Establishment of a murine model of acute-on-chronic liver failure with multi-organ dysfunction

Acute-on-chronic liver failure (ACLF) is a distinct clinical entity with high probability of organ failure and mortality. Since patients generally present late, experimental models are needed to understand the pathophysiology and natural course of the disease. To reproduce the syndrome of ACLF, chronic liver disease was induced in C57BL6 mice (6–8 weeks; approximately 20–24 g weight) by intraperitoneal administration of carbon tetrachloride (CCl4) for 10 weeks followed by an acute injury with acetaminophen (APAP) and lipopolysaccharide (LPS). Blood, ascitic fluid, and organs were collected to study cell death, regeneration, and fibrosis. At 24 h post-APAP/LPS infusion, the liver tissue showed increased hepatocyte ballooning and endothelial cell TUNEL positivity. This was followed by progressive hepatocyte necrosis from perivascular region at day 7 to lobular region by day 11. ACLF (day 7 and day 11) animals showed increase in bilirubin (p < 0.05), prothrombin time (p < 0.0001), blood ammonia (p < 0.001), and portal pressure post-acute hepatocellular injury similar to human ACLF. Ascites was noticed by day 11 with median serum-ascites albumin gradient of 1.2 (1.1–1.3) g/dL. In comparison to cirrhosis, ACLF group (day 7 and day 11) showed significant decrease in Sirius red (p ≤ 0.0001), collagen1 (p < 0.0001), and a-SMA proportionate area (p < 0.0001) with loss of hepatocytes regeneration (p < 0.005). At day 11, ACLF animals also showed significant increase in serum creatinine (p < 0.05) and acute tubular necrosis suggestive of organ failure, compared to cirrhotic animals. The CCL4/APAP/LPS (CALPS) model of ACLF mimics the clinical, biochemical, and histological features of ACLF with demonstrable progressive hepatocellular necrosis, liver failure, impaired regeneration, development of portal hypertension, and organ dysfunction in an animal with chronic liver disease.


Introduction
Acute-on-chronic liver failure (ACLF) is an increasingly recognized clinical entity encompassing an acute liver failure in patients with underlying chronic liver disease or cirrhosis, resulting in cascade of event leading to sepsis, multi-organ failure (MOF), and death [1][2][3][4][5][6]. Clinical management is still challenging due to limited treatment options. Liver transplantation is the only curative treatment for such patients, which is, however, feasible in a small proportion of patients with ACLF [7].
One of the major challenges in the ACLF is the poor understanding of underlying pathophysiology leading to rapid progression of liver injury, sepsis and MOF. There is also a need to understand the dynamic course of hepatic injury with alterations in immune mediators and development of immune-paralysis in this disease state. Hence, a representative experimental model that mimics the course of human ACLF is crucial to decipher the pathophysiological journey of the disease and development of new therapeutic strategies for improved clinical outcomes for ACLF patients.
Several experimental models of ACLF have been reported [8][9][10][11][12] but none of them reproduces the clinical and pathological process of ACLF. In most models of ACLF, the chronic injury has been induced by carbon tetrachloride (CCl 4 ), bile duct ligation or porcine serum whereas the acute injury was induced by D-galactosamine/lipopolysaccharide (LPS) [8][9][10][11][12]. Though, these models showed histological features for the presence of both chronic and acute liver injury, the mean survival period was very short after the acute insult and rarely showed development of portal hypertension, ascites, and MOF; a common feature of human ACLF. Recently, Xiang et al. [13] have developed a mouse model of sepsis induced ACLF that provide the survival 1 3 window to understand the pathophysiology of liver failure in response to infection and liver injury. However, this model did not define whether the secondary organ failure is because of bacteremia or liver failure and does not address hepatic regeneration. Whether sepsis is a precipitant or consequence of ACLF is controversial [14][15][16]. According to the APASL criteria [17,18] sepsis is a complication which follows ongoing hepatocellular necrosis and occurs due to liver failure induced immune-paralysis [1,19,20]. To carefully evaluate the sequence of events, there is a need of an ACLF model which can reproduce this spectrum of human ACLF progression to understand the advancement of liver failure and its impact on organ dysfunction and immune injury.
Acute hepatic insult in ACLF has been thought to occur due to inappropriate and widespread activation of the inflammatory cytokine pathway; release of damage-associated proteins and other cytokines from dying parenchymal and non-parenchymal cells, intestinal endotoxemia as a result of increased bacterial translocation/bacterial products from the gut into the portal circulation [1,21,22]. In the current study, we developed a mouse model of ACLF by chronic administration of CCl 4 to produce chronic liver injury/ cirrhosis followed by an acute insult with acetaminophen (APAP) and a low dose of lipopolysaccharide (LPS); the CALPS model, to mimic the situation of an acute insult on a chronic liver disease along with introduction of intestinal endotoxemia. The present model was developed with an aim to investigate whether the progression of liver injury to liver failure in animals with pre-existing chronic liver disease/cirrhosis leads to subsequent development of portal hypertension, sepsis, and secondary organ dysfunction. The concomitant processes of hepatic regeneration, and injury to other organs was carefully studied in the presence of on-going liver failure. The sole aim was to see whether the sequence of animal experiments and the resulting murine disease model, resembles the syndrome of ACLF seen in human beings. 6-8-week-old male C57Bl6 mice were purchased from LIVEON BioLabs Pvt. Ltd, Karnataka, India. Animals were housed in the Institutional Animal Facility in a clean, temperature controlled environment with a 12 h light and dark cycle and were provided with free access to regular laboratory chow diet and water. Animal care and all experimental procedures were approved by the Institutional Animal Ethics Committee (IAEC) under the project approval (IAEC/ ILB/17/03) from Department of Science and Technology, Government of India.

Induction of chronic liver injury:
Chronic liver injury was induced by intraperitoneal administration of carbon tetrachloride (CCl 4 ; Central Drug House, Delhi, India) in Olive oil (HiMedia Pvt. Ltd., India). We used an increasing dose of CCl4 approach to develop chronic liver injury. Chronic liver fibrosis was induced in 6-8 weeks male C57BL6 mice over a 10 weeks period by twice weekly intraperitoneal injections of CCl4 dissolved in sterile olive oil at a concentration of 0.1 ml/kg for first 3 weeks increasing to 0.2 m/kg for next 3 weeks followed by 0.5 ml/kg of CCl4 till they developed grade 3/4 fibrosis by week 10. [Note: Time for development of grade ¾ fibrosis may vary with mice strain and their housing condition. Hence, duration of CCl4 treatment must be optimized for grade ¾ fibrosis before induction of ACLF injury].
Validation After 10 weeks of chronic liver injury, mice (n = 5) were euthanized to assess the degree of fibrosis. To develop ACLF injury, mice (n = 22) were fasted for 12 h before administrating APAP (350 mg/kg, i.p) followed by LPS (50 µg/kg, i.p) 60 min post-APAP. Post-APAP/LPS injury mice were culled at 24 h, day 7 and day 11. Organs (liver, kidney, and lungs) and blood were collected for histological and biochemical analysis. All animals were given humane care in compliance with the IAEC guidelines.
Detailed methodology of sample collection, histo-pathological and biochemical analysis are mentioned in supplementary data sheet.
Statistics: All the data are expressed as the means ± SD and were analyzed using the GraphPad Prism software (La Jolla, CA). The Student t test was performed to compare values obtained from two groups with two-tailed while the comparison between three or more groups was analyzed by one-way ANOVA. p < 0.05 was considered as significant.

Development of chronic liver disease
To induce chronic liver injury, C57Bl6 male mice (6-8 weeks) were subjected to intraperitoneal injection of successive doses of CCL 4 for 10 weeks (twice a week). After 10 weeks of chronic CCL 4 administration biochemical analysis of blood plasma showed a significant increase in ALT (p < 0.0001), AST (p < 0.0001), total bilirubin (p < 0.001) and decrease in albumin (p < 0.0001) but they did not show any feature of hepatic decompensation like ascites or increase in the level of blood ammonia (figure S1A).
In comparison to healthy, post-CCl 4 injury, liver tissue showed enhanced portal inflammation (with the collection of immature calls, blasts, plasma cells and few PMNs) with thin grade II to III bridging fibrosis and necrosis. Hepatocytes showed eosinophilic cytoplasm with enlarged reactive nucleus (figure S1B). Altogether, these data confirmed the presence of chronic liver disease/cirrhosis after 10 weeks of chronic CCl 4 administration in mice. Further analysis of plasma creatinine and blood urea nitrogen (figure S2A) were comparable to healthy mice after CCl 4 injury and they did not show any injury to kidney (figure S2B) and lung (figure S2C); suggesting no secondary organ dysfunction or injury.

Induction of acute injury
To induce ACLF in chronic liver injured/cirrhotic mice, we initially used three different combinations of acute liver injury. After 10 weeks of CCl 4 injury, animals were divided into three groups (n = 6, each group). Group 1 received high dose of LPS (100 ug/kg), group 2 received high dose of APAP (600 mg/kg), and group 3 received low dose of APAP (350 mg/Kg) + Low dose of LPS (50 ug/kg). While all the animals in groups 1 and 2 died within 48 h of acute injury, the animals in group 3 did not show any short-term mortality and developed features of human ACLF. These animals showed the presence of ascites, liver necrosis, inflammation, and acute tubular necrosis (Fig. S3A, B). Hence, we used this combination further for the development of ACLF and analysis of pathophysiological changes in the progression of ACLF model.
As described in Fig. 1A, after 10 weeks of CCl 4 injury, ACLF (n = 22) was induced by the combination of APAP and LPS and animals were sacrificed at 24 h (n = 5, day 7(n = 5) and day 11 (n = 10) post-LPS treatment. Two animals died between days 3-5. Post-APAP and LPS injury while there was no significant change in plasma AST and albumin [except an increase in AST (p < 0.05) at day 11 in comparison to cirrhosis, there was a progressive increase in plasma ALT (p < 0.001), prothrombin time (p < 0.001), bilirubin (p < 0.01), and blood ammonia (p < 0.001) (Fig. 1B] suggesting progressive liver failure. All the animals sacrificed at day 11 showed the presence of ascitic fluid with serum-ascites albumin gradient (SAAG) of > 1.1 g/dl (Fig. 1C). Further hemodynamic assessment in our set of mice, there was 48% (7.46 ± 0.5 mmHg) increase in portal pressure (PP) in CCl 4 cirrhotic mice compared to healthy (5.04 ± 0.5 mmHg; p < 0.01). 24 h post-APAP/LPS induced ACLF showed further rises in PP + 16% (8.77.46 ± 0.5 mmHg) in comparison to vehicle treated (7.464 ± 0.5 mmHg; p < 0.05). Suggesting aggravation in the PP was due to the second hepatic insult and this pathological increase in PP was maintained after 11 days + 12% (8.15 ± 0.5 mmHg) in comparison to vehicle treated (p = 0.08) (Fig. 1D).
Altogether, the animals showed characteristic clinical and biochemical features of human ACLF as an increase in bilirubin (jaundice), prothrombin time (coagulopathy), blood ammonia, ascites, and portal hypertension.

APAP with LPS induces progressive liver injury and inflammation
Histological analysis of liver tissue at 24 h, day 7 and day 11 post-APAP and LPS treatment showed persistent portal inflammation ( Fig. 2A). In comparison to cirrhotic liver, 24 h post-APAP and LPS treatment, the hepatocytes showed a marked change in their morphology in terms of ballooning and peripheral condensation of organelles with rigid appearance of the cell membrane ( Fig. 2A, S4A), which reduced till day 11. However, there was an increase in necrosis from day 7 to day 11 ( Fig. 2A) and the presence of cholestasis at day 11 (figure S4B), this is quite characteristic in human ACLF as well. To confirm liver injury TUNEL assay was performed. In cirrhotic livers, the TUNEL positivity was mainly confined to immune cells near the fibrotic septa. However, at 24 h post-injury, the presence of apoptosis was seen around the vascular regions having most of the endothelial and liver infiltrating cells. This was seen to increase at day 7 with the hepatocyte death surrounding the vascular niche. However, at day 11, the hepatocyte death further progressed to the lobular areas and showed marked differences at the level of showing blood biochemical analysis of mice sacrificed at 24 h, day 7 and day 11 of analysis in comparison to cirrhosis. c Bar-graph showing the measured SAAG value (> 1 g/dL SAAG is an indicator of ascites development due to liver failure). d Bar-graph showing portal pressure of healthy, 10 weeks CCl4 injury, 24 h and day 11 post-ACLF injury; p = * < 0.05; p = ** < 0.005; p = *** < 0.0005; p = **** < 0.0001 ◂ apoptosis early (peri-nuclear TUNEL +) as well as late-stage (Nuclear TUNEL +) (Fig. 2B, S4C). In further western blot analysis while faint band of cleaved caspage-3 was observed at 24 h post-acute injury there was no Cleaved caspage-3 expression at day 11 (Fig. 2C), this suggests that hepatocyte death in this model is mainly because of necrosis. Similarly, the Ly6g positivity enhanced from 24 h to day 11 ( figure  S4D) suggesting activation of neutrophils at day 11 of postacute injury.
To investigate the changes in inflammatory responses, we analyzed the expression of pro-(IL6, IL-1β, TNF-α) and anti-(IL10) inflammatory cytokines in liver tissue by q-RT-PCR and infiltration of Ly6G+ neutrophil by immunohistochemistry. In comparison to cirrhosis animals, while there was a persistent increase in pro-inflammatory cytokines namely IL6, TNF-α, and IL-1β starting from 24 h (Fig. 2D); the expression of anti-inflammatory cytokines (IL10) (Fig. 2E) showed increase by day 7 post-APAP and LPS treatment. Similarly, the number of Ly6G+ neutrophil significantly increased from 24 h to day 11 ( figure S4D). Altogether, these data show the presence of progressive liver injury, neutrophil infiltration and cytokine storm similar to what is witnessed in human ACLF. With APAP and LPS treatment, liver injury begins with hepatocyte ballooning, vascular injury and portal inflammation leading to and perpetuating hepatocyte death.

APAP with LPS treatment causes regression of liver fibrosis in cirrhosis
APAP and LPS treatment induced a significant reduction in fibrosis measured by MT and Sirius red. In comparison to chronic liver injury/cirrhotic liver, in the post-APAP and LPS treated animals there was reduction in the level of fibrosis from 24 h of treatment till day 11. MT staining showed the reduction from 41.2% ± 2.1 (p < 0.0001) at 24 h to 22.9% ± 2.1 (p < 0.0001) at day 7, and 13.1% ± 1.2 (p < 0.0001) at day 11. Similarly, Sirius red staining reduced to 82.2% ± 2.4 (p < 0.01) at 24 h, 50.6% ± 1.6 (p < 0.0001) at day 7, and 16.7% ± 1.6 (p < 0.0001) at day 11 (Fig. 3A). To validate the reduction in fibrosis, we did collagen I staining and found it to be reduced from 85.2% ± 4.1 (p = 0.02) at 24 h; 40.3% ± 1.3 (p < 0.0001) on day 7 and 33.3% ± 3.9 (p < 0.0001) on day 11 (Fig. 3B). A reduction in the number of α-SMA + myofibroblasts through apoptosis is a key early event during fibrosis resolution. The amount of α-SMA staining post-APAP and LPS treatment significantly reduced to 81.2% ± 1.4, p = 0.01at day 7 and 58.6% ± 2.4, p < 0.0001 at day 11 of cirrhosis (Fig. 3C). Hence, the reduction of α-SMA + myofibroblasts in response to APAP and LPS induced acute injury might seem to have contributed in the regression of fibrosis in this animal model.

ACLF induces regeneration failure in cirrhosis
Acute injury triggers hepatic regeneration [23], however, we observed progressive symptoms of liver failure in terms of increase in bilirubin, prothrombin time, and blood ammonia post-APAP and LPS injury in CCL 4 models of chronic liver disease (Fig. 1B). Hence, we checked the effects of APAP and LPS induced liver injury on hepatic regeneration by analyzing the proliferating cell nuclear antigen (PCNA) positive hepatocyte for hepatocyte proliferation and CK-19 positive ductular cells for hepatic progenitor cell-mediated liver regeneration. At cirrhosis level there were 66 ± 1.5 PCNA + hepatocytes per high power field which were significantly reduced from 24 h to day 11 (46 ± 2.7 by 24 h (p < 0.001); 10 ± 3.2 by day 7 (p < 0.0001) and 3 ± 1.2 by day 11 (p < 0.0001)) post-acute injury (Fig. 4A, B). This suggests a progressive loss of hepatocyte proliferation after post-APAP and LPS induced acute injury. Similar to hepatocyte proliferation, the number of CK-19 + ductular cells also reduced in the cirrhotic animals at 24 h and day 7 with slight increase at day 11 from day 7 (Fig. 4C, D). Our animal model did not show significant ductular reaction, an indicator of progenitor cell-mediated liver regeneration.

APAP with LPS induced liver injury induces systemic organ dysfunctions
Systemic organ dysfunctions are the common features of ACLF in human [1]. ACLF patients are more susceptible to renal dysfunction which is associated with severe outcome. To check extra-hepatic manifestation of ACLF, we analyzed the levels of serum creatinine and blood urea nitrogen (BUN). While there was no significant change in the level of creatinine by day 7 post-APAP and LPS, it was significantly (p < 0.01) increased at day 11. Similarly, the level of BUN also increased significantly (p < 0.05) at Day 11 (Fig. 5A) suggesting renal dysfunction. Further, the analysis of kidney histology showed the marked presence of acute tubular necrosis (ATN) at day 11 (n = 8; 80%), suggesting renal dysfunction in response to APAP/LPS induced ACLF is due to structural damage of kidney (Fig. 5B).  in ACLF and cirrhotic liver tissue (magnification ×20). c Quantitative analysis of PCNA positive hepatocytes per high power field. d Quantitative analysis of ck-19 positive cells per high power field. p = * < 0.05; ** < 0.005; **** < 0.0001, N = 5 in each group

3
The histology of lungs showed no changes at week 10 of chronic liver injury, 24 h and day 7. There was the presence of pulmonary dysfunction (interstitial pneumonia and exudate pneumonia) at day 11 in 3 of 10 (30%) mice samples ( figure S5).
Altogether, these data suggest that liver failure induces systemic organ dysfunctions in this animal model of ACLF.

Discussion
The underlying cause of rapid and progressive liver failure leading to sepsis, MOF and high mortality in ACLF is poorly defined [17,18,25]. This is dominantly because, the patients present days or weeks after the initial hepatic insult. Quite often they present with single or multi-organ failure. Hence, there is need for a representative experimental model that mimics various stages and the course of human ACLF; to decipher the pathophysiological journey of the disease and development of new therapeutic strategies.
In the current study, we developed a mouse model of progressive ACLF, the CALPS model; that resembles the clinical and histological features of human ACLF [17,18]. This model demonstrates that acute liver injury with endotoxemia in cirrhosis leads to progressive hepatocyte necrosis without liver regeneration leading to liver failure and development of secondary organ injury. This model helped to investigate the hypothesis that the perpetual presence of hepatotoxin and intestinal endotoxemia lead to hepatocellular necrosis along with inappropriate and widespread activation of the inflammatory cytokine pathway leading to liver failure and subsequent secondary organ failure/sepsis in ACLF [1,19,21,22].
Several approaches have been reported in the past by various groups to develop the ACLF model using D-galactosamine/lipopolysaccharide (LPS) [8][9][10][11][12]. Though these models showed histological features for the presence of both chronic and acute liver injury, they failed to recapitulate the whole pathological process of this disease due to high short-term mortality and lack the presence of ascites, encephalopathy and secondary organ dysfunction; common features of human ACLF. It is also difficult to understand in these models, whether the high short-term mortality is due to liver failure or endotoxin induced shock. We showed that modest doses of APAP and LPS are sufficient to trigger the liver injury followed by hepatic decompensation (as evident by the presence of jaundice, coagulopathy, ascites and rise in blood ammonia in the current model) and secondary organ injury similar to human ACLF. Our data show that in response to APAP/LPS injury in cirrhotic liver, there is marked increase in hepatocyte ballooning and endothelial/immune cell death within 24 h followed by progressive increase in hepatocyte necrosis from day 7 to 11. This leads to cytokine storm with a persistent increase in pro-inflammatory cytokines (IL6, TNFα and IL-1β) starting from 24 h and followed by an increase in expression of anti-inflammatory cytokines (IL10) by day 7 post-acute injury with increase in neutrophil infiltration. Manifestation of jaundice and coagulopathy followed by the development of ascites and/or encephalopathy post hepatic injury is essential for ACLF as per APASL definition [18]. Similar to human ACLF, our animals showed a progressive increase in bilirubin and blood ammonia post-APAP/LPS induced liver injury. All animals sacrificed on day 11 showed the presence of ascites with SAAG > 1.1 g/dl, hence, conforming to the development of human ACLF. Ballooning degeneration is the precursor of lytic necrosis and has been reported as important histological features in > 50% of ACLF patients [26]. The underlying cause of ballooning degeneration is probably a mitochondrial injury followed by impairment of oxidative phosphorylation, ATP-depletion, loss of energy homeostasis and an increase in permeability of the plasma membrane leading to a ballooning of cells [27]. It reverses if cells repair the loss of damage or may lead to necrosis if the cells fail to restore the energy homeostasis [27]. This raises the possible link between loss of hepatocyte energy and progression of necrosis in response to APAP and LPS in progressive liver injury in ACLF.
Interestingly in the CALPS model, we observed a significant reduction in fibrosis. A reduction in the number of α-SMA + myofibroblasts through apoptosis is a key early event during fibrosis resolution [28]. We observed a significant reduction in α-SMA + myofibroblast post-APAP/LPS treatment. We also observed increased tunnel positive cells in the fibrotic area of the liver; this may be associated with reduction of fibrosis in response to acute injury in the current model. Regression of fibrosis has also been reported in some ACLF patients [18].
The liver has an extraordinary capacity to regenerate on the loss of liver tissue following liver injury due to toxins, surgical resection, infection, or trauma. Compensatory dose dependent increase in liver regeneration with hepatocyte injury has been shown to be associated with spontaneous recovery in various models of drug induced acute live injury [23,29]. This may be quite different in ACLF. In response to APAP and LPS injury, we observed a progressive reduction in hepatocyte proliferation, even though there was a progressive increase in hepatocyte necrosis. Hence, the loss of compensatory regeneration with progressive hepatocyte death indicates impairment in cellular signaling pathways in chronic liver disease which could be responsible for the rapid progression of liver failure in this model of ACLF. Loss of hepatocyte replication has also been shown in the human ACLF [24]; however, unlike human ACLF we did not observe any increased ductular proliferation, an indicator of HPC mediated regeneration. Indeed earlier we have shown that increased HPCs expansion in ACLF does not contribute to outcome and only hepatocyte replication is associated with spontaneous recovery of these patients [24]. Recently, in active infection induced animal model of ACLF, decrease in IL6 and increase in TGF-beta have been shown to be the underlying cause of poor hepatocyte replication. However, we did not find any association of these cytokines with loss of hepatocyte regeneration in our model. In comparison to the cirrhosis, there was a significant increase in hepatic IL6 within 24 h post-injury which remained high till day 11, while the level of TGF-beta was comparable to that observed in cirrhosis except at 24 h it decreases (Fig. 2E).
Systemic organ dysfunctions are the common features of ACLF in human [1] of which kidneys are the most affected organs [25]. In the present animal model of ACLF, while there was little discernible change noticed in the kidney/lung histology between day 0 and 7, at day 11, about 80% of the animals showed features of acute tubular necrosis and 30% of them also showed interstitial and/or exudative pneumonia. This suggests that the kidney damage and renal dysfunction are the consequences and not the precipitant of ACLF. It also shows that the renal dysfunction is due to structural damage of kidney, possibly similar human ACLF [30].
In summary, the CALPS murine model of ACLF shows the clinical and histological features of human ACLF in terms of the presence of jaundice, ascites, and acute tubular necrosis and renal dysfunction. It showed that the perpetual presence of acute liver injury and endotoxemia trigger hepatocyte ballooning followed by massive hepatocyte necrosis in the presence of chronic liver disease. The acute insult in chronic liver disease triggers regression of fibrosis, concomitant impaired hepatocyte regeneration with on-going liver injury results in liver failure and development of secondary organ injury. This model provides the pathophysiological road-map of the development of ACLF in chronic liver disease in response to acute liver damage and endotoxemia. It is also a representative model for understanding the effects of liver failure on renal injury. This model could help in determining the stage specific interventions to prevent development of portal hypertension, renal and lung injury and also to stimulate hepatic regeneration in the presence of on-going liver injury.