• cyclooxygenase-2;
  • prostaglandin-endoperoxide synthase 2;
  • trefoil factor 1;
  • adenoma;
  • celecoxib;
  • carcinogenesis


  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Cyclooxygenase-2 (Cox-2) expression is a marker of reduced survival in gastric cancer patients, and inhibition of Cox-2 suppresses gastrointestinal carcinogenesis in experimental animal models. To investigate the role of Cox-2 in gastric carcinogenesis in vivo, we utilized trefoil factor 1 (Tff1) deficient mice, which model the neoplastic process of the stomach by developing gastric adenomas with full penetrance. These tumors express Cox-2 protein and mRNA, and we have now investigated the effects of genetic deletion of the mouse Cox-2 gene [also known as prostaglandin-endoperoxide synthase 2 (Ptgs2)] and a Cox-2 selective drug celecoxib. Our results show that genetic deletion of Cox-2 in the Tff1 deleted background resulted in reduced adenoma size and ulceration with a chronic inflammatory reaction at the site of the adenoma. To characterize the effect of Cox-2 inhibition in more detail, mice that had already developed an adenoma were fed with celecoxib for 8–14 weeks, which resulted in disruption of the adenoma that ranged from superficial erosion to deep ulcerated destruction accompanied with chronic inflammation. Importantly, mice fed with celecoxib for 16 weeks, followed by control food for 9 weeks, redeveloped a complete adenoma with no detectable inflammatory process. Finally, we determined the identity of the Cox-2 expressing cells and found them to be fibroblasts. Our results show that inhibition of Cox-2 is sufficient to reversibly disrupt gastric adenomas in mice.

Cyclooxygenase-2 (Cox-2) enzyme is critical for the conversion of arachidonic acid to prostanoid precursors, which are converted to biologically active prostanoids by downstream synthases.1 Prostaglandin E2 (PGE2) is the major prostanoid produced by epithelial cells and by neoplastic lesions derived from these cells and it is thought to play an important role in carcinogenesis via modulation of angiogenesis, invasion and metastasis.2–4 Gastric cancer and its precursors often express an elevated level of Cox-2,2, 5, 6 and overexpression of Cox-2 is an independent marker of reduced survival in gastric adenocarcinoma patients.7, 8

In mice, there exist only a few reports on the role of Cox-2 in gastric tumorigenesis utilizing genetic manipulation, and most of them model a non-neoplastic phenotype, such as hyperplasia or hamartoma.9–11 However, one model showed dysplastic tumors at the age of 20 weeks when Cox-2 and microsomal PGE2 synthase-1 (mPGES-1) were overexpressed in combination with Wnt1.12 In another model deletion of trefoil factor 1 (Tff1) lead to dysplastic adenoma formation with full penetrance in the pyloric region of the stomach.13 Tff1 protein is expressed in human and mouse gastric epithelial cells, where it protects the gastrointestinal mucosa,14, 15 and its expression is frequently lost in human gastric adenocarcinomas.16, 17 We have previously shown that Cox-2 is expressed in gastric adenomas of the Tff1 deficient mice, as detected by in situ hybridization and by immunohistochemistry.18 Treatment with a selective Cox-2 inhibitor, which started before the development of the adenoma, totally prevented adenoma formation in these mice. This effect of the drug was adenoma specific, because gastric tissues proximal to the adenoma or intestinal tissues distal to the adenoma did not suffer from side effects including ulcerations or perforations.18

To more specifically characterize the role of Cox-2 in gastric carcinogenesis, we have now investigated the effect of genetic deletion of mouse Cox-2 gene (prostaglandin-endoperoxide synthase 2, Ptgs2) on adenoma formation in the Tff1 deficient mice. The effect of the selective Cox-2 inhibitor, celecoxib, was further characterized on already established adenomas.

Material and Methods

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information


The generation of Tff1 deficient mice (mixed genetic background C57BL/6J/129/SVJ) has been described previously.13 The Cox-2 knockout mice19 were obtained from the Jackson Laboratory (Bar Harbor, Maine) and were kept on a mixed genetic background (C57BL/6J/129xCD1). Tff1 −/− female mice were crossed with male Cox-2 −/− mice to generate heterozygous mice. These mice were further crossed to produce double knockout mice. From crossings of mice heterozygous for Tff1 and Cox-2 328 mice were screened for their genotype and seven double knockouts (Tff1 −/−, Cox-2 −/−) were found. Littermates from these crossings were used as controls. The lower than expected number of double knockout mice is most likely due to a reduced ratio at birth combined with neonatal mortality within 48 hrs after birth often caused by patent ductus arteriosus as has been reported for Cox-2 deficient mice.20 All animals were housed in the animal facility at the Biomedicum Helsinki and had free access to food and water throughout the studies All animal studies have been approved by the local ethics committee (STU729A).


DNA was isolated from the tail with the NucleoSpin Tissue kit (Machery-Nagel, Düren, Germany). Primers used for genotyping are specified in Supporting Information Table 1. For Cox-2 genotyping we applied a modified protocol kindly provided by Prof. S. K. Dey (University of Vanderbilt Medical Center, Nashville, Tennessee). All primers were purchased from Sigma (Sigma–Aldrich, Helsinki, Finland). Amplification protocol: 95°C for 1 min, 60°C for 1 min, and 72°C for 2 min for 40 cycles. The amplified fragments were visualized by agarose gel electrophoresis and ethidium bromide staining.

Table 1. Histology of gastric tissues in Tff1 −/− mice according to Cox-2 genotype and age
inline image

Celecoxib treatment

Celecoxib (1,600 ppm; kindly provided by Pfizer, Groton, CT) was mixed with chow powder (Harlan Tekland, 2018 Global Rodent Breeding), from which a dough was prepared, and ready-to-use 50 g portions were stored at −20°C until use. Control food was prepared in the same way excluding the drug. The mice were fed with celecoxib (1,600 ppm) or the control food for 2–14 weeks beginning at the age of 2 months at which time the mice have a fully developed adenoma with full penetrance.13, 18 A subgroup of mice were fed first with celecoxib for 16 weeks after which the drug was either continued or replaced by the control food for 9 weeks. During all experiments the food was replaced three times a week.

Tissue handling

Immediately after sacrifice, the stomach including the proximal part of the small intestine was filled with 4% paraformaldehyde and fixed for 2 min. Subsequently the stomach was opened, cleaned with PBS and fixed in 2% paraformaldehyde over night at 4°C. The tissues were embedded in paraffin and for histological examination 4 μm thick tissue sections were stained with H&E. Alternatively the unfixed tissue was frozen in Tissue Tek O.C.T. compound (Sakura Finetik Europe, Zoeterwoude, Holland) and stored at −70°C, and 6 μm thick sections were stained with toluidine blue for histological analysis.


The deparaffinized sections were stained with the following antibodies: Ki-67 (1:200, 1 h, RT Novocastra, Immuno Diagnostics, Hämeenlinna, Finland), Cox-2 (1:1200, over night incubation at 4°C, Cayman Chemical, Tallinn, Estonia, no. 160126), macrophage (F4/80, 1:100–1:200, 1 h, RT, Serotec, Kidlington, UK and eBioscience, Hatfiel, UK), B cells (B220, 1:1000, over night incubation at 4°C, Southern Biotech, Birmingham, AL), plasma cells (anti-mouse Ig-HRP, 1:100, 30 min RT, Dako, Glostrup, Denmark), MECA-32 (panendothelial cell antigen, 1:50, over night incubation at 4°C, BioLegend, Uithoorn, The Netherlands), CD45 (leukocyte common antigen, 1:10 over night incubation at 4°C, BD Biosciences, Erembodegem, Belgium), E-cadherin (1:1000, overnight incubation at 4°C, BD Biosciences) and apoptosis (cleaved caspase-3, 1:100, over night incubation at 4°C, Cell Signaling, Danvers, MA). The signal was developed directly (plasma cells), after incubation with anti-rabbit or anti-mouse-HRP (Ki-67, E-cadherin), or after enhancement with a biotinylated secondary antibody (anti-rat-biotin 1:1000 or anti-rabbit-biotin 1:200, 30 min; Vector labs) followed by streptavidin-HRP (1:500, 30 min; Dako). The antibodies for macrophages (F4/80, 1:100; CD206, 1:300, 1 h, RT, biotin-streptavidin enhancement), CD45 (1:500, 1 h, RT), Cox-2 (1:1000, 1 h RT) and T cells (CD3e, 1:200, 1 h, RT, eBioscience, biotin-streptavidin enhancement) were used on frozen sections.

Tissue analysis

Phenotype of the adenomas was scored as a superficial erosion whenever clear mucosal injury with fibrinoid debris was evident on the luminal side of the adenoma. Deep injuries, i.e., ulcers, extended to the muscularis mucosa or beyond. The level of inflammation was evaluated at H&E stained sections and grouped into level 0 to 3, where level 0 represented no additional inflammatory cells when compared to the controls, 1 low level of eosinophilic and/or lymphocytic infiltration, 2 moderate lympho-plasmacytic infiltration and 3 strong lympho-plasmacytic infiltration often combined with germinal center formation. To measure adenoma height pictures of H&E stained sections were taken (20×) and a grid was applied to the picture. The relative height of the adenoma was measured at the pyloric region from the middle of the adenoma or eroded/ulcerated adenoma and the percentage of adenoma reduction was calculated. To measure microvessel density, the MECA-32 stained sections were viewed with a light microscope and mucosal vessels were counted from four high power (400×) fields per slide. B cells were evaluated with the same method. The index for macrophages was evaluated by counting F4/80- or CD206 labeled cells from two high power fields (400×) from the undamaged mucosa. T cells were counted with the same method as macrophages. To measure proliferating cells the percentage of Ki-67 positive cells of 1,000 epithelial cells was determined in 2–3 fields in the adenoma region. Apoptosis was determined by counting all cells staining for cleaved caspase-3 in the mucosal area of the adenoma. A detailed overview of the analyzed number of mice and high power fields is summarized in Supporting Information Table 2.

Table 2. Histology of gastric tissues in Tff1 −/− mice according to treatment time with the Cox-2 selective inhibitor celecoxib or control food
inline image

Detection of senescent cells

Senescent cells were detected with the senescence detection kit (BioVision, Mountain View, CA) that histochemically detects senescence-associated expression of β-galactosidase according to the manufacturer's instructions. Briefly, frozen sections were fixed 10 min with fixative solution, and incubated with staining solution mix for 22 h at 37°C. Sections were counterstained for 5 min with nuclear fast red solution (Sigma). Senescent cells displayed a blue color.

Protein analysis

Proteins were extracted with a Protein/RNA extraction kit (Machery Nagel) and subjected to Western blot analysis. As primary antibodies rabbit anti-Cox-2 (1:1000, Cayman), or goat anti-β-actin (1:1000, both Santa Cruz) were used. Secondary antibodies were goat anti-rabbit (1:2000; Dako) or donkey anti-goat (1:2000; Santa Cruz).


For double fluorescence in frozen sections, CD45 or F4/80 (macrophages) were visualized with a goat anti rat-Alexa Fluor 568 antibody (1:500, 30 min, RT) and vimentin (1:200, 1 h, RT; Santa Cruz) with a donkey anti goat-Alexa Fluor 568 (1:500, 30 min, RT). Cox-2 was visualized with a goat anti rabbit-Alexa Fluor 488 antibody (1:400, 30 min, RT; all Molecular Probes, Invitrogen, Carlsbad, CA). To double stain deparaffinated sections, the sections were treated with Cox-2 (1:800) antibody, goat anti rabbit-Alexa Fluor 488 (1:400, 30 min, RT) and mouse anti-Sma-Cy3 (1:200, 45 min; Sigma). Nuclei were visualized with Hoechst staining (1:200, 10 min, Sigma).

Statistical analysis

The data were analysed by unpaired t-test. A p-value of ≤ 0.05 was considered statistically significant (SPSS 16.0) and all data are shown as means ± SD.


  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Effect of Cox-2 gene deletion in Tff1 deficient mice

To study the role of Cox-2 in the Tff1 deficient mice, we deleted the Cox-2 gene and investigated the morphology of the adenoma site (the pylorus) at the age of 2 months and 4 to 6 months. Deletion of the Cox-2 gene in the Tff1 knockout background resulted in ulceration in all mice (n = 6) at the site of the adenoma with moderate to strong chronic inflammation (Figs. 1a–1c and Table 1). The inflammatory score for the 2 months old mice was 2.3 (range 2–3) and in the 4 to 6 months old group it was 3 for each mouse. The inflammatory infiltrate consisted mainly of lymphocytes and plasma cells, and in two mice also secondary lymphatic follicles with germinal centers were formed (Figs. 1b and 1c). No gastric phenotype has previously been detected in Cox-2 deficient mice or in Tff1 +/− mice.13, 18, 19 In accord with these data, all Tff1 +/− mice (n = 7) had a normal histology of the stomach regardless of their deletion status of the Cox-2 gene (Fig. 1d; three Cox-2 −/− and four Cox-2 +/− mice). Importantly, Tff1 −/− mice that had only one intact Cox-2 allele showed only hyperplasia in the pyloric region of the glandular stomach at the age of 1 month (Fig 1e; n = 2), while intact noninflamed adenomas were present at the age of 2 months or older mice (Fig. 1f and Table 1; n = 8). The height of the adenoma from Tff1 −/− Cox-2 −/− mice was reduced 100% when compared to littermates that were heterozygous for Cox-2 (Figs. 1g–1i; height in fields: 8.8 ± 1.3 Tff1 −/− Cox-2 +/− mice, 0.0 ± 0.0 Tff1 −/− Cox-2 −/− mice; p < 0.005). These data show that Cox-2 is necessary for neoplastic growth in Tff1 deficient mouse gastric tissues, and that only complete deletion of Cox-2 leads to disruption of this mucosal lesion.

thumbnail image

Figure 1. Histology of the gastric tissues of mice according to age and genotype. Panels ac, Tff1 −/− Cox-2 −/− mice. (a) An ulcer with mild inflammation and fibrinoid debris of a 2 months old mouse. (b) An ulcer (arrow) and strong chronic transmural inflammation in a 4 month old mouse. Secondary lymphoid follicles are depicted by asterisks and residual adenoma tissue by an arrowhead. (c) Same sample as in (b) immunostained for plasma cells. (d) Normal pyloric histology in a 4 month old Tff1 +/− Cox-2 −/− mouse. (e) Intact hyperplastic mucosa in a 1 month old Tff1 −/− Cox-2 +/− mouse. (f) Intact and noninflamed pyloric adenoma in a 6 month old Tff1 −/− Cox-2 +/− mouse. The Brunner glands, marking the start of the duodenum, are visible on the left side of panels d–f. Adenoma height in a 6 months old Tff1 −/− Cox-2 +/− mouse (g) and in a Tff1 −/− Cox-2 −/− mouse (h); a marker exemplifies the height detection. (i) Average height of the adenomas of Tff1 −/− Cox-2 +/− mice (height in fields: 8.8 ± 1.3, n = 8) compared to Tff1 −/− Cox-2 −/− mice (0.0 ± 0.0, n = 6); *p < 0.005. Original magnification 100× (a, d–f) and 40× (b, c, g, h).

Download figure to PowerPoint

Treatment with a Cox-2 selective inhibitor of already established adenomas

To characterize the effect of Cox-2 inhibition on an already established adenoma, 2 months old Tff1 deficient mice were treated with a selective Cox-2 inhibitor celecoxib (1,600 ppm). This drug treatment lasted for 2 weeks or 8–14 weeks and resulted in treatment time-dependent morphological changes (Table 2). The 2 weeks treatment (eight mice) lead to only one ulcer and did not chance the height of the adenoma (Fig. 2c; height in fields: 6.5 ± 1.0 control food, 5.9 ± 2.8 celecoxib; p = 0.11), while the longer treatment period (8–14 weeks, 11 mice) resulted in six ulcerations and the height of the adenoma was reduced by 61% when compared to mice that were fed control food (Figs. 2a–2c; height in fields: 7.9 ± 1.6 control food, 3.1 ± 2.8 celecoxib; p < 0.005). Importantly, when the mice were fed with celecoxib for 16 weeks after which the drug was replaced by control food for 9 weeks (n = 3), the intact adenoma reappeared without any inflammation (Fig. 2d). However, the mice (n = 3) treated with celecoxib for full 25 weeks had an ulcerated and chronically inflamed phenotype (Fig. 2e). The mice did not suffer any noticeable side-effects and the weight of the mice in these two groups was almost identical (average weight of the control group 24.7 ± 1.2 g and in the treatment group 23.9 ± 1.1 g). Thus, celecoxib treatment was continued for 25 weeks without any apparent side-effects, but a discontinued treatment lead to reappearance of the adenoma.

thumbnail image

Figure 2. Effect of celecoxib (1,600 ppm) on adenomas in the pyloric region of Tff1 −/− mice. (a,b). Adenoma height in a mouse treated for 12 weeks with control food (a) or with celecoxib (b), a marker exemplifies the height detection. (c) Left panel. Average height of the adenomas treated for 2 weeks with control food (Ctr) (height in fields: 6.5 ± 1.0, n = 8) or celecoxib (Cele) (5.9 ± 2.8, n = 8); p = 0.11. Right panel. Average height of the adenomas treated for 8–14 weeks with control food (height in fields: 7.9 ± 1.6, n = 12) or celecoxib (3.1 ± 2.8, n = 11); *p < 0.005. (d) Mice were fed with celecoxib for 16 weeks after which it was replaced by control food for 9 weeks. The adenoma reappeared without ulceration or inflammation. Adenoma (Ad) and Brunner glands (Br) are marked with a dotted line. (e) The mice treated with celecoxib for 25 weeks had an ulcerated (Ul) and chronically inflamed pyloric region. Original magnification 40×.

Download figure to PowerPoint

To understand the mechanism of Cox-2 inhibition we measured macrophage number, apoptosis, blood vessel density, proliferation, B cell count, T cell count and senescence in the adenomas after 2 weeks treatment with celecoxib or control food. The number of macrophages was significantly higher in the adenomas of the celecoxib treated mice (macrophages per high power field: 36.3 ± 12.0 in mice fed with control food, 50.9 ± 18.4 mice treated with celecoxib, p = 0.021; Figs. 3a–3c). The count for M2 type alternatively activated macrophages was not statistically different between the two treatment groups (CD206 positive cells per high power field: 34.6 ± 6.1 control food, 39.5 ± 6.9 celecoxib; Fig. 3c and Supporting Information Fig. 1A). However, in the celecoxib treated group there were less CD206 positive macrophages than total number of macrophages (Fig. 3c; p = 0.05), while in the control groups the number of F4/80 and CD206 positive macrophages was not different (p = 0.64). Only few apoptotic cells were scattered in the adenomas and no difference between the treatment group was apparent when compared to the control group. B cells were present in very low numbers in both groups (B cells per high power field: 1.4 ± 1.4 control food, 1.3 ± 1.3 celecoxib; Fig. 3c). There was no difference in blood vessel density (vessels per field: 56.7 ± 9.4 control food, 51.8 ± 11.5 celecoxib; Fig. 3c and Supporting Information Fig. 1B), or proliferation index (Ki-67 positive epithelial cells: 59.1% ± 6.3% control food, 62.1% ± 5.3% celecoxib; Fig. 3c and Supporting Information Fig. 1C). The number of T cells in the undamaged mucosa was very low and there was no difference between the treatment groups (T cells per field: 2.4 ± 1.5 control food, 3.1 ± 1.2 celecoxib; Fig. 3c and Supporting Information Fig. 1D; p = 0.11)). There were numerous T cells in the lamina propria of the adjacent duodenal villi, thus confirming the specificity of the staining (Supporting Information Fig. 1D). The only tissue that displayed detectable senescent cells was the duodenal Brunner glands, but no senescent cells were detected in the adenomas of control food or celecoxib treated animals (Supporting Information Fig. 2).

thumbnail image

Figure 3. Count of macrophages, microvessels, proliferative cells, T cells and B cells in the adenomas. Macrophages were stained with F4/80 antibody in Tff1 −/− mice treated for 2 weeks with control food (a) or celecoxib (b). Original magnification 400×. (c) Count for F4/80 positive cells (macrophages), CD206 positive cells (M2 activated macrophages), microvessels (MECA-32), % of Ki-67 positive cells, and T cells (CD3e) and B cells (B220) in mice treated for 2 weeks with control food (Ctr) or celecoxib (Cele). *p = 0.021.

Download figure to PowerPoint

Identity of Cox-2 expressing cells

As described previously,18 Cox-2 is expressed in the stromal cells of the luminal side of the pyloric adenoma in mice deficient for the Tff1 gene. The specificity of the Cox-2 antibody was confirmed with proteins isolated from mouse gastro-intestinal tissues by Western blot analysis (Supporting Information Fig. 3A). There was a strong Cox-2 signal from the pyloric region with adenoma (Tff1 −/−, Cox-2 +/+ mouse) whereas other stomach tissues and intestinal tissues were only weakly positive as well as pyloric tissue from a mouse without adenoma (Tff1 +/+, Cox-2 +/+). No Cox-2 protein was detected by immunoblotting or by immunohistochemistry in Cox-2 −/− mice (Supporting Information Figs. 3A and 3B). Our results show that stromal cells expressing Cox-2 are already present in the hyperplastic lesion in 1 month old mice as detected by immunohistochemistry (Fig. 4a). Importantly, the Cox-2 staining pattern in the reappeared adenomas (Fig. 4b) resembled closely the one observed in untreated adenomas.18 To identify the nature of the Cox-2 expressing cells, we used a double fluorescence staining against Cox-2 and stromal cell markers. We found a colocalization of Cox-2 and vimentin in adenomas that were either treated with control food or with celecoxib for 2 weeks (Fig. 4c and Supporting Information Fig. 4A). To investigate the role of epithelial mesenchymal transition (EMT) we stained the sections with E-cadherin. There was no positivity for E-cadherin in the stromal compartment of the adenomas which contained the Cox-2 positive cells, regardless of treatment with celecoxib or control food (Fig. 4d). Thus it is unlikely that the Cox-2 positive stromal cells would represent epithelial cells that had undergone EMT. We also used the double fluorescence staining technique to stain Cox-2 and smooth muscle actin (Sma), the leukocyte common antigen CD45, or the macrophage marker F4/80, but none of these markers colocalized with Cox-2 (Supporting Information Figs. 4B–4D). These results suggest that Cox-2 is expressed by fibroblasts in the mouse adenomas.

thumbnail image

Figure 4. Identity of Cox-2 expressing cells. (a) Cox-2 staining in the hyperplastic lesion of a 1 month old Tff1 −/− Cox-2 +/− mouse. Tff1 −/− mice are shown in (bd). (b) Regrown adenoma in a mouse treated for 16 weeks with celecoxib and subsequently 9 weeks with control food stained for Cox-2 expression. (c) Double fluorescence staining of pyloric area for Cox-2 (green) and vimentin (red) in a mouse treated for 2 weeks with control food. Cox-2 is expressed in stromal cells in the upper area of the adenoma. Cells that coexpress Cox-2 and vimentin show a clear yellow staining. (d) Staining of Cox-2 (left panel) and E-cadherin (right panel) in a mouse treated for 2 weeks with control food. Original magnification 100× (a), 40× (b), 200× (c), 400× (d).

Download figure to PowerPoint


  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Mice with a homozygous genetic deletion of Tff1 and Cox-2 displayed a phenotype with disrupted gastric adenoma. This finding is similar to our previously reported effect of celecoxib on the adenoma in Tff1 knockout mice,18 and thus confirms that the underlying mechanism of the drug is primarily dependent on inhibition of Cox-2 and not due to an unrelated effect. Whereas deletion or inhibition of Cox-2 led to tumor disruption in our model, transgenic overexpression of Cox-2 and mPGES-1 leads to hyperplastic gastric tumors in mice.9 Additionally, induction of the PGE2 pathway promotes gastric hamartoma development when the signaling of bone morphogenetic protein is inhibited.21 In our experiments, when treatment with the Cox-2 inhibitor was interrupted the adenomas regrew within 2 months. These experiments show the necessity of Cox-2 derived products for tumor integrity and promotion. Importantly, not only genetic inhibition of Cox-2 or an early start of celecoxib treatment disrupted adenomatous growth, but also the treatment of fully developed adenomas with celecoxib led to regression in our current experiments. Jacoby et al. reported similar findings in the Min mouse model for adenomatous polyposis. The mice were treated with celecoxib (highest dose 1,500 ppm) either before adenomas had developed or after most of the adenomas were established and a reduction in tumor multiplicity and size was found in both treatment groups.22 Also in azoxymethane induced colon tumors in rats the start of celecoxib treatment during or after initiation stages inhibited the incidence and multiplicity of these tumors.23 Combined these results indicate that treatment of tumors with celecoxib is not only effective at the initiation stage but also when the tumors have already been established.

Tumor associated macrophages (TAM) have been described to promote carcinogenesis.24 In a hyperplastic gastric tumor model (K19-C2mE), inhibition of PGE2 signaling led to a reduced macrophage number in the tumors.9 Furthermore, active PGE2 signaling has been shown to induce the chemokine CCL2, which recruits macrophages to the tumors of K19-Wnt1/C2mE mice, and CCL2 inhibition suppressed macrophage infiltration.25 However, in our model inhibition of Cox-2 lead to destruction of tumor tissue, whereas in K19-C2mE mice no injury was observed.9 We detected a substantial amount of macrophages in the adenomas of Tff1 −/− mice and their number was increased in the celecoxib treated group. Our analysis suggest that most macrophages in the control food treated group expressed also the marker CD206 and around 70% of all macrophages expressed CD206 in the celecoxib treated group. The CD206 expressing alternatively activated macrophages most likely represent TAM. It has been reported that T cells are not required for TAM recruitment to the gastric mucosa,25 and indeed there were almost no T cells detectable in our samples. The macrophages that did not express CD206 might be inflammatory macrophages (classically activated macrophages) that could possess cytotoxic activity against tumor cells.26

The early mechanism that led to tumor disruption were not due to changes in proliferation, apoptosis, or angiogenesis, as we could not detect any differences between the celecoxib or control food treated mice after 2 weeks. In previous models of xenografted gastric cancer cells the effect of Cox-2 inhibition has often been linked to increased apoptosis or decreased angiogenesis when measured after 2–4 weeks treatment.27, 28 This raises the question, whether the assessment of the adenomas after 2 weeks celecoxib treatment was too early to detect such changes. However, since phenotypic changes were found in some mice already after the 2 week treatment, and treatment for longer time periods resulted in erosive or ulcerative changes in all mice, we feel that in the Tff1 knockout mice the 2 week treatment should be sufficient to detect changes in the aforementioned parameters.

Tff1 is a potent protector of the gastric mucosa and promotes healing after mucosal damage.14 Also Cox-2 plays a role in gastric ulcer healing and inhibition of Cox-2 delays the healing process.29, 30 Depletion of Tff1 alone did not cause mucosal damage. Likewise, Cox-2 selective inhibitors as sole treating agent did not lead to gross gastric injury in experimental models.18, 31 However, combination of Tff1 depletion and Cox-2 inhibition lead to ulcer development at the location of the adenoma, whereas other parts of the gastrointestinal tract remained undamaged. Similar findings have been reported in two rat models, in which high levels of Cox-2 were found in stomachs of adrenalectomized rats or in rats with induced arthritis, and upon inhibition with selective Cox-2 inhibitors gastric lesions developed, but no damage was found in the gastric mucosa of normal rats treated with a Cox-2 inhibitor.32, 33 Interestingly, also Tff2 deficient mice showed an increased susceptibility to gastric ulceration caused by NSAID treatment. In this model the mice were treated with indomethacin, a nonselective Cox-2 inhibitor, that caused gastric injury in wild-type mice, and the severity of injury was significantly increased in Tff2 deficient mice.34 In summary, the sustained ulceration in our model might be the consequence of a reduced cytoprotection and impaired healing that is caused by the combined lack of Tff1 and Cox-2.

When celecoxib was withdrawn after 16 weeks treatment, the adenoma reappeared after 2 months, reaffirming the importance of Cox-2 in promotion of the adenoma growth. Whereas we saw the reappearance of the adenomas only after drug withdrawal, Carothers et al. report an initial regression but then recurrence of intestinal tumors in the Min mouse model after uninterrupted treatment with 1,500 ppm celecoxib for more than 3 months.35 Interestingly, in one case a mouse treated for 5 months with celecoxib developed a colon carcinoma that showed Cox-2 expression in the stromal cells and in the infiltrating tumor cells.35 In contrast, in our model Cox-2 expression was restricted to the stromal cells at all conditions. Whereas in human dysplastic lesions36 and in tumors of interleukin-10 (−/−) mice37 myofibroblasts contributed to Cox-2 expression, we did not detect Cox-2 expressing myofibroblasts in our model. In addition, while Hull et al. showed that Cox-2 was localized to macrophages in the intestine of Min/+ mice,38 macrophages did not express Cox-2 in our model. We found that the stromal cells that expressed Cox-2 were fibroblasts, as they colocalized with cells expressing the fibroblast marker vimentin. Furthermore, these Cox-2 expressing stromal cells did not express E-cadherin and it is thus unlikely that they are epithelial cells that have undergone EMT. Cox-2 expressing fibroblasts have also been reported in intestinal polyps in the ApcΔ716 mouse model,39 and in human gastric precancerous lesions.36 When Cox-2 expressing gastric fibroblasts NF-21 and scirrhous gastric cancer cells OCUM-2M were xenografted together into nude mice, the resulting tumors were larger than with the cancer cells alone, and administration of a selective Cox-2 inhibitor JTE-522 decreased the tumor size.40 These results highlight the importance of Cox-2 expression in the stromal connective tissue cells in gastric tumor growth, at least in mouse models.

The Tff1 knockout mouse model resembles early stages of gastric intestinal carcinogenesis cascade. The lesions progress in both cases through different chronological stages, from hyperplasia to adenoma in the Tff1 knockout mouse model and from intestinal metaplasia to adenoma/dysplasia in intestinal gastric cancer. The intensity of Cox-2 positivity was similar in hyperplastic gastric tissue and in the adenomas of the Tff1 deficient mice, whereas in human precursor lesions Cox-2 immunopositivity increased with the severity of the lesion (reactive epithelium to high grade dysplasia).36 Furthermore, whereas Cox-2 positivity was restricted to stromal cells in the Tff1 knockout mice, the localization in human lesions was mostly epithelial, since stromal expression was only found at sites of epithelial injury.6, 36 A major difference is that intestinal gastric cancer arises usually due to atrophic gastritis caused by H. pylori infection, whereas the hyperplasias in the Tff1 knockout mice arise without H. pylori infection. Interestingly, it has recently been shown that expression of Tff1 was suppressed progressively by promoter hypermethylation in antral intestinal type precancerous lesions and gastric cancer, which is in line that Tff1 protein is an important tumor suppressor of gastric carcinogenesis.41

Due to the increased risk for cardiovascular events, the use of selective Cox-2 inhibitors is not recommended in cancer prevention.42–45 However, in a recent trial where patients with atrophic gastritis, intestinal metaplasia, or dysplasia were treated twice daily with 200 mg celecoxib, no increased risk of gastroduodenal ulcers and cardiovascular events was found.46 When a selective Cox-2 inhibitor was combined with S-1 in treating peritoneal metastasis of scirrhous gastric carcinoma in an in vivo mouse model the survival was significantly prolonged.47 In the recently updated safety analysis of the APC Trial an interaction between a baseline history of atherosclerotic heart disease and the increased risk for cardiovascular toxicity after celecoxib use was shown.48 Thus, providing a careful risk-benefit evaluation, the treatment of patients with a high risk of developing malignant gastric lesions with celecoxib alone or in combination with other drugs might be feasible. Our results indicate that a complete and long lasting inhibition of Cox-2 can suppress gastric tumor growth in mice.


  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

The authors thank Päivi Peltokangas and Tuire Koski for excellent technical assistance; Eero Mervaala and Taru Pilvi for the help with the mouse food preparation; the Biomedicum Helsinki Molecular Imaging Unit for technical assistance on imaging. A.T. is a student in the Helsinki Biomedical Graduate School. Academy of Finland (A.R.), Helsinki University Central Hospital Research Funds (A.R.), Sigrid Juselius Foundation (A.R., C.H.), Finnish Cancer Organisations (A.R., A.T.) and Finnish Cultural Foundation (A.T.).


  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information
  • 1
    Simmons DL, Botting RM, Hla T. Cyclooxygenase isozymes: the biology of prostaglandin synthesis and inhibition. Pharmacol Rev 2004; 56: 387437.
  • 2
    Saukkonen K, Rintahaka J, Sivula A, Buskens CJ, Van Rees BP, Rio MC, Haglund C, Van Lanschot JJ, Offerhaus GJ, Ristimäki A. Cyclooxygenase-2 and gastric carcinogenesis. APMIS 2003; 111: 91525.
  • 3
    Greenhough A, Smartt HJ, Moore AE, Roberts HR, Williams AC, Paraskeva C, Kaidi A. The COX-2/PGE2 pathway: key roles in the hallmarks of cancer and adaptation to the tumour microenvironment. Carcinogenesis 2009; 30: 37786.
  • 4
    Menter DG, Schilsky RL, DuBois RN. Cyclooxygenase-2 and cancer treatment: understanding the risk should be worth the reward. Clin Cancer Res 2010; 16: 138490.
  • 5
    Ristimäki A, Honkanen N, Jänkälä H, Sipponen P, Härkönen M. Expression of cyclooxygenase-2 in human gastric carcinoma. Cancer Res 1997; 57: 127680.
  • 6
    Saukkonen K, Nieminen O, van Rees B, Vilkki S, Härkönen M, Juhola M, Mecklin JP, Sipponen P, Ristimäki A. Expression of cyclooxygenase-2 in dysplasia of the stomach and in intestinal-type gastric adenocarcinoma. Clin Cancer Res 2001; 7: 192331.
  • 7
    Mrena J, Wiksten JP, Thiel A, Kokkola A, Pohjola L, Lundin J, Nordling S, Ristimäki A, Haglund C. Cyclooxygenase-2 is an independent prognostic factor in gastric cancer and its expression is regulated by the messenger RNA stability factor HuR. Clin Cancer Res 2005; 11: 73628.
  • 8
    Park ES, Do IG, Park CK, Kang WK, Noh JH, Sohn TS, Kim S, Kim MJ, Kim KM. Cyclooxygenase-2 is and idependent prognostic factor in gastric carcinoma patients receiving adjuvant chemotherapy and is not associated with EBV infection. Clin Cancer Res 2009; 15: 2918.
  • 9
    Oshima H, Oshima M, Inaba K, Taketo MM. Hyperplastic gastric tumors induced by activated macrophages in COX-2/mPGES-1 transgenic mice. EMBO J 2004; 23: 166978.
  • 10
    Udd L, Katajisto P, Rossi DJ, Lepisto A, Lahesmaa AM, Ylikorkala A, Jarvinen HJ, Ristimaki AP, Makela TP. Suppression of peutz-jeghers polyposis by inhibition of cyclooxygenase-2. Gastroenterology 2004; 127: 10307.
  • 11
    Kanda N, Seno H, Kawada M, Sawabu T, Uenoyoma Y, Nakajima T, Konda Y, Fukui H, Takeuchi T, Chiba T. Involvement of cyclooxygenase-2 in gastric mucosal hypertrophy in gastrin transgenic mice. Am J Physiol Gastrointest Liver Physiol 2006; 290: G51927.
  • 12
    Oshima H, Matsunaga A, Fujimura T, Tsukamoto T, Taketo MM, Oshima M. Carcinogenesis in mouse stomach by simultaneous activation of the Wnt signaling and prostaglandin E2 pathway. Gastroenterology 2006; 131: 108695.
  • 13
    Lefebvre O, Chenard MP, Masson R, Linares J, Dierich A, LeMeur M, Wendling C, Tomasetto C, Chambon P, Rio MC. Gastric mucosa abnormalities and tumorigenesis in mice lacking the pS2 trefoil protein. Science 1996; 274: 25962.
  • 14
    Taupin D, Podolsky DK. Trefoil factors: initiators of mucosal healing. Nat Rev Mol Cell Biol 2003; 4: 72132.
  • 15
    Tomasetto C, Rio MC. Pleiotropic effects of trefoil factor 1 deficiency. Cell Mol Life Sci 2005; 62: 291620.
  • 16
    Machado JC, Carneiro F, Ribeiro P, Blin N, Sobrinho-Simoes M. pS2 protein expression in gastric carcinoma. an immunohistochemical and immunoradiometric study. Eur J Cancer 1996; 32A: 158590.
  • 17
    Milne AN, Carvalho R, Morsink FM, Musler AR, de Leng WW, Ristimaki A, Offerhaus GJ. Early-onset gastric cancers have a different molecular expression profile than conventional gastric cancers. Mod Pathol 2006; 19: 56472.
  • 18
    Saukkonen K, Tomasetto C, Narko K, Rio MC, Ristimäki A. Cyclooxygenase-2 expression and effect of celecoxib in gastric adenomas of trefoil factor 1-deficient mice. Cancer Res 2003; 63: 30326.
  • 19
    Dinchuk JE, Car BD, Focht RJ, Johnston JJ, Jaffee BD, Covington MB, Contel NR, Eng VM, Collins RJ, Czerniak PM. Renal abnormalities and an altered inflammatory response in mice lacking cyclooxygenase II. Nature 1995; 378: 4069.
  • 20
    Yu Y, Fan J, Chen XS, Wang D, Klein-Szanto AJ, Campbell RL, FitzGerald GA, Funk CD. Genetic model of selective COX2 inhibition reveals novel heterodimer signaling. Nat Med 2006; 12: 699704.
  • 21
    Oshima H, Itadani H, Kotani H, Taketo MM, Oshima M. Induction of prostaglandin E2 pathway promotes gastric hamartoma development with suppression of bone morphogenetic protein signaling. Cancer Res 2009; 69: 272933.
  • 22
    Jacoby RF, Seibert K, Cole CE, Kelloff G, Lubet RA. The cyclooxygenase-2 inhibitor celecoxib is a potent preventive and therapeutic agent in the min mouse model of adenomatous polyposis. Cancer Res 2000; 60: 50404.
  • 23
    Reddy BS, Hirose Y, Lubet R, Steele V, Kelloff G, Paulson S, Seibert K, Rao CV. Chemoprevention of colon cancer by specific cyclooxygenase-2 inhibitor, celecoxib, administered during different stages of carcinogenesis. Cancer Res 2000; 60: 2937.
  • 24
    Qian BZ, Pollard JW. Macrophage diversity enhances tumor progression and metastasis. Cell 2010; 141: 3951.
  • 25
    Oshima H, Hioki K, Popivanova BK, Oguma K, Van Rooijen N, Ishikawa TO, Oshima M. Prostaglandin E 2 signaling and bacterial infection recruit tumor-promoting macrophages to mouse gastric tumors. Gastroenterology 2011; 140: 596, 607.e7.
  • 26
    Allavena P, Sica A, Garlanda C, Mantovani A. The yin-yang of tumor-associated macrophages in neoplastic progression and immune surveillance. Immunol Rev 2008; 222: 15561.
  • 27
    Fu SL, Wu YL, Zhang YP, Qiao MM, Chen Y. Anti-cancer effects of COX-2 inhibitors and their correlation with angiogenesis and invasion in gastric cancer. World J Gastroenterol 2004; 10: 19714.
  • 28
    Iwata C, Kano MR, Komuro A, Oka M, Kiyono K, Johansson E, Morishita Y, Yashiro M, Hirakawa K, Kaminishi M, Miyazono K. Inhibition of cyclooxygenase-2 suppresses lymph node metastasis via reduction of lymphangiogenesis. Cancer Res 2007; 67: 101819.
  • 29
    Mizuno H, Sakamoto C, Matsuda K, Wada K, Uchida T, Noguchi H, Akamatsu T, Kasuga M. Induction of cyclooxygenase 2 in gastric mucosal lesions and its inhibition by the specific antagonist delays healing in mice. Gastroenterology 1997; 112: 38797.
  • 30
    Watanabe T, Higuchi K, Taira K, Sasaki E, Shiba M, Tominaga K, Fujiwara Y, Oshitani N, Arakawa T. Rebamipide reduces delay in gastric ulcer healing in cyclooxygenase-2-deficient mice. Dig Dis Sci 2005; 50( Suppl 1): S639.
  • 31
    Laine L, Takeuchi K, Tarnawski A. Gastric mucosal defense and cytoprotection: bench to bedside. Gastroenterology 2008; 135: 4160.
  • 32
    Filaretova L, Tanaka A, Komoike Y, Takeuchi K. Selective cyclooxygenase-2 inhibitor induces gastric mucosal damage in adrenalectomized rats. Inflammopharmacology 2002; 10: 41322.
  • 33
    Kato S, Ogawa Y, Kanatsu K, Okayama M, Watanabe T, Arakawa T, Takeuchi K. Ulcerogenic influence of selective cyclooxygenase-2 inhibitors in the rat stomach with adjuvant-induced arthritis. J Pharmacol Exp Ther 2002; 303: 5039.
  • 34
    Farrell JJ, Taupin D, Koh TJ, Chen D, Zhao CM, Podolsky DK, Wang TC. TFF2/SP-deficient mice show decreased gastric proliferation, increased acid secretion, and increased susceptibility to NSAID injury. J Clin Invest 2002; 109: 193204.
  • 35
    Carothers AM, Moran AE, Cho NL, Redston M, Bertagnolli MM. Changes in antitumor response in C57BL/6J-min/+ mice during long-term administration of a selective cyclooxygenase-2 inhibitor. Cancer Res 2006; 66: 64328.
  • 36
    van Rees BP, Saukkonen K, Ristimaki A, Polkowski W, Tytgat GN, Drillenburg P, Offerhaus GJ. Cyclooxygenase-2 expression during carcinogenesis in the human stomach. J Pathol 2002; 196: 1719.
  • 37
    Shattuck-Brandt RL, Varilek GW, Radhika A, Yang F, Washington MK, DuBois RN. Cyclooxygenase 2 expression is increased in the stroma of colon carcinomas from IL-10(-/-) mice. Gastroenterology 2000; 118: 33745.
  • 38
    Hull MA, Booth JK, Tisbury A, Scott N, Bonifer C, Markham AF, Coletta PL. Cyclooxygenase 2 is up-regulated and localized to macrophages in the intestine of min mice. Br J Cancer 1999; 79: 1399405.
  • 39
    Sonoshita M, Takaku K, Oshima M, Sugihara K, Taketo MM. Cyclooxygenase-2 expression in fibroblasts and endothelial cells of intestinal polyps. Cancer Res 2002; 62: 68469.
  • 40
    Yashiro M, Nakazawa K, Tendo M, Kosaka K, Shinto O, Hirakawa K. Selective cyclooxygenase-2 inhibitor downregulates the paracrine epithelial-mesenchymal interactions of growth in scirrhous gastric carcinoma. Int J Cancer 2007; 120: 68693.
  • 41
    Tomita H, Takaishi S, Menheniott TR, Yang X, Shibata W, Jin G, Betz KS, Kawakami K, Minamoto T, Tomasetto C, Rio MC, Lerkowit N, et al. Inhibition of gastric carcinogenesis by the hormone gastrin is mediated by suppression of TFF1 epigenetic silencing. Gastroenterology 2011; 140: 87991.
  • 42
    Bresalier RS, Sandler RS, Quan H, Bolognese JA, Oxenius B, Horgan K, Lines C, Riddell R, Morton D, Lanas A, Konstam MA, Baron JA, et al. Cardiovascular events associated with rofecoxib in a colorectal adenoma chemoprevention trial. N Engl J Med 2005; 352: 1092102.
  • 43
    Solomon SD, McMurray JJ, Pfeffer MA, Wittes J, Fowler R, Finn P, Anderson WF, Zauber A, Hawk E, Bertagnolli M, Adenoma prevention with celecoxib (APC) Study Investigators. Cardiovascular risk associated with celecoxib in a clinical trial for colorectal adenoma prevention. N Engl J Med 2005; 352: 107180.
  • 44
    Nussmeier NA, Whelton AA, Brown MT, Langford RM, Hoeft A, Parlow JL, Boyce SW, Verburg KM. Complications of the COX-2 inhibitors parecoxib and valdecoxib after cardiac surgery. N Engl J Med 2005; 352: 108191.
  • 45
    Marnett LJ. The COXIB experience: a look in the rearview mirror. Annu Rev Pharmacol Toxicol 2009; 49: 26590.
  • 46
    Feng GS, Ma JL, Wong BC, Zhang L, Liu WD, Pan KF, Shen L, Zhang XD, Li J, Xia HH, Li JY, Lam SK, et al. Celecoxib-related gastroduodenal ulcer and cardiovascular events in a randomized trial for gastric cancer prevention. World J Gastroenterol 2008; 14: 45359.
  • 47
    Tendo M, Yashiro M, Nakazawa K, Yamada N, Sawada T, Ohira M, Hirakawa K. A synergic inhibitory-effect of combination with selective cyclooxygenase-2 inhibitor and S-1 on the peritoneal metastasis for scirrhous gastric cancer cells. Cancer Lett 2006; 244: 24751.
  • 48
    Bertagnolli MM, Eagle CJ, Zauber AG, Redston M, Breazna A, Kim K, Tang J, Rosenstein RB, Umar A, Bagheri D, Collins NT, Burn J, et al. Five-year efficacy and safety analysis of the adenoma prevention with celecoxib trial. Cancer Prev Res 2009; 2: 31021.

Supporting Information

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Additional Supporting Information may be found in the online version of this article.

IJC_27331_sm_SuppFig1.tif10689KSupporting Information Figure 1. Staining of macrophages, blood vessels, proliferative cells, and T cells in gastric tissues. Tff1 -/- mice treated for two weeks with control food (left panel) or celecoxib (right panel) are displayed in A-C. A. M2 activated macrophages were stained with CD206. B. Blood vessels were stained with MECA-32. C. Proliferating cells were stained with Ki-67. D. T cells were stained with CD3e. T cells in an adenoma section are marked with an arrow and control duodenal villi with abundant T cell staining in the lamina propria are displayed in the right panel. Original magnification 400x (A, D left panel), 200x (B), 100x (C, D villi).
IJC_27331_sm_SuppFig2.tif5084KSupporting Information Figure 2. Detection of senescent cells with senescence associated β-galactosidase staining. A. Gastric tissue of a mouse treated for 2 weeks with celecoxib is displayed showing duodenal villi, Brunner glands, and adenoma tissue. Only the Brunner glands display senescent cells as detected by the blue color produced by β-galactosidase. There are no senescent cells in the adenoma. Right panel: Enlargement of adenoma region. Original magnification 40x (left panel), 100x (right panel).
IJC_27331_sm_SuppFig3.tif3208KSupporting Information Figure 3. Expression of Cox-2 in different tissues and genotypes. A. Cox-2 expression was investigated by Western blot analysis in proteins isolated from stomach, pyloric, or small intestinal tissues. Cox-2 was highly expressed in the pyloric region of Tff1 -/- mice. No expression was detected in Cox-2 -/- mice. β-actin served as a loading control. B. Cox-2 staining in the hyperplastic lesion of a one month old Tff1 -/- Cox-2 -/- mouse. No staining was detected. Original magnification 100x.
IJC_27331_sm_SuppFig4.tif13397KSupporting Information Figure 4. Identity of Cox-2 expressing cells. Double fluorescence staining of gastric tissues for Cox-2 (green color) and vimentin (A, red color) leukocyte common antigen CD45 (B, red color), or macrophages F4/80 (C, red color), or myofibroblasts/smooth muscle cells Sma (D, red color) in Tff1 -/- mice treated for two weeks with 1600 ppm celecoxib. Cox-2 is expressed in stromal cells in the upper area of the adenoma. Cells that coexpress Cox-2 and vimentin show a clear yellow staining. Cells that express Cox-2 and CD45 or F4/80 or Sma do not colocalize. Original magnification 100x (B, C, D left panels), 200x (A, left panel; C, B right panels), 400x (A, D right panels).
IJC_27331_sm_SuppTab1.doc30KSupporting Information Table 1. Sequences of primers used for genotyping.
IJC_27331_sm_SuppTab2.doc48KSupporting Information Table 2. Detailed overview for quantitative analysis of markers.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.