Health effects of kiwi wine on rats: an untargeted metabolic fingerprint study based on GC-MS/TOF

Abstract

Kiwi wine is a popular fermentation product of kiwi fruit in Asian countries. To better understand the potential health effects of kiwi wine, an untargeted gas chromatography-mass spectrometer (GC-MS) approach was taken to assess the metabolic fingerprint of rats after dietary ingestion of kiwi wine. 7 differentially expressed endogenous metabolites from serum and 8 from urine were enriched in carbohydrate metabolism, amino acid metabolism pathway, fat metabolism and other metabolisms and selected from the KEGG. The above results showed that kiwi wine mainly led to a pronounced perturbation of energy metabolism (especially carbohydrate metabolism) during the consumption period. After stopping the supply of kiwi wine 30 days later, 6 and 3 endogenous metabolites from serum and urine respectively were screened and involved in a small part of carbohydrate related amino acid metabolism and fat metabolism, which indicated that the effect of kiwi wine sustained a lasting effect on energy metabolism, amino acid metabolism and lipid metabolism after stopping the supply. Thus, kiwi wine might have a positive function on health associated with the metabolism of its constituents. To the best of our knowledge, this study provides a nutrition field view for the development of the kiwi wine agricultural industry via an untargeted GC-MS metabolomic approach.

Fig. 1. Total design of the experiment. The experiment was taken on rats, which were undivided into two groups (kiwi wine, KWG and control group, CG). The serum and urine samples were collected at the 0, 20^th, 40^th, 50^th days and one month (30 days) later after stopping kiwi wine supply. Metabolites were analyzed by GC-MS/TOF and identified by comparing with database. Data were deeper dig for screened significant metabolites and relative pathways from two aspects (time point and dynamic change) based on statistical model.

Fig. 2. Body weight of controls and kiwi wine rats for 80 days. The body weights of rats were recorded every three days. The whole experiment lasted for 80 days. And before the experiment, there was no significant different between Kiwi wine group and control group as the p value = 0.2622.

Fig. 3. Plots of OPLS-DA and validation of the OPLS-Da model (serum). (A) Day 20 KWG (S2) vs. CG (Sc2); (B) day 40 KWG (S3) vs. CG (Sc3); (C) day 50 KWG (S4) vs. CG (Sc4); (D) day 80 KWG (S5) vs. CG (Sc5). R²X and R²Y are the cumulative modelled variation in the X and Y matrix, respectively; Q² is the cumulative predicted variation in the Y matrix; validation: 7-fold cross validation was used to estimate the robustness and the predictive ability of our model, such permutation test was proceeded in order to further validate the model. After 200 permutations, the low values of Q² (Q_max² > Q²) intercept indicate the robustness of the models, and thus show a low risk of over fitting and reliable.

Fig. 4. Plots of OPLS-DA and validation of the OPLS-Da model (urine). (A) Day 20 KWG (U2) vs. CG (Uc2); (B) day 40 KWG (U3) vs. CG (Uc3); (C) day 50 KWG (U4) vs. CG (Uc4); (D) day 80 KWG (U5) vs. CG (Uc5). R²X and R²Y are the cumulative modelled variation in the X and Y matrix, respectively; Q² is the cumulative predicted variation in the Y matrix; the validation was same with the description in Fig. 3.

Fig. 5. Plots of OPLS-DA and validation of the OPLS-Da model (dynamic analysis of serine). (A) Day 20 KWG (S2) vs. day 0 KWG (S1); (B) day 40 KWG (S3) vs. day 0 KWG (S1); (C) day 50 KWG (S4) vs. day 0 KWG (S1); (D) day 80 KWG (S5) vs. day 0 KWG (S1). R²X and R²Y are the cumulative modelled variation in the X and Y matrix, respectively; Q² is the cumulative predicted variation in the Y matrix; the validation was same with the description in Fig. 3.

Fig. 6. Plots of OPLS-DA and Validation of the OPLS-Da model (dynamic analysis of urine). (A) Day 20 KWG (U2) vs. day 0 KWG (U1); (B) day 40 KWG (U3) vs. day 0 KWG (U1); (C) day 50 KWG (U4) vs. day 0 KWG (U1); (D) day 80 KWG (U5) vs. day 0 KWG (U1). R²X and R²Y are the cumulative modelled variation in the X and Y matrix, respectively; Q² is the cumulative predicted variation in the Y matrix; the validation was same with the description in Fig. 3.

Fig. 7. A heat map based on the relative levels of the potential biomarkers in the serum (A) and urine (B) samples in 0–50 days by KW vs. CG analysis. The data set was screened using t-test p value < 0.05, VIP value > 1. Rats in kiwi wine group were treated with kiwi wine with 3.35 mL per kg per day for 50 days, while control group with the same amount water. The data were analysis between KWG and CG.

Fig. 8. A heat map based on the relative levels of the potential biomarkers in the serum (A) and urine (B) samples in 0–50 days by dynamic analysis. The data set was screened using t-test p value < 0.05, VIP value > 1. Rats in kiwi wine group were treated with kiwi wine with 3.35 mL per kg per day for 50 days. The data were analysis between every time point and 0 days in KWG.

Fig. 9. The metabolism network of highest correlation metabolites pathways and relative metabolites in 0–50 days. (A) Urine sample (B) serum. The potential biomarkers that increased are labeled in red and those that decreased are labeled in blue.

Fig. 10. The metabolism network of highest correlation metabolites pathways and relative metabolites in 80 days. (A) KWG vs. CG (B) dynamic analysis. The potential biomarkers that increased are labeled in red and those that decreased are labeled in blue.