Targeted Nutrient Modifications in Purified Diets Differentially Affect Nonalcoholic Fatty Liver Disease and Metabolic Disease Development in Rodent Models

ABSTRACT

Nonalcoholic fatty liver disease (NAFLD) is a complex spectrum of disorders ranging from simple benign steatosis to more aggressive forms of nonalcoholic steatohepatitis (NASH) and fibrosis. Although not every patient with NAFLD/NASH develops liver complications, if left untreated it may eventually lead to cirrhosis and hepatocellular carcinoma. Purified diets formulated with specific nutritional components can drive the entire spectrum of NAFLD in rodent models. Although they may not perfectly replicate the clinical and histological features of human NAFLD, they provide a model to gain further understanding of disease progression in humans. Owing to the growing demand of diets for NAFLD research, and for our further understanding of how manipulation of dietary components can alter disease development, we outlined several commonly used dietary approaches for rodent models, including mice, rats, and hamsters, time frames required for disease development and whether other metabolic diseases commonly associated with NAFLD in humans occur.

The pathophysiology of nonalcoholic fatty liver disease (NAFLD) is still not completely understood, nor are there any approved therapies for treating this disorder, so there is a need for further research. Rodent models are typically used as surrogates for human disease and this article provides examples of how targeted nutrient manipulations in rodent diets influence NAFLD development in mice, rats, and hamsters. This article also focuses on study design details including nutritional components of the diets (and their controls), proper control diet choice, animal model/strain, and length of the study to clearly outline the strengths and drawbacks of the different dietary methods of inducing NAFLD in rodents.

Introduction

Nonalcoholic fatty liver disease (NAFLD) is a spectrum of disorders characterized by excessive lipid accumulation in hepatocytes. Each stage of the disease spectrum has distinctive histopathological characteristics. The beginning stages include simple hepatic steatosis, which is characterized by fat droplet accumulation in hepatocytes and this is usually benign and asymptomatic (1, 2). The disease may progress further to nonalcoholic steatohepatitis (NASH), which may include hepatocellular injury, ballooning (i.e., cellular swelling), and/or inflammation. If left unchecked, NASH can lead to fibrosis, cirrhosis, and ultimately hepatocellular carcinoma (HCC), thus affecting overall liver function (3, 4). A growing body of evidence portrays NASH as the hepatic manifestation of the metabolic syndrome because the majority of patients with NASH (and those with steatosis) suffer from obesity, diabetes, and insulin resistance (IR) (2, 5). However, whether IR causes or is caused by NAFLD/NASH is yet to be clearly defined. Most obese adults have hepatic steatosis, and at least one-third of these individuals will eventually develop worsening NAFLD (6) including NASH, which is projected to be the leading cause for liver transplantation in the upcoming decade (7).

Accumulation of fat droplets in the form of triglycerides (TGs) in hepatocytes constitutes steatosis, which is a hallmark feature of NAFLD (8). NAFLD is histologically diagnosed when TG accumulation occurs in >5% of hepatocytes (9) in the absence of significant alcohol consumption. Although there are many histological scoring methods for NAFLD, some of the common ones used in the literature are those by Kleiner et al. and Brunt et al. (9–12). The extent of steatosis can be graded according to the percentage of steatotic hepatocytes: mild, 5%–33% (score: 1); moderate, 33%–66% (score: 2); and severe, >66% (score: 3). Similarly, lobular inflammation is graded as a score of 0–3 and hepatocyte ballooning as 0–2. The sum of these scores is known as the NAFLD activity score (NAS); it ranges from 0 to 8 and typically scores >5 are considered confirmed cases of NASH. Similar scoring systems are also modified and translated to be used in rodent models and it has been shown that the human scoring system is largely reproducible in rodent NAFLD (13). Fibrosis is typically scored from 0 to 4, with 0 = no fibrosis, 1–2 = mild to moderate fibrosis, 3 = advanced fibrosis, and 4 = cirrhosis, and fibrotic scores are calculated separately from NAS scores. To have a complete understanding of disease development, most researchers typically report both NAS and fibrosis scores (14).

Obesity and particularly IR are tightly associated with the genesis of NAFLD and NASH. IR can lead to an increase in the circulating free fatty acids (FFAs) through adipose tissue lipolysis. Elevated insulin (due to increased secretion) and glucose concentrations (reduced uptake by skeletal muscle) typically found in conditions of IR also promote de novo lipogenesis in the liver via multiple mechanisms (15). IR also directly inhibits β-oxidation, thereby promoting hepatic accumulation of TGs and thereby steatosis (15). IR also can contribute to development of NAFLD via multiple other mechanisms. Patients with NAFLD have IR in liver, muscle, and adipose tissue (16) and this led to NAFLD being called the hepatic manifestation of metabolic syndrome. Day and James (17) proposed the “two hit” model for NASH development, with the first hit being steatosis with IR contributing to development of steatosis. This makes the liver vulnerable to the second hit (oxidative stress, proinflammatory cytokines, etc.), thus promoting NASH and fibrosis development. However, given that nonobese individuals, including those from developing countries like China and India with little to no IR, also have NAFLD, recent research has also explored other possible mechanisms toward NAFLD development independent of IR. Currently, it is thought that there are multiple hits leading to development of advanced liver disease including increased oxidative stress and inflammation, hepatotoxicity by FFAs and ceramides, gut bacterial alterations, and increased gut permeability, compounded by the existence of the metabolic disturbance of IR (18, 19).

In spite of the enormous amount of research in the field of NAFLD/NASH in the past decade, the precise mechanisms underlying the development of NAFLD and its progression to NASH have not been completely elucidated, including its link to metabolic syndrome, requiring additional studies and models to elucidate its pathophysiology. Because of its growing worldwide prevalence, various animal models that mirror both the pathophysiology and the histopathology of each stage of NAFLD/NASH are available. Certain dietary approaches can drive NAFLD/NASH in rodent models to mimic human disease and produce different severities of disease along the NAFLD spectrum, and, depending on the dietary manipulations, likely work by unique mechanisms (Tables 1, 2, Figures 1, 2). This is essential in determining how the disease progresses, and also helps in evaluating different therapeutic approaches toward the treatment of specific stages of NAFLD (5).

TABLE 1.

Summary of commonly used diets and their expected effects on nonalcoholic fatty liver disease and nonalcoholic steatohepatitis in rats and mice¹

Diet	Rodent model	Body weight1	Fasting glucose/insulin	Steatosis	Steatohepatitis	Fibrosis	Time frame (fibrosis)2	References
Methionine- and choline-deficient diet (MCD)	Rats and mice	↓	↓	+++	+++	++	4–8 wk	(24–26)
0.1% Methionine and choline-deficient high-fat diet	Mainly mice	↓*	No change	+++	+++	++	6–12 wk	(27–29)
Choline-deficient amino acid–based diet with 0.17% methionine (CDAA)	Rats and mice	No change	↑ (Mainly mice)	+++	++	++	4–12 wk (rats) 12 wk (mice)	(30–34)
Choline-deficient high-fat diet (CD)	Mainly mice	↑	↑	+++	++	++	12 wk	(35–38)
High-fat diet (HFD)	Rats and mice	↑	↑	+++	++	+ Mild at best	24 wk (rats) 16 wk (mice)	(39–42)
High-fructose diet (HFr)	Mainly rats	No change	↑	+++	++	++	12 wk	(43, 44)
High-fat, high-fructose, high-cholesterol diet	Rats and mice	↑	↑	+++	++	+ (Mainly mice)	16 wk (rats) 20–30 wk (mice)	(45–48)

¹+, mild; ++, modest; +++, severe. Body weight and fasting glucose/insulin for arrows is relative to the control diet.

²The length depends on diet formula; length of the study; and species, strain, and gender of the animal model.

^*Compared with a low-fat, methionine- and choline-sufficient group. Body weight of these animals typically remains unchanged compared with baseline.

TABLE 2.

Study design factors to consider in diet-induced nonalcoholic fatty liver disease rodent models¹

	High-fat diets	High-fructose diets	High-fat, -fructose, and -cholesterol diets	Choline-deficient high-fat diets	Methionine- and choline-deficient diets
Dietary modifications commonly used	30–60 kcal% fat, higher SFAs increase ER stress, higher ω-6 PUFAs increase increase oxidative stress, both increase NASH More sucrose or fructose leads to NASH and mild fibrosis	Usually 60–70 kcal% fructose drives steatosis, NASH The addition of sucrose (50%fructose) also effective effective for steatosis, NASH	40 kcal% fat (trans fat or SFAs), 20–40 kcal% fructose, and 1–2% cholesterol Fat type and cholesterol increase ER and oxidative stress/fibrosis Fructose drives steatosis and inflammation	Fat amount (30–60 kcal% fat) Lard commonly used; typically can drive steatosis, but prolonged feeding (6 mo) can cause fibrosis	Rapid onset of steatosis (1 wk), NASH/fibrosis in 6–8 wk Addition of fat (≤60 kcal% fat) Fat type typically lard, butter (SFAs), or corn oil (PUFAs), addition of sucrose and/or cholesterol drives further NASH/fibrosis
Other metabolic effects	Increases body weight IR/glucose intolerance	Increases body weight IR/glucose intolerance Increased plasma TGs (typically ratsand hamsters)	Increases body weight IR/glucose intolerance Increases plasma lipids	Increases body weight Less IR than choline-sufficient diet	Reduces body weight, but 0.1% methionine maintains weight No IR, reduced plasma lipids
Matched control diet	Low-fat diet with matched amount of sucrose or mostly corn starch	Low-fat diet with 60–70 kcal% as either glucose or corn starch	Low-fat diet with 60–70 kcal% as either glucose or corn starch	Low-fat diet with choline	Methionine- and choline-sufficient diet

¹ER, endoplasmic reticulum; IR, insulin resistance; NASH, nonalcoholic steatohepatitis; TG, triglyceride.

FIGURE 1.

Effects of dietary fructose, total and saturated fat, and cholesterol on mechanisms affecting nonalcoholic fatty liver disease development in rodent models. ER, endoplasmic reticulum; FFA, free fatty acid; IR, insulin resistance; NASH, nonalcoholic steatohepatitis; TG, triglyceride.

FIGURE 2.

Effects of choline and methionine + choline deficiency on mechanisms affecting nonalcoholic fatty liver disease development in rodent models. ER, endoplasmic reticulum; NASH, nonalcoholic steatohepatitis; PC, phosphatidylcholine; PE, phosphatidylethanolamine; SAM, S-adenosylmethionine; TG, triglyceride.

Studies frequently use either mice or rats as animal models for studying fatty liver disease (8, 20). The most common model used for diet-induced NAFLD is the C57BL/6 mouse model, which is likely due to their increased susceptibility relative to other mouse strains (21). Rats [Sprague Dawley (SD) and Wistar] and hamsters are other commonly used models. Many diets including high-fat diets (HFDs) and methionine- and choline-deficient (MCD) diets, of which there are many variations, produce a robust model for different stages of NAFLD/NASH depending on the type of diet and length of feeding. Recently a review was published regarding the influence of diet on NAFLD in rats (22) as well as 1 in mice (23), but none to our knowledge have been done for hamsters. Therefore, we will provide a more comprehensive description of dietary strategies in these 3 rodent models to aid in the selection of the diet choice during the study design phase.

HFDs

The term “HFD” encompasses a wide variety of diet formulas with different fat types and amounts (30–60 kcal% fat), as well as other compositional differences such as low or high sucrose amounts or different forms (i.e., pellet or liquid). HFDs typically (depending on the fat source) increase body weight and body fat and induce IR in rodent models. One of the widely used HFDs for obesity research, D12492 (60 kcal% fat, mainly lard), induced visceral obesity and hepatic steatosis characterized by significantly increased liver and plasma FFA and TG concentrations and plasma alanine aminotransferase (ALT) in C57BL/6 mice fed for 8 wk (49). In another study, C57BL/6 mice fed diet D12492 for 16 wk exhibited increased body weight and adipose tissue weight, widespread hepatic steatosis indicated by Oil Red O staining and increased hepatic TGs, mild fibrosis, and adipose tissue inflammation (39). NAS scores of C57BL/6 animals fed D12492 for 16 wk were ∼2–4 (24). When fed chronically (∼52 wk), an HFD (D12492) induced NASH characterized by inflammation along with excess body weight, hyperinsulinemia, and hypercholesterolemia (50). Even a slightly lower-fat (45 kcal% fat, D12451) containing HFD induced steatosis and steatohepatitis after 6 mo in C57BL/6 mice (40). NAFLD in mice was worsened (mild fibrosis) by the addition of sucrose, as shown in a different study that compared animals fed an HFD (36 kcal% fat as mainly milk fat) with those fed a high-fat, high-sucrose diet (36 kcal% fat, 30 kcal% sucrose) (51). The type of fat in the HFD formulation also plays a role, as evidenced in the study by de Wit et al. (52) which observed that palm oil (∼50% SFAs, mainly palmitic acid) in 45 kcal% fat diet increased liver TGs and body weight and reduced insulin sensitivity more rapidly (as early as week 2) than other fat sources (olive oil, safflower oil) in C57BL/6 mice.

Studies in SD rats observed that certain HFDs increase liver fat concentrations quite rapidly (within days) as well as hepatic IR before significant increases in peripheral fat deposition occur (53). Chronically, HFD-induced liver fat accumulation may not follow a linear progression and liver fat concentrations may actually decrease, then increase again during prolonged HFD feeding in rats (54). Researchers also combine HFD feeding with chemicals such as streptozotocin or carbon tetrachloride (CCl₄) to induce NASH in rat models (24, 55, 56).

Like in mice, fat type should be considered in NAFLD studies with rats, because markers of liver injury in circulation, ALT and aspartate aminotransferase (AST) (at 4 and 24 wk), and markers of endoplasmic reticulum (ER) stress including those involved with increased protein folding (X-box-binding protein 1, glucose regulated protein 78) and apoptosis [caspase-3, C/EBP homologous protein (CHOP)] in the liver were increased in rats fed a 45 kcal% fat diet with lard relative to those fed a diet high in corn oil (57). These markers were associated with higher concentrations of hepatic SFAs (after only 1 wk), and thus ER stress in combination with steatosis may set the stage for increased NASH development (57). Because the ER makes up over half the membrane composition of hepatocytes, in vitro work using primary hepatocytes confirmed that increased amounts of SFAs as palmitic acid caused increased incorporation of these fatty acids into lipotoxic phospholipids and led to induction of ER stress markers such as CHOP (proapoptotic) in a dose-dependent manner (58). In contrast, the addition of oleate reversed this effect and allowed for more palmitic acid to be incorporated in more benign TGs (58).

Increasing the fat content of HFDs (∼70 kcal% fat) is also effective in rodent models. A classic HFD-NAFLD model used SD rats fed a liquid diet composed of 71 kcal% fat (combination of olive oil, corn oil, and safflower oil), 11 kcal% carbohydrates, and 18 kcal% protein for 3 wk, as compared with control rats fed a lower-fat diet with 35 kcal% fat, 47 kcal% carbohydrates, and 18 kcal% protein (59). This diet caused hepatic steatosis, and lipid concentrations were almost 2-fold those in control rats, which was due to the increased dietary contribution of TGs. Similarly to human NAFLD patients, these rats developed IR as suggested by elevated plasma insulin concentrations (59); this diet also developed NASH and IR in male C57BL/6 mice after 16 wk (60). Even in this HFD formulation, the type of fat plays an important role because replacement (two-thirds or more) of long-chain TGs with medium-chain TGs (MCT oil) resulted in lower (or no) steatosis and lower hepatic TG and inflammation (TNF-α) in SD rats (61).

HFDs can also drive the development of NASH in hamsters. A 45 kcal% fat diet (mainly lard, D12451) increased liver and plasma lipids (total cholesterol and TGs) and steatosis score (histology) in golden Syrian hamsters fed for 10 wk relative to a 10 kcal% fat diet (62). In addition, hepatic inflammation as measured by expression of inflammatory cytokines like monocyte chemoattractant protein 1 (MCP-1), TNF-α, IL-1β, and IL-6 was elevated in the HFD group (62). Another study also reported development of steatosis (mainly microvesicular) in a hamster model fed an HFD with ∼30 kcal% fat as mainly lard with 0.2% cholesterol after 8 wk, whereas those fed a low-fat purified diet (AIN-93M) had maintained normal liver histopathology. In addition to elevations in liver lipids, these hamsters also had elevated plasma total cholesterol, LDL cholesterol, and TGs (63).

It is important to note that when fed for equal lengths of time, HFD feeding results in only 10% of liver fat concentrations than what accumulates on an MCD diet (64) and induces only mild steatosis and mild/no liver fibrosis as compared with MCD diets (25, 65), highlighting an important difference between these dietary regimes. Unlike rodents fed an MCD diet, an advantage of using an HFD is that HFD-fed mice will have features of metabolic disease including weight gain, IR, and/or glucose intolerance; however, the practicality of the model is limited by length of the dietary regime, especially for fibrosis induction (41, 42).

MCD Diets

Among the different dietary approaches for diet-induced NAFLD in rodents, MCD diets produce the most severe NASH phenotype in the shortest time frame, which is highly favorable for those studying ways to slow down or reverse this phenotype. The formulation of MCD diets requires the replacement of whole protein (such as casein) in purified diets with crystalline amino acids and the removal of both methionine and choline. These diets typically induce measurable hepatic steatosis (mainly macrovesicular) in mice and rats by 1–4 wk, and if fed longer (1–2 mo), it can progress to inflammation and fibrosis (66, 67). Liu et al. (24) observed that SD rats fed an MCD diet for 7 wk exhibited significant steatosis with lobular inflammation and mild perisinusoidal fibrosis. Surprisingly, this group also observed hepatocellular ballooning, which is typically not seen in rodents fed other NAFLD diet strategies. Steatosis scores in rats consuming MCD diets range around 2–3 (24, 26). The mechanism for steatosis includes impaired VLDL secretion due to lack of phosphatidylcholine synthesis and increased lipolysis of adipose tissue (68). The progression of steatosis to steatohepatitis in MCD mouse models involves downregulation of proteins affecting methionine metabolism and oxidative stress such as peroxiredoxin, which participates in cellular defense against the development of hepatitis (69). Other mechanisms involve mitochondrial S-adenosyl-l-methionine depletion due to reduced methionine intake and this leads to mitochondrial dysfunction and mitochondrial glutathione depletion (70). This leads to increased propensity for hepatocyte injury via oxidative stress and inflammation. Some studies have shown that MCD diets more closely mimic the mechanisms implicated in the pathogenesis of human NASH, including ER stress, oxidative stress, and autophagocytic stress, relative to other dietary models, including HFDs (25).

Within the context of an MCD diet, other dietary components also affect the NASH phenotype. Sucrose is an important component of the MCD diet, because replacing it with cornstarch greatly reduces liver fat accumulation, inflammation, and injury, likely through reductions in sucrose-induced de novo lipogenesis and TG synthesis (71). Although both glucose and fructose can induce hepatic fat accumulation, fructose was more effective than glucose at inducing hepatocellular injury measured by histology, apoptosis staining, and serum ALT, in C3H/HeOuJ mice fed MCD diets for 3 wk (72). The source of dietary fat can also alter the phenotype; for example, relative to high-SFA sources (coconut oil, beef tallow), omega-6 PUFA sources such as corn oil can increase liver fat oxidation and induce expression of proinflammatory genes leading to inflammation, although this does not necessarily correlate with increased liver damage (73). Also, relative to butter fat (high in SFAs), olive oil (high in MUFAs) reduced liver TG accumulation, whereas fish oil (high in ω-3 PUFAs) reduced liver cholesterol concentrations (74). Besides fat type and sucrose, the addition of 1% cholesterol to a low-methionine (0.1%), no-choline HFD (A16092003, 45% kcal, mainly lard) accelerated development of fibrosis (∼6 wk) in Wistar rats (75), as it did also in the context of an MCD (0% methionine) diet with less fat (21 kcal% fat, corn oil) in C57BL/6 mice fed for 12 wk (76).

Unlike humans with NAFLD, or other HFD-induced rodent models of NAFLD which usually are obese and have IR (among other metabolic complications), rats and mice fed MCD diets lose weight owing to a vastly lower caloric intake (24, 77–79), display cachexia, have no IR, and have low fasting serum insulin and glucose, leptin, and TG concentrations (78). This limits the extrapolation of data from this model to human NASH. Despite these differences, MCD diets are still a good candidate for screening experiments (i.e., drug compounds, genetic modifications) which can provide answers regarding how they influence the development or reversal of advanced NASH, mainly liver fibrosis phenotype, in a relatively short time frame.

Owing to its effect on weight loss and to counter this effect, some modifications have been made to traditional MCD diets. Matsumoto et al. (27) removed methionine and choline from an HFD background (60 kcal% fat, A06071301B) with the idea of providing more energy as fat to reduce weight loss. Despite an increased fat amount compared with traditional MCD diets, this diet still caused a reduction in food intake and body weight in both C57BL/6 and A/J mice. However, adding a small amount of methionine (0.1% wt:wt) back in the diet (A06071302) helped better maintain energy intake and body weight to the same level as animals fed a low-fat, methionine- and choline-sufficient diet. This diet also increased levels of liver fat accumulation, inflammatory markers, and AST and ALT within 6 wk and fibrosis progressed from week 3, week 6, and week 14; however, hepatic inflammation, TG concentrations, and ballooning did not change from week 6 to week 14. Although the modified diet arrested weight loss in both A/J and C57BL/6 mice, it increased fibrosis in only C57BL/6 mice after 6 wk, but if fed longer (≥9 wk), both mouse models developed fibrosis on this diet. At week 9, the steatosis score was 3, lobular inflammation score was 1–2, ballooning score was 1, and fibrosis scores were ∼2–3. Similar results were also observed by us (unpublished data) and Chiba et al. (28) in a similar diet containing 45% kcal fat, in C57BL/6 mice. However, the amounts of sucrose in the 45 kcal% diet in the aforementioned studies were 17% compared with 7% in the 60 kcal% fat diet. If the sucrose amounts are kept similar, the higher fat amount (i.e., 60 kcal% fat) provided a more robust degree of fibrosis than 45 kcal% fat or 10 kcal% fat after 8 wk in C57BL/6 mice (80). In another study, Cong et al. (81) used a modified HFD (60 kcal% fat) to simultaneously contain low amounts of methionine (0.15%) and choline (0.06%). C57BL/6 mice fed the diet for 23 wk developed obesity, IR, and dyslipidemia as well as liver steatosis, inflammation, and fibrosis. Therefore, the addition of a low methionine dose in the context of a choline-deficient HFD background can allow for an improved phenotype or, at the very least, reduce the weight loss concern associated with MCD diets. This idea of modifying so-called “standard” HFDs is powerful because it allows the researcher to “fine-tune” the phenotype to meet their needs.

The addition of 0.17% methionine in a choline-deficient amino acid–based diet (commonly known as the CDAA diet) initially formulated by Nakae et al. (30, 82) for studying HCC in rats with ∼30 kcal% fat (mainly contains Primex shortening, a source of trans fat) also induced significant histopathological evidence of the complete NAFLD spectrum in a relatively short time frame, which includes steatosis (score: 3), inflammation (score: 1–2.4), and fibrosis (score: 2.4) (31, 32, 83). AST and/or ALT concentrations in these animals were elevated after 4 (33), 8 (31), 10 (83), or 12 (32) wk; however, these rats had no evidence of IR and gained similar or less weight and had lower blood glucose, TG, and cholesterol concentrations compared with the choline-sufficient groups. De Minicis et al. (34) also used this same CDAA diet in mice (fed for 1–9 mo) and observed relevant NASH (NAS > 5) and fibrosis after 3 mo (associated with increased mRNA expression collagen 1α, α-smooth muscle actin, and metallopeptidase inhibitor-1), which progressed further after 6 and 9 mo. Additionally, unlike in rats, in mice this group observed IR as shown by higher fasting plasma insulin concentrations and lower glucose uptake. Others who fed this diet to mice have observed significant fibrosis at 22 wk, which progressed further after 65 wk and was associated with oxidative DNA damage and HCC (66.7%) after 84 wk (84). It should be noted that the description of the CDAA diet in the methods section of some publications may lack the detail that this diet is lower in methionine than what is typically added in purified diets (typically 0.5% from casein), which is an important driver of the NASH/fibrosis phenotype. The methionine amount in CDAA diets is also lower than the NRC recommendations for mice (0.5%), but may be adequate for rats (NRC recommendation: 0.23% for methionine + cystine) (85). Furthermore, the control diet is unusual as it is also low in methionine (0.17%) and has a very high concentration of choline bitartrate (14.48 g/kg) (33), which is >7-fold higher than what is typically added in purified diets (2–2.5 g/kg) to meet NRC recommendations (85). Although it is unclear why a higher than typical amount of choline was added, the intention may have been to provide more methyl donors to the control group to compensate for the reduced amount of methionine in this diet.

In addition to rats and mice, MCD diets can drive significant steatosis and fibrosis in male F1B hamsters within 8 wk. However, in contrast to rats and mice fed an MCD diet, hamsters had similar body weight and elevated plasma TG compared with animals on a grain-based diet (86). When fed a diet containing 0.13% methionine and no added choline with 21 kcal% fat, F1B hamsters had minimal weight gain, and exhibited a high degree of steatosis (macrovesicular) and hepatocellular ballooning relative to those fed a diet with 25 kcal% fat and normal methionine and choline amounts; however, perhaps due to the study length of only 4 wk, the MCD diet–fed hamsters exhibited no fibrosis (87).

Choline-Deficient Diets

Typically, choline-deficient (CD) diets used in fatty liver disease studies tend to contain higher amounts of fat (30–60 kcal%) and these diets can induce steatosis, inflammation, and mild fibrosis over 10 wk without the reduction in body weight seen in MCD diets (35–37, 88), which makes CD diets more appealing to some researchers. However, choline deficiency, even in the context of a lard-based HFD (45 kcal% fat, mainly lard, D05010402), improved glucose tolerance compared with the choline-sufficient group in C57BL/6 mice fed for 8 wk (36). Time frame may be important as C57BL/6 mice on D05010402 showed impaired glucose tolerance if fed longer (6 mo) when compared with a grain-based diet (38), but it is not possible to conclude that the longer time frame was responsible for the differences in glucose intolerance given the many differences between these diets in addition to choline amounts. However, fibrosis in mice consuming CD diets tends to be minimal unless 1) the diet is fed for a long time period (6–12 mo) (38) or 2) a compound such as ethionine is added to the diet or water (36, 89). This lack of fibrosis found in mice with a choline-deficient HFD (D05010402) when fed for shorter time frames (∼8 wk) may be due to an upregulation of enzymes involved with phosphatidyl choline synthesis (increased phosphate cytidylyltransferase 1, choline, alpha, Pcyt1amRNA expression) and TG synthesis (increased acyl-CoA synthetase long chain family member, Acsl-1, acyl-CoA synthetase long chain family member 4, Acsl-4, and glycerol-3-phosphate acyltransferase, Gpat) and diglyceride esterification (increased diacylglycerol O-acyltransferase 2, Dgat2), which shunt potentially toxic FFAs to TGs (36). Others also found that choline deficiency in the context of an HFD (with 37 kcal% fat as lard, 46 kcal% sucrose) increased phosphatidyl choline synthesis and maintained VLDL secretion relative to those fed a choline-sufficient diet (90). Therefore, a reduced accumulation of toxic lipids and maintenance of VLDL secretion explain why (at least in the shorter term) only steatosis develops without NASH.

When considering a CD diet, significant steatosis occurs only when choline deficiency is combined with a high-fat background (60 kcal% fat, D05010403) because the low-fat CD diet (10 kcal% fat, D05010401) failed to induce any significant steatosis (91), or induced it to a mild degree relative to a higher-fat (45 kcal% fat, D05010402) diet after 8 wk (36). In contrast, an MCD diet with a lower-fat diet background (i.e., 12 kcal% fat as corn oil) was capable of initiating steatosis, NASH, and fibrosis in just 15 d, and methionine deficiency was the main driver of liver injury, likely through reduced S-adenosylmethionine and glutathione (70). After feeding either CD or MCD diets to Wistar rats for 7 wk, the MCD diet group had significantly higher scores of liver inflammation (2.9 in the MCD compared with 2.1 in the CD group) and steatosis (1.9 in the MCD compared with 0.6 in the CD group) and a rapid increase in ALT and presence of fibrosis, all of which were absent in rats fed the CD diet; a similar effect was also found in mice consuming MCD or CD diets for 15 d (26, 70). However, the CD diet–fed rats had IR and gained weight and had higher plasma lipids compared with the MCD group and those fed a grain-based diet (26).

High-Fructose Diet

Humans consuming a significant number of calories from fructose-rich foods are at increased risk of the development of obesity and NASH. Studies in rodent models to evaluate the influence of fructose on NAFLD development in rodents have used various techniques including the addition of high-fructose (HFr) corn syrup to water, which has been used in combination with HFD feeding (92) or with fructose being directly added in the context of a pelleted diet with 60–70 kcal% carbohydrate as fructose (93). In C57BL/6 mice, an HFr diet with 60 kcal% as fructose and 10 kcal% fat can drive more liver TG, steatosis, and inflammation than a low-fat matched control diet with corn starch and also an HFD with 45 kcal% fat (mainly lard) (93). These changes were accompanied by increased lipogenesis and VLDL production, antioxidant pathways suggesting oxidative stress (Nrf2, Nuclear factor erythroid 2-related factor 2), inflammation markers (IL-1β, intercellular adhesion molecule-1), and metabolic disorders such as increased plasma lipids and glucose intolerance relative to those fed the corn starch–based control diet and HFD. Findings obtained with C57BL/6 mice fed an HFD (58 kcal% fat, hydrogenated coconut oil) or high-fat, high-fructose (HFHFr) diet (combination of the same 58 kcal% fat diet and HFr corn syrup in water) for 16 wk suggested that both HFD and HFHFr diets increased body weight, IR, and hepatic steatosis similarly (92). However, only the HFHFr diet allowed for further increases in hepatic oxidative stress and the progression from liver fat deposition to inflammation, transforming growth factor-β1-driven fibrogenesis, and collagen deposition (92). Sodhi et al. (94) reported that an HFr diet (with 20 kcal% protein, 13 kcal% fat, 66 kcal% carbohydrate, 0.1% cholesterol, “high fructose”, but amount not reported) increased IR, blood pressure, markers of oxidative stress and lipogenesis, along with fibrotic markers in C57BL/6 mice fed this diet for 8 wk. However, it is not possible to determine to what extent the differences in cholesterol (only 0.02% in the control diet) or other diet background factors (30 kcal% protein, 57 kcal% carbohydrate in the control diet) contributed to the observed effects.

In the rat model, an HFr (low-fat background, ∼60–70 kcal% fructose) diet may be more potent than an HFD or HFHFr diet. Kawasaki et al. (43) found that an HFr diet (73% kcal fructose) increased hepatic TGs more than high-sucrose (73% kcal), HFD (40% kcal fat), or HFHFr diets (40% kcal fat, 41% kcal fructose) in Wistar rats. In addition, the HFr diet also promoted macrovesicular and microvesicular steatosis (score: 2.6–2.9) and lobular inflammation (score: 2.4); however, fibrotic scores were not significantly different from the control group. HFr diet (70 kcal% fructose) also induced NAFLD/NASH in the SD rat model (compared with a grain-based diet) (44), including increased steatosis (score: 3), inflammation (score: 2.13), hepatic ballooning (score: 1.75), and fibrosis (score: 5.25). In this study, pericellular and perivenular fibrosis were scored separately on a range of 0–4 and then added to depict the final score (range: 0–8) (44).

Similarly to rats, an HFr diet with 60 kcal% fructose increased plasma and liver TG more than diets containing the same amount of corn starch or sucrose in male golden Syrian hamsters fed for 7 wk. In this study, the researchers also found that HFr increased body weight, adiposity, and plasma insulin and reduced glucose disappearance (glucose tolerance test) more potently than in the high-sucrose- or starch-fed groups (95).

High-Fat, High-Fructose, High-Cholesterol Combination Diets

As suggested in earlier sections of this review, dietary cholesterol is a critical factor in the progression of NASH and hepatic inflammation in animal models (76, 96–98). In C57BL/6 mice, a high-cholesterol diet (HCD) on a low-fat diet background (1% cholesterol, 11 kcal% fat) elevated hepatic cholesterol esters and both the HCD and an HFD (33 kcal% fat, added cocoa butter) increased steatosis and mild steatohepatitis (99). However, the combination of the high-fat content and cholesterol amount (HFHC diet) interacted synergistically to drive features of NASH because moderate steatohepatitis and mild fibrosis were observed only when 1% cholesterol was given in conjunction with an HFD. In addition, mice fed the HFHC diet allowed for greater elevations in weight gain, hepatic lipid accumulation, serum ALT concentrations, and fibrosis, and decreased adiponectin concentrations compared with animals fed a low-fat (11 kcal% fat) diet, HFD (33 kcal% fat), or HCD (11 kcal% fat, 1% cholesterol) (99).

In an effort to drive more NASH, some have added cholic acid in the context of an HFHC diet, which has been traditionally used in atherosclerosis studies in rodent models (100). In a recent study, a diet with 60 kcal% fat as mainly lard with 1.25% cholesterol and 0.5% cholic acid (called an “atherogenic diet”) caused increased steatosis and fibrosis after 12 wk in male C57BL/6 mice; however, the extent of fibrosis was lesser than in those fed an MCD diet with 20 kcal% fat as corn oil (101). Although the addition of cholate in combination with higher amounts of fat and cholesterol can drive more liver TG and cholesterol accumulation, NASH, and fibrosis (102), caveats include improved insulin sensitivity and glucose tolerance and reduced liver TGs and weight gain (100, 102). The reduction in these metabolic disorders by cholic acid was independent of the presence of dietary cholesterol, which was observed in mice fed a very-high-fat diet with 60 kcal% fat as safflower oil plus 0.5% cholic acid relative to those fed the same diet without added cholic acid (102).

When C57BL/6 mice were fed a diet containing 40 kcal% fat (of which ∼18% was trans fat), 22% fructose, and 2% cholesterol, they developed 3 stages of NAFLD (steatosis, steatohepatitis with fibrosis, and cirrhosis) as assessed by histological and biochemical methods (e.g., increased collagen content, collagen 1α1 protein, and mRNA expression of collagen-1α2 and MCP-1) in ∼30 wk as shown by Clapper et al. (45) and by Trevaskis et al. (103). Animals on this diet also demonstrated metabolic dysfunction as evidenced by increased total cholesterol, fasting insulin, and HOMA-IR, and lower adiponectin. This diet formulation is widely known as the AMLN diet (D09100301) and was a preferred diet by many researchers because it caused liver damage similarly to a mechanism that occurs in the human condition. The results of the AMLN diet are similar to the ALIOS (American Lifestyle-Induced Obesity Syndrome) model used by Tetri et al. (104) and show that trans fats in the diet could potentially play a role in worsening NASH. The AMLN diet has also been shown to induce NAFLD/NASH in SD rats similarly to mice; however, in the rat model, the NAFLD/NASH development (steatosis, inflammation, ALT) seems to happen in a relatively shorter time frame (46, 105) (∼16–20 wk).

Although the AMLN diet was widely popular, the trans fat source, Primex, a partially hydrogenated fat source (palm oil and partially hydrogenated soybean oil) containing trans fat (∼25% of fat), was banned by the FDA for use in foods in 2018. Alternative fat sources are currently being evaluated and preliminary evidence suggests that replacing Primex with palm oil (D09100310) in the AMLN diet drives a similar phenotype in C57BL/6 mice (106, 107). Researchers have also used other replacement diets, such as a modified AMLN diet with non–trans fat Primex, which contains palm oil and fully hydrogenated soybean oil (D16022301), or with trans fat from corn oil shortening (D16010101), or a mixture of both (108), and observed similar (and in some cases more advanced) NASH development compared with the AMLN diet in C57BL/6 mice (47). Using the non–trans fat Primex worsened IR in the animals, which to our understanding has not been well characterized in the AMLN diet (47). This has also been observed when Primex shortening was replaced with palm oil (D09100310) (48). The palm oil–based diet, also known as the GAN diet, showed remarkable similarity to the AMLN diet in liver histology. Steatosis scores were 3 in both groups, and scores of inflammation (mean: 2.5) and fibrosis (range: 1.6–1.9) were also quite similar between animals in both these diet groups (48, 106). The ability of these diets to drive symptoms of metabolic disease and all stages of NASH in rodent models provides individual researchers with a more human-like model of disease development, and contract research organizations and breeders with a robust commercially available NASH model for drug discovery.

In golden Syrian hamsters, adding 0.05%–0.25% cholesterol to an HFHFr diet (30% fat, 40% fructose, 6 wk) increased liver TG and cholesterol concentrations (109). Addition of cholesterol to an HFHFr diet also worsened glucose tolerance and insulin sensitivity in these animals and this was dose dependent. These data clearly implicate dietary cholesterol as exacerbating the effect of dietary fat and fructose and as a major determinant of the severity of metabolic disturbances in the hamster model, similarly to rats and mice.

Dietary NAFLD-Induced HCC

NASH, if unchecked, can progress to cirrhosis and can be complicated with HCC (110). Although most HCC studies involve chemically induced models, some dietary models of NASH have also shown progression toward HCC when fed for chronic periods. Wolf et al. (38) used a CD diet on an HFD background (45% kcal fat, mainly lard, D05010402) and found that along with steatosis and NASH, some C57BL/6 mice also progressed toward HCC after 1 y. The tumor incidence in HFD mice was only 2.5%, as compared with 25% (19/75 mice) in CD-HFD mice (38). In another study, C57BL/6 mice fed a choline-deficient amino acid–defined diet (CDAA as previously described) developed liver injury with biochemical features of NASH and this led to HCC. Feeding with CDAA induced IR (in ∼1 mo), hepatic steatosis (∼3 mo), inflammation (∼3 mo), and liver damage and fibrosis (∼3–6 mo). HCC developed after 9 mo of feeding (∼35% of the animals) that increased to 100% when given in combination with CCl₄ (34). Using the diet A06071302 (0.1%-methionine, no-choline HFD) in C57BL/6 mice, Ikawa-Yoshida et al. (29) observed NASH/HCC in around the same time frame (36 wk); however, others have found this same diet can drive HCC in a shorter time frame (24 wk) in 8 of 10 C57BL/6 mice (80). NASH leading to HCC development was also observed in DIAMOND mice (Diet Induced Animal Model Of Non-alcoholic fatty liver Disease) (111) consuming a Western diet (typically high-fat: ∼40–45 kcal% fat; high sugar, ∼30%; plus cholesterol at ∼0.2–1%) along with sucrose in water. This model has been shown to develop steatosis in 4–8 wk, NASH in 16–24 wk, and HCC at week 52.

Other Models of Fatty Liver Disease

Nondietary models of fatty liver disease mainly fall into 2 categories: 1) genetic models and 2) chemical models of liver disease induction. Some of the genetic models used earlier include the MATO mice, which have the gene methionine adenosyltransferase-1A removed, and mice with liver-specific phosphatase and tensin homolog (PTEN) deletion. Liver-specific PTEN deletion in mice also helps to study NASH and NASH-associated HCC [reviewed in (112, 113)]. Models which now are widely studied in concert with diet include the ob/ob mice (leptin deficiency), the db/db mice (leptin receptor deficiency), and the foz/foz mice (with a mutated Alms1 gene). These models, particularly the ob/ob mice, have been used in various studies along with HFDs (including the HFHFr diet with added cholesterol, and AMLN/GAN diets) to induce various aspects of NAFLD/NASH and other comorbidities mainly because of the advantage of development of a serious disease phenotype in a relatively short time frame. Chemical models include damaging the liver using chemicals such as streptozotocin, diethylnitrosamine, CCl₄, etc. either alone or in combination with a dietary model. These types of models have been extensively reviewed elsewhere (2, 4, 23, 112, 114).

Control Diets

When designing experiments with animal models, the importance of dietary choice is typically overlooked, especially if diet is not the main focus of the study. The aforementioned diets (as are most used in NAFLD research) are all purified diets, formulated with refined ingredients, and the formulas are “open” to researchers so that they (and those that wish to repeat their work) are aware of the actual amount of each ingredient in the formulation (115, 116). The refined ingredients of purified diets allow complete control over their nutrient compositions (117), which provides an ability to manipulate the nutrients one at a time, critical to formulating diets such as the MCD or CD diets, and modifications of these diets; however, when making modifications, the choice of control diet becomes important (115). Some researchers use grain-based diets (typically available in their animal facilities) as a “control diet” for purified diets. Rather, the use of cereal grains (e.g., ground corn) and animal byproducts (e.g., fish meal) in grain-based diets, which are in stark contrast to the refined ingredients used in purified diets, makes it impossible to determine what is driving NAFLD [see references (115–117) for more information]. The lack of concern regarding the choice of control diet can be seen in the methods sections of many publications, where it is common to read vague terms such as “standard chow” or “regular diet” which do not provide readers with any useful information and typically refer to grain- or cereal-based diets (117). Instead, a defined, purified diet which only differs in the ingredients that are driving NAFLD in the experimental diet should be chosen as the control to allow for proper data interpretation (117). In this review, we cited research using purified control diets in most cases; however, in some cases, we were limited to only studies where grain-based diets were chosen or were unable to conclude what control diet was chosen because of poor reporting.

Conclusions

Owing to the ease of selectively modifying the components in purified diets, a variety of different options can drive NAFLD in rodent models and individual components of these diets can be selectively manipulated to “fine-tune” the phenotype to varying degrees of the NAFLD spectrum (see Tables 1, 2) through varying mechanisms (Figures 1, 2). Although some have limitations in representing humans with NAFLD, purified diets are powerful tools for studying the pathogenesis and progression of this disease and uncovering potential treatment targets. The combination of proper diet choice, matched control diet, and appropriate rodent model will continue to move this field forward toward developing novel prevention or treatment therapies along the NAFLD spectrum.

Notes

The authors’ salaries, office space, computing, and manuscript preparation were provided by Research Diets, Inc., New Brunswick, NJ.

Author disclosures: SR, J-YK, and MAP are employees of Research Diets, Inc., New Brunswick, NJ.

Abbreviations used: ALT, alanine aminotransferase; AST, aspartate aminotransferase; CD, choline deficient; CDAA, choline-deficient amino acid–based diet; CHOP, C/EBP homologous protein; FFA, free fatty acid; HCC, hepatocellular carcinoma; HCD, high-cholesterol diet; HFD, high-fat diet; HFHC, high-fat, high-cholesterol; HFHFr, high-fat, high-fructose; HFr, high-fructose; IR, insulin resistance/insulin resistant; MCD, methionine and choline deficient; MCP-1, monocyte chemoattractant protein 1; NAFLD, nonalcoholic fatty liver disease; NAS, NAFLD activity score; NASH, nonalcoholic steatohepatitis; PTEN, phosphatase and tensin homolog; SD, Sprague Dawley; TG, triglyceride.