PF attenuates the pathological changes in liver injury generated by APAP

By macroscopic pathological examination, we observed that the APAP mice began to manifest body weight loss compared with the control group after APAP was injected, while the physiological changes were similar in each group in the first 5 days (Fig. 1C). Besides, we also measured the liver index and spleen index of mice, and the results demonstrated that both of them increased in APAP group (p < 0.01). By contrast, the HPF and LPF groups could drop them (p < 0.01), meaning the swelling of spleen and liver was reduced by PF (Fig. 1A, B).

Fig. 1

Biochemical and pathological parameters changes by PF on APAP-induced mice. The levels of liver index (A) and spleen index (B) are shown (n = 8). C Weight of mice during the whole experiment (n = 8). D–H Serum levels of ALT, AST, ALP, TBIL, and γ-GT in mice (n = 8). I Necrosis index of mice in each group. J H&E staining in mice (×100, ×400) (n = 3). The red arrow points to the area of hepatic lesion. Data are presented as mean ± SD. *p < 0.05 and **p < 0.01 versus control; #p < 0.05 and ##p < 0.01, versus APAP

If hepatocytes are poisoned, infected, necrotized, and inflamed, as a sensitive indicator of liver injury, serum transaminases will be released into the blood, i.e., ALT, AST, and TBIL [27]. We further measured unique indicators of liver functions. Compared with control group, the serum levels of these indices evidently rose in APAP group (p < 0.01), noticing that the DILI model had been successfully established. By contrast, PF at high and middle doses significantly diminished the levels of ALT, ALP, γ-GT, TBIL, and AST (p < 0.01) compared with the APAP group (Fig. 1D–H). In addition to TBIL (p > 0.05), the LPF group (50 mg/kg) was sensitive to the other indicators and effectively reduced their contents (Fig. 1G).

Similar to the biochemical indices, the pathological results showed that hepatic lobules in the APAP group were disorderly arranged, with a large number of necrotic hepatocytes distributed around the portal area and infiltrated with inflammatory factors, besides, focal necrosis were widely observed in the field of view (× 100). Compared with the APAP group, hepatic lobules in the MPF and HPF group were structured and tightly arranged, and the hepatic edema and inflammatory infiltration were improved by PF, especially at the high dose (Fig. 1J). In addition, the necrotic index of hepatic tissues was calculated. The liver injury was relieved after PF treatment (p < 0.01) (Fig. 1I). In brief, PF exerted a protective effect on the liver and apparently slowed down the further progress of liver injury induced by APAP.

PF inhibits ROS to improve the antioxidant power of APAP model mice

Some of the crucial signal molecules involved in the oxidative stress response are ROS and antioxidant compounds. In APAP-induced liver injury, the excessive accumulation of ROS disrupts cellular homeostasis and further triggers oxidative stress and mitochondrial dysfunction [28]. To elucidate the capacity of PF treatment on the oxidative status, we evaluated the levels of oxidation-related indicators, such as GSH, SOD, and MDA, at first. Compared with control group, the contents of SOD and GSH were significantly abated, if processed with APAP (p < 0.01), while PF restored the antioxidant capacity of the APAP mice at different concentrations, especially at 200 mg/kg (Fig. 2A, B). In contrast to the role of GSH and SOD, changes in the MDA levels occur in response to the oxidative capacity of the body [29]. For MDA, the evident increase occurring in the APAP group and the onset of oxidative effects were mitigated by PF at different doses (p < 0.01) (Fig. 2C). These indicators suggested that PF remitted the occurrence of oxidative stress in the APAP model.

Fig. 2

Effects of PF on oxidative stress with APAP stimulation. The levels of GSH (A), SOD (B), and MDA (C) are shown(n = 6). D Mean density of ROS. E Immunofluorescence of ROS (n = 3). F Western blotting images of CYP3A4 (n = 3). G Relative protein expression of CYP3A4. H, I The expression of CYP2E1 by Immunohistochemistry (×200, ×400) (n = 3). J Transmission electron microscope (n = 3). Data are presented as mean ± SD. *p < 0.05 and **p < 0.01, vs. control; #p < 0.05 and ##p < 0.01, vs. APAP

Concomitantly, ROS expression was observed and measured by immunofluorescence (Fig. 2E). The IF of ROS showed that the fluorescence expression of ROS in MPF and HPF groups was prominently choked compared to APAP group, suggesting that PF at high and middle doses significantly inhibited ROS expression in hepatocytes (p < 0.01) (Fig. 2D). In addition, NAPQI, the metabolite of APAP by the cytochrome P450 enzymes CYP2E1 and CYP3A4, induces major cell death and mitochondrial dysfunction and then leads to damage in the process of accumulation, which means that increased expressions of cytochrome P450 enzymes can be seen in the course of injury by APAP overdoses [30]. Similar results were observed in our experiment. IHC revealed that the positive expression levels of CYP2E1 were higher in the APAP group than in the control and PF groups (Fig. 2I). By contrast, the reactions of CYP2E1 in APAP group predominantly aggravated (p < 0.01). However, given PF at doses of 100 and 200 mg/kg overturned overtly the APAP-stimulated oxidative stress compared with the APAP group (p < 0.01) (Fig. 2H). Meanwhile, the expression of CYP3A4 was detected by western blotting. The expression of CYP3A4 increased after APAP injection (p < 0.05), while this effect was notably reversed by PF at dose of 200 mg/kg (p < 0.01) (Fig. 2G). Moreover, the results of TEM exhibited that the mitochondrial structure in APAP group was fuzzy, and the ridge was not clear. The matrix electron density increased and the rough endoplasmic reticulum expanded. After high dose treatment with PF, the mitochondrial structure was complete and the matrix was uniform, which showed a uniform gray structure under the electron microscope. Moreover, the morphological structure of the rough endoplasmic reticulum was restored (Fig. 2J). This evidence explained that PF blocked the generation of ROS, improved the activity of cytochrome P450 enzymes, and enhanced the antioxidant capacity in mice against damage produced by APAP.

PF refrains the progress of apoptosis on APAP-induced liver injury

Along with variation in the mitochondrial outer membrane permeability caused by ROS accumulation, the Bcl-2 protein family releases pro-/anti-apoptotic factors to control the apoptosis process and then activates caspase 9 to further release caspase 3, triggering apoptosis [31]. The TUNEL staining, visually reflected the occurrence of apoptosis, is exhibited in Fig. 3A. Compared with the control group, plenty of positive apoptotic factors gathered in the liver injury area in APAP group. Moreover, compared with APAP group, the positive expression rate of apoptosis in the treatment group was evidently reduced (p < 0.01), suggesting that PF improved apoptosis in APAP-induced liver injury (Fig. 3D). To further confirm the role of apoptosis in PF treatment for liver damage, IHC was used to examine the positive expression of apoptotic characteristic factors. The positive expression was concentrated in the hepatic portal area, and the expression of caspase 3 in mice livers stimulated by APAP was more severe, suggesting that the process of liver injury was accompanied by apoptosis. Likewise, the caspase 9 expression displayed low levels in control group and high expression in APAP group. PF pretreatment abated the positive expression of pro-apoptotic proteins (caspase 9 and caspase 3) with an optimal effect appearing at a dose of 200 mg/kg (Fig. 3B, C).

Fig. 3

Anti-apoptotic effects of PF in APAP-induced hepatic damage. A The apoptotic expression is shown by TUNEL staining (×100, ×400). The expression of caspase 9 (B) and caspase 3 (C) is shown in the different groups (×100, ×400). D The TUNEL positive rate in mice. E Western blotting images of BAD, BCL-2, and BAX in mice. F Western blotting images of caspase 9 and caspase 3 in mice. The relative protein expression of BAD (G), BCL-2 (H), BAX (I), caspase 9 (K), and caspase 3 (L) is depicted. J Ratio of BCL-2/BAX in mice. The red arrow points to positive expression areas. Data are presented as mean ± SD in each three samples in group. *p < 0.05 and **p < 0.01, versus control; #p < 0.05 and ##p < 0.01, versus APAP

The relative expression of pro-/anti-apoptotic proteins (BAD, BAX, caspase 9, caspase 3, and BCL-2) were assessed by western blotting (Fig. 3E, F). The relative protein expressions of BAD, BAX, and caspase 9 were similar. All of them showed that results were elevated upon the administration of APAP, while the relative expressions of them were distinctly lessoned upon PF exposure (p < 0.05 or p < 0.01) (Fig. 3G, I, K). Caspase 3 was the most sensitive indicator of apoptosis, and different doses of PF were effective in suppressing the increased expression caused by APAP (p < 0.01) (Fig. 3L). Although the result for BCL-2 displayed no difference between the control group and APAP group, an obvious decline in the ratio of BCL-2/BAX was found in APAP group (p < 0.01) compared with the control group, whereas an evident rise occurred in the HPF group (p < 0.01) (Fig. 3H, J). This result indicates that PF promoted the activity of anti-apoptotic proteins and terminated the expression of pro-apoptotic proteins to exert anti-apoptotic effects.

Activation of autophagy by PF protects liver from APAP damage

Relevant studies suggested that autophagy selectively removes impaired mitochondria and NAPQI-protein adducts, thereby blocking the progression of liver damage generated by APAP [32]. TEM was used to observe the number of autophagosomes and the structure of organelles in hepatic tissues (Fig. 4A). Compared with other groups, the rough endoplasmic reticulum in APAP group was extensively swollen and only a few autophagosomes were found. At the same time, apoptosis, nuclear shrinkage, and chromatin concentration were also observed in hepatocytes. Relatively, the number of autophagosomes in PF groups was markedly reinforced, the morphology and structure of hepatocytes were intact, and the chromatin was evenly distributed. LC3, as a downstream manifestation molecule of the autophagy process represents the onset of autophagy by its conversion from type I to type II [33]. Moreover, as the pivot, p62 is closely related to the regulation of autophagy and mitochondrial autophagy [34]. Thus, IHC and WB were used to visualize changes in proteins (LC3 and p62) associated with autophagy regulation (Fig. 4B, C). The results of IHC displayed that PF increased the positive expression of LC3 (p < 0.01), exerting the promoting effect for autophagy (Fig. 4D). On the other hand, the expression of p62 was remarkably boosted in APAP group, while this was alleviated by PF treatment (p < 0.01) (Fig. 4E). Similar to IHC, the results of WB showed an evidently inhibition of p62 expression in the MPF and HPF groups (p < 0.01). For LC3, the ratio of type II to type I was obviously reduced by APAP (p < 0.01), and PF effectively mitigated this process (p < 0.05) (Fig. 4F, G).

Fig. 4

Changes in autophagy representative indices modulated by PF. A Transmission electron microscopy of autophagosomes. B Immunohistochemistry of LC3 and p62 in different groups (×200). C Western blotting images of LC3 and p62. The mean density of LC3 (D) and p62 (E) in immunohistochemistry. F Ratio of LC3II/LC3I in mice. G Relative protein expression of p62. The red arrow points to autophagosomes. Data are presented as mean ± SD in each three samples in group. *p < 0.05 and **p < 0.01, versus control; #p < 0.05 and ##p < 0.01, versus APAP

Furthermore, the immunofluorescence results showed a weak fluorescence in the APAP group and a strong fluorescence in the MPF and HPF groups (p < 0.01) (Fig. 5A, C). The results of WB suggested that both ATG5 and ATG7 were depleted in APAP group, compared with control group (p < 0.01). Instead, PF reversed the effect of APAP and promoted the expression of them (p < 0.05 or p < 0.01) (Fig. 5D, E). We also measured mRNA expression levels to determine autophagic activity changes in APAP-induced mice. The levels of ATG5, ATG7, and BECN1 in the APAP group decreased (p < 0.05 or p < 0.01) compared with the control group. More importantly, HPF improved the mRNA expression pattern in the APAP model (p < 0.05), despite the fact that the effect was hidden in ATG7, and only the LPF group showed an alleviation (Figure S3). These results illustrated that PF effectively remitted the inhibition of autophagy generated by APAP.

Fig. 5

Activation of autophagy by PF on APAP-induced mice. A Immunofluorescence of LC3. B Western blotting images of ATG7 and ATG5. C Mean density of LC3 in immunofluorescence. Relative protein expression of ATG7 (D) and ATG5 (E). Data are presented as mean ± SD in each three samples in group. *p < 0.05 and **p < 0.01, versus control; #p < 0.05 and ##p < 0.01, versus APAP

Transcriptomics reveals in-depth mechanisms of PF on APAP-induced hepatic injury

The current findings indicated that the amelioration of APAP-induced DILI by PF was closely related to the activation of autophagy, inhibition of oxidative stress, and apoptosis. Yet, the mechanisms by which PF regulated the role of these phenotypes were not clear. Therefore, we sequenced hepatic samples of mice to further obtain the potential mechanism of PF against liver injury caused by APAP. Data preprocessing suggested homogeneous gene expression in the samples and correlation analysis informed good correlation of samples (R2 = 0.858–0.994) (Figure S4A, B). Moreover, the results of the principal component analysis (PCA) showed that the samples within the group were consistent and the classification between the groups was obvious, which manifested the prominent discrepancy among these groups and the subsequent differential gene analysis results were feasible (Figure S4C). In summary, the above results indicated that the sequencing data were reliable.

Quantitative analysis of gene expression for each group was compared and presented in Figure S4D. Different from the control group, 2444 genes were upregulated and 1552 genes were downregulated in the APAP group. A total of 508 different genes with 443 downregulated and 65 upregulated were presented between the HPF group and the APAP group (Fig. 6A, B). The APAP group clustered in oxidative stress and cell death, whereas the clustered genes of HPF group were less in these area under the cluster analysis (Figure S5). For the gene set enrichment analysis (GSEA), PF exerted regulatory effects on several signals and physiological processes, including MAPK signaling pathway, mTOR signaling pathway, apoptosis, and glutathione metabolism (Figure S6). In addition, combining the results of GO analysis and KEGG enrichment, the MAPK pathway showed the higher score and performed simultaneously in all three sets of differential gene analysis (Fig. 6C–F). All comparisons between the HPF group and the control group appeared in Figure S7. Therefore, the results of the transcriptomics analysis suggested that the MAPK signaling pathway played a key role for PF’s activity on drug-induced liver injury by APAP, which further regulated mTOR signal, then activated autophagy, and inhibited oxidative stress and apoptosis.

Fig. 6

Transcriptomic analysis of PF in improving APAP-induced liver injury. Volcano map of the control versus APAP groups (A) and HPF versus APAP groups (B). GO analysis of the control versus APAP groups (C) and HPF versus APAP groups (D). KEGG analysis of control verus APAP groups (E) and HPF versus APAP groups (F)

PF blocks MAPK/mTOR signaling pathway by directly targeted p38

Based on the transcriptomic results, we were intrigued about the importance of MAPK signal in DILI development and PF therapy. To further examine the role of PF on MAPK signaling, molecular docking analyses were performed. The p38-PF complex displayed a great binding capacity with a binding energy of − 32.50 ± 0.71 kJ/mol in contrast to the ERK-PF complex with a binding energy of − 21.65 ± 0.98 kJ/mol (Table S5), which suggested that p38 might be a potential target for PF. PF was bound to the MET265, LEU291, and LEU246 residues via one hydrogen bond, respectively, while LYS295 was connected with two hydrogen bonds to PF. In addition, GLY240 was bound by one carbon hydrogen bond to PF (Fig. 7A). Subsequently, the binding ability of PF to p38 was explored by molecular dynamics simulations. The root mean square deviation (RMSD) results showed that the protein–ligand complex reached the equilibrium after 20 ns, indicating that the entire simulation was stable and reliable (Fig. 7B). Both solvent-accessible surface area (SASA) and radius of gyration (Rg) were reflected the tightness of the complex protein structure during simulation, and our results indicated a decreasing trend in both variables, suggesting an increase in protein tightness structure and a good combination of p38-PF complex (Fig. 7C, D). The protein root mean square fluctuation (RMSF) of the p38-PF complex were less than 0.35 nm, showing a stable protein–ligand binding (Fig. 7E). Besides, the hydrogen bonding analysis suggested that the average hydrogen bond number distribution of p38 to PF was 0.51 (Fig. 7F). Meanwhile, the results of the binding energies of p38 to PF listed in Table 1. The total binding free energy was −79.263 kJ/mol, which indicated a remarkable stability in the p38-PF complex, and the main interactions were given by van der Waals forces and electrostatic energy. These results suggested that PF may target p38 rather than ERK in the MAPK signal cascade and that the binding between PF and p38 was stable.

Fig. 7

Targeting of p38 by PF. A Molecular docking analysis of PF to p38. B RMSD of the p38–PF complex. C SASA of the p38–PF complex. D Rg of the p38–PF complex. E RMSF of the p38–PF complex. F Hydrogen bonding analysis of the p38-PF complex

Table 1 Molecular Mechanics Poisson-Boltzmann Surface Area (MMPBSA) analysis of the p38-PF complex

Furthermore, the results of mRNA expression were consistent with the transcriptome analysis. PF (200 mg/kg) dramatically downregulated the relative expression of MAPK1, MAPK14, and mTOR (p < 0.05 or p < 0.01) (Fig. 8A–C). The western blotting results manifested that phosphorylation of p38 and ERK proteins were significantly activated after APAP overdose injection, whereas MPF and HPF effectively inhibited this change (p < 0.01) (Fig. 8E, F). However, for phosphorylation of mTOR, the inhibition was only shown at a concentration of 200 mg/kg PF (p < 0.05) (Fig. 8H). This suggested that PF acted on MAPK signaling which in turn regulated the phosphorylation of mTOR. Besides, the phosphorylation of ULK1 was inhibited by APAP stimulation, whereas PF memorably reversed this effect and restored the phosphorylation of ULK1, which further activated the downstream autophagy proteins (p < 0.05 or p < 0.01) (Fig. 8G, I). In brief, the results revealed that PF initiated autophagy through blocking MAPK/mTOR signaling to protect hepatocytes.

Fig. 8

Regulation of PF on MAPK/mTOR signaling. A–C Relative gene expression of MAPK14, MAPK1 and MTOR. D Western blotting images of p38, p-p38, ERK, p-ERK, mTOR, and p-mTOR. E, F Ratio of p-p38/p38 and p-ERK/ERKin mice. G Western blotting images of ULK1 and p-ULK1. H, I Ratio of p-mTOR/mTOR and p-ULK1/ULK1 in mice. Data are presented as mean ± SD in each three samples in group. *p < 0.05 and **p < 0.01, versus control; #p < 0.05 and ##p < 0.01, versus APAP