IM (intestinal metaplasia) of the stomach is a pre-neoplastic lesion that usually follows Helicobacter pylori infection and that confers increased risk for gastric cancer development. After setting the role played by CDX2 (Caudal-type homeobox 2) in the establishment of gastric IM, it became of foremost importance to unravel the regulatory mechanisms behind its de novo expression in the stomach. In the present paper, we review the basic pathology of gastric IM as well as the current knowledge on molecular pathways involved in CDX2 regulation in the gastric context.
- bone morphogenetic protein (BMP)
- Helicobacter pylori
- intestinal metaplasia
Metaplasia: definition and examples
Metaplasia is classically defined as a reversible change in which one epithelial or mesenchymal adult cell type is replaced by another adult cell type . At least two possibilities for the cellular origin of metaplasia can be envisioned: (i) transdifferentiation of a tissue-specific differentiated cell into another cell type; or (ii) the change in commitment of a tissue-specific stem cell. This suggests a perhaps more accurate definition of metaplasia as the conversion, in postnatal life, of one cell type, differentiated or stem, into another .
In epithelial tissues, metaplasia follows a glandular–squamous, a squamous–glandular or a glandular–glandular pathway. A frequent example of the first type can be found in the respiratory epithelium of cigarette smokers, whereas a squamous–glandular metaplasia is most frequently observed in Barrett's oesophagus, subsequent to gastro-oesophageal reflux disease. IM (intestinal metaplasia) of the stomach is the prototype of a glandular–glandular metaplasia. Nevertheless, the metaplastic lesions are not exclusive to epithelial tissues, and mesenchymal tissues, such as bone, cartilage and adipose tissue, can be encountered ectopically in soft tissues or other organs (e.g. heart).
The present review examines several aspects of gastric IM, with an emphasis on the molecular pathways that lead to the CDX (Caudal-type homeobox) 2-dependent intestinalization of the gastric mucosa.
Intestinal metaplasia of human stomach
Gastric IM is a multifocal lesion characterized, morphologically, by the presence in the gastric mucosa, alone or in combination, of intestinal cell types such as goblet, Paneth and absorptive cells. The metaplastic glands show reorganization of the whole metaplastic crypt, with displacement of the proliferative zone from the neck region to the base of the crypt, and modification of the stroma that surrounds the metaplastic glands [3,4].
Characterization of gastric IM was initially based on histochemical methods that allowed the identification of three distinct types: type I, containing absorptive cells, Paneth cells and goblet cells secreting sialomucins; Type II with goblet and columnar cells secreting sialomucins; and Type III, with goblet and columnar cells secreting sulfomucins .
We and others have contributed to describe the molecular alterations of IM beyond the classical morphological and histochemical characterization. We know more now about the differentiation markers that are lost and gained in the differentiation switch. On the basis of newly identified markers of gastric and intestinal differentiation, namely mucin markers, a new classification of IM emerged. The new approach defines two main IM types: complete, or intestinal IM, and incomplete, or gastrointestinal mixed IM [6–8]. Complete IM is characterized by loss of gastric mucin markers (MUC1, MUC5AC, MUC6) and gain of the intestinal mucin MUC2, whereas incomplete IM is characterized by the maintenance of gastric mucin markers, together with the intestinal mucin MUC2 [6,7]. Interestingly, mixture of the gastric and the intestinal phenotype can also occur at the cellular level. Niwa et al.  showed that single cells can express both gastric and intestinal mucins, such as MUC5AC and MUC2.
Common to both types of IM is the de novo expression of CDX2, a transcription factor that, in normal conditions, is expressed exclusively in the intestine, but is present in every focus of aberrant intestinal differentiation occurring in the human body. It is thus considered the master gene for intestinal differentiation and it is reviewed in detail below.
Pathophysiology of gastric IM
The pathophysiology of metaplasia in general, and gastric IM in particular, involves diverse aggression mechanisms, but it is generally perceived that the common ground to all metaplasia types is chronic inflammation. Several epidemiological studies, pioneered by Pelayo Correa , show that IM occurs in gastric carcinogenesis pathway following Helicobacter pylori infection, which induces a chronic gastritis with atrophy and IM. Epidemiological data are supported by Mongolian gerbil models, where IM is also induced by H. pylori infection [11–13] and follows the same cascade of events. Mouse models vary considerably in the susceptibility to infection by different Helicobacter strains. Those that are susceptible do not develop the classical Cdx2-dependent IM seen in humans, but rather a different type of metaplasia characterized by the presence of antral morphology in the corpus, with expression of spasmolytic polypeptide, a trefoil peptide expressed in normal intestinal mucosa. From this observation was coined the name SPEM (spasmolytic polypeptide-expressing metaplasia) which, nevertheless, progresses from atrophy and evolves to dysplasia and cancer, closely recapitulating the human disease [14,15].
IM of the stomach is generally considered to be a pre-cancerous lesion, and the association with H. pylori infection is indisputable. However, IM develops in a small proportion of infected human subjects. Only 30% of infected individuals will develop IM and, among these, only approx. 7% will develop gastric cancer . These values are particularly important in countries where infection with H. pylori is still very high, such as Portugal  and South American countries , with approx. 80% of the population infected. Classification of IM in subtypes is particularly relevant because different risk impacts are attributed to each subtype, with incomplete IM being more associated with development of gastric carcinoma .
IM is the outcome of an adaptive response to an adverse environment that affects some, but not all, individuals in similar contexts. Evidence suggests that interactions between individual genetic variations, for instance IL (interleukin)-1β polymorphisms , H. pylori virulence factors, such as cagA+ strains [21,22], and exposure to other environmental carcinogens, are determinant for single individual susceptibility to IM.
H. pylori is probably not the only risk factor for IM development, but the results obtained this far on other environmental risk factors are still conflicting. However, the overall epidemiological information gathered shows that smoking plays a major deleterious role , and salty foods are probably harmful [23,24]. On the other hand, studies regarding the protective effect of dietary antioxidants and anti-inflammatory nutrients remain inconclusive .
Role of CDX genes in the development of IM
Intestinal differentiation is strongly dependent on the expression of the homeobox gene of the ParaHox cluster CDX2. This is evident from in vitro studies with undifferentiated intestinal cell lines where de novo expression of CDX2 induces intestinal differentiation [26–28]. Furthermore, Cdx2+/− heterozygous mice develop non-cancerous polyp-like heteroplasias in the intestine, characterized by a total loss of Cdx2 expression, that arise early during embryonic development  and correspond to an anterior homoeotic shift with appearance of gastric and oesophageal mucosa in the intestine, as well as an anterior homoeotic shift in the vertebrae [30,31]. Conversely, mouse models demonstrated that forced expression of Cdx2 is also involved in intestinal differentiation of the gastric mucosa. Indeed, transgenic mice with ectopic expression of Cdx2 in the stomach, either in embryonic development or postnatally, develop gastric IM [32,33], which means that CDX2 is responsible not only for the normal intestinal differentiation, but also for intestinal differentiation in aberrant locations. Several studies have shown that CDX2 is expressed in human intestinal metaplastic lesions of the stomach [34–38], oesophagus [39–41], liver  and gall bladder . Nonetheless, another gene has also been implicated in intestinal differentiation, which is the CDX2 homologue CDX1. The role of CDX1 in intestinal differentiation is not as well defined as that of CDX2 [44,45]. Cdx1-knockout mice, in contrast with the Cdx2-knockout mouse, and the Cdx1-overexpressing mice in the gut are viable and do not exhibit any intestinal malformation [46,47]. Cdx1 expression is turned on after Cdx2 in the developing gut endoderm and is absent from the foregut-type heteroplasia resulting from loss of Cdx2 , indicating that Cdx2 expression is before and is required for Cdx1 expression. However, CDX1 is also expressed in intestinal metaplastic lesions of several locations such as stomach , oesophagus  and liver , and a transgenic mouse with Cdx1 expression in the stomach also develops IM .
IM directed both by CDX1 and CDX2 exhibits not only a morphological transformation, but also a functional transformation. They act as transcription factors for a large number of intestinal genes that play major roles in intestinal function, such as those encoding sucrase-isomaltase , lactase–phlorizin hydrolase  and MUC2 . Moreover, ectopic Cdx2 expression in the gastric mucosa not only drives the intestine-type differentiation of the epithelium, but also has an impact on the underlying mesenchymal sheet .
Regulation of CDX2 expression in gastric IM
The evidence showing that CDX2 is a master gene involved in intestinal metaplasia of the gastric mucosa is very convincing. The question that is now being addressed is how it becomes ectopically expressed in gastric mucosa. Considering the context where this occurs, several mechanisms can be envisioned: CDX2 can be positively regulated by transcription factors whose expression is triggered by molecular pathways that become activated in the complex milieu where IM arises, namely the inflammatory milieu; CDX2 can be activated directly by H. pylori proteins; and/or repression of CDX2 by gastric transcription factors can be released in the molecular context that precedes IM. There is evidence supporting all of these mechanisms, suggesting that CDX2 regulation is rather complex and most probably involves interaction of several signalling pathways.
BMPs (bone morphogenetic proteins)
We and others [55,56] observed that BMPs, specifically BMP2 and BMP4, become abundantly expressed in the gastric mucosa infected with H. pylori, either in inflammatory cells  or in epithelial cells . The pathway is active in epithelial cells as demonstrated by the expression of key elements of the BMP pathway: BMP receptors, SMAD4 and the phosphorylated form of SMAD1/5/8. In addition, we demonstrated, in a gastric carcinoma cell line, that BMP2/4, through its canonical signalling transducer SMAD4, up-regulates CDX2 expression . In accordance, in an in vitro model of Barrett's oesophagus , oesophageal cells exposed to BMP4 begin to express columnar cytokeratins, not normally found in squamous epithelium. However, in the Barrett's oesophagus model, CDX2 expression was not assessed.
The precise function of the BMP pathway in the adult intestine is not fully understood, but growing evidence shows that the BMP pathway plays an important role in the maintenance of the intestinal proliferative and differentiated compartments in Xenopus , as well as in mammalian models . Furthermore, there is a distinct pattern of expression of BMP pathway elements and inhibitors along the crypt–villus axis . Mutations in members of this pathway have been associated with diseases such as juvenile polyposis syndrome [61,62], which is an hamartomatous syndrome characterized by the appearance of distorted crypts with excessive proliferation , but also focal loss of intestinal differentiation, and appearance of gastric markers . Ultimately, complete loss of expression of the BMP pathway in the intestine could lead to gastric metaplasia . Recently, two mouse models of BMP pathway inhibition in the intestine were developed: mice with ablated Bmpr1a in the intestine showed a mild proliferative defect, and an obvious differentiation defect of the secretory cell lineage with loss of goblet, Paneth and enteroendocrine cell maturation ; mice with overexpression of the BMP inhibitor Noggin in the intestine presented numerous ectopic crypts and altered proliferation together with polyp-like growth [59,66]. Altogether, this demonstrates that an intact BMP signalling in the intestinal epithelium is important for the balanced production of epithelial cell lineages, as well as maintenance of the proliferative/differentiation homoeostasis. Furthermore, the BMP pathway has been shown to cross-talk with other active signalling pathways in the intestine, such as the Wnt  and the Hedgehog  pathways.
H. pylori and SOX2 [SRY (sex-determining region Y) box 2]
Evidence exists that H. pylori is involved in the aberrant expression of CDX2 in the stomach, but very little is known about the molecular mechanisms involved in this process. Several studies have demonstrated that H. pylori infection alone  or in combination with GORD (gastro-oesophageal reflux disease)  leads to expression of CDX2 in the gastric mucosa, mainly in areas of IM, but also in foci of non-metaplastic cells. We have shown in vitro that gastric cell lines co-cultured with H. pylori have an increase in CDX2 expression, suggesting that the direct interaction of the bacteria with epithelial cells is able to induce CDX2 expression . In the same model, there was an increase in BMP2 expression . Recently, Asonuma et al.  showed that H. pylori inhibits IL-4-mediated SOX2 induction, in a co-culture model similar to the previous one . They also show, as other groups previously had , that SOX2 and CDX2 are inversely expressed in gastric mucosa and intestinal metaplasia. Finally, SOX2 is also suggested as a CDX2 repressor, since down-regulating its levels led to up-regulation of CDX2 expression . Accordingly, a study characterizing the Cdx2 promoter, revealed interactions of multiple transcription factors, and Sox2 appeared to be a repressor, impairing the positive regulation exerted by other transcription factors, whereas HNF4α (hepatic nuclear factor 4α), β-catenin, Tcf4 (T-cell factor 4) and, to a lesser extent, GATA6, all stimulated promoter activity of a Cdx2–luciferase construct in co-transfection experiments in a colonic cell line . SOX2 is crucial for foregut formation during embryogenesis and it has also been suggested as a transcription factor involved in gastric differentiation . However, the role of SOX2 in gastric differentiation remains to be better clarified.
Other regulatory mechanisms
Other CDX2 regulatory mechanisms have been studied in different models, such as intestine and Barrett's oesophagus, but not demonstrated to be relevant in gastric cells. In pancreatic and intestinal cell lines, OCT-1 was implicated in CDX2 promoter activation , and an autoregulation of CDX2 was proposed  in an intestinal cell line. The former hypothesis was tested in the gastric model, showing that, in fact, OCT-1 is overexpressed, together with CDX2, in intestinal metaplastic lesions, but it is unable to regulate its expression at the promoter level . The putative CDX2 autoregulation remains to be clarified in the gastric context. However, this regulatory pathway deserves further investigation, since if CDX2 is in fact able to regulate its own expression in IM, this might contribute to the stability of the phenotype and impair reversion of this gastric cancer precursor lesion.
It has been described that acid and bile acids could activate CDX2 in oesophageal epithelial cells through promoter demethylation, leading to Barrett's oesophagus ; however, in gastric cancer cell lines, in normal gastric mucosa and in IM foci, we could not detect a functional relation between the methylation status of the proximal CDX2 promoter and CDX2 expression , suggesting a tissue-specific role for CDX2 promoter methylation.
Gastric IM is a tissue phenotype switch initiated by H. pylori infection that, in susceptible individuals subjected to particularly aggressive H. pylori strains and lifestyle exposures, will trigger CDX2 expression. This process involves inhibition of gastric transcription factors, with SOX2 being a good candidate, and activation of CDX2 transcription factors, with the BMP pathway playing a central stimulatory role. It is known that, upon CDX2 expression, a whole set of intestinal genes is up-regulated with the appearance of an intestinal phenotype. The progress in identification of molecular pathways involved in CDX2 regulation has been enormous, but is still limited by the lack of an in vitro model of gastric IM and by the lack of a mouse model that recapitulates the CDX2-dependent metaplasia upon H. pylori infection. Animal models such as the Mongolian gerbil model, which reproduces the human disease, deserves further study to validate and extend observations with gastric cell lines.
We thank Fundação para a Ciência e a Tecnologia and the European Funding Fonds Européen de Développement Régional [Project PTDC/SAU-OBD/64490/2006] and Fundação Calouste Gulbenkian [Projects FC-54918 and FC-68697].
Barrett's Metaplasia: A Biochemical Society Focused Meeting held at University of Bath, Bath, U.K., 17–18 September 2009. Organized and Edited by Mark Farrant (Bath, U.K.), Rebecca Fitzgerald (Cambridge, U.K.) and David Tosh (Bath, U.K.).
Abbreviations: BMP, bone morphogenetic protein; CDX, Caudal-type homeobox; IL, interleukin; IM, intestinal metaplasia; SOX2, SRY (sex-determining region Y) box 2
- © The Authors Journal compilation © 2010 Biochemical Society