Effect of HNK on the hepatic biomarkers in the tested mice
Hepatic biomarkers of the mice were detected by commercial test kits. The results showed that HNK had no effect on weight loss and hepatic index of mice fed by MCD diet ((Fig. 1a and b). Compared with MCS group, serum ALT, AST and liver TG, MDA levels in MCD group were significantly increased. After HNK treatment, serum ALT activity, liver TG and MDA content decreased remarkably, AST level was reduced similarly, but there was no significant difference compared with MCD group. In addition, the levels of liver TG, MDA and serum AST in OCA group were obviously lower than those in MCD group (Fig. 1c–f).
Effect of HNK on the hepatic biomarkers in the tested mice. Mice were fed with a methionine-choline sufficient (MCS) diet, a methionine-choline deficient diet (MCD), MCD plus HNK (20 mg/kg/d) or OCA (6.5 mg/kg/d) for 8 weeks. Changes in the (a) Body weight and (b) Hepatic index. Serum levels of (c) ALT, (d) AST; Liver levels of (e) TG, (f) MDA. Data were expressed as mean ± SD (n = 7). ***p < 0.001 versus MCS group; #p < 0.05, ##p < 0.01 versus MCD group. MCD, methionine- and choline-deficient diet; MCS, methionine- and choline-sufficient diet; HNK, honokiol; OCA, obeticholic acid.
HNK attenuated hepatic steatosis, inflammation and fibrosis of NASH mice
Hepatic lipid accumulation, steatosis, inflammation and fibrosis of the mice were analyzed by Oil red O staining, HE staining and Masson staining (Fig. 2). Compared with MCS group, the liver cells of MCD mice were disordered and showed ballooning changes, with a large number of lipid droplets and fat vacuoles, accompanied by inflammatory cell infiltration and fibrosis. After treatment with HNK and OCA, the number of lipid droplets and fat vacuoles in liver cells were reduced significantly, the degree of inflammatory cell infiltration and fibrosis decreased, and the liver tissue morphology was closer to MCS group.
The hepatic histological changes of the tested mice evaluated by Oil Red O, HE and Masson staining (200 ×). HE, hematoxylin and eosin; MCD, methionine- and choline-deficient diet; MCS, methionine- and choline-sufficient diet; HNK, honokiol; OCA, obeticholic acid.
Effects of HNK on serum lipids of NASH mice
The principal component analysis (PCA) showed the separation trend of serum lipid composition in each group. As shown in Fig. 3a, there was an obvious separation trend between the MCS group and the MCD group, indicating that MCD diet caused significant changes of serum lipid composition in mice. Compared with MCD group, the sample distribution of HNK group showed no change.
Model analysis of serum lipid metabolic profile in the tested mice (n = 7). (a) PCA score plot. (b) OPLS-DA score plot between MCS and MCD groups (R2X(cum) = 0.706,Q2(cum) = 0.967). (c) OPLS-DA score plot between MCD and HNK groups (R2X(cum) = 0.458,Q2 (cum) = 0.502). (d) Volcano plots of lipid metabolites between MCS and MCD groups. (e)Volcano plots of lipid metabolites between MCD and HNK groups. (f) Analysis of lipid metabolism pathway based on the significant differences of MCD verus HNK groups. (g) Lipid metabolic pathway analysis of differential lipid species between HNK and MCD groups (www.kegg.jp/kegg/kegg1.html). MCD, methionine- and choline-deficient diet; MCS, methionine- and choline-sufficient diet; HNK, honokiol.
To further clarify the differential lipids among each group, we analyzed the serum lipid composition of MCS, MCD and HNK groups using orthogonal partial least squares discriminant analysis (OPLS-DA). As shown in Fig. 3b, MCS and MCD group were significantly separated. R2Y = 0.997, Q2 = 0.967, and both R2Y and Q2 were greater than 0.4, indicating that the OPLS-DA model was effective. Figure 3c showed that HNK group and MCD group were significantly differentiated, with R2Y = 0.998 and Q2 = 0.719, indicating that the OPLS-DA model was effective.
Variable importance in projection (VIP) (VIP ≥ 1) in OPLS-DA analysis and fold change (fold change ≥ 2 or ≤ 0.5) of univariate analysis were used to select differential lipid metabolites. As shown in Fig. 3d, MCD group showed 448 different lipid metabolites compared with MCS group, including 24 fatty acyls (FA), 297 glycerophospholipids (GP), 49 sphingolipids (SP), 15 sterol lipid (ST), 63 glycerides (GL). Compared with MCD group, HNK group showed 34 different lipid metabolites, 18 of them (including 17 FA and 1 GP) were down-regulated, and 16 of them (including 2 ST, 10 GP, 1 FA, 1 prenol lipids, 1 SP and 1 GL) were up-regulated (Fig. 3e).
In order to intuitively observe the change of the relative content of different lipid metabolites, we standardized the 34 screened significantly different lipid metabolites between MCD and HNK groups, and showed in the heat map (Fig. 3f). Compared with MCS group, 3 eicosanoids (12,13-EPOME, 9,10-DIHOME and 14(S)-HDHA) and 3 FFAs (16:2, 18:3 and 22:6) were significantly up-regulated, while TxB3, PA(18:0_22:6), SE(28:1_22:6) and 8 PEs (20:1_18:0, O-18:0_16:0, O-20:0_16:0, O-16:1_16:0, O-20:0_22:6, O-15:1_22:6, P-18:0_22:6 and P-16:0_16:0) were down-regulated in MCD group. HNK treatment significantly back-regulated the levels of these lipid metabolites. Furthermore, HNK treatment also increased the levels of taurolithocholic acid-3-sulfate (TLCA-3-S), PE(20:2_18:0), Coenzyme Q10, TG(O-20:0_16:0_18:1) and Cer(t18:0/24:0), and decreased the levels of PC(20:4_22:6) and 11 FFAs (14:0, 16:1, 17:1, 18:1, 19:1, 20:2, 20:3, 22:4, 22:5, 24:6 and 16:3), as compared to MCD mice.
KEGG pathway (www.kegg.jp/kegg/kegg1.html) enrichment analysis was used to determine the metabolic pathways related to the 34 different lipids mentioned above. Figure 3g showed that fatty acid biosynthesis and unsaturated fatty acid biosynthesis pathways were related to HNK alleviating lipid metabolism disorders in MCD mice.
Effects of HNK on serum BAs levels and hepatic mRNA expression of BAs metabolism-related genes in NASH mice
The serum levels of 50 BAs were shown in Table S2 and the significant changes of serum BAs levels in mice were shown in Table 1. Compared with MCS group, the levels of 23-nor-deoxycholic acid (23-DCA), hyocholic acid (HCA), ursocholic acid (UCA), hyodeoxycholic acid (HDCA), deoxycholic acid (DCA), 7-ketodeoxycholic acid (7-KDCA), 3β-ursodeoxycholic acid (3β-UDCA) and 3-oxodeoxycholic acid (3-oxo-DCA) were increased in MCD group. HNK treatment markedly decreased the levels of 23-DCA, HDCA, glycocholic acid (GCA) and taurodeoxycholic acid (TDCA), as compared with MCD group.
The mRNA expression of BAs metabolism-related genes in mice liver were shown in Fig. 4. Compared with MCS group, the mRNA expression of CYP7A1, CYP27A1, Bsep, Mrp2, Ntcp and Oatp1b2 in MCD group were significantly decreased. HNK treatment markedly increased Oatp1b2 mRNA expression of MCD mice (Fig. 4f), whereas there were no significant effect on the mRNA expression of CYP7A1, CYP27A1, Bsep, Mrp2 and Ntcp (Fig. 4a–e).
Effects of HNK on the mRNA expression of BAs metabolism-related genes in mice liver. The mRNA expression of CYP7A1(a), CYP27A1(b), Bsep(c), Mrp2(d), Ntcp(e) and Oatp1b2(f) were analyzed by RT-qPCR. The PCR products were normalized to β-actin. Data were expressed as mean ± SD (n = 7). ***p < 0.001, **p < 0.01, *p < 0.05 versus MCS group; #p < 0.05 versus MCD group. MCD, methionine- and choline-deficient diet; MCS, methionine- and choline-sufficient diet; HNK, honokiol.
Effects of HNK on the composition of GM in MCD mice
The Exponential dilution curve showed that the amount of 16Sr DNA sequencing data in this experiment was progressive and reasonable, and the sequencing depth was reliable (Fig. 5a). The statistical analysis of the Alpha diversity index of different samples under the 97% consistency threshold were shown in Table S3. There were no significant differences in Chao1, ACE, Simpson and Shannon indexes of GM among the three groups, indicating that HNK had no significant effect on the diversity of GM. Beta diversity analysis was applied to further study the composition of GM in each group, and non-metric multi-dimensional scaling (NMDS) showed significant differences in the composition of GM among the three groups (Fig. 5b).
Effects of HNK on the composition of gut microbiota in the tested mice (n = 7). (a) Exponential dilution curve. (b) Non-metric multidimensional scaling (NMDS) on the OTU level. (c) Microbiota composition at the phylum level. (d) Microbiota composition at the family level. (e) Changes in gut microbiota of mice at the genus level. (f) LDA bar plot of LEFSe analysis on gut microbiota. MCD, methionine- and choline-deficient diet; MCS, methionine- and choline-sufficient diet; HNK, honokiol.
The composition of GM of the tested mice at phylum, family and genus levels were shown in Fig. 5c–e. Compared with MCS group, the abundance of Firmicutes and Turicibacter were markedly increased, and the abundance of Eggerthellaceae was significantly decreased in MCD group. MCD mice treated by HNK significantly increased the relative abundance of Ruminococcaceae, Caulobacteraceae and Brevundimonas, and the abundance of Firmicutes and Dubosiella significantly decreased. Based on LEFSe analysis, species with LDA > 4 were considered as the set value, and the microbiota with statistical differences among three groups were screened out and showed in Fig. 5f.
Spearman correlation analysis
Spearman correlation analysis was used to explore the relationship between the abundance of GM and the contents of serum lipid or BAs significantly changed by HNK treatment in MCD mice. At phylum level, Firmicutes was negatively correlated with TLCA-3-S and TxB3 (Fig. 6a). At family level, Ruminococcaceae was negatively correlated with the levels of 14 FFAs (16:1, 16:2, 14:0, 18:1, 17:1, 22:4, 20:3, 18:3, 20:2, 19:1, 16:3, 22:6, 22:5, 24:6) (Fig. 6a), Caulobacteraceae was negatively correlated with 9 FFAs (18:1, 17:1, 22:4, 20:3, 18:3, 16:3, 22:6, 22:5 and 24:6) and positively related with 6 PEs (O-16:1_16:0, P-16:0_16:0, O-18:0_16:0, O-20:0_16:0, O-20:0_22:6 and P-18:0_22:6) (Fig. 6a). At genus level, Dubosiella was positively related with 4 FFAs (18:1, 17:1, 22:4, 18:3 and 24:6) and negatively correlated with multiple 4 PEs (20:1_18:0, 20:2_18:0, P-16:0_16:0 and O-16:1_16:0), Brevundimonas was negatively correlated with 2 oxidized lipids (12, 13-EPOME and 9,10-DiHOME) and 9 FFAs (18:1, 17:1, 16:3, 20:3, 22:4, 18:3, 22:5, 22:6 and 24:6) (Fig. 6a). At phylum level, Firmicutes was positively correlated with 23-DCA, 7-KDCA, DCA, UCA, 3-oxo-DCA, HCA, 3β-UDCA (Fig. 6b). At family level, Caulobacteraceae was negatively correlated with 23-DCA and TDCA, Ruminococcaceae was negatively correlated with 23-DCA (Fig. 6b). At genus level, Brevundimonas were negatively correlated with 23-DCA and TDCA, Dubosiella was positively related with 23-DCA, HCA, 7-KDCA, 3β-UDCA (Fig. 6b).
Spearman correlation analysis between gut microbiota and serum lipids or BAs in MCD mice. (a) Correlation between gut microbiota and serum lipids. (b) Correlation between gut microbiota and serum BAs. *Indicates that the p value of the correlation coefficient significance test is < 0.05, and **Indicates that the p value is < 0.01. MCD, methionine- and choline-deficient diet; MCS, methionine- and choline-sufficient diet; HNK, honokiol.