Immunoglobulin A and liver diseases

In the intestine, naïve B cells are found in the lamina propria immediately underlying the epithelium and are clustered in the gut-associated lymphoid follicles, such as the Peyer’s patches (PP). By presentation of bacterial antigens and aids from T cells in the germinal center of the gut-associated lymphoid follicles, naïve B cells are activated and become IgA-producing plasma cells. In addition, activation of B cells was found to also occur independently from T cells and specific antigens. In general, T cell-dependent IgA production is related to higher specificity and affinity to bacterial antigens than T cell-independent IgA [23]. There are several extensive reviews about this B cell activation and IgA production process in the gut [24, 25].

IgA-producing plasma cells translocate to the lamina propria and work as a main source of the gut IgA. In the lamina propria, plasma cells produce predominantly dimeric IgA and the dimeric IgA is transcytosed across gut epithelium by a membrane receptor, pIgR that is expressed on the basolateral surface of epithelial cells. At the apical surface of epithelial cells, the IgA-pIgR complex is cleaved and IgA bound to an outer membrane part of pIgR, which is known as secretory component (SC), is released as a secretory IgA (SIgA) [14]. Although SIgA produced in the lamina propria constitutes a major part of total IgA in the adult gut, there are two other sources of gut IgA: maternal milk IgA and liver-derived IgA. The milk SIgA is the only source of IgA in the gut of newborn mammals and contributes to the composition of the first gut microbiome [26], instead of endogenous SIgA that starts to be produced several months after birth in humans and slowly increases [27, 28]. On the other hand, liver-derived IgA is produced by plasma cells that colonize portal regions and the submucosa of biliary tracts, and is secreted to bile by pIgR on hepatocytes and cholangiocytes [29, 30]. In the human liver, however, pIgR is expressed only on cholangiocytes [29]. The proportion of the liver-derived SIgA to total gut SIgA varies between species: low in human and dog, moderate in mouse, and high in rabbit and rat [29]. There is a comprehensive review discussing liver IgA [29].

Although precise mechanisms how SIgA maintains gut homeostasis and protects the host from pathogenic stimuli of bacteria are not known, several mechanisms have been suggested [14, 15]. SIgA prevents the access of bacteria to the apical surface of epithelial cells by agglutinating bacteria and/or entrapping bacteria in the mucus layer. Moreover, binding of immunoglobulins can alter gene expression of some bacteria to reduce their virulence or motility [31, 32]. In addition to direct effects on bacteria, SIgA neutralizes bacterial toxins and protects epithelial cells from the toxic stimulation [14, 15]. Meanwhile SIgA binding inhibits bacterial access and invasion to epithelial cells, SIgA binding to luminal bacteria also promote sampling of the bacteria to PPs, leading to induction of a strong response against the bacteria by B and T cells [33]. In general, SIgA exerts these functions without induction of deleterious inflammation at the site [25]. Although precise functions of SIgA that regulate host-microbiota homeostasis in the gut are not well defined, deficiency of SIgA in mice led to altered microbiome in the gut [17, 18] and to increased susceptibility to both pathogens and toxic reagents [17, 34].

There are several genetically modified mice which lack or have decreased SIgA in the gut lumen, and these models allow us to investigate the role of SIgA on gut homeostasis and host health. The effects of SIgA deficiency on models of infectious disease have been reviewed [14]. Here, we will focus on IgA influences on gut microbiota composition and gastrointestinal infection of bacteria. Iga ^−/− mice that lack the IgA switch region and the IgA constant region showed altered microbiota shift from newborn to adult [17]. Iga ^−/− mice showed increased and persistent colonization of Enterobacteriaceae family, which is predominant (> 75%) in newborn mice and decreased (< 1%) during the first 6 weeks [17]. It has been reported that the Enterobacteriaceae bacteria family is increased in patients with Crohn’s disease [2, 35], children with type 1 diabetes [36], and cirrhotic patients [21]. In addition, Iga ^−/− mice exhibited sustained intestinal inflammation and increased susceptibility to dextran sodium sulfate (DSS)-induced injury [17]. However, Iga ^−/− mice showed similar susceptibility to oral infection of Citrobacter rodentium infection compared with wild-type mice [37].

There are two other animal models that lack SIgA in the gut: pIgR-deficient and ‘joining’ chain (J chain)-deficient mice. These two mice cannot properly secret IgA or IgM into the gut lumen [14]. Decreased abundance of Bifidobacterium spp., which are known as beneficial bacteria in general, and increased abundance of Helicobacter spp. were observed in the cecum of the Pigr ^−/− mice [38]. Moreover, the Pigr ^−/− mice were susceptible to Salmonella typhimurium infection via the fecal–oral route [39] and more vulnerable to DSS-induced colitis than wild-type mice [34, 38]. However, Pigr ^−/− exhibited similar susceptibility to Citrobacter rodentium [37] and Clostridium difficile as compared with wild-type mice [40]. Immunized J chain-deficient mice were not protected from Cholera toxin [41], while they exhibited similar clearance of Citrobacter rodentium as wild-type mice [37]. There is no clear reason why these mice, lacking gut SIgA, showed normal susceptibility to several bacterial infection models. It is speculated that SIgA effects on simple bacterial infection model could be compensated by other type of immunoglobulins, such as SIgM and IgG, and/or antimicrobial peptides. Indeed, Iga ^−/− mice exhibited increased SIgM level in the gut [42] and mRNA of antimicrobial peptides were increased in the colon epithelium of Pigr ^−/− mice [38].

In humans, individuals with selective IgA deficiency have a tendency to develop infections of the gastrointestinal tract (e.g., Giardia infections), celiac disease, and inflammatory bowel diseases [43]. Although selective IgA deficiency showed association with these diseases, 85–90% of IgA-deficient people are asymptomatic. However, there are few reports investigating the characteristics of gut microbiome of human IgA deficiency. Friman’s group demonstrated that IgA-deficient individuals more often had the genes involved in virulence of Escherichia coli in their rectal flora, although only several virulent factors of E. coli were examined [44]. Future studies that investigate IgA-deficiency on the human gut microbiome, will be important for this field of research.

Dietary factors including alcohol directly and indirectly influence the gut microbiota [45‐48]. Intestinal bacterial overgrowth and bacterial dysbiosis after chronic alcohol consumption were observed in animals and human [47]. Intragastric alcohol feeding was associated with bacterial overgrowth in the large intestine as early as 1 week after feeding in mice [49]. Binge drinking of alcohol leads to elevation of blood endotoxin level. Alcohol and its metabolite acetaldehyde disrupts tight junction of epithelial cells and increase intestinal permeability [46]. Mice fed with high-fat diet exhibited altered gut microbiota composition [50, 51] and the change was independent from obesity [51], indicating high-fat diet itself influences gut microbiome composition. In addition, high-fat diet increased intestinal permeability through reduction of tight-junction proteins and induction of intestinal inflammation, leading to elevated blood endotoxin [45].

The mechanisms how altered gut microbiome contributes to development and progression of liver diseases were previously reviewed [20, 21, 47, 48, 52, 53]. Lipopolysaccharide (LPS), known as endotoxin, is a cell-wall component of gram-negative bacteria and interacts with Toll-like receptor 4 (TLR4). As mentioned above, increased level of circulating LPS was observed in patients with ALD and rodent models of ALD [54]. Increased endotoxin is also observed in alcoholic patients with minimal symptoms of ALD [55] and healthy subjects with single binge drinking [56]. In addition, the LPS levels correlate with disease severity [57, 58]. In mice, a LPS increase was observed both by acute binge gavage and chronic feeding with ethanol [59]. In addition to LPS, other bacterial components, such as bacterial 16S DNA and peptidoglycan which is a cell-wall component of gram-positive bacteria, were also increased in the circulating blood [56, 60]. In general, liver-resident macrophages, Kupffer cells, are tolerant to LPS-induced TLR4 activation and eliminate microbial components without inflammatory response. However, prolonged and excessive exposure to LPS can make Kupffer cells sensitive to LPS [20]. Moreover, ethanol-induced hepatocyte damage is associated with the liver macrophage activation through the damage associated molecular pattern production or signaling molecule-containing exosome production [61, 62]. Thus, sensitized Kupffer cells by ethanol-stimulated hepatocytes respond to gut-derived bacterial and fungal components and promote local inflammation in the liver. Other types of microbial products that are derived from the gut, such as bacterial DNA and cell-wall components, are also possible mediators of liver inflammation [20]. Recently, we found 1,3-β-glucan, which is a cell-wall component of fungi, is also increased in alcohol-fed mice and anti-Saccharomyces cerevisiae IgG antibody is significantly higher in alcoholic cirrhosis patients than in healthy individuals or viral cirrhosis patients [63]. Translocating fungi cell-wall component promoted IL-1β processing and local inflammation via its receptor, dectin-1, on liver macrophage cells [63].

Diet-induced endotoxemia is observed in rodent models of obesity and NAFLD/NASH [64‐67]. In humans, NAFLD/NASH patients have higher level of systemic endotoxin when compared with healthy people [68, 69]. Infiltration of LPS into the liver has been shown to exacerbate liver injury in NAFLD/NASH mice, possibly via TLR4 activation [70]. Predisposition to obesity, which is a common factor attributed to NAFLD/NASH development, has been shown to be influenced by gut microbiota [71, 72]. Germ-free mice, which do not have any microorganisms in their body, are tolerant to high-fat diet-induced weight gain. By transplantation of gut microbiota from obese mice, but not from lean mice, germ-free mice gained more fat by high-fat diet [73]. That could be mediated by increased energy extraction from diet via altered microbiome. In addition, the production of fasting-induced adipocyte factor (Fiaf) is suppressed by enteric bacteria, resulting in increased activity of lipoprotein lipase and increased accumulation of triglycerides in the liver [21, 71, 72]. High-fat diet altered microbiota composition toward metabolizing dietary choline to methylamines, resulting in reduced level of plasma phosphatidylcholine [21, 74]. Choline-deficient diet, which also reduces phosphatidylcholine, is used to create rodent model of NASH. Reduced level of phosphatidylcholine, which is necessary for assembly of very-low-density lipoprotein (VLDL), results in triglyceride accumulation in the liver [21]. In addition, obese patients and children with NASH showed increased blood ethanol concentration [10, 75]. Increased ethanol produced by altered gut microbiome could possibly be a mediator of liver damage [21, 52].

IgA has been studied in the field of alcoholic liver disease for a long time [76]. Brown and Kloppel reviewed IgA levels in serum and the liver of ALD patients [29]. Systemic IgA levels are increased in patients with alcoholic liver disease [29] and in a rodent model of alcoholic liver disease [77]. In addition, liver IgA deposits appear specific for alcoholic liver disease, correlate with the amount of alcohol consumption, but not with serum IgA concentrations [78‐80]. Moro-Sibilot and colleagues demonstrated that mouse and human liver contain IgA-secreting cells. Increased amount of IgA-secreting cells in mice fed with ethanol was diminished by inhibition of IgA-plasmablast egress from Peyer’s Patches with FTY720 that modulates sphingosine-1-phosphate receptor 1 and inhibits the egress of lymphocytes form secondary lymphoid organs, indicating that IgA deposits in the liver of ALD patients is attributed, at least partially, to increased entry of IgA-secreting cells from Peyer’s Patches to the liver [30].

SIgA levels in the gut of chronic liver diseases seems more important than systemic and liver IgA changes because SIgA regulates gut homeostasis. Studies that examined IgA levels in the gut of alcoholic patients and rodent models of ALD and NAFLD/NASH are summarized in Table 1. The number of IgA-secreting plasma cells in the intestinal lamina propria [30] and fecal IgA levels [77], which reflect the amount of colonic IgA, are reduced in experimental models of ethanol-induced liver disease of mice. These decreased SIgA and/or IgA-secreting cells in the intestine are reproducibly observed in rodent models of chronic alcohol feeding [81, 82], but not in rat model of single administration with ethanol [83‐85]. Studies investigating intestinal IgA of patients with alcoholics are very limited [86‐88]. Pelletier et al. observed decreased intestinal secretion of SIgA in patients with severe alcoholic cirrhosis when compared to patients with compensated alcoholic cirrhosis [86]. In contrast, Colombel et al. showed similar levels of jejunal SIgA secretion among alcoholics with cirrhosis, alcoholics without cirrhosis, and control subjects [87]. Maier and colleagues observed similar number of IgA-secreting cells in duodenum between alcoholics and controls, but a slightly decreased IgA amount in IgA-secreting cells of alcoholics with recent alcohol intake [88]. These conflicting reports may result from subjects in different stages of the disease or different methods of IgA measurement. Recently, we investigated the role of IgA in a mouse model of ethanol-induced liver disease by using Iga ^−/− mice. Unexpectedly, Iga ^−/− mice fed with ethanol showed similar degrees of liver injury (plasma ALT), hepatic triglyceride accumulation, and liver inflammation as compared with wile-type littermates [77]. Interestingly, we observed increased bacterial translocation to mesenteric lymph node in Iga ^−/− mice fed with control diet, but no further increase of the translocation in Iga ^−/− mice fed with ethanol. The result suggests that exposure of the liver of Iga ^−/− mice to gut-derived bacterial component is increased under physiological conditions and the increased translocation is further regulated by other factors than SIgA during disease. One possible reason why loss of IgA did not affect disease development is a compensation by SIgM. The Iga ^−/− mice fed with ethanol exhibited increased level of fecal IgM [77]. Further studies investigating compensatory roles of intestinal IgM and/or antimicrobial peptides are necessary.

Unlike the case of ALD, reports that investigated an association between serum IgA level and NAFLD/NASH are limited [89‐91]. Tomita et al. showed that serum IgA concentration is increased in severe-NASH patients when compared with early-stage NASH patients, although IgA concentrations were comparable among control, simple steatosis, NAFLD, and early-stage NASH [89]. McPherson’s group showed that serum IgA was elevated in NASH patients as compared with simple steatosis patients [90]. In these studies, serum IgA levels were positively associated with severity of liver fibrosis [89‐91]. However, associations between increased serum IgA and advanced stage are also observed in other types of chronic liver diseases. Thus, increased systemic IgA levels in NAFLD/NASH patients may not be a specific marker for the disease progression. Although there is no study that investigated intestinal or fecal IgA levels of NAFLD/NASH patients, several studies using rodent models of NAFLD/NASH showed altered intestinal IgA secretion [92‐96]. In the study of Matsumoto’s group, mice with NASH induced by methionine-choline-deficient diet, exhibited decreased fecal IgA and IgA-positive cells in ileal and colonic tissues [94]. In addition, Li et al. showed decreased IgA in small intestinal mucosa of rats fed with high-fat diet, and a negative correlation between the intestinal IgA levels and systemic endotoxin levels [95]. However, similar IgA levels of feces between high-fat and control groups or increased fecal IgA in high-fat group were observed in several studies [92, 93, 96]. Different feeding protocol might explain these results. In the study from Li’s group, decreased SIgA in the small intestine was not observe in 8-week feeding group, but observed in 12-week feeding group [95]. Although the mechanisms how increased systemic IgA and/or reduced intestinal SIgA are involved in disease development and progression are not clear yet, some gut dysbiosis and bacterial overgrowth observed in ALD and NAFLD/NASH might contribute to decreased SIgA levels in the gut.

The study of gut SIgA functions in liver diseases is the field that started recently, and SIgA behavior in the gut of human alcoholic patients has not been intensively investigated yet. However, several studies revealed shared features of gut microbiome changes between SIgA-deficient mice and human patients with alcoholic liver disease. As mentioned above, Iga ^−/− mice exhibited increased colonization of Enterobacteriaceae family (in phylum Proteobacteria) [17]. Increased abundance of phylum Proteobacteria is a signature of intestinal dysbiosis associated with metabolic and inflammatory diseases [97]. Moon et al. found dichotomous fecal IgA levels among wild-type mice in the same facility and the IgA-low phenotype was driven by transmissible fecal bacteria [98]. This IgA-low phenotype of mice was associated with enriched Enterobacteriaceae family. Although increased Enterobacteriaceae family was not observed in IgA-low mice from a different facility in the same study, the phylum Proteobacteria was increased in the mice [98]. These results indicate decreased SIgA may drive the expansion of Enterobacteriaceae family (or phylum Proteobacteria) in the gut of chronic liver diseases, such as ALD and NAFLD/NASH. Increased members of the Enterobacteriaceae family or phylum Proteobacteria is also observed in chronic liver diseases [21]. Zhu et al. and colleagues revealed that abundance of Escherichia spp. (in Enterobacteriaceae family) and Enterobacteriaceae family are increased in stool of children with NASH when compared with healthy children or children with obesity [10]. Increased Gammmaproteobacteria class, which includes Enterobacteriaceae family, was also observed in children with NAFLD when compared with healthy children in a different study [99]. A study in which gut microbiota composition were compared between 53 NAFLD patients and 32 healthy subjects showed increased Escherichia spp. in NAFLD patients [100]. Increased abundance of phylum Proteobacteria and/or Enterobacteriaceae family were reproduced in rodent models of NAFLD [101‐104]. However, the enrichment of Enterobacteriaceae and Escherichia spp. was not reproduced in human studies [105, 106]. Taken together, decreased amount of SIgA in the gut lumen, which was observed in rats fed with high-fat diet [95], may contribute to expansion of phylum Proteobacteria and/or Enterobacteriaceae family of bacteria and disease pathogenesis.

The prominent feature of dysbiosis observed in alcoholic patients and rodent model of alcoholic liver disease is characterized by a decreased phylum Firmicutes and lower amount of Lactobacillus spp. (in phylum Firmicutes) [49, 107‐110]. Probiotic treatment with Lactobacillus spp. and Bifidobacterium spp. improved bacterial dysbiosis and ameliorated liver injury and/or inflammation in both humans and animal models (previously reviewed by Bluemel et al. [111]). In addition to these bacteria, alcoholic patients with dysbiosis showed a higher relative abundance of Gammmaproteobacteria class (in phylum Proteobacteria) than alcoholic patients without dysbiosis [112]. Relative abundance of Enterobacteriaceae family was higher in the stool of patients with alcohol hepatitis than alcoholic patients without hepatitis [113]. Increased Enterobacteriaceae is reproducibly observed in human cirrhotic patients [9, 11, 114]. Bajaj et al. proposed the cirrhosis dysbiosis ratio (CDR). The ratio of “good” bacteria, such as Lactobacillus spp., against potentially “bad” bacteria, such as Enterobacteriaceae was lower in advanced cirrhotic patients as compared with that in compensated-cirrhotic patients [58]. Especially, cirrhotic patients with alcoholic etiology showed increased Enterobacteriaceae family [58]. Although these facts suggest a possible link between reduced SIgA and development of the disease, studies investing direct effects of SIgA on the diseases are limited. Further studies investigating the change of gut microbiome and disease status in SIgA-deficient mice are required.