Chronic Xiao-chai-hu Decoction Therapy Slows the Progression of Acute-on-chronic Liver Failure via Prokinetic Effects on Gastrointestinal Motility

the serum levels of aspartate aminotransferase and time (PT); survival rate and rat-adapted model for end-stage liver disease (MELD) score; the hepatic hyperplastic collagen bers with Masson’s trichrome staining; 4) the serum level of TNF-α and endotoxin induced hepatocyte apoptosis expression; 5) the ileum slow waves, gastric emptying, and small intestinal transit after XCHD treatment.


Background
Acute-on-chronic liver failure (ACLF) is a syndrome in patients with chronic liver disease with or without previously diagnosed cirrhosis. It is characterized by acute hepatic decompensation resulting in liver failure and one or more extrahepatic organ failures. Despite the application of extracorporeal liver support systems, such as plasma exchange, molecular adsorbent recirculating system, and liver transplantation in recent years (Finkenstedt et al., 2013), the mortality rate of ACLF is still 30-50% (Solé et al., 2018). Notably, 25% of the ACLF patients are suitable for a liver transplant (Finkenstedt et al., 2013). Moreover, a randomized study showed a signi cant increase in the survival of extracorporeal liver support systems  Huebert et al., 2014) and colony-stimulating factor (CSF) (Khanam et al., 2014), are rather challenging in both developing and developed countries due to heavy medical expenses.
Chinese herbal medicine has been adopted in the treatment of liver diseases for a long time in China (Qi et al., 2013). Xiao Chai Hu decoction (XCHD) is widely recognized as a medicine in Asia for the treatment of liver diseases. It is a well-known prescription in traditional Chinese medicine (TCM), and is composed of seven herbs: Radix bupleuri, Scutellaria baicalensis, Panax ginseng, Rhizoma Pinelliae preparata, honey-fried Glycyrrhizae Radix, Zingiber o cinale, and Ziziphus jujube (Xiong et al., 2011). It was believed that XCHD alleviates the symptoms of fever, abdominal distension, poor appetite, nausea, vomiting, and jaundice. Previous studies con rmed that XCHD inhibits the progression of hepatitis to liver brosis (Dou et al., 2005) and the replication of hepatitis B and C virus (Li et al., 2017). Moreover, it prevents liver damage via immune regulation Liu et al., 2013). Bupleurum and Scutellaria baicalensis are the core paired components of this prescription. Previous studies have shown that Scutellaria root is effective an antiendotoxic (Cheng et al., 2007), anti-bacterial (Miyasaki et al., 2013), anti-hepatic brosis (Miyasaki et  Moreover, XCHD alleviates chronic liver damage that might decrease the mortality risk of patients. Thus, we hypothesized that XCHD prevents ACLF-induced death via prokinetic effects on the gastrointestinal motility against tumor necrosis factor-alpha (TNF-α) and endotoxin-induced hepatocyte apoptosis.

Experimental animals
A total of 120 healthy and clean male Sprague-Dawley rats (90-120 g) were purchased from the Laboratory Animal Center of Xi'an Jiao Tong University (Xi'an, Shaanxi, China), acclimated to the research laboratory for ve days before experiments, and maintained in a light-controlled room ( 2) XCHD (Su et al., 2014): It consisted of seven crude herbs (herbs implementing standards: Pharmacopoeia of the People's Republic of China 2015 rst edition). These herbs were mixed and decocted three times (100 g/1000 mL the rst time; 100 g/400 mL the second time; 100 g/400 mL the third time) with water for 30 min, as described previously. The decoction was ltered through the gauze; the ltrate was concentrated as 4.45 g/mL. After an overnight fast, under anesthesia with the inhalation of 1.5-2.0% iso urane, the hair was shaved, the skin on top of the corresponding position was cut open, and then one pair of cardiac pacing wires was implanted in the serosal surface of the ileum about 4 cm proximal to the ileocecum for ileum slowwave recording. The distance between the two electrodes in the pair was 0.3 cm. After the isolated lead wires were tunneled and externalized on the rat's neck, buprenorphine (0.05 mg/kg) and cefazolin (30 mg/kg) was administered for 2 days to alleviate postoperative pain and prevent infection, respectively. The rats were housed individually to avoid the wires and tubes from being chewed off by other rats. The experiments were initiated after the rats were completely recovered from the surgery, usually 7 days after the operation. After 7 days of conventional adaptive feeding, 110 rats were used to establish the model. Then, 10 rats were used as the normal control group. During the initial 4-week treatment, 40% CCL4 olive oil solution was injected subcutaneously at 1.5 mL/kg body weight every three days. Then, in the following two weeks, the 40% CCL4 olive oil solution was injected subcutaneously at 2 mL/kg body weight every three days. After the establishment of the cirrhotic rat model, lipopolysaccharide (LPS)-and D-galactosamine (D-Gal)-induced ACLF models were constructed. On day 45, the rats were injected LPS at a dose of 10 mg/kg and D-Gal at a dose of 700 mg/kg intraperitoneally, which induce acute liver failure based on chronic liver injury.
The normal control rats were administered physiological saline at the time points same as in the model rats. The animal behavior was closely monitored during the modeling.

Method of administration
During modeling, 110 rats were randomly divided into model control group with 30 rats and four treatment groups with 20 rats in each group: polyene phosphatidylcholine (100 mg/kg; PP) group, highdose (44.5 g/kg; HXCHD), middle-dose (26.5 g/kg; MXCHD), and low dose (8.5 g/kg; LXCHD) Xiao Chai Hu decoction. In the model and control groups, rats were given an equivalent volume of water by gavage daily.

Specimen collection and preparation
On day 44, ve rats were selected randomly in the ACLF model group, and their liver histopathology (included) was detected to determine the cirrhosis. After an acute attack, 0.6 mL blood samples withdrawn at 2 h, 10 h, and 18 h were collected from the inner canthus in all rats to detect the serum biochemical levels. After 47 days of treatment, the colon slow-wave was measured. Then, all the surviving rats were sacri ced for specimen collection. One part of the liver tissue was cut and embedded in para n for hematoxylin and eosin (H&E) and Masson's trichrome staining.

Analysis of liver disease progression 1) Serum levels of liver functionality measurements
The serum was obtained by centrifugation of blood samples at 1500 ×g for 10 min at 4 °C and stored at room temperature for 1 h and then at 20 °C until further analysis. The activity of AST, ALT, TB, ALB, and PT was measured using commercially available kits (Sigma-Aldrich), according to the manufacturer's instructions.
2) Survival rate and rat-adapted model of end-stage liver disease (MELD) score (Said et al., 2004) Based on the literature, the rats' MELD was adapted as follows: Rat-adapted MELD score = 0.957×Loge (creatinine mg/dL) + 0.378×Loge (bilirubin mg/dL) + 1.120×Loge ( INR) + 0.643. MELD is a scale system to score the severity of liver disease and predict death. A high MELD score indicates a high probability of death.
In Masson staining, blue color indicated hyperplastic collagen bers. Five low-power (×40) elds were randomly assessed for each slice. The ratio of the area of hepatic hyperplastic collagen bers (AHHCF%) was calculated by Image Pro-Plus 6.

Mechanisms of XCHD involving endotoxin-induced hepatocyte apoptosis
1) Detection of serum endotoxin and TNF-α levels with two-step double-antibody sandwich enzyme-linked immunosorbent assay (ELISA).
According to the manufacturer's description 2.5-80 pg/mL was used in ELISA to detect and quantitatively analyzed the serum endotoxin and TNF-α levels of rats in all the groups. A volume of 100 μL samples was dispensed in each well in 96-well plates after pre-coating with rat TNF-α/endotoxin antibody for 30 min at 37 ℃. Then, a protein-speci c biotinylated antibody was incubated for 2 h at 37 ℃. Between each reaction, the unbound proteins were washed away from the wells. Then, substrate solutions A and B were added to each well, and the reaction incubated for 20 min at 37 ℃. The absorbance was measured at 450 nm in a microplate reader, and the concentration data expressed as a mean value.
2) Detection of hepatocyte apoptosis by terminal deoxynucleotide transferase dUTP nick end labeling (TUNEL) According to the manufacturer's instructions, TUNEL method was used to detect cellular apoptosis on the liver tissue. The sections were xed in ethanol-acetic acid (2:1), incubated with proteinase K (100 μg/mL), rinsed in PBS, incubated in 3% H 2 O 2 , and washed with phosphate-buffered saline (PBS) for 10 min, followed by permeabilization (0.1% Triton X-100, 0.1% sodium citrate) for 5 min. Subsequently, the sections were washed and incubated in TUNEL reaction mixture. Converter-POD with 0.02% 3,3'diaminobenzidine (DAB) was used for visualization, and Mayer's hematoxylin was used for counterstaining. TUNEL-positive cells were dyed as the yellow nucleus. The liver cells were counted under high-power (×400) eld. Five high-power elds were examined in each case, and the number of TUNEL-positive cells was counted. Finally, 500 cells were counted in each eld, and the apoptotic index (AI%) of each group was calculated.

Mechanisms of XCHD involving gastrointestinal motility
After overnight fasting, each rat was fed 2 g of solid dry food at the end of the experiment. A volume of 1.5 mL phenol red (0.5 mg/mL) was mixed with 1.5% methylcellulose was delivered by gavage after 10 min. Then, the rat was euthanized, the content of the stomach and the small intestine was collected for measuring the gastric emptying (GE) (Lin et al., 2018) and small intestinal transit (SIT) (Ohno et al., 2006), using a previously established method.
The percentage of GE was described as the ratio between the amount of undigested food in the stomach and 2 g of solid dry food. The small intestine was cut into 10 equal segments. The SIT was assessed using the geometric center based on the amount of phenol red in each of the segments (Scarpignato et al., 1980).

Statistical methods
SPSS16.0 statistical analysis software package (IBM, Armonk, NY, USA) was utilized. Continuous data were expressed as mean ± standard. One-way analysis of variance (ANOVA) was performed, and q test (Student-Newman-Keuls test, S-N-K) was used for pairwise comparisons. Kaplan-Meier method was used to calculate the survival rate. Two-sided P-values < 0.05 were considered statistically signi cant.

Veri cation of ACLF rat model
After the abdominal cavity was opened, it was over owed with turbid yellow liquid. Varying degrees of adhesion was observed between the abdominal tissues. The surface of the liver was rough and uneven, and dense nodules were detected in the cirrhosis and ACLF model ( Figure 2B, C). Also, white fur or pus was on the surface of the liver in the ACLF model than the cirrhosis model.
H&E staining and Masson's trichrome staining detected clear hepatic lobule structures in the normal rats ( Figure 3A1, A2). In the model group, the hepatic lobule structures were damaged, and liver cell cords were arranged disorderly with obvious swelling and fatty degeneration ( Figure 3B1, B2, C1, C2). A large number of blue hyperplastic thick collagen bers was observed in the portal area and lobules in Masson staining. The hepatic cell cords of the liver lobule were disordered, and a large amount of unequal-sized pseudo lobules were formed. Also, a large number of necrotic liver cells were detected, and in ammatory cell in ltration was observed ( Figure 3C1, C2).
After acute attachment for 2 h, the serum level of ALT, AST, and TNF-α increased sharply and decreased in 10 h. However, the level of total bilirubin (TBIL) increased progressively from 2-18 h ( Figure 4A), indicating liver failure. Together, these ndings demonstrate an experimental model of ACLF.

Comparison of the liver disease progression 1) Comparison of the serum levels of liver functionality measurements
The activity of AST, ALT, TB, ALB, and PT was measured using commercially available kits (Sigma-Aldrich), according to the manufacturer's instructions.  Figure 4B, 4C). These data suggested that XCHD treatment improves hepatobiliary function. Importantly, the serum ALB levels of the rat model demonstrated a signi cant decrease, whereas the animals treated with HXCHD and MXCHD showed a similar increase compared to those of the PP group ( Figure 4D). This demonstrated that HXCHD promotes liver functionality as a protein producer. The rats treated with XCHD showed decreased PT ( Figure 4E). Conversely, no differences were detected in the PT among the PP, HXCHD, and MXCHD groups.
2) Comparison of the survival rate and rat-adapted MELD score Two stages were detected in the ACLF model. The rst lasted 4 weeks to develop progressive chronic liver failure; six rats died in the model group, ve rats died in LXCHD, four rats died in MXCHD, and none of the rats died in the HXCHD groups. In addition to chronic liver failure, acute liver failure resulted in rat mortality. After 48 h of acute attack, the survival rate of the various groups was as follows: 100% in the normal control group, in the model and LXCHD groups (χ 2 = 0.078, P < 0.05 vs. model), 77% in the MXCHD group (χ 2 = 1.99, P < 0.05 vs. model), and 60% in the HXCHD group (χ 2 = 11.37, P < 0.05 vs. model; χ 2 = 5.976, P < 0.05 vs. MXCHD). We also observed the time of death. In the model group, the death mainly occurred during the chronic liver failure stage to a relatively high mortality rate of 30%, while early survival was 100% in the HXCHD-treated rats ( Figure 5A). After administration of D-gal/LPS, the HXCHD group succumbed to mortality at 20 h compared to that in the model group at 12 h, suggesting that treatment with XCHD effectively slowed the evolution of the disease.
In addition, the XCHD treated groups showed a signi cantly lower MELD score than the model group ( Figure 5B), indicating healthier livers after XCHD treatment. Taken together, this phenomenon suggested that XCHD treatment slowed the evolution of the disease. Compared to the model rats ( Figure 5C), the INRs were decreased in the rats treated with XCHD compared to the model group similar to PP. These data indicated an improvement on liver failure due to XCHD treatment.
3) Comparison of the ratio of the area of hepatocyte apoptosis in each group Hepatocyte apoptosis in normal rats was detected occasionally ( Figure 6A). The hepatocyte apoptosis of ACLF rats increased markedly. The positive nuclei with brown or black color were scattered diffusely in the eld and were pyknotic-like, or chromatin aggregated to the surrounding, or the nuclei were fragmented ( Figure 6B).

4) Comparison of the ratios of the area of hepatic hyperplastic collagen bers in each group
The model rats showed 29.8% area of brosis in the pathological livers, while improved hepatic histology was observed in XCHD rats. The hepatic brogenesis brosis was signi cantly decreased in the XCHDtreated rats as follows ( Figure 7A): in the HXCHD (10.88 ± 0.88; P < 0.05 vs. model), MXCHD (15.32 ± 0.55; P < 0.05 vs. model), and LXCHD (19.18 ± 1.12; P < 0.05 vs. model) groups. Moreover, the percentage of brosis was signi cantly lower in the HXCHD rats than in the rats treated with PP (12.14 ± 0.90; P < 0.05 vs. model).
The pooled biochemical and histopathology data indicated that the administration of HXCHD slowed the progression of the disease. Subsequent studies investigated whether the low survival rate of XCHD was associated with TNF-α and endotoxin-induced hepatocyte apoptosis.

5) Comparison of the serum levels of TNF-α and endotoxin in each group
Compared to the normal control group, the TNF-α and endotoxin levels of all the experimental groups increased signi cantly (P < 0.05, F TNF-α = 4.37; F endotoxin = 3.31; Figure 7B, C). The serum TNF-α and endotoxin levels of the model group were signi cantly higher than those of all the treatment groups. The serum endotoxin level of the HXCHD treatment group was signi cantly lower than that of the PP treatment group, (P < 0.05; q TNF-a = 3.98; q endotoxin = 2.79).
Compared to the normal group, the percentage of normal ISW (intestinal slow wave) was substantially reduced in the ACLF model. The HXCHD treatment prevented the ACLF-associated decline in the percentage of normal ISW (Figure 9).

Discussion
ACLF is one of the leading causes of deaths in patients with chronic liver diseases. In recent years, some progress has been made to decrease the mortality rate of ACLF, such as using the combination of antivirus, arti cial liver, and liver transplantation. However, medical expenses, expensive medical equipment, and limited donor organ hinder development, especially in the developing country.
Thus, new strategies are urgently required for high mortality. In the present study, XCHD was bene cial in treating liver brosis, which improved the serum parameters of liver disease. In addition, the preventive therapy with XCHD satisfactorily improved the cirrhosis and ACLF survival in rats. Furthermore, XCHD prevented ACLF via a decline in TNF-α and endotoxin-induced hepatocyte apoptosis through the prokinetic effects on gastric motility. In this model, we demonstrated that liver brosis/cirrhosis is induced by CCl4, which is con rmed by liver histology results. CCl4 leads to the necrosis of hepatocytes, induces in ammation, and further promotes the progression of hepatic brogenesis in rats. Subsequent D-gal/LPS treatment caused acute massive hepatic necrosis, and a large white fur or pus was observed on the surface of the liver in ACLF rats.
ACLF progresses rapidly, and its mortality rate in patients is 30-40% (Arroyo et al., 2015). Reportedly, during the brosis-induction period, the CCL4-induced model showed high mortality of 30%, and about 90% of the rats died from acute liver failure after administration of D-gal/LPS within 13 h (Liu et al., 2007). In this study, a high mortality rate was observed within a short period following D-gal/LPS treatment, starting after 12 h and reaching 100% at 48 h. The subsequent ndings on the signi cantly increased levels of ALT and AST that decreased afterwards, and the markedly increased TB in the ACLF model indicated liver parenchymal damage. Taken together, these observations suggested that ACLF rat model re ects the typical clinical manifestations of ACLF.
ACLF with cirrhosis shows a higher mortality rate in acute decompensation than without cirrhosis (Ruiz- proposed to remove these pro-apoptotic factors. Hitherto, no study has described a molecular adsorbent recirculation system dialysis to reduce these toxic molecules in the presence of liver failure plasma (Saich et al., 2007). In the current study, with different doses of XCHD treatment, the hepatocyte apoptosis rate was signi cantly reduced, while the survival rate was increased after XCHD treatment within 48 h. However, it was challenging to determine whether XCHD removed the pro-apoptotic factors from the plasma directly and whether hepatocyte apoptosis is closely related to the survival from ACLF.
TNF-α is a major cytokine produced by in ammatory cells (Baggiolini, 2001). Among the clinical TCMs, XCHD has been widely used in treating chronic hepatitis, cirrhosis, and acute and chronic cholecystitis (Li, 2020). Accumulating experimental and clinical evidence suggested that XCHD affects liver diseases via an anti-in ammatory mechanism, suppressing the progression of liver damage and via immune regulation (Qi et  . In the present study, we found that GE and SIT were delayed in ACLF rats. XCHD treatment accelerated the GE and SIT compared to the ACLF group. Similarly, the normal percentage of ISW was signi cantly improved after than before the XCHD treatment. These ndings indicated XCHD improves GE and SIT by improving dysrhythmia of GSW.
Based on the current study design, it was di cult to determine whether XCHD could directly prohibit intestinal bacterial overgrowth and reduce the permeability of the intestinal mucosa. It was also possible that XCHD would have some effects on the intestinal translocation of bacteria or bacterial products via prokinetic effects on gastrointestinal motility. Therefore, additional follow-up studies are needed to prove this hypothesis. Although XCHD is a promising drug for treating ACLF in the future, in-depth studies are essential.

Conclusions
XCHD prevents the progression of acute-on-chronic hepatic failure partially via the prokinetic effects on gastrointestinal motility to reduce the levels of TNF-α and endotoxin and hepatocyte apoptosis. Although in-depth studies are needed, the current results suggested that preventive therapy with XCHD satisfactorily improves cirrhosis and ACLF survival in rats.

Consent for publication
Written informed consent for publication was obtained from all participants Availability of data and materials The datasets used or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.       Comparison of the serum levels of TNF-α and endotoxin in each group. A. area of hepatic brogenesis; B. serum levels of TNF-α; C. serum levels of endotoxin. Asterisk: signi cant at p<0.05.

Figure 8
Prokinetic effects of XCHD on motility in each group. A: gastric emptying; B. geometric center. Asterisk: signi cant at p<0.05.