Molecular Pathways: Pathogenesis and Clinical Implications of Microbiome Alteration in Esophagitis and Barrett Esophagus

Esophageal adenocarcinoma, the malignant transformation at the end of a spectrum of diseases related to gastroesophageal reflux, is now the most rapidly increasing cancer in the Western world. Barrett esophagus is defined as the metaplastic columnar epithelium that replaces squamous mucosa and predisposes to cancer development (1). The rate of progression from Barrett esophagus to esophageal adenocarcinoma is approximately 0.12% to 0.4% per patient-year (2–4). For unclear reasons, the incidence of esophageal adenocarcinoma in the United States has increased approximately 600% since the 1970s (5–7), and because its development is not universal among patients with Barrett esophagus, it is important to understand and to gauge the factors that influence risk of progression to dysplasia and cancer. The current review aims to highlight new insights into mechanisms underlying the host–bacterium interaction in the context of reflux-induced inflammation and esophageal carcinogenesis. In particular, the influence of microbial lipopolysaccharides (LPS) on the molecular pathways involved in inflammation-associated esophageal tumorigenesis is examined.

The term “microbiome,” coined by Joshua Lederberg, refers to the collection of all members in a complex microbial community (8). The host relationship with the microbiome can be commensal, symbiotic, or pathogenic. Bacterial mutualists within the gastrointestinal tract are beneficial to the host, as they aid digestion benefit from the host, assist in the synthesis of vitamins, promote development of the gut immune system, and provide competitive barriers to pathogen invasion. This complex microbial population influences an estimated 10% of all metabolites in our body (9). In return, the host provides bacteria with safe housing and food during lean times. For this symbiotic relationship to be sustained, the immune system has to balance permissive, tolerogenic responses to food antigens and commensal microbes with potentially damaging, inflammatory responses to ward off pathogens. This delicate balance is maintained by the constant interplay among the microbiome, the gastrointestinal barrier, and the mucosal immune system, which is a prerequisite for normal gut homeostasis. Imbalance of this system may lead to innate immune (inflammation) and adaptive immune (infectious pathology) responses.

Reflux esophagitis and Barrett esophagus represent phenotypes of inflammation of the esophageal mucosa induced by long-term gastric acid and bile reflux into the esophagus. The gastroesophageal reflux impairs the mucosal barrier and exposes the squamous epithelium and lamina propria to (i) the microbes swallowed from the oral cavity, colonized in the esophagus, and regurgitated from the stomach; (ii) acidic gastric contents; and (iii) bile from the duodenum.

A recent study of the human distal esophageal microbiome linked inflammation and Barrett esophagus to the change in the microbiome. The study used a 16S rRNA gene survey to characterize the bacterial communities in biopsy samples taken from the distal esophagus (10). With an unsupervised approach, samples of the microbiome form 2 distinct clusters or 2 microbiome types, type I and II, on the basis of combined genetic distance among samples. Although neither of the 2 types of clusters correlated exclusively with esophageal phenotypes, the type I microbiome is more closely associated with normal esophagus (11/12, 91.7%), whereas the type II microbiome is mainly associated with abnormal esophagus (13/22, 59.1%; P = 0.0173 among group comparison), including both esophagitis (7/12, 58.3%; OR = 15.4) and Barrett esophagus (6/10, 60.0%; OR = 16.5). Thus, alteration of the microbiome from type I to type II in the distal esophagitis is associated with host phenotypes and its disease progression. The type I microbiome is dominated by gram-positive bacteria representing the Firmicutes phylum. In contrast, the type II microbiomes are composed of larger numbers of gram-negative bacteria in phyla Bacteroidetes, Proteobacteria, Fusobacteria, and Spirochaetes. Streptococcus is the most dominant genus in the esophageal microbiome, and its relative abundance is significantly higher in the type I microbiome (78.8%) than in the type II microbiome (30%). In the type II microbiome, the decrease in the relative abundance of Streptococcus is compensated for by an increase in the relative abundance of 24 other genera. Specifically, the most prominent increase involves Veillonella, Prevotella, Haemophilus, Neisseria, Rothia, Granulicatella, Campylobacter, Porphyromonas, Fusobacterium, and Actinomyces, many of which are gram-negative anaerobes or microaerophiles and are putative pathogens for periodontal disease. Overall, gram-negative bacteria compose 53.4% of the type II microbiome, but only 14.9% of the type I microbiome.

The type II microbiome, with its larger content of gram-negative bacteria, might engage innate immune functions of the epithelial cells in a different way than the type I microbiome, owing to their production of larger amounts of gram-negative microbial components, for instance LPS. The bacterial products may directly or indirectly stimulate pattern receptors [i.e., Toll-like receptors (TLR)] in the epithelial or inflammatory cells, to promote expression of proinflammatory cytokines and persistent innate immune responses in the esophagus. Because many of the periodontal pathogens in the type II microbiome are known to cause inflammation in the mouth, it is plausible that they may similarly contribute to the development and maintenance of chronic inflammation in the esophagus. The change from microbiome type I to type II might, thus, prove to be an important step in the pathogenesis of esophageal tumorigenesis in progression of reflux esophagitis and Barrett esophagus and development of esophageal adenocarcinoma.

The current strategy for Barrett esophagus screening and surveillance may not be cost-effective and has not been shown to reduce esophageal adenocarcinoma incidence or mortality (4, 11). Indeed, data about risk factors for Barrett esophagus have not been systematically applied to screening guidelines, and the current state of the art for screening focuses primarily on endoscopic evaluation of individuals with chronic reflux symptoms. A surveillance interval of 3 to 5 years has been suggested for individuals without dysplasia, 6 to 12 months for low-grade dysplasia, and every 3 months for high-grade dysplasia (1). Much investigation is currently under way to identify prognostic biomarkers that may determine the best diagnostic and therapeutic course in esophageal adenocarcinoma.

LPS, the major outer membrane component present in gram-negative bacteria, consist of a lipid core and polysaccharide side chain joined by a covalent bond. LPS act as the prototypical endotoxin to promote the secretion of proinflammatory cytokines in many cell types. Host responses to gram-negative LPS are mediated mainly through activation of TLR 4 (Fig. 1). LPS molecules first bind plasma-derived LPS-binding protein and then interact with cluster of differentiation 14 (CD14), expressed mainly by macrophages, neutrophil granulocytes, dendritic cells, and local gastrointestinal epithelial cells to form a ternary complex, LPS:LPS-binding protein:CD14, which further transfers LPS to TLR4 accessory protein MD2 complex (12). LPS stimulation of monocytes or epithelial cells leads to the activation of TLR4 and the downstream NF-κB pathway, to evoke an inflammatory response. LPS might indirectly activate the NF-κB pathway of the epithelial cells by inducing inflammatory cells to produce interleukin-1β (IL-1β) or TNF-α, which may then engage cytokine receptor(s) through an alternative pathway (noncanonical; Fig. 1; refs. 13, 14).

NF-κB activation is important in the initial cellular response to chemical, bacterial, or viral stimuli. It is a major transcription factor that regulates genes responsible for both the innate and adaptive immune response. It is normally predominantly located in the cytoplasm, but translocates to the nucleus upon activation. Although the normal esophagus has no detectable active NF-κB, high levels of active NF-κB are found in esophageal adenocarcinoma in the setting of reduced levels of IκB-α (a known inhibitor of NF-κB; ref. 15). There is a stepwise increase in the activation of NF-κB pathway along the spectrum of reflux esophagitis (16, 17), Barrett epithelium (18, 19), and adenocarcinoma (16, 19, 20), parallel to an increase in IL-1β, IL-6, IL-8, and TNF-α.

The NF-κB pathway can be triggered by exposing cells to LPS from gram-negative bacteria, peptidoglycan from gram-positive bacteria, inflammatory cytokines (such as TNF-α or IL-1β), or by other physiologic and nonphysiologic stimuli (Fig. 1). Microbial components activate the NF-κB pathway via TLRs signaling through the classical pathway (canonical), whereas cytokines stimulate the pathway via cytokine receptors (such as IL–1Rs, TNFRs, and other TNFR-like receptors) through the alternative pathway (noncanonical; ref. 21). Peptidoglycan can also activate the NF-κB pathway by stimulating the nucleotide-binding oligomerization domain (NOD)–like receptors 1 (22, 23) and 2 (24–26). Moreover, the observations of NF-κB activation in response to inflammatory signaling through the mutated NOD2 gene in Crohn disease (24, 25) and through NOD1 in infection with Helicobacter pylori (23) and Chlamydophila pneumoniae organisms (22) raise some interesting possibilities in relation to human cancers developed in these inflammatory diseases. Despite evidence of TLR 1, 2, 3, 4, 5, 7, and 9 expression in human esophageal epithelial cells (27, 28), their degree of expression varies among individuals and their roles in NF-κB activation in reflux disorders remain poorly defined.

NF-κB is important in reflux disorders because of its broad role in upregulating its downstream target gene expressions involved in inflammation, innate immune responses, adaptive immune responses, apoptosis blocking, cell proliferation, and cell differentiation. It directly contributes to innate immune responses in reflux esophagitis, Barrett esophagus, and esophageal adenocarcinoma, and may ultimately determine the rate of progression to esophageal adenocarcinoma. Proinflammatory cytokines regulated by NF-κB pathway, such as IL-1β and IL-8, stepwise increased in reflux esophagitis, Barrett esophagus toward esophageal adenocarcinoma (14, 18, 24), and IL-4 and IL-6 are also increased in reflux esophagitis and Barrett esophagus (14, 19). It is notable that Fitzgerald and colleagues found no difference in the levels of IL-1β and IL-8 between noninflamed squamous mucosa and Barrett esophagus in the same patient (29). The secreted cytokines, including TNF and IL-1β, may also start a feedback loop for a second phase of NF-κB activation that continues the induction of robust innate immune responses. The cellular pattern recognition receptors such as TLRs, retinoic acid–inducible gene-I-like receptors (RLR), and NOD-like receptors, which all sense microbial opportunistic pathogens or pathogens, and pathogen-associated molecular patterns, use distinct signaling pathways that eventually converge to activate NF-κB, leading to the production of inflammatory mediators (Fig. 1; ref. 21).

NF-κB activation also upregulates the expression of genes encoding proinflammatory enzymes, such as COX-2 and inducible nitric oxide synthase (iNOS; Fig. 1). COX-2 protein expresses in the epithelial cells in Barrett metaplasia, and its level of expression is elevated in esophageal adenocarcinoma (30–33). The elevation in expression occurs along the progression from low-grade dysplasia to high-grade dysplasia in Barrett esophagus and esophageal adenocarcinoma (31). iNOS expression is also increased in esophageal adenocarcinoma (32) and in the lower esophageal sphincter in mouse model (34). Increased expressions of both iNOS and COX-2 have been shown in inflammatory bowel disease, such as ulcerative colitis (35) and Crohn's disease (24–26).

Two major opposing factors stand out among the mechanisms that determine the development of pathologic reflux: One is the lower esophageal sphincter, which serves as a gatekeeper against reflux, and the other is increased intragastric pressure, which promotes reflux. Studies in a mouse model for sepsis illustrate that in normal mice, the lower esophageal sphincter maintains a basal tone, but LPS cause a dose-dependent decrease in the basal tone, and the effect of LPS can be blocked by l-canavanine, which is a selective iNOS inhibitor (34). In this rodent model, LPS caused a selective increase in iNOS protein and mRNA in both the lower esophageal sphincter and internal anal sphincter without significant changes in the expression of other NOS isozymes. In LPS-treated mice, the increased iNOS activates the mitogen-activated protein kinase signaling pathway by phosphorylation of mitogen-activated protein kinases (36). Mitogen-activated protein kinases, including a family of serine and threonine kinases of extracellular signal-regulated kinase, c-jun-NH₂-kinase, and p38, convert external stimuli into a wide range of cellular responses, such as proliferation, survival, differentiation, and migration. Because of these critical functions, deregulated mitogen-activated protein kinases are often found to contribute to the development of many cancers (37). The increased gram-negative bacteria in the type II microbiome in reflux esophagitis and Barrett esophagus could serve as the trigger of the NF-κB pathway and be responsible for the upregulation of the iNOS gene. Thus, the type II microbiome might cause abnormal relaxation of the lower esophageal sphincter and contribute to the etiology of gastroesophageal reflux and carcinogenesis of esophageal adenocarcinoma.

Forward flow of gastric contents reduces the pressure and reduces the opportunity for reflux. In normal mice, the stomach is mostly empty between meals but is nearly full in LPS-treated mice (38). The LPS-delayed gastric emptying can be blocked by NS398, which is a selective COX-2 inhibitor (38). Thus, by reducing gastric emptying, the type II microbiome might cause an increase in the intragastric pressure that contributes to the development of gastroesophageal reflux.

The type II microbiome could serve as a marker and an important target of intervention in clinical practice. If it is proved to play a critical role in disease progression from reflux esophagitis to esophageal adenocarcinoma, the use of the type II microbiome as a biomarker might help to improve stratification of patients with reflux esophagitis and Barrett esophagus into high- versus low-risk groups, which would improve the sensitivity as well as specificity of a surveillance strategy for early detection of esophageal adenocarcinoma. Furthermore, the cancer risk may be reduced by reversion from the type II microbiome to type I microbiome with the use of selective antibiotics or probiotics.

Because of its association with a number of inflammatory and neoplastic diseases, the NF-κB pathway has been the target of drug development (39). Several drugs have been reported to block the pathway at various steps, and their clinical use has been described (40, 41). Many of these inhibitors could be effective in reduction of inflammation caused by NF-κB activation in reflux disorders. Drugs that are currently used in treating inflammatory diseases, such as glucocorticoids, nonsteroidal anti-inflammatory drugs, sulfasalazine, and immunosuppressive agents (cyclosporin A and tacrolimus), often interfere with the NF-κB pathway at multiple steps (Fig. 1). Because they interfere with normal cellular function necessary to mount immune responses, inhibitors of NF-κB may also cause significant side effects, such as increased susceptibility to infections and liver dysfunction. Although limited by clinical side effects, interest remains in the therapeutic potential of NF-κB inhibitors to halt the metaplastic progression of Barrett esophagus or to treat esophageal adenocarcinoma by inhibiting inflammation. Curcumin (diferuloylmethane), a naturally occurring NF-κB inhibitor, was recently shown to increase apoptosis in 2 esophageal adenocarcinoma cell lines and to enhance their sensitivity to chemotherapeutic agents (42).

iNOS has been an attractive drug target, as it is related to a variety of human diseases (43–46). Although l-canavanine, a selective iNOS inhibitor present in alfalfa, can block the LPS-induced lower esophageal sphincter relaxation in rodents (34), its relationship with lupus-like autoimmunity limits its direct use in humans (47). The prodrug L-N6-(1-iminoethyl)lysine 5-tetrazole amide (SC-51), another selective iNOS inhibitor, causes marked suppression of exhaled breath nitrous oxide levels both in healthy control subjects and in patients without significant side effects (48). It may have therapeutic potential for controlling reflux and/or microbiome-induced esophageal inflammation.

COX-2–selective inhibitors represent a form of nonsteroidal anti-inflammatory drug, such as aspirin, that directly targets COX-2, an enzyme expressed in inflammation of reflux esophagitis, Barrett esophagus, and esophageal adenocarcinoma (Fig. 1). Ingestion of nonsteroidal anti-inflammatory drugs decreases the inflammatory complications of gastroesophageal reflux disease (49) and may reduce the risk of neoplastic progression in patients with Barrett esophagus (50, 51). Thus, aspirin and other nonsteroidal anti-inflammatory drugs could protect against esophageal adenocarcinoma by either preventing the development of its primary precursor (i.e., Barrett esophagus) or by diminishing the likelihood of Barrett esophagus progressing to esophageal adenocarcinoma. The mechanism of potential risk reduction is related to these agents' inhibition of the COX-2 enzyme, which is expressed in reflux esophagitis, Barrett esophagus, and the early stages of esophageal carcinomas. Like NF-κB inhibitors, COX-2 inhibitors represent a potential therapeutic option in controlling reflux and/or microbiome-induced esophageal inflammation, which might trigger a progressive cascade to esophageal adenocarcinoma.

With these data, we can speculate on the roles of the type II microbiome in the diseases of reflux esophagitis, Barrett esophagus, and esophageal adenocarcinoma. The type II microbiome with stepwise increase in gram-negative bacteria in reflux esophagitis, Barrett esophagus, and probably in esophageal adenocarcinoma, could contribute to carcinogenesis by induction of chronic inflammation and cause gastric reflux by induction of abnormal relaxation of the lower esophageal sphincter and increase in intragastric pressure by delaying gastric emptying. These pathologic effects could be explained in part by the activation of the LPS/TLR4/NF-κB pathway. The type II microbiome could be used as a novel biomarker for risk assessment in clinical management. Antibiotic and/or probiotic treatment could reverse the type II microbiome back to the type I microbiome and decrease the detrimental effects of gram-negative bacteria on the LPS/TLR4/NF-κB pathway. The negative effects could also be alleviated using specific inhibitors to NF-κB and/or the downstream components, such as COX-2 and iNOS.

No potential conflicts of interest were disclosed.

Conception and design: L. Yang, Z. Pei

Development of methodology: L. Yang

Acquisition of data: L. Yang

Analysis and interpretation of data: L. Yang

Writing, review, and/or revision of the manuscript: L. Yang, F. Francois, Z. Pei

Study supervision: L. Yang

Supported by grants from the National Cancer Institute and the National Institute for Allergy and Infectious Diseases UH3CA140233, R01CA159036, R01AI063477, U19DE018385, K23CA107123, as well as the RWJ Amos Medical Faculty Development Program.