Among its many functions, the liver detoxifies and metabolizes components in blood arriving from the hepatic portal vein and hepatic artery, stores glycogen, secretes bile to aid digestion, and produces cholesterol as well as major plasma proteins such as albumin and fibronectin. These synthetic and secretory capacities make the liver the largest gland in the human body (1). Anatomically, the liver is in the upper right abdomen, beneath the right hemidiaphragm, and it is protected by the rib cage. It is separated by visible fissures, the most prominent of which is the umbilical fissure, which is lined by the falciform ligament. Such divisions allow identification of different lobes, for example, the large right lobe and a smaller left lobe (2, 3). The right lobe is further divided into a quadrate lobe and a caudate lobe; these are functional sections where the gallbladder and the inferior vena cava reside (4, 5) (Figure 1A).

To carry out its specialized functions, the liver is made up of a variety of cell types organized in lobular structures. One lobule is roughly hexagonal in shape and formed around the central vein with portal triads that demarcate the six corners of each lobule containing bile ducts, sinusoids, branches of the hepatic portal vein, and the hepatic artery (Figure 1B). Within its lobular structure are several cell types that cooperate to carry out physiological functions and maintain homeostasis in the tissue microenvironment. The different cell types can be generally divided into parenchymal cells, hepatocytes, and non-parenchymal cells which include hepatic stellate cells (HSC), Kupffer cells (KC), and liver sinusoidal endothelial cells (LSEC) (6) (Figure 1C). These cells carry out essential functions in the liver (Table 1) and are crucial in responding to injury (7).

Focusing on the immune compartment, KCs are the main immune cells that populate the liver in steady state. They extensively interact with hepatocytes, HSCs, and LSECs. They can stimulate leukocyte chemotaxis and adherence and produce cytokines that influence their activation (10, 11). KCs phagocytose foreign particles and microorganisms in portal circulation coming from the gut (12). These cells are the first line to defense against pathogens, they remove abnormal cells and cellular debris, participate in recycling erythrocytes, and their signaling can dictate inflammation and the immune response (13). KCs also contribute to liver disease when they are dysfunctional (14), this is discussed in more detail below.

Non-alcoholic fatty liver disease (NAFLD) refers to a spectrum of conditions affecting the liver. These range from benign steatosis to steatohepatitis (NASH), fibrosis, and cirrhosis. Importantly, NASH and fibrosis often occur simultaneously and are the last reversible steps of the condition. At the stage of cirrhosis, liver function is impaired, and patients are at high risk of developing hepatocellular carcinoma (HCC). Liver transplantation is the only possible intervention strategy at or beyond the stage of cirrhosis. Many individuals have the early stages of NAFLD, benign ectopic lipid storage in the form of steatosis, without any signs or symptoms; this is because the presence of fat in the form of simple steatosis does not cause damage to the liver (15–17) (Figure 2A). NAFLD and its progressive form NASH have similar risk factors, including overweight or obesity, insulin resistance, high levels of fats, particularly triglycerides, and hyperglycemia, indicating prediabetes or type 2 diabetes (T2D) (17).

The excessive increase of free fatty acids and triglycerides increases lipid oxidation and the production of reactive oxygen species (ROS) and lipotoxicity. This causes cellular damage and the secretion of cytokines, triggering KC activation and the recruitment of monocytes and other immune cells from the circulation. Ensuing hepatocyte apoptosis and HSC activation results in the deposition of extracellular matrix, which can lead to fibrosis if excessive (Figure 2B). At this stage, there is obvious persistent scar tissue in the liver and in the blood vessels around the liver. Nonetheless, the liver may still work well, and treating the cause of the inflammation may inhibit further development or even partially reverse the damage (18). However, when the replacement of normal liver tissue by fibrous septa exceeds a certain point, this develops into cirrhosis, the late stage of NAFLD. At this stage, there is extensive scarring and loss of function. Symptoms will include variceal bleeding, ascites, fatigue, and jaundice (19, 20). As a result, the liver stops functioning, urgent medical care and liver transplant surgery may be required.

Insulin resistance can in part be attributed to abnormal hepatic insulin processing; this would otherwise physiologically regulate glucose and lipid metabolism. The liver produces glucose using glycogen; this happens with the use of glucogenic precursors in the presence of a high glucagon-to-insulin ratio during a fasting period (28). During the fed state, reductions in glucagon and raised insulin levels signal the liver to raise glucose uptake, cease glucose production, and store the remaining nutrients as glycogen and lipids (29). In pathological conditions of insulin resistance, insulin fails to regulate liver metabolism, resulting in excessive glucose production alongside increased lipid synthesis (30); consequently, NAFLD is associated with insulin-resistance (31).

In obese individuals with T2D and NAFLD, dyslipidemia and hyperinsulinemia are more severe than in individuals without NAFLD (32). Insulin resistance develops in overweight or obese individuals when insulin action on its target tissues is compromised. Excess fatty acids, due to lifestyle factors as well as lipogenesis, accumulate in peripheral tissues such as the liver and AT, which play key roles in lipid storage, synthesis, and metabolism (33). Whilst independently affected by insulin resistance, significant crosstalk also occurs between the liver and AT, this crosstalk has been reported to influence the liver’s immune compartment. Indeed, it has been demonstrated that transplanting donor visceral AT (vAT) from obese mice increased high cholesterol diet (HCD)-induced liver macrophage content, worsening liver damage, compared to mice receiving transplants from lean donors. Adipose tissue macrophage (ATM) depletion prior to vAT transplantation abrogated this effect. On normal chow diets, vAT transplantation increased circulating and hepatic neutrophil numbers in obese-transplanted mice, similarly ATM depletion prior to vAT transplantation reversed this effect (34). Microarray analysis of sorted CD11c+ and CD11c− macrophages isolated from donor adipose tissue showed that obesity increased expression of genes involved in chemotaxis from CD11c+ ATMs. CD11c+ ATMs are known to increase in proportion in obese and insulin resistant conditions (34). These findings indicate that secreted molecules from ATMs act in an endocrine manner and can affect the liver’s immune compartment, influencing susceptibility to, and progression of, NAFLD.

AT also secretes adipokines, a special class of cytokine-like messenger molecules, such as leptin and adiponectin (35). Adiponectin regulates lipid accumulation both in the liver and in AT by inhibiting fatty acid oxidation (36). Moreover, it controls glucose homeostasis through regulating hepatic insulin sensitivity (37). Individuals with NAFLD have lower serum adiponectin levels than the healthy population (38). Hypoadiponectinemia promotes chronic inflammation of the liver that results from dysregulation of fatty acid metabolism associated with insulin resistance (39). Therefore, maintaining the adiponectin level in individuals with NAFLD may prevent progression to NASH. In contrast, leptin levels have been reported to be higher in individuals with NAFLD, and higher levels of circulating leptin are associated with increased disease severity (40). However, in vivo studies do not support a causal association since ob/ob mice that lack leptin and develop hyperphagic obesity have severe steatosis (41). Administration of leptin reverses the obese phenotype and improves steatosis. Hence, leptin may play multiple roles in NAFLD (42).

KCs are the dominant immune population in the liver; they are bona fide resident macrophages that stem from the yolk sac and thus have an embryonic origin (43). KCs are self-renewing and are phenotypically distinct from monocyte-derived macrophages that can infiltrate the liver during disease (44). Monocyte-derived macrophages differentiate in situ from circulating monocytes, and thus they originate from bone marrow hematopoiesis (45). Under healthy conditions, rodent models have between 20 and 40 macrophages for every 100 hepatocytes in the liver, making it the organ with the highest proportion of tissue macrophages compared to other tissues. This supports a functional role in homeostasis and normal physiology of the liver (46). They perform a number of tasks, including clearance of metabolic waste and cellular debris (46), regulation of iron homeostasis via red blood cell phagocytosis and iron recycling (47), maintenance of cholesterol homeostasis (48), maintenance of immune tolerance (49), and promotion of antimicrobial defense (50).

Once thought to be a homogenous population, recent studies reveal the heterogeneity of liver macrophages, beyond their origin, and evidence continues to accumulate to indicate that certain subpopulations are necessary in maintaining different aspects of homeostasis. Liver macrophage heterogeneity is increasingly deconvoluted with the emergence of precision sorting, single-cell/-nucleus RNA sequencing (scRNA-seq), and macrophage targeting technologies (51). For example, one study demonstrated that there are distinct populations of liver macrophages that function in an inflammatory and non-inflammatory or regulatory manner (52) Another study reported that diet-induced steatohepatitis impairs differentiation of myeloid cells in the liver and bone marrow (53). The diversity and subtypes of liver macrophages are further discussed in a dedicated section below, here we focus on the physiological distinction and roles of liver macrophages.

Interestingly, liver macrophages constitute a separate immune surveillance niche and form a cellular network distinct from those in the hepatic capsule (54). The capsular macrophages are phenotypically and developmentally different from KCs. They arise from circulating monocytes and express the macrophage markers F4/80 and CD64 yet are negative for the canonical resident macrophage markers Clec4F and T-cell immunoglobulin mucin domain containing 4(Tim4) (55). In addition, they can express markers also associated with dendritic cells or M1-like polarization, such as CD11c and MHCII. These cells sense bacteria in the peritoneum and promote neutrophil recruitment to restrict peritoneal bacteria spread to the liver (44). KCs on the other hand are prone to exposure to nutrients and microbial products in the blood rising from portal circulation. KCs encounter this rich venous blood and blood rich in insulin and oxygen from the hepatic artery in sinusoids (56). Mice deficient in KCs have weakened survival following infection with Listeria monocytogenes, indicating that KCs plays an important role in the control of bloodborne bacteria. Because of their high exposure to microbial products and nutrient-rich blood, KCs are generally tolerogenic and promote immune tolerance in the liver microenvironment (57). This is achieved by the expression of regulatory or anti-inflammatory molecules like interleukin-10 and prostaglandins, such signaling is important in maintaining immune tolerance and can also promote the differentiation of regulatory T cells (57) (Figure 2B).

Although KCs are generally tolerogenic, they retain the capacity to raise an alarm following the detection of danger signals from neighboring cells. KCs will sense disturbances in homeostasis from lipotoxic hepatocytes and endothelial cells; their subsequent signaling places them at the center of an intense cellular crosstalk (Figure 2B). This crosstalk allows the recruitment of other immune cells, the activation of HSCs, and the induction of apoptosis and phagocytosis of damaged cells. NAFLD is characterized by an early infiltration of monocytes; these monocytes differentiate into macrophages in situ and contribute to the inflammatory response. KC and other macrophage-derived cytokines can target HSCs. TNFα, IL-1β and TGFβ can all induce HSC activation. In turn, HSCs up-regulate several ligands, like CCL2, which are able to attract and regulate the activity of macrophages and other immune cells in NASH (57). Fibrosis progression is largely mediated by liver macrophage-HSC crosstalk (58). Once activated, HSCs will deposit extracellular matrix to create fibrous septa. Thus, macrophages amplify inflammation, support fibrogenic phenotypes and facilitate the survival of HSCs through the chemokines and cytokines they release (57–59). The initial inflammatory hit is key in the progression of the disease (60). KCs can also influence inflammation by binding of bacterial products in the liver’s portal vein. Activation of macrophages is also accompanied by the generation of malondialdehyde because of oxidative stress within the liver. However, KC ablation decreased steatosis and insulin resistance in the liver supporting the notion that KCs can also drive the development of early NAFLD (61). Indeed, an increase in CD68+ cells has been reported in biopsies from patients with NASH compared to those with steatosis (62). In animal studies, high-fat diet (HFD) and methionine-choline-deficient (MCD) diet-fed mice have more liver macrophages that produce pro-inflammatory cytokines (63). In addition, mice fed a MCD diet produced these inflammatory mediators at a higher level after 4 weeks and decreased later, suggesting that the release of these mediators may specifically contribute to progression and that macrophages may play different roles once NASH is established (64). Furthermore, clodronate injections, which deplete phagocytes, decrease steatosis and NASH severity (65). In the absence of KCs, fatty acid oxidation genes and peroxisome proliferator-activated receptor (PPAR) expression were increased in the liver. This mechanism was reported to be dependent on interleukin-1β-mediated suppression of PPAR-α (66, 67). Further, Zhang et al (68) demonstrated that p38α expression is increased in biopsies of patients with NAFLD, relative to control individuals. The importance of macrophage expression of p38α was demonstrated using macrophage-specific knockouts, in which mice deficient for p38α developed less severe liver disease and insulin resistance upon multiple dietary models of NAFLD and NASH. Hence, progression to steatohepatitis is promoted in a p38-dependent manner; p38α promotes macrophage polarization and proinflammatory cytokine release (57).

Liver macrophages can be categorized in multiple ways, for example, based on their origin, their polarization states, their functional specificities, or the expression of phenotypic markers. These ways of categorizing liver macrophages are a subject of continuous debate, with new categories being continuously proposed as we learn more about population and subpopulation specificities. When origin is used, broadly speaking, liver macrophages are grouped into resident and recruited cells, resident cells being KCs, that are embryonically derived, and recruited macrophages differentiate from monocytes that originate from bone marrow hematopoiesis (69) (Figure 3A). In mice, KCs are distinguishable by their variable expression of typical macrophage markers as F4/80^hi, CD68+, and CD11b^int cells, alongside specific liver resident markers Tim4 and Clec4F (70). Recruited macrophages will not carry the latter to markers, and their monocyte progenitors are Cx3cr1+, CD11b+, Ly6c+, and CC-chemokine receptor 2 (CCR2+) (43). CCR2 in particular plays an important role as it regulates the recruitment of monocytes and macrophages. Genetic deficiency of Ccr2 in mice fed a HFD reduced their food intake and decreased their development of obesity compared to wild-type mice. Ccr2 deficiency led to a decrease in AT macrophage content and inflammatory profiles, a substantial rise in adiponectin expression, decreased severity of hepatic steatosis and improved insulin sensitivity in obese mice. An antagonist of CCR2 was found to have significant effects in enhancing insulin sensitivity and lowering macrophages number in adipose tissue in obese mice. In obesity and its related metabolic consequences, CCR2 plays a role in recruiting monocytes to tissues under metabolic stress, sustaining and amplifying inflammation in AT and the liver, and promoting insulin resistance (71). Polarization state can also be used to classify macrophages, it refers to a final stage of differentiation that macrophages reach in response to signals from their immediate environment. Such signals include cytokines, growth factors, exosomes, fatty acids, or various PAMPs and DAMPs. Historically, macrophage polarization states were designated as either M1 or M2 (72). This M1-M2 classification is now considered outdated, macrophages in-fact exhibit enormous diversity, they can adopt intermediate phenotypes that lay on a sliding scale, where M1 and M2 are extremes (Figure 3B). When considering phenotypic extremes, M1 macrophages are proinflammatory, releasing an elevated level of proinflammatory cytokines and generating reactive oxygen species, which stimulate inflammatory responses (57, 59). M1 macrophages are activated by GM-CSF, TNF-α, and IFN-γ (73) and are involved in triggering the Th1 response via the production of cytokines, including TNF-α (74). They are fueled via glycolysis, which has been directly linked to IL-1β production (75). On the other hand, M2 macrophages express molecules that play anti-inflammatory and reparative roles (76). Furthermore, M2 macrophages utilize oxidative metabolism to fuel their functions (75). M2 macrophages are induced by IL-13 or IL-4 (77). They produce polyamines, ornithine, and arginase-1 and induce a Th2 response, promoting immune tolerance and tissue repair (78). M2 macrophages have been found to be protective in the context of inflammatory and metabolic disease, such as insulin resistance; however, they can play deleterious roles in certain conditions with extensive tissue remodeling, such as atherosclerosis and cancer (79). M2 macrophages can be further categorized into four subtypes based on the stimuli and activated transcriptional states: alternatively activated macrophages stimulated by IL-13 or IL-4 (M2a), type 2 macrophages activated via IL-10 or IL-1RA (M2b), deactivated macrophages stimulated via IL-10 or glucocorticoids (M2c), and M2-like macrophages induced by adenosines or IL-6 (M2d) (79). The canonical transcriptional regulators for M1-like and M2-like polarizations are respectively IRF5, NFkb, STAT1 and IRF4, PPARG, and other members of the STAT and SMAD families of transcription factors (80, 81). In terms of phenotypic identification of these subtypes, both M1-like and M2-like macrophages will express typical macrophage markers (e.g., CD68), in addition to CD11c or CD206 to denote polarization as M1 or M2, respectively (75, 77). Double positivity has been reported for these two markers, as well as associating other functional markers to denote subpopulations (82). Depending on stimulus and context, other markers have been employed to denote M1-like and M2-like polarization. These include CD40, CD86 and high expression of MHC-II for M1-like macrophages, and Dectin-1, CD36 and CD163 for M2-like macrophages (76, 83). The most robust phenotyping strategies should include a combination of appropriate surface markers and the use of intracellular markers that indicate function, i.e., transcription factors responsive to the condition studied.

Using the M1-M2 model, not only are the two extremes functionally distinct, they are also distinct in terms of their cellular metabolic profiles (84). The activation of M1 macrophages leads to the induction of aerobic glycolysis that generates ATP and lactate (85, 86). The pentose phosphate pathway (PPP), that branches from the early stages of glycolysis, is also induced under the same conditions, by IFN-γ or LPS, and produces NADPH and substrates for nucleotide synthesis (87). Additionally, the M1 phenotype is regulated by carbohydrate kinase-like protein (CARKL), implicated in the catalysis of sedoheptulose-7-phosphate (88). CARKL is involved in the metabolic control of pro-inflammatory immune responses by causing a redox shift in M1 macrophages. Consequently, stimulation of an M1-like state by LPS inhibits CARKL and leads to generation of NADH and GSH, while the stimulation of an M2-like state positively regulates CARKL (88). In terms of fatty acid oxidation and oxidative metabolism, IL-4 triggers STAT-6, which results in increased mitochondrial respiration (89, 90). M2 activation facilitates the entrance of pyruvate into the Krebs cycle, supporting the Electron Transport Chain (ETC) and providing the energy required for the remodeling and repair of tissues (84). The activation of M1 macrophages results in increased glycolysis, which aids in microbicidal activity and the management of hypoxia that occurs within the microenvironment of the tissue (91). Glucose, in the cytosol, is converted into L-lactate by glycolysis, in which hexokinase (HK), 6-phosphofructokinase 1 (PFK1) and pyruvate kinase (PK) are key enzymes. During glycolysis, glucose-6-phosphate can be propelled to PPP pathway promoting pentose NADPH and phosphate production. Pentose phosphates are used for the synthesis of amnio acid and nucleotide, while NADPH contributes to the production of ROS and NO (92, 93). Pyruvate is stimulated to lactate in hypoxic conditions, while decarboxylated into acetyl-CoA in the mitochondria in normoxic conditions. Here, acetyl-CoA goes into the TCA cycle, bringing reducing agents to the ETC to generate energy. The TCA metabolite, Citrate, participates in fatty acid synthesis when exported to the cytoplasm, this process can support membrane such functions as membrane synthesis (94). Two breakpoints in M1 macrophages cause the generation of itaconate and the increase of succinate, decreasing pH and stabilizing HIF-1a. HIF-1a results in upregulation of glycolysis and M1-like functions via GLUT1 expression and IL-1b production (84). What’s more, the production of NO inhibits the ETC. Nonetheless, M2 macrophages can acquire sufficient ATP from the ETC and via the TCA cycle (95).

Beyond the above standard classifications, new subpopulations of macrophages are constantly being discovered, they can functionally lay along the M1-M2 spectrum or have significant phenotypic differences from previously described populations. For example, the recently described functionally distinct KC1 and KC2 populations (96) or Trem2+ lipid-associated macrophage (LAM) (97, 98). The more abundant KC1s are identifiable by their low expression of CD206 and are endothelial cell-selective adhesion molecule negative (ESAM-), whereas KC2s highly express CD206, ESAM, CD36 and the lymphatic vessel endothelial hyaluronan receptor-1 (LYVE1) (52). LAMs were first described in adipose tissue, they enhance the degradation and processing of lipids via lipoprotein lipase (Lpl), fatty acid transporter Cd36, and fatty acid binding proteins 4 and 5 (Fabp4, Fabp5) (99). LAMs have also been described in other lipid-rich environments such as in atherosclerotic plaques, here LAM expression of Spp1 and Cd9 is correlated with lesion calcification and a downregulation of pro-inflammatory genes (79). With regards to NAFLD, LAMs can localize to fibrotic areas, macrophage aggregates or to hepatic crown-like structures (hCLS) (100, 101), and have been reported to mitigate liver fibrosis (100). Daemen et al ( 101) revealed a number of LAM specificities, that they express Trem2, Cd63, Cd9 and Gpmnb; and that they are recruited through Ccr2. Their failure to accumulate in the liver of Ccr2-deficient mice increases liver fibrosis upon high-fat, high-sucrose feeding, indicating that they play an important role in tissue remodeling (101). Interestingly, Trem2 that is expressed on LAMs also exists in a soluble form, and levels in plasma reflect recruitment and expansion of LAMs in the liver (98) (Figure 3B).

Whilst significant progress has been made in deciphering the heterogeneity of liver macrophages, under physiological and pathophysiological conditions, most of these studies have been carried out in murine models. More recently, several resources have been developed using novel single cell or single nucleus sequencing to translate findings to humans and shed light on conserved subsets and population features in liver macrophages from human biopsies (e.g., www.livercellatlas.org) (102–104). Thanks to these studies that expand the knowledgebase on human liver macrophages, the translatability of findings from murine studies is increasing with time.

Among the dietary models, a common approach has been to use diets deficient in one or more nutrients. The MCD diet has been adopted for some studies and has the advantage of inducing NASH within a few weeks, but it fails to model the human disease in terms of its often-concomitant obesity. Animals on this diet lose weight, but they still develop hepatic insulin resistance (127). A modification of this diet, the choline-deficient-L-amino-acid-defined (CDAA) diet, overcomes the weight loss issue with the MCD diet; though NASH develops later (20-22 weeks), the diet still elicits limited metabolic syndrome-like characteristics compared to NAFLD in humans (128–131). The established atherogenic diet has also been investigated for its usefulness in NAFLD mouse models: the diet induces steatosis, inflammation, and other histological features of NASH and NAFLD in the mouse liver but does not cause weight gain or insulin resistance (132, 133). Also, simply adding fructose to water was found to result in steatosis, increased weight, and hyperglycemia, though no definitive NASH (134). More commonly used currently are high-fat diets (HFD), which can vary widely in the amounts and types of fat used. These generally induce steatosis and inflammation to a lesser degree than in mice fed an MCD diet. However, they have the advantage of inducing a metabolic syndrome-like pathology that is more similar in profile to human NAFLD, with obesity, insulin resistance, and hyperglycemia (135). Modifications of the HFD, including a combination with fructose (High Fat High Fructose Diet) or sucrose (High Fat High Sucrose Diet) added, either to water or food, result in a phenotype more closely resembling human NAFLD, but not the full extent of liver damage (136, 137). Mice on other HFD variations, the High Fat High Cholesterol Diet (111) and the American Lifestyle Induced Obesity Syndrome (ALIOS) diet which is rich in trans fats (138), also showed additional NAFLD compared with mice on a typical HFD. A large disparity exists between individuals in terms of evolution from NAFLD to NASH, and between mouse strains used for in vivo modelling. In one study, an MCD diet was used to induce NASH in C57Bl6/J and Balb/c mice, which exhibit very different responses in terms of innate and adaptive immune responses.C57Bl6/J mice that are more prone to Th1 and M1-like responses, showed more steatosis and lobular inflammation following 4 weeks on the MCD diet than Balb/c mice, which are more prone to Th2 and M2-like responses. Accordingly, neither the Th1/Th2 bias nor IL-4 (interleukin 4) or GATA-3 expression in the liver of either strain is significantly modified by MCD feeding (indicating that innate immune polarization plays a more important role). As a result of MCD feeding, liver mRNA expression of macrophage activation markers M1 (iNOS), MLD (inducible NO synthase), and MCD (CXC chemokine ligand 10) were significantly higher in C57BL6/J mice, however, M2 polarization markers IL-10 and MGL-1 (macrophage galactose-type C-type lectin-1) remained the same. In addition, C57Bl6/J mice fed MCD had a higher level of circulating IL-12 than Balb/c mice fed MCD. Compared with mice fed MCD, macrophages isolated from the livers of C57Bl6/J mice expressed significantly more M1 markers (139).

As well as dietary fat, the form, content and delivery of sugar can be manipulated to influence susceptibility to NAFLD and a NASH-like phenotypes in rodents (140). The considerations have been whether sugar should be delivered in the form of glucose, fructose or sucrose (or in combination), and whether this is provided as part of the solid diet or in drinking water. There are mechanistic advantages and practical considerations for each of these options, for example quantity consumed or calories per gram are easier to control when incorporated into solid food. Fructose alone making up to 60% (w/v) of drinking water can reliably induce steatosis, when fructose is combined with glucose in a solid diet (30% w/v of each), this also results in steatosis (140). Interestingly, when prolonged the 60% (w/v) fructose in drinking water can induce fibrosis, and the same concentration in a solid diet (kcal/w) can induce inflammation and fibrosis. Up to 50% (w/v) sucrose in drinking water, can induce inflammation and fibrosis. In studies of these diet, fructose-supplemented drinking water causes steatosis after 8 weeks, and leads to a significant rise in body weight, and glucose and plasma triglyceride levels (141). Moreover, intestinal bacterial overgrowth is detected after 8 weeks of treatment along with high levels of endotoxin in the portal blood and activation of KCs (134). However, weight gain, followed by development of fat deposits in the abdomen, is not essentially prognostic of steatosis. It has been revealed that fructose causes greater fat accumulation in the liver than sucrose and glucose, despite of the weight gain from glucose and sucrose (142).

Among the chemical mouse models of NAFLD is the well-used streptozotocin induction of T2D. A low dose of intraperitoneal or subcutaneous streptozotocin shortly after birth leads to inflammation and the destruction of pancreatic islets; when combined with a HFD, animals progress quickly through steatosis, NASH, and increased fibrosis to hepatocellular carcinoma (HCC) around 20 weeks of age (143). The use of carbon tetrachloride (CCl₄) in multiple IP doses leads to significant but reversible fibrosis (144); when combined with HFD steatosis, NASH and increasing fibrosis are seen (145);however, the systemic metabolic aspects of NAFLD are absent (146). One-time administration of diethylnitrosamine (DEN) has been used to model HCC; when combined with a HFD, animals do gain weight and develop NASH, as well as some other aspects of the metabolic syndrome associated with NAFLD (147). In each case, the value of toxicity or drug-induced NASH is questionable because it most likely doesn’t parallel the progression of the disease in humans and the mechanisms leading to fibrogenesis may be completely independent.

Leptin is an adipokine that centrally controls appetite. The leptin knockout mice, Lepob/Lepob (ob/ob), are obese, insulin resistant, and hyperglycemic, and show some degree of steatosis. Likewise, leptin receptor knockout mice, Leprdb/Leprdb (db/db), have a similar phenotype to ob/ob, and in both cases, an additional stimulus is required for the development of NASH. Several groups have explored the combination of genetic models with dietary or chemical stimulus to induce NAFLD with varying degrees of success (148–151). Mice with a mutation in Alms1, which encodes a protein involved in control of satiety via the hypothalamus, also require a dietary stimulus (HFD) to induce symptoms similar to NAFLD, with the lipid profile not mirroring the human disease (152). Sterol regulatory element-binding protein (SREBP-1c) transgenic mice also develop insulin resistance and steatosis (123). Better for examining the metabolic syndrome are the established atherosclerosis models such as ApoE-/- mice (apolipoprotein E deficient) and Ldlr-/- mice (LDL receptor deficient); these animals are predisposed to hypercholesterolemia, atherosclerosis, and obesity, and on a HFD or Western diet will also develop NAFLD/NASH (153).