2.1 Animal models of ACLF

Healthy male Sprague Dawley (SD) rats (150-170 g) fed in the medical research centre of Beijing You-an Hospital, Capital Medical University, were used in our experiments. They were acclimated for 2 weeks before experimentation. The ACLF animal model was established according to a published method.13 Lentiviruses encoding the Mfn2 (LV-Mfn2) and control lentivirus were constructed by JI KAI Gene Technology Co. Ltd. Animals were randomly divided into 5 groups (n = 10). The grouping and treatment regimens were as follows: Group I: normal control, the rats receiving normal saline (1.5 mL/Kg), twice per week, 10 weeks; Group II: ACLF model, intraperitoneally receiving vegetable oil and 40% CCL4 mixture (1.5 mL/kg), twice per week, 10 weeks; Group III: LV-Mfn2 group, receiving vegetable oil and 40% CCL4 mixture (1.5 mL/kg), twice per week, and LV-Mfn2 (1X10⁹ IU/mL), 200 µL per one rat, once per 3 weeks, 10 weeks; Group IV: LV-Con group, receiving vegetable oil and 40% CCL4 mixture (1.5 mL/kg), twice per week, and LV-Con (1X10⁹ IU/mL), 200 µL per one rat, once per 3 weeks, 10 weeks; Group V: LV-Mfn2 + rapamycin group, receiving vegetable oil and 40% CCL4 mixture (1.5 mL/kg), twice per week, and LV-Mfn2 (1X10⁹ IU/mL), 200 µL per one rat, once per 3 weeks, beginning from the sixth week, rapamycin (1 mg/kg), once per day, 10 weeks. Groups II–V, against the background of chronic liver injury, were then challenged intraperitoneally with LPS (100 µg/kg) (Sigma Chemical Co.) combined with D-gal (0.5 g/kg) to induce acute liver failure.

The investigation was conducted according to the ethical standards and the Declaration of Helsinki. This study has also been approved by Beijing You-an Hospital, Capital Medical University. All methods and procedures including animals were performed according to the guidelines of the Animals Committee.

2.2 Hepatocyte autophagy model

The human normal liver HL-7702 cell line was a gift from the cell laboratory of the Beijing You-an Hospital, Capital Medical University. The 7702 cells were selected and incubated with 95% air and 5% CO₂ at a temperature of 37°C (normal control). The Binder three-gas incubator was used, with a temperature of 37°C, a CO₂ concentration of 5% and an O₂ concentration of 0.3% to provide a hypoxic environment, and serum-free DMEM was used to induce starvation for 6 hours.

2.3 Adenovirus-Mfn2 treatment of cell lines

Adenovirus encoding the Mfn2 open reading frame (Ad-Mfn2) and control adenovirus were constructed by JI KAI Gene Technology Co. Ltd. The grouping and treatment regimens were as follows: Group I: normal control; Group II: hepatocyte autophagy model; Group III: Ad-Mfn2 group, where the 7702 cells were incubated with Ad-Mfn2 at a multiplicity of infection (MOI) of 200 pfu per cell at 37°C for 24 hours, hypoxia and starvation for 6 hours; Group IV: Ad-Mfn2 NC group, where the 7702 cells were incubated with Ad-Mfn2 NC at a multiplicity of infection (MOI) of 200 pfu per cell at 37°C for 24 hours, hypoxia and starvation for 6 hours; and Group V: Ad-Mfn2 + rapamycin group, where the 7702 cells were incubated with Ad-Mfn2 at a multiplicity of infection (MOI) of 200 pfu per cell at 37°C for 12 hours, rapamycin for 12 hours, hypoxia and starvation for 6 hours.

2.4 Subcellular fractionation

The mitochondrial and cytosolic fractions were obtained using a ProteoExtract subcellular proteome extraction kit (Calbiochem).

2.5 Western blotting analysis

Western blotting analysis was performed as described previously.14 Samples of 50 μg of total protein were used for Western blotting. The list of primary antibodies is shown in Table S1 online. Protein bands were visualized using the SuperSignal West Pico Chemiluminescent Substrate (Thermo Fisher Scientific). The average intensities of each standard protein band were quantified using ImageJ and these results were normalized using GAPDH. The results were column-plotted using GraphPad Prism 8.0 software.

2.6 Electron microscopy

Glutaraldehyde (3%) in 0.2 M sodium cacodylate was used to fix liver tissue samples. The samples were washed three times with PBS. After dehydration using serial concentrations of ethanol, the cells were embedded in Epon. The images were acquired with a transmission electron microscope (JEM-1200; Jeol Ltd.) at 80 kV. The autophagic vacuole was identified as double-membrane-bound vesicle within residual. The quantitative morphometric analysis was also done independently and blindly by two experienced researchers under electron microscopy.15

2.7 Liver function test

Serum levels of alanine aminotransferase (ALT), aspartate aminotransferase (AST) and serum total bilirubin (TBIL) were measured with an AU400 automatic biochemical analyser.

2.8 Haematoxylin-Eosin (H&E) Staining

Livers were fixed in 10% buffered formalin, dehydrated with graded ethanol and embedded in paraffin for sectioning. Five-micron paraffin sections were mounted on glass slides, rehydrated with distilled water and stained with haematoxylin and eosin for light microscopy examination. Routine H&E staining was performed as described previously.16 The staining results were assessed independently and blindly via two experienced pathologists under light microscope.

2.9 Quantitative real-time RT-PCR

Guided by the manufacturer's instructions, an RNeasy Mini Kit (QIAGEN) was used to isolate total RNA. The RT-PCR assay was performed as described previously.16 GAPDH gene was used as an internal standard. The list of primers is shown in Table S2 online.

2.10 Statistical analyses

Statistical significance was calculated via Student's t test (SPSS 19.0; SPSS Inc). Quantitative variables are shown as the mean ± SD on the basis of at least three separate experiments. P < .05 was considered significant.

3.1 Morphological characteristics of the liver

In the normal control group (Group I), the livers were rosy and smooth with intact lobule structure and no adhesions to surrounding tissues (Figure 1A). In the ACLF group (Group II) and the LV-Mfn2 negative control group (Group IV), liver tissues were congested and hard, with severe adhesions to surrounding tissues and small nodules and granules on the surface (Figure 1B,D). In the LV-Mfn2 (Group III) and LV-Mfn2 + Rap groups (Group V), the surfaces of livers were smoother, softer in texture and less adherent to the surrounding tissue (Figure 1C,E). The summary for the detailed morphological characteristics of liver is shown in Figure 1F.

3.2 Characteristics of HE staining in the liver

Compared with the normal control group (Group I), other groups showed various degrees of damage. In the ACLF group (Group II) and the LV-Mfn2 negative control group (Group IV), the lobular structure collapsed, with a large number of fibrous tissues proliferated and pseudolobules formed. The hepatocytes showed massive fused necrosis, with the necrotic cells disappearing and the surrounding inflammatory cells infiltrated substantially (Figure 2B,D). In the LV-Mfn2 group (Group III), the hepatic lobule structure was basically intact, with occasional pseudolobules formed, showing small amounts of necrosis, cell balloon-like changes or vacuolar degeneration (Figure 2C). The LV-Mfn2 + Rap group (Group V) had a milder degree of fibrosis, intact hepatic lobule structure and occasional small focal necrosis; hepatic cord and hepatic sinus structure were basically intact (Figure 2E). The pathological results showed that Mfn2 overexpression had a protective effect on ACLF. With Mfn2 overexpression, the addition of rapamycin significantly reduced the degree of ACLF injury and improved the liver status of ACLF. The summary for the detailed histological characteristics in the liver is shown in Figure 2F.

3.3 Effect of Mfn2 on liver function in ACLF

As shown in Table 1, compared with the normal control group (Group I), the levels of ALT, AST and TBIL»in the other groups increased, indicating that the liver function of each treatment group (Groups II–V) was damaged. The ALT, AST and TBIL in the LV-Mfn2 group (Group III) were lower than that of the ACLF group (Group II) and the LV-Mfn2 negative control group (Group IV) (P < .05), suggesting that Mfn2 overexpression attenuated the degree of liver injury in ACLF. The liver function indexes of the LV-Mfn2 + Rap group (Group V) were significantly better than those of the other treatment groups (P < .05), suggesting that rapamycin-activated autophagy during Mfn2 overexpression significantly improved liver function in ACLF.

3.4 Effect of Mfn2 on apoptosis in ACLF

In the ACLF animal and hepatocyte autophagy models, for assessing apoptosis with Mfn2 overexpression, the expression of Bcl-2 and Bax was measured using RT-PCR analysis. Bax levels were significantly lower in the Mfn2 overexpression group. In addition, Mfn2 significantly increased Bcl-2 levels compared with the ACLF group. Compared with the Mfn2 overexpression group, combined Mfn2 and rapamycin administration further decreased apoptosis (Figure 3).

3.5 The effect of Mfn2 on cell autophagy in ACLF

Electron microscopy was used as the gold standard for ultrastructure. In the ACLF animal model, under the electron microscope, the normal control group (Group I) had normal liver cell morphology, with rounded nuclei, scattered nuclear inclusions, clear nuclear membranes, as well as abundant mitochondria, endoplasmic reticulum and other organelles in the cytoplasm, without autophagosomes (Figure 4A). The ACLF model group (Group II) and the LV-Mfn2 negative control group (Group IV) showed severe apoptosis with irregular shape and pyknosis of the nuclei. A large number of autophagosomes or autophagic lysosomes were observed (Figure 4C). Compared with Groups II and IV, the LV-Mfn2 group (Group III) had relatively intact nuclei, mitochondria and endoplasmic reticulum, with less apoptosis and more autophagosomes, suggesting that Mfn2 overexpression promoted autophagy so as to protect hepatocytes. The LV-Mfn2 + Rap group (Group V) had more complete hepatocyte structure than did the LV-Mfn2 group (Group III), with a large amount of autophagosomes or autophagic lysosomes (Figure 4E). Furthermore, in the LV-Mfn2 + Rap group (Group V), the typical phenomenon of apoptosis can be seen when autophagy is excessive (Figure 4E). Quantitative analysis of electron microscopy images captured is shown in Figure 4G.

We also observed mitochondrial morphological changes under electron microscopy (Figure 4F). Compared with the ACLF group, autophagosomes mainly appeared in the cytoplasm, and there was no significant difference for the count of mitochondria enveloped by autophagic vesicles after LV-Mfn2 treatment in ACLF.

In the ACLF animal model, Western blotting analyses showed that the expressions of LC3-II/LC3-I and Atg5 were greater in the LV-Mfn2 group (Group III) than in the ACLF group (Group II). In the LV-Mfn2 + Rap group (Group V), there were further significantly greater expressions of LC3-II/LC3-I and Atg5. Meanwhile, there were also less expressions of Mfn2 and P62 in the ACLF group (Group II) than in the normal control group (Group I) (Figure 5A). In the hepatocyte autophagy model, there were the same effects of Mfn2 on autophagy via RT-PCR analysis (Figure 5C).

Meanwhile, the Western blots were also performed with mitochondrial fractions in the ACLF animal model. Compared with the ACLF group, there was no significant difference for the expressions of LC3 II/I and P62 in the LV-Mfn2 group (Figure 5D). All these data indicated that Mfn2 mainly promotes autophagy instead of mitophagy during ACLF.

3.6 Mfn2 enhances cell autophagy via the PI3K/AKT/mTOR signalling pathway

Next, in the ACLF animal model, Western blotting analyses showed that Mfn2 significantly decreased the expression of phosphorylated-PI3K (p-PI3K), phosphorylated-Akt (p-Akt) and phosphorylated-mTOR (p-mTOR) (Figure 6A). Given that rapamycin induces autophagy by inhibiting mTOR expression, we believe that like rapamycin, Mfn2 induces autophagy via the PI3K/AKT/mTOR signalling pathway. In the hepatocyte autophagy model, there were similar results of Mfn2 on autophagy via the PI3K/AKT/mTOR signalling pathway, according to RT-PCR analysis (Figure 6C).