Pharmacometabolomics in Endogenous Drugs: A New Approach for

Subscriber access provided by JAMES COOK UNIVERSITY LIBRARY

Article

Pharmacometabolomics in endogenous drugs: a new approach for predicting the individualized pharmacokinetics of cholic acid Zhixin Zhang, Hao Gu, Huizhen Zhao, Yuehong Liu, Shuang Fu, Meiling Wang, Wenjuan Zhou, Ziye Xie, Honghong Yu, Zhenghai Huang, and Xiaoyan Gao J. Proteome Res., Just Accepted Manuscript • DOI: 10.1021/acs.jproteome.7b00218 • Publication Date (Web): 25 Aug 2017 Downloaded from http://pubs.acs.org on August 26, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Journal of Proteome Research is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

Pharmacometabolomics in endogenous drugs: a new approach for predicting the individualized pharmacokinetics of cholic acid

Zhixin Zhang1, Hao Gu2, Huizhen Zhao1, Yuehong Liu1, Shuang Fu1, Meiling Wang1, Wenjuan Zhou1, Ziye Xie1, Honghong Yu1, Zhenghai Huang1, Xiaoyan Gao1, *

1

School of Chinese Materia Medica, Beijing University of Chinese Medicine, South of Wangjing

Middle Ring Road, Chaoyang District, Beijing 100102, P. R. China 2

Institute of Basic Research in Clinical Medicine, China Academy of Chinese Medical Sciences,

No. 16, Nanxiao Road, Dongzhimen, Dongcheng District, Beijing 100007, P. R. China

* Corresponding author: Prof. Dr. Xiaoyan Gao, School of Chinese Materia Medica, Beijing University of Chinese Medicine, South of Wangjing Middle Ring Road, Chaoyang District, Beijing 100102, P. R. China. Tel./Fax: +86-010-84738618. E-mail: [email protected].

1

ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ABSTRACT The evaluation of individual variability in endogenous drugs’ metabolism and disposition is a very challenging task. Here, we developed and validated a metabotype to pharmacokinetics (PK) matching approach by taking cholic acid as an example to predict the individualized PK of endogenous drugs. The stable isotope-labeled cholic acid was selected as the substitute analyte of cholic acid to ensure the accurate measurement of blood concentration. First, large-scale metabolite profiling studies were performed on the predose urine samples of 28 rats. Then, to examine the individualized PK of deuterium 4-cholic acid (d4-cholic acid) in these rats, we determined its plasma concentrations and calculated the differential AUC values. Subsequently, we conducted a two-stage partial least squares analysis in which 31 baseline metabolites were screened initially for predicting the individualized AUC values of d4-cholic acid using the data of predose urine metabolites. Finally, network biology analysis was applied to give the biological interpretation of the individual variances in cholic acid metabolism and disposition, and the result further narrowed the selection of baseline metabolites from 31 to 2 (sarcosine and s-adenosyl-l-homocysteine) for such prediction. Collectively, this pharmacometabolomics research provided a new strategy for predicting individualized PK of endogenous drugs. KEYWORDS: Pharmacometabolomics, metabotype, pharmacokinetics, endogenous drugs, cholic acid.

2


Page 2 of 27

Page 3 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


INTRODUCTION Endogenous drugs are the substances that naturally exist in vivo, including vitamins, hormones, amino acids, bile acids and so on. Assessment of their metabolism and disposition is complicated by the presence of endogenous levels of said drugs that cannot be distinguished from the exogenous compounds following drug administration. Besides, the individual variances in their baseline level also make it very difficult to measure the precise blood concentration of endogenous drugs. Furthermore, the body feedback, circadian rhythm, food and some other factors all have a great influence on the blood concentration of endogenous drugs1. In fact, same with the common drugs, the inter-individual variances in response to endogenous drugs are strongly influenced by the subject’s biochemical state2. The drugs and body interact and influence with each other: drugs act on the body will yield the corresponding pharmacological responses; oppositely, the body act on the drugs will affects their absorption, distribution, metabolism and elimination (ADME) in vivo. Since the individual variances in the baseline levels and biochemical states, huge inter-individual variances exist in the metabolism and disposition of endogenous drugs, and their pharmacokinetics (PK) studies cannot be completely studied using classic methods. Therefore, it is preferable to explore a strategy that can effectively predict the individualized PK of endogenous drugs. This is expected to promote the rational use of endogenous drugs in clinical, and therefore benefits the patients. Currently, most of the studies about investigating individual variances in drug metabolism and disposition are focusing on the assessment of genetic polymorphism and genetic variation3, 4. However, since the genetic3, 5, environmental6 and gut microbiome7 factors all contribute to individual variances, the individualized drug response cannot be fully explained by genomic 3



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

information alone. As a complementary strategy, a pharmacometabolomics approach has been suggested to predict the individual variation in drug response. Pharmacometabolomics, also known as pharmacometabonomics, is a new and rapidly growing field in life science, it was first demonstrated by Clayton et al. that a predose urine metabotype can predict the metabolism and toxicity of paracetomol in rats8. The metabotype is an individual’s metabolic state that the changes of species and quantities of metabolites can be influenced by genetics, environment and gut microbiome9. Therefore, the metabotype analysis of baseline metabolites can serve as a promising way for predicting the individualized drug response. Since the concept of pharmacometabolomics was proposed, it has been used to predict the individualized response of some drugs10-13, but has never been applied in endogenous drugs. Therefore, application of pharmacometabolomics methods to predict the individual variances in endogenous drug metabolism and disposition is of great significance to the rational clinical use of this kind of drugs. Here we present, for the first time, a metabotype to PK matching approach that can be applied to predict individualized PK of cholic acid, which belongs to bile acids. In our previous work, it has been found that, cholic acid, as an endogenous drug, has great individual variances in its baseline level, besides, the level can be influenced by body feedback, circadian rhythm, food and other factors14, which greatly challenges the establishment of an accurate quantitative method for this drug. In this study, in order to measure the accurate blood concentration of cholic acid, deuterium 4-cholic acid (d4-cholic acid), which is a stable isotope labeled compound, was selected as the substitute of cholic acid. A metabotype to PK matching approach was applied for predicting the individualized PK of d4-cholic acid. The overall scheme for the processes of this pharmacometabolomics approach is shown in Figure 1. First, large-scale metabolite profiling 4


Page 4 of 27

Page 5 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


studies were performed on rats’ predose urine samples with the metabolomic tool of ultra performance liquid chromatography coupled with quadrupole time-of-flight mass spectrometry (UPLC Q-TOF/MS) and the individualized PK of d4-cholic acid was evaluated. Then, partial least squares (PLS) analysis was conducted in which associated baseline metabolites were screened initially for predicting the individualized PK of d4-cholic acid using the data of predose urine metabolites. At the same time, network biology analysis was applied to establish the relationship between predose urine metabolites and cholic acid, give the biological interpretation of the individual variances in cholic acid metabolism and disposition. Finally, by using the selected metabolites, we generated a prediction index based on PLS analysis that can be used for predicting the individualized PK of cholic acid. The study provided a new strategy for predicting individual differences in PK of endogenous drugs.

Figure 1. The overall scheme for the processes of this pharmacometabolomics approach.

EXPERIMENTAL SECTION 5



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Materials Isorhamnetin (internal standard, IS) was supplied by National Institute for the Control of Pharmaceutical and Biological Products (Beijing, China), d4-cholic acid (the structure of d4-cholic acid is shown in Figure S3) was purchased from Sigma-Aldrich Co., Ltd (St. Louis, MO, USA). LC/MS-grade methanol and acetonitrile were purchased from Fisher Scientific (Fair Lawn, NJ, USA), HPLC-grade formic acid was supplied by ROE Scientific (Newcastle, Delaware, USA), and ultra high purity water was generated from a MilliQ system Millipore (France).

Animals and Sample Collection All animal experiments were approved by the Animal Ethics Committee of the Beijing University of Chinese Medicine (Beijing, China), and all procedures were performed in accordance with the Helsinki Declaration. The 28 male Sprague–Dawley rats (180 − 220 g) were purchased from Beijing Weitonglihua Laboratory Animal Technology Co., Ltd. (Beijing, China). All the rats were kept under a 12 h light-dark cycle in controlled rooms with a constant temperature (23 ± 2°C) and humidity (60 ± 5%). The rats were free access to water and standard laboratory chow for 1 week and then were kept separately in metabolic cages and allowed to acclimatise for an additional 3 days. After fasted for 12 h, the 24 h period predose urine samples were collected before drug administration. Urine was centrifuged at 14000 rpm for 10 min at 4°C, and then the supernatant was stored at -20°C until analysis. After the predose urine samples were collected, blood samples (0.3 mL) of these 28 rats were collected just before (0 h) oral administration of d4-cholic acid (60 mg/kg) and at 0.033 h, 0.083, 0.17, 0.25, 0.5, 1, 2, 4, 6, 8, 10, 12, 24, 48 and 72 h thereafter. All blood samples for plasma 6


Page 6 of 27

Page 7 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


preparation were subjected into heparinized tubes and centrifuged at 3000 rpm for 10 min at 4°C, and then immediately stored at -80°C until analysis.

Analysis of D4-cholic Acid Concentrations D4-cholic acid was extracted from the plasma samples by liquid-liquid extraction, and a liquid chromatography−hybrid triple quadrupole linear ion trap mass spectrometry (LC−QTRAP/MS) technique was used for quantification of d4-cholic acid. D4-cholic acid and isorhamnetin were detected as their most abundant ions [M-H]- in negative ESI mode by monitoring their parent/daughter transitions at m/z 411.3/347.3 and 315.2/300, respectively. The detailed information of the sample preparation, LC−QTRAP/MS analysis and method validation is shown in Supplementary Information.

Pharmacokinetic Analysis WinNonlin software Ver.6.2 ×& 6.3 (Pharsight) was used to perform the pharmacokinetic analysis of d4-cholic acid and graph the concentration-time curves. Also, the area under curves (AUC0–72, denoted hereafter as “AUC”) values were calculated.

Metabolic Profiling of Metabolites in Baseline Urine Urine samples were thawed at room temperature and centrifuged to remove particulate matter. Each predose urine sample (100 µL) was vortexed for 1 min with 200 µL of acetonitrile for protein precipitation. After being incubated at 4°C for 5 min, the samples were centrifuged at 12000 rpm for 10 min, and then the supernatant was transferred to an injection vial for analysis. The sample 7



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

analysis was performed on a UPLC Q-TOF/MS approach and quality control samples15 were injected at regular intervals during the analytical process to validate the quality of data acquisition (see Supplementary Information for details). Micromass MarkerLynx Application Manager Ver. 4.0 (Waters Corp., Milford, USA) was employed for peak alignment of the UPLC Q-TOF/MS raw data to obtain the data list that containing the retention time, m/z, and peak intensity of each sample.

Data Analysis All computations occurred during PLS multivariate analysis were performed by SIMCA P+ (version 14.0; Umetrics, Uppsala, Sweden) software. Two methods were used to validate the second PLS model: (i) cross-validation using the leave-one-out approach and (ii) internal validation applying 20 permutations tests (see Supporting Information for the details of modeling and validation). The metabolite features screened from the second PLS models for high VIP (variable importance to the projection) values (VIP > 1) were identified. The metabolite peaks were detected by MSE technique and some available databases such as METLIN (http://metlin.scripps.edu/), KEGG (http://www.kegg.com/) and HMDB (http://www.hmdb.ca/) were used to identify these potential biomarkers. The regression analysis between the identified metabolites and the AUC values was accomplished using the SPSS16.0 software.

Network Construction TCM grammar systems16 was used to construct the metabolic network for illustrating the relations between the selected metabolites and bile acids. The metabolite-related proteins were derived from the STITCH database (http://stitch.embl.de/). The parameter of the required confidence score was 8


Page 8 of 27

Page 9 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


set higher than 0.1 to obtain more overall results. Bile acids were collected from HMDB database. The network was visualized using the software Cytoscape 3.2.1.

RESULTS AND DISCUSSION Metabolic Profiling of Metabolites in Baseline Urine Since there is rich metabolome information in urine samples, it is necessary for applying large-scale metabolic profiling to screen the metabolites that can be used for predicting the individualized PK of drugs. LC-MS analysis can provide a wide coverage of metabolome required to achieve this goal with its high sensitivity17, 18. In the study, predose urine samples of 28 rats collected over 24 h before they were dosed with d4-cholic acid were analyzed using UPLC Q-TOF/MS to generate the global urine metabolic profiles (see Supplementary Information for the details). As a result, 3,510 and 1,923 metabolic features were detected from the 28 datasets in positive mode and negative mode, respectively. Figure 2 displays the typical base peak intensity (BPI) chromatograms detected from a predose urine sample by UPLC Q-TOF/MS analysis. As shown in the figure, many kinds of urinary metabolites were detected, including organic acids, amino acids, nucleosides, lipids and other metabolites. In each sample, the relative intensity of monitored features was represented by their measured peak area.

9



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 2. Typical base peak intensity (BPI) chromatograms of rat predose urine in positive (a) and negative (b) mode with some identified metabolites. The locally magnified inset shows the MS spectra at 0.6-1.4 min of the chromatograms.

Pharmacokinetic Analysis To evaluate the pharmacokinetic response, d4-cholic acid was selected as the substitute of cholic acid, and after oral administration (60 mg/kg) of this drug, we measured its dynamic plasma concentration at various time points. As shown in Figure 3, the plasma concentration profiles of d4-cholic acid in the 28 rats revealed huge variation in their individualized PK responses. According to these profiles, the AUC values of d4-cholic acid were calculated by WinNonlin software (Pharsight, USA), and their maximum and minimum values have a difference for about 10-fold. As shown in the figure, multiple peaks plasma concentration-time profiles were observed following 10


Page 10 of 27

Page 11 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


drug administration. This phenomenon might be explained by the mutual transformation of bile acids, enterohepatic recirculation19-21 and other complex reasons. Besides, as AUC values represent the total exposure of drug and can provide the most suitable evaluation of the variation in individualized PK of d4-cholic acid. Therefore, during the PLS analysis, AUC values were used as a response variable. Using the stable isotope labeled drugs as an alternative analyte is a safe and accurate research method to study the in vivo behavior of the said drugs. This method has many advantages: (i) Providing accurate measurement results, it can provide the true concentration of the said drug in the body; (ii) It can dynamically describe the trends of drugs in the body; (iii) Safe, non-radiological hazards, and therefore has wide range of applications. In this study, stable isotope tracer technique was performed on cholic acid for the first time to its pharmacokinetic studies, and predicted its individual PK in rats based on a pharmacometabolomics approach.

Figure 3. Plasma concentration–time plots of d4-cholic acid after oral administration in rate (n = 28) (the mean plot was drew by red line with blue points). Inset: table of pharmacokinetic parameter AUC, the pharmacokinetic parameter of d4-cholic acid is highly variable among individuals.

Prediction of the Individualized PK of D4-cholic acid 11



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

A two-stage PLS analysis was employed here to select the baseline urinary metabolites that can be used for predicting the individualized PK of d4-cholic acid. The first stage PLS analysis was to reduce the dimensionality of X data; the second stage was to select the associated metabolites that contributing to PK prediction. PLS analysis can construct a supervised model (PLS model) which can be used to find the relationships between two groups of variables and screen the X variables (baseline urinary metabolites) that can predict the response variable (Y variable, AUC). The first stage PLS analysis was carried out using all the 3,510 and 1,923 baseline urine peak intensities (prediction variables, X block) related to the AUC (response variable, Y block) in positive and negative mode, respectively. According to this analysis, the baseline urinary metabolites that have a huge contribution for predicting the AUC values of d4-cholic acid were preliminary selected. Using this metabolites, a second PLS analysis was then performed for predicting the individualized AUC values. In Figure 4a and b, the score plots of first stage PLS analysis show a good correlation between baseline urinary metabolites and AUC values with an R2 value of 0.86 and 0.81 in positive and negative mode, respectively. In Figure 4c and d, the corresponding loading plots show the metabolites that were either positively (dots on the upper right corner) or negatively (dots on the lower left corner) correlated to AUC in positive and negative mode. As a result, according to the VIP values (VIP > 1.5), 116 and 74 metabolites (enclosed by red boxes) that have strong correlations with AUC values were screened in positive and negative mode, respectively. The second PLS model was then built based on these selected variables to predict the individualized AUC values. Here, two methods were employed to validate the reliability of this PLS model. Figure 5a and b show the plots that obtained by “cross-validation” using leave-one-out method, the 12


Page 12 of 27

Page 13 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


predicted AUC values exhibited excellent correlation with actual measured AUC with the regression line of 0.9513 and 0.8084 in positive and negative mode, respectively. Simultaneously, “internal validation” was performed to verify the prediction ability of the second PLS model without overfitting risk. As can be seen in Figure 5c and d, the goodness-of-fit (R2) and predictability (Q2) values (left) obtained from 20 random permutation experiments were all smaller than those (right) of the second PLS model, and the regression line of Q2 was intersect the Y-axis below zero, which means the model was not overfitting. Therefore, the second PLS model was proved to be reliably for our research. Finally, 17 and 14 key baseline urinary metabolites were selected from the second PLS model with VIP > 1 in positive and negative mode, respectively, which can be used to characterize the individualized AUC values of d4-cholic acid. Table 1 summarizes the details of the selected 31 metabolites, and 28 of them were identified (see Supporting Information for the details of identification methods). The correlation coefficients between those identified metabolites and the AUC are shown in Figure S4 of Supporting Information, and the variable selection method was validated by the strong correlations between these metabolites and AUC values in this 28 samples.

13



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 4. PLS score (a, b) and loading (c, d) plots of the first stage PLS model. (a, b) The score plot of the first stage PLS model, which demonstrated good correlation between baseline urinary metabolites and AUC values with an R2 value of 0.86 and 0.81 in positive (a) and negative (b) mode. (c, d) The loading plots of the first stage PLS model, the selected metabolites with high VIP values (VIP > 1.5) are marked in red squares in (c) positive mode and (d) negative mode.

14


Page 14 of 27

Page 15 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 5. The second PLS model to predict the AUC of d4-cholic acid. (a, b) Plots of predicted AUC vs. actual measured AUC from second PLS model by ‘cross-validation’ method. Color from blue to brown indicates increasing AUC values in positive mode (a) and negative (b) mode. (c, d) Internal validation of the second PLS model using 20 permutation tests in positive mode (c) and negative mode (d).

Network Biology Analysis Network biology analysis providing an integrated view of biological processes at the molecular level, are useful tools for understanding how molecules and the interactions between the molecules determine cell functions22. The results of linear correlation regression analysis gave a description that the metabolites that selected from the two stages PLS analysis correlated well with AUC. In this paper, in order to illustrate the relationship between the selected metabolites and cholic acid in molecular level, network biology analysis was used to construct a metabolic network for describing 15



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 27

the selected metabolites interactions with cholic acid. Figure 6 shows the baseline metabolites sarcosine (VIP = 1.71) and s-adenosyl-l-homocysteine (VIP =

1.1)

could

regulate

the

synthesis

of

four

molecules

in

vivo,

including

3alpha-Hydroxy-5beta-cholanate, 3alpha,12alpha-dihydroxy-5beta-cholanate, taurolithocholate and chenodeoxycholate, these four molecules all belongs to bile acids. Figure S5 (shown in Supporting Information) shows the transforming relationship between the four bile acids and cholic acid. It was reported that at least several dozen bile acids have been identified in humans, rodents and other animals23, 24. Bile acids are the derivative of cholanic acid, the difference in their structure is small, and they can mutual transform with each other through a series of pathways in vivo. Therefore, in bile acid system, the disturbance of some bile acids is bound to affect the in vivo concentration of other bile acids. In summary, the variation in the baseline level of sarcosine and s-adenosyl-l-homocysteine is bound to significantly affect the pharmacokinetic response of cholic acid, and the two metabolites can be used for predicting the individualized PK of cholic acid before drug administration.

16


Page 17 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 6. The interaction network of the selected metabolites and bile acids. The selected metabolites are shown as red diamond. The related bile acids are green rhombus. Orange triangles denote metabolic reactions, and pink squares denote the intermediate metabolites. Edges with arrows represent the directions of reactions.

AUC Prediction Based on Levels of Sarcosine and S-adenosyl-l-homocysteine in Baseline Urine After the network biology analysis, we further narrowed the selection of baseline metabolites from 31 to 2 (sarcosine and s-adenosyl-l-homocysteine) for predicting the individualized PK of cholic acid. As described above, the 2 baseline metabolites were both correlated well with AUC values (r = 0.633 and 0.611) in the 28 samples and also have relationships with cholic acid in molecular level. A PLS analysis was applied to the 2 metabolites and an equation which able to be used for predicting the AUC of cholic acid was developed: AUC = 3.66 Sarcosine + 2.73 S-Adenosyl-L-homocysteine. Figure 7 shows the scatterplot of the predicted vs. measured AUC values for the 28 samples using the equation. The prediction power (R2 = 0.75) was 84% of that of the first stage PLS model, which 3310 metabolites were used, while that of the second PLS model was 88%, which 116 metabolites were used. Internal validation was performed to check the model ability of prediction without risk of overfitting (shown in Figure S6 of Supporting Information). Therefore, meaning that the prediction capability of the equation was really well.

17



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 7. Scatterplot of the predicted vs. measured AUC values for the 28 samples using the equation AUC = 3.66 Sarcosine + 2.37 S-Adenosyl-L-homocysteine.

CONCLUSIONS In this study, we have shown that predose metabolic profile of urine that can predict the individualized responses of endogenous drugs in rats based on a pharmacometabolomics approach. The approach was used for predicting individualized PK of cholic acid. First, UPLC Q-TOF/MS analyses was used to generate metabolic profiles of predose urine samples of 28 rats, resulting in 3,510 and 1,923 detected metabolic features in positive and negative mode, respectively. Then, PLS model was built using these baseline urinary metabolites which could predict the AUC values of cholic acid. Subsequently, based on the PLS model, 31 associated metabolic features were selected, and 28 of them were identified by MSE fragment information along with searching database. Finally, a metabolic network was reconstructed using these 28 identified metabolites to describe the metabolites interactions with cholic acid. The network revealed the baseline molecules sarcosine and s-adenosyl-l-homocysteine were correlated to the pharmacokinetic response after cholic acid treatment, and the two metabolites can be used for predicting the individualized PK of cholic acid before drug administration. In conclusion, a novel approach was proposed here to predict the 18


Page 18 of 27

Page 19 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


individualized PK of endogenous drugs, and by applying this approach to cholic acid, we identified two baseline urine metabolites which can predict the individualized PK of cholic acid. This work will also contribute to developing novel strategies in the prediction of the individual responses of other endogenous drugs.

ASSOCIATED CONTENT Supporting Information Table S1. Regression data and LLOQ of d4-cholic acid determined. Table S2. Accuracy and precision of d4-cholic acid in rat plasma. Table S3. Matrix effect and extraction recovery of d4-cholic and IS in rat plasma (n=6). Table S4. Stability of d4-cholic in rat plasma (n=6). Figure S1. Multiple reaction monitoring chromatograms for d4-cholic acid and IS in rat plasma samples: (A) a blank plasma sample; (B) a blank plasma sample spiked with the d4-cholic acid and IS in LLOQ; (C) a plasma sample from a rat at 10 min after a single oral administration of d4-cholic acid. Figure S2. Product ion spectra of the two potential biomarkers. (a) trigonellinamide; (b) pantothenic acid. Figure S3. The structure of the stable isotope labeling reagent, d4-cholic acid. Figure S4. Correlations between AUC0-72 and baseline urine metabolites. Figure S5. Transforming relationships between the related bile acids and cholic acid. Figure S6. ‘Internal Validation’ plot by 20 permutation tests for final PLS model built by using only 2 selected metabolites. 19



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

CONFLICT OF INTEREST DISCLOSURE The authors declare no competing financial interest.

ACKNOWLEDGMENTS This work was financially supported by National Natural Science Foundation of China (81173649/H2817).

20


Page 20 of 27

Page 21 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


REFERENCES (1) Zhang, X.; Xie, X. Q.; Liu, T. L.; Song, D. M. Quantitative determination and bioequivalence assessment of the drugs as endogenous compounds. Chin. J. New Drugs 2011, 20, 2222-2228. (in Chinese) (2) Wilson, I. D. Drugs, bugs, and personalized medicine: pharmacometabonomics enters the ring. Proc. Natl. Acad. Sci. USA 2009, 106, 14187–14188. (3) Spear, B. B.; Heath-Chiozzi, M.; Huff, J. Clinical application of pharmacogenetics. Trends mol. med. 2001, 7, 201-204. (4) Pagliarulo, V.; Datar, R. H.; Cote, R. J. Role of genetic and expression profiling in pharmacogenomics: the changing face of patient management. Curr. issues mol. Boil. 2002, 4, 101-110. (5) Zhou, S. F.; Di, Y. M.; Chan, E.; Du, Y. M.; Chow, V. D.; Xue, C. C.; Lai, X.; Wang, J. C.; Li, C. G.; Tian, M.; Duan, W. Clinical pharmacogenetics and potential application in personalized medicine. Curr. Drug Metab. 2008, 9, 738–784. (6) Bronson, S. L.; Ahlbrand, R.; Horn, P. S.; Kern, J. R.; Richtand, N. M. Individual differences in maternal response to immune challenge predict offspring behavior: contribution of environmental factors. Behav. Brain Res. 2011, 220, 55–64. (7) Alam, A. N.; Saha, J. R.; Dobkin, J. F.; Lindenbaum, J. Interethnic variation in the metabolic inactivation of digoxin by the gut flora. Gastroenterology 1988, 95, 117–123. (8) Clayton, T. A.; Lindon, J. C.; Cloarec, O.; Antti, H.; Charuel, C.; Hanton, G.; Provost, J. P.; Le Net, J. L.; Baker, D.; Walley, R. J.; Everett, J. R.; Nicholson, J. K. Pharmaco-metabonomic phenotyping and personalized drug treatment. Nature 2006, 440, 1073–1077. 21



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(9) Fiehn, O. Metabolomics--the link between genotypes and phenotypes. Plant Mol. Bio. 2002, 48, 155–171. (10) Kantae, V.; Krekels, E. H.; Esdonk, M. J.; Lindenburg, P.; Harms, A. C.; Knibbe, C. A.; Van der Graaf, P. H.; Hankemeier, T. Integration of pharmacometabolomics with pharmacokinetics and pharmacodynamics: towards personalized drug therapy. Metabolomics 2017, 13, 9. (11) Burt, T.; Nandal, S. Pharmacometabolomics in Early-Phase Clinical Development. Clin. Transl. Sci. 2016, 9, 128–138. (12) Everett, J. R. Pharmacometabonomics in humans: a new tool for personalized medicine. Pharmacogenomics 2015, 16, 737–754. (13) Everett, J. R.; Loo, R. L.; Pullen, F. S. Pharmacometabonomics and personalized medicine. Ann. Clin. Biochem. 2013, 50, 523–545. (14) Peng, L.; Liu, H. Y.; Gao, X. Y. An exploration for pharmacokinetic studies of cholic acid and hyodeoxycholic acid. Eur. J. Integr. Med. 2014, 6, 739–740. (15) Dunn, W. B.; Wilson, I. D.; Nicholls, A. W.; Broadhurst, D. The importance of experimental design and QC samples in large-scale and MS-driven untargeted metabolomic studies of humans. Bioanalysis 2012, 4, 2249–2264. (16) Yan, J.; Wang, Y.; Luo, S. J.; Qiao, Y. J. TCM grammar systems: an approach to aid the interpretation of the molecular interactions in Chinese herbal medicine. J Ethnopharmacol. 2011, 137, 77–84. (17) Sreekumar, A.; Poisson, L. M.; Rajendiran, T. M.; Khan, A. P.; Cao, Q.; Yu, J.; Laxman, B.; Mehra, R.; Lonigro, R. J.; Li, Y.; Nyati, M. K.; Ahsan, A.; Kalyana-Sundaram, S.; Han, B.; Cao, X.; Byun, J.; Omenn, G. S.; Ghosh, D.; Pennathur, S.; Alexander, D. C.; Berger, A.; Shuster, J. R.; Wei, J. 22


Page 22 of 27

Page 23 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


T.; Varambally, S.; Beecher, C.; Chinnaiyan, A. M. Metabolomic profiles delineate potential role for sarcosine in prostate cancer progression. Nature 2009, 457, 910–914. (18) Want, E. J.; Nordström, A.; Morita, H.; Siuzdak, G. From exogenous to endogenous: the inevitable imprint of mass spectrometry in metabolomics. J. Proteome Res. 2007, 6, 459–468. (19) Ezzet, F.; Krishna, G.; Wexler, D. B.; Statkevich, P.; Kosoglou, T.; Batra, V. K. A population pharmacokinetic model that describes multiple peaks due to enterohepatic recirculation of ezetimibe. Clin. Ther. 2001, 23, 871–885. (20) Funaki, T. Enterohepatic circulation model for population pharmacokinetic analysis. J. Pharm. Pharmacol. 1999, 51, 1143–1148. (21) Plusquellec, Y.; Efthymiopoulos, C.; Duthil, P.; Houin, G. A pharmacokinetic model for multiple sites discontinuous gastrointestinal absorption. Med. Eng. Phys. 1999, 21, 525–532. (22) Barabasi, A. L.; Oltvai, Z. N. Network biology: understanding the cell's functional organization. Nat. Rev. Genet. 2004, 5, 101–113. (23) Hofmann, A. F.; Hagey, L. R. Bile acids: chemistry, pathochemistry, biology, pathobiology, and therapeutics. Cell. Mol. Life Sci. 2008, 65, 2461−2483. (24) Hagey, L. R.; Møller, P. R.; Hofmann, A. F.; Krasowski, M. D. Diversity of bile salts in fish and amphibians: evolution of a complex biochemical pathway. Physiol. Biochem. Zool. 2010, 83, 308−321.

23



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 24 of 27

Table 1. The selected metabolites and their VIP values in the second PLS modes. No.

Compounds

VIP

tR (min)

m/z

Ion mode

HMDB ID

1

Creatine

3.66

0.8885

132.0776

ESI+

HMDB00064

2

3-Methoxytyrosine

3.17

1.1298

212.1038

ESI+

HMDB01434

+

HMDB00562

3

Creatinine

3.08

0.8626

114.067

ESI

4

Prolylhydroxyproline

2.08

1.1339

229.1192

ESI+

HMDB06695

203.1109

ESI

+

HMDB29026

+

HMDB00630

5

Prolyl-Serine

1.9

6.7484

6

Cytosine

1.8

1.1554

112.0512

ESI

7

Sarcosine

1.71

0.8877

90.0556

ESI+

HMDB00271

+

HMDB00699

+

HMDB00870

8

1-Methylnicotinamide

1.68

0.8238

137.0718

ESI

9

Histamine

1.64

2.169

112.0512

ESI

10

Ectoine

1.55

1.1423

143.1188

ESI+

220.1186

+

HMDB00210

+

11

Pantothenic acid

1.5

4.0589

ESI

—

12

Leucyl-proline

1.46

1.1773

229.1553

ESI

HMDB28937

13

Phosphodimethylethanolamine

1.36

6.7999

170.0608

ESI+

HMDB60244

455.1885

ESI

+

—

+

HMDB02721

14

Unknow

1.32

1.1484

15

1-Methylinosine

1.31

5.9334

283.1158

ESI

16

Trigonelline

1.18

0.9078

138.0558

ESI+

HMDB00875

385.1179

+

HMDB00939

17

S-Adenosyl-L-homocysteine

1.1

4.2605

ESI

-

18

p-Cresol sulfate

3.71

6.0411

187.0067

ESI

HMDB11635

19

Indoxyl Sulfate

2.96

5.2642

212.002

ESI-

HMDB00682

20

p-Cresol glucuronide

2.76

6.1266

283.0823

ESI-

HMDB11686

-

21

Phenol Sulphate

1.92

4.7791

172.9911

ESI

HMDB60015

22

cis-4-Decenedioic acid

1.91

7.2267

199.0972

ESI-

HMDB00603

165.0552

-

HMDB41683

-

23

4-Hydroxyphenyl-2-propionic acid

1.74

6.4187

ESI

24

Pseudouridine

1.73

5.5235

242.9965

ESI

HMDB00767

25

4-Sulfobenzoate

1.59

6.9763

201.0222

ESI-

—

158.0818

-

HMDB00339

-

26

2-Methylbutyrylglycine

1.58

4.9908

ESI

27

Vanillin 4-Sulfae

1.56

5.3719

245.0123

ESI

HMDB41789

28

Unknow

1.54

7.1165

343.0854

ESI-

—

199.1333

-

HMDB37179

-

29

3-Nonanon-1-yl acetate

1.46

7.5383

ESI

30

5-Phosphoribosylamine

1.45

3.801

227.9967

ESI

HMDB01128

31

Unknow

1.38

7.3399

357.101

ESI-

—

24


Page 25 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure legends Figure 1. The overall scheme for the processes of this pharmacometabolomics approach. Figure 2. Typical base peak intensity (BPI) chromatograms of rat predose urine in positive (a) and negative (b) mode with some identified metabolites. The locally magnified inset shows the MS spectra at 0.6-1.4 min of the chromatograms. Figure 3. Plasma concentration–time plots of d4-cholic acid after oral administration in rate (n = 28) (the mean plot was drew by red line with blue points). Inset: table of pharmacokinetic parameter AUC, the pharmacokinetic parameter of d4-cholic acid is highly variable among individuals. Figure 4. PLS score (a, b) and loading (c, d) plots of the first stage PLS model. (a, b) The score plot of the first stage PLS model, which demonstrated good correlation between baseline urinary metabolites and AUC values with an R2 value of 0.86 and 0.81 in positive (a) and negative (b) mode. (c, d) The loading plots of the first stage PLS model, the selected metabolites with high VIP values (VIP > 1.5) are marked in red squares in (c) positive mode and (d) negative mode. Figure 5. The second PLS model to predict the AUC of d4-cholic acid. (a, b) Plots of predicted AUC vs. actual measured AUC from second PLS model by ‘cross-validation’ method. Color from blue to brown indicates increasing AUC values in positive mode (a) and negative (b) mode. (c, d) Internal validation of the second PLS model using 20 permutation tests in positive mode (c) and negative mode (d). Figure 6. The interaction network of the selected metabolites and bile acids. The selected metabolites are shown as red diamond. The related bile acids are green rhombus. Orange triangles denote metabolic reactions, and pink squares denote the intermediate metabolites. Edges with arrows represent the directions of reactions. 25



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 7. Scatterplot of the predicted vs. measured AUC values for the 28 samples using the equation AUC = 3.66 Sarcosine + 2.37 S-Adenosyl-L-homocysteine.

26


Page 26 of 27

Page 27 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


For TOC only

27


Pharmacometabolomics in Endogenous Drugs: A New Approach for

Recommend Documents