgfrΔepi mice compared with Egfrfl/fl mice (figure 3D). DNA damage in gastric tissues of infected mice was also significantly reduced in EgfrΔepi mice compared with Egfrfl/fl mice (figure 3E).

Figure 3

Effect of Egfr deletion on pEGFR, MPO, pH2AFX and chemokine expression with Helicobacter pylori PMSS1 infection for 4 weeks. (A) Western blot analysis for epidermal growth factor receptor (EGFR) in gastric epithelial cells (GECs) isolated from naive Egfrfl/fl and EgfrΔepi mice. (B) Immunohistochemistry for pEGFR in gastric tissues of infected Egfrfl/fl and EgfrΔepi mice. (C) Quantification of MPO-positive cells assessed by immunohistochemistry. Scale bar=100 µm. (D) Cxcl1 (upper panel) and Cxcl2 (lower panel) mRNA expression assessed by real-time PCR in isolated GECs. (E) Number of pH2AFX-positive nuclei by immunohistochemistry. *p<0.05, **p<0.01 vs uninfected mice; §p<0.05, §§p<0.01 vs PMSS1 infected on control diet.

Gefitinib decreases development of neoplastic lesions and inflammation in INS-GAS mice infected with H. pylori

To determine the effect of EGFR on H. pylori-induced gastric carcinogenesis, hypergastrinemic INS-GAS mice that can develop dysplasia and carcinoma with H. pylori infection15 were given the control or gefitinib-supplemented diet during a 2-month infection with H. pylori strain PMSS1. Colonisation was increased in INS-GAS mice on gefitinib (see online supplementary figure S2C). As shown in figure 4A, on the control diet, 79% developed dysplasia, and 41% developed intramucosal carcinoma; when mice were given the gefitinib diet, there were significant decreases in development of dysplasia (to 55%) and carcinoma (to 3%). Extent of dysplasia and carcinoma as a percentage of the gastric mucosa was increased with infection and significantly attenuated by gefitinib (figure 4B). The modest effects of gefitinib on uninfected mice in figure 4A, B were not statistically significant. Figure 4C depicts H&E staining of the gastric mucosa of an infected INS-GAS mouse on the control diet, showing intramucosal carcinoma, with irregular and angulated glands, infiltration of the lamina propria by tumour cells and desmoplastic stroma; in contrast, the gefitinib-treated mouse shown exhibited only low-grade dysplasia. When INS-GAS mice on the gefitinib diet were compared with those on the control diet, there were decreases in gastric tissue MPO-positive cells (figure 4D), Cxcl1 and Cxcl2 mRNA levels (figure 4E) and CXCL1 protein levels (figure 4F) in isolated GECs, and pH2AFX-positive cells in the GECs in situ (figure 4G). In infected INS-GAS mice there was an increase in infiltrating T cells, detected by CD3 immunostaining, which was decreased by gefitinib (see online supplementary figure S3).

Figure 4

Friend Virus B Type NIH (FVB/N) insulin-gastrin (INS-GAS) mice infected for 8 weeks with PMSS1 and maintained on control or gefitinib-supplemented diet. (A) Frequency of diagnoses in INS-GAS mice. The adjacent numbers correspond to the number of mice in each section of the bar. ¶¶¶p=0.0003 for PMSS1 infected on gefitinib vs control diet, comparing intramucosal carcinoma frequency; #p=0.044, comparing frequency of carcinoma+low-grade dysplasia (red+blue bars). LGD, low-grade dysplasia. (B) Extent of dysplasia and carcinoma quantified as a percentage of the total area of the gastric mucosa in H&E-stained tissues. (C) H&E-stained stomach tissues, showing intramucosal carcinoma (indicated by the circle) on the control diet, vs low-grade dysplasia with gefitinib. (D) Quantification of gastric tissue MPO-positive cells by immunohistochemistry. (E) Cxcl1 and Cxcl2 mRNA expression by real-time PCR on isolated gastric epithelial cells (GECs). (F) CXCL1 protein levels quantified by ELISA from GEC lysates. (G) Number of pH2AFX-positive nuclei in gastric tissues. In B–G, *p<0.05, **p<0.01 vs uninfected mice; §p<0.05, §§p<0.01 vs PMSS1 infected on control diet.

Gefitinib treatment reduces DNA damage and gastric carcinogenesis in Mongolian gerbils infected with H. pylori

To establish the role of EGFR activation in a model of H. pylori-induced cancer, we used Mongolian gerbils infected with H. pylori 7.13, known to induce gastric carcinoma in these animals.15 ,24 There was increased H. pylori colonisation in gefitinib-treated gerbils after 8, but not 12 weeks of infection (see online supplementary figure S2D–E). Figure 5A illustrates representative H&E staining of the gastric antrum of infected gerbils. Gerbils on the control diet exhibited invasive adenocarcinoma, characterised by irregular and angulated glands infiltrating the submucosa. In contrast, the depicted example from a gerbil on the gefitinib diet shows only low-grade dysplasia characterised by elongated, branched, irregular glands without infiltration of the submucosa. On the control diet, infection resulted in invasive gastric adenocarcinoma in 60% and 87.5% of gerbils at 8 and 12 weeks postinfection, respectively (figure 5B). Administration of gefitinib resulted in a significant decrease in gastric cancer, to 31.25% and 62.5%, at these time points (figure 5B).

Figure 5

Mongolian gerbils infected with Helicobacter pylori strain 7.13 for 8 or 12 weeks and maintained on control or gefitinib-supplemented diet. (A) H&E staining, showing invasive adenocarcinoma with control diet and low-grade dysplasia with gefitinib diet. Scale bar=100 µm. (B) Frequency of invasive adenocarcinoma. (C) Quantification of MPO-positive cells assessed by immunohistochemistry in gastric tissues. (D) Cxcl1 and Il1b mRNA expression in gastric tissues by real-time PCR. (E and F) pH2AFX quantification, number of positive nuclei per high power field (HPF), and representative staining in gastric tissues. Scale bar=100 µm in upper panel. Scale bar=50 µm in lower panel. In B–F, gerbils were infected for 12 weeks. Scale bar=100 µm. **p<0.01, ***p<0.001 vs uninfected mice; §p<0.05, §§p<0.01 vs 7.13 infected on control diet.

Infection of gerbils by H. pylori was also associated with increased MPO-positive cells in the gastric mucosa (figure 5C), which was decreased in animals treated with gefitinib (figure 5C). In parallel, a significant increase in mRNA expression of the pro-inflammatory markers Cxcl1 and Il1b was observed in GECs from infected gerbils on the control diet, and these mRNA levels were significantly reduced by gefitinib treatment (figure 5D). Moreover, there was a significant decrease in pH2AFX-positive cells in infected gerbils on the gefitinib versus control diet (figure 5E, F).

To evaluate gefitinib after H. pylori infection was established, we started treatment 2 weeks postinfection (see online supplementary figure S4A). Gefitinib treatment either before or after infection resulted in the same reduction in gastric adenocarcinoma (see online supplementary figure S4B) and a similar decrease in MPO-positive cells (see online supplementary figure S4C).

DNA damage and chemokine production in GECs is mediated by EGFR-dependent ERK/AP-1 activation

To recapitulate effects of EGFR inhibition in vitro, we infected ImSt cells with H. pylori strains PMSS1 or 7.13 and showed that gefitinib efficiently inhibits EGFR activation (see online supplementary figure S5A). mRNA levels of Cxcl1 and Cxcl2 were induced in response to both H. pylori strains (figure 6A,B), and gefitinib significantly decreased expression of these genes (figure 6A, B). We also assessed CXCL1 secretion by ELISA, and found that levels of this chemokine were upregulated in response to H. pylori and markedly inhibited when cells were treated with gefitinib (figure 6C). When DNA damage was assessed by flow cytometry, there was an increased percentage of pH2AFX-positive cells after infection with either H. pylori strain, and a significant decrease in the cells treated with gefitinib (figure 6D).

Figure 6

Effect of gefitinib on chemokine and pH2AFX levels in mouse gastric epithelial cells (ImSt). (A) Cxcl1 and (B) Cxcl2 mRNA expression by real-time PCR in cells cocultured with Helicobacter pylori strains PMSS1 or 7.13 for 6 hours. (C) CXCL1 levels quantified by ELISA in supernatants of ImSt cells cocultured with H. pylori for 24 hours. (D) Percentage of pH2AFX-positive cells by flow cytometry after infection for 24 hours. *p<0.05, **p<0.01, ***p<0.001 vs uninfected cells; §p<0.05, §§p<0.01 vs H. pylori-infected cells.

In the human cell line AGS, expression of the gene CXCL8, also known as IL-8, the human homologue of the murine Cxcl1, was induced in AGS cells infected with H. pylori PMSS1 or 7.13 (figure 7A). We confirmed the reduction of pEGFR by gefitinib in AGS cells (see online supplementary figure S5B). Also, EGFR activation and CXCL8 expression occurred on infection with lower bacterial densities (see online supplementary figure S6E–F). There was significant inhibition of CXCL8 mRNA expression in infected cells treated with gefitinib (figure 7A; see online supplementary figure S6F), as well as with the MAPK1/3 (ERK1/2) inhibitor, PD98059 or the AP-1 inhibitor, SR11302 (figure 7A).

Figure 7

Effect of epidermal growth factor receptor (EGFR) inhibition on chemokine production and mitogen-activated protein kinase (MAPK)/activator protein 1 (AP-1) pathway activation. (A) CXCL8 mRNA expression in AGS cells pretreated with gefitinib (G), the MEK inhibitor PD98059 (P) or the AP-1 inhibitor SR11302 (S) and infected with Helicobacter pylori strains PMSS1 or 7.13 for 2 hours. ***p<0.001 vs uninfected cells; §p<0.05, §§p<0.01, §§§p<0.001 vs H. pylori-infected cells. (B) Western blot analysis for JUN and fibrillarin (FBL) in nuclear lysates of AGS cells pretreated with inhibitors and infected with H. pylori for 1 hour. (C) Western blot analysis for pMAPK1/3 and pRELA in AGS cells pretreated with gefitinib and infected with H. pylori for 1 hour. (D) Immunohistochemistry for pMAPK1/3 and pJUN in gastric tissues from C57BL/6 mice infected for 4 weeks. (E) Immunohistochemistry for pMAPK1/3, pJUN and pEGFR in gastric tissues from insulin-gastrin mice infected for 8 weeks. Scale bar=100 µm in pMAPK1/3 and pEGFR photomicrographs. Scale bar=50 µm in pJUN photomicrographs.

Consistent with these results, we found that infection of AGS cells with H. pylori-induced Jun proto-oncogene (JUN) translocation from the cytoplasm to the nucleus, indicating activation of AP-1 (figure 7B); nuclear translocation of JUN was markedly reduced in cells with EGFR or MAPK1/3 inhibition (figure 7B); as a positive control, SR11302 blocked JUN activation (figure 7B). Cytoplasmic JUN was not reduced by the inhibitors, when assessed as a control (see online supplementary figure S6G). There was induction of nuclear factor-kappaB (NF-κB) activation, by western blot analysis of phosphorylated RELA (NF-kappaB p65 subunit) with infection by H. pylori PMSS1 or 7.13, but there was no decrease in NF-κB activation with EGFR inhibition (figure 7C). NF-κB activation was not altered in gefitinib-treated cells after infection with H. pylori (see online supplementary figure S6H). This indicates that activation of JUN/AP-1, but not NF-κB, requires signalling involving EGFR and MAPK1/3. In addition, gefitinib reduced MAPK1/3 phosphorylation in response to infection with H. pylori PMSS1 or 7.13 in AGS cells (figure 7C), demonstrating that MAPK1/3 activation by H. pylori is mediated through EGFR.

Gefitinib may target other receptor kinases, like RIPK2.25 H. pylori induced significant RIPK2 phosphorylation in AGS cells (see online supplementary figure S6A).26 However, gefitinib did not affect pRIPK2 in infected cells (see online supplementary figure S6A–B).

CagA and VacA have been reported to modulate EGFR activation.27 ,28 However, similar levels of EGFR activation were induced by wild-type, or cagA and vacA isogenic mutants of H. pylori (see online supplementary figure S6C–D).

MAPK1/3 phosphorylation was increased in vivo in infected C57BL/6 and INS-GAS mice and reduced by gefitinib, when assessed by western blot analysis and densitometry (see online supplementary figure S7A–D). By immunohistochemistry, pMAPK1/3 levels were attenuated in gastric tissues of H. pylori-infected mice on gefitinib (figure 7D, E), along with pJUN levels (figure 7D, E; see online supplementary figure S8A–B). Reduced pEGFR levels with gefitinib were confirmed in INS-GAS mice by immunohistochemistry (figure 7E).

MAPK1/3-JUN signalling pathway activation and chemokine production are decreased in cultures of GEC isolated from EgfrΔepi mice

Using immunofluorescence, we confirmed that H. pylori stimulation ex vivo induced a marked increase in pEGFR in two-dimensional monolayers of primary GEC isolated from Egfrfl/fl mice, and this activation was not observed in monolayers from EgfrΔepi mice (figure 8A). H. pylori also induced marked MAPK1/3 and JUN activation in cells from Egfrfl/fl mice, which was prevented in the cells from EgfrΔepi mice (figure 8A and see online supplementary figure S9). Cxcl1 and Cxcl2 expression was significantly increased in Egfrfl/fl GEC after H. pylori infection, which did not occur in GEC from EgfrΔepi mice (figure 8B).

Figure 8

Effect of Egfr deletion on pEGFR, pMAPK1/3, pJUN and chemokine expression induced by Helicobacter pylori in gastric epithelial cell (GEC) monolayers. (A) pEGFR, pMAPK1/3 and pJUN immunofluorescent staining of GEC monolayers isolated from Egfrfl/fl and EgfrΔepi mice, and infected with PMSS1 for 1 hour at multiplicity of infection 10. pEGFR, pMAPK1/3 and pJUN in green, actin in red and nuclei in blue, imaged by confocal microscopy. Scale bar=20 µm. (B) Cxcl1 and Cxcl2 mRNA levels by real-time PCR in gastric glands isolated from Egfrfl/fl and EgfrΔepi mice, and infected for 24 hours with PMSS1. **p<0.01, ***p<0.001 vs uninfected cells; §§p<0.01, §§§p<0.001 vs H. pylori-infected cells.

EGFR is still activated after H. pylori eradication in humans

EGFR activation was assessed by immunohistochemistry in gastric tissues from a cohort of H. pylori-infected Colombian patients preantibiotic and postantibiotic eradication therapy. Variable levels of pEGFR staining were observed in pretreatment biopsies (figure 9A). EGFR activation was not attenuated after H. pylori eradication (figure 9A). Epithelial and immune cell staining for pEGFR was observed in pretreatment biopsies (figure 9B, left); there was a loss of inflammation post-treatment, but pEGFR was still present in epithelial cells (figure 9B, right).

Figure 9

(A) Effect of Helicobacter pylori eradication on pEGFR in human gastric biopsies. All individuals were confirmed to be H. pylori-negative on follow-up. (B) Immunohistochemistry for pEGFR in pre-eradication and posteradication biopsies. (C) Frequency of diagnoses in insulin-gastrin mice. The number of mice in each section of the bar is noted. p<0.05, ¶¶p<0.01, for carcinoma vs PMSS1 only; #p<0.05, for carcinoma vs PMSS1+antibiotics (ABX). When dysplasia+carcinoma is considered, for gefitinib+ABX, p<0.05 vs ABX only, and p<0.01 vs PMSS1 only. (D) Representative H&E staining of gastric tissues from infected mice in the different treatment groups. PMSS1 only: intramucosal carcinoma; gefitinib or ABX: dysplastic glands; gefitinib+ABX: non-dysplastic tissue. (E) Quantification of pEGFR staining. *p<0.05 vs PMSS1 only. (F) Representative pEGFR immunohistochemistry. Scale bar=100 µm.

Gefitinib treatment reduces development of neoplastic lesions even after H. pylori eradication

To determine the effect of gefitinib after H. pylori eradication, INS-GAS mice were infected for 4 weeks, and then antibiotics were administered for 1 week, followed by 3 weeks of gefitinib. All mice were verified to have elimination of H. pylori infection by antibiotics. As shown in figure 9C, intramucosal carcinoma was significantly reduced by gefitinib treatment during the last 3 weeks of infection; the incidence of intramucosal carcinoma went from 50% in the untreated group to 8.3% in gefitinib-treated animals. Antibiotics reduced the incidence of carcinoma to 23%, but this did not reach statistical significance. Gefitinib treatment after antibiotics showed a further benefit; none of the mice in this group developed intramucosal carcinoma, and 80% showed no signs of dysplasia, a significant reduction versus both untreated and antibiotic-treated animals. Figure 9D illustrates representative H&E staining of the gastric mucosa of infected mice in each group.

pEGFR was evaluated by immunohistochemistry and the intensity of staining was quantified (figure 9E). Abundant pEGFR staining in INS-GAS mice infected with H. pylori was detected and was reduced in mice treated with gefitinib during the last 2 weeks of infection. Antibiotic treatment did not reduce pEGFR staining compared with the untreated group. Gefitinib treatment after antibiotics also led to less pEGFR compared with the untreated group. Representative images of pEGFR staining are shown in figure 9F.


H. pylori infection is accompanied by the potent induction of both innate and adaptive mucosal responses, which play a fundamental role in development of the chronic active inflammation that characterises the disease phenotype.29 ,30 The dysregulated synthesis of immune effectors, including reactive oxygen or nitrogen species, also has deleterious and procarcinogenic effects on the gastric epithelium.31 ,32 In the present report, we have implicated activation of epithelial EGFR in gastric inflammation, chemokine expression, DNA damage and development of cancerous lesions. Our findings implicate MAPK1/3-mediated and AP-1-mediated chemokine production by the gastric epithelium that leads to recruitment of immune cells, and thus induction and maintenance of inflammation during H. pylori infection. Our data showing that EGFR inhibition by gefitinib either as pretreatment or when is administered after the infection is established led to decreased inflammatory cells in the infected mucosa, phosphorylation of H2AFX and gastric dysplasia and carcinoma, have been obtained in both mice and gerbils. The inhibitory effect of gefitinib on H. pylori-induced MAPK1/3 phosphorylation and chemokine expression was recapitulated in human GECs, emphasising that this mechanism is not species-dependent. EGFR inhibitors are in current use in patients with breast and lung cancer; in the case of gastric cancer the efficacy of these agents appears to be more limited, but is under continued investigation.33 ,34 Importantly, our studies were not designed to investigate gefitinib as an agent to treat gastric cancer, but rather, suggest that it may prove useful to prevent development of neoplastic lesions by blocking inflammation-related DNA damage.

We focused our in vivo work on CXCL1 in mice and gerbils because it is the rodent equivalent of human CXCL8, a chemoattractant for neutrophils and T cells.35 It is established that expression of CXCL8 mRNA and production of CXCL8 is enhanced in gastric tissues of H. pylori-infected versus uninfected individuals.36 In H. pylori-positive patients, serum CXCL8 levels were further elevated in those with gastric cancer,37 and CXCL8 has been implicated in the development, severity and spread of gastric carcinoma.38 ,39 In the present report, CXCL1 was associated with dysplasia/carcinoma in mice and gerbils infected with H. pylori, and gefitinib treatment reduced both parameters. Similarly, studies in hepatocellular carcinoma and in non-small cell lung cancer cell lines have demonstrated that EGFR activation leads to increased expression of CXCL8, and this supports the tumour microenvironment.40 ,41 Patients with hepatocellular carcinoma have increased levels of serum CXCL8, and this correlates with larger tumour size, more advanced stage and worse survival.42 In a mouse model of carcinogen-induced lung tumourigenesis, gefitinib reduced tumour multiplicity and burden, and this was associated with decreased expression of pro-inflammatory genes in a microarray analysis.13

Our results indicate that EGFR signalling in epithelial cells is critical for inflammation after H. pylori infection. GECs isolated from infected mice and gerbils treated with gefitinib and mice with epithelial-specific deletion of Egfr express less chemokines and cytokines, associated with decreased MPO-positive cells and DNA damage. It should be noted that myeloid cells are also a source of CXCL8 in H. pylori infection.43 Because we reported that myeloid-specific knockout of EGFR also leads to a diminution of chemokine and cytokine expression in the gastric tissues of H. pylori-infected mice,17 gefitinib may also reduce EGFR signalling in gastric macrophages, thus amplifying the protective effect of EGFR inhibition in gastric carcinogenesis.

Eradication therapy for H. pylori has been proposed as a strategy to reduce gastric cancer incidence.2 However, this approach may not be effective in patients with premalignant lesions. Our data indicate that pEGFR is still abundant in mice as well in humans after H. pylori has been eradicated. Thus, persistent activation of EGFR even after the bacterium has been eliminated can play an important role in the carcinogenic process. Indeed, H. pylori eradication in INS-GAS mice did not completely prevent development of carcinoma. However, double treatment with antibiotics plus gefitinib completely prevented development of intramucosal carcinoma.

The p38/NF-κB pathway44 ,45 and/or MAPK1/3 signals46 ,47 have been linked to H. pylori-induced CXCL8 transcription. It has been suggested that both AP-1 and NF-κB are required for maximal induction of CXCL8 mRNA expression.48 In our report, inhibitors of EGFR, MAPK1/3 or AP-1 decreased CXCL8 mRNA expression in infected human AGS cells. Furthermore, gefitinib also blocked MAPK1/3 and AP-1 activation, and MAPK1/3 inhibition decreased JUN nuclear translocation, demonstrating that EGFR signals through a MAPK1/3-AP-1 transduction pathway to stimulate CXCL8 gene expression.

Supporting these data, it has been demonstrated that the type IV secretion system pilus component, CagL, activates EGFR and ERBB3, as well as focal adhesion kinase and SRC (Src Proto-Oncogene), by mimicking the function of fibronectin.49 Additionally, CagL can lead to ADAM-17-mediated release of EGF, resulting in MAPK1/3 activation.50 In addition, induction of human β-defensin-3 or matrix metalloproteinase-10 expression in H. pylori-infected GECs requires an EGFR/MAPK1/3 transduction pathway.28 ,51 Notably, chronic in vitro infection with H. pylori for 24 hours has been reported to suppress the ability of EGF ligand to activate EGFR.28 However, we have shown herein that experimental H. pylori infection in vivo leads to increased pEGFR beginning at 1 day postinfection and continuing out to chronic time points (4 and 8 weeks), and that humans exhibit epithelial pEGFR staining after many years of infection that can persist even after H. pylori eradication. Growth factors like EGF are capable of triggering NF-κB activation in breast, prostate and lung cancer cell lines through EGFR phosphorylation.52 However, we did not observe an effect of gefitinib on NF-κB activation. This is not surprising, considering that H. pylori possesses a variety of virulence factors that can activate NF-κB through multiple cellular receptors, including toll-like receptors and NOD1.53 ,54 Importantly, we recapitulated the concordance between EGFR and MAPK activation in vivo: C57BL/6 and INS-GAS mice infected with H. pylori had concomitantly increased levels of pEGFR and pMAPK1/3, which were inhibited with gefitinib treatment.

We also showed that gefitinib dampens the innate immune response of GECs and consequently their ability to recruit PMNs, and inhibits H. pylori-induced DNA damage in GECs. This effect further supports the potential benefits of gefitinib. Our studies provide new insight into the mechanism by which H. pylori mediates gastric carcinogenesis, and we propose that gefitinib could represent an alternative therapeutic approach to limit gastric cancer development by decreasing inflammation and by inhibiting DNA damage in GECs. Related to this concept, H. pylori strains isolated from patients in regions with low risk of developing gastric cancer induce less CXCL8 and DNA damage than those from high-risk regions.21 ,55 ,56 In addition, the interaction of the host and H. pylori genetics, according to their phylogeographic variations, is a critical feature for the development of gastric diseases, such that loss of coevolution increases cancer risk.55 ,57 Therefore, determination of EGFR activation in individuals with different genetic origins and/or infected with H. pylori of varying phylogeographic origin is a future direction for our studies. It could prove important to consider human and bacterial genetic factors and specific host signalling responses in developing personalised chemoprevention strategies, such as utilisation of EGFR inhibition in populations at high risk for H. pylori-induced gastric cancer.


Whole slide imaging and quantification of immunostaining were performed in the Digital Histology Shared Resource at Vanderbilt University Medical Center ( Immunofluorescence confocal imaging was performed in the Vanderbilt Cell Imaging Shared Resource.



  • Correction notice This article has been corrected since it published Online First. Figure 9 has been amended.

  • Contributors JCS performed the experiments, analysed the data and drafted the manuscript. MA, TGV, GS, JR-G, AGD and LEW performed experiments. MBP analysed and scored all of the tissue sections. DPB designed and generated epithelial-specific knockout mice. RMP provided funding and commented on the manuscript. APG designed experiments and revised the manuscript. KTW designed the experiments, analysed the data, supervised the studies, obtained funding and also wrote the paper.

  • Funding This study was funded by National Institutes of Health (NIH) grants R01DK053620, R01AT004821, R01CA190612 and P01CA028842 (to KTW), and P01CA116087 (to KTW and RMP), a Department of Veterans Affairs Merit Review grant I01BX001453 (to KTW), the Thomas F. Frist Sr. Endowment (to KTW), the Vanderbilt Center for Mucosal Inflammation and Cancer (to KTW and APG), the Vanderbilt Digestive Disease Research Center, funded by NIH grant P30DK058404 and the Vanderbilt Ingram Cancer Center, funded by NIH grant P30CA068485.

  • Competing interests None declared.

  • Ethics approval All animal study protocols were approved by the Vanderbilt University Institutional Animal Care and Use Committee.

  • Provenance and peer review Not commissioned, externally peer reviewed.