PHYTOTHERAPY RESEARCH Phytother. Res. 30: 654–662 (2016) Published online 25 January 2016 in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/ptr.5575
Serum Metabolomic Profiling in a Rat Model Reveals Protective Function of Paeoniflorin Against ANIT Induced Cholestasis Zhe Chen,1,2† Yun Zhu,3† Yanling Zhao,1* Xiao Ma,1,2 Ming Niu,4 Jiabo Wang,4 Haibin Su,5 Ruilin Wang,3 Jianyu Li,3 Liping Liu,1 Zhenman Wei,1 Qingguo Zhao,1 Hongge Chen1 and Xiaohe Xiao4* 1
Department of Pharmacy, 302 Hospital of People’s Liberation Army, Beijing 100039, China College of Pharmacy, Chengdu University of Traditional Chinese Medicine, Chengdu 611137, China 3 Department of Integrative Medical Center, 302 Hospital of People’s Liberation Army, Beijing 100039, China 4 China Military Institute of Chinese Medicine, 302 Hospital of People’s Liberation Army, Beijing 100039, China 5 Liver Failure Treatment and Research Center, 302 Hospital of People’s Liberation Army, Beijing 100039, China 2
Cholestasis is a leading cause of hepatic accumulation of bile acids resulting in liver injury, fibrosis, and liver failure. Paeoniflorin displays bright prospects in liver protective effect. However, its molecular mechanism has not been well-explored. This study was designed to assess the effects and possible mechanisms of paeoniflorin against alpha-naphthylisothiocyanate-induced liver injury. Ultraperformance liquid chromatography coupled with quadrupole time-of-flight combined with principle component analysis and partial least squares discriminant analysis were integrated to obtain differentiating metabolites for the pathways and clarify mechanisms of disease. The results indicated that paeoniflorin could remarkably downregulate serum biochemical indexes and alleviate the histological damage of liver tissue. Different expression of 14 metabolites demonstrated that paeoniflorin mainly regulated the dysfunctions of glycerophospholipid metabolism and primary bile acid biosynthesis. Moreover, several pathways such as arginine and proline metabolism, ether lipid metabolism, and arachidonic acid metabolism were also related to the efficacy. In conclusion, paeoniflorin has indicated favorable pharmacological effect on serum biochemical indexes and pathological observation on cholestatic model. And metabolomics is a promising approach to unraveling hepatoprotective effects by partially regulating the perturbed pathways, which provide insights into mechanisms of cholestasis. Copyright © 2016 John Wiley & Sons, Ltd. Keywords: metabolomics; cholestasis; paeoniflorin; pattern recognition approaches; mechanism.
INTRODUCTION As one of the chronic liver disease, cholestasis arises from impaired hepatobiliary production and excretion of bile, which is one of the top 15 causes of death in the USA (Yang et al., 2009). In these disorders, injuries to bile ducts or liver cells can lead to a range of clinical presentations, such as isolated abnormalities in liver biochemistry, liver failure, hepatobiliary malignancy, congenital, immunologic, and toxic factors, which can contribute to disease (Hirschfield et al., 2010). As one of the most common and devastating manifestations among hereditary and acquired liver diseases, cholestasis ultimately leads to liver failure and cirrhosis (Hofmann, 2002; Watanabe et al., 2007). In humans, it frequently occurs because of various endogenous and exogenous
* Correspondence to: Yanling Zhao, Department of Pharmacy, 302 Hospital of People’s Liberation Army, Beijing 100039, China; Xiaohe Xiao, China Military Institute of Chinese Medicine, 302 Hospital of People’s Liberation Army, Beijing 100039, China. E-mail: [email protected] (Y. Zhao); [email protected] (X. Xiao) Contract/grant sponsor: National Science and Technology; contract/grant number: 2012ZX10005010-002-002. Contract/grant sponsor: National Natural Science Foundation Project of China; contract/grant numbers: 81303120, 81173571. † These authors are thought to have equal contributions.
Copyright © 2016 John Wiley & Sons, Ltd.
factors, such as dysregulation of bile acid transporters, oxidation stress, inflammation in hepatocytes, several drugs, alcohol abuse, and virus infection (Padda et al., 2011). While these factors are thought to be involved in the pathogenesis of cholestasis, the remarkable and important metabolites in cholestasis are not yet precisely recognized. As an omic science in systems biology, metabolomics represents a powerful discipline concerned with the comprehensive analysis of small molecules and provides a powerful approach to discovering biomarkers in biological systems (Wang et al., 2013). It has recently demonstrated significant potential in many fields such as toxicology, disease diagnosis, drug mechanism and development, responses to environmental stress, nutrition, new targets, and natural product discovery (Wang et al., 2011a). Based on analyzing the specific early biomarkers during disease or drug treatment, metabolomics provides a holistic insight into the relationship between substance and metabolic pathways (Wang et al., 2011b; Sun et al., 2012). As the newest of the ‘omics’ sciences, metabolomics has brought much excitement to the field of life sciences as a potential translation tool. Especially, its method and design resemble those of traditional Chinese medicine (TCM), indicating that metabolomics has the potential to impact our understanding of the TCM (Wang et al., Received 26 August 2015 Revised 25 November 2015 Accepted 22 December 2015
SERUM METABOLOMIC
2013). Yinchenhao decoction, a Chinese famous formula for treating jaundice, was proved to be effective on regulating multiple perturbed pathways to normal state in jaundice model by metabolomics (Sun et al., 2014). Paeoniflorin is one of the main bioactive components of the dried root of Paeonia lactiflora pall and Paeonia veitchii Lynch (Jiang et al., 2011). Clinical studies have demonstrated that Chi Shao can confer excellent therapeutic effects on hepatitis with severe cholestasis (Zhao et al., 2013). In this study, we assessed the hepatoprotective effects of paeoniflorin against liver injury induced by alpha-naphthylisothiocyanate (ANIT) through serum biochemical indexes and pathological observation. Metabolomics applying ultraperformance liquid chromatography coupled with quadruple time-of-flight (UPLC-Q-TOF) characterized the metabolic profiles of paeoniflorin protective efficacy on cholestasis. Principle component analysis (PCA) and partial least squares discriminant analysis (PLS-DA) were integrated to obtain differentiating metabolites for the pathways and clarify the liver protection mechanisms of paeoniflorin on cholestasis, which provided deep insights into mechanisms of cholestasis.
MATERIALS AND METHODS Chemicals and reagents. Paeoniflorin was purchased from the National Institute for the Control of Pharmaceutical and Biological Products (Beijing, China). ANIT was purchased from Sigma-Aldrich Corporation (St Louis, MO, USA). Ursodeoxycholic acid (UDCA) was purchased from Shanghai Zhongxi Pharmaceutical Factory (Shanghai, PR China). Total bilirubin (TBIL), direct bilirubin (DBIL), aspartate transaminase (AST), alanine transaminase (ALT), alkaline phosphatase (ALP), γ-glutamyltranspeptidase (γ-GT), and total bile acid (TBA) assay kits were obtained from Nanjing Jiancheng Bioengineering Institute (Nanjing, PR China). All the other chemicals of analytical grade were purchased from commercial sources.
Animals and treatments. Male Sprague-Dawley rats weighing 200 ± 20 g were obtained from the laboratory animal center of the Military Medical Science Academy of the PLA (Permission No. SCXK-(A) 2012-0004). All of the animals were acclimated for 1 week prior to the experiment and were kept under the same temperature (25 ± 2 °C) and lighting (12:12 h light : dark cycle) conditions. The whole studies were performed in accordance with the guidelines of the Council on Animal Care of Academia Sinica. The animals were randomly divided into six groups with six rats in each group. The rats in the normal group serving as normal control were given saline each day and treated with the vehicle (olive oil) alone. The model group was treated with 50 mg/kg ANIT dissolved in an equal volume of olive oil by gavage, and this dosage is known to induce cholestasis based on a previous study. UDCA (60 mg/kg), the positive control, was treated to rats with the same condition of model group. Paeoniflorin was dissolved in saline, and it was given to experimental rats Copyright © 2016 John Wiley & Sons, Ltd.
655
at doses of 0.2 (PaeH), 0.1 (PaeM), or 0.05 (PaeL) g/kg body weight, respectively, for six times before and four times after they were treated with 50 mg/kg ANIT by gavage. In this study, paeoniflorin doses that were adopted were based on preliminary experiment.
Sample preparation. Before the rats were sacrificed, they were fasted for 12 h after the last paeoniflorin administration. Blood samples were collected and centrifuged at 3000 × g for 10 min to obtain the serum. All of the serum samples were sterile, hemolysis-free, and stored at 80 °C before determining the biochemical indices and metabolomics analysis. The serum levels of TBIL, DBIL, AST, ALT, ALP, and TBA were measured by commercial test kits. Liver tissues were excised and fixed in 10% phosphate buffer solusion (PBS) buffered formalin. Three or four paraffin-embedded sections (4- to 5-μm thick) per specimen were prepared and stained with hematoxylin-eosin (HE staining). The stained sections were examined under Nikon microscope (Nikon Instruments Corporation, Shanghai, China) and analyzed by Image-Pro Plus 7200 software (Media Cybernetics, Inc., Rockville, MD, USA).
Metabolic profiling. Chromatography was performed using an Agilent 1290 series UPLC system equipped with quaternary pump, online degasser, auto sampler, and thermostat column compartment. The injection volume was fixed at 4 μL. All the samples were maintained at 4 °C during the analysis. The separation was performed on a ZORBAX RRHD 300 SB-C18 column (2.1 mm × 100 mm, 1.8 μm; Agilent, USA). The column temperature was set at 30 °C. The mobile phases were composed of 0.1% formic acid in acetonitrile (solvent A) and 0.1% formic acid in water (solvent B). The flow rate was set as 0.30 mL/min. The gradient was used as follows: a linear gradient of 95% A over initial– 1.0 min; 95–60% A over 1.0–9.0 min; 60–10% A over 9.0–19.0 min; 10–0% A over 19.0–21.0 min; and 0% A over 21.0–25.0 min. Then the eluent was introduced to the mass spectrometer directly. After every 10 samples injected, a pooled sample as the QC sample followed by a blank was injected to ensure the stability and repeatability of the liquid chromatography–mass spectrometry (LC-MS) systems. For MS, an Agilent 6550 Q-TOF/MS with an electrospray ionization source (ESI) in both positive and negative mode was used. Ionization was achieved using electrospray. The electrospray source parameters were fixed as follows: electrospray capillary voltage was 3.0 kV in negative ionization mode and 4.0 kV in positive ionization mode. The mass range was set from m/z 80 to 1000. Gas temperature was 200 °C in negative ionization mode and 225 °C in positive ionization mode. Gas flow was 11 L/min. Nebulizer was set to 35 pisg (negative) and 45 pisg (positive). Sheath gas temperature was 350 °C, and sheath gas flow was 12 L/min. Nozzle voltage was 500 V in both negative and positive mode.
Data extraction and pattern recognition analysis. Sample data were extracted by MassHunter Profinder software (Agilent, California, USA) for peak detection Phytother. Res. 30: 654–662 (2016)
656
Z. CHEN ET AL.
and alignment. Full scan mode was employed in the mass range of 80 to 1000 m/z. The initial and final retention time were set for data collection. The resultant data matrices were introduced to SIMCA-P 11.0 software (Umetrics AB, Umea, Sweden) for PCA and PLS-DA. Prior to PCA, all variables obtained from data matrix were mean-centered and scaled to Pareto variance. PCA, an unsupervised pattern recognition approach, is used to reduce the dimension of UPLC-MS data and to disclose intrinsic clustering of samples. To maximize the differences in inter-class discrimination and minimize the differences in inter-class discrimination, the data were further analyzed using the PLS-DA method. These variables with a higher VIP value (VIP ≥1.0) in PLS-DA method were considered to be further evaluated by one-way analysis of variance (ANOVA). Biomarkers identification and pathway enrichment analysis. Only the variables (p < 0.05 in ANOVA) were selected as potential biomarkers and subjected to further identification of the molecular mechanism. The potential biomarkers were identified by online biochemical database service METLIN (http://metlin.scripps.edu/). The compound names and KEGG (http://www.kegg.jp/) numbers of potential biomarkers were performed with MetaboAnalyst 3.0 (http://www.metaboanalyst.ca/) for further enrichment and pathway analysis. Statistical analysis. Data were expressed as the Mean ± SD and analyzed with the SPSS software program (version 13.0, Chicago, USA). The differences between the group means were calculated by ANOVA. The differences were considered to be statistically significant when p < 0.05 and highly significant when p < 0.01. The resulting three-dimensional matrix containing peak index, sample name, and peak area were introduced into SIMCA-P 11.0 software (Umetrics AB, Umea, Sweden), which was used for pattern recognition analysis including PCA and PLS-DA.
RESULTS Biochemical analysis and pathological observations The serum ALT and AST activities were measured as indexes of liver cell damage. As shown in Fig. 1A and B, the rats that were given ANIT (50 mg/kg i.g.) alone displayed profound hepatotoxicity, and serum level of ALT and AST increased more than sevenfold and sixfold, respectively. The administration of 0.2 (PaeH) and 0.1 (PaeM) g/kg paeoniflorin significantly reduced the serum levels of ALT and AST activities, respectively, which was almost equal to UDCA (p < 0.01). As shown in Fig. 1C–G, the ALP, TBIL, DBIT, TBA, and γ-GT levels, which were the indexes of character cholestasis, were significantly higher in ANIT-treated rats than in the normal group. ANIT alone significantly increased the serum TBIL (approximately threefold) and ALP levels (over threefold). In the rats treated with 0.2, 0.1, and 0.05 (PaeL) g/kg paeoniflorin, all the levels of ALP, TBIL, DBIL, γ-GT, and TBA, which were initially induced by ANIT, reduced significantly. Pathological observations provided a direct evidence of protective effects of paeoniflorin on cholestasis. As shown in Fig. 2A, the hepatic tissues in the rats of the normal group exhibited normal structure with no abnormal morphological change. The specimens in the model group (treated with ANIT) showed acute infiltration with polymorphous-clear neutrophils, edema, sinusoid congestion, severe demolition, or loss of the interlobular ducts and hepatic necrosis (Fig. 2B). The administration of UDCA and 0.2 g/kg paeoniflorin exhibited a milder degree of bile duct epithelial damage and hepatocyte hydropic degeneration with less neutrophil infiltration (Fig. 2C and D), which was similar to the normal group. The specimens that were treated with 0.1 g/kg paeoniflorin displayed a moderate reduced severity of inflammatory cell infiltration and other ANIT-induced histological damage (Fig. 2E). The liver damage in the specimens that were treated with 0.05 g/kg of paeoniflorin
Figure 1. Effects of paeoniflorin on serum biochemistry. The liver function markers in the serum biochemistry were (A) alanine transaminase (ALT); (B) aspartate transaminase (AST); (C) alkaline phosphatase (ALP); (D) total bilirubin (TBIL); (E) direct bilirubin (DBIL); (F) total bile acid ## (TBA); and (G) γ-glutamyltranspeptidase (γ-GT). Data are expressed as the Mean ± SD (n = 6 in each group). p < 0.01 compared with the ** * normal group; p < 0.01, p < 0.05 compared with the model group. Copyright © 2016 John Wiley & Sons, Ltd.
Phytother. Res. 30: 654–662 (2016)
657
SERUM METABOLOMIC
Figure 2. Effects of paeoniflorin on histological evaluations. (A) Normal group; (B) alpha-naphthylisothiocyanate-treated model; (C) model + UDCA 60 mg/kg; (D) model + paeoniflorin 0.2 g/kg; (E) model + paeoniflorin 0.1 g/kg; (F) model + paeoniflorin 0.05 g/kg (hematoxylin-eosin stained, 200× magnification). This figure is available in colour online at wileyonlinelibrary.com/journal/ptr.
was mildly attenuated with respect to portal tract edema, cholangitis, and bile duct epithelial damage (Fig. 2F). Multivariate statistical analysis All the data containing the retention time, peak intensity, and exact mass were imported into SIMCA-P 11.0 software (Umetrics AB) for multiple statistical analyses. Pattern recognition approaches including PCA and PLS-DA were usually used to classify metabolic phenotype and identify different metabolites in order to evaluate variation among complex data sets. At first, we specifically compared the normal, model, and 0.2 g/ kg paeoniflorin (PaeH) groups in order to obtain an explicit classification. Before multivariate statistical analysis, inadequate data of variables in LC-MS were excluded because of a high degree of analyzing interference. PCA was used to visualize general clustering, trend, or outliers among the observations. Score plot revealed a direct image of observational clusters. As seen in Fig. 3A and B, there was a distinguished classification between the clustering of normal and model groups and normal and PaeH groups in both positive and negative modes. However, when coming to clustering the model and PaeH, it was unable to classify these two groups. The results of PCA analysis indicated that further multivariate statistical analysis was needed to discern the relationship among the normal, model, and PaeH groups. Then, PLS-DA was applied to better understand the different metabolic patterns and find out potential biomarkers. In metabolomics approach, we constructed PLS-DA model, which has been widely used in metabolomics studies, to determine whether the metabolite fingerprints in the serum differed among the normal, model, and PaeH groups. As shown in Fig. 3C and D, there was a distinguished classification among the clustering of normal, model, and PaeH in both positive and negative models. Commonly, R2X, R2Y, and Q2 (cum) provided an estimate of how well the model fits the data. In our positive model, the R2X, R2Y, and Q2 (cum) of PLS-DA were 0.62, 0.989, and 0.789, respectively. Copyright © 2016 John Wiley & Sons, Ltd.
In addition, the R2X, R2Y, and Q2 (cum) of PLS-DA were 0.694, 0.996, and 0.838 in the negative model. It indicated that both of these two models obtain good quality and prediction characteristics. To validate the model, permutation tests with 100 iterations were further performed. Permutation tests compared the goodness of fit of the original model with the goodness of fit of randomly permuted models. As shown in Fig. 4E and F, the validation plots indicated that the original models were valid. In our results, the VIP value above 1.0 was one of the screening standards for the potential metabolites selection. Potential metabolites identification in cholestasis treatment The VIP value reflects the influence of every metabolite ion on the classification. Variables with a VIP value >1 influence more than the average one on the explanation of the Y matrix (classification). Therefore, the metabolite ions with a VIP value >1 were kept for further study. According to the VIP, 1008 variables among the normal, model, and PaeH groups were selected as the candidates for ANOVA analysis selection. Then, candidates that differed among the groups with a significant p value below 0.05 were identified as candidate biomarkers for METLIN and MetaboAnalyst identification. Among the metabolites acquired earlier, there were 14 potential biomarkers for distinguishing the difference among the normal, model, and PaeH groups. The detailed changes of potential biomarkers were also identified among two respective groups (Table 1). Biological pathway analysis With pattern recognition analysis of metabolites, 14 potential biomarkers for distinguishing the normal, model, and PaeH group were achieved. Then, more detailed analyses of pathways influenced among different groups were performed by MetaboAnalyst 3.0, which is a free web-based tool that combines results from powerful pathway enrichment analysis with the topology analysis. Phytother. Res. 30: 654–662 (2016)
658
Z. CHEN ET AL.
Figure 3. Multivariate statistical analysis. (A) Principal component analysis (PCA) score plot of the normal, model, and 0.2 g/kg paeoniflorin + groups in ESI model; (B) PCA score plot of the normal, model, and 0.2 g/kg paeoniflorin groups in ESI model; (C) partial least squares + discriminant analysis (PLS-DA) score plot of the normal, model, and 0.2 g/kg paeoniflorin groups in ESI model; (D) PLS-DA score plot of the normal, model, and 0.2 g/kg paeoniflorin groups in ESI model; (E) 100-permutation test of the normal, model, and 0.2 g/kg paeoniflorin + groups in ESI model; and (F) 100-permutation test of the normal, model, and 0.2 g/kg paeoniflorin groups in ESI model. This figure is available in colour online at wileyonlinelibrary.com/journal/ptr.
Metabolic pathway analysis revealed that 14 pathways were responsible for regulating cholestasis, including primary bile acid biosynthesis, glycerophospholipid metabolism, arachidonic acid metabolism, arginine and proline metabolism, ether lipid metabolism, and so on (Table 2 and Fig. 4). Signaling networks In order to reveal the internal relationships among these signal pathways, identified metabolites and pathways were imported into the KEGG to find interactions. The networks were primarily related to primary bile acid biosynthesis and glycerophospholipid metabolism based on the impact (value >0.1) and hits (value >3) in Table 2. According to the flow of the pathways, glycerophospholipid metabolism, linoleic acid metabolism, and GPI-anchor Copyright © 2016 John Wiley & Sons, Ltd.
biosythesis were considered to be the upstream signaling network. Then, arachidonic acid metabolism, ether lipid metabolism, glycine, serine and threonine metabolism, sphingolipid metabolism, and primary bile acid biosynthesis were considered to be the downstream signaling network (Fig. 5). From the prototypes of cholestasis, 0.2 g/kg paeoniflorin group demonstrated its significant efficacy. We applied the changes of 14 potential metabolites in cholestasis to catch a deep insight into the difference caused by different doses of paeoniflorin treatment. Compared with the model group, 0.2 g/kg paeoniflorin significantly decreased lyso PAF, creatine, inosine, and taurocholic acid. However, 0.1 g/kg and 0.05 g/kg paeoniflorin obtained a mild effect in these metabolites, indicating its limited ability to regulate bile acid synthesis or secretion (Fig. 6J–M). Furthermore, 15 (S)-HPETE, cortexolone, glycocholic acid, and Phytother. Res. 30: 654–662 (2016)
659
SERUM METABOLOMIC
Table 1. Identification and trends of change for differential metabolites
No
R.T. (min)
Mass (m/z)
Metabolites
KEGG
Formula
Trend in the model groupa
Trend in the PaeH groupb
1 2 3 4 5 6 7 8 9 10 11 12 13 14
10.5 9.4 10.58 9.45 0.85 11.94 21.31 16.88 0.97 3.95 15.84 9.35 16.35 10.65
336.229 346.213 449.311 465.305 174.113 833.595 785.617 183.066 131.069 145.052 509.393 515.291 805.553 499.293
15(S)-HPETE Cortexolone Glycochenodeoxycholic acid Glycocholic Acid L-Arginine Lecithin Phosphatidylethanolamine Phosphocholine Creatine Inosine Lyso PAF Taurocholic acid Lactosylceramide Taurochenodeoxycholic acid
C05966 C05488 C05466 C01921 C00062 C00157 C00350 C00588 C00300 C00294 C04317 C05122 C01290 C05465
C20H32O4 C21H30O4 C26H43NO5 C26H43NO6 C6H14N4O2 C10H18NO8PR2 C7H12NO8PR2 C5H15NO4P C4H9N3O2 C10H12N4O5 C8H20NO6PR C26H45NO7S C31H56NO13R C26H45NO6S
↓## ↓## ↑## ↑## ↓## ↓## ↓## ↓## ↑## ↑## ↑## ↑## ↓## ↑##
↑** ↑** ↓* ↓* ↑** ↑** ↑** ↑** ↓** ↓* ↓** ↓** ↑* ↓*
a
Change trends compared with the normal group. Changes trend compared with the model group. The levels of differential metabolites were marked with (↓) downregulated and (↑) upregulated. *p < 0.05 compared with the model group. **p < 0.01 compared with the model group. ## p < 0.01 compared with the model group. b
lactosylceramide were upregulated significantly in 0.2 g/kg and 0.1 g/kg paeoniflorin groups, but failed to be regulated in 0.05 g/kg paeoniflorin groups (Fig. 6A B, D, and N). Whereas, glycochenodeoxycholic acid, L-Arginine, lecithin, taurochenodeoxycholic acid, phosphatidylethanolamine, and phosphocholine could be upregulated significantly in all paeoniflorin groups (Fig. 6C, E–H, and O).
Figure 4. Summary of pathway analysis with MetaboAnalyst 3.0. (1) Primary bile acid biosynthesis; (2) glycerophospholipid metabolism; (3) arachidonic acid metabolism; (4) linoleic acid metabolism; (5) arginine and proline metabolism; (6) taurine and hypotaurine metabolism; (7) alpha-Linolenic acid metabolism; (8) ether lipid metabolism; (9) glycosylphosphatidylinositol (GPI)-anchor biosynthesis; (10) sphingolipid metabolism; (11) glycine, serine, and threonine metabolism; (12) aminoacyl-tRNA biosynthesis; (13) purine metabolism; and (14) steroid hormone biosynthesis. This figure is available in colour online at wileyonlinelibrary.com/journal/ptr. Copyright © 2016 John Wiley & Sons, Ltd.
DISCUSSION Cholestasis results in the intrahepatic retention of bile acid, which is one of the leading causes of death worldwide (Kakizaki et al., 2011). Its mechanism remains unclear completely despite of extensive studies concerning about it. Metabolomics focuses on a limited number of single pathways and aims at attempting to capture the complexity of metabolic networks (Wang et al., 2013). It has reported that metabolomics approach has been used to identify the disease states (Zhang et al., 2010; Zhang et al., 2012). In Addition, the high-throughput metabolomics makes it ideal to perform biomarker screens for diseases or drug efficacy (Wang et al., 2010). In this study, we assessed the effects and possible mechanisms of paeoniflorin against ANIT-induced liver injury. A metabolomics approach combined with pattern recognition approaches including PCA and PLS-DA was integrated to obtain differentiating metabolites for the pathways and clarify mechanisms of disease. We gave an illustrative example to show that the novel biomarker identification can be effectively solved by metabolomics method and then apply it to a paeoniflorin-related metabolic pathway in a global view. From the results of the serum biochemistry and pathological observations, it indicated that paeoniflorin could remarkably downregulate serum biochemistry indexes and alleviate the damage of liver tissue, which demonstrated that paeoniflorin could have therapeutic efficacy on ANIT-induced cholestasis. Furthermore, we have built the metabolomic feature profile and metabolites interaction network of paeoniflorin against cholestasis. The results showed that paeoniflorin presented protective effects on cholestasis by reversing potential biomarkers to normal levels. Especially, 14 metabolites were regulated in the pathway of primary bile acid biosynthesis, glycerophospholipid metabolism, linoleic acid metabolism, GPI-anchor biosynthesis, arachidonic acid Phytother. Res. 30: 654–662 (2016)
660
Z. CHEN ET AL.
Table 2. Result from ingenuity pathway analysis with MetaboAnalyst 3.0 Pathway name Primary bile acid biosynthesis Glycerophospholipid metabolism Arachidonic acid metabolism Linoleic acid metabolism Arginine and proline metabolism Taurine and hypotaurine metabolism alpha-Linolenic acid metabolism Ether lipid metabolism Glycosylphosphatidylinositol (GPI)-anchor biosynthesis Sphingolipid metabolism Glycine, serine, and threonine metabolism Aminoacyl-tRNA biosynthesis Purine metabolism Steroid hormone biosynthesis
Total
Expected
Hits
Raw p
log(p)
46 30 36 5 44 8 9 13 14 21 32 67 68 70
0.45934 0.29957 0.35949 0.049929 0.43937 0.079886 0.089872 0.12981 0.1398 0.2097 0.31954 0.66904 0.67903 0.699
4 3 2 1 2 1 1 1 1 1 1 1 1 1
0.00079939 0.0027476 0.048072 0.04901 0.068977 0.077335 0.086602 0.12281 0.13165 0.19126 0.27731 0.49784 0.50311 0.51349
7.1317 5.897 3.035 3.0157 2.674 2.5596 2.4464 2.0971 2.0276 1.6541 1.2826 0.69747 0.68695 0.66653
Impact 0.11904 0.275 0.09646 0 0.09426 0 0 0.21429 0.0439 0 0 0 0.0026 0.02409
The total are the number of compounds in the pathway; the hits are the actually matched number from the user uploaded data; the raw p is the original p value calculated from the enrichment analysis; and the impact is the pathway impact value calculated from pathway topology analysis.
metabolism, arginine and proline metabolism, ether lipid metabolism, glycine, serine and threonine metabolism, sphingolipid metabolism, and so on. Network reconstruction led to the integration of metabolites associated with the caused perturbation of multiple pathways. We observed that the significant downregulation components included creatine, inosine, lyso PAF, taurocholic acid, and the significant upregulation components included glycochenodeoxycholic acid, 15(S)HPETE, cortexolone, glycocholic acid, L-arginine, lecithin, phosphatidylethanolamine, lactosylceramide, taurochenodeoxycholic acid, and phosphocholine. The levels of 14 metabolites exhibited the marked changes among the normal, model, and PaeH group with high specificity and sensitivity as markers. It indicated that these metabolites might be the biomarkers, which were related to the action mechanism of peaoniflorin. Furthermore, we constructed the signaling network and found that primary bile acid biosynthesis and
glycerophospholipid metabolism might play important roles in cholestasis regulation based on the impact (value >0.1) and hits (value >3). Firstly, glycerophospholipid metabolism focuses on lecithin, phosphocholine, and phosphatidylethanolamine, which are the intermediates of the glycerophospholipid metabolism. Compared with the model group, the value of the three components was increased. It has been reported that the three components are the main lipid components of biomembranes, which indicates that the organism tries to repair the lesions of cell membranes and decrease the cell damage (Wright et al., 2004). Recent research has also illustrated that increase of phosphocholine could decrease cholesterol, which could decrease the liver damage of the organism (Volinsky et al., 2012). Phosphocholine also indirectly affects the ether lipid metabolism. Lecithin participates in the linoleic acid metabolism, which is related to the protection of cholestasis (Minich
Figure 5. Signaling networks. The red solid box represented metabolite higher in paeoniflorin-treated group than in the normal group; the blue solid box represented metabolite lower in paeoniflorin-treated group than in the normal group. AST, aspartate transaminase; ALT, alanine transaminase. This figure is available in colour online at wileyonlinelibrary.com/journal/ptr. Copyright © 2016 John Wiley & Sons, Ltd.
Phytother. Res. 30: 654–662 (2016)
SERUM METABOLOMIC
661
Figure 6. Changes of potential metabolites in cholestasis with paeoniflorin treatment. (A) 15(S)-HPETE; (B) cortexolone; (C) glycochenodeoxycholic acid; (D) glycocholic acid; (E) L-arginine; (F) lecithin; (G) phosphatidylethanolamine; (H) phosphocholine; (J) creatine; (K) inosine; (L) lyso PAF; (M) taurocholic acid; (N) lactosylceramide; and (O) taurochenodeoxycholic acid. **p < 0.01, *p < 0.05 compared ## # with the model group; p < 0.01, p < 0.05 compared with the normal group.
et al., 2000). Then, lecithin indirectly affects the arachidonic acid metabolism, and the increase tendency of the 15(S)-HPETE is the same as lecithin. As we all know, arachidonic acid metabolism is related to the inflammation, and there are many studies indicated that inhibit inflammation could reduce cholestasis (Lu et al., 2014). Lecithin also indirectly affects glycine, serine and threonine metabolism, purine metabolism, and sphingolipid metabolism and eventually indirectly affects the Kreb’s cycle. Kreb’s cycle is of vital importance for the survival of life and plays an important role in gluconeogenesis, transamination, deamination, and lipogenesis (Moffett and Namboodiri, 2003). Acetyl-CoA is the starting point for the Kreb’s cycle and obtained from various sources (Melendez-Hevia et al., 1996). In this study, many metabolites eventually affect the Kreb’s cycle, which is the bridge between the primary bile acid biosynthesis and glycerophospholipid metabolism. Primary bile acid biosynthesis focuses on taurocholic acid, glycocholic acid, taurochenodeoxycholic acid, and glycochenodeoxycholic acid, which are recognized as the potential metabolites in bile acid secretion. The enhancement of four components is commonly presented in cholestatic patients (Woolbright et al., 2015). Recent research has also illustrated that accumulation of several hydrophobic bile acids during cholestasis results in hepatotoxicity via apoptosis and inflammation (Dilger Copyright © 2016 John Wiley & Sons, Ltd.
et al., 2012). In our study, the four acids were significantly increased, and PeaH group remarkably downregulated the levels in the serum, which may partially reveal the probable therapeutic mechanism through bile acid synthesis and secretion. In urea cycle, L(D)-arginine and its downstream substrate 4-guanidinobutanoate were found to decline in the model group, whereas PeaH group significantly enhanced the contents. Previous studies demonstrate that modulation of L-arginine metabolism displays protective effects on cholestatic rats through NO synthasis and attenuating oxidative stress (Tain et al., 2013). In order to differentiate diseased and health states, it is a new powerful technology for the metabolomics study, which helps us have a better understanding of the disease, which lead to the discovery of new targets and the mechanism of the diseases (Tain et al., 2013). Application of metabolomic technologies for the study of cholestasis increases our understanding of the pathophysiological processes, and it should help us to identify potential biomarkers to develop new therapeutic strategies. System analysis of metabolic networks that are a central paradigm in biology will help us to identify new biomarkers, which will generate more in-depth understanding of cholestasis mechanism and thus provide better guidance for drug efficacy. In this study, our data have showed that paeoniflorin works by adjusting multiple metabolic pathways to the normal state. Additionally, Phytother. Res. 30: 654–662 (2016)
662
Z. CHEN ET AL.
the application of metabolite profiling tools can provide an efficient means for early diagnosis of cholestasis and offer a practical method for monitoring therapeutic intervention. Indeed, future metabolomic studies in human populations with cholestasis are still needed to validate the biomarkers found in the animal model.
CONCLUSION Our study highlights the importance of the power of metabolomics to elucidate metabolic characters of cholestasis and therapeutic effects of paeoniflorin at the global metabolomics levels. Analyzing the topology of the network, we detected 14 potential biomarkers and predicted the major metabolites network of cholestasis using the validated pattern recognition methods. These biomarkers that represent molecular targets for cholestasis disease patterns were identified and related to the acutely perturbed pathway of primary bile acid biosynthesis, glycerophospholipid metabolism,
arachidonic acid metabolism, arginine and proline metabolism, ether lipid metabolism, and so on. Of note, the results indicated that paeoniflorin could provide satisfactory effects on cholestasis through regulating the multiple perturbed networks to their normal state. Taken together, the present study proves that metabolomics is a promising approach to unraveling hepatoprotective effects of paeoniflorin by partially regulating the perturbed pathways, which provides deep insights into the mechanisms of cholestasis.
Acknowledgements The current work was supported by Major Projects of the National Science and Technology (no. 2012ZX10005010-002-002) and the National Natural Science Foundation Project of China (nos. 81303120 and 81173571).
Conflict of Interest The authors have declared that there is no conflict of interest.
REFERENCES Dilger K, Hohenester S, Winkler-Budenhofer U, et al. 2012. Effect of ursodeoxycholic acid on bile acid profiles and intestinal detoxification machinery in primary biliary cirrhosis and health. J Hepatol 57: 133–140. Hirschfield GM, Heathcote EJ, Gershwin ME. 2010. Pathogenesis of cholestatic liver disease and therapeutic approaches. Gastroenterology 139: 1481–1496. Hofmann AF. 2002. Cholestatic liver disease: pathophysiology and therapeutic options. Liver 22: 14–19. Jiang FJ, Zhao YL, Wang JB, et al. 2011. Comparative pharmacokinetic study of paeoniflorin and albiflorin after oral administration of Radix Paeoniae rubra in normal rats and the acute cholestasis hepatitis rats. Fitoterapia 83: 415–421. Kakizaki S, Takizawa D, Tojima H. 2011. Nuclear receptors CAR and PXR: therapeutic targets for cholestatic liver disease. Front Biosci 16: 2988–3005. Lu X, Tian Y, Lian X, et al. 2014. Integrated systems toxicology approaches identified the possible involvement of ABC transporters pathway in erythromycin estolate-induced liver injury in rat. Food Chem Toxicol 65: 343–355. Melendez-Hevia E, Waddell TG, Cascante M. 1996. The puzzle of the Krebs citric acid cycle: assembling the pieces of chemically feasible reactions, and opportunism in the design of metabolic pathways during evolution. J Mol Evol 43: 293–303. Minich DM, Havinga R, Stellaard F, Vonk RJ, Kuipers F, Verkade HJ. 2000. Intestinal absorption and postabsorptive metabolism of linoleic acid in rats with short-term bile duct ligation. Am J Physiol Gastrointest Liver Physiol 279: G1242– G1248. Moffett JR, Namboodiri MA. 2003. Tryptophan and the immune response. Immunol Cell Biol 81: 247–265. Padda MS, Sanchez M, Akhtar AJ, Boyer JL. 2011. Drug-induced cholestasis. J Hepatol 53: 1377–1387. Sun H, Zhang A, Wang X. 2012. Potential role of metabolomic approaches for Chinese medicine syndromes and herbal medicine. Phytother Res 26: 1466–1471. Sun H, Zhang AH, Zou DX, Sun WJ, Wu XH, Wang XJ. 2014. Metabolomics coupled with pattern recognition and pathway analysis on potential biomarkers in liver injury and hepatoprotective effects of yinchenhao. Appl Biochem Biotechnol 173: 857–869. Tain YL, Chen CC, Lee CT, et al. 2013. Melatonin regulates L-arginine transport and NADPH oxidase in young rats with bile duct ligation: role of protein kinase C. Pediatr Res 73: 395–401.
Copyright © 2016 John Wiley & Sons, Ltd.
Volinsky R, Paananen R, Kinnunen PK. 2012. Oxidized phosphatidylcholines promote phase separation of cholesterol-sphingomyelin domains. Biophys J 103: 247–254. Wang P, Sun H, Lv H, et al. 2010. Thyroxine and reserpine-induced changes in metabolic profiles of rat urine and the therapeutic effect of Liu Wei Di Huang Wan detected by UPLC-HDMS. J Pharm Biomed Anal 53: 631–645. Wang X, Sun H, Zhang A, Sun W, Wang P, Wang Z. 2011a. Potential role of metabolomics approaches in the area of traditional Chinese medicine: as pillars of the bridge between Chinese and Western medicine. J Pharm Biomed Anal 55: 859–868. Wang X, Wang H, Zhang A, Sun W, Wang P, Wang Z. 2011b. Metabolomics study on the toxicity of aconite root and its processed products using ultraperformance liquid-chromatography/ electrospray-ionization synapt high-definition mass spectrometry coupled with pattern recognition approach and ingenuity pathways analysis. J Proteome Res 11: 1284–1301. Wang X, Wang Q, Zhang A, et al. 2013. Metabolomics study of intervention effects of Wen-Xin-formula using ultra high-performance liquid chromatography/mass spectrometry coupled with pattern recognition approach. J Pharm Biomed 74: 22–30. Watanabe N, Takashimizu S, Kojima S, et al. 2007. Clinical and pathological features of a prolonged type of acute intrahepatic cholestasis. Hepatol Res 37: 598–607. Woolbright BL, Dorko K, Antoine DJ, et al. 2015. Bile acid-induced necrosis in primary human hepatocytes and in patients with obstructive cholestasis. Toxicol Appl Pharmacol 283: 168–177. Wright MM, Howe AG, Zaremberg V. 2004. Cell membranes and apoptosis: role of cardiolipin, phosphatidylcholine, and anticancer lipid analogues. Biochem Cell Biol 82: 18–26. Yang H, Ramani K, Xia M, et al. 2009. Dysregulation of glutathione synthesis during cholestasis in mice: molecular mechanisms and therapeutic implications. J Hepatol 49: 1982–1991. Zhang A, Sun H, Wang Z, Sun W, Wang P, Wang XJ. 2010. Metabolomics: towards understanding traditional Chinese medicine. Planta Med 76: 2026–2035. Zhang A, Sun H, Wang X. 2012. Serum metabolomics as a novel diagnostic approach for disease: a systematic review. Anal Bioanal Chem 404: 1239–1245. Zhao Y, Zhou G, Wang J, et al. 2013. Paeoniflorin protects against ANIT-induced cholestasis by ameliorating oxidative stress in rats. Food Chem Toxicol 58: 242–248.
Phytother. Res. 30: 654–662 (2016)
Serum Metabolomic Profiling in a Rat Model Reveals Protective Function of Paeoniflorin Against ANIT Induced Cholestasis Zhe Chen,1,2† Yun Zhu,3† Yanling Zhao,1* Xiao Ma,1,2 Ming Niu,4 Jiabo Wang,4 Haibin Su,5 Ruilin Wang,3 Jianyu Li,3 Liping Liu,1 Zhenman Wei,1 Qingguo Zhao,1 Hongge Chen1 and Xiaohe Xiao4* 1
Department of Pharmacy, 302 Hospital of People’s Liberation Army, Beijing 100039, China College of Pharmacy, Chengdu University of Traditional Chinese Medicine, Chengdu 611137, China 3 Department of Integrative Medical Center, 302 Hospital of People’s Liberation Army, Beijing 100039, China 4 China Military Institute of Chinese Medicine, 302 Hospital of People’s Liberation Army, Beijing 100039, China 5 Liver Failure Treatment and Research Center, 302 Hospital of People’s Liberation Army, Beijing 100039, China 2
Cholestasis is a leading cause of hepatic accumulation of bile acids resulting in liver injury, fibrosis, and liver failure. Paeoniflorin displays bright prospects in liver protective effect. However, its molecular mechanism has not been well-explored. This study was designed to assess the effects and possible mechanisms of paeoniflorin against alpha-naphthylisothiocyanate-induced liver injury. Ultraperformance liquid chromatography coupled with quadrupole time-of-flight combined with principle component analysis and partial least squares discriminant analysis were integrated to obtain differentiating metabolites for the pathways and clarify mechanisms of disease. The results indicated that paeoniflorin could remarkably downregulate serum biochemical indexes and alleviate the histological damage of liver tissue. Different expression of 14 metabolites demonstrated that paeoniflorin mainly regulated the dysfunctions of glycerophospholipid metabolism and primary bile acid biosynthesis. Moreover, several pathways such as arginine and proline metabolism, ether lipid metabolism, and arachidonic acid metabolism were also related to the efficacy. In conclusion, paeoniflorin has indicated favorable pharmacological effect on serum biochemical indexes and pathological observation on cholestatic model. And metabolomics is a promising approach to unraveling hepatoprotective effects by partially regulating the perturbed pathways, which provide insights into mechanisms of cholestasis. Copyright © 2016 John Wiley & Sons, Ltd. Keywords: metabolomics; cholestasis; paeoniflorin; pattern recognition approaches; mechanism.
INTRODUCTION As one of the chronic liver disease, cholestasis arises from impaired hepatobiliary production and excretion of bile, which is one of the top 15 causes of death in the USA (Yang et al., 2009). In these disorders, injuries to bile ducts or liver cells can lead to a range of clinical presentations, such as isolated abnormalities in liver biochemistry, liver failure, hepatobiliary malignancy, congenital, immunologic, and toxic factors, which can contribute to disease (Hirschfield et al., 2010). As one of the most common and devastating manifestations among hereditary and acquired liver diseases, cholestasis ultimately leads to liver failure and cirrhosis (Hofmann, 2002; Watanabe et al., 2007). In humans, it frequently occurs because of various endogenous and exogenous
* Correspondence to: Yanling Zhao, Department of Pharmacy, 302 Hospital of People’s Liberation Army, Beijing 100039, China; Xiaohe Xiao, China Military Institute of Chinese Medicine, 302 Hospital of People’s Liberation Army, Beijing 100039, China. E-mail: [email protected] (Y. Zhao); [email protected] (X. Xiao) Contract/grant sponsor: National Science and Technology; contract/grant number: 2012ZX10005010-002-002. Contract/grant sponsor: National Natural Science Foundation Project of China; contract/grant numbers: 81303120, 81173571. † These authors are thought to have equal contributions.
Copyright © 2016 John Wiley & Sons, Ltd.
factors, such as dysregulation of bile acid transporters, oxidation stress, inflammation in hepatocytes, several drugs, alcohol abuse, and virus infection (Padda et al., 2011). While these factors are thought to be involved in the pathogenesis of cholestasis, the remarkable and important metabolites in cholestasis are not yet precisely recognized. As an omic science in systems biology, metabolomics represents a powerful discipline concerned with the comprehensive analysis of small molecules and provides a powerful approach to discovering biomarkers in biological systems (Wang et al., 2013). It has recently demonstrated significant potential in many fields such as toxicology, disease diagnosis, drug mechanism and development, responses to environmental stress, nutrition, new targets, and natural product discovery (Wang et al., 2011a). Based on analyzing the specific early biomarkers during disease or drug treatment, metabolomics provides a holistic insight into the relationship between substance and metabolic pathways (Wang et al., 2011b; Sun et al., 2012). As the newest of the ‘omics’ sciences, metabolomics has brought much excitement to the field of life sciences as a potential translation tool. Especially, its method and design resemble those of traditional Chinese medicine (TCM), indicating that metabolomics has the potential to impact our understanding of the TCM (Wang et al., Received 26 August 2015 Revised 25 November 2015 Accepted 22 December 2015
SERUM METABOLOMIC
2013). Yinchenhao decoction, a Chinese famous formula for treating jaundice, was proved to be effective on regulating multiple perturbed pathways to normal state in jaundice model by metabolomics (Sun et al., 2014). Paeoniflorin is one of the main bioactive components of the dried root of Paeonia lactiflora pall and Paeonia veitchii Lynch (Jiang et al., 2011). Clinical studies have demonstrated that Chi Shao can confer excellent therapeutic effects on hepatitis with severe cholestasis (Zhao et al., 2013). In this study, we assessed the hepatoprotective effects of paeoniflorin against liver injury induced by alpha-naphthylisothiocyanate (ANIT) through serum biochemical indexes and pathological observation. Metabolomics applying ultraperformance liquid chromatography coupled with quadruple time-of-flight (UPLC-Q-TOF) characterized the metabolic profiles of paeoniflorin protective efficacy on cholestasis. Principle component analysis (PCA) and partial least squares discriminant analysis (PLS-DA) were integrated to obtain differentiating metabolites for the pathways and clarify the liver protection mechanisms of paeoniflorin on cholestasis, which provided deep insights into mechanisms of cholestasis.
MATERIALS AND METHODS Chemicals and reagents. Paeoniflorin was purchased from the National Institute for the Control of Pharmaceutical and Biological Products (Beijing, China). ANIT was purchased from Sigma-Aldrich Corporation (St Louis, MO, USA). Ursodeoxycholic acid (UDCA) was purchased from Shanghai Zhongxi Pharmaceutical Factory (Shanghai, PR China). Total bilirubin (TBIL), direct bilirubin (DBIL), aspartate transaminase (AST), alanine transaminase (ALT), alkaline phosphatase (ALP), γ-glutamyltranspeptidase (γ-GT), and total bile acid (TBA) assay kits were obtained from Nanjing Jiancheng Bioengineering Institute (Nanjing, PR China). All the other chemicals of analytical grade were purchased from commercial sources.
Animals and treatments. Male Sprague-Dawley rats weighing 200 ± 20 g were obtained from the laboratory animal center of the Military Medical Science Academy of the PLA (Permission No. SCXK-(A) 2012-0004). All of the animals were acclimated for 1 week prior to the experiment and were kept under the same temperature (25 ± 2 °C) and lighting (12:12 h light : dark cycle) conditions. The whole studies were performed in accordance with the guidelines of the Council on Animal Care of Academia Sinica. The animals were randomly divided into six groups with six rats in each group. The rats in the normal group serving as normal control were given saline each day and treated with the vehicle (olive oil) alone. The model group was treated with 50 mg/kg ANIT dissolved in an equal volume of olive oil by gavage, and this dosage is known to induce cholestasis based on a previous study. UDCA (60 mg/kg), the positive control, was treated to rats with the same condition of model group. Paeoniflorin was dissolved in saline, and it was given to experimental rats Copyright © 2016 John Wiley & Sons, Ltd.
655
at doses of 0.2 (PaeH), 0.1 (PaeM), or 0.05 (PaeL) g/kg body weight, respectively, for six times before and four times after they were treated with 50 mg/kg ANIT by gavage. In this study, paeoniflorin doses that were adopted were based on preliminary experiment.
Sample preparation. Before the rats were sacrificed, they were fasted for 12 h after the last paeoniflorin administration. Blood samples were collected and centrifuged at 3000 × g for 10 min to obtain the serum. All of the serum samples were sterile, hemolysis-free, and stored at 80 °C before determining the biochemical indices and metabolomics analysis. The serum levels of TBIL, DBIL, AST, ALT, ALP, and TBA were measured by commercial test kits. Liver tissues were excised and fixed in 10% phosphate buffer solusion (PBS) buffered formalin. Three or four paraffin-embedded sections (4- to 5-μm thick) per specimen were prepared and stained with hematoxylin-eosin (HE staining). The stained sections were examined under Nikon microscope (Nikon Instruments Corporation, Shanghai, China) and analyzed by Image-Pro Plus 7200 software (Media Cybernetics, Inc., Rockville, MD, USA).
Metabolic profiling. Chromatography was performed using an Agilent 1290 series UPLC system equipped with quaternary pump, online degasser, auto sampler, and thermostat column compartment. The injection volume was fixed at 4 μL. All the samples were maintained at 4 °C during the analysis. The separation was performed on a ZORBAX RRHD 300 SB-C18 column (2.1 mm × 100 mm, 1.8 μm; Agilent, USA). The column temperature was set at 30 °C. The mobile phases were composed of 0.1% formic acid in acetonitrile (solvent A) and 0.1% formic acid in water (solvent B). The flow rate was set as 0.30 mL/min. The gradient was used as follows: a linear gradient of 95% A over initial– 1.0 min; 95–60% A over 1.0–9.0 min; 60–10% A over 9.0–19.0 min; 10–0% A over 19.0–21.0 min; and 0% A over 21.0–25.0 min. Then the eluent was introduced to the mass spectrometer directly. After every 10 samples injected, a pooled sample as the QC sample followed by a blank was injected to ensure the stability and repeatability of the liquid chromatography–mass spectrometry (LC-MS) systems. For MS, an Agilent 6550 Q-TOF/MS with an electrospray ionization source (ESI) in both positive and negative mode was used. Ionization was achieved using electrospray. The electrospray source parameters were fixed as follows: electrospray capillary voltage was 3.0 kV in negative ionization mode and 4.0 kV in positive ionization mode. The mass range was set from m/z 80 to 1000. Gas temperature was 200 °C in negative ionization mode and 225 °C in positive ionization mode. Gas flow was 11 L/min. Nebulizer was set to 35 pisg (negative) and 45 pisg (positive). Sheath gas temperature was 350 °C, and sheath gas flow was 12 L/min. Nozzle voltage was 500 V in both negative and positive mode.
Data extraction and pattern recognition analysis. Sample data were extracted by MassHunter Profinder software (Agilent, California, USA) for peak detection Phytother. Res. 30: 654–662 (2016)
656
Z. CHEN ET AL.
and alignment. Full scan mode was employed in the mass range of 80 to 1000 m/z. The initial and final retention time were set for data collection. The resultant data matrices were introduced to SIMCA-P 11.0 software (Umetrics AB, Umea, Sweden) for PCA and PLS-DA. Prior to PCA, all variables obtained from data matrix were mean-centered and scaled to Pareto variance. PCA, an unsupervised pattern recognition approach, is used to reduce the dimension of UPLC-MS data and to disclose intrinsic clustering of samples. To maximize the differences in inter-class discrimination and minimize the differences in inter-class discrimination, the data were further analyzed using the PLS-DA method. These variables with a higher VIP value (VIP ≥1.0) in PLS-DA method were considered to be further evaluated by one-way analysis of variance (ANOVA). Biomarkers identification and pathway enrichment analysis. Only the variables (p < 0.05 in ANOVA) were selected as potential biomarkers and subjected to further identification of the molecular mechanism. The potential biomarkers were identified by online biochemical database service METLIN (http://metlin.scripps.edu/). The compound names and KEGG (http://www.kegg.jp/) numbers of potential biomarkers were performed with MetaboAnalyst 3.0 (http://www.metaboanalyst.ca/) for further enrichment and pathway analysis. Statistical analysis. Data were expressed as the Mean ± SD and analyzed with the SPSS software program (version 13.0, Chicago, USA). The differences between the group means were calculated by ANOVA. The differences were considered to be statistically significant when p < 0.05 and highly significant when p < 0.01. The resulting three-dimensional matrix containing peak index, sample name, and peak area were introduced into SIMCA-P 11.0 software (Umetrics AB, Umea, Sweden), which was used for pattern recognition analysis including PCA and PLS-DA.
RESULTS Biochemical analysis and pathological observations The serum ALT and AST activities were measured as indexes of liver cell damage. As shown in Fig. 1A and B, the rats that were given ANIT (50 mg/kg i.g.) alone displayed profound hepatotoxicity, and serum level of ALT and AST increased more than sevenfold and sixfold, respectively. The administration of 0.2 (PaeH) and 0.1 (PaeM) g/kg paeoniflorin significantly reduced the serum levels of ALT and AST activities, respectively, which was almost equal to UDCA (p < 0.01). As shown in Fig. 1C–G, the ALP, TBIL, DBIT, TBA, and γ-GT levels, which were the indexes of character cholestasis, were significantly higher in ANIT-treated rats than in the normal group. ANIT alone significantly increased the serum TBIL (approximately threefold) and ALP levels (over threefold). In the rats treated with 0.2, 0.1, and 0.05 (PaeL) g/kg paeoniflorin, all the levels of ALP, TBIL, DBIL, γ-GT, and TBA, which were initially induced by ANIT, reduced significantly. Pathological observations provided a direct evidence of protective effects of paeoniflorin on cholestasis. As shown in Fig. 2A, the hepatic tissues in the rats of the normal group exhibited normal structure with no abnormal morphological change. The specimens in the model group (treated with ANIT) showed acute infiltration with polymorphous-clear neutrophils, edema, sinusoid congestion, severe demolition, or loss of the interlobular ducts and hepatic necrosis (Fig. 2B). The administration of UDCA and 0.2 g/kg paeoniflorin exhibited a milder degree of bile duct epithelial damage and hepatocyte hydropic degeneration with less neutrophil infiltration (Fig. 2C and D), which was similar to the normal group. The specimens that were treated with 0.1 g/kg paeoniflorin displayed a moderate reduced severity of inflammatory cell infiltration and other ANIT-induced histological damage (Fig. 2E). The liver damage in the specimens that were treated with 0.05 g/kg of paeoniflorin
Figure 1. Effects of paeoniflorin on serum biochemistry. The liver function markers in the serum biochemistry were (A) alanine transaminase (ALT); (B) aspartate transaminase (AST); (C) alkaline phosphatase (ALP); (D) total bilirubin (TBIL); (E) direct bilirubin (DBIL); (F) total bile acid ## (TBA); and (G) γ-glutamyltranspeptidase (γ-GT). Data are expressed as the Mean ± SD (n = 6 in each group). p < 0.01 compared with the ** * normal group; p < 0.01, p < 0.05 compared with the model group. Copyright © 2016 John Wiley & Sons, Ltd.
Phytother. Res. 30: 654–662 (2016)
657
SERUM METABOLOMIC
Figure 2. Effects of paeoniflorin on histological evaluations. (A) Normal group; (B) alpha-naphthylisothiocyanate-treated model; (C) model + UDCA 60 mg/kg; (D) model + paeoniflorin 0.2 g/kg; (E) model + paeoniflorin 0.1 g/kg; (F) model + paeoniflorin 0.05 g/kg (hematoxylin-eosin stained, 200× magnification). This figure is available in colour online at wileyonlinelibrary.com/journal/ptr.
was mildly attenuated with respect to portal tract edema, cholangitis, and bile duct epithelial damage (Fig. 2F). Multivariate statistical analysis All the data containing the retention time, peak intensity, and exact mass were imported into SIMCA-P 11.0 software (Umetrics AB) for multiple statistical analyses. Pattern recognition approaches including PCA and PLS-DA were usually used to classify metabolic phenotype and identify different metabolites in order to evaluate variation among complex data sets. At first, we specifically compared the normal, model, and 0.2 g/ kg paeoniflorin (PaeH) groups in order to obtain an explicit classification. Before multivariate statistical analysis, inadequate data of variables in LC-MS were excluded because of a high degree of analyzing interference. PCA was used to visualize general clustering, trend, or outliers among the observations. Score plot revealed a direct image of observational clusters. As seen in Fig. 3A and B, there was a distinguished classification between the clustering of normal and model groups and normal and PaeH groups in both positive and negative modes. However, when coming to clustering the model and PaeH, it was unable to classify these two groups. The results of PCA analysis indicated that further multivariate statistical analysis was needed to discern the relationship among the normal, model, and PaeH groups. Then, PLS-DA was applied to better understand the different metabolic patterns and find out potential biomarkers. In metabolomics approach, we constructed PLS-DA model, which has been widely used in metabolomics studies, to determine whether the metabolite fingerprints in the serum differed among the normal, model, and PaeH groups. As shown in Fig. 3C and D, there was a distinguished classification among the clustering of normal, model, and PaeH in both positive and negative models. Commonly, R2X, R2Y, and Q2 (cum) provided an estimate of how well the model fits the data. In our positive model, the R2X, R2Y, and Q2 (cum) of PLS-DA were 0.62, 0.989, and 0.789, respectively. Copyright © 2016 John Wiley & Sons, Ltd.
In addition, the R2X, R2Y, and Q2 (cum) of PLS-DA were 0.694, 0.996, and 0.838 in the negative model. It indicated that both of these two models obtain good quality and prediction characteristics. To validate the model, permutation tests with 100 iterations were further performed. Permutation tests compared the goodness of fit of the original model with the goodness of fit of randomly permuted models. As shown in Fig. 4E and F, the validation plots indicated that the original models were valid. In our results, the VIP value above 1.0 was one of the screening standards for the potential metabolites selection. Potential metabolites identification in cholestasis treatment The VIP value reflects the influence of every metabolite ion on the classification. Variables with a VIP value >1 influence more than the average one on the explanation of the Y matrix (classification). Therefore, the metabolite ions with a VIP value >1 were kept for further study. According to the VIP, 1008 variables among the normal, model, and PaeH groups were selected as the candidates for ANOVA analysis selection. Then, candidates that differed among the groups with a significant p value below 0.05 were identified as candidate biomarkers for METLIN and MetaboAnalyst identification. Among the metabolites acquired earlier, there were 14 potential biomarkers for distinguishing the difference among the normal, model, and PaeH groups. The detailed changes of potential biomarkers were also identified among two respective groups (Table 1). Biological pathway analysis With pattern recognition analysis of metabolites, 14 potential biomarkers for distinguishing the normal, model, and PaeH group were achieved. Then, more detailed analyses of pathways influenced among different groups were performed by MetaboAnalyst 3.0, which is a free web-based tool that combines results from powerful pathway enrichment analysis with the topology analysis. Phytother. Res. 30: 654–662 (2016)
658
Z. CHEN ET AL.
Figure 3. Multivariate statistical analysis. (A) Principal component analysis (PCA) score plot of the normal, model, and 0.2 g/kg paeoniflorin + groups in ESI model; (B) PCA score plot of the normal, model, and 0.2 g/kg paeoniflorin groups in ESI model; (C) partial least squares + discriminant analysis (PLS-DA) score plot of the normal, model, and 0.2 g/kg paeoniflorin groups in ESI model; (D) PLS-DA score plot of the normal, model, and 0.2 g/kg paeoniflorin groups in ESI model; (E) 100-permutation test of the normal, model, and 0.2 g/kg paeoniflorin + groups in ESI model; and (F) 100-permutation test of the normal, model, and 0.2 g/kg paeoniflorin groups in ESI model. This figure is available in colour online at wileyonlinelibrary.com/journal/ptr.
Metabolic pathway analysis revealed that 14 pathways were responsible for regulating cholestasis, including primary bile acid biosynthesis, glycerophospholipid metabolism, arachidonic acid metabolism, arginine and proline metabolism, ether lipid metabolism, and so on (Table 2 and Fig. 4). Signaling networks In order to reveal the internal relationships among these signal pathways, identified metabolites and pathways were imported into the KEGG to find interactions. The networks were primarily related to primary bile acid biosynthesis and glycerophospholipid metabolism based on the impact (value >0.1) and hits (value >3) in Table 2. According to the flow of the pathways, glycerophospholipid metabolism, linoleic acid metabolism, and GPI-anchor Copyright © 2016 John Wiley & Sons, Ltd.
biosythesis were considered to be the upstream signaling network. Then, arachidonic acid metabolism, ether lipid metabolism, glycine, serine and threonine metabolism, sphingolipid metabolism, and primary bile acid biosynthesis were considered to be the downstream signaling network (Fig. 5). From the prototypes of cholestasis, 0.2 g/kg paeoniflorin group demonstrated its significant efficacy. We applied the changes of 14 potential metabolites in cholestasis to catch a deep insight into the difference caused by different doses of paeoniflorin treatment. Compared with the model group, 0.2 g/kg paeoniflorin significantly decreased lyso PAF, creatine, inosine, and taurocholic acid. However, 0.1 g/kg and 0.05 g/kg paeoniflorin obtained a mild effect in these metabolites, indicating its limited ability to regulate bile acid synthesis or secretion (Fig. 6J–M). Furthermore, 15 (S)-HPETE, cortexolone, glycocholic acid, and Phytother. Res. 30: 654–662 (2016)
659
SERUM METABOLOMIC
Table 1. Identification and trends of change for differential metabolites
No
R.T. (min)
Mass (m/z)
Metabolites
KEGG
Formula
Trend in the model groupa
Trend in the PaeH groupb
1 2 3 4 5 6 7 8 9 10 11 12 13 14
10.5 9.4 10.58 9.45 0.85 11.94 21.31 16.88 0.97 3.95 15.84 9.35 16.35 10.65
336.229 346.213 449.311 465.305 174.113 833.595 785.617 183.066 131.069 145.052 509.393 515.291 805.553 499.293
15(S)-HPETE Cortexolone Glycochenodeoxycholic acid Glycocholic Acid L-Arginine Lecithin Phosphatidylethanolamine Phosphocholine Creatine Inosine Lyso PAF Taurocholic acid Lactosylceramide Taurochenodeoxycholic acid
C05966 C05488 C05466 C01921 C00062 C00157 C00350 C00588 C00300 C00294 C04317 C05122 C01290 C05465
C20H32O4 C21H30O4 C26H43NO5 C26H43NO6 C6H14N4O2 C10H18NO8PR2 C7H12NO8PR2 C5H15NO4P C4H9N3O2 C10H12N4O5 C8H20NO6PR C26H45NO7S C31H56NO13R C26H45NO6S
↓## ↓## ↑## ↑## ↓## ↓## ↓## ↓## ↑## ↑## ↑## ↑## ↓## ↑##
↑** ↑** ↓* ↓* ↑** ↑** ↑** ↑** ↓** ↓* ↓** ↓** ↑* ↓*
a
Change trends compared with the normal group. Changes trend compared with the model group. The levels of differential metabolites were marked with (↓) downregulated and (↑) upregulated. *p < 0.05 compared with the model group. **p < 0.01 compared with the model group. ## p < 0.01 compared with the model group. b
lactosylceramide were upregulated significantly in 0.2 g/kg and 0.1 g/kg paeoniflorin groups, but failed to be regulated in 0.05 g/kg paeoniflorin groups (Fig. 6A B, D, and N). Whereas, glycochenodeoxycholic acid, L-Arginine, lecithin, taurochenodeoxycholic acid, phosphatidylethanolamine, and phosphocholine could be upregulated significantly in all paeoniflorin groups (Fig. 6C, E–H, and O).
Figure 4. Summary of pathway analysis with MetaboAnalyst 3.0. (1) Primary bile acid biosynthesis; (2) glycerophospholipid metabolism; (3) arachidonic acid metabolism; (4) linoleic acid metabolism; (5) arginine and proline metabolism; (6) taurine and hypotaurine metabolism; (7) alpha-Linolenic acid metabolism; (8) ether lipid metabolism; (9) glycosylphosphatidylinositol (GPI)-anchor biosynthesis; (10) sphingolipid metabolism; (11) glycine, serine, and threonine metabolism; (12) aminoacyl-tRNA biosynthesis; (13) purine metabolism; and (14) steroid hormone biosynthesis. This figure is available in colour online at wileyonlinelibrary.com/journal/ptr. Copyright © 2016 John Wiley & Sons, Ltd.
DISCUSSION Cholestasis results in the intrahepatic retention of bile acid, which is one of the leading causes of death worldwide (Kakizaki et al., 2011). Its mechanism remains unclear completely despite of extensive studies concerning about it. Metabolomics focuses on a limited number of single pathways and aims at attempting to capture the complexity of metabolic networks (Wang et al., 2013). It has reported that metabolomics approach has been used to identify the disease states (Zhang et al., 2010; Zhang et al., 2012). In Addition, the high-throughput metabolomics makes it ideal to perform biomarker screens for diseases or drug efficacy (Wang et al., 2010). In this study, we assessed the effects and possible mechanisms of paeoniflorin against ANIT-induced liver injury. A metabolomics approach combined with pattern recognition approaches including PCA and PLS-DA was integrated to obtain differentiating metabolites for the pathways and clarify mechanisms of disease. We gave an illustrative example to show that the novel biomarker identification can be effectively solved by metabolomics method and then apply it to a paeoniflorin-related metabolic pathway in a global view. From the results of the serum biochemistry and pathological observations, it indicated that paeoniflorin could remarkably downregulate serum biochemistry indexes and alleviate the damage of liver tissue, which demonstrated that paeoniflorin could have therapeutic efficacy on ANIT-induced cholestasis. Furthermore, we have built the metabolomic feature profile and metabolites interaction network of paeoniflorin against cholestasis. The results showed that paeoniflorin presented protective effects on cholestasis by reversing potential biomarkers to normal levels. Especially, 14 metabolites were regulated in the pathway of primary bile acid biosynthesis, glycerophospholipid metabolism, linoleic acid metabolism, GPI-anchor biosynthesis, arachidonic acid Phytother. Res. 30: 654–662 (2016)
660
Z. CHEN ET AL.
Table 2. Result from ingenuity pathway analysis with MetaboAnalyst 3.0 Pathway name Primary bile acid biosynthesis Glycerophospholipid metabolism Arachidonic acid metabolism Linoleic acid metabolism Arginine and proline metabolism Taurine and hypotaurine metabolism alpha-Linolenic acid metabolism Ether lipid metabolism Glycosylphosphatidylinositol (GPI)-anchor biosynthesis Sphingolipid metabolism Glycine, serine, and threonine metabolism Aminoacyl-tRNA biosynthesis Purine metabolism Steroid hormone biosynthesis
Total
Expected
Hits
Raw p
log(p)
46 30 36 5 44 8 9 13 14 21 32 67 68 70
0.45934 0.29957 0.35949 0.049929 0.43937 0.079886 0.089872 0.12981 0.1398 0.2097 0.31954 0.66904 0.67903 0.699
4 3 2 1 2 1 1 1 1 1 1 1 1 1
0.00079939 0.0027476 0.048072 0.04901 0.068977 0.077335 0.086602 0.12281 0.13165 0.19126 0.27731 0.49784 0.50311 0.51349
7.1317 5.897 3.035 3.0157 2.674 2.5596 2.4464 2.0971 2.0276 1.6541 1.2826 0.69747 0.68695 0.66653
Impact 0.11904 0.275 0.09646 0 0.09426 0 0 0.21429 0.0439 0 0 0 0.0026 0.02409
The total are the number of compounds in the pathway; the hits are the actually matched number from the user uploaded data; the raw p is the original p value calculated from the enrichment analysis; and the impact is the pathway impact value calculated from pathway topology analysis.
metabolism, arginine and proline metabolism, ether lipid metabolism, glycine, serine and threonine metabolism, sphingolipid metabolism, and so on. Network reconstruction led to the integration of metabolites associated with the caused perturbation of multiple pathways. We observed that the significant downregulation components included creatine, inosine, lyso PAF, taurocholic acid, and the significant upregulation components included glycochenodeoxycholic acid, 15(S)HPETE, cortexolone, glycocholic acid, L-arginine, lecithin, phosphatidylethanolamine, lactosylceramide, taurochenodeoxycholic acid, and phosphocholine. The levels of 14 metabolites exhibited the marked changes among the normal, model, and PaeH group with high specificity and sensitivity as markers. It indicated that these metabolites might be the biomarkers, which were related to the action mechanism of peaoniflorin. Furthermore, we constructed the signaling network and found that primary bile acid biosynthesis and
glycerophospholipid metabolism might play important roles in cholestasis regulation based on the impact (value >0.1) and hits (value >3). Firstly, glycerophospholipid metabolism focuses on lecithin, phosphocholine, and phosphatidylethanolamine, which are the intermediates of the glycerophospholipid metabolism. Compared with the model group, the value of the three components was increased. It has been reported that the three components are the main lipid components of biomembranes, which indicates that the organism tries to repair the lesions of cell membranes and decrease the cell damage (Wright et al., 2004). Recent research has also illustrated that increase of phosphocholine could decrease cholesterol, which could decrease the liver damage of the organism (Volinsky et al., 2012). Phosphocholine also indirectly affects the ether lipid metabolism. Lecithin participates in the linoleic acid metabolism, which is related to the protection of cholestasis (Minich
Figure 5. Signaling networks. The red solid box represented metabolite higher in paeoniflorin-treated group than in the normal group; the blue solid box represented metabolite lower in paeoniflorin-treated group than in the normal group. AST, aspartate transaminase; ALT, alanine transaminase. This figure is available in colour online at wileyonlinelibrary.com/journal/ptr. Copyright © 2016 John Wiley & Sons, Ltd.
Phytother. Res. 30: 654–662 (2016)
SERUM METABOLOMIC
661
Figure 6. Changes of potential metabolites in cholestasis with paeoniflorin treatment. (A) 15(S)-HPETE; (B) cortexolone; (C) glycochenodeoxycholic acid; (D) glycocholic acid; (E) L-arginine; (F) lecithin; (G) phosphatidylethanolamine; (H) phosphocholine; (J) creatine; (K) inosine; (L) lyso PAF; (M) taurocholic acid; (N) lactosylceramide; and (O) taurochenodeoxycholic acid. **p < 0.01, *p < 0.05 compared ## # with the model group; p < 0.01, p < 0.05 compared with the normal group.
et al., 2000). Then, lecithin indirectly affects the arachidonic acid metabolism, and the increase tendency of the 15(S)-HPETE is the same as lecithin. As we all know, arachidonic acid metabolism is related to the inflammation, and there are many studies indicated that inhibit inflammation could reduce cholestasis (Lu et al., 2014). Lecithin also indirectly affects glycine, serine and threonine metabolism, purine metabolism, and sphingolipid metabolism and eventually indirectly affects the Kreb’s cycle. Kreb’s cycle is of vital importance for the survival of life and plays an important role in gluconeogenesis, transamination, deamination, and lipogenesis (Moffett and Namboodiri, 2003). Acetyl-CoA is the starting point for the Kreb’s cycle and obtained from various sources (Melendez-Hevia et al., 1996). In this study, many metabolites eventually affect the Kreb’s cycle, which is the bridge between the primary bile acid biosynthesis and glycerophospholipid metabolism. Primary bile acid biosynthesis focuses on taurocholic acid, glycocholic acid, taurochenodeoxycholic acid, and glycochenodeoxycholic acid, which are recognized as the potential metabolites in bile acid secretion. The enhancement of four components is commonly presented in cholestatic patients (Woolbright et al., 2015). Recent research has also illustrated that accumulation of several hydrophobic bile acids during cholestasis results in hepatotoxicity via apoptosis and inflammation (Dilger Copyright © 2016 John Wiley & Sons, Ltd.
et al., 2012). In our study, the four acids were significantly increased, and PeaH group remarkably downregulated the levels in the serum, which may partially reveal the probable therapeutic mechanism through bile acid synthesis and secretion. In urea cycle, L(D)-arginine and its downstream substrate 4-guanidinobutanoate were found to decline in the model group, whereas PeaH group significantly enhanced the contents. Previous studies demonstrate that modulation of L-arginine metabolism displays protective effects on cholestatic rats through NO synthasis and attenuating oxidative stress (Tain et al., 2013). In order to differentiate diseased and health states, it is a new powerful technology for the metabolomics study, which helps us have a better understanding of the disease, which lead to the discovery of new targets and the mechanism of the diseases (Tain et al., 2013). Application of metabolomic technologies for the study of cholestasis increases our understanding of the pathophysiological processes, and it should help us to identify potential biomarkers to develop new therapeutic strategies. System analysis of metabolic networks that are a central paradigm in biology will help us to identify new biomarkers, which will generate more in-depth understanding of cholestasis mechanism and thus provide better guidance for drug efficacy. In this study, our data have showed that paeoniflorin works by adjusting multiple metabolic pathways to the normal state. Additionally, Phytother. Res. 30: 654–662 (2016)
662
Z. CHEN ET AL.
the application of metabolite profiling tools can provide an efficient means for early diagnosis of cholestasis and offer a practical method for monitoring therapeutic intervention. Indeed, future metabolomic studies in human populations with cholestasis are still needed to validate the biomarkers found in the animal model.
CONCLUSION Our study highlights the importance of the power of metabolomics to elucidate metabolic characters of cholestasis and therapeutic effects of paeoniflorin at the global metabolomics levels. Analyzing the topology of the network, we detected 14 potential biomarkers and predicted the major metabolites network of cholestasis using the validated pattern recognition methods. These biomarkers that represent molecular targets for cholestasis disease patterns were identified and related to the acutely perturbed pathway of primary bile acid biosynthesis, glycerophospholipid metabolism,
arachidonic acid metabolism, arginine and proline metabolism, ether lipid metabolism, and so on. Of note, the results indicated that paeoniflorin could provide satisfactory effects on cholestasis through regulating the multiple perturbed networks to their normal state. Taken together, the present study proves that metabolomics is a promising approach to unraveling hepatoprotective effects of paeoniflorin by partially regulating the perturbed pathways, which provides deep insights into the mechanisms of cholestasis.
Acknowledgements The current work was supported by Major Projects of the National Science and Technology (no. 2012ZX10005010-002-002) and the National Natural Science Foundation Project of China (nos. 81303120 and 81173571).
Conflict of Interest The authors have declared that there is no conflict of interest.
REFERENCES Dilger K, Hohenester S, Winkler-Budenhofer U, et al. 2012. Effect of ursodeoxycholic acid on bile acid profiles and intestinal detoxification machinery in primary biliary cirrhosis and health. J Hepatol 57: 133–140. Hirschfield GM, Heathcote EJ, Gershwin ME. 2010. Pathogenesis of cholestatic liver disease and therapeutic approaches. Gastroenterology 139: 1481–1496. Hofmann AF. 2002. Cholestatic liver disease: pathophysiology and therapeutic options. Liver 22: 14–19. Jiang FJ, Zhao YL, Wang JB, et al. 2011. Comparative pharmacokinetic study of paeoniflorin and albiflorin after oral administration of Radix Paeoniae rubra in normal rats and the acute cholestasis hepatitis rats. Fitoterapia 83: 415–421. Kakizaki S, Takizawa D, Tojima H. 2011. Nuclear receptors CAR and PXR: therapeutic targets for cholestatic liver disease. Front Biosci 16: 2988–3005. Lu X, Tian Y, Lian X, et al. 2014. Integrated systems toxicology approaches identified the possible involvement of ABC transporters pathway in erythromycin estolate-induced liver injury in rat. Food Chem Toxicol 65: 343–355. Melendez-Hevia E, Waddell TG, Cascante M. 1996. The puzzle of the Krebs citric acid cycle: assembling the pieces of chemically feasible reactions, and opportunism in the design of metabolic pathways during evolution. J Mol Evol 43: 293–303. Minich DM, Havinga R, Stellaard F, Vonk RJ, Kuipers F, Verkade HJ. 2000. Intestinal absorption and postabsorptive metabolism of linoleic acid in rats with short-term bile duct ligation. Am J Physiol Gastrointest Liver Physiol 279: G1242– G1248. Moffett JR, Namboodiri MA. 2003. Tryptophan and the immune response. Immunol Cell Biol 81: 247–265. Padda MS, Sanchez M, Akhtar AJ, Boyer JL. 2011. Drug-induced cholestasis. J Hepatol 53: 1377–1387. Sun H, Zhang A, Wang X. 2012. Potential role of metabolomic approaches for Chinese medicine syndromes and herbal medicine. Phytother Res 26: 1466–1471. Sun H, Zhang AH, Zou DX, Sun WJ, Wu XH, Wang XJ. 2014. Metabolomics coupled with pattern recognition and pathway analysis on potential biomarkers in liver injury and hepatoprotective effects of yinchenhao. Appl Biochem Biotechnol 173: 857–869. Tain YL, Chen CC, Lee CT, et al. 2013. Melatonin regulates L-arginine transport and NADPH oxidase in young rats with bile duct ligation: role of protein kinase C. Pediatr Res 73: 395–401.
Copyright © 2016 John Wiley & Sons, Ltd.
Volinsky R, Paananen R, Kinnunen PK. 2012. Oxidized phosphatidylcholines promote phase separation of cholesterol-sphingomyelin domains. Biophys J 103: 247–254. Wang P, Sun H, Lv H, et al. 2010. Thyroxine and reserpine-induced changes in metabolic profiles of rat urine and the therapeutic effect of Liu Wei Di Huang Wan detected by UPLC-HDMS. J Pharm Biomed Anal 53: 631–645. Wang X, Sun H, Zhang A, Sun W, Wang P, Wang Z. 2011a. Potential role of metabolomics approaches in the area of traditional Chinese medicine: as pillars of the bridge between Chinese and Western medicine. J Pharm Biomed Anal 55: 859–868. Wang X, Wang H, Zhang A, Sun W, Wang P, Wang Z. 2011b. Metabolomics study on the toxicity of aconite root and its processed products using ultraperformance liquid-chromatography/ electrospray-ionization synapt high-definition mass spectrometry coupled with pattern recognition approach and ingenuity pathways analysis. J Proteome Res 11: 1284–1301. Wang X, Wang Q, Zhang A, et al. 2013. Metabolomics study of intervention effects of Wen-Xin-formula using ultra high-performance liquid chromatography/mass spectrometry coupled with pattern recognition approach. J Pharm Biomed 74: 22–30. Watanabe N, Takashimizu S, Kojima S, et al. 2007. Clinical and pathological features of a prolonged type of acute intrahepatic cholestasis. Hepatol Res 37: 598–607. Woolbright BL, Dorko K, Antoine DJ, et al. 2015. Bile acid-induced necrosis in primary human hepatocytes and in patients with obstructive cholestasis. Toxicol Appl Pharmacol 283: 168–177. Wright MM, Howe AG, Zaremberg V. 2004. Cell membranes and apoptosis: role of cardiolipin, phosphatidylcholine, and anticancer lipid analogues. Biochem Cell Biol 82: 18–26. Yang H, Ramani K, Xia M, et al. 2009. Dysregulation of glutathione synthesis during cholestasis in mice: molecular mechanisms and therapeutic implications. J Hepatol 49: 1982–1991. Zhang A, Sun H, Wang Z, Sun W, Wang P, Wang XJ. 2010. Metabolomics: towards understanding traditional Chinese medicine. Planta Med 76: 2026–2035. Zhang A, Sun H, Wang X. 2012. Serum metabolomics as a novel diagnostic approach for disease: a systematic review. Anal Bioanal Chem 404: 1239–1245. Zhao Y, Zhou G, Wang J, et al. 2013. Paeoniflorin protects against ANIT-induced cholestasis by ameliorating oxidative stress in rats. Food Chem Toxicol 58: 242–248.
Phytother. Res. 30: 654–662 (2016)
