Carbon tetrachloride is a powerful hepatotoxin which is used to induce liver fibrosis/cirrhosis through oral administration or injection (34, 35). CCl₄ induced liver fibrosis can be reproduced in both rats and mice, even in rabbits and dogs. Liver injury caused by repeated injection of hepatoxic reagents, such as CCl₄, would lead to the regeneration of hepatocytes, formation of fibrosis, and collapse of reticulin, and finally, result in liver architectural distortion and cirrhosis (34, 36). Besides, CCl₄ is relevant with cell metabolism, the dysregulation of cations such as Ca²⁺, Na⁺, and K⁺ in cells and the activation of cytochrome 450 (CYP450), which also plays an important role in inducing liver steatosis (37). Single injection of CCl₄ would result in acute hepatocytes damage and centrilobular necrosis (36, 38), which can be used to mimic acute hepatic insult for ACLF animal model. CCl₄ can be given in several different routes including subcutaneous, intramuscular or intraperitoneal injections, oral administration and inhalation (39–43).

Carbon tetrachloride is the most commonly used reagent to induce acute liver injury and liver fibrosis due to its convenience and low cost. CCl₄ induced liver fibrosis in mice can be developed in 6–8 weeks with continuous injection and is similar to clinical patients in pathophysiology. However, the hepatic fibrosis in mice induced by CCl₄ is easily to reverse, and CCl₄ would definitely cause damage to other organs. In addition, considering the toxicity and volatility of CCl₄, this reagent should be carefully used in fume cupboard.

D-Galactosamine is a powerful hepatotoxic reagent. Interfering with the uridine pool in the cell is the underlining mechanism of D-GalN in inducing liver injury. It induces lethal liver injury at large dose and would enhance the sensitivity of liver to LPS, an agonist of TLRs, playing synergetic liver damaging effects. Thus, D-GalN is widely used in combination with LPS in acute liver failure or endotoxemia animal models (44).

Thioacetamide, an indirect hepatotoxin, exerts toxic effect via a two-step biotransformation mediated mainly by CYP450 2E1 to thioacetamide sulfoxide and further to thioacetamide sulfur dioxide (TASO₂). TASO₂, the dominating reactive metabolite of TAA, leads to hepatic cellular damage, apoptosis and necrosis via oxidative stress and downregulation of catabolism enzymes (45). TAA is applied to induce acute or chronic liver disease in experimental animal models (46). It is reported that the main features of clinical chronic liver disease, such as hepatic encephalopathy, metabolic acidosis, elevated transaminases, abnormal coagulopathy, and centrilobular necrosis, could be induced after TAA administration (47). However, the carcinogenicity of TAA to humans (class 2B rating) limits its extensive use.

Acetaminophen N-acetyl-p-APAP, the most widely used antipyretic and analgesic drug, would cause severe liver injury even acute liver failure in the case of overdose in human (48). In mice, acute liver injury or failure can be induced following APAP overdose. Generally, at therapeutic dose, the majority of APAP will be metabolized in the liver to non-toxic metabolites (APAP-sulfate or APAP-glucuronide) and excreted via the bile and urine, whereas at toxic dose, the excess APAP will be oxidized in hepatocytes by CYP450 isoforms to highly toxic metabolite N-acetyl-p-benzoquinone imine (NAPQI) (49). The accumulation of NAPQI that causes hepatocellular necrosis and subsequent DAMPs secreted by damaged hepatocytes that activate innate inflammatory response eventually leads to acute liver injury/failure (50).

Concanavalin A is a lectin isolated from Jack beans (also called Canavalia ensiformis). Lectins are proteins that bind to carbohydrates, and the specific binding structures for Con A are α-Mannose and α-Galactose structures found in sugars, glycoproteins and glycolipids (51). Con A is a well-known T cell mitogen that can activate the immune system, recruit lymphocytes and elicit cytokine production (52). Unlike the hepatoxic reagents, Con A induced acute liver injury in mice is mainly based on the activation of CD4 + T cells and the subsequent secretion of proinflammatory cytokines, mainly IFN-γ and TNF. The mouse model of Con A induced liver injury is commonly adopted for investigating the mechanisms of autoimmune hepatitis (AIH) (53).

Human serum albumin, the most abundant serum protein in blood with a half-life of 19 days in humans (54), is a typical constituent of heterologous serum for murine. HSA is often used to develop the immunologic reaction induced chronic liver injury models in rats and mice (13, 55, 56). Immune mediated chronic liver injury induced by repeated administration of HSA would lead to typical liver fibrosis in mice or cirrhosis in rats. Subsequently, D-GalN plus LPS are administrated to establish ACLF model (56, 57). However, it is reported that the mortality of HSA administration during chronic liver injury or fibrosis-induction period is relatively high at 23% (56). The high mortality limits the application of HSA in establishing chronic liver fibrosis models.

Immune-mediated hepatic injury models are easily developed via the administration of heterologous serum constituent such as HSA. But the high mortality during the period of HSA induced chronic liver injury in murine models impels the usage of other kinds of heterologous serum. Porcine serum has been used to induce hepatic fibrosis for a long time, but the mechanism is uncertain until 1996. In order to investigate whether the hepatic fibrosis is caused by immune responses, Bhunchet et al. (58) divided rats into two groups, the porcine serum tolerant group and control group. Rats in the tolerant group had been injected with porcine serum peritoneally from the day of birth for 18 weeks while 8 weeks old rats in the control group received porcine serum injection for 10 weeks peritoneally. And antibody against porcine albumin level in the tolerant group is extremely lower than the control group, which suggests that no immune responses exist in tolerant group. Besides, no rats in the tolerant groups developed hepatic fibrosis. Based on this study, the mechanism of porcine serum induced hepatic fibrosis can be verified. Porcine serum is a suitable candidate for inducing the immune mediated liver injury models because of the low mortality reported (9). Compared with CCl₄, immune metabolism disorder is the basis of PS induced liver fibrosis, which mainly used to mimic the chronic liver injury caused by HBV infection or autoimmune liver diseases mediated cirrhosis (59–62). From histological perspective, the infiltration of monocytes and the formation of fibrosis around portal vein are the remarkable features of this model (60). PS induced immune mediated chronic liver cirrhosis demonstrates great popularity due to its economic efficiency and practicability.

Bile duct ligation is a typical surgical approach established since 1930s to simulate extrahepatic biliary obstruction that leads to biliary cirrhosis in rats or mice (63–65). The core procedure for BDL surgery is that rats or mice are subjected to double ligation of the common bile duct with section between the two ligatures, then hyperbilirubinemia would be mimicked in these BDL rats (64, 65). In BDL models, acute obstructive jaundice occurs and the expression of pro-inflammatory cytokines (such as TNF, IL-6, and IL-17) and pro-fibrotic proteins (such as collagen-α1, MMP-2, and TIMP-1) are induced in portal areas, which would progress to cirrhosis (66–68). Though liver inflammation and fibrosis are well displayed in the BDL models, the surgical procedures are difficult to handle that limits its wide application.

Lipopolysaccharide is the main component of the outer membrane of all Gram-negative bacteria, which is mainly consist of three parts, the lipid A (or endotoxin), a core phosphorylated oligosaccharide, and a variable specific long polysaccharide chain composed of repeating oligosaccharide (or O-antigen) (69, 70). LPS, one of the classical PAMPs, is a powerful mediator of systemic inflammation and septic shock via activating the PRRs-TLR4/TLR2 signaling pathways (4, 71–73). Normally, LPS first binds to LPS-binding protein to form an activated receptor complex with myeloid differentiation factor 2 (MD2), the CD14, and TLR4. Signals are transduced to intracellular proteins (MyD88, IRAKs, TRAFs, and NIK) by the activated receptor complex, generating an intricate network of cellular responses, activation of the NF-kB pathway, and secretion of a large amount of pro-inflammatory cytokines (74). Usually, LPS is co-administrated with D-GalN to induce acute liver injury models or fibrosis models in rats or mice which has been widely used and extensively studied (75, 76).

In order to investigate sepsis and sepsis-associated multiorgan failure, several experimental animal models with polymicrobial infection have been established to mimic the pathophysiological changes in septic patients (77). Cecal ligation and puncture (CLP) in murine is the most widely used and typical model for experimental sepsis which has been developed more than 30 years. Moreover, the CLP model is considered to be an ideal model for the induction of polymicrobial sepsis (77, 78). The surgical procedure features of CLP include midline laparotomy, ligation below the ileocecal valve, and needle puncture of the cecum (79). The severity of CLP model can be tailored via the ligation length of cecum and the needle puncture size.

Since the major problem for CLP-based polymicrobial sepsis model is consistency of the surgery, cecal slurry (CS) injection based polymicrobial peritoneal sepsis model is developed to solve the consistency problem and simplify the surgical procedure (80). CS-induced sepsis model is an infectious model with bacterial colonization, systemic inflammation and dose-dependent mortality without surgery, which is widely accepted and now considered as the “gold standard” model for murine neonatal sepsis study (80, 81). The advantages of CS-induced polymicrobial sepsis are no surgical procedures, a single CS donor can be administrated in a large number of animals, and easy to perform.

To study the role of liver during bacterial infection in different organs, several experimental bacterial animal models have been developed. For systemic single bacterial infection model, mice or rats are injected intraperitoneally with a certain dose [colony-forming unites, (CFU)] of K.P. or Escherichia coli (E. Coli) or Salmonella typhimurium (S. Typhimurium) directly (82–84). For the lung bacterial infection model, animals are given K.P. through a non-invasive intratracheal intubation (85).

Hepatotoxic reagents induced ACLF models are the most commonly used models and suitable for mimicking most clinical cases.

Immune responses induced ACLF models are mostly used to imitate autoimmune liver cirrhosis or hepatitis virus induced liver cirrhosis. The occurrence of autoimmune disease mainly due to the dysregulation of immune response, which will result in the breakdown of immune tolerance, and further, lead to the immune mediated organ or tissue damage caused by host itself. The mechanism of autoimmune hepatitis induced liver cirrhosis is still unclear, but there are several hypotheses may account for it. Molecular mimicry is a process that immune system responses to self-components which are similar to external pathogens such as HBV or hepatitis C virus (HCV). Autoantibody like anti-nuclear Antibody (ANA) and smooth muscle antibody (SMA) can be found in these patients, indicating that HBV and HCV may play an important role in autoimmune hepatitis induced liver cirrhosis (99). Besides, genetic factors may also participate in the occurrence of autoimmune hepatitis. Donaldson (100) revealed that major histocompatibility complex (MHC) is associated with autoimmune hepatitis to a large degree. However, no matter what the trigger is, a mass of activated inflammatory cells, especially the CD4 + T helper/inducer cells such as Th1, Th2, and Th17 cells, should be responsible for this immune mediated organ or tissue damage via the secretion of IL-2, IL-6, IFN-γ, and TGF-β.

Surgical procedure such as bile duct ligation is appropriate to mimic clinical cases suffer from cholestasis.

According to the current understanding and findings on the mechanism ACLF, the clinical course of ACLF could divide into three major stages, including chronic liver injury, acute hepatic/extrahepatic insult, and the excessive systemic inflammatory response caused by over-reactive immune system especially bacterial infection (14). However, due to the lack of optimal experimental animal model for ACLF, the progress of basic study on ACLF is limping. Though the above-mentioned experimental ACLF models were established, none of them can recapitulate and simulate the whole pathological process of ACLF patients.

Recently, we have developed a novel mouse model for ACLF combining chronic liver injury (injection of CCl₄ for 8 weeks, 0.2 ml/kg), acute hepatic insult (injection of a double dose CCl₄, 0.4 ml/kg), and bacterial infection (intraperitoneal injection of a single dose K.P., 1,000 CFU/mouse) (Figure 2), recapitulating the major clinical features of patients with ACLF worsened by bacterial infection (14). Moreover, this ACLF model includes chronic liver injury, acute hepatic insult, bacterial infection, renal injury, high short-term mortality, which could simulate the major pathological course of ACLF patients (14). To our knowledge, we introduced for the first time an easy double dose of CCl₄ injection as acute hepatic insult and a single dose of viable K.P. injection to mimic bacterial infection that occurred in most ACLF patients. In addition, systemic inflammatory responses induced by both PAMPs and DAMPs were fully simulated in this model. Importantly, the survival period of this ACLF model has been prolonged to 5–7 days after acute insult, which provide appropriate time for preclinical interventional researches, such as drug screening.

The establish process of this ACLF model was not go well in the beginning. In brief, we first combined CCl₄ injection with CLP surgery to test if an ACLF model could be developed. Repeated CCl₄ injection combined with CLP surgery could generate a model with high short-term mortality and sepsis like symptoms, however, no ALT or AST elevation was found even the mice died. Thus, when a double dose of CCl₄ was added as the acute hepatic insult in the chronic CCl₄ treated mice, followed by CLP surgery, an ideal ACLF model was established with the three major stages, including chronic liver injury (0.2 ml/kg, CCl₄ injection twice a week), acute hepatic (a double dose of CCl₄ injection) insult and polymicrobial infection (CLP surgery) (14). However, the defects of CLP-based ACLF model are obvious. First, it is hard to accurately control the spillage of cecal contents into the peritoneal cavity. Second, the surgical wounding has influence on the pathogenesis of the end stage liver failure. Third, the CLP surgical procedure is more time consuming to some extent.

To improve the CLP-based ACLF model, CS injection was adopted for the substitution of CLP induced polymicrobial infection. We prepared the CS solution according to a recently published protocol and administrated a suggested high dose of 200 μl/mouse in mice (121). However, no mortality was observed in mice injected with CS though the blood cultures showed positive results of bacteriotoxemia (122).

Subsequently, we turned to use single bacterial infection to replace the CLP or CS induced polymicrobial infection. Clinically, Escherichia coli (E. Coli) and K.P. are in the top rank of pathogens in ACLF patients (123). Different doses of E. Coli were first sent to the test. Surprisingly, mortality in mice was able to be observed after E. Coli injection till at a dose of 10⁸ CFU/mouse (14), which is too high to apply. Subsequently, different doses of K.P. were tested and an optimal mortality with appropriate survival period was found at a dose of 1,000 CFU/mouse (14). Meanwhile, Salmonella at a dose of 8,000 CFU/mouse was found similar results like K.P. (84). Therefore, a three-step ACLF model has been developed in mice, which could not only recapitulate the major three stages of ACLF, but also prolong the animal survival period with longer observation and interventional time for screening drugs and mechanism studies (Figure 2).