Emerging evidence shows that an intestinal barrier disorder leads to the translocation of the GM and metabolites, which can reach the liver directly along the gut–liver axis. Patients with NAFLD have decreased expression of junctional adhesion molecule A and zonula occludens-1, and increased intestinal permeability, which might be important factors in disease progression (Kolodziejczyk et al., 2019). Compared with simple hepatic steatosis, the correlation of increased intestinal permeability was stronger in patients with NASH, suggesting that inflammatory persistence and exacerbation might be caused by destruction of the intestinal barrier (Luther et al., 2015). In a mouse model, compared with the control group, the zonula occludens-1-deficient mouse group (intestinal barrier disorder model) fed a high fat, fructose, and cholesterol diet, developed more severe NASH (Rahman et al., 2016). Similarly, destruction of the intestinal barrier is considered an early event in the pathogenesis of NAFLD. Mice fed an HFD suffered from diet-induced intestinal dysbacteriosis after only 1week, resulting in intestinal barrier damage and bacterial translocation to the liver (De Santis et al., 2015; Mouries et al., 2019). Thus, an impaired intestinal barrier secondary to dysbacteriosis might be a prerequisite for diet-driven NAFLD.

It is unclear whether NAFLD is the cause or the result of intestinal barrier disruption. However, it can be confirmed that no matter what the triggering event is, the transfer of pro-inflammatory products, such as LPS, secondary to intestinal barrier damage, to the systemic circulation, will aggravate NAFLD and lead to poor prognosis (Gabele et al., 2011; Ipsen et al., 2018; Mouries et al., 2019). As shown in Figure 1B, excessive LPS translocation to the liver can interact with toll-like receptor (TLR)-4 on Kupffer cells and stellate cells, thus activating the nuclear factor kappa B (NF-κB) signaling pathway, and ultimately promoting the release of pro-inflammatory cytokines, such as interleukin (IL)-1, IL-6, and tumor necrosis factor, aggravating hepatic steatosis, inflammation, and fibrosis (Henao-Mejia et al., 2012; Plociennikowska et al., 2015; Carpino et al., 2020). TLR signaling in the mucosa also leads to the production of NOD-like receptor family, pyrin domain containing 3 (NLRP3), which results in the production of hepatic pro-inflammatory and pro-fibrotic mediators (e.g., caspase-1, IL-1β, and IL-18; Mridha et al., 2017; Carpino et al., 2020).

Human SCFAs (e.g., acetate, propionate, and butyrate) are generated mainly from the fermentation of polysaccharides by the GM, and play a pivotal role in energy metabolism and inflammation regulation (Gomes et al., 2018). The different phenotypes of the GM and different dietary factors will affect the type and quantity of SCFAs synthesized in the gut. A high-fiber or resistant starch diet, the Mediterranean diet, and the enrichment of specific bacteria, such as Akkermansia municiphilla (producing propionate), Ruminococcus, Faecalibacterium, and Eubacterium (producing butyrate), can induce SCFA production (Morrison and Preston, 2016; Gomes et al., 2018).

Many polysaccharides cannot be hydrolyzed by the host, but can be realized by specific microbiota, finally generating SCFAs. If the excess SCFAs are not metabolized by colon cells, they will enter the liver and peripheral circulation through the portal vein, where they can be used as the substrates for fat synthesis and glycogenesis (Rau et al., 2018). This enables the host to obtain excess energy from food more efficiently, and to synthesize and store more fat to the liver (Harris et al., 2020). In a cohort study, with the development of NAFLD, higher abundances of SCFA-producing bacteria and intestinal acetate and propionate levels were observed (Rau et al., 2018). Interestingly, elevated peripheral levels of pro-inflammatory T cells (lower numbers of resting regulatory T-cells and higher numbers of Th17 cells) were observed simultaneously, which suggested that SCFAs are involved in the development of NAFLD, not only by affecting metabolism, but also by influencing immune and inflammatory responses (Rau et al., 2018). NAFLD is most associated with obesity. In mouse models and human studies, obese subjects have more carbohydrate metabolism genes in the intestinal microbiome and a higher concentration of SCFAs in the cecum, indicating that their production is excessive or their absorption is disrupted (Schwiertz et al., 2010). Zhao et al. (2020) indicated that liver lipid synthesis triggered by dietary fructose is dependent on metabolizing fructose to acetic acid and then to Acetyl Coenzyme A through the GM rather than via ATP citrate lyase. The hepatic metabolism of fructose promotes the transcription of genes related to hepatic lipid synthesis, and the metabolite acetate provides the raw material for this process. In addition, SCFAs can induce differentiation of T-cells into Th1 or Th17 cells, depending on the cytokine milieu and the epigenetic activity of histone deacetylases. IL-17 secreted by Th17 cells might play a pro-HCC role by promoting tumor angiogenesis (Liao et al., 2013; Park et al., 2015).

However, emerging evidence suggests the potential protective effect of SCFAs in NAFLD. Sodium butyrate can alleviate HFD-induced intestinal dysbacteriosis and endotoxemia, and thus inhibit NAFLD, by regulating intestinal and liver immune responses (Zhou et al., 2017a). One of the mechanisms is to affect the nutrient intake and metabolism of the host. Butyrate and propionate can activate free fatty acid receptor-3, and thus upregulate the production of the hormones intestine peptide YY and glucagon-like peptide (GLP)-1, which can increase satiety and reduce the intake of energy (Lin et al., 2012). Similarly, activation of GLP-1 has been proven to contribute to the recovery of hepatocyte function, the inhibition of hepatic steatosis and fibrosis, and the prevention of NAFLD developing into NASH (Tang et al., 2015a). SCFAs can also inhibit insulin signal transduction in adipocytes by activating G-protein receptor-43, thereby promoting glucose and unbound lipid metabolism, and inhibiting fat accumulation in liver and adipose tissue (Kimura et al., 2013). Another possible mechanism of SCFAs limiting NAFLD is to maintain a healthy intestinal barrier and to weaken inflammatory signals. SCFAs can prevent intestinal mucosal atrophy mediated by GLP-2 (Cani et al., 2009). Increasing the level of GLP-2 through microbial intervention can reduce the intestinal permeability and the levels of LPS and cytokines, thus reducing oxidative stress and liver inflammation (Cani et al., 2009). SCFA supplementation also showed beneficial effects in several inflammatory diseases (e.g., colitis; Rau et al., 2018). By inhibiting colitis, the intestinal barrier can be improved, thereby reducing the liver damage caused by bacterial translocation and the liver transfer of metabolites.

The effects of SCFAs are diverse and extensive, and different kinds and contents of SCFAs in different hosts show different and even contradictory biological effects; therefore, it is difficult to clarify their overall impact (Harris et al., 2020; Martin-Gallausiaux et al., 2020). In view of the close and complex relationship between SCFAs and host nutrient intake and metabolism, inflammation, and immunity, an in-depth study is needed to determine the specific mechanism by which SCFAs affect the occurrence and development of NAFLD.

Lipotoxicity promotes the progression of NASH, fibrosis, cirrhosis, and even HCC (Ioannou, 2016). Among liver lipids, cholesterol is the most important lipotoxic molecule in the development of NAFLD (Ioannou, 2016). Abnormal liver cholesterol homeostasis has been confirmed in both animal models and in humans with NASH. Zhang et al. (2021) revealed the GM-mediated mechanism of dietary cholesterol leading to the progression of NASH, that is, long-term high dietary cholesterol can induce an increase in taurocholic acid and the decrease of 3-indolepropionic acid by changing the GM (decreased levels of Bifidobacterium and Bacteroides and increase levels of Desulfovibrionaceae Anaerotruncus, Desulfovibrio, and Mucispirillum), thus promoting liver lipid accumulation and cell proliferation, leading to the occurrence of NAFLD-HCC (Zhang et al., 2021). In a mouse model, anti-cholesterol treatment eliminated completely the onset of NAFLD-HCC induced by dietary cholesterol (Zhang et al., 2021). This suggests that some of the mechanisms remain unknown: how the related pathogenesis inducing factor (e.g., HFD) causes inflammation, and how to accelerate the transformation of simple hepatic steatosis to NASH, which might be explained by the GM and its metabolites.

BAs are synthesized from cholesterol in the liver and play an important role in the digestion, absorption, and metabolism of fat. The GM is involved in the transformation and metabolism of BAs (Sanchez, 2018). The interaction between BAs and the GM plays an important role in the pathogenesis of NAFLD (Chiang and Ferrell, 2020). BAs participate in the pathogenesis of NAFLD through the farnesoid X receptor (FXR). By binding to FXR, BAs increase insulin sensitivity and reduce hepatic gluconeogenesis and triglyceride in the circulation (Chiang and Ferrell, 2020). Under the intervention of an HFD, the GM promotes weight gain and liver steatosis in an FXR-dependent manner, and the improvement of hepatic steatosis associated with antibiotic therapy depends on FXR signal transduction (Jiang et al., 2015; Parseus et al., 2017). In a large cohort of patients with NASH, although Obecholate (an FXR agonist) did not improve NASH, it significantly improved liver fibrosis compared with that in the control group (Neuschwander-Tetri et al., 2015). BAs also activate Takeda G-protein-coupled receptor 5 (TGR5) in muscle and adipose tissue, thereby increasing energy expenditure (Pols et al., 2011). In addition, activation of TGR5 in the intestine can promote the release of GLP-1, and then positively regulate the secretion of insulin (Pols et al., 2011; Kumar et al., 2016). TGR5 is also expressed in Kupffer cells, which are involved in the regulation of liver inflammation. Activation of TGR5 seems to induce anti-inflammatory effects by inhibiting the NF-κB signaling pathway and cytokine production (Perino and Schoonjans, 2015). The BA levels in liver, serum, and urine were increased in patients with NAFLD (Arab et al., 2017). In a phase II clinical trial in patients with NASH, the BA synthesis inhibitor, Aldafermin, reduced liver inflammation, steatosis, and fibrosis significantly (Harrison et al., 2021). The GM is likely to affect the BA pool, and regulates the metabolism of host cells through the transformation of BAs, including the homeostasis of lipids and glucose; however, its role in the pathogenesis of NAFLD remains controversial, which requires further in-depth study.

Choline deficiency is closely related to the induction and promotion of NAFLD, and is often used to construct animal models of NAFLD (Sherriff et al., 2016). Compared with those in the healthy group, patients with NAFLD generally showed lower levels of serum choline and higher levels of trimethylamine (TMA; Sherriff et al., 2016). In the absence of choline in human body (e.g., because of a choline deficient diet or gut dysbiosis), the synthesis of phosphatidylcholine is insufficient and the level of very-low-density lipoprotein is downregulated, which leads to liver lipid transfer disorder, enhanced mitochondrial β-oxidative damage, and oxidative stress in hepatocytes, eventually leading to liver steatosis, and aggravating liver inflammation and fibrosis (Smallwood et al., 2016). Some intestinal bacteria (e.g., Desulfovibrio desulfuricans and Escherichia coli) can convert choline to TMA and then to trimethylamine-N-oxide (TMAO) in the liver, which reduces the bioavailability of choline (Sohlenkamp et al., 2003). In addition, TMAO can promote insulin resistance by destroying blood glucose homeostasis and increasing the level of serum inflammatory cytokine C-C motif chemokine ligand 2 (CCL2), and affect lipid metabolism and BA homeostasis by reducing the conversion of cholesterol to BAs, which suggests that TMAO might affect NAFLD indirectly (Tang et al., 2015b). In fact, strategies to reduce TMA and/or TMAO have been used in the clinical treatment or prevention of NAFLD. For example, 3,3-dimethyl-1-butanol, a structural analog of choline, inhibits TMA and TMAO production by inhibiting microbial TMA lyase. However, based on the different microbial characteristics of individuals, it might only be effective for some patients (Wang et al., 2015). Therefore, additional genotyping of the NAFLD cohort is needed to identify patients that would respond to TMA and/or TMAO inhibitors.

Endogenous ethanol is produced by some intestinal bacteria via carbohydrate fermentation. Although obese mice with NAFLD did not ingest any alcohol, alcohol could still be detected in their breath (Cope et al., 2000). Compared with healthy individuals or patients with simple hepatic steatosis, the blood ethanol concentration in patients with NASH is higher, which is associated with increased liver inflammation and liver damage (Baker et al., 2010). Recent studies revealed that about 60% of patients with NAFLD have high alcohol producing K. pneumoniae in their intestines and their abundance is related to the severity of the disease (Yuan et al., 2019). The pathogenesis of NAFLD caused by endogenous ethanol is similar to that of alcoholic fatty liver disease (Parlesak et al., 2000; Seitz et al., 2018; Jennison and Byrne, 2021): (1) the induction of mitochondrial damage and enhanced oxidative stress; (2) destruction of the intestinal barrier and aggravation of liver damage through the gut–liver axis; (3) the induction of cytokines, chemokines, Th17, and other immune cells to intensify liver inflammation; and (4) the induction of liver cell damage through acetaldehyde-mediated cytotoxicity, metabolic disorder, and fat accumulation. These findings not only explain many of the similarities of the pathological features between the two diseases, but also provide a feasible method for clinical diagnosis and treatment of fatty liver caused by such bacteria (Brown and Kleiner, 2016).

Phenylacetic acid (PA) is mainly produced by the metabolism of aromatic amino acids (e.g., phenylalanine) by Bacteroides (Cook, 2019). A multi-omics study showed that PA levels were high in the serum of patients with NASH. At the same time, aromatic amino acids and branched chain amino acids increased, and bacterial diversity decreased (Hoyles et al., 2018). These characteristics have great potential as biomarkers for the clinical diagnosis and prediction of this disease.

Indole, one of the products of tryptophan metabolism by the GM, is generally considered to have anti-inflammatory effects (Yang et al., 2020). Clinical sample analysis, and mouse and cell experiments, showed that indole correlated negatively with NAFLD. Indole supplementation could reduce diet-induced NAFLD, liver fat accumulation, and the inflammatory response in mice. This protective effect was mediated by 6-phosphofructo-2-kinase/fructose-2,6-biphosphatase 3 (PFKFB3), a glycolysis regulatory factor of bone marrow cells (Ma et al., 2020). Mimicking or specifically activating PFKFB3 expression in macrophages using indole might be a feasible method to prevent and treat NAFLD and other inflammatory related diseases.