Introduction

Humans are colonized by residential microbes including bacteria, archaea, eukaryotes, and viruses [1]. Until recently, the properties of the microbiome of humans were largely unknown. The proportion of bacterial cells and genes present in the human body are estimated to be ∼90 and >99 %, respectively. Initial colonization occurs at the time of birth, and humans progressively acquire ∼1014 bacterial cells at equilibrium, which remain for life. Recent studies of the human microbiome revealed that even healthy individuals differ remarkably in the microbes that occupy the gut, oral cavity, skin, and vagina. The Human Microbiome Project by the US National Institutes of Health has produced a 2.3-TB 16S ribosomal RNA metagenomic data set of over 35 billion reads taken from 690 samples from 300 US subjects, across 15 body sites [1]. Large-scale endeavors (for example, the Human Microbiome Project and the European project, MetaHIT) are already providing a preliminary understanding of the biology and medical significance of the human microbiome and its collective genes [3]. The aim of these projects is to characterize the compositional range of the “normal” microbiome of healthy individuals. Important questions concerning the commonalities and differences among healthy individuals in both microbial taxa and functional pathways are being addressed.

Nonalcoholic fatty liver disease (NAFLD) is an important cause of chronic liver injury in many countries [37]. Data collected from the USA showed that the prevalence of NAFLD has increased steadily in recent years, despite other diseases remaining at steady states [8]. NAFLD ranges from benign simple steatosis to steatohepatitis (including nonalcoholic steatohepatitis, NASH). This latter condition includes progressive fibrosis [9] and hepatocellular carcinoma [1012]. The prevalence of obesity and its associated disorders, metabolic syndrome, have increased risk for many diseases such as NAFLD, atherosclerosis, and certain cancers. Moreover, studies on the relationship of intestinal microbial flora with obesity have identified profound changes in the composition and metabolic function of the intestinal microbiota in obese individuals [1, 1315], which appear to enable the “obese microbiota” to extract more energy from the diet [16]. These facts suggest that microbiota could play an important role in obesity, NAFLD, and its related diseases. However, the relationship between this “obese microbiota” and the pathogenesis of NAFLD has not yet been elucidated. In this article, we will discuss current evidence regarding the potential role of microbiota in the development of NAFLD, with a special focus on inflammation and obesity-related metabolic disorders.

Inflammation and NASH

An interesting working model known as the “two-hit” theory postulates the progression from simple steatosis to NASH, fibrosis, or cirrhosis. The “first hit” consists of the accumulation of excessive hepatic fat owing to insulin resistance and inflow of free fatty acid (FFA) (Fig. 1). Often, this step is present in patients with metabolic syndrome, and although it is not sufficient to cause NASH, it can predispose the liver to chronic inflammation. Oxidative stress caused by reactive oxygen species (ROS), gut-derived lipopolysaccharide (LPS), and soluble mediators synthesized from immune system cells and adipose tissue cells have been indicated as risk factors for the “second hit” [17, 18] (Fig. 1). Although the model of the “two-hit theory” quickly spread through the scientific world, it seems obvious that different factors are necessarily interacting. In the last few years, studies in animal models of NAFLD have provided new insights into the molecular and physiologic alterations responsible for the switch from steatosis to steatohepatitis. In this regard, several groups demonstrated that in NAFLD progression, complex multifactorial interactions between genetic determinants, nutritional factors, and dysmetabolism participate in the development of hepatocellular damage and progressive liver disease. Lipotoxic effects of FFAs and lipid intermediates impair the normal functions of liver cell organelles involved in the production of ROS, the activation of pro-inflammatory defense programs, and ultimately apoptosis, by mechanisms that are not fully understood. Toxic lipids and cytokine release contribute to impaired insulin signaling, which in turn causes diminished very-low-density lipoprotein (VLDL) assembly and liver secretion, involving insufficient regulation of important transcription factors required for lipogenesis [19].

Fig. 1
figure 1

Multifactorial complex interactions leading to hepatic steatosis and its progression. A liver loaded with lipids consisting primarily of triglycerides (first hit) might reflect a benign process because they might exert mostly protective effects. When the capacity of peripheral and central organs to detoxify aggressive lipids, including free fatty acid (FFA) (second hit) fails, lipotoxic attack of the liver might begin. Moreover, adipocytokine, lipopolysaccharide (LPS), and reactive oxygen species (ROS) may enhance the first and/or second hits, resulting in progression of steatohepatitis pathogenesis

Inflammation is a physiological response of an organism to harmful physical, chemical, or biological stimuli. The response usually concludes with the reestablishment of homeostasis. It involves the coordinated action of many cell types and mediators, whose intervention depends on the nature of the initial stimulus and ensuing responses. The normal acute inflammatory response involves the delivery of plasma components and leukocytes to the site of insult and is initiated by tissue-resident macrophages and mast cells leading to a production of different types of inflammatory mediators (cytokines, chemokines, vasoactive amines, eicosanoids, and products of proteolytic cascades) [20]. The inflammatory state that accompanies the metabolic syndrome shows a quite peculiar presentation, as it is not accompanied by infection or sign of autoimmunity, and no massive tissue injury seems to occur. Furthermore, the dimensions of inflammatory activation are not large, and so it is often called “low-grade” chronic inflammation. Other researchers have attempted to name this inflammatory state “metainflammation,” meaning metabolically triggered inflammation [21] or “parainflammation” as a term to define an intermediate state between basal and inflammatory states [22]. Whatever the term used, the inflammatory process that characterizes the metabolic syndrome has unique features, and its mechanisms are far from being fully understood [24]. Metabolic syndrome, particularly NASH, is thought to develop through “metainflammation.” Indeed, chronic low-grade inflammation is an important contributing factor in NASH pathogenesis [24, 25].

The liver is continually exposed to gut-derived factors including bacteria and bacterial components because the portal vein is the direct venous outflow of the intestine. The liver is an important site for bacterial phagocytosis and clearance as it contains the largest population of tissue macrophages. Activated Kupffer cells, resident macrophages of the liver, exposed to pro-inflammatory mediators such as LPS, membrane components of gram-negative bacteria, or other bacterial products, are the major source of inflammatory mediators including pro-inflammatory cytokines, chemokines, and reactive oxygen/nitrogen species, which contribute to liver injury [26]. Through pattern recognition receptors, including Toll-like receptors (TLRs), the innate immune system recognizes conserved pathogen-associated molecular patterns (PAMPs) [27]. The healthy liver expresses low mRNA levels of TLRs such as TLR1, TLR2, TLR4, TLR6, TLR7, TLR8, TLR9, and TLR10, implying a high tolerance of the liver to TLR ligands from the microbiota, to which it is constantly exposed. Signaling through TLRs plays a major role in the physiology and pathophysiology of the liver [28]. LPS is a potent activator of innate immune responses through binding to the TLR4 complex. TLR4 is expressed by Kupffer cells, hepatic stellate cells, hepatocytes, biliary epithelial cells, sinusoidal endothelial cells, and hepatic dendritic cells, which are consequently responsive to LPS [28]. There is a positive correlation between liver dysfunction and the occurrence of bacterial translocation and increased LPS. Furthermore, the clearance of LPS from the circulation is decreased in states of hepatic dysfunction, such as cirrhosis [29]. Downstream targets of TLR4 signaling are determined by selective recruitment of cytosolic sorting and signaling adaptor proteins via interactions between Toll-IL-1 receptor (TIR) domains [3032]. Thus, TLR4 activation may engage myeloid differentiation factor 88 (MyD88) and TIR domain-containing adaptor protein or MyD88 adaptor-like factors, leading to the activation of nuclear factor kappa-B (NF-κB) and AP-1 transcription factors [3335]. There is substantial evidence that TLR4-mediated cellular events escalate liver injury in steatosis [33, 35]. Recent studies indicated that TLR4 sorting specificity might reflect the etiology of fatty liver disease. Due to the ubiquitous presence of TLR4 among various types of liver cells, the specific role of Kupffer cells in differential activation of TLR4 pathways remains to be determined. It must also be noted that endogenous ligands such as certain FFA and other alarmins may also be linked to TLR4 sorting specificity, a question particularly relevant to NAFLD.

Kupffer cells produce tumor necrosis factor-α (TNF-α) and interleukin (IL)-10 in response to physiological concentrations of LPS [3638]. TNF-α is a proinflammatory cytokine that activates various signal transduction cascades, including many of the pathways that are discussed below, as critical inhibitors of insulin action. It is overexpressed in the adipose tissues and livers of obese rodents and humans, and its concentration is reduced after weight loss. Geoffrey et al. showed that TNF-α and NF-κB activation was essential for hepatic inflammatory recruitment in steatohepatitis by using methionine choline-deficient (MCD) diet-fed TNF-α and TNF-receptor 1 knockout animals, and in vivo transfection of wild-type (WT) mice with nondegradable mutant-IκB [39]. Furthermore, such NF-κB activation occurred independently of TNF-α. Other studies using the MCD dietary model produced conflicting findings; curcumin, which blocks oxidative stress-mediated NF-κB activation, provided protection [40], but TNF-α antiserum reduced liver injury in rats administered the MCD diet [41]. Tomita et al. demonstrated that TNF-receptor knockout mice were protected against liver fibrosis in their MCD experiments [42].

Alternatively, activated macrophages (also termed M2 as opposed to the classical M1 or proinflammatory phenotype), including Kupffer cells, represent another critical pathway for the resolution of inflammatory responses [43]. The coordinated program of alternative activation is primarily stimulated by Th2 cytokines IL-4 and IL-13 and characterized by the cell surface expression of M2 signature genes such as the mannose receptor, arginase-1, and dectin-1 [43]. There is evidence that steatosis promotes Th1 polarization of the cytokine balance, which favors the innate or classic activation of macrophages in NAFLD. Thus, in experimental and human NAFLD alike, the pool of hepatic natural killer T cells (NKT) is reduced, and liver tissue levels of Th1 cytokines, such as TNF-α, IL-12, IL-18, and interferon-γ, are elevated [4447]. These studies suggest that microbiota-induced “metainflammation” associated with pro-inflammatory cytokines is critical in the pathogenesis of NASH, but a complex and still poorly characterized interaction between microbiota and the innate immune system for “metainflammation” may be involved in metabolic dysfunction.

Intestinal microbiota and NASH

The human gut contains an immense number of microorganisms, collectively known as the intestinal microbiota. This community consists of at least 1013 citizens, is dominated by anaerobic bacteria, and includes 500–1,000 species whose collective genomes are estimated to contain 100 times more genes than our own human genome [47, 48]. The duodenum and proximal jejunum normally contain small numbers of bacteria, usually lactobacilli and enterococci, gram-positive aerobes or facultative anaerobes (<104 organisms/mL). Coliforms may be transiently present (<103 bacteria/mL), and anaerobic Bacteroides are not found in the jejunum in healthy people. Up to one third of jejunal aspirates might be sterile in healthy volunteers. The distal ileum is a transition zone between sparse populations of aerobic bacteria of the proximal small intestine and very dense populations of anaerobic microorganisms in the large bowel [4951]. The epithelial surface of the small intestine in a healthy human is not colonized. Occasional groups of bacteria can be found in low concentrations within the lumen. Bacteria do not form clusters and spatial structures, and the luminal contents are separated from the mucosa by a mucus layer [52].

Important studies on the relationship of intestinal microbial flora with obesity have identified profound changes in the composition and metabolic function of the intestinal microbiota in obese individuals (obese microbiota), as described above. Moreover, these studies demonstrated that intestinal microbiota interact with host epithelial cells to indirectly control energy expenditure and storage [13], and activated inflammatory responses in NASH pathogenesis. Several reports demonstrating an association between microbiota and NASH are shown in Table 1.

Table 1 Animal and clinical studies regarding microbiota and NAFLD

Intestinal microbiota and metabolic disorders

Intestinal microbiota benefits the host in numerous ways, among them the capability to extract calories from otherwise indigestible common polysaccharides in the diet via enzymes such as glycoside hydrolases and others that are not encoded within the human genome (Fig. 2) [5355]. The first evidence that intestinal microbiota may be associated with body weight and composition was reported by Bachhed et al. [13]. They analyzed germ-free mice and conventionally raised mice that were allowed to acquire intestinal microbiota from birth to adulthood [13]. Compared to germ-free mice, conventional mice developed more adipose tissue, a higher percentage of body fat, and a twofold increase in hepatic triglyceride content, accompanied by increased hepatic mRNA expression of sterol-responsive element-binding protein (SREBP-1) and carbohydrate-responsive element-binding protein (ChREBP), two nuclear positive regulators of lipogenic enzymes [13], despite eating smaller amounts of the same diet (Fig. 2). Mice hosting intestinal microbiota also had higher levels of leptin and insulin resistance [58]. When they transplanted normal microbiota harvested from the distal intestine of conventionally raised mice into adult germ-free mice, a 57 % increase in body fat content and insulin resistance was observed [56] associated with decreased intestinal expression of fasting-induced adipose factor (Fiaf), also known as angiopoietin-like factor, a circulating lipoprotein lipase (LPL) inhibitor. This favored fatty acid uptake and adipose tissue expansion (Fig. 2) [56]. Conventionalization of germ-free mice suppressed the expression of Fiaf in gut epithelial cells [13]. Increased adipocyte LPL activity resulted in increased cellular uptake of fatty acids and adipocyte triglyceride accumulation (Fig. 2). Germ-free Fiaf−/− mice have the same total body fat weight as conventionalized mice, suggesting Fiaf is a mediator of microbial regulation of energy storage [13]. In contrast, mice fed a high-fat diet (HFD) complemented with Lactobacillus paracasei exhibited significantly reduced body fat, paralleled by increased circulating levels of Fiaf [57]. L. paracasei upregulated Fiaf expression in colonic epithelial cell lines, and oral inoculation of germ-free mice with this species resulted in increased circulating Fiaf levels [57]. Fiaf also induces peroxisomal proliferator activated receptor gamma coactivator 1α (PGC1-α) that regulates the expression of enzymes involved in fatty acid oxidation (Fig. 2) [58]. Therefore, Fiaf appears to have an important role in central regulation of energy metabolism, leading to exacerbation of insulin resistance.

Fig. 2
figure 2

Proposed mechanisms of the effects of intestinal microbiota on host metabolic disorder. (1) The digestion of polysaccharides by microbial enzymes leads to increased hepatic de novo lipogenesis. (2) The decreased expression of fasting-induced adipose factor (Fiaf) leads to increased lipoprotein lipase (LPL) activity, resulting in increased hepatic uptake of free fatty (FFA) acid and decreased expression of peroxisomal proliferator-activated receptor coactivator-1a (PGC-1α). These cause decreased hepatic oxidation of FFA. (3) The hepatic expression of sterol-responsive element-binding protein (SREBP-1) and carbohydrate-responsive element-binding protein (ChREBP) is increased through insulin resistance, following increased de novo lipogenesis. These effects lead to obesity and hepatic lipid accumulation

Insulin resistance and hyperinsulinemia were the most closely associated laboratory findings with the presence of NAFLD in a large series of patients [5961]. NAFLD prevalence is increased in individuals with impaired glucose tolerance (43 %) and those with newly diagnosed diabetes mellitus (DM) (62 %) [62]. In a prospective study of 100 patients with type 2 DM, the incidence of hepatic steatosis was 49 %, confirming this strong independent risk factor for NAFLD [63]. In Japan, the prevalence of metabolic syndrome and DM among patients with NAFLD was estimated at 25.4 and 16.2 %, respectively [64]. The association of both conditions is related with more aggressive disease and increasing mortality [65, 66]. Targher et al. described increased prevalence of NAFLD and its association with cardiovascular disease in diabetic patients, independent of other confounding factors [67]. The occurrence of insulin resistance in the liver, characterized by reduced insulin-suppressing effects in hepatic glucose production, aggravated peripheral insulin resistance and contributed to hepatic lipogenesis [68]. Consequently, hyperinsulinemia caused increased hepatic synthesis of FFAs and decreased synthesis of apolipoprotein B-100, leading to triglyceride accumulation in the liver [69]. Thus, elevated FFA levels caused by insulin resistance in adipocytes lead to the decreased suppression of lipolysis by insulin [70]. The liver is not only an end organ for the effects of insulin resistance, but also regulates carbohydrate homeostasis, under the control of insulin signaling through its specific receptor [71]. The insulin receptor belongs to a subfamily of tyrosine kinase receptors including insulin-like growth factor-1 and insulin-related receptor [72]. When insulin binds its receptor, insulin receptor substrate (IRS)-1 and IRS-2 are phosphorylated and mediate insulin signaling in the liver, leading to the recruitment and activation of IRS by tyrosine phosphorylation, which recruits signaling molecules containing Src homology-2 domains, including phosphatidylinositol 3-kinase, growth factor receptor-bound protein 2, and SH2 domain containing tyrosine phosphatase. In turn, this activates downstream effectors that mediate the metabolic effects of insulin [73, 74]. In addition, IRS proteins undergo serine phosphorylation, which attenuates insulin signaling by decreasing insulin-stimulated tyrosine phosphorylation [75, 76]. Tyrosine phosphatases and serine kinases including c-Jun N-terminal kinase [71, 77], protein kinase C [78], inhibitor κB kinase complex [76], and mitogen-activated protein kinase are involved in these pathways [79]. Depending on which pathway is impaired, elevated circulating levels of insulin and glucose may play a crucial role in pathogenesis of NASH by hepatocyte injury.

Small intestinal bacterial overgrowth

Qualitative or quantitative imbalances of complex intestinal microbiota might have serious health consequences for a macroorganism, including small intestinal bacterial overgrowth syndrome (SIBO). Initial evidence for an altered microflora associated with obesity came from studies in the leptin-deficient ob/ob mouse model. 16S rRNA sequencing of the distal intestinal microbiota of ob/ob mice, lean ob/+, and wild-type siblings and their ob/+ mothers, all fed the same diet, revealed that ob/ob mice exhibit a major reduction in the abundance of Bacteroidetes and a proportional increase in Firmicutes [14]. Feeding of a high-fat/high-polysaccharide diet to genetically WT rodents led to similar microbial changes and SIBO [80]. Confounding factors affecting microbial composition and function may include diet; the use of antibiotics, which substantially reduces bacterial diversity [81]; and effects related to the genetic background of animal models [82]. Consistent with animal models, Ley, et al. observed analogous differences with an increase in the ratio of Firmicutes/Bacteroidetes in the distal intestinal microbiota in human obesity [83]. Another study demonstrated that Firmicutes were dominant in lean and obese individuals and decreased in three patients undergoing Roux-en-Y gastric bypass surgery [84]. In contrast to earlier studies, Zhang et al. [85] described that Prevotellaceae, a subgroup of Bacteroidetes, were significantly enriched in obese individuals, again raising the potentially important issue of diet as a confounding factor. Patients in the Ley study [83] were either on a fat-restricted or carbohydrate-restricted diet, whereas in the Zhang study, researchers did not limit dietary components. Another study also described a decrease of Bacteroidetes in obesity and an increase in Firmicutes [86]. Overweight pregnant patients also have reduced numbers of Bifidobacteria and Bacteroidetes, but increased numbers of certain Firmicutes or Proteobacteria [87].

The overall prevalence of SIBO in the public is unknown. In general, SIBO is substantially underdiagnosed. Some patients may not seek health care, or SIBO may not be properly diagnosed by medical investigations. SIBO might be asymptomatic or with nonspecific symptoms only, or all symptoms might be incorrectly ascribed to the underlying disease (leading to SIBO). Of course, diagnostic yield also depends on the methods used for investigation. According to different studies investigating small sets of clinically healthy people as controls, findings consistent with SIBO were found in 2.5 to 22 % [8896]. Thus, SIBO may coexist with NASH (Table 1). Wan et al. reported that excessive multiplication of Escherichia coli coexisted in NASH rats, consistent with previous studies. This suggested SIBO may be one of many factors important in the pathogenesis of NASH, as antibacterial treatment could alleviate the severity of NASH (Table 1) [97]. In addition, levels of ALT increased or decreased relative to serum levels of TNF-α [97]. This strongly supported TNF-α as an important mediator for the promotion of NASH by SIBO. Generally, endotoxemia is thought to be a link between SIBO and elevated TNF-α levels [98, 99]. Furthermore, Wigg et al. found a higher prevalence of SIBO (11/22, 50 %) in patients with NASH, compared to healthy control subjects (5/23, 22 %) (Table 1) [97]. Higher values for the xylose–lactulose test in patients with NASH correlated with higher serum levels of TNF-α. However, they were not associated with increased serum endotoxin [97]. In another study of NASH, SIBO was diagnosed in half of the patients (6/12) but only in one subject (1/11, 9 %) in the healthy control group [100]. Treatment with ciprofloxacin suppressed bacterial overgrowth, increased serum insulin, and decreased endogenous ethanol production but did not influence serum acetylated ghrelin levels (half values compared to controls). Changes in fasting insulin and ethanol following ciprofloxacin suggested these parameters might be influenced by small intestinal bacterial activity. In an experimental rat model of NASH, there was a slower transit time and higher quantity of coliform bacteria (E. coli) present. Treatment with gentamicin (cidomycin) accelerated the transit time, decreased TNF-α levels, and alleviated severity of liver involvement in experimental animals [101]. Thus, SIBO might play an important role in the pathogenesis of NASH (Fig. 3). However, despite these findings, the impact of increased SIBO during NASH progression remains controversial. Sabate et al. reported that SIBO was not associated with the frequency of NASH in clinical study [95]. Thus, additional studies are required to clarify the involvement of SIBO in NASH pathogenesis.

Fig. 3
figure 3

Proposed mechanisms of intestinal microbiota-induced inflammation in the liver. (1) Small intestinal bacterial overgrowth and (2) increased intestinal permeability induce the translocation of bacteria and endotoxin. (3) Obesity-mediated increased leptin levels induce hyperreactivity against low-level bacteria and endotoxin, causing increased inflammation in steatosis. (4) Tissue damage and dying cells result in the release of damage-associated molecular patterns (DAMPs) that stimulate local intravascular sentinel cells, including Kupffer cells, to produce interleukin (IL)-1β, which induces upregulation of intercellular adhesion molecule-1 (ICAM-1) on sinusoidal endothelial cells (5). Neutrophils are then recruited via β2 integrin (Mac-1)-dependent adhesion to endothelial ICAM-1. (6) Finally, neutrophils are recruited to the liver

Intestinal permeability and endotoxin

The idea that increased intestinal permeability and intestinal flora might contribute to the development of several diseases was first suggested in 1890 [102]. The occurrence of cross-talk between the gut and the liver is an intriguing hypothesis that could explain the hepatobiliary changes associated with several inflammatory and infectious intestinal diseases such as inflammatory bowel disease, celiac disease, and infections caused by Salmonella and Yersinia amongst others [103]. Evidence supporting a role for the liver–gut axis in the pathogenesis of NASH has been slowly accumulating [104106].

Obesity may increase intestinal permeability via disorders of intestinal barrier integrity [106108]. A HFD may increase intestinal translocation of endotoxin in mice (the so-called metabolic endotoxemia) and reduce enteric Bifidobacteria [109, 110], a group of bacteria that may lower intestinal LPS levels to improve mucosal barrier function [107]. Mechanisms regulating intestinal barrier integrity may also modulate the extent of endotoxinemia [106108, 111]. Glucagon-like peptide (GLP)-2, a 33-amino-acid peptide with intestinotrophic functions, cosecreted with GLP-1 by enteroendocrine L cells, may be a key modulator of intestinal barrier function. Recently, Cani et al. assessed the effect of a prebiotic fermentable oligofructose on intestinal microbiota composition, intestinal permeability, and hepatic and systemic inflammation in obese ob/ob mice (Table 1) [112]. The prebiotic-supplemented diet increased the intestinal proportions of Lactobacilli and Bifidobacteria, increased the expression of epithelial tight-junction proteins occludin and zonulin-1 (restoring normal intestinal permeability), and reduced systemic endotoxinemia as well as hepatic inflammation and oxidative stress. These effects were associated with increased intestinal GLP-2 levels, were mimicked by the administration of a GLP-2 agonist, and were prevented by pretreatment with a GLP-2 antagonist. These findings suggested GLP-2 might link intestinal microbiota, intestinal permeability, and systemic endotoxinemia and inflammation.

An alternative pathway for LPS absorption from the gut may involve chylomicron secretion from enterocytes, rather than loss of intercellular tight-junction integrity. Studies using cell cultures or animal models suggested that endotoxin is actively secreted into the blood with chylomicrons, and that inhibition of chylomicron synthesis blocks blocked endotoxin secretion [113]. Collectively, these data strongly connect gut-derived endotoxin, via disorder of intestinal barrier integrity and increase in chylomicron secretion from enterocytes, to the pathogenesis of NASH (Fig. 3).

Hyperresponsivity to endotoxin

Previous studies showed that gut-derived bacterial endotoxin may play a role in the progression of disease from simple steatosis to steatohepatitis in NASH [97, 98, 100, 106108, 114116]. Despite these findings, the impact of increased endotoxinemia during NASH progression remains controversial. Namely, it is unclear whether serum endotoxin levels are significantly higher in NASH patients than in control subjects or patients presenting with simple steatosis. Harte et al. reported that serum endotoxin levels were elevated in NAFLD patients compared with control subjects. However, Loguercio et al. showed endotoxemia was absent in all NAFLD patients tested [115, 117]. At present, there is general agreement that mild portal endotoxemia can be detected in healthy subjects due to gut-derived bacterial endotoxin [118]. However, the levels of portal endotoxemia observed under healthy conditions do not usually cause liver dysfunction [119]. We therefore propose it may be necessary to consider a low-level endotoxin-mediated mechanism for the progression of NASH. Here, we hypothesize that responsiveness against gut-derived bacterial endotoxin might be enhanced under simple steatosis compared with that of the healthy liver. Indeed, our own data show that responsivity against low-dose LPS was enhanced in HFD-induced murine steatosis and that low-dose LPS led to liver injury and severe hepatic fibrosis in HFD-fed mice, but not in chow-fed mice (Table 1) [120]. Previous studies also showed that a high cholesterol diet increased the sensitivity of mice to LPS without affecting plasma levels, supporting our suggestion (Table 1) [121]. Next, we investigated the mechanisms for enhancing responsivity against low-dose LPS in HFD-induced murine steatosis compared with chow-fed murine healthy livers, and showed that CD14 is overexpressed in the liver of HFD-fed WT mice compared to chow-fed WT mice [120]. Our data clearly indicate that CD14-positive cells are specifically identified as Kupffer cells and that increased CD14-positive Kupffer cell numbers contributed to enhanced responsivity against low-dose LPS in simple steatosis. This suggests that hyperresponsivity against low-dose bacterial endotoxin is regulated by the expression of CD14 in Kupffer cells during simple steatosis [120]. CD14 is an important regulatory factor in LPS-induced inflammation and enhances the LPS effects in Kupffer cells [122127]. In addition, a previous report showed that promoter polymorphisms of CD14 are a risk factor for human NASH [128]. Therefore, increased expression of CD14 is closely related to the pathogenesis of NASH. Indeed, our data showed that CD14 mRNA expression levels were much higher in NAFLD patients, including NAFL and NASH, than those in control subjects [120]. Therefore, hepatic CD14 may be an important factor in the development of NASH by enhancing hepatic inflammation against gut-derived bacterial endotoxin. We also investigated leptin-dependent increases in hepatic CD14 expression using leptin-deficient ob/ob mice and leptin receptor-deficient db/db mice. We found that both leptin-deficiency and leptin receptor deficiency led to decreased hepatic CD14 expression, resulting in decreased responsivity to LPS, despite mice exhibiting severe obesity and steatosis. In contrast, HFD-fed WT mice exhibited severe obesity and steatosis with increased hepatic CD14 expression via increased serum leptin levels and activation of leptin receptor and signal transducer and activator of transcription 3 (STAT3) signaling in the liver, causing enhanced LPS responsivity. Furthermore, exogenously administered leptin in chow-fed WT mice caused overexpression of hepatic CD14 despite a lack of obesity and steatosis, resulting in enhancing LPS responsivity. Increased CD14 expression was also observed in RAW264.7 cells, a murine monocyte/macrophage cell line, and Kupffer cells isolated from chow fed WT mice treated with leptin in vitro. Furthermore, the increase in CD14 expression in RAW246.7 cells was decreased after STAT3 inhibitor administration. Thus, leptin and STAT3 signaling may increase responsivity to gut-derived, low-dose bacterial endotoxin even in the healthy liver via an increase in CD14-positive Kupffer cells, irrespective of steatosis.

In humans, increased serum leptin levels are generally associated with obesity, visceral fat accumulation, and steatosis [129, 130], and patients with NAFLD are generally often obese and with hyperleptinemia. We also observed that serum leptin and hepatic CD14 expression levels in NAFLD patients were much higher than in control subjects. Similarly, the levels in NASH patients were much higher than in patients with simple steatosis. Significant positive correlations between serum leptin levels and hepatic CD14 expression levels were observed [120]. These results clearly indicate that serum leptin levels and hepatic CD14 expression are closely associated with the onset of NAFLD and its progression to NASH. In general, simple steatosis is considered to have progressed to NASH when factors such as oxidative stress, proinflammatory cytokines, and gut-derived high-level endotoxin are present in addition to conditions that enhance responsivity to endotoxin via increased hepatic CD14 expression. Under these circumstances, hepatic TNF-α expression may play an important role in liver inflammation and fibrosis. We observed that hepatic TNF-α expression was greater in NASH than simple steatosis, but not between simple steatosis and control subjects [120]. Thus, leptin-induced hepatic CD14 expression may enhance hepatic responsivity to gut-derived bacterial endotoxins, even at low levels, resulting in the progression from simple steatosis to NASH (Fig. 3).

Oral microbiota and NASH

Microorganisms found in the human oral cavity have been referred to as the oral microflora, oral microbiota, or more recently as the oral microbiome. The oral cavity, or mouth, includes several distinct microbial habitats, such as teeth, gingival sulcus, attached gingiva, tongue, cheek, lip, hard palate, and soft palate. Contiguous with the oral cavity are the tonsils, pharynx, esophagus, Eustachian tube, middle ear, trachea, lungs, nasal passages, and sinuses. Studies demonstrated different oral structures and tissues are colonized by distinct microbial communities [131]. Approximately 280 bacterial species from the oral cavity have been isolated in culture and formally named. It was estimated that less than half of bacterial species present in the oral cavity can be cultivated using anaerobic microbiological methods and that there are likely 500 to 700 common oral species [132]. Cultivation-independent molecular methods, primarily using 16S rRNA gene-based cloning studies, have validated these estimates by identifying approximately 600 species or phylotypes [132]. Microorganisms colonizing one area of the oral cavity have a significant probability of spreading along contiguous epithelial surfaces to neighboring sites. Microorganisms from the oral cavity can cause a number of oral infectious diseases, including periodontitis (gum disease), caries (tooth decay), endodontic (root canal) infections, alveolar osteitis (dry socket), and tonsillitis. Furthermore, accumulating evidence has linked oral bacteria to a number of systemic diseases [133], including DM [134], cardiovascular disease [135, 136], stroke [137], intracranial hemorrhage [138], preterm birth [139], ulcerative colitis [140], and pneumonia [141]. However, the relationship between oral microbiota and pathogenesis of NASH has not been demonstrated.

Periodontitis and insulin resistance

Periodontal disease is the most common infectious disease affecting tooth-supporting structures. Although a relationship has been reported between infection with periodontal bacteria and the onset of type 2 DM, it is not well recognized in the medical community [142, 143]. In dentistry, understanding the changes in oral microbiota that foretell the early stages of periodontitis and dental caries, prevalent chronic oral diseases, may allow better diagnosis and treatment before the appearance of the characteristic disease manifestations (such as tissue damage in periodontal pockets or dental hard tissue loss). The emergence and evolution of antibiotic resistance in periodontal pathogens has affected the therapeutic success rates for this disease [144, 145]. New approaches are urgently required to control periodontal disease, and microbiome studies offer a promising new angle of attack.

Periodontal inflammation often leads to superficial ulcers on the gingival sulcus, where blood capillaries are exposed to microbial biofilms [146]. Periodontal pathogens are translocated and released from the sulcus into the bloodstream, and clinical trials have demonstrated such transient bacteremia occurring after preventive dental procedures and periodontal therapy, including tooth brushing, chewing, subgingival irrigation, periodontal treatment, and dental extractions, at reported frequencies ranging from 17 to 100 % in infected individuals [147149]. Porphyromonas gingivalis is a major causative agent of periodontitis [150152]. Recent reports suggest infection with P. gingivalis is associated with several systemic diseases, including cardiovascular diseases, preterm low birth weight, rheumatoid arthritis, and DM [150152]. Moreover, increased serum levels of LPS and TNF-α associated with P. gingivalis infection can induce insulin resistance leading to the development of type 2 DM (Fig. 4) [153]. In addition, our colleagues reported a relationship between the fimbrial type of the periodontal bacteria causing periodontitis and the risk of type 2 DM development [154]. Taken together, periodontitis may lead to the development of NASH pathogenesis via exacerbation of insulin resistance.

Fig. 4
figure 4

Proposed mechanisms of periodontitis-induced progression of nonalcoholic steatohepatitis (NASH). (Direct effect) Periodontal diseases also involve activation of the broad axis of innate immunity through upregulation of pro-inflammatory cytokines from macrophages including Kupffer cells, resulting in liver inflammation. (Indirect effect) Increased pro-inflammatory cytokines released into the systemic circulation affect adipose tissue, leading to insulin resistance, which results in the progression of NASH pathogenesis

Periodontitis and NASH

Yoneda et al. demonstrated that the prevalence of P. gingivalis infection was significantly higher in NAFLD patients than in healthy subjects. Multiple regression analysis in NAFLD patients and control subjects revealed a significantly higher prevalence of P. gingivalis infection in NAFLD patients compared with control subjects, even after adjusting for age, history of DM, and body mass index (BMI). This suggested P. gingivalis infection might be involved in the independent risk of onset of NAFLD, because P. gingivalis itself or endotoxin and cytokines released from the bacteria can easily enter the blood circulation (Table 1) [155]. These results may indicate that both DM and P. gingivalis infection may be involved in the progression of simple steatosis to NASH through direct and/or indirect effects (Fig. 4).

Sterile inflammation and NASH

In addition to infectious insults, sterile tissue injury and cell necrosis can induce profound neutrophilic inflammation. In these conditions, neutrophils do not function as antimicrobial effectors but clear debris and initiate wound healing [156]. Indeed, neutrophils are emerging as central orchestrators of resolution and restitution following tissue injury and may even contribute to the avoidance of autoimmunity following sterile inflammation (SI) [157, 158]. However, excessive or prolonged neutrophil infiltration can exacerbate tissue injury, leading to disease.

The liver is a target of pathogens that induce rapid inflammatory responses via PAMPs [159]. However, less intuitive is that injury in the absence of pathogens can result in inflammation. SI, however, occurs in all tissues after injury of various etiologies. In the liver, SI is particularly important because it is a major component of the pathology of a wide range of diseases, such as alcoholic steatohepatitis (ASH), NASH, drug-induced liver injury, and ischemia/reperfusion [160162]. Although SI is a well-characterized phenomenon, the mechanisms that initiate and regulate innate immune responses and cell death have only recently begun to be described [163, 164]. The theoretical basis for this originated in the inability of self-recognition and non-self-recognition theories to explain the selectivity of the adaptive immune system. This led to the proposal that danger is used as a criterion for immune activation, and molecules termed “damage-associated molecular patterns” (DAMPs) released from damaged cells are a molecular trigger for inflammation [165]. Liver-resident macrophages (Kupffer cells) and dendritic cells respond to DAMPs by creating an inflammatory milieu that incites the influx and/or activation of T cells, monocytes, and neutrophils, the chief effectors of liver injury (Fig. 3) [166, 167]. This was experimentally validated and explained many aspects of the sterile inflammatory response [166]. Recent advances have identified DAMPs, their receptors, and the cellular machinery required for processing this into a full inflammatory response. Some general concepts have emerged. Firstly, SI is a bona fide inflammatory response, with features including redness, swelling, heat, neutrophilic infiltrate, cytokine production, and tissue damage (Fig. 3). SI is induced in minutes using innate immune pathways but is also present chronically, for example in NASH and ASH. Conceptually, pathogen-driven inflammation and SI are distinct, but functionally, there are many areas of overlap. Many receptors and pathways initially identified as activated by PAMPs are also activated by DAMPs [168, 169]. One example is TLR, which can be activated by LPS as well as cellular high-mobility group protein B1 (HMGB1), a ubiquitously expressed protein that normally binds to the minor groove of nuclear DNA to control the docking and activity of transcription factors by structurally modifying the DNA double helix [170]. By this route, HMGB-1 can increase the activity of cell cycle regulator p53 while negatively affecting transcription of the pro-inflammatory transcription factor NF-κB [170]. During liver injury, HMGB-1 is predominantly released by excessively ROS-generating hepatocytes, which actively secrete HMGB-1 into the circulation without dying [171]. Additionally, in vitro studies have shown that HMGB-1 can be actively released during apoptosis [172] and passively leaks from necrotic cells [173]. Thus, HMGB-1 acts as a proximal danger signal. Due to the unique vascular supply of the liver, PAMPs of intestinal origin and DAMPs of hepatocyte origin collectively contribute to inflammation in a number of diseases. There is also direct interaction with PAMPs, such as LPS stimulating the release of HMGB1 [174].

Probiotics, prebiotics, antibiotics, and periodontitis treatment for NASH

The current standard of care for treating patients with NAFLD and/or NASH focuses on lifestyle interventions, particularly diet and exercise. There is a lack of consensus regarding the most effective and appropriate pharmacologic therapy for NASH, especially without DM, dyslipidemia, hypertension, or obesity. Alterations in intestinal or oral flora and responses to endotoxin may contribute to liver inflammation and the development of NASH, suggesting that reduction of microbiota and endotoxin may improve NASH. Next, we will review the current options of probiotics, prebiotics, antibiotics, and periodontal treatment available for treating NASH.

Probiotics and prebiotics

Probiotics are live microbes that modulate the intestinal microflora and enhance body health. The most common probiotics in the market are Lactobacilli, Streptococci, and Bifidobacteria. Prebiotics are indigestible carbohydrates that stimulate the growth and activity of beneficial bacteria, particularly Lactobacilli and Bifidobacteria. Some examples of prebiotics are lactulose, which increases the number of Bifidobacteria, and fructo-oligosaccharides such as oligofructose and inulin, Lactobacillus rhamnosus G, and Bifidobacteria lactis Bb12 [175].

Intestinal microbiota manipulation with probiotics in rodents with fatty liver reduced intestinal inflammation and improved epithelial barrier function [109, 176]. Therefore, probiotics could represent a new treatment for human NAFLD. Loguercio et al. showed that probiotics might reduce NAFLD liver injury and improve liver function tests (Table 2) [177, 178]. However, the Cochrane review [179] and a subsequent pediatric meta-analysis [180] highlighted that probiotic treatment in patients with NAFLD and nonalcoholic steatohepatitis could not be recommended because of the lack of randomized clinical trials. More recently, a double-blind randomized controlled trial (RCT) showed that treatment with 500 million Lactobacillus bulgaricus and Streptococcus thermophiles/day in adults with biopsy-proven NAFLD caused a reduction of liver transaminase levels (Table 2) [181]. In children, another double-blind RCT study showed that obese NAFLD patients (mean age 10.7 years) treated with Lactobacillus GG (12 billion CFU/day for 8 weeks), irrespective of changes in BMI z score and visceral fat, demonstrated significant decreases (up to normalization in 80 % of cases) in ALT and in antipeptidoglycan-polysaccharide antibodies (a SIBO marker) (Table 2) [182]. Serum TNF-α levels and bright liver parameters remained stable [182]. Additionally, a study in obese mice demonstrated that treatment with the probiotic VSL#3 resulted in improved liver histology and decreased ALT levels (Table 2) [177].

Table 2 Probiotics and periodontitis treatment studies in patients with NAFLD

Prebiotics are nondigestible food ingredients that beneficially affect the host by selectively stimulating growth and/or modifying metabolic activity of selected intestinal bacteria [183]. The health effects of prebiotics are likely conveyed to the host via two chief mechanisms that include improved glucoregulation and modified lipid metabolism as well as selective modulation of intestinal microbiota [183]. Studies in obese rats demonstrated that prebiotic fibers improved or normalized intestinal microbiota dysbiosis by increasing Firmicutes and decreasing Bacteroidetes phylae [184].

These promising preliminary results are strongly indicative of the great potential of probiotics and prebiotics for the prevention or treatment of NASH. However, as recently stated in a meta-analysis, further clinical studies are necessary to better define this innovative strategy [180]. The large amount of experimental data currently available demonstrating beneficial effects of probiotics and prebiotics will likely drive the design of future clinical trials.

Antibiotics

Prophylactic use of antibiotics in patients with chronic liver disease is an established method of preventing infections or innate immune dysfunction in acute liver failure (ALF) and has been reviewed by Antoniades et al. [185]. The systemic inflammatory response is crucial for the development of multiorgan dysfunction and mortality in NAFLD. A cascade of events starts with a local inflammatory response in the liver, followed by systemic inflammatory and compensatory anti-inflammatory responses, ultimately leading to an unbalanced immune response with refractory multiorgan failure. Monocytes and macrophages play a central role within that cascade. After activation by bacterial products, pro- and anti-inflammatory cytokines are produced in large quantities. Monocytes also present antigens to T cells by human leukocyte antigen (HLA)-DR, which triggers an adaptive immune response. After prolonged exposure to endotoxins, monocytes develop an endotoxin-tolerant phenotype with decreased HLA-DR expression and an anti-inflammatory cytokine expression profile, consisting mainly of IL-10 [186]. Defects in complement, neutrophils, and Kupffer cells are also observed in ALF. Patients exhibit qualitative and quantitative changes in both the classic and alternative complement pathways. Their neutrophils show impaired oxidative burst, phagocytosis, and bacterial killing capacity, and Kupffer cells showed impaired clearance function [185].

Lichtman et al. showed that antibiotics (metronidazole and tetracycline) reduced hepatic injury in rats with surgically induced intestinal bacterial overgrowth [187]. Drenick et al. and Vanderhoof et al. also showed that antibiotics prevented and reversed hepatic steatosis and liver injury after intestinal bypass for patients with morbid obesity [188, 189]. Additionally, Bergheim et al. showed that antibiotics could reduce hepatic steatosis and endotoxinemia in a fructose-induced rodent NAFLD model [190]. These findings imply a critical role for small bowel flora, suggesting that intestinal bacterial overgrowth treatment might reduce ethanol and LPS levels. However, direct evidence is currently lacking, and thus, antibiotics cannot be routinely recommended to treat NASH at this time, although further study is indicated.

Periodontitis treatment

Periodontitis is the most common localized dental inflammatory disease related to several pathological conditions such as inflammation of gums (gingivitis), degeneration of periodontal ligaments, and dental cementum and alveolar bone loss. Various approaches have been used for treatment including surgical intervention, mechanical therapy, and use of pharmacological agents [193196]. Medications are specifically used to better manage periodontitis and include antimicrobials that change microbial flora in periodontal milieu, host response modulating agents that modify the reduction of excessive enzyme levels, cytokines, prostaglandins, and osteoclast activity [194].

Yoneda et al. confirmed the efficacy of periodontal treatments in improving liver function parameters such as serum aspartate aminotransferase and ALT in NAFLD patients [156]. This study indicated that periodontitis caused by P. gingivalis in NAFLD patients may be a risk factor for the aggravation of NAFLD and that periodontal treatment may be a useful supportive measure in the management of patients with NAFLD. Further large-scale clinical practice for the periodontal treatments in NAFLD patients will be required in the future (Table 2).

Conclusion

In conclusion, the microbiome, including intestinal or oral microbiota, may influence body composition and liver inflammation, contributing to the development of NASH pathogenesis. NASH subjects not only contain a specific microbiota that harvests energy from the diet more effectively, but also have hyperresponsivity to endotoxin in the liver. Additionally, microbiota-induced cell death in the liver may induce SI via the effect of DAMPs, which are critical instigators of sterile inflammation, eliciting an intricate cytokine and chemokine network that, in terms of the involved cell types, is more elaborate than previously assumed.

Obesity and DM have been closely correlated with increased risk of several malignancies, specifically hepatocellular carcinoma (HCC) [195196]. In addition, NASH is a well-recognized cause of not only cirrhosis but also HCC [197]. The influence of NASH and its causative factors on the development of HCC has drawn attention in recent years. It was reported that the development of HCC in patients with features of metabolic syndrome has distinct characteristics and occurs, in most cases, in the absence of significant liver fibrosis when compared with those with other liver diseases including hepatitis C virus infection [198]. Development of HCC in the absence of significant liver fibrosis has also been observed in etiologies other than metabolic syndrome, especially in patients with hepatitis B virus (HBV) infection [199]. HBV has direct oncogenic potential secondary to integration of HBV DNA into the genome of hepatocytes [200]. In line with this observation, it is suggested that metabolic syndrome could also have direct oncogenic effects, such as insulin, lipid peroxidation, oxidative stress, and gut-derived bacterial endotoxin. Recently, obesity-induced gut microbial metabolite was reported to be associated with development of HCC [201]. These findings suggest that the microbiome may contribute to obesity-associated HCC through the direct oncogenic effects. The mechanisms underlying obesity-associated HCC is urgently needed to be elucidated.

Future research should focus on adequately powered, randomized, placebo-controlled, human clinical trials to determine the effect of probiotics, prebiotics, antibiotics, and periodontitis treatment on composite histological markers of NASH. Finally, further research into potential mechanisms is required to simultaneously assess the effects of these treatments on the microbial composition and function in patients with NASH, as well as host energy balance, inflammatory cytokines, and regulators of metabolism.