The intestinal epithelium is a continuous monolayer of specialized cells generated by intestinal stem cells at the base of the crypt. IEC differentiation from the crypt to the villus tip (small intestine), or surface epithelium (large intestine), and their subsequent turnover is critical for the chemical and physical functions of the barrier. IEC apoptosis and shedding into the lumen is tightly regulated to promote continuous turnover of the epithelium and maintain barrier integrity [17-19]. In a healthy gut, only 1% to 2% of the epithelia are undergoing apoptosis at any one time, and neighboring IECs act to seal gaps arising from apoptosis or cell “shedding events” in order to prevent unregulated access across the epithelium by luminal contents or microbes. Indeed, the pro-inflammatory cytokine TNF-α increases the number of shedding events however, this is not associated with a barrier defect due to the “gap-sealing” actions of adjacent epithelial cells [18,19]. The life cycle of an IEC is driven by changes in enterocyte function and epithelial differentiation. Villus or surface enterocytes display an absorptive phenotype while IECs at the crypt base secrete Cl^-ions and water that aid in hydrating the mucus layer and flushing bacteria out of the crypt [20]. These functions are critical for the principal roles of the intestine (digestion, nutrient absorption, and waste excretion) and maintaining balanced interactions with luminal bacteria.

In addition to their absorptive and secretory functions, IECs also differentiate into specialized cells that contribute to intestinal barrier homeostasis. Paneth cells that are situated in the crypt base of the small intestine secrete anti-microbial factors such as α-defensins (or cryptidins in mice), RegIIIγ, lysozyme, and phospholipase A2 in response to microbial components [21]. These secreted factors aid in preventing pathogenic bacterial attachment and promote growth of beneficial bacteria. Positioned along the length of crypts, goblet cells secrete mucin proteins to establish a protective mucus layer overlying the intestinal epithelium that also functions as a diffusive medium for metabolites and nutrients. The composition and physical properties of the mucus layer differ by location in the GI tract. The small intestine has a loose, non-adherent mucus layer which facilitates nutrient absorption. In the colon, the mucus layer is composed of 2 sublayers–an outer layer much like that of the small intestine and a thick inner adherent layer that is impermeable to bacteria in the distal colon and somewhat permeable in the proximal colon [22]. Commensal microbes aid in maintaining this mucus layer through stimulation of mucin secretion. This observation has also been supported by analysis of the mucus layer in germ-free mice where the colonic inner layer was less developed [23]. Finally, the mucus binding capacity of probiotic bacteria is correlated with their colonization capacity, providing a selective advantage over pathogenic bacteria [24]. Studies on cystic fibrosis transmembrane conductance regulator knockout mice, which display accumulations of mucus, have shown an increase in IEC inflammatory gene expression and bacterial overgrowth, suggesting an abnormal mucus layer can promote dysbiosis [25,26].

A primary function of IECs is their ability to act as mediators of innate immunity. Similar to other cell types, IECs express cell-surface and intracellular pattern recognition receptors such as Toll-like receptors (TLRs) and NOD-like receptors, respectively. TLRs expressed by IECs function to respond primarily to commensal bacteria and different regions of the intestinal tract exhibit distinct patterns of TLR expression [27]. Accordingly, TLR expression is highest in the colon where their activation promotes expression of host defense genes [27]. In addition, the presence of gut flora or inflammation can also alter the expression profile of IEC TLRs. Activation of TLRs by commensal bacteria promotes tight junction protein expression to fortify the barrier and increases secretion of anti-microbial peptides [28-30].

Immune cells located in the mucosa and submucosa also function in shaping the microbial environment and hampering colonization by pathogenic bacteria. Lymphoid follicles called Peyer’s patches are located at various intervals in the ileum and consist of macrophages, lymphocytes, and M or micro-fold cells. These sites function in maintaining immune tolerance of commensal bacteria. In addition, secretion of IgA and components of the complement system aid in inhibiting pathogenic bacterial growth. Finally, certain bacteria are also necessary for immune maturation, such as segmented filamentous bacteria which promote Th17 cell development and Bacteroides fragilis which influences regulatory T cell development, and Th2 and Th1 cell balance [31-38].

Secretory IgA is a mucosal Ig produced by intestinal lamina propria B cells that captures microbial pathogens and their toxins to prevent adherence and invasion. Induction of IgA secretion occurs at Peyer’s patches and isolated lymphoid follicles within the GI wall where signals from T cells, dendritic cells, and IECs cause class switching of B cells to IgA-producing plasma cells [39]. Importantly, IgA is able to protect the barrier from bacteria without increasing inflammation. Commensal bacteria can promote IgA production directly through bacterial proteins binding host receptors on IECs, and indirectly through microbial metabolites such as short-chain fatty acids (SCFAs) binding the GPR43 receptor [40,41]. In turn, IgA binding to commensals can promote their colonization and stability within their colonization niche [42-44]. A continuing question in the field is how host responses differentiate between commensal and pathogenic microorganisms. It is thought that IgA binding to commensals is low-affinity whereas increased coating of bacteria by IgA indicates the increased pathogenic potential of the bacterium. This has been supported by transfer of highly IgA-coated bacteria isolated from IBD patients to germ-free mice which led to increased susceptibility to dextran sulfate sodium (DSS) colitis [45]. Thus, IgA is an important discriminator of commensal and pathogenic bacteria and promotes mucosal homeostasis through increased colonization of healthy commensals.

Normal antigen-sampling of the GI luminal contents by IECs and submucosal immune cells elicits a tolerogenic immune response essential for shaping GI microbial composition. The role of IECs as mediators of innate immunity is underscored by studies of single-nucleotide polymorphisms (SNPs) associated with IBD, in which genes critical for antigen recognition and bacterial destruction have been identified, such as nucleotide-binding oligomerization domain 2 (NOD2) and autophagy-related protein 16-like 1 (ATG16L1) [46-50]. Deficiency of these genes impairs normal maintenance of commensal populations and restriction of pathobionts, leading to dysbiosis [51-55]. The influence of IBD candidate genes on epithelial barrier function and bacterial sensing are discussed in greater detail in several comprehensive review articles [56,57].

The mucus layer is composed of mucins with oligosaccharide chains whose terminal ends cannot be metabolized by the majority of gut bacteria [58]. However, certain bacteria termed mucin specialists have evolved specific enzymes that allow degradation of mucin oligosaccharide chains for sugar and protein, thereby facilitating their own colonization as well as growth of other commensals by increasing nutrient availability [59-62]. In fact, commensals such as Faecalibacterium prausnitzii rely on these mucin specialists to facilitate their colonization to the mucus layer [63-68]. Akkermansia muciniphila is a widely studied mucin specialist capable of tolerating the hypoxic environment near the epithelial surface and that utilizes host-derived proteins for survival [69-71]. A. muciniphila colonization increases mucin production and mucus thickness, and improves barrier function [70,72-74]. Butyrate production by A. muciniphila is also beneficial to the host by upregulating energetic pathways, such as increased β-oxidation and intestinal gluconeogenesis in colonocytes that maintain a commensal-dominated microenvironment and protect from metabolic impairment [75-78]. Finally, increased abundance of A. muciniphila at sites of intestinal injury where oxygen is depleted, allows expanded colonization by anaerobic bacteria. In this setting, A. muciniphila also promotes intestinal epithelial formyl peptide receptor 1 (FPR1) signaling and NADPH oxidase (NOX1) activity to increase enterocyte migration and proliferation to facilitate mucosal healing [79].

In addition, A. muciniphila has been inversely associated with intestinal inflammation [62,80-84]. Recent evidence has also shown that A. muciniphila-secreted extracellular vesicles protect the epithelium from colitis and diet-induced barrier dysfunction [85]. Another muciniphile Peptostreptococcus russellii, is also protective against DSS-induced chemical injury colitis model in mice. In addition to its mucin-degrading activity, P. russellii expresses aromatic amino acid metabolic enzymes that utilize tryptophan to produce indoleacrylic acids, which have anti-inflammatory effects and promote goblet cell function [86]. The phenyllactate dehydratase gene cluster responsible for expression of these enzymes was decreased in IBD patients compared to healthy controls [86]. Collectively, these findings indicate both host and commensal act synergistically to encourage mucosal homeostasis and repress inflammation.

Another commensal microbe that is consistently decreased in UC patients is F. prausnitzii [87-89]. F. prausnitzii is a highly abundant butyrate-producer in the intestinal tract, encompassing 5% of bacteria in feces [90,91]. Since increased F. prausnitzii is correlated with improved outcomes following surgical resection, it was thought that this microbe could directly affect host immunity although its presence is not required for patient recovery [92,93]. In accordance with this, Quevrain et al. [94] identified a 15 kDa anti-inflammatory protein was secreted by F. prausnitzii. Subsequently, a study by Breyner et al. [95] demonstrated F. prausnitzii secreted peptides termed “microbial anti-inflammatory molecules” can inhibit nuclear factor (NF)-κB promoter activity and interleukin (IL)-6 production, leading to protection from experimental colitis in mice through alteration of T cell immune responses. Interestingly, co-culture studies have demonstrated F. prausnitzii does not enhance barrier function but can induce activation of NF-κB signaling downstream of IL-1β in Caco-2 cells [92,96]. In contrast, Caco-2 cells treated with F. prausnitzii supernatant decreases basal and IL-1β-induced NF-κB activity [92].

L. rhamnosus Gorbach-Goldin (LGG) has been well-studied as a probiotic used in the production of dairy yogurt. This commensal microbe can indirectly and directly regulate epithelial function to repress inflammation and resist enteric infections. Administration of LGG culture supernatant to neonatal rats was associated with increased IEC proliferation, decreased paracellular barrier permeability, and increased expression of mucin-2, zonula occludens-1 (ZO-1), and IgA, together promoting resistance to neonatal E. coli K1 infection [97]. In addition, LGG enhances wound healing through enterocyte activation of FPR1 and NOX1, similar to A. muciniphila [98]. Direct actions of LGG are performed p40, were found to inhibit pro-inflammatory cytokine-induced IEC apoptosis [99]. In addition, p40 can also modulate IgA production by increased expression of a proliferation-inducing ligand in IECs [41]. More recently, p40 was found to exert its responses through the epidermal growth factor receptor in neonatal mice, ultimately promoting IgA production and T regulatory cell (Treg) differentiation in adult mice [100].

B. fragilis is a commensal found within the glycocalyx and in colonic crypts of mice that exhibits single-strain stability or the ability of a single strain of bacteria to persist in the microbial community for years [44]. Its colonization of the mucosa is facilitated by IgA binding, which allows the bacteria to aggregate and contribute to colonization resistance against pathogenic bacteria [44]. In addition, its expression of cytochrome bd oxidase permits its growth in low concentrations of oxygen, such as those present in the colon [101]. B. fragilis secretes an immunomodulatory molecule, polysaccharide A (PSA) that regulates immune cells such as induction of Foxp3+ Treg cell differentiation to reduce intestinal inflammation and repression of IL-17 secretion [36-38]. Furthermore, B. fragilis strains expressing PSA are able to prevent colonization by the pathobiont Helicobacter hepaticus and subsequent inflammation, indicating commensals can prevent pathobiont expansion [37]. Intriguingly, B. fragilis also accounts for a majority of Bacteroides infections in the body and the enterotoxigenic B. fragilis (ETBF) strain is a causative pathogen of diarrhea [102]. Recently, Chan et al. [103] demonstrated stable colonization of non-enterotoxigenic B. fragilis (NTBF) ameliorated disease severity following ETBF infection, in a PSA-independent manner. This study confirmed that B. fragilis predominance alone can be protective against mucosal inflammation. Whether increased virulence results from changes in the GI microenvironment and microbial exchange of genetic information is not well understood, although this is an emerging topic among research groups [102].

Under healthy conditions, an oxygen gradient in the colonic mucosa exists as a result of the high oxygen consumption required to maintain the inward Na⁺ gradient driving fluid absorption. However, bacteria can also contribute to the colonic oxygen gradient. Butyrate-producing bacteria increase levels of CO₂ and together these mechanisms promote a hypoxic environment in the colonic lumen. Decreased levels of O₂ promote anaerobic growth of commensal bacteria such as B. fragilis and F. prausnitzii closer to the surface of IECs [101,110]. However, in a chronic inflammatory state, oxygen levels near IECs increase due to secretion of reactive oxygen species and increased blood flow. This allows facultative anaerobic, or “aerotolerant,” bacteria to predominate near the intestinal epithelium such as Actinobacteria and Proteobacteria, phyla that are also associated with IBD [119]. For example, mice exposed to DSS had increased numbers of aerobic bacteria compared to healthy controls, a phenomenon that has also been observed in newly-diagnosed CD patients [109,120]. Depletion of butyrate-producers such as Clostridia by antibiotic therapy induces a subsequent increase in luminal oxygen, allowing increased growth of Enterobacteriaceae such as Salmonella serovar typhimurium [121]. The concept of increased epithelial oxygenation acting as a driver of facultative anaerobe expansion at the expense of butyrate-producing obligate anaerobes is elegantly described in a recent review by Litvak et al. [78]

A primary method of preventing pathobiont expansion is colonization resistance by homeostatic microbes. Efficient nutrient assimilation provides a selective advantage to certain bacteria that can hinder growth of pathogens. In addition, metabolites in the GI lumen can also activate signaling processes within IECs that promote selective microbial growth in both the healthy and inflamed gut. During inflammation, secreted reactive nitrogen species such as nitric oxide degrade to nitrate in the lumen where they are utilized by microbial nitrogen respiration pathways. Similarly, production of reactive oxygen species promotes alternative sources of electron acceptors that are then utilized in microbial respiration. For example, hydrogen sulfide (H₂S) produced by commensals is oxidized to thionate during inflammation which is then utilized by S. typhimurium, although there has been some evidence indicating H₂S is also anti-inflammatory and promotes mucosal healing [111,122]. Another metabolite increased by intestinal inflammation includes ethanolamine (EA), which is produced from phosphatidylethanolamine. Enzymes that utilize EA as a nutrient source are found in the genomes of many enteric pathogens so it is unsurprising that EA promotes predominance of S. typhimurium and enterohemorrhagic E. coli in mice [112,123]. Similarly, the bacterial fermentation product 1,2-propanediol is exploited by S. typhimurium during inflammation with increased expression of virulence factors [124]. Finally, GPR43, a Gprotein coupled receptor bound by bacterial-produced SCFAs, is important for resolution of intestinal inflammation in mice, and thus indicates a link between gut microbiota and inflammation resolution [125].

Availability of dietary transition metals can also restrict pathogenic bacteria growth. In a recent study Zhu et al. [114] confirmed that microbial molybdenum cofactor-dependent respiration is a metabolic signature of DSS-induced dysbiosis and promotes Enterobacteriaceae expansion during colitis [126]. Consistently, pathobiont expansion was reversed by tungstate treatment [114]. Competition for iron by the E. coli strain Nissle 1917 restricts S. typhimurium during colitis and infection [115]. Interestingly, the host protein lipocalin-2 is required for this effect, indicating an intricate interaction between host and commensal bacteria that promotes a healthy gut environment. In addition to colonization resistance, dietary metals can also alter bacterial gene expression. Zinc deficiency elevates the risk of pediatric diarrhea and is associated with pathogenic E. coli. [127]. Indeed, low zinc causes increased virulence of enteroaggregative E. coli, the pathogen responsible for traveler’s diarrhea [116].

Another factor affecting the gut microbial community is luminal pH, which can range from 5.7 to 6.8 in human cecum and pH 6.1 to 7.5 in the colon [128]. There are major shifts in the gut microbiota between luminal pH values of 5.5 and 6.5. In a healthy colon, a gradient from a mildly acidic to a neutral environment exists from the proximal to distal colon. The acidic pH in the proximal colon is maintained by substrate fermentation where acetate and other SCFAs are produced from the fermentation of indigestible polysaccharides facilitating expansion of butyrate-producing species such as Roseburia [129,130]. During spontaneous colitis, accumulation of lactate drives a decrease in luminal pH, potentially due to repressed growth of lactateutilizing bacteria [131,132].

From these examples, it is clear that environmental changes of the microbial niche can alter composition of the gut microbiota, occurring as both small perturbations and large displacement of commensal bacteria. Both diet and inflammatory status contribute to these effects. Apart from these shifts in population, changes in pathogenicity of seemingly innocuous bacteria are also of great interest since chronic GI conditions are associated with increased numbers of more virulent bacteria including “pathobionts.”

Pathobionts can interact with IECs to exert their effects both directly through adherence and translocation and indirectly through modulation of IEC functions (Fig. 1). Pathobionts such as Bilophila wadsworthia and Enterobacteriaceae faecalis are able to adhere to and translocate through IEC monolayers in vitro and in vivo [137-139]. Adherence and invasion of pathobionts usually requires expression of specific genes such as proteases in E. faecalis and GipA (growth in Peyer’s patches) in AIEC [140-143]. In addition to attachment and invasion, pathobiont species also impair normal IEC functions for intestinal homeostasis. C. concisus, a gram-negative microbe that is increased in pediatric and adult IBD patients, inhibits expression of the tight junction-associated proteins ZO-1 and occludin, as well as their association with tight junction complexes leading to a less regulated intestinal barrier [144-147]. Pathobionts can also alter the life cycle of IECs. C. concisus increased apoptosis which left large gaps (apoptotic lesions) in the monolayer that persisted longer than those in control HT-29/B6 IECs [148]. In addition, colibactin-producing E. coli strains (which are also associated with inflammation-associated colorectal cancer) activate senescent secretory intestinal cells to increase growth factor production, leading to increased local cell proliferation and tumor growth [149-151]. Finally, expansion of Bacteroides vulgatus in Nod2-deficient mice was associated with goblet cell dysfunction and subsequently increased numbers of interferon-γ-expressing lymphocytes [152]. Together these studies demonstrated the ability of pathobionts to hijack the proliferation-differentiation process in the intestine in order to reduce important mediators of host defense and promote bacterial invasion.

Pathobiont expansion can also increase pro-inflammatory signaling in IECs leading to dysbiosis and colitis. As mentioned above, B. vulgatus also increased pro-inflammatory cytokine production in Nod2-deficient mice [153]. The capacity of pathobionts to drive a pro-inflammatory response does not appear to be uniformly dependent on the presence of a host defect in innate immunity as E. faecalis activates NF-κB, p38 MAPK, and ERK1/2 pro-inflammatory signaling via TLR2 resulting in induction of IL-6 and IP-10 secretion in wild-type murine IECs [154]. AIEC secretion of outer membrane vesicles leads to activation of the host ER stress response protein Gp96 [155,156]. In addition, AIEC infection in mice also increased Il6 and Lcn2 mRNA expression, fecal lipocalin-2 content, and spleen weight [108]. Bretin and colleagues recently suggested that transient AIEC infection in mice can promote inflammatory signaling after clearance of the pathobiont [136]. These data demonstrate that pathobionts are able to induce stress responses and inflammatory signaling in IECs and may continue to promote inflammation long after infection.

While much work has focused on how pathobionts induce inflammation, we have only begun to understand how these initially innocuous microbes become virulent and expand. As chronic GI disease is prevalent in developed countries, it has been hypothesized that environmental and behavioral factors such as diet contribute to dysbiosis. In support of this, studies have demonstrated that individuals relocating to developed countries begin to exhibit a more Western-like microbiome composition, similar to that of non-human primates who are captured in the wild and transferred to captivity [157]. B. wadsworthia is a gram-negative bacterium usually found at low abundance and has sulfite-reducing activity. Mice that were fed a high milk fat diet with saturated fats developed expanded B. wadsworthia leading to colitis, however this did not occur in mice fed other types of saturated fats or polyunsaturated fats [117]. Milk fats have elevated levels of hydrophobic stearate that requires higher concentrations of bile salts for emulsification. Milk fat feeding led to a higher ratio of the bile salt taurocholate, which is a more efficient emulsifier and a source of organic sulfur [117]. The subsequent increase in luminal organic sulfur facilitated a bacterial bloom of B. wadsworthia in these mice [117]. In addition to dietary components, inflammatory mediators can also promote pathobiont growth. For example, when IL-22RA1 KO mice were infected with Citrobacter rodentium they developed sepsis and had increased mortality due to expansion and increased translocation of E. faecalis. IL-22RA increases Fut2 expression and subsequently increases intestinal fucosylation. This promotes increased diversity of anaerobic commensal symbionts, restricting expansion of E. faecalis [113]. Pathobionts can also promote their own survival by escaping or resisting host defense mechanisms. For example, E. faecalis can resist antimicrobials targeting its cell envelope through expression of IreK, which is a kinase important for long-term colonization, potentially through inhibition of antagonistic enterococci proteins [158]. Thus, it is clear that a variety of host and environmental factors can precipitate selective expansion of pathobionts and that inflammatory responses can not only alter populations of pathobionts but also modify their level of pathogenicity, in conjunction with increasing host epithelial susceptibility to pathobionts.