Correlation of drug-induced and drug-related ultra-high performance liquid chromatography-mass spectrometry serum metabolomic profiles yields discovery of effective constituents of Sini decoction against myocardial ischemia in rats

Guangguo Tan a, Xin Wang b, Kui Liu c, Xin Dong d, Wenting Liao *b and Hong Wu *a
aSchool of Pharmacy, Fourth Military Medical University, Xi'an 710032, China. E-mail: wuhongfmmu@163.com; Tel: +86-29-84776823
bSchool of Pharmacy, China Pharmaceutical University, Nanjing 210009, China. E-mail: lwting84@163.com; Tel: +86-25-83271038
cStudent Brigade, College of Basic Medicine, Fourth Military Medical University, Xi'an 710032, China
dSchool of Pharmacy, Second Military Medical University, Shanghai 200433, China

Received 19th June 2018 , Accepted 24th August 2018

First published on 31st August 2018


Screening active constituents of traditional Chinese medicines (TCMs) is vital for lead compound discovery. Sini decoction (SND) is a well-known TCM formula for relieving myocardial ischemia (MI) in clinic. Due to complex nature, the effective compounds of SND are still unknown. In this study, a novel “system to system” strategy based on the correlation of drug-related and drug-induced ultra-high performance liquid chromatography coupled with quadrupole-time-of-flight mass spectrometry (UHPLC-Q-TOFMS) serum metabolomic profiles was developed to discover bioactive compounds of SND against isoproterenol-induced MI. Thirteen SND-induced metabolites and 19 SND-related metabolites were identified by UHPLC-Q-TOFMS coupled with S-plot and SUS-plot of orthogonal projection to latent structure-discriminant analysis (OPLS-DA) models, respectively. Canonical correlation analysis between the SND-induced and SND-related metabolites revealed that 12 compounds had strongly correlated relationship with the protective effect of SND on MI, and these compounds include isotalatizidine, songorine, fuziline, neoline, talatizamine, 14-acetyltalatizamine, liquiritigenin, benzoylmesaconitine, isoliquiritin, benzoylaconitne, benzoylhypaconitine and 6-gingerol. Combination functional enrichment analysis and network topology analysis revealed that the targeted metabolic pathways of these correlated compounds were involved in valine, leucine and isoleucine biosyntheses, tryptophan metabolism, glycerophospholipid metabolism and sphingolipid metabolism. The results demonstrated that the “system to system” strategy may be a high-throughput method to discover potentially effective compounds from TCMs.


1. Introduction

Traditional Chinese medicines (TCMs) have been used in Asia for a long time, and their therapeutic effects are generally considered credible by laboratory and clinical studies.1 Therefore, it may be hypothesized that there is a higher possibility of finding bioactive natural products from TCMs. However, due to the complexity of TCMs and the “system to system” interaction between the TCM system and a biological system, determining the active ingredients that play a therapeutic role presents a great challenge.2,3 Therefore, it is essential to develop a novel model to determine the constituents that contribute to the therapeutic effects.

The emergence of metabolomics based on ultra-high performance liquid chromatography coupled with quadrupole-time-of-flight mass spectrometry (UHPLC-Q-TOFMS), which can profile the global metabolic state of entire organisms, provides a systemic assessment of the holistic efficacy of TCMs.4–6 On the other hand, this technique, which allows for the capture of all metabolic changes due to treatment, is fully adapted for the determination of drug-related (exogenous) and drug-induced (endogenous) metabolomic profiles in biological systems.7 The drug-induced metabolite biomarkers from UHPLC-Q-TOFMS metabolomic profiles can be considered as possible pharmacodynamic indices.8,9 In addition, according to the serum pharmacochemistry of TCMs, the drug-related compounds in the blood have the potential of being effective constituents.10 Therefore, correlation analysis between drug-induced and drug-related UHPLC-Q-TOFMS serum metabolite profiles from TCMs can provide a “system to system” approach for the screening and discovery of effective components associated with the therapeutic effects of TCMs.

Sini decoction (SND), which consists of Aconitum carmichaelii Debeaux lateral root, Zingiber officinale Roscoe and Glycyrrhiza uralensis Fisch with a mass ratio of 3[thin space (1/6-em)]:[thin space (1/6-em)]2[thin space (1/6-em)]:[thin space (1/6-em)]3, has shown significant effect in the treatment of cardiovascular diseases for many years.11,12 Previous studies based on myocardial ischemia (MI) of rats showed that SND can improve cardiovascular function and reduce the myocardial infarct area.13,14 However, similar to most TCMs, the effective components of SND responsible for anti-MI bioactivities are unknown. Therefore, as a demonstrative study, we developed a “system to system” strategy by correlating drug-related and drug-induced UHPLC-Q-TOFMS serum metabolomic profiles in an isoproterenol-induced MI model to screen potentially effective compounds found in SND that act on MI. In comparison with the conventional screening for bioactive compounds based on a single target, the presented “system to system” strategy is advantageous for the characterization of the interaction relationship between the biological system and the TCM intervening system by means of global metabolic regulation.

2. Materials and methods

2.1 Chemicals and materials

Chromatographic grade methanol, acetonitrile and formic acid were purchased from Burdick & Jackson (USA). Isoprenaline hydrochloride (ISO) was purchased from MedChem Express. Lysophosphatidylcholine (18[thin space (1/6-em)]:[thin space (1/6-em)]2) was purchased from Larodan AB (Malmö, Sweden). Sphingosine 1-phosphate and palmitoylcarnitine were purchased from Sigma-Aldrich (St Louis, MO, USA). Valine, tyrosine, isoleucine, Indoleacetic acid, tryptophan, hippuric acid, and L-2-chlorophenylalanine (internal standard) were obtained from Shanghai Jingchun Reagent Co. Liquiritigenin, isoliquiritin, benzoylmesaconitine, benzoylaconitine, benzoylhypacoitine, 6-gingerol and glycyrrhetic acid were obtained from the Control of Pharmaceutical and Biological Products (Beijing, China). Ultrapure water was prepared with a Milli-Q water purification system (Millipore, Bedford, MA, USA).

Aconitum carmichaelii Debeaux lateral root (Sichuan province), Zingiber officinale Roscoe (Guizhou province) and Glycyrrhiza uralensis Fisch (Xinjiang province) were obtained from Tongrentang drugstore (Xian, China). The crude herbs met the quality standards recorded in Chinese Pharmacopoeia (2015 Edition).

2.2 Preparation of SND samples

The preparation of SND samples used in this study was identical to the method used in our previous study.13 Briefly, the dry crude drugs of Aconitum carmichaelii Debeaux lateral root (90 g), Zingiber officinale Roscoe (60 g), and Glycyrrhiza uralensis Fisch (90 g) were extracted twice with boiling water (1[thin space (1/6-em)]:[thin space (1/6-em)]10 and then 1[thin space (1/6-em)]:[thin space (1/6-em)]8) for 2.0 and 1.0 h and filtered through a gauze. The merged mixtures were then condensed under decompression. Finally, the extraction solution was made to a concentration of 1.0 g crude drugs per mL.

2.3 Ethical statement

All animal procedures were performed in accordance with the Guidelines for Care and Use of Laboratory Animals of China Pharmaceutical University and approved by the Animal Ethics Committee of China Pharmaceutical University.

2.4 Animal study and sampling

Thirty-five male Sprague–Dawley rats (weighing 200 ± 15 g) were purchased from the Sino-British SIPPR/BK Lab Animal Ltd (Shanghai, China) and housed in standard conditions with temperature range from 20 °C to 25 °C and humidity range from 50% to 60% as well as 12 h light/dark cycle.

The animals were randomly divided into three groups: normal control group (CTR, n = 10), ISO-induced MI (MI, n = 10) and SND pretreatment in ISO-induced model group (SND, n = 15). The rats in the SND group were given SND via intragastric administration at a daily dose of 10 g per kg body weight (equal to 10 mL per kg body weight) for 10 consecutive days with subcutaneous injections of ISO saline solution (85 mg kg−1) 5 min after SND pretreatment on the 9th and 10th days. The control and MI groups received water via an intragastric tube during the ten consecutive days. On the 9th day and 10th day, the control and MI groups were injected subcutaneously with normal saline and ISO saline solution, respectively. At 2 h and 4 h after the second ISO injection, blood samples from each rat were collected via the retro-orbital plexus in standard vials. The blood samples were allowed to clot at room temperature for 1 h; then, the serum was separated by centrifugation at 3000g for 10 min at 4 °C and stored at −70 °C for biochemistry assays within three days and metabolomic study within seven days. Meanwhile, the heart tissues were quickly removed and treated with phosphate buffer saline at room temperature for 3 minutes and then frozen at −20 °C for determining myocardial infarct size.

2.5 Biochemical and histological assay

The myocardial-specific enzymes lactate dehydrogenase (LDH), creatine kinases (CK), and aspartate transaminase (AST) as well as the lipid peroxidation product malondialdehyde (MDA) and antioxidant enzyme superoxide dismutase (SOD) in serum were determined using standard kits supplied by Jiancheng Tech Co., Ltd (Nanjing, China). To determine the size of the myocardial infarct, duplicate 1 mm mid-left ventricular sections of other frozen hearts were cut and incubated with 1% 2,3,5-triphenyltetrazolium chloride (TTC) for 20 min at 37 °C. The infarct size was determined via planimetry of the infarct zone and was expressed as a percentage of the total left ventricular area.

2.6 Sample preparations for metabolomic and SND analysis

Serum samples for metabolomic analysis were thawed and vortexed for 5 s at room temperature. Then, 400 μL methanol containing 12.5 μg ml−1L-2-chlorophenylalanine (internal standard) was added to 100 μL serum. After vortexing for 1 min and incubating on ice for 10 min, the mixture was centrifuged at 14[thin space (1/6-em)]000g for 15 min at 4 °C. All supernatants were transferred to an auto-sampler vial. A pooled sample obtained from aliquots of the whole sample set was prepared and used as quality control (QC) samples.15 For the sample preparation for SND analysis, 0.1 mL SND (1.0 g crude drugs per mL) was diluted and made up to 5 mL by the addition of methanol. The diluted sample solution was filtered through a 0.22 μm filter. The 5 μL filtrate was used as the UHPLC sample.

2.7 UHPLC-Q-TOFMS analysis

UHPLC-Q-TOFMS analysis was performed on the Agilent 1290 Infinity LC system coupled with an Agilent 6530 Accurate-Mass Quadrupole Time-of-Flight (Q-TOF) mass spectrometer. An ACQUITY UPLC HSS T3 C18 column (2.1 mm × 100 mm, 1.7 μm, Waters, Milford, MA) was used for sample separation at 40 °C. The mobile phase consisted of 0.1% formic acid (A) and acetonitrile modified with 0.1% formic acid (B) with gradient elution of 5% B at 0–2 min, 5–70% B at 2–16 min, 70–95% B at 16–17 min, and 95% B at 17–19 min. The post time was 5 min. The flow rate was 400 μL min−1, and the injection volume was 5 μL. The detailed mass spectrometric conditions were carried out according to our previously published paper.16

2.8 Data analysis

The raw UHPLC-Q-TOFMS data were exported in mzData format and then processed by the XCMS package, as described in a previous publication.17 XCMS parameters were default settings except for the following: full width at half maximum (fwhm) = 10, bandwidth (bw) = 10 and signal-to-noise ratio (snthersh) = 5. The output data matrix consists of a collection of values (observation, retention time, m/z, intensity). The variables with less than 30% relative standard deviation (RSD) of peak intensity in QC samples and present in more than 80% of QCs were then retained for multivariate data analysis.18 The resulting data matrix was then normalized to the intensity of the internal standard. After normalization, the variable from the internal standard was removed so that the data set used for the multivariate data analysis consisted of extracted compounds. To perform principal component analysis (PCA), partial least-square-discriminant analysis (PLS-DA), and orthogonal projection to latent structure-discriminant analysis (OPLS-DA), the data set was introduced to the SIMCA-P software version 14.1 (Umetrics, Umea, Sweden), in which the data were mean-centered and pareto-scaled for these models. OPLS-DA together with the S-plot and SUS-plot, shared and unique structure (SUS) plot,19 was employed to mine MI-related and SND-related features from the complex dataset, respectively.

2.9 Characterization of MI-related and SND-induced endogenous metabolites

Endogenous metabolite identification was performed according to our previous method with slight modification.14 Briefly, the quasi-molecular ions of endogenous biomarkers were judged based on the positive and negative ESI-Q-TOFMS scans. To search the detected molecular weights from open-access databases METLIN and HMDB within a mass accuracy of 30 ppm, the chemical structures of the preliminarily identified metabolites were obtained. To narrow the scope of target metabolites, the quasi-molecular ions were then subjected to MS/MS analysis. Finally, these metabolites were structurally confirmed by comparing them with commercial standards available in our lab or fragment models in METLIN and HMDB databases.

2.10 Characterization of exogenous SND-related metabolites

The SUS-plot revealed exogenous variables (SND-related metabolites) from endogenous variables. Compared with the total ion chromatogram of an SND sample in vitro, ±20 ppm narrow mass window-extracted ion chromatograms of exogenous variables from UHPLC-Q-TOFMS serum metabolomic profiles were used for further confirmation. Formulas were generated using the MassHunter Software Qualitative Analysis (Agilent Technologies, California, United States) for each variable that was detected with a mass accuracy of 5 ppm. According to our previous studies on chemical composition of SND and the MS fragmentation rules,16,20 the exogenous variables were identified based on formula-matching from an in-house database of SND and MS/MS fragments.

2.11 Correlation analysis of SND-reversed and SND-related metabolites

Based on the premise that SND has a significant protective effect on ISO-induced MI rats, a canonical correlation analysis between SND-induced and SND-related metabolites’ standardization intensities was used to discover constituents of SND that were effective against myocardial infarction. The correlation coefficient (r), depicting the degree of correlation between SND-induced and SND-related metabolites, was used to determine which constituents contributed most to the therapeutic effect. A cross-correlation heat map was used to visualize the relationship, in which the SND-related compounds having |r| values greater than 0.60 in this study were considered as highly correlated ingredients that contributed to disease treatment. The canonical correlation analysis and heat map analysis were performed in the R environment (http://www.r-project.org) with CCA and pheatmap packages, respectively.

3. Results and discussion

3.1 Biochemical and histological assay

When myocardial infarction occurred, the myocardial-specific enzymes LDH, CK, and AST were released from damaged myocardial cells into the blood. Therefore, they are important indicators for evaluating the MI model and protective effect of SND. As shown in Fig. 1A–C, the LDH, CK, and AST levels at 2 and 4 h were significantly elevated in the MI model rats, and their levels were markedly improved by SND (2 h and 4 h). MDA and SOD are important parameters in the evaluation of lipid peroxidation and oxidative stress. As shown in Fig. 1D and E, the significantly increased level of MDA and decreased level of SOD were observed after injections of ISO at 2 and 4 h in the MI group relative to those in the control group, but these changes were significantly ameliorated in the SND group. As shown in Fig. 1F, compared with the results of the MI group, the infarct areas of SND group were significantly reduced. Taken together, these findings showed that myocardial infarction occurred in all rats except in control rats, and pre-treatment with SND had a cardioprotective effect on MI rats and reduced myocardial damage.
image file: c8fo01217b-f1.tif
Fig. 1 Effect of SND on serum LDH, CK, AST, MDA and SOD levels and the myocardial infarct size in ISO-induced model rats. (A) Lactate dehydrogenase (LDH); (B) creatine kinase (CK); (C) aspartate transaminase (AST); (D) malondialdehyde (MDA); (E) superoxide dismutase (SOD); (F) myocardial infarct size evaluation. The infarct size (%) was expressed as myocardial infarct size/left ventricular area × 100%. Data shown are mean ± SD, n = 10, 10 and 15 for control, MI and SND-treated groups, respectively. *P < 0.05 vs. control; #P < 0.05 vs. MI model.

3.2 Quality control of UHPLC-Q-TOFMS analysis

Quality checking for UHPLC-Q-TOFMS fingerprinting data set was conducted by using QC samples. Metabolic profiles from rat serum samples and QC samples were aligned together and then filtered to obtain a data set that contained 1555 variables present in more than 80% of QC samples and with a relative standard deviation (RSD) of peak intensity less than 30% in the QC samples. This data set covered 86.4% variables in fingerprinting data set, which indicated good repeatability of the established UHPLC-Q-TOFMS analysis. An unsupervised PCA model was further studied to assess the stability of the analytical system. The close clustering of the QC samples and no outliers were observed in the PCA score plot (Fig. 2A), which also confirmed that the UHPLC-Q-TOFMS analysis has good repeatability.
image file: c8fo01217b-f2.tif
Fig. 2 PCA and PLS-DA score plots of the control, MI and SND-treated rats for 2 h and 4 h. (A) PCA score plot of real samples and QC samples. The QC cluster is highlighted within the red ellipses. (B) PLS-DA score plot for both 2 h and 4 h; (C) PLS-DA score plot for 2 h (D) PLS-DA score plot for 4 h. C: control group; M: MI group; S: SND-treated group; 2: 2 h; 4: 4 h.

3.3 Protective effects of SND against MI

Pattern recognition analysis based on the UHPLC-Q-TOFMS data set was first employed to evaluate the protective effects of SND on MI. To cross-validate the effect of SND, a supervised PLS-DA model was established, and the collected samples were divided into six sub-groups: control rats at 2 h and 4 h (C2 and C4), MI rats at 2 h and 4 h (M2 and M4), and SND pretreated rats at 2 h and 4 h (S2 and S4). As shown in the score plot of PLS-DA (Fig. 2B), serum samples from the same group demonstrated a cluster tendency. However, samples from different groups showed a separate tendency to different extents. It was found that the ISO-induced MI sub-groups at 2 h (M2) and 4 h (M4) exhibited clear shifts from the corresponding control sub-groups at 2 h (C2) and 4 h (C4), whereas the SND pretreated sub-groups at 2 h (S2) and 4 h (S4) deviated from the MI sub-groups toward the control sub-groups. These metabolic data indicated that the administration of SND for 2 h had an impact on the metabolic profiles of the ISO-induced MI rats and shifted the score plots closer to that of the corresponding controls at 2 h (C2) than that in the sub-model of the samples collected at 4 h (Fig. 2C and D). To quantitatively assess the protective and regulatory effects, the relative mean distance values (RDVs) of the three sub-groups at 2 h and 4 h were calculated based on the PLS-DA sub-models at 2 h and 4 h.21 Based on RDVs, the normalized RDVs between the SND group and the MI group decreased from 0.75 (2 h) to 0.52 (4 h) (ESI Table S2), which suggested that pre-administration of SND could reverse the pathological process of ISO-induced MI, and pre-administration of SND for 2 h had a greater effect on the perturbed metabolism than pre-administration of SND for 4 h.

3.4 Characterization of MI-related and SND-induced endogenous metabolites

To identify the metabolites that contributed to the metabolic distinctions at two sub-models, two supervised OPLS-DA models between the MI sub-groups and the control sub-groups at 2 h and 4 h were created. The OPLS-DA score plots displayed significant differences between the MI group and the control group (Fig. S1AB). The S-plot from the OPLS model was applied to find MI-related metabolites (Fig. S1CD), where the ions furthest away from the origin may be regarded as potential biomarkers (shaded areas of Fig. S1CD), and they contributed to the differences between MI and control groups. The selected metabolites were further filtered based on the threshold values of variable importance in the projection (VIP) (>1) generated from OPLS-DA model and validated by using one-way ANOVA (p < 0.05). According to these criteria, 21 metabolites were found to be the common characteristics at 2 h and 4 h and were identified as MI-related metabolites. With these perturbed metabolites as possible targets of SND, one-way ANOVA analysis revealed that 13 of the 21 identified MI-related metabolites were significantly reversed by SND. The details of the fragments in each MS/MS spectrum for each identified endogenous metabolite are presented in ESI (Fig. S2).

3.5 Characterization of exogenous SND-related metabolites

A SUS-plot (shared and unique structure), a new statistical approach enabling the comparison of different treatments (MI and SND groups in this study) with the same control and allowing clear separation of exogenous variables (drug metabolites) from endogenous variables,19 was used to screen the SND-related metabolites (SND-absorbed constituents). In the SUS-plots from the two sub-models at 2 h and 4 h, the variables from SND-related metabolites were exactly located on or close to the axis in the upregulated direction (red rectangles in Fig. 3). By cross-validating the results from the two SUS-plots at 2 h and 4 h, 21 ions were identified as SND-absorbed constituents. Finally, by comparing the chromatograms of dosed samples with those of control and SND samples in vitro with ±20 ppm narrow mass window-extracted ion chromatograms from raw chromatograms and excluding in-source fragments and adducts (Fig. S3), 19 SND-related metabolites were identified in serum sample.
image file: c8fo01217b-f3.tif
Fig. 3 SUS-plot between ISO-induced MI group and SND-treated group at 2 h (A) and 4 h (B). SUS plot uses the combination of two predictive loadings obtained from two different OPLS models, both sharing a common control group. Each variable has then one coordinate from each model. This approach allows in particular the discrimination between endogenous potential biomarkers and drug metabolites, the later being exactly located on one axis in the upregulated direction (red rectangles on both plots).

3.6 Correlation analysis of SND-induced and SND-related metabolites

Canonical correlation analysis was used for determining the relationships between the relative intensities of the 13 SND-induced metabolites and the 19 SND-related metabolites in each of the SND-treated MI rats at 2 h and 4 h. To reveal which compounds contributed to the therapeutic effect of SND against MI, the correlation coefficient (r) data when r ≥ 0.6 were extracted using the pheatmap package. To combine and cross-validate the correlation coefficient at 2 h and 4 h, 12 absorbed components were identified as highly positively (r ≥ 0.6) and negatively correlated (r ≤ −0.6) with the protective effects of SND on MI (Fig. 4), and these include isotalatizidine, songorine, fuziline, neoline, talatizamine, 14-acetyltalatizamine, liquiritigenin, benzoylmesaconitine, isoliquiritin, benzoylaconitne, benzoylhypaconitine and 6-gingerol.
image file: c8fo01217b-f4.tif
Fig. 4 Correlation heatmap between SND-induced metabolites and SND-related compounds at 2 h (A) and 4 h (B). image file: c8fo01217b-u1.tif, Highly positively correlated (r ≥ 0.6); image file: c8fo01217b-u2.tif, highly negatively correlated (r ≤ −0.6); image file: c8fo01217b-u3.tif, low correlation (|r| < 0.6).

3.7 Metabolic pathways of the correlated compounds

Based on this study, 12 absorbed compounds with a high correlation coefficient value contributed to the protective effects of SND on MI. To explore the possible metabolic pathways that were regulated by these effective constituents on ISO-induced MI rats, the relative intensities of 13 significantly reversed endogenous metabolites from both MI and SND-pretreated groups at 2 h and 4 h were introduced into MetaboAnalyst,22 in which the most relevant pathways were identified based on the pathway enrichment analysis combined with the topology analysis. With the threshold of impact-value (more than zero) generated from a pathway topology analysis and the p value (less than 0.05) generated from an enrichment analysis, four metabolic pathways were identified as targeted metabolic pathways (Fig. 5) including valine, leucine and isoleucine biosyntheses, tryptophan metabolism, glycerophospholipid metabolism and sphingolipid metabolism. Therefore, it can be concluded that the correlated compounds from SND can modulate disordered homeostasis of valine, leucine and isoleucine biosyntheses, tryptophan metabolism, glycerophospholipid metabolism and sphingolipid metabolism to alleviate ISO-induced MI.
image file: c8fo01217b-f5.tif
Fig. 5 Summary of pathway analysis of significantly reversed metabolites between MI and SND-treated rats.

4. Conclusion

In this study, a “system to system” strategy based on the correlation of drug-related and drug-reversed UHPLC-Q-TOFMS serum metabolomic profiles was developed to screen and discover effective components associated with the therapeutic effect of TCMs. Under the premise that SND has a significant protective effect on ISO-induced MI rats, 13 SND-induced metabolites and 19 SND-related metabolites were identified by using UHPLC-Q-TOFMS coupled with S-plot and SUS-plot of OPLS-DA models, respectively. Canonical correlation analysis showed that 12 compounds had a strongly correlated relationship with the protective effect of SND on MI. Modulating the disordered homeostasis of valine, leucine and isoleucine biosyntheses, tryptophan metabolism, glycerophospholipid metabolism and sphingolipid metabolism were the targeted metabolic pathways of these bioactive compounds. The activity and metabolic pathways of these compounds need to be verified by subsequent assays. The “system to system” strategy can provide a high-throughput method to discover potentially effective compounds from TCMs.

Conflicts of interest

The authors declare no competing financial interests.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (No. 81402888 and 81773677).

References

  1. B. He, C. Lu, M. L. Wang, G. Zheng, G. Chen, M. Jiang, X. J. He, Z. X. Bian, G. Zhang and A. P. Lu, Drug discovery in traditional Chinese medicine: From herbal fufang to combinatory drugs, Science, 2015, 350, S74–S76 Search PubMed.
  2. Q.-L. Liang, X.-P. Liang, Y.-M. Wang, Y.-Y. Xie, R.-L. Zhang, X. Chen, R. Gao, Y.-J. Cheng, J. Wu, Q.-B. Xu, Q.-Z. Xiao, X. Li, S.-F. Lv, X.-M. Fan, H.-Y. Zhang, Q.-L. Zhang and G.-A. Luo, Effective components screening and anti-myocardial infarction mechanism study of the Chinese medicine NSLF6 based on “system to system” mode, J. Transl. Med., 2012, 10, 26 CrossRef PubMed.
  3. S. Li, Mapping ancient remedies: Applying a network approach to traditional Chinese medicine, Science, 2015, 350, S72–S74 CrossRef PubMed.
  4. X. Wang, A. Zhang, H. Sun, Y. Han and G. Yan, Discovery and development of innovative drug from traditional medicine by integrated chinmedomics strategies in the post-genomic era, TrAC, Trends Anal. Chem., 2016, 76, 86–94 CrossRef.
  5. X. Liang, X. Chen, Q. Liang, H. Zhang, P. Hu, Y. Wang and G. Luo, Metabonomic Study of Chinese Medicine Shuanglong Formula as an Effective Treatment for Myocardial Infarction in Rats, J. Proteome Res., 2011, 10, 790–799 CrossRef PubMed.
  6. X. Wang, H. Sun, A. Zhang, W. Sun, P. Wang and Z. Wang, Potential role of metabolomics apporoaches in the area of traditional Chinese medicine: As pillars of the bridge between Chinese and Western medicine, J. Pharm. Biomed. Anal., 2011, 55, 859–868 CrossRef PubMed.
  7. V. Croixmarie, T. Umbdenstock, O. Cloarec, A. Moreau, J. M. Pascussi, C. Boursier-Neyret and B. Walther, Integrated Comparison of Drug-Related and Drug-Induced Ultra Performance Liquid Chromatography/Mass Spectrometry Metabonomic Profiles Using Human Hepatocyte Cultures, Anal. Chem., 2009, 81, 6061–6069 CrossRef PubMed.
  8. X. Wang, A. Zhang, X. Zhou, Q. Liu, Y. Nan, Y. Guan, L. Kong, Y. Han, H. Sun and G. Yan, An integrated chinmedomics strategy for discovery of effective constituents from traditional herbal medicine, Sci. Rep., 2016, 6, 18997 CrossRef PubMed.
  9. G. Xie, S. Wang, H. Zhang, A. Zhao, J. Liu, Y. Ma, K. Lan, Y. Ni, C. Liu, P. Liu, T. Chen and W. Jia, Poly-pharmacokinetic Study of a Multicomponent Herbal Medicine in Healthy Chinese Volunteers, Clin. Pharmacol. Ther., 2018, 103, 692–702 CrossRef PubMed.
  10. C. X. Zhang, R. X. Lian, M. Mahmoodurrahman, S. S. Lai, Z. X. Zhao and Y. Yu, Serum pharmacochemistry for tracking bioactive components by UPLC-Q-TOF-MS/MS combined chromatographic fingerprint for quality assessment of Sanziguben Granule, J. Chromatogr. B: Anal. Technol. Biomed. Life Sci., 2016, 1029, 128–136 CrossRef PubMed.
  11. J. H. Zheng, Y. Q. Mo, X. Y. Zhuo, X. C. Wu, X. C. Zhang, J. H. Li and J. Zhang, Literature-based metabolism mechanism study of Sini Decoction on myocardial infarction model, J. Am. Coll. Cardiol., 2016, 68, C15–C15 CrossRef.
  12. S. Chen, H. L. Jiang, Y. Cao, Y. Wang, Z. H. Hu, Z. Y. Zhu and Y. F. Chai, Drug target identification using network analysis: Taking active components in Sini decoction as an example, Sci. Rep., 2016, 6(14), 24245 CrossRef PubMed.
  13. G. Tan, W. Liao, X. Dong, G. Yang, Z. Zhu, W. Li, Y. Chai and Z. Lou, Metabonomic Profiles Delineate the Effect of Traditional Chinese Medicine Sini Decoction on Myocardial Infarction in Rats, PLoS One, 2012, 7, e34157 CrossRef PubMed.
  14. G. Tan, Z. Lou, W. Liao, X. Dong, Z. Zhu, W. Li and Y. Chai, Hydrophilic interaction and reversed-phase ultraperformance liquid chromatography TOF-MS for serum metabonomic analysis of myocardial infarction in rats and its applications, Mol. BioSyst., 2012, 8, 548–556 RSC.
  15. H. G. Gika, G. A. Theodoridis, J. E. Wingate and I. D. Wilson, Within-day reproducibility of an HPLC-MS-based method for metabonomic analysis: application to human urine, J. Proteome Res., 2007, 6, 3291–3303 CrossRef PubMed.
  16. G. Tan, M. Liu, X. Dong, S. Wu, L. Fan, Y. Qiao, Y. Chai and H. Wu, A strategy for rapid analysis of xenobiotic metabolome of Sini decoction in vivo using ultra-performance liquid chromatography-electrospray ionization quadrupole-time-of-flight mass spectrometry combined with pattern recognition approach, J. Pharm. Biomed. Anal., 2014, 96, 187–196 CrossRef PubMed.
  17. C. A. Smith, E. J. Want, G. O'Maille, R. Abagyan and G. Siuzdak, XCMS: processing mass spectrometry data for metabolite profiling using nonlinear peak alignment, matching, and identification, Anal. Chem., 2006, 78, 779–787 CrossRef PubMed.
  18. S. Bijlsma, I. Bobeldijk, E. R. Verheij, R. Ramaker, S. Kochhar, I. A. Macdonald, B. van Ommen and A. K. Smilde, Large-scale human metabolomics studies: a strategy for data (pre-) processing and validation, Anal. Chem., 2006, 78, 567–574 CrossRef PubMed.
  19. S. Wiklund, E. Johansson, L. Sjostrom, E. J. Mellerowicz, U. Edlund, J. P. Shockcor, J. Gottfries, T. Moritz and J. Trygg, Visualization of GC/TOF-MS-based metabolomics data for identification of biochemically interesting compounds using OPLS class models, Anal. Chem., 2008, 80, 115–122 CrossRef PubMed.
  20. G. Tan, Z. Zhu, J. Jing, L. Lv, Z. Lou, G. Zhang and Y. Chai, Characterization of constituents in Sini decoction and rat plasma by high-performance liquid chromatography with diode array detection coupled to time-of-flight mass spectrometry, Biomed. Chromatogr., 2011, 25, 913–924 CrossRef PubMed.
  21. J. Aa, F. Shao, G. Wang, Q. Huang, W. Zha, B. Yan, T. Zheng, L. Liu, B. Cao, J. Shi, M. Li, C. Zhao, X. Wang and Z. Wu, Gas chromatography time-of-flight mass spectrometry based metabolomic approach to evaluating toxicity of triptolide, Metabolomics, 2011, 7, 217–225 CrossRef.
  22. J. Chong, O. Soufan, C. Li, I. Caraus, S. Li, G. Bourque, D. S. Wishart and J. Xia, MetaboAnalyst 4.0: towards more transparent and integrative metabolomics analysis, Nucleic Acids Res., 2018, 46, W486–W494 CrossRef PubMed.

Footnotes

Electronic supplementary information (ESI) available. See DOI: 10.1039/c8fo01217b
These authors contributed equally to this work.

This journal is © The Royal Society of Chemistry 2018