A mini-review of the rodent models for alcoholic liver disease: shortcomings, application, and future prospects

Abstract

Rodents are the most common models in studies of alcoholic liver disease (ALD). Although several rodents ALD models have been established and multiple mechanisms have been elucidated based on them, these models have some non-negligible shortcomings, specifically only inducing early stage (mainly steatosis, slight to moderate steatohepatitis) but not the whole spectrum of human ALD. The resistance of rodents to advanced ALD has been suggested to be due to the physiological differences between rodents and human beings. Previous studies have reported significant interstrain differences in the susceptibility to ethanol-induced liver injury and in the manifestation of ALD (such as different alteration of lipid profiles). Therefore, it would be interesting to characterize the manifestation of ethanol-induced liver damage in various rodents, which may provide a recommendation to investigators of ALD. Furthermore, more severe ALD models need to be established for the study of serious ALD forms, which may be achieved by using genetic modified rodents.

Introduction

Alcoholic liver disease (ALD), a progressively aggravated liver disease with a wide spectrum ranging from reversible steatosis to irreversible fibrosis and cirrhosis, remains to be one of the leading causes of alcohol-related death in both developed and developing countries [1–3]. The investigation of the hepatotoxicity of ethanol can date back to 60 years ago, when the natural aversion of rodents to ethanol was overcome and high blood alcohol concentration (BAC) was achieved by incorporating ethanol into the liquid diet [4]. Since then, a huge number of studies have revealed multiple mechanisms underlying the onset of ALD such as the toxicity of acetaldehyde, reactive oxygen species (ROS), mitochondrial dysfunction, activation of cytochrome P4502E1 (CYP2E1), disturbance of lipid homeostasis, endoplasmic reticulum stress, and the gut/adipose-liver axes [5–14]. Unfortunately, effective therapeutic strategy for ALD is still lacking, which is at least partially due to the absence of ideal animal models. Although primary hepatocytes and various liver cell lines have been used as in vitro models, they could not mimic the crosstalk between hepatocytes and other intrahepatic cells (such as Kupffer cells, KCs) and the communication between liver and gut/adipose. Currently, rodents are the most commonly used animals in ALD study. However, the available rodent models mainly produce early stage of ALD (mainly steatosis, slight to moderate steatohepatitis). This mini-review focuses on the current available rodent ALD models in which ethanol is delivered via gastrointestinal tract, and discusses the limitation and application of these rodent models. Although some studies achieve high BAC by utilizing other methods such as inhalation, intravenous and intraperitoneal injection, these models are not discussed here as they fail to mimic the natural alcohol drinking pattern of human ALD patients.

ALD: a progressive liver disease with complicated mechanisms

Ethanol is primarily metabolized into acetaldehyde and then to acetate in hepatocytes by cytosolic alcohol dehydrogenase (ADH) and mitochondrial acetaldehyde dehydrogenase (ALDH), respectively. Acetaldehyde, a well-known hepatotoxicant, is considered as one of the principal culprits of ALD, as it could binds nonenzymatically to free amino groups in the proteins of the liver cells, leading to the functional impairments of key proteins [15–17]. ALD is also closely related with the inducible CYP2E1, which has been demonstrated by using CYP2E1 inhibitors and Cyp2e1 ablated/knockin mice [14, 18–21]. In addition to the direct effects on hepatocytes, the activation of KCs and hepatic stellate cells (HSCs) contribute to the inflammation and fibrosis/cirrhosis in ALD. Specifically, ethanol abuse leads to the disturbance of intestine microbiome and impairment of the gastrointestinal mucosa, favoring the translocation of gut lipopolysaccharide (LPS) to liver followed by the activation of KCs and the release of proinflammatory cytokines and chemokines [22]. Elimination of KCs or blocking the Toll-like receptor (TLR)-mediated signaling suppressed ethanol-induced liver damage, demonstrating the crucial roles of innate immune mechanism in ALD [23–28]. HSCs, the major producers of extracellular matrix in fibrotic liver, could be activated by proinflammatory cytokines, profibrotic cytokines, danger-associated molecular patterns (DAMPs) released by injured hepatocytes, and ROS [6, 9, 12]. The accumulation of collagen deposition will lead to the portal tract-septal fibrosis and eventually result in the formation of fibrous septate and scar tissue [29]. Furthermore, other intrahepatic cells such as invariant natural killer T (iNKT) and extrahepatic cells including adipocytes may be also involved in the development of ALD [30, 31]. A brief scheme of pathogenesis of ALD is presented in Figure 1.

Figure 1.

A mechanism scheme of alcoholic liver disease (ALD). The onset of ALD is initiated by the metabolism of ethanol in hepatocytes, which could lead to production of toxic intermediate acetaldehyde, the overproduction of reactive oxygen species (ROS), and mitochondrial dysfunction. The impairment of the intestinal mucosa and disturbance of gut microbiome lead to translocation of gut lipopolysaccharide (LPS) to liver, resulting in the activation of hepatic resident macrophage, Kupffer cells. The activated Kupffer cells release a great amount of proinflammatory cytokines and chemokines, inducing the infiltration of peripheral macrophages and neutrophiles. The activated hepatic stellate cells (HSCs) produce a great amount of extracellular matrix (ECM). Furthermore, ethanol consumption results in the decline of adiponectin secretion, further aggravating liver damage.

Rodents utilized in ALD study: species, strains, and gender differences

Rats and mice are the most popular rodents in ALD studies, whereas hamsters are used in some earlier studies [32, 33]. Almost all the available strains of rats (including Sprague–Dawley (SD) rats, Wistar rats, Long–Evans rats, and Fischer-344 rats, Brown Norway rats, and Lewis rats) and mice (such as C56BL/6, BALB/c, ICR, 129S1/SvlmJ, C3H) have been used, and significant interstrain differences in sensitivity to ethanol-induced liver injury has been reported. One study showed that Long-Evans rats were more susceptible to ALD compared with SD and Fischer-344 rats after receiving same volume of liquid diet for 8 weeks, which was thought to be due to the differences in expression of ethanol metabolizing enzymes [34]. Interestingly, another study found that ethanol induced mainly pericentral lipid deposition pattern in Fisher-344 rats (similar to human ALD) but hepatic midzonal steatosis in Wistar rats, suggesting that Fischer-344 rats were better suited for lipid studies in early stage of ALD [35]. In regard with mice, C57BL/6 mice were found to consume highest volume of ethanol voluntarily compared with other 21 inbred mice under normal circumstances (>10 g/kg/d) [36]. Another study compared the sensitivity of 14 inbred strains of mice (129S1/SvImJ, AKR/J, BALB/cJ, BALB/cByJ, BTBR T + tf/J, C3H/HeJ, C57BL/10J, DBA/2J, FVB/NJ, KK/HIJ, MOLF/EiJ, NZW/LacJ, PWD/PhJ, and WSB/EiJ) using an intragastric intubation model, and found profound interstrain differences in ethanol-induced steatohepatitis in spite of consistently high urine ethanol level [37]. This study revealed that NZW/Lacj was most susceptive to ethanol-induced liver injury, whereas Wsb/Eij was most resistant [37]. However, C57BL/6 strain is not involved in this study, although it is the most popular strain used in ALD studies. Apparently, much more works are needed to identify the optimal strains of rodents for the study of ALD.

Epidemiological studies and animal studies have demonstrated that females are more susceptible to ethanol-induced liver injuries than males [38–41]. Although the exact mechanisms remain to be elucidated, available evidences suggest that gender differences in ALD susceptibility may be related with the differences in bioavailability of ethanol, the sex hormone levels, and the activation status of Kupffer cells between male and female drinkers [42–45]. Although female rats are more susceptible to ALD, both male and female rodents have been used in previous studies. Interestingly, female C57BL/6 mice were found to be less sensitive to the high fat-plus-binge-induced liver injury than the male counterparts, probably because female mice gained much less body weight post high-fat feeding [46].

Binge drinking model

One “binge” is defined as consumption of five and four drinks for men and women, respectively, in 2 h to achieve BAC over 80 mg/dl by the National Institute on Alcohol Abuse and Alcoholism (Bethesda, MD, USA) [47, 48]. One single binge model (4–6 g/kg, by gavage) developed by Carson and Stephen could mimic blood alcohol levels, behavioral effects, and physiological changes in human binge drinkers, and has been wildly used [49–53]. In a more commonly used multiple binge model, three doses of ethanol bolus (usually 5 g/kg) lead to marked increase of serum aminotransferase activities, hepatocyte apoptosis, liver steatosis, liver lipid peroxidation, induction of CYP2E1, and increased production of hepatic tumor necrosis factor α (TNF-α) [54–58]. In addition to the dose of ethanol and the number of binges, several other critical factors may influence the effects of ethanol such as the concentration of ethanol solution and the time points of sample collection [48]. Ethanol of higher concentration and fasting before ethanol gavage would lead to damage of gastrointestinal tract, more severe liver damage, and even mortality of mice [49, 59]. Interestingly, previous studies have reported biphasic effect of acute ethanol/binge drinking on the activation of KCs and the innate immune response, which might be associated with the time of examination [22, 60, 61]. Therefore, the time point of sampling is very important, especially when the aim of study is to investigate the effects of ethanol on innate immune system. Furthermore, maltodextrin is usually selected as a control solution which could not be neglected as it could provide isocaloric intake in control animals.

Chronic rodent ALD models

Voluntary drinking model

Rodents are allowed free access to chow diet and ethanol-containing drinking water (0–40% (v/v)) to accurately simulate human drinking pattern. However, a pioneer study showed that rats received 15% (v/v) ethanol-containing water for 177 days had no obvious liver injury [62]. Similarly, mice exposed to 10–20% ethanol-containing drinking water only developed slight steatosis [63, 64]. Interestingly, moderate ethanol treatment through drinking water exerted beneficial effects on nonalcoholic fatty liver in mice fed a high-fat diet [65]. The absence of liver toxicity of ethanol is attributed to the lower BAC due to the natural aversion of rodents to ethanol and the higher rate of ethanol metabolism compared with human being [66–68]. Although this “voluntary drinking” mode could induce damage after long time treatment with high ethanol concentration [69–71], it cannot exclude the contribution of malnutrition in the toxicity observed in ethanol group.

Lieber–DeCarli liquid diet model

The liquid diet model established by Lieber and Decarli is a breakthrough in the study of ALD. This elegant model is designed to overcome the aversion characteristic of rodents to ethanol by feeding animals only with ethanol-containing liquid diet. Typically, the control liquid diet is composed of protein (18% of total of calories), fat (35% of total calories), carbohydrate (47% of total calories), vitamins and salts mixes, whereas 36% of total energy provided by carbohydrate is replaced by ethanol in ethanol-containing liquid diet [72]. The animals are fed with gradually increased ethanol-containing diet (from 1% to 5%, w/v) during 1-week period, and then with 5% ethanol diet for 4–8 weeks. The daily ethanol intake could reach to 12–18 g/d for rats and 24–34 g/d for mice [72, 73]. The amount of liquid diet given to animals in control group is equal to those ingested by the ethanol-treated littermate, and thus accurately control energy balance between ethanol-exposed rodents and the control animals [72, 74–76]. The advantage of this model is easy handling, inexpensive, short time-consuming, low mortality of animals, and convenient (liquid diet commercially available), and thus applicable for all laboratories. However, no serious liver injury was induced in this model, which might be attributed to the low BAC levels possibly due to the long time needed to consume the ethanol-containing diets (eating slowing) [77]. Anyway, this model provides a useful tool for studying the early stage of ALD.

Tsukamoto–French intragastric infusion model

Tsukamoto and French developed a rat ALD model by directly injecting alcohol and nutrients via a cannula inserted in the stomach [78, 79]. This model achieves high BAC level, as rodents passively receive ethanol-containing liquid diets and investigators can exactly control the diet balance between control and ethanol-fed animals [80]. Ethanol delivered using this method could induce steatosis, megamitochondria, apoptosis, central lobular and pericellular fibrosis, portal fibrosis, bridging fibrosis, central necrosis, and mixed inflammatory infiltrate, which closely resembled human ALD [77]. Although initially developed in rats, the intragastric infusion model has been successfully established in mice [81, 82]. The advantage of this model is that, for the first time, the control and ethanol-fed animals were in exactly the same state of nutrition, as the diet and ethanol intake are completely under the control of investigator [83]. As higher BAC level and more severe liver injuries can be achieved, this model can serve as a useful tool for studying advanced ALD. Furthermore, this model provides a method for the study of multifactorial liver disease, such as the synergy or antagonism between the environment/nutrients and alcohol. However, the application of this model is restricted due to complicated operating techniques, difficulty in postoperative animal health maintenance, and expensive equipment.

Chronic-plus-binge model

Binge drinking after chronic ethanol consumption is one of the important factors contributing to the progression of steatosis to steatohepatitis. Chronic-plus-binge model simulates the “long-term drinking history and recent alcoholism” drinking pattern observed in ALD patients. Aroor et al. developed a chronic-plus-binge rat model in which chronic ethanol-containing (5%, w/v) liquid diet feeding rats were gavaged with single dose of ethanol (5 g/kg) [84]. A similar model was established in mice by Gao group, and named as NIAAA model or Gao–Binge model. In the Gao–Binge model, the ethanol group mice receive an alcoholic liquid diet for 10 days followed by an acute ethanol gavage (5 g/kg), and sacrificed 9 h later for liver injury examination [85]. In both models, single binge significantly increased BAC level (from 100 mg/dl to 175 mg/dl in rats, and from 180 mg/dl to 400 mg/dl in mice) and augmented liver injuries. Extension of chronic ethanol feeding period or multiple binges resulted in more serious neutrophile infiltration and aggravated liver damage [84, 85]. The advantage of this model is flexible and easy to operation, suitable for exploring the pathological mechanism of hepatitis.

High fat-plus-binge/chronic drinking models

High fat-plus-binge/chronic drinking models were developed based on the concept that obesity could exacerbate the hepatotoxicity of ethanol. Tsukamoto et al. established a high-fat diet feeding and intragastric ethanol infusion model by feeding mice 2 weeks of high-fat diet (1% cholesterol, 20% calories from lard, 17% calories from corn oil) followed by intragastric infusion of ethanol (27 g/kg/d) and high-fat liquid diet (60% of total calories) for 8 weeks. The results showed that high-fat diet dramatically enhanced BAC level in ethanol-fed mice (90 mg/dl and 340 mg/dl in chow diet plus ethanol group mice and high-fat diet plus ethanol-fed mice, respectively) and significantly augmented liver injury (15-fold increase of alanine aminotransferase levels and severe steatohepatitis, perisinusoidal and pericellular fibrosis) [86, 87]. High fat-plus binge drinking model was established and characterized in a following study, in which mice were fed with a high-fat diet before exposure to a single binge or multiple binge (5 g/kg). It was found that as little as 3 days of high-fat feeding could significantly exacerbate single binge-induced neutrophilia, hepatic neutrophil infiltration, and liver injury [88]. The deleterious effects of ethanol on liver could be aggravated if the feeding time of high-fat diet was extended to 3 months [46, 88]. These models serve as useful tools to study the synergistic effect of a high-fat diet and alcohol on liver injury. The detailed factors affecting chronic-plus-binge model have been reviewed recently [46]. One specific point needed to pay attention is that binge drinking (5 g/kg bw) will lead to high mortality in mice of higher weight, which can be avoided by reducing ethanol dose or shortening the period of chronic high-fat feeding (reducing the weight of mice) [46].

“Second or multiple hit” model

Ethanol with a “second hit” (another hepatotoxicant such as LPS, carbon tetrachloride, diethyl nitrosamine) could achieve more severe liver damage, providing a model for the study of serious lesions in the end-stage ALD [89]. However, it is obvious that certain differences exist in the pathological mechanisms between liver damage induced by ethanol per se and those by combination of ethanol and a second hepatotoxicant.

The shortcomings of available rodent ALD models

The primary shortcoming of the above rodent ALD models is that they all fail to cover the whole spectrum of human ALD. Even the aversion of rodents can be overcome by incorporating ethanol into liquid diet or by gavage, however, most of these models only induce early stage of ALD. For example, none of the above ALD model could develop hepatocellular carcinoma (HCC), although ethanol is classified as group 1 carcinogen (known to be carcinogenic to humans) by the International Agency for Research on Cancer (IARC) [90]. Although early HCC could be induced by combing ethanol and other hepatotoxicants (such as diethyl nitrosamine, carbon tetrachloride, and urethane) [91–94], the molecular and cellular mechanisms observed in these models cannot be directly ascribed to ethanol per se [95].

The resistance of rodents to advanced ALD may be related with the physiological differences between human beings and rodents. Firstly, the catabolic rate of ethanol in rodents is significantly higher than that in human beings, suggesting that higher BAC level is needed to produce similar liver injuries in rodents as observed in human beings [96]. Second, the innate immune system, which plays critical roles in pathogenesis of ALD, has been demonstrated to vary significantly between mice and human [96]. Specifically, rodents are resistant to LPS challenge, which might be due to the lower proportion of neutrophiles in blood leukocytes [97, 98]. Thirdly, genetic factors has been suggested to be involved in the development of human ALD, as only about 30% of continued drinkers develop fibrosis or cirrhosis and about 5–15% of abstainers develop fibrosis [99].

Conclusion and future research perspective

Several rodent ALD models have been established for the studies of ALD mechanisms and the evaluation of potential medicines. The versatile Liber–DeCarli liquid diet model is convenient, flexible, and applicable for most laboratories, whereas the NIAAA model could mimic the “chronic-plus-binge” pattern of human ALD. However, all ALD models have drawbacks and specifically only induce early stage of ALD, which may be related with the physiological differences between human beings and rodents. Although the Tsukamoto–French intragastric infusion model can cause aggravated liver damage such as fibrosis, its application is limited due to complicate operation, expensive equipment, time-consuming characteristic and difficulty in postoperative animal health maintenance. The “second hit” model can induce more severe liver damage (liver fibers, cirrhosis, and liver cancer), but the addition of another hit makes it hard to ascertain the contribution of ethanol per se in the onset of liver injury (Table 1).

Table 1.

Summary of common rodent models of alcoholic liver disease

Models	Characteristics and application	Special attention and disadvantage	Phenotypes of ALD
Binge drinking model	Mimic the binge drinking pattern of human	The time point of sampling; control solution.	Elevation of ALT and AST; steatosis; mild inflammation
Voluntary drinking model	[1] Simulation of human drinking pattern [2] Simple with low mortality rate [3] Inexpensive	Natural aversion of rodents cannot be overcome; unbalanced nutrition status between control and ethanol group rodents	Mild elevation of ALT and AST; usually no severe liver injuries
Lieber-DeCarli liquid diet model	[1] Simulation of human drinking pattern [2] Easy to control effect of nutrients [3] Overcoming the aversion of rodents to alcohol [4] Flexible application [5] Time and cost efficient [6] High blood alcohol concentration	No fibrosis and end-stage injuries	Various degrees of steatosis; mild inflammation
Tsukamoto–French intragastric infusion model	[1] Nutrition balance between pair-feed animals [2] Overcoming the aversion of experimental animals to alcohol [3] Flexible application	Complicated operating technique; difficulty in postoperative animal health maintenance; expensive equipment	Steatosis; inflammation; mild fibrosis; focal liver necrosis
Chronic-plus-binge model	[1] Mimicking the chronic-plus-binge drinking pattern [2] Flexible application [3] Time and cost efficient [4] High blood alcohol concentration	No fibrosis and end-stage injuries	Steatosis; inflammation
High fat-plus-binge model	Mimicing the deleterious effects of ethanol in obesity population	High mortality in overweight mice	Steatosis; inflammation
“Second hit” model	Induction of end-stage liver injuries	Difficulty in analysis of experiment result	Advanced liver injury (cirrhosis, hepatocellular carcinoma)

Future researches on ALD models may focus on two aspects: mapping the manifestation of ethanol-induced liver damage in various rodents and establishing models of advanced ALD. Previous studies have suggested that rodents of different strains may have different sensitivity to ALD and discrepant alteration of lipid profiles after ethanol exposure [34, 35]. Thus, it would be interesting to map the manifestation of ethanol-induced liver damage in various rodents, which may finally provide a recommendation to investigators of ALD. Besides, more severe ALD models need to be established for the study of serious form of human ALD, which may be achieved by using genetic modified rodents. Mechanisms studies have suggested that CYP2E1 was responsible for oxidative stress, hepatotoxicity, and carcinogenic ethno-DNA lesions in ALD [11, 21, 100], whereas Aldh2 deficiency promoted alcohol-associated liver cancer [91]. Interestingly, people with a homozygous c2c2 genotype of Cyp2e1 (higher CYP2E1 activity) or with *2 allele of Aldh2 gene (decreased ALDH2 activity) were suggested to have increased susceptibility to ALD [91, 101]. Results of these studies suggest that genetic modified mice may serve as invaluable tools to explore novel mechanisms, develop diagnostic biomarkers, and screen potential medicines of advanced ALD.

Conflict of interest statement

None declared.