Branched-Chain Amino Acids as Predictors for Individual Differences

Subscriber access provided by UB + Fachbibliothek Chemie | (FU-Bibliothekssystem)

Article

Branched-Chain Amino Acids as Predictors for Individual Differences of Cisplatin Nephrotoxicity in Rats: A Pharmacometabonomics Study Pei Zhang, Wei Li, Jiaqing Chen, Ruiting Li, Zunjian Zhang, Yin Huang, and Fengguo Xu J. Proteome Res., Just Accepted Manuscript • DOI: 10.1021/acs.jproteome.7b00014 • Publication Date (Web): 08 Mar 2017 Downloaded from http://pubs.acs.org on March 9, 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 36

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

Branched-Chain Amino Acids as Predictors for Individual Differences of Cisplatin Nephrotoxicity in Rats: A Pharmacometabonomics Study Pei Zhang1,2,3, Wei Li1,2,3, Jiaqing Chen1,2,3, Ruiting Li1,2,3, Zunjian Zhang1,2,3, Yin Huang1,2,3**, and Fengguo Xu1,2,3* 1

Key Laboratory of Drug Quality Control and Pharmacovigilance (Ministry of Education),

China Pharmaceutical University, Nanjing 210009, P. R. China; 2

Jiangsu Key Laboratory of Drug Screening, China Pharmaceutical University, Nanjing 210009,

P. R. China; 3

State Key Laboratory of Natural Medicine, China Pharmaceutical University, Nanjing 210009,

P. R. China.

ACS Paragon Plus Environment

1


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 2 of 36

ABSTRACT Nephrotoxicity is the dose-limiting adverse effect of cisplatin with large individual differences. Up to now, little has been done on how to recognize and predict the individual differences either in preclinical or clinical research. In the present study, important post-dose indicators were screened out first and integrated into a grouping factor, according to which rats were recognized as low or high sensitive individuals. Then, mass spectrometry-based untargeted metabolomics approach was performed to dissect the metabolic differences in pre-dose serum of the two groups. Eventually, branched-chain amino acids (BCAAs) were found to be most significant with the lowest p value of Mann-Whitney U test and the highest area under receiving operating characteristic curve (AUC-ROC). The findings were further confirmed by absolute quantitation of BCAAs using liquid chromatography-tandem mass spectrometry (LC-MS/MS). Binary logistic regression (BLR) showed that in the discovery set absolute BCAA contents in rat predose serum could predict cisplatin nephrotoxicity with the accuracy of 85%. This result was validated by another two independent external validation sets with the accuracy of 81.8% and 78.8%, respectively. This study could provide new insight into cisplatin nephrotoxicity and may help expedite personalized medicine of cisplatin or other antitumor drugs in future clinical studies.

KEYWORDS: cisplatin; nephrotoxicity; individual differences; prediction; BCAA; metabolomics


2

Page 3 of 36

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


1. INTRODUCTION Nephrotoxicity is the chief side-effect of cisplatin [cis-diamminedichloroplatinum(II)] with the characteristic of large individual differences.1-4 It is consistently a major limitation to the use and efficacy of cisplatin in cancer therapy and is sometimes life-threatening, especially to the elderly with weakened renal function. Statistically, about 25-35% patients experienced varying degrees of acute kidney injury after receiving a single dose of cisplatin.5,

6

Hence, it is of

significant importance to develop methods for recognizing and predicting the individual differences of cisplatin nephrotoxicity. In the individual differences recognition, blood urea nitrogen (BUN) and serum creatinine (SCr) are commonly used to evaluate renal function in clinical settings. However, both of them have the problems of low sensitivity and specificity.7, 8 Pathological examination remains the gold standard in the assessment of kidney toxicity in preclinical studies, whereas it is hard to use in clinical practice generally due to its invasiveness, time-consumption and low sensitivity. Recent mechanism studies revealed strong associations between cisplatin nephrotoxicity and oxidative stress and inflammation.9-11 But up to now, there is no single parameter or method can accurately and specifically recognize the individual differences of cisplatin nephrotoxicity. In case of individual differences prediction, many post-dose indexes such as kidney weight, urinary vanin-1, urinary neutrophil gelatinase-associated lipocalin and kidney injury molecular-1 have been tested for their ability to forecast early cisplatin nephrotoxicity.12-15 But they cannot be used to predict the risk of nephrotoxicity before cisplatin administration. Recently, a nuclear magnetic resonance-based pharmacometabonomics study was conducted to predict cisplatin nephrotoxicity in rats using pre-dose urine.16 The prediction accuracy was 66%, and four urinary metabolites (allantoin, creatinine, succinate and oxoglutarate) were discriminated as important


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

Page 4 of 36

biomarkers. Although the prediction ability was not further validated by additional experiments, this study pioneered the prediction of cisplatin nephrotoxicity using pharmacometabonomics approach. Urine was often studied in metabolomics to discover non-invasive biomarkers. However, the results are susceptible to many factors such as food and water consumption, temperature, urine preservation, sample collecting strategy, and individual healthy condition in urine metabolomics.

17-21

In order to minimize these effects, normalization using protein/creatinine

ratio was routinely employed in data preprocessing procedure. However, consistent protein/creatinine ratio among individuals is the prerequisite for utilizing this method. Thus, it is inapplicable to nephrotoxicity, in which urinary protein and creatinine will increase disproportionately.22 As it is important to limit the variances introduced by external factors in prediction studies, serum with higher stability might be a better choice. Hence, in the present study, we dedicated to integrate a proper parameter to recognize the individual differences of cisplatin nephrotoxicity first. Then, untargeted metabolomics was performed to discover predictive metabolites in pre-dose serum. After confirmation by targeted determination, those metabolites were used to construct the prediction model, which was further validated by another two experiments. 2. MATERIALS AND METHODS 2.1 CHEMICALS AND REAGENTS Cisplatin injections were purchased from Haosen Pharmaceutical (Lianyungang, China). OMethoxyamine hydrochloride, N-methyl-N-trifluoroacetamide (MSTFA) and pyridine, sodium diethyldithiocarbamate trihydrate (DDTC), cis-diammineplatinum (II) (cisplatin), BCAAs


4

Page 5 of 36

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


including L-leucine, L-isoleucine, L-valine and L-13C1-leucine, branched-chain α-keto acids (BCKA) including α-ketoisocaproic acid (KICA), α-ketomethylvaleric acid (KMVA) and αketoisovaleric acid (KIVA), and salicylic acid were purchased from Sigma-Aldrich (St Louis, MO, USA). Trans-diamminedichloropalladium (II) (TDPD) was purchased from J & K chemical (Beijing, China). Liquid chromatography-mass spectrometry (LC-MS) grade reagents including methanol, acetonitrile and ethyl acetate were obtained from Merck (Germany). Formic acid was obtained from Nanjing Chemical Reagent (Nanjing, China). Deionized water was produced using a Milli-Q system (Millipore, Bedford, MA, USA). 2.2 ANIMAL EXPERIMENT AND SAMPLE COLLECTION All animal experiment protocols were made according to the guide for the care and use of laboratory animals (8th edition) released by the National Research Council of the National Academies and approved by the Animal Ethics Committee of China Pharmaceutical University (License Number: SYXK 2012-0035). In the present study, totally three animal experiments were conducted. The objective of the first experiment was to discover predictive metabolites and build a model to forecast cisplatin nephrotoxicity. Therefore, dataset derived from this experiment was named as discovery set. The rest two experiments were carried out to validate the predictive model and named as validation sets. In the first experiment, totally 60 male Sprague-Dawley rats, 6-7 weeks old were allowed to acclimatize for a week (Nanjing, China). The animals were fed a standard commercial diet while housed in a light- and temperature-controlled room (12/12h light/dark, 22-26 °C, 45-55% humidity). After the acclimatization, pre-dose blood were collected before cisplatin administration at day 0. At day 1, rats were intravenously administered with cisplatin (n=45, 8 mg/kg). This dosage was made according to clinical dose, results of our pre-experiment and


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

Page 6 of 36

existing literature. Rats in control group (n=15, group C) were intravenously administered with equivalent volumes of normal saline. At day 6, rats were sacrificed after blood collection, and the left kidneys were removed, weighted and kept at -80 °C until analysis. The right kidneys were removed, weighted and fixed in 10% neutral-buffered formalin for hematoxylin and eosin staining. Body weight of the rats over the course of study was measured daily. In the other two experiments, animal raising condition, dosing time and sample collection strategy were unchanged. 11 and 18 rats were intravenously administered with 8 and 5 mg/kg cisplatin in the second and third experiment, respectively. The overall experimental strategy is shown in Figure 1. 2.3 POST-DOSE PARAMETERS DETERMINATION Routine renal function indicators including BUN, SCr, uric acid and cystatin C were determined in post-dose serum. Routine blood consisting of 24 indexes were tested in post-dose whole blood. This work was performed at Zhongda Hospital Southeast University. After fixed overnight in 10% neutral-buffered formalin, kidneys were dehydrated in alcohol and then embedded in paraffin. Paraffin sections were prepared and stained using standard hematoxylin and eosin staining methods. This part was completed by a pathologist of Nanjing Medical University. Oxidative stress indicators in kidney including malondialdehyde (MDA) and super oxide dismutase (SOD) were determined by using commercial detection kits (Jianchen, Nanjing). Inflammation factors in kidney including TNF-α and IL-10 were determined by means of enzyme-linked immunosorbent assay using commercial detection kits (Raybiotech, USA). Protein content in kidney tissue was determined using bicinchoninic acid protein assay kit (Jianchen, Nanjing). In addition, body weight and kidney coefficient were record or calculated. 2.4 INDIVIDUAL TOXIC RESPONSE GROUPING


6

Page 7 of 36

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


Unsupervised pattern recognition method principal component analysis (PCA) was firstly constructed based on all the determined post-dose indicators. According to the natural distribution of samples in PCA score plot, supervised pattern recognition orthogonal partial least squares-discriminant analysis (OPLS-DA) was then employed. To check whether the model was overfitting, permutation test (200 times) was performed on the corresponding original partial least squares-discriminant analysis (PLS-DA) model.23 S-plot was used for the extraction of statistically signiﬁcant indicators.24 Besides, important indicators screening was supported by regression coefficient plot.25 These were performed in SIMCA-P software (Umetrics, Sweden). After screening, important indicators with different order of magnitude were firstly scaled by range-scaling method for achieving value comparability. Then, the scaled values were summed up within each sample and named as grouping factor. Samples were then sorted according to grouping factor by the order of the lowest to the highest. The foremost 30% samples with the lowest grouping factor values were defined as low sensitive individuals (LS group), and the aftermost 30% were categorized as high sensitive individuals (HS group). As there is no sharp boundary of grouping factor for LS and HS groups, this method taking both ends was applied referring to existing pharmacometabonomics literatures.26-28 2.5 CISPLATIN DETERMINATION IN KIDNEY Cisplatin determination method was modified referring to previous study.29 TDPD and DDTC were chosen as internal standard and derivatization reagent, respectively. Approximately 50 mg frozen kidney were firstly placed into pre-cooled 2 mL homogenization tubes containing ceramic beads. Then, precooled normal saline was added (4 µL/mg tissue), and the samples were homogenized for three times (5.5 m/s for 30 s) with 60 s intervals between homogenization steps. After two centrifugations (14000 rpm, 5 min, 4 °C), the supernatant was removed and


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

Page 8 of 36

named as kidney homogenate. 80 µL kidney homogenate were added with 200 µL normal saline and 300 µL methanol. The mixture was vortex-mixed and centrifuged twice. 250 µL supernatant were removed and added with 20 µL TDPD (1 µg/mL) and 5 µL DDTC (dissolved in 0.2 M NaOH, 5%). The mixture was vortex-mixed and then placed in a water bath (40 °C) for 4 h. After that, 1 mL hexane-ethyl acetate (V/V, 1/1) was added. The mixture were vortex-mixed for 5 min and then 800 µL of the supernatant were removed and evaporated. The residues were dissolved in 100 µL mobile phase and centrifuged twice. Finally, the supernatant was removed for instrumental analysis. In the final calculation, cisplatin content was normalized to the kidney weight and expressed as ng/mg kidney weight. A calibration curve covering the entire range of expected cisplatin concentrations was generated. Other method validation items including matrix effect and stability were also conducted. LC-MS/MS (Shimadzu, Japan) was utilized to analyze cisplatin concentration in kidney. Chromatographic separation was achieved on a Phenomenex Kinetex C18 column (100×2.1 mm, 2.6 µm) (Phenomenex, USA). The isocratic elution involved a mobile phase consisting of 20% water (0.4% formic acid) and 80% methanol. The flow rate was set at 0.3 mL/min, and the column oven was maintained at 40 °C. Multiple reaction monitoring mode was applied and the ion pairs were 640.10→492.05 for TDPD and 401.95→116.25 for cisplatin. Dwell time was set as 300 msec. Nebulizing gas flow and drying gas flow were 3.0 and 15.0 L/min, respectively. Electrospray-ionization was used as the interface with the voltage of 4.5 kV. The temperature for curved desolvation line and heat block were 250 °C and 400 °C, respectively. The pressure for collision induced dissociation gas was 230 kPa. The detector voltage was 1.96 kV. 2.6 UNTARGETED METABOLOMICS ANALYSIS


8

Page 9 of 36

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


Methods and instruments for untargeted metabolomics study including sample pretreatment, gas chromatography-mass spectrometry (GC-MS) and LC-MS analysis, data preprocessing, differential features screening and metabolites identification were similar to our previous studies.30,

31

The details are also provided in Supporting Information (SI) for the readers’

convenience. After metabolic information collecting and data preprocessing, OPLS-DA model was constructed between LS and HS groups regarding pre-dose serum. Differential features were screened out following our strict filtering criteria. For each identified differential metabolite, receiver operating characteristic (ROC) curve was drawn and AUC-ROC was calculated (GraphPad Prism, USA). To explore the variations in metabolic profiles of the two groups before and after cisplatin administration, PCA and OPLS-DA models were also built based on post-dose serum between LS and HS groups. 2.7 DATA QUALITY EVALUATION FOR UNTARGETED METABOLOMICS Samples from control and cisplatin-treated groups were randomly analyzed during the whole instrumental analysis in order to avoid inter-batch differences. Quality control (QC) samples were made by pooling equal aliquot of each sample, from which equal volumes were then taken and treated consistently with study samples. In order to stabilize the analytical system, the first 10 QC samples were injected before the formal analytical batch. For evaluating the robustness of sample preparation, analytical method and the stability of instruments, one QC sample was inserted every 10 real samples in the analytical batch. Data quality evaluation involved the following three aspects: 1) PCA was constructed based on all samples. By investigating the cluster degree of QC samples in PCA score plot, data quality could be evaluated directly and visually;32 2) within-group relative log abundance (RLA) plot was employed to assess the coefficient of variation within QC samples using R program (https://www.r-


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

Page 10 of 36

project.org/).33, 34 If the coefficient of variations were small, the boxplot of features for each QC sample would have a median value close to zero and similar lower and upper quartiles between samples; and 3) variations in retention time and mass spectrometry response for all the differential metabolites were checked in all QC samples. 2.8 BCAAS AND BCKAS DETERMINATION BCAAs and BCKAs were simultaneously determined using multiple reaction monitoring mode in LC-MS/MS. Instruments and methods for the determination referred to our previous study.35 The details were also provided in the SI. Method validation with respect to linearity, matrix effect, stability, precision, accuracy, and recovery, limit of quantification and limit of detection were carried out. 2.9 PREDICTION MODEL CONSTRUCTION BLR was applied to evaluate the predictive accuracy of BCAAs in discovery set (SPSS, Chicago, USA). In order to achieve comparability among different batches of experiments, dosing strategies or laboratories, imported BCAA contents were normalized first. The model diagnosis parameters including -2 log likelihood and Nagelkerke R Square were observed to evaluate the goodness of fit. Probability is a parameter derived from BLR, and its default cut value was 0.5 (full name ‘probability’ is used here to avoid confusion with p value in statistical test). In the present study, individuals were predicted to be high sensitive if probability>0.5 and low sensitive if probability0.99) was obtained with the limit of detection of 0.1 µg/mL. The recovery was more than 80% with relative standard deviation (RSD) less than 15.0%. The matrix effect was from 85% to 115% with RSD lower than 15%. Stability tests, inter- and intra-day precision were also qualified. Concentrations of leucine, isoleucine and valine in pre-dose serum were 16.96±1.25, 13.45±0.78 and 16.89±2.85 µg/mL for LS group. While in HS group, the contents were 20.26±2.74, 15.68±2.00 and 20.68±2.50 µg/mL, respectively. Statistical analysis demonstrated that there were significant differences (p

Branched-Chain Amino Acids as Predictors for Individual Differences

Recommend Documents