Ink4a/Arf-Dependent Loss of Parietal Cells Induced by Oxidative Stress Promotes CD44-Dependent Gastric Tumorigenesis

Chronic inflammation induces histopathologic progression of the stomach epithelium leading to the development of metaplasia followed by gastric adenocarcinoma (1, 2). Oxyntic glands, the predominant type of gastric gland, comprise several types of fully differentiated epithelial cells, including acid-secreting parietal cells and pepsinogen-secreting chief cells. Inflammation of the gastric epithelium results in a gradual loss of parietal cells and their replacement with proliferative metaplastic cells derived from transdifferentiated chief cells (3, 4), suggesting that chronic inflammation disrupts homeostasis of the gastric epithelium by altering the viability or differentiation status of these differentiated cells. Spasmolytic polypeptide–expressing metaplasia (SPEM) is triggered by the parietal cell loss, and is recognized as a precancerous lesion (5, 6). Parietal cells, thus, play a key role in the homeostasis of gastric glands, with the loss of these cells being considered an early and critical event in the histopathologic progression of the stomach epithelium and eventual development of gastric cancer. The molecular mechanism underlying parietal cell loss has remained unclear, however.

Chronic inflammation followed by carcinogenesis is associated with the production of reactive oxygen species (ROS; refs. 7, 8). ROS function as activators of MAPK signaling, including the Ras–Raf–MEK1/2–ERK1/2 and p38^MAPK signaling pathways, and they play opposing roles in tumor promotion and suppression (9, 10). In general, ROS-mediated activation of Ras–Raf–MEK1/2–ERK1/2 signaling is associated with carcinogenesis as a result of its promotion of cell survival and proliferation, whereas ROS-mediated activation of p38^MAPK signaling inhibits cell proliferation and induces cell senescence through induction of the tumor-suppressor proteins p16^INK4a and p19^ARF (11). The functional relevance of ROS-mediated signaling in the development of metaplasia and gastric cancer has not been elucidated, however.

CD44 exists in numerous isoforms that are generated through alternative splicing of CD44 precursor mRNA (12–14). CD44 variant (CD44v) isoforms, which contain additional insertions in the membrane-proximal extracellular region, are highly abundant in epithelial-type carcinomas (15). We previously showed that interaction of CD44v with xCT (SLC7A11), a subunit of the cystine/glutamate antiporter system xc(–), stabilizes the latter protein, and thereby promotes glutathione synthesis and potentiates the ability of cancer cells to defend themselves against ROS (16). In gastric carcinogenesis, de novo expression of CD44v is associated with the development of SPEM (17) and gastric tumors (18), and CD44v was recently identified as a cell surface marker of gastric cancer stem cells (19). Consistent with these observations, we recently showed that potentiation of ROS defense by the CD44v-xCT system plays a key role not only in tumor cells, but also in the expansion of SPEM cells in the K19-Wnt1/C2mE mouse, a transgenic model of gastric cancer induced by activation of Wnt and prostaglandin E₂ (PGE₂) signaling pathways (17). The populations of CD44v-expressing metaplastic cells and tumor cells in which ROS defense is bolstered by the CD44v-xCT system might, thus, be able to expand preferentially in inflamed gastric epithelium exposed to high levels of ROS. However, the role of ROS and its downstream signaling in CD44-dependent gastric tumorigenesis has remained unknown.

K19-Wnt1/C2mE transgenic mice, Ink4a/Arf^−/− mice and CD44^−/− K19-Wnt1/C2mE mice were described as previously (16, 20, 21). K19-Wnt1/C2mE transgenic mice were crossed with Ink4a/Arf^−/− mice to generate Ink4a/Arf^−/− K19-Wnt1/C2mE mice. All animal experiments were performed according to protocols approved by the Ethics Committee of Keio University. All efforts were made to minimize the suffering of animals used in this study.

Tamoxifen (Sigma-Aldrich) was injected i.p. at a daily dose of 250 mg/kg in WT or Ink4a/Arf^−/− mice for 3 days, and tissue was dissected for analysis at 3 days after the last injection. K19-Wnt1/C2mE mice at 20 weeks of age were injected i.p. with tamoxifen (250 mg/kg) once a week for 5 weeks. Tamoxifen was dissolved in a vehicle consisting of 10% ethanol and 90% sunflower seed oil (Sigma-Aldrich), and control mice were injected with vehicle alone.

Immunohistochemistry was performed as previously reported (16, 17). TFF2 was detected with rabbit monoclonal antibodies (diluted 1:100; Proteintech).The TFF2-positive area in five randomly selected microscopic fields (magnification, ×40) per section was measured with the use of analysis software (BZ-9000; Keyence), and the mean percentage positive area was calculated. Parietal cells were detected with mouse monoclonal antibodies to the β subunit of H⁺,K⁺-ATPase (MBL). The mean percentage of H⁺,K⁺-ATPase–positive cells per gland in five microscopic fields per mouse was calculated. CD44v8-10 was detected with a rat monoclonal antibody (diluted 1:100; ref. 16), and the proportion of positive cells was determined with the use of a TissueFAXS cell analysis system (Novel Science).

Immunoblot analysis was performed as described previously (16, 17).

Gastric gland units were isolated as previously described (22). Other experimental procedures are described in Supplementary Materials and Methods.

Data are presented as means ± SD and were analyzed with the unpaired Student t test as performed with the use of Microsoft Excel 2007. A P value of <0.05 was considered statistically significant.

To explore the relevance of ROS in gastric carcinogenesis, we orally administered ascorbic acid, a potent water-soluble antioxidant, to K19-Wnt1/C2mE mice from 12 to 30 weeks of age. Tumors were markedly smaller in the ascorbic acid–treated K19-Wnt1/C2mE mice than in untreated control animals at 30 weeks of age (Fig. 1A and B), suggesting that gastric tumorigenesis is promoted by ROS in these mice.

Given that CD44v promotes ROS defense in tumor cells in this model (17), expansion of the CD44v-expressing tumor cells might be the result of selection by ROS in the tumor microenvironment. We therefore investigated CD44v expression in gastric tumors of K19-Wnt1/C2mE mice treated with ascorbic acid. The relative area occupied by CD44v-expressing cells in tumors was markedly reduced by ascorbic acid treatment (Fig. 1C), suggesting that CD44v-expressing tumor cells expand preferentially compared with CD44v-negative cells in the presence of ROS. The expression of CD44v in epithelial cells of the normal foregut was not diminished by ascorbic acid treatment, however, suggesting that ROS depletion suppresses the selective expansion of ROS-resistant CD44v-expressing tumor cells without affecting normal epithelial cells in K19-Wnt1/C2mE mice.

Given that SPEM is a premalignant lesion (4), we next investigated whether CD44v-expressing SPEM cells might be the cells of origin for CD44v-expressing tumor cells. We thus examined whether ROS depletion by ascorbic acid treatment was able to reduce not only the number of CD44v-expressing tumor cells, but also that of CD44v-expressing SPEM cells in K19-Wnt1/C2mE mice. Formation of trefoil factor 2 (TFF2)–expressing SPEM was significantly suppressed in ascorbic acid–treated mice (Fig. 1D and E), indicating that ROS promotes CD44v-expressing SPEM formation as well as tumor formation.

Given that the parietal cell loss is a key event in SPEM development, we next examined the expression of the β subunit of H⁺,K⁺-ATPase, a marker of parietal cells. Immunohistochemical analysis revealed that the loss of parietal cells was significantly attenuated in ascorbic acid–treated mice (Fig. 1D and F), implicating ROS in the parietal cell loss apparent in untreated K19-Wnt1/C2mE mice. Analysis of the expression of CD44v in epithelium adjacent to tumors of K19-Wnt1/C2mE mice revealed that the emergence of CD44v-expressing SPEM cells was markedly suppressed by ascorbic acid treatment (Fig. 1D). Thus, ROS play a key role in parietal cell loss leading to the development of SPEM comprising CD44v-expressing SPEM cells.

To examine whether oxidative stress–induced signaling is operative in parietal cells of K19-Wnt1/C2mE mice, we investigated the activation status of p38^MAPK, a major target of ROS. The phosphorylated (activated) form of p38^MAPK (phospho-p38^MAPK) was found to be abundant in parietal cells of K19-Wnt1/C2mE mice (Fig. 2A), suggesting that ROS accumulate in these cells. To address this possibility further, we measured ROS levels in parietal cells and nonparietal gastric cells isolated from wild-type (WT) and K19-Wnt1/C2mE mice. Flow cytometric analysis of cells stained with 2′,7′-dichlorofluorescein diacetate (DCFH-DA) revealed that DCFH-DA fluorescence intensity was much higher in parietal cells of K19-Wnt1/C2mE mice than in nonparietal cells of the same animals or in parietal cells or nonparietal cells of WT mice (Fig. 2B), suggesting that ROS activates p38^MAPK signaling in parietal cells.

To examine further whether p38^MAPK is activated selectively in parietal cells of the stomach epithelium in the presence of high ROS levels, we exposed WT mice to ionizing radiation at a dose of 10 Gy to induce ROS production. The abundance of phospho-p38^MAPK was increased selectively in parietal cells of the irradiated mice (Fig. 2C), suggesting that parietal cells are highly sensitive to oxidative stress. We also examined the sensitivity of p38^MAPK activity in parietal cells to ROS with the use of a gastric organ culture system (Fig. 2D). Exposure of normal gastric mucosa to the oxidative stressor hydrogen peroxide (H₂O₂) induced p38^MAPK activation specifically in parietal cells, and this activation was attenuated in the additional presence of ascorbic acid or the specific p38^MAPK inhibitor SB203580 (Fig. 2E). Together, these results indicated that parietal cells have a low capacity for ROS defense, and therefore might be a major target for ROS during gastric carcinogenesis.

Given that activated p38^MAPK has been shown to induce gene expression at the Ink4a/Arf locus, which encodes p16^INK4a and p19^ARF, in cells exposed to oxidative stress (10), we next examined whether the expression of these proteins is increased in SPEM lesions of K19-Wnt1/C2mE mice. The abundance of p16^INK4a and p19^ARF as well as that of phospho-p38^MAPK were markedly increased in SPEM of K19-Wnt1/C2mE mice compared with normal gastric mucosa of WT mice (Fig. 3A), suggesting that the p38^MAPK−p16^INK4a/p19^ARF pathway is activated in SPEM.

To investigate the functional relevance of p16^INK4a and p19^ARF to parietal cell loss, we examined the effects of i.p. injection of tamoxifen, a potent inducer of parietal cell loss (23), in Ink4a/Arf knockout (Ink4a/Arf^−/−) mice. Immunohistochemical analysis revealed that the tamoxifen-induced parietal cell loss was markedly suppressed in Ink4a/Arf^−/− mice compared with WT mice (Fig. 3B and C), suggesting that the Ink4a/Arf locus is essential for this effect of tamoxifen. The area of TFF2-positive mucosa was also substantially smaller in tamoxifen-treated Ink4a/Arf^−/− mice than in tamoxifen-treated WT mice (Fig. 3B and D). Together, these results suggested that the Ink4a/Arf locus is required for the parietal cell loss, which plays a role in the development of SPEM in tamoxifen-treated mice.

We next investigated the abundance of phospho-p38^MAPK and p19^ARF in tamoxifen-treated mice. Immunofluorescence analysis revealed that phospho-p38^MAPK and p19^ARF were highly expressed in the few remaining parietal cells of tamoxifen-treated WT mice (Fig. 3E). The number of phospho-p38^MAPK–positive parietal cells in Ink4a/Arf^−/− mice after tamoxifen treatment was much higher than that in WT mice (Fig. 3E), suggesting that activation of p38^MAPK and downstream p19^ARF expression contribute to tamoxifen-induced parietal cell loss.

To examine whether the parietal cell loss induced by tamoxifen results in the generation of CD44v-expressing metaplastic cells in the gastric epithelium, we performed immunofluorescence analysis of CD44v and the proliferation marker Ki67. Injection of tamoxifen effectively depleted parietal cells in WT mice (Supplementary Fig. S1). Proliferative CD44v-expressing cells were found to emerge from the base of gastric glands in tamoxifen-treated WT mice, whereas this phenomenon was greatly attenuated in tamoxifen-treated Ink4a/Arf^−/− mice (Fig. 3F). These results suggested that the induction of Ink4a/Arf expression plays a key role in the parietal cell loss, and that such parietal cell loss triggers the generation of proliferative CD44v-expressing metaplastic cells.

Given that CD44v is a gastric cancer stem cell marker (19) and plays a key role in SPEM development (16), we next examined whether SPEM induced by tamoxifen treatment contains stem-like cells that possess the ability to reconstitute a gastric gland. To assess the stem cell potential of gastric epithelial cells, we isolated a 5-mm square piece of glandular mucosa from the stomach of mice treated (or not) with tamoxifen for evaluation of the ability to form gastric organoids (Fig. 4A; ref. 24). The number of gastric organoids generated by the TFF2-expressing gastric mucosa resected from WT mice treated with tamoxifen (Fig. 4B) was markedly increased compared with that for the gastric mucosa of nontreated mice (Fig. 4C), suggesting that SPEM tissue contains a much higher number of stem-like cells than does normal stomach mucosa. We next investigated CD44v expression in the gastric organoids. Immunofluorescence analysis revealed that organoids derived from both tamoxifen-treated and untreated WT mice contained CD44v-expressing stem-like cells, although the number of such cells was greatly increased for organoids formed by SPEM compared with those formed by normal gastric mucosa (Fig. 4D). These results suggested that CD44v-expressing stem-like cells are generated by the mechanically damaged epithelial tissue derived from normal gastric mucosa, and that parietal cell loss further increases the number of such stem-like cells in tamoxifen-induced SPEM.

Given that tamoxifen-induced parietal cell loss and subsequent SPEM development were suppressed in the stomach of Ink4a/Arf^−/− mice, we next examined the role of the Ink4a/Arf locus in tamoxifen-induced gastric organoid formation. The number of organoids formed by gastric mucosa from tamoxifen-treated mice was significantly reduced by genetic ablation of the Ink4a/Arf locus as well as that of CD44 (Fig. 4E and F), suggesting that Ink4a/Arf–dependent parietal cell loss induces the expansion of CD44v-expressing stem-like cells.

Given that tamoxifen treatment promoted the development of SPEM concomitant with an increase in the number of stem-like cells in WT mice, we next administered tamoxifen to K19-Wnt1/C2mE mice at 20 weeks of age to examine whether tamoxifen-induced parietal cell loss might also accelerate tumor development. I.p. injection of tamoxifen once a week for 5 weeks promoted the tumor growth in these mice (Fig. 5A and B). Furthermore, tamoxifen treatment enhanced the parietal cell loss as well as the production of CD44v-expressing tumor cells in K19-Wnt1/C2mE mice (Fig. 5C and D). Together, these results suggested that loss of parietal cells might play a role in the production of CD44v-expressing tumor cells and tumor growth in K19-Wnt1/C2mE mice.

To examine the functional relevance of the Ink4a/Arf locus in the development of gastric tumors in K19-Wnt1/C2mE mice, we generated Ink4a/Arf^−/− K19-Wnt1/C2mE mice. Tumors of Ink4a/Arf^−/− K19-Wnt1/C2mE mice were markedly smaller than those of K19-Wnt1/C2mE mice (Fig. 6A and B) and were similar in size to those of CD44^−/− K19-Wnt1/C2mE mice (16), suggesting that the Ink4a/Arf locus is required for CD44-dependent gastric tumor development in K19-Wnt1/C2mE mice. The number of parietal cells remaining in the gastric epithelium adjacent to tumors was higher for Ink4a/Arf^−/− K19-Wnt1/C2mE mice than for K19-Wnt1/C2mE mice or CD44^−/− K19-Wnt1/C2mE mice (Fig. 6C and D), suggesting that the parietal cell loss in the initial stage of gastric tumorigenesis is regulated by the Ink4a/Arf locus but not by CD44 expression. On the other hand, the development of SPEM was significantly suppressed in both Ink4a/Arf^−/− K19-Wnt1/C2mE mice and CD44^−/− K19-Wnt1/C2mE mice compared with K19-Wnt1/C2mE mice (Fig. 6E and F), suggesting that the suppression of SPEM development in Ink4a/Arf^−/− K19-Wnt1/C2mE mice and CD44^−/−K19-Wnt1/C2mE mice is mediated by different mechanisms. The Ink4a/Arf locus is essential for the parietal cell loss, whereas CD44 is required for the expansion of SPEM after parietal cell loss.

Our results indicated that the stem-like cells in SPEM might be the cells of origin for CD44v-expressing tumor cells in K19-Wnt1/C2mE mice; therefore, we examined whether the suppression of CD44v-expressing SPEM development might reduce the production of CD44v-expressing tumor cells in Ink4a/Arf^−/− K19-Wnt1/C2mE mice. The area of CD44v-expressing tumor cells was found to be greatly reduced in Ink4a/Arf^−/− K19-Wnt1/C2mE mice compared with K19-Wnt1/C2mE mice (Fig. 6G and H). Given that SPEM is a precancerous lesion of gastric cancer, these results suggest that CD44v-expressing metaplastic cells produced as a result of parietal cell loss might subsequently undergo conversion to CD44v-expressing tumor cells in K19-Wnt1/C2mE mice (Fig. 6I).

Gastrointestinal malignancies are strongly linked to chronic inflammation (25–29). Gastric tumorigenesis in K19-Wnt1/C2mE mice was also recently shown to be suppressed by knockout of either TNFα or its receptor TNFR1, suggesting that TNFα-dependent inflammation plays a key role in the development of gastric cancer (30). Furthermore, the development of SPEM in response to acute parietal cell loss is dependent on the recruitment of macrophages to the gastric epithelium (31). Together, these findings suggest that inflammation triggers the histopathologic alterations of gastric epithelium that leads to tumor development. We have now shown that the inflammation-associated ROS plays a key role in the parietal cell loss that is an early and critical step in the development of SPEM, and the subsequent formation of gastric tumors. The accumulation of ROS triggers the p38^MAPK–p16^INK4a/p19^ARF signaling pathway selectively in parietal cells and the consequent induction of parietal cell loss. In fact, parietal cells have been reported to contain abundant mitochondria (32), a major source of intrinsic ROS production, and have been considered to be highly susceptible to oxidative stress (33). Together, oxidative stress–dependent p38^MAPK–p16^INK4a/p19^ARF signaling is activated selectively in parietal cells of the stomach epithelium in the presence of high ROS level.

The Ink4a/Arf locus encodes the cyclin-dependent kinase inhibitor p16^INK4a as well as p19^ARF, both of which have been shown to induce cell-cycle arrest, senescence, and apoptosis in response to oncogenic stimulation (34, 35). These proteins are thought to serve as key regulators in tumor-suppressor networks that are often disabled in human cancers (35). However, our present results indicate that ROS-induced expression of p16^INK4a and p19^ARF promotes gastric tumorigenesis by inducing parietal cell loss and the emergence of CD44v-expressing SPEM cells. The activation of tumor-suppressor networks might, thus, promote carcinogenesis in some cases through induction of the loss of gatekeeper cells that play an important role in epithelial homeostasis.

Carcinogenesis is often coupled to the generation of ROS, and cancer cells have, therefore, evolved mechanisms to protect themselves from oxidative stress through upregulation of both antioxidants and prosurvival molecules (36, 37). Expression of CD44v promotes cystine transport via system xc(–), and thereby contributes to ROS defense by increasing the synthesis of reduced glutathione in cancer cells (16). We previously showed that the de novo expression of CD44v in metaplasia cells potentiates ROS defense in these cells, and thereby promotes the development of SPEM (17). Depletion of CD44 or treatment with the xCT inhibitor sulfasalazine markedly suppressed the development of SPEM and gastric tumors in K19-Wnt1/C2mE mice (16, 17), and the expression of SPEM-related genes was found to be greatly increased in CD44v-expressing tumor cells compared with CD44-negative tumor cells in this model (17). We have now found that, compared with normal gastric mucosa, SPEM induced as a result of Ink4a/Arf–dependent parietal cell loss contains an increased number of CD44v-expressing stem-like cells that are capable of initiating gastric organoid formation. Gastric organoids established from SPEM were found to express high level of Troy mRNA (data not shown), suggesting that Troy-expressing chief cells, which act as the reserve stem cells in response to gastric tissue damage, might be the candidate source of CD44v-expressing stem-like cells. Our results thus suggest that, in gastric tumorigenesis, it is likely that CD44v-expressing stem-like cells in SPEM undergo conversion to CD44v-expressing tumor cells, possibly as a result of the activation of oncogenic Wnt signaling or other genetic or epigenetic events (Fig. 6I).

In conclusion, our present data provide evidence that ROS promote gastric tumorigenesis through the induction of p38^MAPK–p16^INK4a/p19^ARF signaling in parietal cells—cells that serve as gatekeepers for homeostasis in oxyntic glands.

No potential conflicts of interest were disclosed.

Conception and design: R. Seishima, J.R. Goldenring, H. Saya, O. Nagano

Development of methodology: T. Wada, K. Tsuchihashi

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): R. Seishima, T. Wada, M. Yoshikawa

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): R. Seishima, T. Wada, K. Tsuchihashi, M. Yoshikawa, H. Hasegawa

Writing, review, and/or revision of the manuscript: R. Seishima, J.R. Goldenring, O. Nagano

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S. Okazaki, H. Oshima, M. Oshima, T. Sato

Study supervision: H. Hasegawa, Y. Kitagawa, O. Nagano

The authors thank I. Ishimatsu, Y. Suzuki, and S. Hayashi for technical assistance as well as M. Sato for help in preparation of the article.

This work was supported by grants (to H. Saya) from, as well as in part by the Project for Development of Innovative Research on Cancer Therapeutics (P-Direct; to O. Nagano) of, the Ministry of Education, Culture, Sports, Science, and Technology of Japan. Department of Veterans Affairs Merit Review award and NIH RO1 DK071590 (to J.R. Goldenring).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.