Acid and the Basis for Cellular Plasticity and Reprogramming in Gastric Repair and Cancer

Abstract

Subjected to countless daily injuries, the stomach still functions as a remarkably efficient digestive organ and microbial filter. Here, we follow the lead of the earliest gastroenterologists who were fascinated by the anti-septic and digestive power of gastric secretions. We propose that it is easiest to understand how the stomach responds to injury by stressing the central role of the most important gastric secretion, acid. The stomach follows two basic patterns of adaptation. The superficial response is a pattern whereby the surface epithelial cells migrate and rapidly proliferate to repair erosions induced by acid or other irritants. The stomach can also adapt through a glandular response when the source of acid is lost or compromised (i.e., the process of oxyntic atrophy). We primarily review the mechanisms governing the glandular response, which is characterized by a metaplastic change in cellular differentiation known as Spasmolytic Polypeptide-Expressing Metaplasia, or SPEM. We propose that the stomach, like other organs, exhibits marked cellular plasticity: the glandular response involves reprogramming mature cells to serve as auxiliary stem cells that replace lost cells. Unfortunately, such plasticity may mean that the gastric epithelium undergoes cycles of differentiation and de-differentiation that increase the risk for accumulating cancer-predisposing mutations.

INTRODUCTION

Historical Insights Into the Stomach

The human stomach is an exocrine and endocrine organ that initiates digestion. Some of the earliest scientific work on the digestive tract focused on the exocrine function of the stomach. This was likely because the live workings of most internal organs were mysteries; however, the secretions of the stomach were accessible with a little ingenuity. For example, in the early part of the 18^th century, the pioneering French scientist Antoine Ferchault de Réaumur had animals swallow food in containers that allowed access to their digestive juices but resisted the stomach’s mechanical contractions (reviewed in ¹). Réaumur’s work was expanded upon by the Italian Lazzaro Spallanzani in the late 1700s. Spallanzani showed that he could extract gastric juice and observe its digestive effects ex vivo over several days when these gastric secretions were mixed with food². In so doing, he helped to prove that gastric secretions could turn food into an “impalpable mass” of chyme. By inducing injury in animal stomachs following the forced ingestion of various caustic (and sometimes sharp!) substances, he also was one of the first to learn of the stomach’s unique adaptive capacity.

Thus, from a historical perspective, it can be argued that the stomach first made gastroenterology a field worthy of careful scientific study. Most research in gastroenterology over the past few decades, however, has not focused on the stomach, and gastric cancer, though the third leading cause of cancer-related deaths worldwide^3,4, remains the most poorly funded cancer of the gastrointestinal tract⁵. Moreover, we still have a rudimentary understanding of how gastric epithelial cells produce the secretions that so fascinated early physiologists. We are just beginning to understand how gastric epithelium develops, how it is maintained in homeostasis and in injury, and how unresolved injury can ultimately lead to disease. The stomach is subjected to countless chemical and microbial injuries on a daily basis while managing to maintain its epithelial integrity (as well as its digestive and anti-septic functions). As we will discuss, the stomach’s ability to withstand these insults is largely due to the interaction between its prodigious acid production and the plasticity of its epithelium.

We will focus on the epithelial cells in the stomach that both produce and protect against the powerful secretions that have intrigued scientists for centuries. How is the stomach organized at an anatomic and glandular level, and how does this organization change during disease? How is gastric epithelium replenished following different forms of injury? We propose a novel classification, based on known responses of the stomach to injury, comprising two distinct (though not mutually exclusive) types of repair mechanisms: 1) the superficial response, fueled by changes in the rapidly recycling surface epithelium lining the stomach lumen, and 2) the glandular response, involving adaptations by cells deeper in the gastric unit (acid-secreting parietal cells and digestive enzyme-secreting chief cells). In particular, we will highlight recent literature illustrating the remarkable plasticity of the chief cell lineage, demonstrating how various forms of injury can cause these cells to be recalled from their post-mitotic, differentiated state back into the cell cycle to initiate and fuel repair. This process of cellular reprogramming affords the stomach remarkable flexibility during repair but may also increase the risk of developing gastric cancer, further emphasizing the delicate balance between re-establishing homeostasis and progressing to gastric neoplasia.

Functional Organization of the Stomach and the Importance of Acid in Gastric Injury

Generally speaking, the human stomach can be divided into two anatomic regions (Figure 1) that are developmentally and functionally distinct based on the organization of their glands^6–10. The gastric corpus^*, or body, makes up the majority of the stomach and is defined by oxyntic glands. Oxyntic glands are characterized by parietal cells that secrete hydrochloric acid and intrinsic factor (in humans) and chief cells that secrete digestive enzymes like pepsinogen. In contrast, the gastric units of the antrum, the distal portion of the stomach, are largely devoid of parietal cells and chief cells¹¹. Antral units consist predominantly of mucous cells that express lower amounts of zymogenic proteins along with specific mucins like MUC6^12,13 and protective factors like TFF2¹⁴. Gastric units in the corpus and antrum also contain scattered hormone-secreting endocrine cells¹⁵. For example, G cells, which are exclusive to the antrum, produce gastrin¹⁶, a hormone that stimulates acid secretion in the corpus. Tuft cells are rarer cells, whose functions are still being defined^17,18, though they are known to expand during gastric injury^19–21. Finally, the pit/foveolar cells line the surface of both the corpus and antrum and produce abundant mucus^22,23. It is worth noting that the morphologic distinction between corpus and antral units is much sharper in common experimental animals like mice but less clear-cut in the human stomach, where few purely antral units and a significant proportion of mixed-type units, containing cells characteristic of both corpus (parietal cells) and antrum (G cells), exist²⁴.

Figure 1. The anatomic and glandular organization of the human stomach.

The human stomach can be divided into two anatomic regions, the corpus (purple) and the antrum (orange). These regions are characterized by the cellular composition of their glands, with corpus units (purple inset) defined by acid-producing parietal cells (blue) and zymogenic chief cells (red). The antrum unit (orange inset) is largely devoid of parietal cells and chief cells and instead comprises mucous cells (green) that extend down to the gland base. Both corpus and antrum units contain an isthmus region, made up of proliferative stem cells (white), and a surface/pit region, with pit cells (purple) extending up from the isthmus to the luminal surface. Pre-parietal (light blue), pre-pit (light purple), and pre-mucous (light green) cells are also shown. Note that endocrine and tuft cells in both the corpus and antrum have been omitted. The gastric fundus has been outlined. Corpus and antrum units have been adapted from Willet and Mills⁹. The human stomach was modified from an original image obtained from turbosquid.com.

Our understanding of the mechanisms regulating the anatomic and cellular specification of the gastric corpus and antrum are still limited relative to what is known about organogenesis in most other organs. However, what is known has been recently summarized^6,9 and will not be further discussed in this review. Instead, we will focus on how the stomach responds to injury. In a sense, the stomach is in a constant state of injury: large concentrations of ingested toxins, microbes, and physically damaging objects can spend hours in the stomach^25,26. A rich network of capillaries^27,28, with their inherent risk for hemorrhage, is separated from all gastric contents by as little as a single epithelial cell layer and the mucus elaborated by those cells. The potential insults to the stomach are largely neutralized by stomach acid, which can kill up to 10 billion microbes per hour²⁹. Early physiologists like Spallanzani were fascinated by the “purifying” nature of stomach secretions and noted that these secretions could even degrade injurious objects like ingested needles². However, gastric acid in itself is potentially harmful, so the stomach has evolved mechanisms to protect against its own principal weapon. As the stomach essentially uses the same mechanisms to respond to any potential irritant, we propose here that simply focusing on acid is an effective method for understanding patterns of injury and response in the stomach. We believe that this approach is easier than trying to categorize responses to any of the myriad potential environmental insults or even using the common pathological concepts of acute versus chronic inflammation^30–33. We condense the response to gastric injury into two basic patterns, one in which the acid that usually protects the stomach inappropriately damages the stomach lining and one in which acid production is impaired.

Production of and Protection against Stomach Acid: the Superficial Response

Gastric acid secretion occurs through the acetylcholine-, histamine-, and gastrin-stimulated release of hydrochloric acid by parietal cells in the gastric corpus³⁴. Gastric acid provides a highly effective, innate microbial filter that can regulate the microbiota of the entire gut³⁵. Indeed, pathophysiologic or iatrogenic increases in gastric pH can increase susceptibility to certain enteric infections and alter gastric³⁶ and enteric flora^37–40. Gastric acid can be equally self-injurious, however, as it can breach gastric, esophageal, or duodenal epithelial integrity to cause bleeding and/or perforation^41,42. In addition to this chemical injury, disruption of the mucosal barrier in the setting of sustained acid likely triggers a local inflammatory response, though this has been more definitively demonstrated in the esophagus in a gastroesophageal reflux disease model⁴³. Regardless, the gastric mucosa must balance gastric acidity while protecting against acid-induced damage. These same defense mechanisms protect not only against the corrosive effects of gastric acid but also against endogenous (e.g., pepsin^44,45, bile^32,46,47) and exogenous (e.g., alcohol^48,49, smoking⁵⁰) agents. The gastric mucosa maintains its protective barrier against these insults as part of a pattern of adaptation that we refer to as the superficial response. The main mechanisms that constitute the superficial response are the secretion of topical defenses, the regulation of local blood flow, and the rapid regeneration of surface epithelium.

Gastric epithelium elaborates a variety of protective factors that act to topically neutralize or limit acid-induced damage (Figure 2). Gastric mucus provides a viscous gel matrix composed of water, mucin, electrolytes, and host and bacterial cellular components that serves to neutralize local acid production⁵¹. In addition to the bicarbonate and non-bicarbonate⁵² buffers that are retained in the mucus network⁵³ and are primarily derived from the surface epithelium⁴⁵, phospholipids within the mucus layer hinder the back diffusion of secreted protons⁵⁴. Among the major constituents of the mucus layer, mucins, such as MUC5AC⁵⁵, are glycoproteins that are predominantly secreted by surface/pit cells, and their production is regulated by acid secretagogues (acetylcholine, gastrin, histamine) as well as paracrine factors (NO, EGF, HGF) through distinct mechanisms⁵¹. Trefoil factor family proteins (TFFs) are co-secreted with mucins⁴⁵ and work to enhance the viscoelastic properties of the mucus gel⁵⁶. Mucin expression profiles also correlate with stages of mucosal regeneration following acid-induced injury^57–59.

Figure 2. The superficial and glandular responses in the gastric corpus.

The corpus unit (left) responds to gastric injury through two main mechanisms, the superficial response (a) and the glandular response (b). (a) The surface epithelium (left) consists of pit cells (purple) that produce a viscous mucus barrier that protects against endogenous (e.g., acid, denoted by H⁺) and exogenous (not shown) injury. Breaches in the surface epithelium resulting from acid oversecretion, decreased mucus production, and/or ischemia can lead to ulcers and/or bleeding (middle). The superficial injury response restores the protective barrier of the surface epithelium by increasing mucus production, restoring local blood flow, and re-establishing epithelial integrity through restitution and cellular proliferation (right). (b) The glandular injury response correlates with the loss of acid-producing parietal cells (blue) and a replacement of the deeper glandular epithelium with metaplastic cells (yellow) that co-express markers of mucous neck cells (green) and chief cells (red). This metaplastic response (middle) is often termed Spasmolytic Polypeptide-Expressing Metaplasia (SPEM) and represents a transient response to re-establish homeostasis (right). Pre-pit (light purple), pre-parietal (light blue), and pre-mucous neck (light green) cells are also shown. The superficial and glandular responses are not mutually exclusive and can occur simultaneously within the same corpus unit.

Prostaglandins promote topical gastric protection by inhibiting acid production and stimulating mucus and bicarbonate secretion. Andre Robert first coined the term “adaptive cytoprotection,”⁶⁰ referring to the protective effect of endogenous prostaglandins in response to mild gastric irritants that induced acid damage. These agents stimulate a cross-protective response, such that prostaglandins generated from mild gastric injury confer additional protection against subsequent, more severe injury⁶¹. For example, prostaglandin E2 (PGE₂) is produced by the constitutively expressed enzymes cyclo-oxygenase 1 (COX-1 or PTGS) and cyclo-oxygenase 2 (COX-2 or PTGS2), whose expression is greatly increased following injury. PGE₂ works through its EP3 receptors on parietal cells to directly inhibit acid secretion and indirectly on enterochromaffin-like (ECL) cells by blocking release of the acid secretagogue histamine⁶². It also stimulates mucus and bicarbonate secretion through EP1 and EP4 receptors⁶³. The importance of constitutive and inducible production of PGE2 has been shown in numerous types of injury. The stomachs of cyclooxygenase-1-deficient (Ptgs1^−/−) mice were more prone to two photon-induced gastric mucosal damage and less able to mount an epithelial protective response compared to Ptgs1^+/− mice⁶⁴. This defect could be corrected with exogenous dimethyl-PGE₂ administration. Finally, non-steroidal anti-inflammatory drugs (NSAIDs) that inhibit constitutive prostaglandin production are one of the most common etiologies of peptic injury, and this has been extensively studied^42,65–67.

An additional mechanism for protecting against the detrimental effects of acid is the regulation of mucosal blood flow to the surface epithelium⁶⁸. Mucosal blood flow affects acid balance in various ways. For one, parietal cells secrete angiogenic factors, like vascular endothelial growth factor B^69,70 (VEGFB), that help to maintain ample capillary networks to supply oxygen to their abundant mitochondria and provide the energy to efficiently pump acid. On the other hand, acute ischemia induces aberrant acid production and can cause focal erosions in the mucosal surface of corpus glands. In some cases, prolonged ischemia can lead to ulcers, as can be seen following central nervous system injuries (Cushing ulcers)⁷¹ or massive burns (Curling ulcers)⁷². Other angiogenic factors, such as basic fibroblast growth factor^73,74 and vascular endothelial growth factor A⁷⁵, typically help to limit acid-induced damage by promoting the re-establishment of the microvascular network. The restoration of blood flow also allows byproducts of acid production and other toxic metabolites to be carried away⁷⁶. However, significant epithelial damage can occur during reperfusion of the gastric mucosa following an ischemic injury, as the acute restoration of oxygen and the infiltration of immune cells can result in the local production of reactive oxygen species⁷⁷.

The vasodilator nitric oxide (NO) also plays a critical role in the superficial response. On the one hand, it maintains resting gastric blood flow to limit cellular damage⁷⁸. On the other, NO provides additional protection through mechanisms that parallel those of prostaglandins^79–81, including enhancing mucus production and limiting acid secretion. More importantly, the protective effects of NO and prostaglandins appear to be cooperative⁸², with NO providing a compensatory level of superficial protection in the setting of NSAID-induced gastric mucosal damage. Indeed, the injurious effects of inhibiting prostaglandin synthesis could be attenuated by exogenous administration of NO donors^83,84. As a result, novel hybrid drug derivatives consisting of NSAIDs coupled to NO donor moieties have been developed as more gastro-protective therapeutic alternatives^85,86, and some of these have shown promise in reducing the risk for peptic ulcer bleeding among chronic NSAID users⁸⁷.

The final wing of the superficial response to damage is the rapid adaptation of the pit/surface cells themselves. In addition to limiting the extent and duration of peptic injury through alterations in the mucus layer, acid secretion, and local blood flow, the acute adaptation to acid-induced damage equally relies on reconstituting mucosal integrity by restoring cellular junctions (restitution) and through the rapid regeneration of surface epithelial cells (proliferation). Restitution, which occurs within minutes, involves the rapid migration of pit cells to cover exposed basal lamina following acute ethanol injury⁸⁸, for example. In addition to rapid cellular migration, pit cells can adapt to acid-induced injury by increasing the rate of proliferation of their progenitors. The ultimate progenitors for pit/surface mucous cells are thought to be stem cells within the isthmus, the region between the gastric pits and the deeper gastric glands (Figure 1). We note, however, that it has not been formally proven that pit cells come from a multi-potent stem cell as opposed to from a long-lived progenitor that is dedicated to making only pit cells. We have discussed the uncertainty in the field in a recent review⁹. Regardless, what is clear is that pit cell precursors emerge from their progenitors in the isthmus and subsequently differentiate during their upward migration to the luminal surface, a process that occurs within days⁸⁹. Surface mucous cells then undergo cell death and are either phagocytosed by a neighboring cell or extruded to the surface, a shedding process that may serve to prevent gastric micro-organisms from gaining a foothold within the gastric gland⁹⁰.

The cellular regeneration and proliferation of surface epithelium are also enhanced by topical and paracrine factors secreted by gastric epithelium. For example, prostaglandins exert proliferative effects in addition to their locally protective topical properties. Early studies in acutely injured rat stomachs demonstrated that gastric epithelial regeneration occurred more rapidly and efficiently in rats pretreated with prostaglandins⁹¹. Growth factors derived from mesenchymal components, though not addressed in this review, have also been shown to play a role in pit cell differentiation in vitro²³. Mitogenic factors of the epidermal growth factor family (e.g., EGF^92,93, TGF-α⁹³, amphiregulin^93,94, HB-EGF⁹⁵) promote the proliferation of pit progenitor cells and the regeneration of surface epithelium. Some of the earliest studies on TGF-α, for example, identified its enrichment in mucous neck cells and parietal cells⁹³, and subsequent evidence has illustrated its paracrine effects on the gastric gland, including inhibiting acid secretion, increasing gastric mucin levels, and stimulating epithelial restitution⁹⁶. Its expression is upregulated following acute mucosal injury in mice^97,98 and promotes cell migration and cellular proliferation during ulcer healing^99–101. When transgenically expressed under the regulation of the metallothionein promoter, TGF-α induced severe hypertrophic gastropathy and achlorhydria in mouse stomachs¹⁰², highly reminiscent of Ménétrier’s disease, a rare, human gastric premalignant condition that can be treated with anti-EGF/TGF-α therapy¹⁰³.

Taken together, the stomach’s superficial response entails a rapid adaptation to protect against acid-induced breaches in epithelial integrity (Figure 2a). This response relies on the secretion of topical protective factors to neutralize acid’s corrosive effects, the regulation of mucosal blood flow to limit the duration and extent of injury, and the regeneration of surface epithelium through restitution and proliferation.

Defining the Glandular Response: the Initiation of SPEM

Under certain types of gastric injury, the gastric mucosa can undergo a second type of adaptation, which we will refer to as the glandular response. In contrast to the superficial response, which aims at protecting against the corrosive effects of endogenous acid, the glandular response occurs when acid production is compromised or lost. Specifically, this injury response is characterized by the loss (atrophy) of acid-producing (oxyntic) parietal cells. In extreme cases of oxyntic atrophy, all parietal cells and chief cells within a gastric unit simply die¹⁰⁴, resulting in gland dropout with only surface/pit cells remaining. The principal histological pattern seen in response to oxyntic atrophy, however, is a repopulation of the gland, depleted of mature parietal cells and chief cells, with metaplastic cells¹⁰⁵. The metaplastic cells constitute a hybrid phenotype, co-expressing proteins like TFF2, normally expressed by the chief cell progenitor mucous neck cells, and pepsinogen, a digestive enzyme normally expressed by mature chief cells (Figure 2b). This pattern of corpus glandular differentiation has been termed Spasmolytic Polypeptide-Expressing Metaplasia (SPEM)^106–109, as it is defined by cells deep in the gland (where chief cells normally reside) that express spasmolytic polypeptide (also known as TFF2). SPEM represents a metaplastic response inherent to the gastric corpus, as the gastric antrum is largely devoid of parietal cells and chief cells¹¹⁰.

Though oxyntic atrophy technically means the loss of acid-secreting cells, both parietal cells and mature chief cells seem to be invariably absent. Accordingly, in multiple murine injury models^107,111, SPEM has been assumed to be specifically triggered by parietal cell death. Indeed, two of these acute injury models (DMP-777¹¹² and high-dose tamoxifen¹¹³) likely act through the H⁺/K⁺-ATPase, and their effects can be mitigated by pre-treatment with the proton pump inhibitor, omeprazole. However, a recent study in mice showed that targeted ablation of parietal cells failed to initiate SPEM¹¹⁴. Thus, the initiation of SPEM may require parietal cell loss as well as additional injury along the glandular axis and/or specific immune or mesenchymal cell changes. Various epithelial (e.g., NOTCH1/2¹¹⁵, Shh^116,117, gastrin^112,118,119), mesenchymal (e.g., BMP-4¹²⁰, noggin¹²¹), and immune factors (e.g., IFNγ¹²², IL-1β^123,124, IL-33¹²⁵, IL-13¹²⁶) contribute to the establishment of a metaplastic milieu and may play a role either in loss of parietal cells or induction of SPEM. Regardless, the glandular response during oxyntic atrophy represents a unique mechanism for adaptation and repair that highlights epithelial plasticity in the gastric corpus.

Superficial Versus Glandular Injury: Implications for Gastric Disease

The utility of distinguishing superficial from glandular injury is not academic. Previous methods for categorizing injury in the stomach have relied on the duration of injury (acute versus chronic) in the presence (gastritis) or absence (gastropathy) of mucosal inflammation^31,127. Acute gastritis, to a pathologist, is identified by local infiltration of polymorphonuclear leukocytes (neutrophils) to the area of injury, whereas a chronic gastritis is characterized by a mononuclear infiltrate¹²⁸. These distinctions are based on histopathology but are confounded by the fact that the same injurious agents can produce overlapping effects. Chronic conditions, like infection with the stomach-adapted bacterium Helicobacter pylori (Hp), can cause repeated bouts of acute inflammation¹²⁹. Acute foci of inflammation from regional biopsies can be superimposed on a background of chronic inflammatory cell infiltrate. Agents that typically induce gastropathy (e.g., refluxed bile, alcohol, non-steroidal anti-inflammatory drugs) can also lead to gastritis¹³⁰. In the case of chronic inactive gastritis secondary to portal hypertension, for example, the resulting injury is characterized by mucosal damage both with and without inflammation¹³¹. Inflammatory responses to chronic gastropathic diseases, like Ménétrier’s disease¹⁰³ and Zollinger-Ellison syndrome¹³² (gastropathy secondary to a gastrin-producing endocrine tumor), can be variable.

As a result, we would argue that a clearer method for categorizing gastric injuries is to group them as problems of either excessive or insufficient acid. The stomach’s response to injury is best understood, as discussed above, as either superficial or glandular adaptations. This system cleaves closer to the underlying etiologies and observed responses than do other systems that force often overlapping disease patterns into categories based on temporal (acute versus chronic) or inflammatory (gastropathy versus gastritis) properties. The superficial response, as previously described, describes a mucosal adaptation to a pathophysiologic increase in acid. It enhances acid-protective mechanisms predominantly in the superficial cells to protect the epithelium from the corrosive effects of exogenous (e.g., ingested acids¹³³) or endogenous (e.g., acid, bile⁴⁷) substances, mechanical trauma (e.g., nasogastric tube placement¹³⁴), and/or microbial infections^135,136.

The glandular response, on the other hand, is an adaptation of epithelial lineages to an injury affecting the gastric gland, the portion of the gastric unit extending from the isthmus down to the base. Though it is possible that the stomach can undergo a glandular adaptation following substantial injury to any of the glandular cell lineages, current data have shown that parietal cell loss correlates with the glandular response¹⁰⁵. In animal models, glandular adaptation is not as rapid as the superficial injury response, though it can still occur on the order of days¹⁰⁷. It has been hypothesized that focal glandular responses might occur sporadically throughout the human stomach¹³⁷, for example, but these have only been seen in cases of sustained damage to glandular cells. The types of damage that can induce a glandular response can be broadly classified into infectious and non-infectious, with chronic Hp infection accounting for the majority of infectious cases. The most representative example of a non-infectious, purely glandular injury is autoimmune gastritis, an autoimmune disease that targets parietal cells, ultimately leading to hypo/achlorhydria and pernicious anemia¹³⁸. Interestingly, some have regarded autoimmune gastritis as an autoimmune manifestation of Hp-induced chronic gastritis^139–142.

Distinguishing superficial from glandular adaptation also helps us to understand the etiologies and mechanisms underlying the responses to the disparate insults that the stomach faces. One key example is the pathophysiology of Hp infection. Approximately half of the world is infected with Hp¹⁴³, though the majority of those infected will remain asymptomatic and histologically manifest as focal acute gastritis with superficial injuries. Approximately 10–15% will at one point develop clinically significant peptic ulcer disease^144,145: Hp infection causes more than half of gastric and essentially all duodenal ulcers^146,147. In the case of a gastric ulcer, the balance between endogenous acid production and mucosal protection is disrupted¹⁴⁸. For an ulcer to develop, this disruption must be relatively chronic, yet the injury response largely involves superficial mechanisms with little effect on cells deeper in the gland. Indeed, a gastric ulcer would be far less likely to occur if it induced a large-scale glandular adaptation in the stomach, as the glandular response is characterized by the loss of parietal cells, the source of acid. Note that focal glandular adaptations in the few gastric units bordering a chronic ulcer may occur, but the glandular adaptation during peptic ulcer disease is not extensive¹⁴⁹. In contrast, a minority (~1%) of patients infected with Hp will develop gastric adenocarcinoma¹⁵⁰. In these cases, the tumors nearly invariably arise in patients who have developed an extensive, glandular adaptation of the corpus with oxyntic atrophy and metaplastic chief cell differentiation^105,151. Interestingly, epidemiological studies have shown that those patients who develop ulcers from Hp, specifically duodenal ulcers secondary to acid oversecretion, may have a lower incidence of gastric cancer compared to the general population¹⁵², though some have argued against this¹⁵³. The oncogenic risk therefore correlates with the type of injury response to Hp infection, regardless of the histologic presence of acute or chronic inflammation. Patients who respond to chronic Hp infection with exclusively superficial mechanisms are less likely to progress to gastric cancer than those whose stomachs undergo glandular adaptations¹⁵⁴. The reasons behind the inherent oncogenic potential of these glandular/metaplastic changes will be discussed later.

In summary, the adaptive mechanisms that underlie the superficial and glandular injury responses are fundamentally different at the cellular level and confer distinct oncogenic risks. Our overall understanding of the superficial injury response is more comprehensive, perhaps because this response is mechanistically less complex. Simplistically, the superficial injury response relies on replenishing the pit cell lineage through cellular death, restitution, and proliferation. While the surface epithelium can respond to glandular injury signals (e.g., EGF^92,93, TGF-α⁹³, gastrin¹¹²), this usually results in pit cell proliferation and an expansion of the pit region, a process known as foveolar hyperplasia⁹⁶. Glandular injury, on the other hand, induces a deeper epithelial response involving multiple lineages along the gastric unit. The glandular response can be transient or prolonged, and the mechanisms for its regulation are currently unclear. Our understanding of the initiation, expansion, and pathologic significance of the glandular response is therefore still evolving. It is worth noting that, because glandular adaptation is a response to parietal cell loss and the metaplastic differentiation of chief cells^105,108, it is a process that is best understood in the gastric corpus, whereas the superficial response can occur throughout the stomach. Though it has not been formally studied, it is possible that, because human antral units can harbor parietal and chief cells, some aspects of the glandular response might also occur in these antral units¹⁵⁵.

The remainder of the review will illustrate our current understanding of the glandular response. As we are concerned with the cellular and molecular mechanisms underlying the glandular adaptation to injury, we will refer from here on out to this pattern of injury as SPEM. Which cells are important for repopulating the gland during SPEM? How might SPEM progress through the stomach, and why might this be pathologically relevant? More importantly, what is unique about the establishment and expansion of SPEM that harbors its oncogenic potential?

Stem Cell Dynamics and the Cell of Origin for Metaplasia

The cellular origin(s) of SPEM remains an area of debate, in part because a resident stem cell population has not been definitively identified in the adult gastric corpus the way other specific stem cell populations have been confirmed in organs like skin and intestine⁹. If it is not clear which cells are actually performing this quotidian progenitor role, how can there be consensus on the cell of origin in a pathophysiologic pattern of differentiation? Here, we highlight various independent lines of evidence supporting a model whereby the cellular origins of SPEM may be fluid, illustrating that the stomach may be plastic with regard to the source of cells that respond to injury and fuel metaplasia. Such plasticity has become recognized as an established feature of other organs like skin^156,157 and intestine^156,158. Hypotheses related to the roles of stem cells in gastric corpus and antrum have been recently reviewed^6,9,159,160. There is certainly consensus supporting the conclusions from the landmark pulse-chase labeling experiments of Karam and Leblond, which identified that the most actively proliferating and ultra-structurally least differentiated cells (granule-free) reside within the presumptive isthmus zone between the surface/pit and the gland¹⁶¹. The isthmus has the most morphologically undifferentiated, mitotically active cells in both the corpus and antrum, though its location along the gastric unit varies, as it is found about a third of the way down from the lumen in corpus units and about a third of the way up from the base in antral units (Figure 1). In continuous labeled nucleotide infusion experiments, labeled cells arise in the isthmus and eventually spread to all cell lineages along the gland unit^89,162,163. Based on these studies, it has been the canon that the isthmus harbored a constitutively active multi-potent stem cell that could replenish all of the mature lineages within the gland on a daily basis. While there is broad consensus that labeled nucleotides do indeed spread bi-directionally at varying rates from the isthmus, some have questioned the interpretation of certain aspects of these studies^9,164, and further experiments are warranted.

There is more definitive information about the behavior of stem cells in the antrum. Lgr5 was initially identified as a stem cell marker in the gastric antrum¹⁶⁵, and subsequent lineage tracing experiments identified Cck2r as a marker of an antral stem cell with the potential to give rise to Lgr5+ antral cells¹⁶⁶. We discuss the advantages and caveats with using lineage tracing to characterize stem cell dynamics in the BOX. Glandular injury following Hp colonization was recently shown to activate and expand Lgr5+ cells in the antrum¹⁶⁷. However, Lgr5+ antral cells are located at the gland base (and not in the isthmus region). Most of the cells in the antral gland base are differentiated, deep mucus cells that happen to be the most differentiated cells within the gastric unit, lacking the “granule-free,” primitive morphological characteristics described by Karam and Leblond in the corpus¹⁶¹ and by Lee and Leblond in the antrum¹⁶⁸. It is possible that Lgr5+ undifferentiated stem cells hide out in this niche, but this has not been definitively shown; interpretation may be compounded by the fact that the original tool used to identify Lgr5+ cells (Lgr5^{CreERT2-iRes-EGFP}) shows mosaic expression¹⁶⁹. Unlike in the corpus, turnover of all antral cells, except endocrine cells, has long been known to be rapid¹⁷⁰. In addition, multiple promoters that drive expression of reporter genes can show “stem cell” activity upon lineage tracing^171–173. For example, in addition to promoters for Lgr5 and Cck2r, it has been shown that even promoter elements for the intestine-specific gene Villin¹⁷⁴, or for a gene that is expressed in many adult gastric cell lineages and critical to stomach development (Sox2^115,175,176), can lineage trace in the antrum.

BOX.

Lineage tracing: utility and pitfalls

Lineage tracing occurs when a single, index progenitor cell is identified and its progeny traced¹⁷². The most common lineage tracing system uses genetic marking, wherein Cre recombinase expression is induced in an index cell also harboring a reporter gene, whose expression is induced when Cre recombines loxP sites in its promoter. Cre expression leads to an irreversible induction of the reporter (e.g., a fluorescent gene like dsTomato), which will thereafter be expressed in all progeny of the index cell. The inducible Cre-loxP system has become indispensable for the in vivo study of tissue development in homeostasis and in injury, though results must be interpreted with possible confounders in mind. One should consider: 1) how the mouse pedigree with a stem cell-specific gene promoter that governs Cre expression is generated (e.g., via traditional transgenic vs. BAC transgenic vs. recombination of Cre into the endogenous stem cell gene locus²⁴¹); 2) the system used to activate Cre in the index cell, for example by using a Cre construct designed to travel to the nucleus only when it binds an exogenous agent like the estrogen mimetic tamoxifen²⁴². In this case, tamoxifen is possibly confounding because it can also cause dose-dependent toxicity, especially in the stomach¹¹³; 3) that stem cells often transcribe low levels of differentiation-specific genes that can be sufficient to induce reporter expression even if the cells do not normally express detectable levels of the gene whose promoter was used to drive Cre expression²⁴³; 4) the efficiency of reporter gene induction; 5) that cells can migrate between the administration of the Cre-activating agent and when the reporter gene is transcribed and translated²⁴⁴; and 6) that there is likely inherent cellular plasticity, with differentiated cells being able to revert to a progenitor cell fate and subsequently lineage trace¹⁵⁸. Taken together, these limitations in lineage tracing experiments that rely on inducible Cre recombinase should be considered and appropriately controlled.

The literature, our own unpublished work, and anecdotal evidence in the field all support the conclusion that it is rare for a cell population in the antrum to not have stem cell potential in a lineage tracing experiment; to our knowledge, only some of the endocrine cell-specific promoters do not trace readily into other lineages⁹. Also, it has become clear that other organs, like intestine, are remarkably plastic, with fluid cell identities that are readily adaptable to different environmental conditions¹⁵⁶. Thus, we conclude that the antrum harbors cells with varying degrees of stochastic stem cell potential. Perhaps the isthmal cells, the cells showing the least differentiated morphology and the most propensity for proliferation, are the most likely to serve as stem cells on a daily basis, with other populations having varying likelihood of being recruited for that function. As in the intestine, injury can promote the recruitment of more differentiated cells back into the cell cycle^156,158. Recent work from Meyer and coworkers supports this assertion¹⁷⁷, as they demonstrate that the undifferentiated, isthmal antral cells are highly Wnt responsive and express the Wnt-responsive co-receptor Lgr4 at much higher levels than Lgr5. The Lgr5-high expressing cells are Lgr4-negative and located at the base of the gland. Both isthmal Lgr4+ and basal Lgr5+ populations can serve as stem cells, but the activity of Lgr4+ isthmal cells is more rapid at baseline. To fully understand stem cells and differentiation in the antrum, we will have to examine, as this group did, the relative potential for various cell populations to serve as stem cells under homeostasis and injury.

One method that could be used to quantitatively assess stem cell potential is to isolate individual cells and use them to grow gastric organoids^178–180 (“gastroids”) ex vivo, though this method has its caveats¹⁸¹. Single CCK2R+¹⁶⁶, LGR5+¹⁶⁵, and LGR4+¹⁷⁷ cells have all been shown to have the capacity to serve as stem cells for gastroid formation. We reason that the raison d’être for stem cell activity and plasticity in the antrum is to support the superficial response to injury described above, as there is currently only scant evidence to suggest that the antrum undergoes a glandular-like response akin to that of the corpus (i.e., a metaplastic response involving a change in differentiation state of the glandular cells)¹⁵⁵. Applying antral stem cell dynamics as a guide for understanding stem cell behavior during the glandular response in the corpus may therefore not be accurate. However, there is considerable evidence that the corpus exhibits plasticity, specifically in the setting of glandular adaptation to injury. Indeed, metaplastic glands have been clearly observed in settings of corpus glandular injury, including Hp infection¹⁸², autoimmune gastritis^183,184, and various murine models for acutely inducing metaplasia¹⁰⁷. As a result, some cell within the corpus gastric unit must be able to exhibit plasticity by changing its identity, a process termed “trans-differentiation¹⁵⁶” by some. Though one of two candidate cells - either the undifferentiated cells in the isthmus (the presumptive stem cell discussed above) or the differentiated chief cell in the base of the unit - has been proposed, it is equally possible that both isthmal and chief cells can serve as the cell of origin for SPEM to varying degrees under different injury conditions.

Evidence of isthmal cell regenerative capacity has been implied in ex vivo gastric organoid models¹⁸⁰. In addition, acute injury models that rapidly and reversibly induce SPEM show increased proliferation in the isthmus of corpus glands^114,171,185. If the isthmal stem cell is the exclusive cell of origin for SPEM, as has been recently proposed^160,186, the model would imply several features of how metaplasia must unfold. The aspects of an isthmal cell serving as the cell of origin for SPEM have been schematized in Figure 3a. One aspect of this model depends on the fact that nearly all of the SPEM glands in mice and in humans are depleted of parietal cells, lack normal chief cells, and are instead populated nearly exclusively with metaplastic cells. This would assume that, for the stem cell to populate the gastric unit with metaplastic cells, the parietal cells, chief cells, and (likely) mucous neck cells must die and be replaced by metaplastic cells derived from a trans-differentiated stem cell. If any mucous neck cells remained, the cell fueling the more basal SPEM cells would have to migrate from their origin in the isthmus through this neck cell region. In murine models for acutely inducing SPEM, where SPEM peaks by 3 days^111,113,187, an isthmal cell of origin model would mean that the isthmal cell would have to produce SPEM progeny to replenish the entire neck and base of the unit at a remarkable rate of cell division. Moreover, a focus of proliferation can be seen at the gland base within one to two days after SPEM induction^113,169, implying that the isthmal cell would need to migrate from the isthmus to the gland base within this time frame.

Figure 3. Distinguishing the possible cellular origins of SPEM.

Three possible mechanisms for the initiation of SPEM and the replacement of depleted glandular epithelium are presented. For the isthmal stem cell (white) to serve as the unique cell of origin for SPEM (a)¹⁸⁶, all of the glandular cells - parietal cells (blue), chief cells (red), and mucous neck cells (green) - must die and be replaced with cells derived from a proliferative, isthmal stem cell (yellow) that has trans-differentiated to become a stem cell for metaplastic cells. (b) Others^108,195 have hypothesized that SPEM arises from the trans-differentiation of chief cells into metaplastic cells following the loss of parietal cells. These basal metaplastic cells (yellow and orange) repopulate the corpus unit from the base up, while isthmal proliferation results only in an expansion of pit cells (purple), a process known as foveolar hyperplasia. Another hybrid model can also be envisioned wherein both the isthmal stem cells and chief cells can contribute to SPEM (c). In this scenario, two foci of proliferation may be dedicated to generating different cells along the gland axis, with the isthmal stem cell giving rise to pit cell precursors (light purple) and/or mucous neck cell precursors (light green), and the chief cells giving rise to metaplastic cells (yellow and orange). Bacteria represent Helicobacter pylori.

The other extreme for the cellular origin of SPEM argues that all metaplastic cells are derived from chief cells that undergo cellular reprogramming (Figure 3b). This model has been proposed and substantiated by Goldenring and colleagues over the last decade^{169,188–190}. Along with the Goldenring group, we and others have also independently observed that chief cells can reprogram to serve as proliferating cells^{114,190–192}. In particular, the evidence for chief cell plasticity has been demonstrated by lineage tracing using Cre^ERT2 driven by the differentiated chief cell marker Mist1¹⁷¹, elements of the Runx1 promoter (known as eR1)¹⁹⁰, Troy (Tnfrsf19)¹⁹¹, and most recently Lgr5¹⁶⁹. Using an expression cassette that better matches endogenous expression of the Lgr5 gene, Barker and colleagues discovered that LGR5 labels chief cells deep within the gland^169,193. Additionally, Mist1, eR1, and Lgr5 have all been shown to cause metaplasia and even dysplasia if these promoters drive expression of an oncogenic K-Ras. In those studies, SPEM was clearly characterized by both an increase in proliferation of the isthmal cells and an induction in proliferation of chief cells at the base. We have also noted that, in human pathology sections, cells with phenotypes that are transitional between chief and SPEM cells tend to occur in the base¹⁹⁴. Because the chief cells reside at the gland base, we interpret those observations as indicating that the transition in differentiation between normal and SPEM occurs in chief cells. Finally, in unpublished studies, we find that treatment with 5-fluorouracil, an inhibitor of DNA synthesis, has little effect on the induction of SPEM, indicating that SPEM can occur in the absence of isthmal cell proliferation (Radyk and Mills, unpublished).

The preponderance of evidence therefore indicates that chief cells serve as the predominant cell of origin for SPEM¹⁹⁵. That the post-mitotic, differentiated chief cell has the ability to reprogram and behave as a proliferative, metaplastic cell is not unique to stomach, given the emerging parallel literature of cellular reprogramming in other gastrointestinal tissues^156,158,196. For example, in the pancreas, which does not have a resident stem cell, the zymogenic acinar cell exhibits plasticity in response to injury by co-expressing ductal markers and becoming proliferative, a process known as acinar-to-ductal metaplasia^197–199. Recent data suggest that shared cellular mechanisms may exist to allow post-mitotic cells, like chief cells and acinar cells, to be recruited back into the cell cycle. This process is initiated by an auto-degradative phase in which existing secretory architecture is recycled, followed by the induction of metaplastic genes, and concludes with the proliferation of metaplastic cells^158,192,200.

The magnitude and extent of gastric injury likely dictate the glandular response, and a single focus of proliferation and metaplasia may not represent the most efficient method to fuel repair. Most of the murine models for inducing SPEM result in a rapid, simultaneous, pan-gastric injury that results in the appearance of two distinct foci of proliferative activity in each gastric unit¹⁰⁷. In addition to a focus of proliferation at the gland base, proliferation at the isthmus is clearly seen, and these isthmal cells could contribute to metaplasia migrating down from the isthmus (Figure 3c). Two foci of epithelial regeneration within the gland would allow for a more rapid and efficient reconstitution of injured epithelium. In the setting of a chronic injury which gradually injures parietal cells and chief cells, it is possible that two regenerative foci would not be required. Another possibility is that the two zones of proliferation during the glandular response are devoted to generating different cells. The isthmus will continue to generate pit cells and may also generate new parietal cells and/or mucous neck cells, whereas proliferating cells from deeper in the gland may be dedicated to regenerating chief cells. It should also be mentioned that, while the chief cell has been identified as a cell capable of reprogramming and answering the call for repair, other cell types along the gland axis (i.e., the mucous neck cell) may be capable of serving a similar role, though this has never been demonstrated. This would imply that multiple lineages of cells along the corpus glandular unit, similar to the previously described antral unit, exhibit an inherent stemness that allows the gland to efficiently respond to injury.

Why Hp Might Benefit from SPEM

As we continue to explore the cellular mechanisms behind the initiation of SPEM, we must equally consider the significance of the expansion of SPEM. In human patients, SPEM appears to slowly expand gland-by-gland along a front extending proximally from the corpus/antrum transition. The leading edge of metaplasia has sometimes been referred to as the “atrophic front,”¹⁵³ a progressing zone of inflammation that moves proximally from the antrum to the corpus along the lesser curvature. The atrophic front blurs the normal, sharp, histological transition between corpus and antrum. As a result, this transition becomes dynamic following chronic glandular adaptation. While the corpus and antrum are functionally and developmentally distinct⁷, glandular adaptation in the corpus (SPEM) causes the corpus to histologically resemble the antrum, a process that has been termed “antralization” of the corpus^201–204. Glandular adaptation to injury in the corpus has taken on various names (atrophic gastritis^30,205,206, oxyntic atrophy¹⁰⁵, pseudopyloric metaplasia^207,208, SPEM¹⁰⁹) which, in our opinion, represent the same cellular and molecular processes. Antralization of the corpus does not mean that the corpus is converted into antrum²⁰⁹, but morphologic and molecular similarities can be seen between corpus units, that have lost their parietal cells and chief cells, and antral units.

How and why might chronic Hp infection lead to an antralization of the corpus? In chronically infected patients that eventually develop extensive atrophy and metaplasia, the natural course of chronic Hp infection mimics the pattern of extension of the atrophic front. It has been proposed that the corpus/antrum transition is initially colonized by Hp and serves as a critical niche for the establishment of a chronic infection²¹⁰. Why Hp may target or hone in on this region is unclear, but, teleogically, this relatively short span of hybrid glands²¹¹ may represent the first hospitable micro-environment that this bacterium encounters in terms of favorable intra-gastric pH. In the setting of Hp infection and its associated inflammation, we propose that SPEM first arises from this corpus/antrum transition and progresses towards the corpus as more and more glands undergo antralization (Figure 4a). It is also possible, though less likely, that several areas of Hp colonization distributed throughout the corpus can each give rise to foci of SPEM expansion. Over decades, Hp can expand its niche along this atrophic front, progressing into the corpus and resulting in a pan-gastritis, a risk factor for the development of gastric adenocarcinoma²¹². The topographic spread of Hp may represent the bacterium’s unique adaptation to a changing environment.

Figure 4. The expansion of SPEM and the Cyclical Hit model: a possible mechanism for dysplasia.

(a) The topographic expansion of SPEM during sustained glandular injury results in the “antralization” of the corpus, such that corpus units become metaplastic and morphologically resemble antral units. The antralization of the corpus likely emerges from an initial focus of metaplasia (boxed red area) at the transition between corpus (purple) and antrum (orange) and expands proximally along the lesser curvature before spreading to the greater curvature. The transition zone (light purple) represents a dynamic, hybrid region that progresses along the leading edge of antralization. The earliest sites of antralization will have the longest history of metaplasia with associated dedifferentiation-redifferentiation cycles, increased risk of accumulation of mutations, and an increased likelihood that those mutations will seed dysplasia/neoplasia: this increased risk is denoted by an increasing color saturation of the boxed region at the corpus-antrum border. (b) The Cyclical Hit Model for the development of gastric dysplasia is presented. As post-mitotic chief cells become metaplastic and re-enter the cell cycle to proliferate and fuel metaplasia, they can accumulate genetic mutations (shown with yellow outlined symbol) through replicative stress. Chief cells within the initial focus of metaplasia (boxed area in a), a region that would have sustained the longest duration of glandular injury, harbor genetic mutations that can become unmasked and prevent re-differentiation, either leading to apoptosis or potentially serving as a cell of origin for dysplasia.

The host and microbial factors that determine these patterns of Hp colonization and spread remain largely unexplored, however. In the setting of chronic Hp infection, inflammation likely indirectly promotes gastric cellular reprogramming, but Hp colonization may also actively drive the expansion of SPEM through the elaboration of virulence factors like CagA^213,214 and VacA²¹⁵ that promote inflammation and glandular reorganization. From the point of view of Hp, the glandular response of the corpus (i.e., SPEM) may be a method for the bacterium to expand its niche. The reorganization of the corpus glands renders them more like antrum, for which Hp has an affinity, at least early in its pathogenesis. The metaplastic cells in SPEM are also proliferative far deeper into the gland than the normal isthmus²¹⁶, and Hp has been shown to actively interact with this proliferative zone²¹⁷. In short, Hp may exhibit a tropism for SPEM glands and hijack the very alterations in the gastric landscape that it has induced.

The Oncogenic Potential of SPEM

The initiation and expansion of SPEM likely have clinical significance in terms of explaining the distribution of gastric cancer. A conundrum in the gastric cancer field has been the disconnect between the topographic distribution of glandular injury and the anatomic location of gastric tumors. Given the strong epidemiological link between pan-gastritis and extensive metaplasia in the corpus, it has been assumed that loss of parietal cells and chief cells in the corpus is a prerequisite for the development of gastric adenocarcinoma²¹⁸. What has been confounding, however, is that the majority of human gastric adenocarcinomas seem to arise within the antrum or at the corpus/antrum transition^219–221, suggesting that parietal/chief cell loss and metaplasia may simply be a surrogate marker for the overall state of chronic inflammation in the stomach. Perhaps this inflammatory state promotes tumors in the antrum, but this does not necessarily mean that the metaplastic/atrophic tissue is itself the origin of gastric cancer.

Rather than being epiphenomenological, we propose that these metaplastic, transitional areas represent regions where many gastric cancers likely arise. Hp primarily colonizes the antrum during the initial establishment of infection. As it expands its niche and promotes the progression of an atrophic front, the first glands undergoing metaplasia would be those with parietal cells and chief cells along the border of the corpus and antrum (Fig. 4a). What makes metaplastic tissue particularly prone to becoming neoplastic? We must first acknowledge that SPEM is a normal, transient, glandular response to injury for re-establishing homeostasis. We believe that SPEM is fueled by differentiated cells re-entering the cell cycle, proliferating, and re-differentiating. One could imagine that repeated rounds of proliferation/differentiation cycles increase the chances of acquiring mutations via replicative stress (Figure 4b). Studies in pancreas and other models of tumorigenesis indicate that certain oncogenic mutations, like constitutively active K-Ras, do not have an effect in differentiated cells but can be unmasked when they are expressed in proliferating (i.e., metaplastic) cells^222–224. In the stomach, if these mutations do not block re-differentiation as the gland recovers from injury, these mutations can be harbored in quiescent, seemingly normal, differentiated chief cells. Over time, glands that have sustained the longest duration of injury will have accumulated substantial mutational burdens (Figures 4a and 4b). The corpus/antrum transition exemplifies one such area, as it likely represents the initial focus of Hp colonization and hence metaplasia. We also know that this region shifts as the corpus gradually becomes antralized during chronic Hp infection²⁰¹. Endoscopic mapping studies looking at the patterns of metaplasia that predicted progression to gastric adenocarcinoma found that the extension of metaplasia along the lesser curvature, known as the “magenstrasse” (German for “narrow street”) pattern, carried a significantly greater cancer risk than other observed patterns²²⁵. This high-risk pattern would correlate with the progression of the atrophic front and expansion of SPEM, arising from the corpus/antrum transition, the initial focus of metaplasia.

A parallel argument could be made for cancers at the gastro-esophageal junction, where a similar transition zone between squamous mucosa of the esophagus and glandular mucosa of the stomach may represent an area of increased metaplastic (and oncogenic) potential^226–228. Indeed, gastro-esophageal cancers are increasingly common²²⁹ and appear to be more similar to gastric cancer at the molecular level²³⁰. One could similarly imagine the distal expansion (i.e., toward stomach, away from esophagus) of this gastro-esophageal transition zone, with a focus of metaplasia at the gastro-esophageal transition (referred to as the cardia in humans)²³¹ serving as a possible starting point for the expansion of metaplasia. Unlike metaplasia at the corpus/antrum transition, metaplasia in the gastro-esophageal transition zone is not usually caused by Hp but rather correlates with acid and/or bile exposure that can be linked to metaplasia of the distal esophagus (i.e., Barrett’s metaplasia)^232,233. The cellular origins of Barrett’s metaplasia^234–236 and the possible expansion of metaplastic glands derived from the stomach into the esophagus have been proposed^237,238. As in the stomach, the focus of metaplasia at this transition zone and the subsequent spread of atrophy and metaplasia into the surrounding tissue may carry a similar risk for cancer, in this case proximal gastric adenocarcinomas.

Conclusions and Remaining Questions

In addition to serving as a highly efficient immune barrier for the gut, the stomach is a versatile organ that is capable of recovering from various forms of injury by relying on mechanisms for superficial and glandular adaptation. The superficial response, which largely centers on how the stomach protects itself from its own acid, has been relatively well studied. Here, we have focused on the less understood glandular response, which is of particular import because it epitomizes the delicate balance between re-establishing homeostasis and progressing to neoplasia, a theme common to multiple organs. We have presented evidence for the mechanisms and cellular origins of SPEM, an evolutionarily conserved mechanism for responding to glandular injury, though we still do not fully understand the cellular signals and mechanisms that regulate metaplasia in the stomach. While acute injury models for inducing SPEM suggest that this is a reversible process^107,113, how does reversion to homeostasis occur? Is there a point at which metaplasia is irreversible?

More importantly, what is the clinical significance of SPEM as a pre-neoplastic lesion? Does gastric dysplasia arise from SPEM? While this review has predominantly focused on the initial injury response and speculated on mechanisms for the expansion of SPEM, it should be noted that most of the pathology literature related to gastric cancer focuses on gastric intestinal metaplasia, a precursor lesion to gastric adenocarcinoma that, like SPEM, emerges after the development of oxyntic atrophy²¹⁸. Intestinal metaplasia involving the gastric corpus carries a significant oncogenic risk that has formed the basis for endoscopic surveillance guidelines²³⁹. Compared to SPEM, however, the cellular origin of intestinal metaplasia is less understood, largely due to a lack of adequate animal models. It remains to be seen whether SPEM gives rise to intestinal metaplasia²⁴⁰ or whether the two precursor lesions can independently give rise to gastric adenocarcinoma¹³⁷. Regardless, it is becoming more evident that the zymogenic chief cell plays a crucial role in the initiation of SPEM and in repairing glandular injury¹⁶⁹, though the precise mechanisms by which the chief cell undergoes cellular reprogramming to fuel metaplasia warrant further investigation. This reprogramming likely constitutes a sequence of intra-cellular mechanisms that are shared across other exocrine organs as part of a conserved response to glandular injury. As we begin to uncover the molecular players involved in this reprogramming sequence, we can begin to identify how specific mutations contribute to the development and progression of gastric cancer for millions of patients at risk.

Key points.

• The stomach is a versatile organ that protects against countless forms of endogenous and exogenous injury, mainly through the production of acid.

• The stomach’s injury response can be classified into two main patterns, one that protects against endogenous acid (superficial response) and one that adapts when the source of acid is lost or compromised (glandular response).

• The glandular response is a process that is best understood in the gastric corpus and involves a replacement of injured epithelium with metaplastic cells, a process known as Spasmolytic Polypeptide-Expressing Metaplasia, or SPEM.

• Recent studies highlight the epithelial plasticity of the gastric corpus, in particular the ability of post-mitotic zymogenic chief cells to re-enter the cell cycle and fuel repair of injured epithelium.

Glossary

Surface Epithelium

mucus-secreting cells that line the surface of the stomach; also referred to as surface, foveolar, or pit cells.

Gastric Pit

The surface epithelium invaginates into gastric units that are funnel-shaped and dive downward towards the gastric muscular wall. The mouth-like opening of each gastric unit represents the gastric pit. The zone where the pit narrows into the gland harbors actively dividing stem cells and is called the isthmus.

Gastric Gland

We use here the human pathology definition of the gastric gland as being separate from the gastric pit. The glandular portion of a gastric unit is located between the base (i.e., nearest the stomach muscular wall) and extends up to the isthmus. In the corpus, the gastric gland comprises parietal, chief, mucous neck, and endocrine cells. In the antrum, the gastric gland contains mucous and endocrine cells.

Superficial Response

how the stomach (both corpus and antrum) repairs erosive injury (most commonly due to acid) on the epithelial surface.

Glandular Response

how the stomach adapts to injury involving loss of acid-secreting parietal cells and digestive enzyme-secreting chief cells from gastric glands in the corpus.

Oxyntic Atrophy

a process characterized by the loss of acid-producing, or oxyntic, glands from the corpus.

Atrophic Front

The stomach-adapted bacterium Helicobacter pylori is known to cause atrophy and metaplasia of the corpus in a subset of chronically infected patients. This atrophy spreads along a “front” from the antrum into the corpus along the lesser curvature.

Cyclical Hit Model of Tumorigenesis

a proposal that mutations may accumulate and be stored in differentiated cells. Following injury, differentiated can re-enter the cell cycle and proliferate. During their proliferative phase, these mutations can be acquired. As the cells re-differentiate, the acquired mutations are stored. These stored mutations may accumulate with little effect until the cells either undergo apoptosis or become trapped in a (proliferative) dysplastic state.

Biography

Jason C. Mills is an expert in epithelial stem cells, metaplasia, and the cells of origin in cancer. He is director of research and co-director of the Digestive Disease Center in the Division of Gastroenterology at Washington University. He is also a Professor in the Departments of Medicine, Pathology and Immunology, and Developmental Biology at the Washington University School of Medicine. José B. Sáenz is an Instructor of Medicine in the Division of Gastroenterology at the Washington University School of Medicine and is particularly interested in how Helicobacter pylori adapts to alterations in the gastric landscape.