UPLC-Q/TOF-MS-Based Serum Metabolomics Reveals Hypoglycemic Effects of Rehmannia glutinosa, Coptis chinensis and Their Combination on High-Fat-Diet-Induced Diabetes in KK-Ay Mice

Abstract

Diabetes is a worldwide severe health issue which causes various complications. This study aimed to evaluate the hypoglycemic effects of Rehmannia glutinosa (RG), Coptis chinensis (CC) alone and their combination on high-fat-diet-induced diabetes in mice via biochemical assays and UPLC-Q/TOF-MS-based serum metabolomic analysis. Diabetic KK-Ay mice were induced by high-fat diet and treated for eight weeks, separately with RG, CC and their combination and the positive control drug metformin. Administration of RG and CC alone, and their combination could decrease the fasting blood glucose level, ameliorate the tolerance of glucose, and recover the levels of total cholesterol (TC), triglycerides (TG), high-density lipoprotein cholesterol (HDL-C), and low-density lipoprotein cholesterol (LDL-C) in sera of diabetic mice. Orthogonal partial least squares discriminant analysis (OPLS-DA) on serum metabolomes revealed that 79 ESI⁺ and 76 ESI^- metabolites were changed by diabetes mellitus (DM) compared to the normal control. Heatmaps on these diabetes-related metabolites showed that CC and RG/CC were clustered closer with the normal control, indicating that they had the better antidiabetic effects at the metabolite level. Fifteen of the differential metabolites in DM serum were annotated and their related metabolic pathways were lipid metabolism. These data suggested that RG and CC alone and in combination treatment had the antidiabetic activity in lowering glycemia and improving lipid metabolism. UPLC-Q/TOF-MS-based metabolomics shed light on the differential metabolite effects of RG and CC in DM treatment. However, it should be noted that some differential metabolites were possibly generated or not detected due to our groupwise run order, which possibly contributed to or covered the group difference in our experiment. They need to be further discriminated in the future work.

Keywords: Coptis chinensis; Rehmannia glutinosa; UPLC-Q/TOF-MS; diabetes; high-fat diet; hypoglycemic effect; metabolomics.

Figure 1

Changes of body weight (A) and fasting blood glucose levels (B) in different treatments of mice over 56 days of the experiment. ^## p < 0.01, and ^### p < 0.001 meant that the FBG value of the indicated group showed significant difference compared with the normal control group (C); * p < 0.05, ** p < 0.01, and *** p < 0.001 meant the FBG value of the indicated group was significantly different from the diabetic model control group (M). MET, RG and CC, RG/CC here and the below indicated diabetic mice groups administered separately with metformin (MET), decoctions of Rehmannia glutinosa (RG), Coptis chinensis (CC) and their combination (RG/CC).

Figure 2

Results of a glucose tolerance test in diabetic KK-Ay mice performed at the end of 56 days of treatment. (A) blood glucose level after glucose loading; (B) AUC: area under the curve of OGTT; (C) iAUC: incremental area under the curve of OGTT. ^# p < 0.05, ^## p < 0.01 and ^### p < 0.001 meant significant difference of the measurement in the indicated group compared with the normal control group (C); * p < 0.05 and ** p < 0.01 meant significant difference of the measurement in the indicated group compared with the model control group (M).

Figure 3

PCA line score plots of different injections of quality control (QC) sample. X-axis represented the run order of QC sample; Y-axis represented standard deviation (A,B) and Hotelling’s T2 range (C,D), separately. (A,C) for ESI⁺ mode; (B,D) for ESI⁻ mode.

Figure 4

PCA and OPLS-DA score plots derived from UPLC-Q/TOF-MS profiling of the sera metabolomics of control and model groups. ESI⁺ mode: (A,B); ESI⁻ mode: (C,D); NC represented the number of components. R2X and R2Y were cumulative variation of all R2Xs and R2Ys, separately. Q2 was the cumulative predicted fraction. p_CV-ANOVA values for OPLS-DA models (B,D) were 5.20008 × 10⁻⁴ and 2.14075 × 10⁻⁵, respectively.

Figure 5

Different groups of serum metabolites visualized using PCA and OPLS-DA score plots. ESI⁺ mode: (A,B); ESI⁻ mode: (C,D). NC represented the number of components. R2X and R2Y were cumulative modelled variation of all R2Xs and R2Ys, separately, and Q2 was the cumulative predicted fraction. p_CV-ANOVA values for OPLS-DA models (B,D) were 4.42718 × 10⁻⁸ and 0.0105384, respectively.

Figure 6

Hierarchical clustering analysis of HFD-induced differential metabolites among different treatments. The heatmaps were generated in MetaboAnalyst 4.0 (http://www.metaboanalyst.ca/) using the differential biomarkers of importance identified between the normal control group (C) and diabetic model control group (M). (A) ESI⁺ mode; (B) ESI⁻ mode. The results showed that CC and its combination with RG were clustered closer to the undiabetic control group, indicating that CC served as a sovereign drug and had a better modulation on diabetes-related metabolites in serum than RG alone and the positive drug metformin.

Figure 7

Venn diagrams showing unique or overlapped modulation of RG, CC, and RG/CC on HFD-induced biomarkers in sera of diabetic mice. The number in () was the modulated variables of the herbal medication on HFD-induced biomarkers compared to the model group. The numerical values in the diagrams depicted the metabolites that are unique to or shared between three medications. (A) ESI⁺ mode; (B) ESI⁻ mode.

Figure 8

SUS-plots showing the shared and unique metabolites between herbal treatments. (A–C): based on OPLS-DA models under ESI⁺ mode. (D–F): based on OPLS-DA models under ESI⁻ mode. MRG: Model vs. RG; MCC: Model vs. CC; MRG/CC: Model vs. RG/CC. The displayed metabolites were selected based on the |p(corr)| ≥ 0.5 and the jack-knifed confidence interval and showed the significant difference (adjusted p < 0.5) between at least one compared treatment and the model control group. To identify the herbal treatment-unique metabolites, the |p(corr)| threshold of such metabolites was ≤0.1 in its opposite OPLS-DA model.