Article pubs.acs.org/jpr
Liquid Chromatography−Mass Spectrometry-based Metabolomic Analysis of Livers from Aged Rats Nari Son,† Haeng Jeon Hur,† Mi Jeong Sung,† Myung-Sunny Kim,† Jin-Taek Hwang,† Jae Ho Park,† Hye Jeong Yang,† Dae Young Kwon,† Suk Hoo Yoon,† Hae Young Chung,‡ and Hyun-Jin Kim*,† †
Research Division for Emerging Innovation Technology, Korea Food Research Institute, Republic of Korea College of Pharmacy, Pusan National University, Pusan, Republic of Korea
‡
ABSTRACT: We used UPLC−Q-TOF MS to analyze hepatic metabolites of rats aged 6, 12, 18, and 24 months; the MS data were processed by partial least-squares discriminant analysis (PLS-DA) to investigate the discrimination among sample groups. Rats were significantly separated with increasing age, except those aged between 6 and 12 months. We identified only 25 of 120 metabolites contributing to the separation: lipid metabolites (glycerol-3-phosphate, linolenic acid, lysophosphatidylcholines [lysoPCs]), energy metabolism intermediates (betaine, carnitine, acylcarnitines, creatine, pantothenic acid), nucleic acid metabolites (inosine, xanthosine, uracil, hypoxanthine, xanthine), and tyrosine. Aging accumulated energy metabolism intermediates, hypoxanthine, xanthine, and 2 major lysoPCs (C18:0 and C22:6). The NAD level and NAD/NADH ratio decreased with age. It was indicated that aging might decrease energy production through β-oxidation because of a decrease in NAD despite the accumulation of lipid energy metabolism intermediates. In addition to energy dysregulation, hypoxanthine and xanthine, which are elevated with age, might accumulate reactive oxygen species in the liver. These results strongly support two aging theories: those of energy dysregulation and free radicals. Additionally, we propose a metabolic pathway related to aging based on these hepatic metabolites. These metabolites and the proposed aging pathway could be used to understand aging and related diseases better, and increase the predictability of aging risk. KEYWORDS: aging, betaine, carnitine, energy dysregulation, hypoxanthine, lysophosphatidylcholine, metabolomics, NAD, UPLC−Q-TOF, xanthine
■
INTRODUCTION The complex biological process of aging results in changes of biological and physiological functions.1 Accumulated research data indicate that these changes are eventually the most important risk factor for age-related diseases, including cardiovascular disease, type 2 diabetes, cancer, and Alzheimer’s disease.2,3 Given that these diseases have strongly positive correlations with mortality rates in older people, the sum of all age-related diseases is the best biomarker of aging.3 Thus, most aging studies have been focused on identifying the biomarkers related to a number of these diseases using various biotechnologies.4 However, in spite of many scientific efforts to understand the mechanism of aging and age-related diseases, and the development of several biological theories such as the free radical and misrepair-accumulation theories,5,6 this biological mechanism remains unknown. Recently, to improve understanding of aging, metabolomics (also referred to as metabonomics), an emerging technology for the systematic comprehensive study of small molecules in a biological cell, tissue, organ, or organism that are the end products of cellular processes,7 has been used to identify potential biomarkers associated with aging and age-related diseases. Although no single instrument can analyze all metabolites, several © 2012 American Chemical Society
aging metabolomic studies on Caenorhabditis elegans, Drosophila, rodents, humans, and other model organisms based on nuclear magnetic resonance (NMR) and mass spectral methods have identified many metabolites associated with age.8,9 In particular, the blood, urinary, or brain tissue metabolome of acceleratedaging mouse models or young, adult, and old rats suggested that aging might be related with mitochondrial dysfunction and lipid and glucose dysmetabolism.10−16 A similar result was reported in a human plasma metabolomic study.17 Human plasma metabolomes analyzed using gas chromatography−mass spectrometry (GC−MS) and high performance liquid chromatography−mass spectrometry (HPLC−MS) indicated that significant changes in the relative concentration of more than 100 metabolites were associated with age, and changes in protein, energy and lipid metabolism as well as oxidative stress were observed with increasing age. Although these aging metabolomic studies provided a new approach for studies on aging, to the best of our knowledge, hepatic metabolomic study of normal aged rats without mediated aging acceleration has not been studied except in peroxisome Received: December 27, 2011 Published: March 2, 2012 2551
dx.doi.org/10.1021/pr201263q | J. Proteome Res. 2012, 11, 2551−2558
Journal of Proteome Research
Article
Figure 1. Characteristics of the livers from rats of different ages. Triglyceride (TG) content (A), total cholesterol content (B), NAD content (C), and NAD/NADH ratio (D). The contents of TG, cholesterol, NAD, and NADH were analyzed by quantitative assay kits. Different letters above bars indicated significant differences at p < 0.05. All experiments were replicated six times.
an NAD/NADH Quantification Kit (BioVision Inc., Mountain View, CA, USA).
proliferator-activated receptor (PPAR)-α-null mice using NMR and GC−MS.18 Therefore, in this study, metabolic profiling of livers from aged rats (6, 12, 18, and 24 months old [m/o]) was analyzed by ultra-performance liquid chromatography− quadrupole-time-of-flight mass spectrometry (UPLC−Q-TOF MS), and the resultant data was applied to multivariate statistical analysis to find potential aging biomarkers for improving understanding of the age-based change of the liver mechanism.
■
UPLC−Q-TOF MS Analysis of Liver Extracts
A UPLC system (Waters, Milford, MA, USA) equipped with a binary solvent delivery system and an autosampler was used to analyze the liver metabolites. The liver extracts were injected into an Acquity UPLC BEH C18 column (2.1 × 50 mm, 1.7 μm; Waters) connected to the UPLC system and equilibrated with water containing 0.1% formic acid (FA). The sample was eluted in a gradient with ACN containing 0.1% FA at a flow rate of 0.35 mL/min for 15 min and the metabolites separated by C18-UPLC were analyzed and assigned by Q-TOF MS (Waters). The Q-TOF was operated in positive electrospray ionization (ESI) mode. The capillary and sampling cone voltages were set at 3 kV and 30 V, respectively. The desolvation flow was set to 700 L/h at a temperature of 300 °C, and the source temperature was set to 110 °C. The TOF MS data were collected in the m/z 100−1000 range with a scan time of 0.2 s and interscan delay time of 0.02 s. All analyses were acquired using lock spray to ensure accuracy and reproducibility; leucine-enkephalin (556.2771 Da in positive ESI mode) was used as the lock mass at a concentration of 200 ρmole and a flow rate of 5 μL/min. The lock spray frequency was set at 10 s. For quality control (QC), a mixture of 5 standard compounds (caffeine, sulfadimethoxine, terfenadine, 4-acetoaminophenol, and reserpine) was injected after every 6 samples. The MS/MS spectra of metabolites were obtained by a collision energy ramp from 10−30 eV. The accurate mass and composition for the precursor and fragment ions were calculated and sequenced by MassLynx (Waters), incorporated in the instrument. All MS data, including retention time, m/z, and ion intensity, were extracted with MarkerLynx software
EXPERIMENTAL SECTION
Animals and Sample Preparation
The liver tissues used in this study were obtained from the Aging Tissue Bank (Pusan National University, South Korea). Male Sprague−Dawley (SD) rats were housed in the Aging Tissue Bank at a constant temperature (22−26 °C) under 12-h light/dark cycles per day. Rats had ad libitum access to autoclaved water and a normal pellet diet for 6, 12, 18, and 24 months (6 rats per group). After sacrifice, the rat livers were collected and immediately placed in liquid nitrogen. All liver samples were stored at −70 °C until analysis. The freeze-dried liver powder (50 mg) was extracted with 1 mL of cold acetonitrile (ACN) by a homogenizer. After centrifuging, the supernatant was dried and the dried sample was dissolved with 40% aqueous methanol containing terfenadine as a standard compound for UPLC−Q-TOF analysis.19 Triglyceride, Cholesterol, and NAD/NADH Ratio Assay
Triglyceride (TG) and cholesterol contents in the liver extracted by a solvent mixture of chloroform/methanol (2:1) were determined by a TG Assay Kit (Asan Pharm., Co. Korea) and a Cholesterol/Cholesteryl Ester Quantitation Kit (Asan Pharm.), respectively. The hepatic NAD/NADH ratio was measured with 2552
dx.doi.org/10.1021/pr201263q | J. Proteome Res. 2012, 11, 2551−2558
Journal of Proteome Research
Article
Figure 2. Ultra-performance liquid chromatography−quadrupole-time-of-flight mass spectrometry (UPLC−Q-TOF MS) profiles of livers from 6-, 12-, 18-, and 24-month-old (m/o) rats (A) and partial least-squares discriminant analysis (PLS-DA) scores (B) and loadings plots (C) obtained from the MS data. Outlying samples of the ellipse region with the 95% confidence interval were excluded by Hotelling’s T2 test. The scores plots showed significant separation among samples on the basis of the model quality parameters: R2X, R2Y, and Q2Y. The PLS-DA model was validated by a permutation test (n = 200): P-values and intercepts of R2 (Ri) and Q2 (Qi). The numbers for the metabolites are as given in Table 1.
quantified by R2X and R2Y and the predictive ability was indicated by Q2Y. A 3-fold validation was applied to PLS-DA models to validate them, and the models’ reliabilities were further rigorously validated by a permutation test (n = 200). To identify metabolites contributing to the discrimination, the metabolite intensity differences of the sample groups were tested by one-way analysis of variance (ANOVA) with Duncan’s test (P < 0.05) (2003; SPSS Inc., Chicago, IL, USA). We used the loadings plot of w*C[1] and [2] and the S-plot showing a combination of covariance p(1) and correlation p(corr) from the PLS-DA model to better visualize the metabolites contributing to the discrimination. Identified metabolites with significant differences (P < 0.05) were also visualized in a heat map drawn by R with g-plots.
(Waters), also incorporated in the instrument, and were assembled into a data matrix. Data Processing
LC−MS data of the liver extracts, including retention time, m/z, and ion intensity were extracted using MarkerLynx software (Waters) and assembled into a data matrix, and were deconvoluted, aligned, and normalized by MarkerLynx. Peaks were collected using a peak width at 5% height of 1 s, a noise elimination of 6, and an intensity threshold of 50. Data were aligned with 0.04 Da mass tolerance and a retention time window of 0.15 s. All spectra were aligned and normalized to an external standard. Assignment of metabolites contributing to the observed variance was carried out using the ChemSpider (www.chemspider.com) and the Human Metabolome Database (www.hmdb.ca). Authentic standards were used to confirm the assignments and their quantitative analysis.
■
RESULTS
Animal Characteristics
Data Analysis
The characteristics of rats according to age are shown in Figure 1. The hepatic TG levels of 12-m/o rats (449.5 ± 33.1 mg/g) was 22% higher than that of 6-m/o rats, but further increase was not observed up to 24 months. The total cholesterol content in the liver increased with age. The total cholesterol content of 24-m/o rats (3.7 ± 0.5 mg/g) was 32% higher than that of 6-m/o rats. In addition, the hepatic NAD/NADH ratio of rats decreased with
The mean-centered LC−MS data sets were analyzed by multivariate statistical analysis using SIMCA-P+ version 12.0.1 (Umetrics, Umeå, Sweden). Partial least-squares discriminant analysis (PLS-DA) was used to visualize discrimination among samples. The quality of PLS-DA models was assessed by 3 parameters: R2X, R2Y, and Q2Y. The goodness of fit was 2553
dx.doi.org/10.1021/pr201263q | J. Proteome Res. 2012, 11, 2551−2558
Journal of Proteome Research
Article
Figure 3. PLS-DA scores of 6- and 12-month (A), 12- and 18-month (B), 18- and 24-month (C), and 6- and 24-month models (D), and S-plots associated with the PLS-DA scores plots (E−H) of liver metabolite profiles derived from UPLC−Q-TOF MS. PLS-DA models were fitted by quality parameters (R2X, R2Y, and Q2Y) and validated by a permutation test (n = 200) (P-values, Ri, and Qi). The numbers for the metabolites are as given in Table 1.
age, consistent with a decline in NAD levels. The ratio (0.65 ± 0.49) and the NAD level (20.6 ± 13.0 pmol/mg) of 24-m/o rats were 1.9-fold lower than those of 12-m/o rats.
contributing to the discrimination (Figure 2C). The plot revealed that the metabolites far from the cross intersection of w*C[1] and [2] were the more relevant ions for explaining the discrimination. The metabolite contents placed in the positive w*C[1] were decreased with age, while the opposite was true for those in the negative w*C[1]. To investigate the differences between rats according to age further, the discrimination between the models for 6 and 12 months, 12 and 18 months, 18 and 24 months, and 6 and 24 months, respectively, was visualized on the PLS-DA scores plots (Figure 3A−D). Except for the 6- and 8-month model, all groups of samples were significantly separated from each other with high parameter values (Figure 3E−H). To identify the metabolites contributing to the discrimination, S-plots of all models were generated using Par scaling (Figure 3E−H). Except the 18- and 24-month models, the metabolite levels in all models placed in the positive p[1] and p(corr)[1] decreased with age, whereas those in the negative section were increased. Inversely, in the 18- and 24-month model, the metabolite levels with positive p[1] and p(corr)[1] values increased with age, but that of negative values decreased. The number-marked metabolites in the plots were identified (Table 1) and their fold changes were calculated.
Metabolomic Analysis of Hepatic Metabolites
The liver metabolites from rats according to age (6, 12, 18, and 24 months) were analyzed by UPLC−Q-TOF (Figure 2A) and PLS-DA was applied to the data. The QC data showing the retention time shift of
