Acute liver failure (ALF) is a serious disease defined as destruction of parenchymal tissue accompanied by the impairment of protein synthesis and detoxifying function, reflected by the presence of jaundice, coagulopathy and liver encephalopathy within short period of time after first symptoms appeared. Liver dysfunction results from potentially reversible liver parenchyma necrosis which is most often caused by viral infection or acetaminophen overdose [1]. Despite reversibility of the disease, it is associated with a high mortality rate. Depending on the aetiology, in the mild course of ALF, it can be effectively treated causally as in acetaminophen overdose, where acetylcysteine can be administered as an antidote. In acute hepatotropic viral infections and different drug overdoses, there is no effective causal treatment, therefore symptomatic treatment is applied [2]. Patients with mild to moderate onset of ALF are usually treated in internal medicine or paediatric wards, where medical management includes prevention of liver failure complications, fluid therapy, and pharmacological liver supportive treatments, such as ornithine administration [3]. In fulminant or hyperacute course of the disease, the only therapeutic option is liver transplantation [4]. However, organ transplantation has important drawbacks including high cost, low organ availability, the necessarily of using life-long immunosuppression treatment and risk of multiple complications [1, 5, 6]. In view of the donor shortage, the need for immunosuppressive treatment, and post-transplant complications, there is imperative to develop a new alternative therapy capable of improving the clinical condition of patients, reducing mortality, or prolonging patient survival [7]. Stem cell-based therapy may be a promising alternative or adjuvant treatment to the currently used methods [8, 9, 10, 11]. Cell therapies with bone marrow mesenchymal stem cells (BM-MSCs) [12], Wharton’s jelly mesenchymal stem cells (WJ-MSCs) [13], hepatocyte-like cells derived from human amniotic epithelial cells (hAEC-HLCs) [14], and hepatocyte-like cells derived from induced pluripotent stem cells (iPS-HLCs) [15] have been proven effective in animal models of ALF. Implanted cells could partially take over the function of damaged liver parenchyma (e.g., hepatocyte implantation) [16], increase liver regenerative ability (e.g., mesenchymal stem cells implantation) [17, 18] or suppress destructive inflammatory reaction (e.g., amniotic epithelial and mesenchymal cells implantation) [19, 20]. It is also suggested that the mechanism of therapeutic action of the above-mentioned cells may be more related to the secreted paracrine factors and microvesicles than to the transplanted cells themselves [21, 22].

Experimental mouse and rat models of ALF using carbon tetrachloride (CCl${}_4$) and D-galactosamine (D-GalN) allow observation of the progression of liver diseases, particularly those observations that cannot be performed in patients for ethical and medical reasons [23, 24]. Therefore, animal models of induced hepatotoxicity can be used in the preclinical assessment of acute and chronic liver injury therapies. Unfortunately, the various mechanisms of xenobiotic toxicity and different sensitivity of animal species and strains to intoxication result in different intoxication effects and, as a consequence, some difficulties in data interpretation and extrapolation to humans [24]. Additionally, different approaches to liver injury assessment make it difficult to perform comparative analysis and choose the optimal in vivo model [23].

The main mechanism of liver injury by CCl${}_4$ involves its biotransformation resulting in free radical generation, i.e., CCl${}_3$, cell membrane oxidation, and DNA damage [25]. Free radicals are generated by cytochrome P450 (CYP) enzymes, in particular the CYP2E1 isozyme, which reduce CCl${}_4$ to CCl${}_3$ and then covert it to superoxides. The oxidative metabolites damage the cell membrane, leading to the cytoplasmic ion imbalance and necrotic cell death. Histological evidence of CCL${}_4$-induced liver damage can be observed mainly in zone 3 of the hepatic acinus [26]. The CCl${}_4$ model of liver injury is one of the most widely used, but unfortunately, in terms of its mechanism of action, it has no direct equivalent in human liver injury. Histopathological changes in acute CCl${}_4$ intoxication are very similar to acetaminophen overdose, which is one of the causes of acute liver injury in humans [27, 28]. To induce ALF in mice and rats, the CCl${}_4$ dose should be 40–750 $\mu$L per 100 g b.w. [29, 30, 31].

D-GalN is a highly hepatospecific compound. Unlike other hepatotoxins, D-GalN does not directly damage other organs and does not cause irritation when injected. In hepatocytes, D-GalN is eliminated by the pathway responsible for galactose metabolism. During the first phase, D-GalN is phosphorylated to galactosamine-1-phosphate (GalN-1-P), then converted to UDP-galactosamine, which has a higher affinity for UDP than for galactose. GalN-1-P is an inhibitor of UDP-glucose pyrophosphorylase-catalysed reaction. Together with UDP, the trapping effect leads to uridine deficiency and inhibition of RNA synthesis. As a consequence, inhibition of protein synthesis leads to apoptotic cell death [32]. In mice and rats, a single D-GalN intraperitoneal injection of 26.6 mg and 20 mg, respectively, changes gene expression related to injury but does not cause visible histopathological changes [33]. Rats are more susceptible to D-GalN intoxication than mice. To develop full-blown acute liver failure, D-GalN dose should be 80–140 mg in rats [34, 35] and 150–270 mg per 100 g b.w. in mice [36, 37]. Furthermore, the effect of D-GalN intoxication is highly dependent on the animal strain.

In the present pilot study, we compared the dynamics of development of acute liver failure induced in Sprague Dawley rat and BALB/c mouse experimental models by classical hepatotoxins, CCl${}_4$, and D-GalN at doses described as effective [38, 39, 40] or modified by us due to excessive mortality in CCl${}_4$-intoxicated mice. We adopted the following criteria of dose selection: signs of acute liver injury in histopathological examination and liver panel; low mortality rate; wide therapeutic window; good animal condition in therapeutic window. We analysed the usefulness of these models with a view to their future introduction into preclinical experiments on stem cell therapy. In particular, it allowed us to identify potential experimental time cut-off points based on histopathological and molecular changes and determine when stem cell therapeutic intervention could be effective as a first-line or adjuvant therapy in the inhibition of progression of these changes in ALF. It is important to define window of treatment in each ALF model, because too late or too early cell implantation could be ineffective in terms of cell homing and suppression of inflammatory reaction and may have negative influence on the final effect of cell therapy.

There is little information available in the literature describing the toxic effects, including histopathology and gene expression, induced by the classical hepatotoxins CCl${}_4$ and D-GalN in rat and mouse experimental models of hepatotoxicity at early time points in the context of potential cell therapy. In our study we focused on earlier time-points when the process of liver injury develops, and is not affected by spontaneous recovery of liver acini architecture and function. We wanted to determine experimental time cut-off points for an effective stem cell therapeutic intervention, that should be introduced before pathological changes become advanced or irreversible. Therefore, we analysed the dynamics of ALF development during 48 hours after intoxication with xenobiotic doses described as effective [38, 39, 40] or modified by us due to excessive mortality.

In rats treated with a single dose of 200 $\mu$L/100g CCl${}_4$, we observed early stages of hepatitis after 12 h, namely ballooning degeneration of hepatocytes, necrosis around central veins, and increased blood parameters. These microscopic observations were similar to those made by Janakat and Al-Merie [40]. The highest activity of aminotransferases and ALP was observed at 48 h after intoxication.

In mice, a dose of CCl${}_4$ that is effective in producing hepatotoxicity can be several times lower than in rats. Doses up to 750 CCl${}_4$$\mu$L/100 g b.w. were described in the literature [29]. High doses were verified by us in preliminary experiments due to mice excessive mortality (about 66%; not published). Finally, in mice treated intraperitoneally with a single injection of 50 $\mu$L/100 g CCl${}_4$, we observed rapid development of severe hepatitis with massive parenchymal necrosis in zone 3 of acinus and interface hepatitis, accompanied by highly elevated aminotransferase activities at 12 h. In mice injected with 50 $\mu$L/100 g CCl${}_4$, there was a massive liver necrosis observed up to 96 h [29]. At an even higher dose of CCl${}_4$ (100 $\mu$L/100g), Yang et al. [53] observed parenchyma necrosis as early as 6 h after intoxication and found the largest area of necrosis at 24 h.

A common characteristic of the changes occurring in rats and mice administered with CCl${}_4$ was the rapid progression of histopathological changes in the liver, concentrated mostly in pericentral zone of the liver acini, and signs of toxicity observed in the blood (12 h–24 h). Some of these changes as well as the proliferative activity of the liver cells continued an upward trend in the subsequent period up to 48 h after intoxication. On the other hand, it should be considered that the toxic effect of a single dose of CCl${}_4$ is not permanent in rats and mice and decreases with time after injection [29]. In animals treated with CCl${}_4$, 7 days after intoxication, we did not observe any significant pathological changes except of minor inflammatory infiltrate in mice, which indicates achieving spontaneous recovery in CCl${}_4$ model. Previous data indicated that liver regeneration in this model starts between 72 h and 120 h. Bizzaro et al. [29] observed up to 2-fold deacrese in necrotic area at 5th day in comparison to 3rd day of intoxication.

We observed similar trends both in mice and rats in the expression of some important liver genes, i.e., CYP2E1 and PPAR$\alpha$, after a single CCl${}_4$ injection. It is known that a strong inhibitory effect of CCl${}_4$ on cytochrome P450E1 begins from a few minutes [54] to 5–9 h after administration [55]. Inactivation of CYP2E1 during intoxication can be explained by the interaction between cytochrome P450E1 and its substrate - CCl${}_4$. The latter acts as a ‘suicide substrate’ that causes cytochrome degeneration and the formation of ubiquitin-conjugated microsomal protein [55, 56]. We observed this effect of the CCl${}_4$ action as a decrease in CYP2E1 gene expression from 12 h. Furthermore, we observed decreasing PPAR$\alpha$ gene expression at the same time. These changes are representative of liver cell injury because both CYP2E1 and PPAR$\alpha$ are important components of the hepatic metabolism of drugs and lipids, respectively [57]. Downregulation of PPAR$\alpha$ causes more severe steatohepatitis due to impaired lipid metabolism [58]. The changes in gene expression correlated with advanced histopathological changes in the pericentral areas of the liver acini, observed from 12 h to 48 h. Both CYP2E1 and PPAR$\alpha$ expression achieved control values on day 7.

We observed a similar pattern of changes (a decrease between 12 h and 48 h/an increase on day 7) induced by CCl${}_4$ in cMet and HGF gene expression in rats and, to a lesser extent, in mice (cMet). HGF is a ligand for the cellular MET (cMet) receptor and together they are responsible for proliferation, cell migration, morphogenesis, angiogenesis, and liver regeneration [59, 60]. Overexpression of cMet and HGF can be observed in liver tumours [61]. Our experiments only partially (increased HGF expression in CCl${}_4$-treated mice) corroborate previous results, where liver injury induced by CCl${}_4$ or D-GalN was associated with overexpression of HGF and c-Met [62, 63]. Upregulation of HGF gene expression was observed at the early stage of liver failure, and was followed by a drop-off at the late stage [62].

In rats, a dose of D-GalN that can be effective in achieving hepatotoxicity is relatively lower as compared to mice (50 vs 150 mg/100 g). We observed gradually progressing liver injury in rats after a single 50 mg/100 g D-GalN injection. Aminotransferase activity in the blood and apoptosis of liver cells increased gradually between 12 h and 48 h. Our findings on blood parameters are consistent with some other studies confirming that hepatocyte apoptosis is one of the earliest signs of liver injury in the D-GalN model [39]. Apoptotic cells were more numerous than in CCl${}_4$ model. In rats, this was observed in numerous hepatic cells as early as 6 h after 50 mg/100 g GalN injection [39].

In contrast to rat D-GalN model and the rat and mouse CCl${}_4$ model, in mice administered with 150 mg/100 g D-GalN, the first signs of liver injury, such as patches of ballooning degeneration and minor lymphocytic inflammatory infiltrates were visible only at 48 h, but not earlier. Another difference between the D-GalN mouse model and other tested models was that in mice the increases in aminotransferase activities were insignificant, which may indicate a considerably weaker hepatotoxic impact of the administered D-GalN on mice as compared to rats. Different aminotransferase activities observed after 150 mg/100 g D-GalN injection in our study and in some studies by other researches may be due to species-specific sensitivity [38, 64, 65]. In rats and mice treated with D-GalN, apoptotic activity of hepatocytes intensified gradualy between 24 h and 48 h. Although D-galactosamine is well known to induce apoptosis/necrosis by forming toxic metabolites, also other authors observed rather late apoptotic activity between 48 h and 72 h [64]. These results, together with spontaneous recovery at 7 day, strongly suggest that regeneration phase in this model is shifted to the later time points. Some authors indicate that after D-GalN injection regeneration phase starts at 72 h [66].

In addition to the delayed histopathological signs of liver injury induced in rats and mice by D-GalN, the expression pattern of CYP2E1 and PPAR$\alpha$ genes was partially (excluding PPAR$\alpha$ genes in rats) shifted to later time points. However, in general this pattern was similar to the corresponding CCl${}_4$ models and was characterized by decreasing expressions at 24 h and/or 48 h, reaching control values after 7 days. Our results are consistent with previous reports, showing that CYP2E1 expression levels decrease 36 h after intoxication and return to normal 1 week after intoxication [67]. Moreover, HGF gene expression was stable in rats and mice between 12 h and 7 days, whereas c-Met expression tended to increase after 48 h, which may be related to the regeneration of the injured liver in this model.

The duration of this short-term experiment was probably not long enough to observe advanced fibrotic changes in the livers of acutely intoxicated rats and mice. The time needed to induce fibrosis in CCl${}_4$ models usually varies from 4–12 weeks, therefore fibrosis-related gene expression and fibrosis-related histopathological changes at early time points after single hepatotoxin injection are not well described in the literature [68, 69]. The absence of fibrosis was expected and generally confirmed in this study by histopathological observations and molecular analysis. Nevertheless, as early as day 7, we observed a slight progression in liver fibrosis in histopathological sections, supported by increasing expression of the COL3A1 gene in mice intoxicated with CCl${}_4$. These data are consistent with some previous studies, where after a single or several injections of CCl${}_4$, the researchers obtained a self-limiting fibrosis model characterized by activation of HSCs and upregulation of fibrotic genes [70, 71].

The knowledge of ALF dynamics in animal models seems to be crucial in answering the question of when is the best time to intervene with cell therapy and which time cut-off points are most appropriate. Since liver damage progresses rapidly after CCl${}_4$ intoxication, there are basically two approaches to the timing of cell therapy intervention. Based on our findings and the results of some other studies, early intervention to 6 h–12 h seems to be optimal for preventive application of cell therapy [14, 72]. Therapeutic intervention during this time seems to be most effective due to the likely mechanism of suppression of the inflammatory process at its early stage. However, cell implantation shortly after intoxication may expose these cells to the toxic effects of the administrated xenobiotic. Due to the ability of CCl${}_4$ to create free radicals, the risk of interaction between CCl${}_4$ and stem cells is high, especially if the routes of cells and hepatotoxin administration are the same [73]. Unlike CCl${}_4$, D-GalN does not directly damage cells other than hepatocytes, so the risk of a negative interaction between D-GalN and the injected cells is lower.

It is very common for researchers to perform cell therapy 24 h or 48 h after CCl${}_4$ administration [74, 75, 76, 77]. However, after this period, we observed severe histopathological changes in mice and rats, corresponding to the acute phase of full-blown liver failure. Cell implantation at the above-mentioned time points could potentially increase the regenerative potential of the liver and shorten the recovery period. Cell administration between 24 h and 48 h seems to be more related to clinical practice, where therapeutic intervention coincides with the diagnosis of full-blown liver failure.

The D-GalN liver injury model is characterized by a longer time (especially in mice—up to 48 h) between toxin injection and the phase of full-blown hepatitis. This provides a wide therapeutic window for cell therapy intervention. In this model, cell therapy is usually conducted prior to development of full-blown liver failure, 24 h after intoxication [13, 78, 79].

In conclusion, we propose the following potential experimental time points appropriate for cell therapy intervention in animal models of CCl${}_4$ and D-GalN-dependent acute liver failure: rat and mouse CCl${}_4$ model—12 h, rat D-GalN model—24 h, and mouse D-GalN model—48 h. In the presented models, cell administration seems to be ineffective after 48 h due to the self-limiting properties of liver injury and spontaneous liver parenchyma regeneration after this time point.

CCl${}_4$, Carbon tetrachloride; D-GalN, D-galactosamine; ALF, acute liver failure; i.p., intraperitoneal; ALT, alanine transaminase; AST, aspartate transaminase; ALP, alkaline phosphatase; TP, total protein; Cas-3, Cleaved Caspase 3; TNF-$\alpha$, Tumor necrosis factor alpha; IL-6, Interleukin 6; COL1A1, Type I collagen; COL3A1, Type III collagen; TGF-$\beta$, Transforming growth factor beta; C-met, Tyrosine-protein kinase Met; HGF, Hepatocyte growth factor; CYP2E1, Cytochrome P450 2E1; PPAR-$\alpha$, Peroxisome proliferator-activated receptor alpha.

PC and MK designed the research study. PC supported the studies financially (grands), provided help and advice. MK, ŁL, EK, AS-S, EB, MH, MM, BS, AP, AG, JP performed the research. PC, MK, ŁL, AS-S, BS analyzed the data. PC, MK, AS-S wrote the manuscript. All authors contributed to editorial changes in the manuscript. All authors read and approved the final manuscript.

Animal experiments were approved by the Animal Experiments Ethical Committee of Medical University of Silesia, Katowice, Poland (decision no. 18/2018).

We thank the Silesian Analytical Laboratory (Katowice; Poland) for performing serum biochemistry analysis and blood morphology assessment. We thank Katarzyna Lorek from Students Scientific Society SUM Katowice for her participation in administration of xenobiotics to laboratory animals.

The studies were supported by institutional grands (SUM Katowice) no: KNW-1-103/N/8/0 and KNW-1-100/K/9/0.

The authors declare no conflict of interest.