Good fences make good neighbors

Abstract

The gastrointestinal lumen is home to over 400 species of microorganisms. The composition of this microbial community varies along the length of the gastrointestinal tract as a function of regional epithelial secretory activity as well as diet and other defined and undefined determinants. Improved understanding of the factors that impact luminal microbial populations and development of means to modulate gut microbes for therapeutic benefit hold great promise. The gastrointestinal epithelium, which regulates interactions between microbes and the mammalian host, is the topic of this review.

Introduction

Like the skin, the gastrointestinal tract creates a barrier between external and internal environments. However, the gastrointestinal tract is also responsible for nutrient absorption and digestion, and in many areas is densely colonized by microbes. Because of these facts, the management of this barrier requires mechanisms far more complex than those required by the skin. The discussion that follows will address common and divergent approaches used to balance these requirements in different regions of the gastrointestinal tract.

The overall structure of the gastrointestinal wall is preserved throughout the tract and, with the exception of the esophagus, the gut is lined by a single layer of columnar epithelial cells. In contrast, stratified (multi-layered) squamous cells line the esophagus (Fig. 1A). Like the skin, these squamous cells form a significant barrier and, with the exception of lubricating secretions from the submucosal glands, the esophagus acts as a simple conduit that delivers masticated food from the oral cavity to the stomach. However, colonization by fungi and bacterial organisms is relatively common, particularly if the surface epithelium is superficially eroded or more deeply ulcerated. While pathological, such infections are rarely invasive and, therefore, do not typically result in systemic consequences. Other infections, including by herpes virus and cytomegalovirus, also occur, particularly in immunosuppressed individuals.

Figure 1.

Epithelial organization. (A) The esophageal (and anal) mucosae are covered by a stratified squamous epithelium. Stem cells reside in the basal zone (arrow) and proliferate to give rise to cells that divide and populate the epithelium. Mitoses are rare above the bottom one-third of the stratified epithelium. Squamous cells terminally differentiate and are shed at the luminal surface (arrowhead). Hematoxylin and eosin stain. (B) The colonic mucosa, and the majority of the gastrointestinal tract, is covered by a simple columnar epithelium. Stem cells reside in the crypts (arrow) and proliferate as they migrate towards the surface. Mitoses are rare beyond this transit-amplifying zone, which occupies the bottom one-third of the mucosal thickness. Terminally-differentiated surface cells (arrowhead) are shed at the luminal surface. The loose connective tissue surrounding the crypts is termed the lamina propria. Hematoxylin and eosin stain.

Within the remainder of the gastrointestinal tract, the simple columnar epithelium (Fig. 1B) rests on a basement membrane and the loose connective tissue of the lamina propria. Lymphocytes, mast cells, macrophages and polymorphonuclear leukocytes populate the lamina propria, which is supported by a thin layer of smooth muscle termed the muscularis mucosa. Together, the epithelium, lamina propria and muscularis mucosa make up the mucosa, which overlays the submucosa composed of fibroconnective tissue, nerves, vessels and lymphatics. While the submucosa supports the mucosa, the muscularis propria—two, or in the stomach three, layers of smooth muscle—supports the submucosa. In most areas, a thin layer of fibrofatty tissue and the surface mesothelial cells form a serosa that encases the gastrointestinal tract. However, in some regions of the esophagus and the distal colorectum, fibrofatty tissue beneath the muscularis propria extends directly into regional connective tissue and no serosa is present.

As the microscopic anatomy of the gastrointestinal tract differs along its length, the commensal and infectious microbes that inhabit each region vary. Thus, an appreciation of region-specific anatomy is necessary for one to understand interactions between microbes and the gut.

Stomach

The gross and microscopic anatomies of the stomach reflect its two major functions: mixing of food with acid, digestive enzymes and fluid to transform it into semi-liquid chyme, and regulation of the passage of this partially-digested material into the duodenum. The mucosa and submucosa of the contracted stomach are piled into folds called rugae. As the stomach expands with food content, the rugae flatten themselves to accommodate the increased volume. On a cellular level, the mucosa contains a network of gastric glands that empty into overlying pits. The glands and pits are extensive in the acid secreting, or oxyntic, fundus and body of the stomach. In these regions, the glands and pits are largely populated by parietal cells, which secrete acid and intrinsic factor, as well as the exocrine chief cells that secrete pepsinogen (Fig. 2A). Once secreted, pepsinogen is cleaved to form pepsin, a protease that contributes to the initial phase of digestion.

Figure 2.

Stomach. (A) The pits house the specialized cells that are responsible for gastric function. The pale, pink parietal cells (arrowhead) secrete acid, while the blue chief cells (arrow) secrete digestive enzymes. Hematoxylin and eosin stain. (B) Helicobacter pylori micro-colonies (arrowhead) grow in the mucous layer secreted by the gastric surface foveolar cells. Steiner silver stain.

The gastric surface is populated by foveolar cells that secrete mucin and bicarbonate. These components work together with the unstirred layer to form a region that is markedly less acidic than the lumen,¹ which typically has a pH between 1 and 2. Thus, this mucus-rich layer is also an ecological niche that allows growth of the most common gastric pathogen, Helicobacter pylori (Fig. 2B). Bacterial production of the enzyme urease buffers and elevates local pH by cleaving urea to generate ammonia. While many bacteria may synthesize urease, the enzyme produced by H. pylori has a low Km that allows function in the low urea concentrations present in the stomach. This permits identification of H. pylori by the CLO (Campylobacter-like organism) test, in which a gastric mucosal biopsy is cultured in a urea-containing gel. Change in color of the pH indicator included within the gel indicates the presence of urease-containing bacteria. H. pylori urease may also regulate epithelial paracellular permeability via myosin light chain kinase activation and occludin internalization² using mechanisms that are thought to be similar to tumor necrosis factor-induced regulation of intestinal epithelial tight junctions (described below).

H. pylori are most often found in the gastric antrum and cardia, likely reflecting the absence of parietal cells and active mucosal acid secretion in these areas. However, in patients with gastric mucosal atrophy, which may occur in chronic gastritis, parietal and chief cells may be lost, allowing H. pylori colonization of the fundus and body. Repopulation of parietal and chief cells is difficult, as these cells are normally replaced every 1 to 3 years, while foveolar cells repopulate rapidly, with the entire surface epithelium renewed every 4 to 8 days. Thus, mucosal atrophy is a longstanding condition that is associated with reduced gastric acid secretion and may proceed to intestinal metaplasia, in which foveolar mucous cells are replaced by intestinal-type goblet cells. Perhaps because of cell type-specific bacterial adhesins, H. pylori are unable to bind to goblet cells and are rarely found in areas of extensive intestinal metaplasia. Similarly, H. pylori are never found in the duodenum. Thus, although intestinal metaplasia is a risk factor for the ultimate development of adenocarcinoma, it may have its origins as an adaptive response to H. pylori infection.³

Small Intestine

The small intestine is divided anatomically into three segments: the duodenum, jejunum and ileum. While these are similar anatomically, there is a gradient of specialized function, such that water soluble vitamins and fatty acids are absorbed in the duodenum, glucose and Na⁺ are most effectively absorbed in the jejunum, and vitamin B12 and bile salts are absorbed in the ileum. Although incompletely defined, there is also a gradient of bacterial colonization in the small intestine, with the duodenal contents containing far fewer organisms than those of the ileum. The ileocecal valve limits reflux of colonic contents, which includes an even greater bacterial load, into the ileum. However, the ileocecal valve is not absolutely required, as patients who have undergone right hemicolectomy for cancer, in which the distal ileum is anastamosed directly to the colon, are not at increased risk for infection or inflammatory disease.

Throughout the small intestine, the mucosa and submucosa are specialized to increase surface area and maximize the absorptive capabilities of the mucosa. The submucosa is organized into regular mucosal folds as well as villi that increase the overall surface area 30-fold. At the core of each villus (Fig. 3A) lies a central arteriole from which capillaries fan out and descend just beneath the basement membrane. This intimate association between central arteriole and descending capillaries allows for the countercurrent exchange of solutes, comparable to that which occurs in the renal medulla. Thus, the villous lamina propria becomes hypertonic during active nutrient absorption, and this serves to enhance bulk fluid absorption.⁴ The center of each villus also contains a lacteal that allows chylomicrons and other lipids to be trafficked after transepithelial absorption. The cellular components of the lamina propria normally include lymphocytes and plasma cells (Fig. 4A). However, these are replaced by dense macrophage infiltrate in mycobacterial infection (Fig. 4B), which results in lymphatic compression, disruption of lipid absorption, and steatorrhea (excessive stool fat content). The rich lymphatic network in the small intestine can also have a dark side, as it may also provide a route for small, superficially invasive cancers to metastasize or be a conduit that delivers metastatic tumors from other sites (Fig. 4C).

Figure 3.

Small intestine. (A) This low magnification view shows the overall organization of crypts (arrow) and villi. Cells are shed from the villus tip (arrowhead). Hematoxylin and eosin stain. (B) High magnification view of villus tip. Note the refractile microvillus brush border (arrowhead). Hematoxylin and eosin stain. (C) High magnification view of a crypt. Paneth cells have large red granules (arrow) within the apical cytoplasm, while enteroendocrine cells (arrowhead) have much smaller, basally oriented granules. Hematoxylin and eosin stain.

Figure 4.

Lamina propria. (A) The small intestinal lamina propria is populated by vessels, lymphatics, lymphocytes and plasma cells (arrowhead). Hematoxylin and eosin stain. (B) In mycobacterial infection, macrophages (arrowhead) become filled with organisms (reddish in acid fast stain, inset). Hematoxylin and eosin stain. (C) This neuroendocrine carcinoma (arrowhead) has filled the lamina propria lymphatics. Hematoxylin and eosin stain.

Like the specialization from duodenum to jejunum, there is functional compartmentalization from crypt to villus. The crypts, which are derived from a single progenitor cell,⁵ are responsible for cell renewal and both exocrine and endocrine-paracrine secretion. Each crypt contains four to six stem cells,⁶ which can be identified by their expression of Lgr5 (leucine-rich-repeat-containing G-protein-coupled receptor 5) or olfactomedin-4.^7,8 While the functions of these and other proteins⁹ in stem cell maintenance is not yet clear, they are useful as markers of this niche,^6,10 which may be the source of transformed cells in cancer.^11,12 Stem cells divide and differentiate into Paneth cells and enteroendocrine cells that also populate the intestinal crypt (Fig. 3C). Recent studies have suggested that Paneth cells are essential components of the crypt cell niche,⁶ and this may explain their increased numbers in chronic inflammatory disease. Paneth cells also release defensins, antimicrobial peptides that regulate microbial populations within the gut¹³ and provide protection from bacterial toxins.¹⁴ Together with defensins, fluid secretion (discussed below) allows the crypt lumen to be a relatively sterile site.

The progeny of stem cells emerge from the basal aspect of the crypt as undifferentiated, transit-amplifying cells. These cells secrete chloride into the lumen in response to a broad spectrum of extracellular stimuli, including adenosine and cholera toxin, using calcium and cyclic nucleotides as intracellular mediators.^15–18 This chloride secretion establishes an osmotic gradient that draws sodium and water across the tight junction, resulting in net fluid secretion.^19–22 Under normal conditions, the undifferentiated crypt cells mature into mucin-secreting goblet cells and absorptive enterocytes that express nutrient and ion transporters as they proliferate and migrate to the villus tip (Fig. 3B).^23–25 These cells have a well developed brush border covered by a dense forest of ∼1 µm-tall microvilli that increase the absorptive surface area 20-fold, thereby significantly increasing the density of transmembrane transporter proteins and efficiency of nutrient absorption. Disruption of microvilli by attaching and effacing organisms, such as enteropathogenic E. coli, is one means by which these pathogens may cause malabsorption and diarrhea.

Each microvillus is by supported a cytoskeletal core of actin filaments and associated proteins, including myosin I and villin. As these cytoskeletal cores protrude into the cell body, they meet the terminal web, a dense mesh of microfilaments enriched in myosin II. At the periphery of the cell, the terminal web condenses to form a perijunctional actomyosin ring that is closely associated with the adherens and tight junctions. As detailed below, the perijunctional actomyosin ring is critical to tight junction assembly, maintenance and regulation.

Ultimately, epithelial cells are shed into the lumen, allowing the intestinal epithelial to turn over completely every 4 to 8 days. Debate continues as to whether apoptosis triggers shedding or detachment from the basement membrane triggers anoikis.²⁶ In either case, it is clear that the epithelial barrier is preserved during cell shedding.^27,28 The mechanisms by which barrier maintenance and epithelial apoptosis are coordinately regulated are an area in need of further study.

Due to the diversity of dietary and microbial antigens present in the lumen, the gut is the primary site at which the immune system encounters the external environment. The small intestine includes several specialized features that facilitate this sampling. The first, M (microfold-bearing) cells are present in the epithelium that covers lymphoid follicles.^29–32 While most prevalent in the distal ileum overlying the Peyer's patches, M cells are present throughout the small intestine and colon. M cells cannot be recognized by light microscopy, but transmission electron microscopy reveals irregular, short and sparse microvilli and a basal membrane that retracts from the basement membrane to form a space. Lymphocytes and macrophages migrate into this space, where the M cells release the contents of apically derived endocytic vesicles. This transcytosis is a rapid and efficient conduit for infectious organisms, such as Salmonella spp., Listeria monocytogenes and Shigella flexneri, and has also been exploited for mucosal vaccine delivery.²⁹ Within the distal ileum, an additional mechanism of luminal sampling is used by a specialized population of dendritic cells that extend processes across the epithelial tight junction.^33–35 The overall contribution of these processes to mucosal immune function is poorly understood.

Colon

The colon absorbs excess water and salts and serves as a reservoir to hold fecal material until defecation. Microbial populations are far denser than in the small intestine, and can exceed 10¹² bacteria per milliliter. As is true throughout the gut, the majority of these bacteria are anaerobes, reflecting the anoxic lumen.

Colonic bacteria are symbiotic with the mammalian host and can break down complex carbohydrates that cannot be metabolized by humans.^36–38 Moreover, fermentation by bacteria generates short chain fatty acids that serve as an important energy source for colonic epithelial cells.³⁹ Deprivation of the fecal stream and resultant alterations of luminal microbial populations may, in part, explain the colitis that develops in excluded segments of distal colon when a proximal ostomy is created.⁴⁰ This diversion colitis generally abates when fecal flow is restored.

The colonic mucosa is organized into parallel crypts that mature into a flat surface that lacks villi. Both crypt and surface epithelium are composed primarily of goblet cells. While goblet cell function is not thought to vary greatly along the crypt-surface axis, the scattered columnar colonocytes tend to be secretory towards the bottom of the crypt and absorptive at the surface. As in the small intestine, the colonic crypts contain stem cells, endocrine cells and Paneth cells. In humans, Paneth cells are normally limited to the right colon; their presence in the left colon, which represents metaplasia, is one indicator of chronic disease.

Diarrhea and Paracellular Barriers

Diarrheal disease may be caused by small intestinal or colonic infection. Organisms that preferentially affect the small intestine include Giardia, V. cholera, S. typhi, enterotoxigenic E. coli, and mycobacteria. In contrast, C. difficile, Shigella spp., enterohemorrhagic E. coli, and Campylobacter spp. tend to infect the colon, while Yersinia spp. and Salmonella spp. can affect both small intestine and colon. Detailed discussion of the means by which each of these organisms causes disease is beyond the scope of this article, but mechanisms of pathogenesis include toxin production, bacterial secretion, and invasion. E. coli, Salmonella and Shigella all use a Type III secretion system to transfer bacterial proteins to host cells. In many cases, the injected proteins inhibit small GTPases within host cells, which causes actin reorganization and may enhance endocytic uptake of bacteria. In the case of enteropathogenic E. coli, actin reorganization leads to the characteristic attaching and effacing lesions that concentrate microfilaments beneath bacterial microcolonies. While C. difficile does not use a Type III secretion system, the two major toxins that this organism releases both trigger glucosylation of the small GTPase rho, a critical regulator of cytoskeletal function.^41–46 Thus, the actin cytoskeleton is a common target of bacterial pathogens.

The perijunctional actomyosin ring (Fig. 5A) is essential for maintenance of epithelial intercellular junctions. Thus, the massive actomyosin disorganization induced by toxin-mediated rho inhibition results in disassembly of adherens and tight junctions and nearly complete loss of paracellular barrier function.^47–49 However, most cases of barrier dysregulation are less extreme and represent altered function of structurally-intact junctions.^50–57 This allows increased paracellular flux of hydrophilic solutes, including dietary and microbial antigens and adjuvants, and may contribute to mucosal immune activation and disease.^58,59

Figure 5.

Apical enterocyte ultrastructure. (A) Caco-2 cells are frequently used as a model of small intestinal enterocytes due to their extensive differentiation and well-developed microvillus brush border. Each microvillus contains an actin core (arrowhead). A dense perijunctional actomyosin ring excludes organelles, such as mitochondria (arrow) from most apical cytoplasm. Electron micrograph of polarized Caco-2_BBe cells⁹⁸–¹⁰⁰ grown on a Transwell support. (B) This high magnification view of the boxed region in (A) shows the apical junctional complex, composed of the tight junction (arrowhead) and adherens junction (arrow). The desmosomes (asterisk) are located more basally.

Tight junctions (Fig. 5B) are composed of multi-protein complexes that include transmembrane proteins, peripheral membrane proteins, and regulatory molecules including kinases and phosphatases.^60–65 While the functions of many of these proteins are only beginning to be understood, it is now clear that trans-membrane proteins of the claudin family, which are expressed in a tissue-specific manner, directly impact tight junction permeability.^66,67 Claudin proteins form the anastamosing strands that are apparent by freeze-fracture electron microscopic examination of tight junctions,^67,68 and individual claudin family members differentially impact paracellular charge selectivity.^69–75 Claudins can also be the targets of bacterial toxins. For example, C. perfringens enterotoxin binds to specific claudin proteins and triggers their removal from the tight junction to disrupt barrier function.^76–80 Intriguingly, the specific claudin proteins expressed within intestinal epithelia are altered in disease.^81–83 Whether this is a beneficial or harmful response remains to be determined.

The roles of occludin, a transmembrane tight junction protein that interacts directly with ZO-1 and actin, are less well understood.^84–86 However, removal of occludin from the tight junction by endocytosis is commonly observed in pathological tight junction disruption.^{2,19,22,52,56,87–91} In vivo, occludin internalization within jejunal enterocytes is elicited by T cell activation as well as systemic administration of recombinant TNF or LIGHT (a member of the TNF core family).^19,22 This occludin internalization is associated with subtle reorganization of the scaffold protein ZO-1 at the tight junction,²² increased phosphorylation of perijunctional myosin II regulatory light chain (MLC), and increased paracellular permeability to large hydrophilic solutes.^19,22,92 All of these changes, including diarrhea, are prevented by enzymatic or genetic inhibition of myosin light chain kinase (MLCK).²² Pathogenic bacteria, such as enteropathogenic E. coli^50,53 and H. pylori;^2,93 bacterial products, such as LPS;⁹⁴ and parasites, such as Giardia may also activate MLCK.⁵⁴ Thus, MLCK-dependent MLC phosphorylation is a common means of barrier regulation by cytokines and infectious agents.

A recent series of studies in mice has shown that MLCK-dependent barrier loss is necessary for immune-mediated diarrhea^19,22 and that constitutive MLCK activation accelerates onset of chronic disease.⁵⁸ However, these studies also show that tight junction barrier loss alone is insufficient to cause diarrhea; in fact, isolated increases in paracellular permeability enhanced water absorption.¹⁹ While this may seem counterintuitive, it is critical to recall that paracellular transport is a passive process driven by transepithelial gradients. Thus, if tight junction permeability to water increases without disruption of the transepithelial osmotic gradient, water absorption may be enhanced.^19,95,96 The experimental diarrhea that follows TNF administration or T cell activation requires activation of both MLCK and PKCα.^19,22 While MLCK increases paracellular barrier function, PKCα inhibits transcellular sodium absorption. The loss of sodium absorption disrupts the transepithelial osmotic gradient that drives paracellular water absorption, allowing water to flow from mucosa to lumen.^19,97

Summary

Although overall themes are maintained, mucosal cellular composition, commensal microbiota, and pathogens vary widely along the length of the gastrointestinal tract. It is, therefore, critical to consider the region involved when assessing disease in humans or animals and to maintain these site-specific relationships when developing reductionist models. Understanding commensal and pathogenic organisms as well as epithelial and immune components of the mucosa is advancing at a rapid pace. Definition of these complex relationships promises to significantly enhance human health.