Practical Analytical Approach for the Identification of Biomarker

Practical Analytical Approach for the Identification of Biomarker Candidates in Prediabetic State Based upon Metabonomic Study by Ultraperformance Liquid Chromatography Coupled to Electrospray Ionization Time-of-Flight Mass Spectrometry Haruhito Tsutsui,† Toshio Maeda,‡ Toshimasa Toyo’oka,*,† Jun Zhe Min,† Shinsuke Inagaki,† Tatsuya Higashi,† and Yoshiyuki Kagawa‡ Laboratory of Analytical and Bio-Analytical Chemistry, Laboratory of Clinical Pharmaceutics and Pharmacy Practice, Graduate School of Pharmaceutical Sciences, and Global COE Program, University of Shizuoka, 52-1 Yada, Suruga-ku, Shizuoka 422-8526, Japan Received February 9, 2010

The number of diabetic patients has recently been increasing worldwide. Thus, the discovery of potential diabetic biomarker(s), leading to the early detection and/or prevention of diabetes mellitus, is strongly required. The diagnosis of the prediabetic state in humans is a very difficult issue because of the lifestyle differences in each person and ethical consideration. Upon the basis of these considerations, animal experiments using ddY strain mice (ddY-H), which undergo naturally occurring diabetes along with age, were carried out in this study. Biomarker discovery based upon a metabonome study is now quite common, the same as that in the proteome analysis. Reversed-phase liquid chromatography-mass spectrometry (LC-MS) has mainly been used for the extensive analysis of low-molecular mass compounds including metabolites. The metabolites in the plasma of diabetic mice (ddY-H) and normal mice (ddY-L) were exhaustively separated and detected by ultraperformance liquid chromatography along with electrospray ionization time-of-flight mass spectrometry (UPLC-ESI-TOF-MS) using T3C18 and HS-F5 columns. The biomarker candidates related to diabetes mellitus were extracted from the metabolite profiling of ddY-H and ddY-L at 5, 9 13, and 20 weeks old using a multivariate statistical analysis such as orthogonal partial least-squares-discriminant analysis (OPLS-DA). Various metabolites and unknown compounds were detected as biomarker candidates related to diabetic mellitus. Furthermore, the concentration of several metabolites on Lysine biosynthesis and Lysine degradation pathways were remarkably changed between the 9-week old ddY-H and ddY-L mice. Because a couple of biomarker candidates related to the prediabetic state were identified using the present approach, the metabolite profiling study could be helpful for understanding the abnormal state of various diseases. Keywords: ultraperformance liquid chromatography (UPLC) • electrospray ionization time-of-flight mass spectrometry (ESI-TOF-MS) • metabonomics • metabolite profiling • diabetes mellitus • principal component analysis (PCA) • orthogonal partial least-squares-discriminant analysis (OPLS-DA) • ddY mice

Metabonome research is considered as the next pillar of system biology, following the genome, transcriptome and proteome studies. The term “metabonome” research generally means the understanding of biological systems based upon the total analysis of a few thousand endogenous metabolites in the end points of biological cascades. The metabonomics approach, which is a top-down and nonhypothesis-driven analysis, is also used to detect the profile difference in the metabolites reflecting the disease state in biological specimens.1,2 The search of biomarker candidates, which are endogenous metabolites and/

or biological molecules found in biological specimens indicating a normal or abnormal state, can be attempted by a metabolite profiling study using appropriate analytical tools.3-15 The metabolic analysis platform must integrate various steps, such as sample handling, sample preparation, sample measurement, data assessment, biomarker candidate extraction and biomarker identification, because a wide variation of metabolites occurs in the samples and the concentration range is very broad, pM-mM in general.16 The significant variety and concentration of the distributed metabolites require a sophisticated analytical method to cover the extended compounds.

* To whom correspondence should be addressed. Tel.: +81-54-264-5656. Fax: +81-54-264-5593. E-mail: [email protected]. † Laboratory of Analytical and Bio-Analytical Chemistry. ‡ Laboratory of Clinical Pharmaceutics and Pharmacy Practice.

A widely used analytical technique in metabonome studies is nuclear magnetic resonance spectroscopy (NMR)17-21 and mass spectrometry (MS).3,22-24 The use of the NMR method is recently minor because of its limited sensitivity and number

Introduction

3912 Journal of Proteome Research 2010, 9, 3912–3922 Published on Web 06/18/2010

10.1021/pr100121k

 2010 American Chemical Society

research articles

Identification of Biomarker Candidates in Prediabetic State of detected compounds. Various types of MS combined with liquid chromatography (LC), gas chromatography (GC)25 or capillary electrophoresis (CE)26 are now commonly applied as an alternative analytical tool for the metabonome research. Atmospheric pressure ionization techniques, such as ESI and APCI, are generally employed for MS detection. In addition, the use of positive and negative ionization modes is recommended to maximize the compound coverage.27-29 The use of multicolumns, possessing different separation modes (e.g., reversed-phase and normal-phase LC), is useful for the peak coverage and to ensure the reliability. Furthermore, the steadily progress in MS instruments (e.g., Q-TOF-MS and Orbitrap-MS), LC instruments (e.g., UPLC and UFLC)22 and related techniques (e.g., HILIC mode)30 has accelerated the metabonome research. The introduction of ultrahigh speed LC systems, such as UPLC, which greatly improves the chromatographic resolution, sensitivity and peak capacity (detectable peak number) in a short run time, increased the speed of the LC-driven metabonomic analysis. The statistical data evaluation and database search systems are another important aspect for the identification of potential biomarkers. The data collection by LC-MS followed by a multivariate statistical data analysis are the most straightforward steps in a metabonomics study. Nowadays, the biomarker candidates can easily be extracted from metabolite profiling. Once the potential biomarker(s) have been selected, precise identification based upon the LC-MS analysis of authentic compounds and various database searches, for example, ChemSpider, KEGG, METLIN and Human metabonome database (HMDB), including the exact mass and elemental composition data, etc., is essentially required. Diabetes mellitus is one of the diseases increasing worldwide, not only in the western world.31,32 In many cases, diabetes is asymptomatic for a long period and the patient is not aware of the disease. Early diagnosis of this prediabetic state, lifestyle modification and/or medical treatment are effective for preventing the development of Type II diabetes.33,34 The Type II diabetes in humans is a multifactorial disorder based on environmental factors and genetic background. Diabetes usually occurs after middle age. However, the diagnosis of the prediabetic state in humans is a very difficult issue, because the lifestyle and food intake, etc. are variable in each person. Therefore, we planned animal experiments using ddY mice that undergo naturally occurring diabetes with aging. In general, the glucose levels and the duration of hyperglyceamia in ddY mice are variable in individual mice. Thus, two strains, namely spontaneous insulin-resistant mice (ddY-H) and noninsulin resistant mice (ddY-L),35 isolated from the ddY strain mice by inbreeding based on the induction of serious hyperglycemia by serum glucose levels after the refeeding, were used in this study. The extraction of biomarker candidate(s) related to prediabetes and diabetes in mice plasma was performed based upon the metabolite profiling technique. The low-molecular mass metabolites (m/z 50-650) in the plasma of ddY-H and ddY-L mice at fixed weekly intervals were determined by UPLC separation using 2 different columns and ESI-TOF-MS detection. The collected peak data involving the retention time, intensity and m/z values were evaluated by the most common multivariate statistical methods, such as principal components analysis (PCA) and orthogonal partial least-squares-discriminant analysis (OPLS-DA).36 The biomarker candidate(s) extracted from the differential analysis of the multivariate sta-

tistical methods were identified through a database search. The metabolic pathways related to diabetes were also discuss based upon the detected metabolites.

Experimental Section Materials and Chemicals. Leucine-enkephalin (used as reference of m/z value) was purchased from Sigma-Aldrich (St. Louis, MO). Acetonitrile (CH3CN) and formic acid (HCOOH) were of LC-MS grade (Wako Pure Chemicals, Osaka, Japan). Other chemicals, such as the metabolites, were of special reagent grade. Deionized and distilled water (H2O) was used throughout the study (Aquarius PWU-200 automatic water distillation apparatus, Advantec, Tokyo, Japan). Mice. Four-week old ddY mice were purchased from SLC, Inc. (Hamamatsu, JAPAN), and the ddY-H and ddY-L mice in our colony were used. The mice were maintained on 12 h light/ dark cycles with free access to the standard chow pellets (MF diet, Oriental Yeast Co., Ltd.) and water ad lib. The animal care and experiments were performed in accordance with the guidelines for the care and use of laboratory animals at the University of Shizuoka. Measurements of Glucose, TG and Body Weight. Samples for the analyses were obtained without fasting at 13:00 from the mice at 5, 9, 13, and 20 weeks of age. Twenty microliters of blood was obtained from the caudal vein for the determination of the glucose and triglyceride (TG). Under anesthesia with pentobarbital, blood was withdrawn from the central vein and kept at -80 °C until analyzed. The body weight of each mouse was also determined at the same time. The serum glucose and TG were measured by the Glucose CII Test-Wako (Wako Pure Chemical Industries, Ltd.) and Triglyceride E Test-Wako (Wako Pure Chemical Industries, Ltd.), respectively. Blood Collection and Pretreatment for Metabonomics Study. Blood for the metabonomics study was obtained without fasting from the mice at 5, 9, 13, and 20 weeks of age. The heparinized blood was immediately centrifuged at 3000 rpm for 5 min. The separated plasma was then stored at -80 °C just before the analysis. To 200 µL of each plasma sample, 600 µL of acetonitrile was added and deproteinized using the Sirocco Protein Precipitation Plate (Waters, Milford, MA), and the filter plate was vortexed for 10 min. Next, after connecting the filter plate to a vacuum manifold and vacuuming up at 10 Hg for 15-20 min, the collection (650 µL) was transferred to an auto sampler vial. Each 10 µL aliquot was injected into the UPLC-ESI-TOF-MS system. UPLC-ESI-TOF-MS Analysis. The UPLC-ESI-TOF-MS system consisted of an ACQUITY ultraperformance lipid chromatograph and a Micromass LCT Premier XE mass spectrometer (high-sensitivity orthogonal time-of-flight instrument; Waters, Milford, MA). An ACQUITY UPLC HSS T3 C18 column (1.8 µm, 150 mm × 2.1 mm i.d., Waters) [T3-C18 column] and Discovery HS-F5 HPLC column (3 µm, 150 mm ×2.1 mm i.d, SUPELCO) [HS-F5 column] were used as the analytical columns. The columns were maintained at 40 °C. The flow rate of the mobile phase was 0.4 mL/min (T3-C18 column) or 0.3 mL/min (HS-F5 column). The TOF-MS was operated in the positive ion mode using an electrospray-ionization source (ESI+). The optimized conditions for the UPLC separation and ESI-TOF-MS detection are shown in Table 1. Statistical Analyses. The raw data were analyzed for peak detection and alignment and exported for a PCA and orthogonal signal correction partial least-squares discriminant analysis Journal of Proteome Research • Vol. 9, No. 8, 2010 3913

research articles

Tsutsui et al.

Table 1. UPLC-ESI-TOF-MS Conditions

Column temperature Flow rate Injection volume

UPLC (separation conditions) ACQUITY UPLC HSS T3-C18 column (1.8 µm, 150 mm ×2.1 mm i.d., Waters) 0.1% HCOOH in H2O 0.1% HCOOH in CH3CN B% ) 0% maintained (0-1 min), 100% linearly increased (1-11 min) and maintained (11-13 min), 0% maintained (13-18 min) 40 °C 0.4 mL/min 10 µL

Polarity Capillary voltage Sample cone voltage Desolation gas flow Cone gas flow Source temperature Desolvation temperature MS range

TOF-MS (LCT Premier XE conditions) ESI positive (W mode) 3000 V 50 V 700 L/h 50 L/h 120 °C 350 °C 50-650 m/z

Column Mobile phase A Mobile phase B Gradient elution

Discovery HS-F5 HPLC Column (3 µm, 150 mm ×2.1 mm i.d., SUPELCO) B% ) 0% maintained (0-10 min), 100% linearly increased (10-25 min) And maintained (25-30 min), 0% maintained (30 min-40 min) 0.3 mL/min

Table 2. Physical Data of ddY-H and ddY-L Mice

(OSC-PLS-DA) by MarkerLynx XS (Waters, Milford, MA). The method parameters were set as follows: Mass tolerance ) 0.05 Da. Apex Track Peak Parameters: Peak width at 5% height (seconds) ) 15/Peak-to-peak baseline noise ) 50, Apply smoothing ) Yes. Collection Parameters: Intensity threshold (counts) ) 100/Mass window ) 0.05/Retention time window ) 0.10, Noise elimination level ) 6, Deisotope data ) Yes. Biomarker Candidate Search. Metabolites were identified based on the accurate mass, retention time, and matching the MS spectra of the unknowns to the standard model compounds. MarkerLynx XS combined lists of the biomarker candidates were extracted from three metabonomic databases: i.e., ChemSpider, KEGG (http://www.kegg.com/) and the human metabolite database (http://www.hmdb.ca/). A mass tolerance of 5.0 mDa was set as well as the maximum elemental composition of C ) 500, H ) 1000, N ) 200, O ) 200, S ) 10, P ) 10, and Cl ) 10. The software automatically filters compounds from different libraries that have the same KEGG.

Results and Discussion Diabetes mellitus is rapidly increasing not only in western areas but also worldwide. Thus, early diagnosis of the prediabetic state is required to prevent the progress of the disease. The fasting glucose concentration in blood is generally used for the diagnosis; however, information about the blood concentration of the metabolites in the prediabetic state is rare. Wilson et al.37-41 actively investigated the identification of biomarker candidates in diabetes mellitus using Zucker rat strains (normal wild, fa/fa and lean/fa). According to the papers, some kinds of compounds are identified as the potential biomarker candidates of diabetes in Zucker rats. However, the biomarker discovery in prediabetic state is not mentioned. The aim of this study is to search for new biomarker(s) in the blood that clearly changed in the prediabetic state. However, the detection of biomarker candidates in human is very difficult because of the lifestyle differences (e.g., food, work, environment and living area) in each person. Another difficulty is the ethical consideration of the experiments using human samples. Therefore, an animal experiment was planned for possible identification of the biomarker(s) to detect during the early stage of diabetes. 3914

Journal of Proteome Research • Vol. 9, No. 8, 2010

age

/

5 9 13 20

?

5 9 13 20

ddY

body weight (g) Av ( SD

TG (mg/dL) Av ( SD

glucose (mg/dL) Av ( SD

H L H L H L H L H L H L H L H L

33.84 ( 2.44 29.32 ( 3.92 49.63 ( 3.30 37.55 ( 3.69 61.35 ( 3.48 47.83 ( 2.95 59.84 ( 2.28 48.58 ( 4.01 27.81 ( 1.48 29.61 ( 1.07 38.85 ( 2.26 34.35 ( 1.46 48.99 ( 3.14 39.95 ( 2.11 61.98 ( 2.81 42.67 ( 2.83

116 ( 47.1 88 ( 23.0 116 ( 28.1 73 ( 16.8 141 ( 22.3 77 ( 8.5 121 ( 21.4 71 ( 8.7 83 ( 14.2 75 ( 13.4 88 ( 22.2 60 ( 14.4 90 ( 13.8 64 ( 9.6 83 ( 22.1 69 ( 13.8

137 ( 17.0 126 ( 11.6 148 ( 16.7 129 ( 8.9 182 ( 17.9 128 ( 7.4 175 ( 20.6 114 ( 8.7 126 ( 10.5 123 ( 11.0 127 ( 9.3 129 ( 14.0 129 ( 11.1 122 ( 12.4 129 ( 17.7 112 ( 13.9

The animals tested in this study are ddY strain mice (ddY-H and ddY-L) that were isolated from ddY mice by selective breeding over 20 generations based on their serum glucose levels after the fasting-refeeding treatment.35 The physiological data of both mice groups at each age are shown in Table 2. The serum TG and glucose levels in the ddY-H mice were remarkably high, and those in the ddY-L mice were relatively low. The drastic change in the glucose levels occurs between 9 and 13 weeks. The striking differences in the serum glucose levels between the ddY-H and ddY-L mice were observed in the males, but not in the females. These characteristics are recessively inherited. No clear difference in the physical data in the male mice at the same week was observed in each group, except for the glucose concentration. The ddY-H mice are naturally attacked by the disease at around 13 weeks, while the ddY-L group is healthy up to 20 weeks of testing. The male ddY-H mice develop insulin resistance at 12-15 weeks of age and show diabetic symptoms after six months (20 weeks more).35 Since the symptoms were spontaneously induced without any loading, such as nutritional stress, the ddY-H mice seem to be a useful diabetic model species. In contrast, the ddY-L mice are normal without any diabetic symptoms and

Identification of Biomarker Candidates in Prediabetic State

research articles and ddY-H at 5, 9, 13, and 20 weeks were applied to the UPLC-ESI-TOF-MS and used for the multivariate statistical analysis.

Figure 1. Typical TIC obtained from mice plasma. (A) ddY-L at 5 weeks; (B) ddY-H at 5 weeks; (C) ddY-L at 20 weeks; (D) ddY-H at 20 weeks. The UPLC-ESI-TOF-MS conditions are the same as those in Table 1.

used as the control mice. Because the glucose levels of the female ddY-H mice were comparable to those of the ddY-L mice, male mice were only used in the following studies. The plasma sets (total 80 samples) of each of 10 male mice of ddY-L

Determination of Low-Molecular Mass Compounds in Mice Plasma by UPLC-ESI-TOF-MS. More than a thousand compounds including metabolites are contained in biological specimens, such as plasma. The use of different mode columns seems to be effective to increase the coverage of these compounds.30 In this study, the chromatographic separation of lowmolecular mass compounds in mice plasma was carried out using different columns: one is an ODS column (T3-C18) packed with small particles, and the other is an HS-F5 column which contains pentafluorobenzene in the silica particles. Other type columns, such as normal-phase columns (silica, amide, and hydroxyl, etc.), seem to be equally useful for this purpose. However, the theoretical plate numbers of these columns are generally lower than the reversed-phase (RP) columns, such as ODS. As a result, the peak capacity (detectable peak number) for the same run time is also low. Reproducibility of the retention times after the multisample separations is another concern. Although a hydroxyl interaction liquid chromatograph (HILIC), which uses water and water miscible organic solvents, such as methanol and acetonitrile as the mobile-phase, seems to be compatible with the RP chromatography, the peak number detected for the same run time (peak capacity) was lower than that of the RP chromatography in our experiments.

Figure 2. PCA score plots of ddY-L and ddY-H groups at 5-20 weeks. (A) ddY-L by T3-C18 column; (B) ddY-H by T3-C18 column; (C) ddY-L by HS-F5 column; (D) ddY-H by HS-F5 column. Weeks, 5 (black), 9 (red), 13 (green), and 20 (blue). The other UPLC-ESI-TOF-MS conditions are the same as those in Table 1. Journal of Proteome Research • Vol. 9, No. 8, 2010 3915

research articles

Tsutsui et al.

Figure 3. OPLS-DA score plot and S-plot of ddY-H (black) and ddY-L (red) at 9 weeks. (A) and (B), T3-C18 column; (C) and (D), HS-F5 column. (A) and (C), score plot; (B) and (D), S-plot. Metabolites denoted by (red open square) could be selected as biomarker candidates.

On the basis of these observations, the HS-F5 column was employed together with the T3-C18 column. After various trials, the gradient elution using a mixture of water-acetonitrile containing 0.1% formic acid provided the best separation for the plasma sample in both columns (Table 1). Figure 1 shows the typical examples of the total-ion chromatograms (TIC) in the plasma of the ddY-L and ddY-H mice. As shown in the chromatograms, the reproducibility of the peaks seemed to be good enough for the metabolite profiling study. More than 10 000 ions were detected by ESI-TOF-MS during a chromatographic run. The number seems to be comparable with the report by Wilson et al.42 Extraction of Biomarker Candidates. To understand the change in the metabolites in the prediabetic state, lowmolecular mass compounds including metabolites in the mice plasma at each age (5, 9, 13, or 20 weeks) were determined by UPLC-ESI-TOF-MS using two columns. Several TICs in the positive-ion mode, obtained from the ddY-L and ddY-H mice plasma at 5 and 20 weeks old are shown in Figure 1. The metabolic change in both groups and/or for each age is not clear from the chromatograms. In order to observe the difference in the metabolite profiling, the preprocessed UPLC-ESITOF-MS data were further investigated by a multivariate statistical analysis. Figure 2A and B show the principal component analysis (PCA) score plots of the first and second components using the T3-C18 column, while Figure 2C and D 3916


are the results of the HS-F5 column. The goodness of fit was greater than 0.35. The Hottelling’s T2 range was used to provide the number of outliers among the samples at the level of 0.05. As shown in Figure 2, each group was classified into 4 groups, but the difference at weeks 5 and 9 was not clear in the ddY-L groups (Figure 2A and C). In the ddY-H group mice, the difference at weeks 9 and 13 was not clear (Figure 2B and D). The PCA results in the ddY-L normal mice suggest that the biomarker candidate(s) concerning aging should be identified by a further investigation. Since the purpose of this study is the identification of biomaraker(s) in the prediabetic state, the results are not shown in this paper. Although the classification between ddY-L and ddY-H was also satisfactory from the PCA score plots (data not shown), a partial least-squares-discriminant analysis (PLS-DA) with the orthogonal signal collection (OSC) data filter of the ddY-L and ddY-H groups was carried out at each age (5, 9, 13, and 20 weeks). Figures 3 and 4 show the OPLS-DA score plots and S-plots of the ddy-L and ddy-H at 9 or 20 weeks. The 20-week old mice were in a diabetic state, while the mice at 9-13 weeks were in the prediabetic state as shown in Table 2. The ddY-H and ddY-L groups were clearly classified by component [1] with a goodness of fit of greater than 0.95 (data not shown). A similar good classification was also observed in the groups at 5 and 13 weeks. The results indicate that the OPLS-DA has a good capability for the classification of the prediabetic state and/or diabetic state. The

research articles


Figure 4. OPLS-DA score plot and S-plot of ddY-H (black) and ddY-L (red) at 20 weeks. (A) and (B), T3-C18 column; (C) and (D), HS-F5 column. (A) and (C), score plot; (B) and (D), S-plot. Metabolites denoted by (red open square) could be selected as biomarker candidates. Table 3. Metabolites in Lysine Biosynthesis Pathway, Increased and Decreased by ddY-Ha age (week) no.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

name L-Homoserine L-Aspartate L-2-Aminoadipate

6-semialdehyde 2-Oxoglutarate L-Lysine 2-Oxoadipate L-2-Aminoadipate L-2,3-Dihydrodipicolinate 2,3,4,5-Tetrahydrodipicolinate Homocis-aconitate N2-Acetyl-L-lysine L-2-Amino-6-oxopimelate N2-Acetyl-L-aminoadipate N6-Acetyl-LL-2,6-diaminopimelate Saccharopine

mass

elemental composition (mDa, i-FIT)

5

9

13

20

119.0582 133.0375 145.0739 146.0215 146.1055 160.0372 161.0688 169.0375 171.0532 188.0321 188.1161 189.0637 203.0794 232.1059 276.1321

C4H9NO3 C4H7NO4 C6H11NO3 (0.7, 0.3) C5H6O5 (0.0, 1.4) C6H14N2O2 (0.1, 0.0) C6H8O5 C6H11NO4 (0.9, 1.3) C7H7NO4 (2.3, 0.7) C7H9NO4 (0.0, 0.8) C7H8O6 C8H16N2O3 C7H11NO5 C8H13NO5 C9H16N2O5 C11H20N2O6 (4.1, 4.8)

f f v f f V v V V f v V v f vvv

f f f f v f f VVV VVV vv v VV v VV vvv

f f f V f V f

v v f VVV v f f

V f f f f V

V V v f v V

a The arrows of increase (v, vv, and vvv) denote the ratio (RI) of MS intensities of each compound in ddY-H to ddY-L [RI ) (ddY - H)/(ddY - L)]. The arrows of decrease (V, VV, and VVV) denote the ratio (RD) of MS intensities of each compound in ddY-L to ddY-H [RD ) (ddY - L)/(ddY - H)]. f, RI or RD ) 0.8-1.2; v, RI ) 1.2-2.0; vv, RI ) 2.0-3.0; vvv, RI ) >3.0; V, RD ) 1.2-2.0; VV, RD ) 2.0-3.0; VVV, RD ) >3.0.

possible metabolic differences between normal and abnormal states seem to be thus clearly demonstrated. To detect the potential biomarker candidate(s) related to diabetes, the difference in the metabolites between the ddY-L and ddY-H at each age were extracted by the S-plot of the OPLS-DA. The

compounds possessing a higher reliable and difference in all the samples at each age were selected as the biomarker candidate(s). Identification of Potential Biomarker Candidates. As the potential biomarker candidates for the prediabetic and/or Journal of Proteome Research • Vol. 9, No. 8, 2010 3917

research articles

Tsutsui et al. a

Table 4. Metabolites in Lysine Degradation Pathway, Increased and Decreased by ddY-H

age (week) no.

name

mass


5

9

13

20

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21

Piperideine Cadaverine 5-Aminopentanoate delta1-Piperideine-2-carboxylate delta1-Piperideine-6-L-carboxylate L-Pipecolate Glutarate 6-Amino-2-oxohexanoate L-2-Aminoadipate 6-semialdehyde 5-Acetamidopentanoate 2-Oxoadipate L-2-Aminoadipate Carnitine N6-Hydroxy-L-lysine 6-Acetamido-2-oxohexanoate N6-Acetyl-L-lysine N6,N6,N6-Trimethyl-L-lysine N6-Acetyl-N6-hydroxy-L-lysine 3-Hydroxy-N6,N6,N6-trimethyl-L-lysine Saccharopine Aerobactin

83.0735 102.1157 117.079 127.0633 127.0633 129.079 132.0423 145.0739 145.0739 159.0895 160.0372 161.0688 161.113 162.1004 187.0845 188.1161 188.1525 204.111 205.1552 276.1321 564.2279

C5H9N (0.1, 0.0) C5H14N2 C5H11NO2 (0.3, 0.0) C6H9NO2 C6H9NO3 C6H11NO2 (0.3, 0.0) C5H8O4 C6H11NO3 (0.7, 0.3) C6H11NO3 (0.7, 0.3) C7H13NO3 (0.4, 0.2) C6H8O5 C6H11NO4 (0.9, 1.3) C7H15NO3 (1.3, 0.3) C6H14N2O3 C8H13NO4 C8H16N2O3 (2.4, 0.3) C9H20N2O2 (3.8, 0.5) C8H16N2O4 C9H21N2O3 C11H20N2O6 (4.1, 4.8) C22H36N4O13

f v v f f v f v v V f f f f f f vv f vvv vvv f

vv vvv v v v vv v VVV f VVV f f vv v f v f v f vvv f

vvv f f f f vv v f f V f f v f v f f vv f

v v v v v v f f f f V f v v v f f f f

v

f

a The arrows of increase (v, vv, and vvv) denote the ratio (RI) of MS intensities of each compound in ddY-H to ddY-L [RI ) ddY - H/ddY - L]. The arrows of decrease (V, VV, and VVV) denote the ratio (RD) of MS intensities of each compound in ddY-L to ddY-H [RD ) ddY - L/ddY - H]. f, RI or RD ) 0.8-1.2; v, RI ) 1.2-2.0; vv, RI ) 2.0-3.0; vvv, RI ) >3.0; V, RD ) 1.2-2.0; VV, RD ) 2.0-3.0; VVV, RD ) >3.0.

Table 5. Other Metabolites and Unknown Compounds, Increased and Decreased by ddY-Ha ret. time no.

ID

1 2

T3-C18 HS-F5

T3 m/z


1-Amino-2-propanol 4-aminobenzoate or Anthranilate 3 2-Hydroxyquinoline 4 2-aminooctanoic acid 5 4-heptylphenol 6 Sebacic acid 7 Bisnorbiotin 8 polyribitol phosphate 9 Heptyl 4-hydroxybenzoate 10 ubiquinol

1 1.1

1.68 1.68 7.95 7.99

17.07 146.06 C9H7NO (0.1, 0.1) 14.04 160.13 C8H17NO2 (0.7, 0.0) 193.16 C13H20O (-4.2, 3.4) 16.82 203.13 C10H18O4 (0.6, 0.9) 1.88 217.07 C8H12N2O3S (4.3, 5.6) 233.04 C5H13O8P (0.7, 5.1) 17.58 237.15 C14H20O3 (0.8, 0.7) 18.32 253.15 C14H20O4 (0.9, 0.9)

11 pyrraline 12 fructose 6-phosphate

7.68 1.68

18.15 255.16 C14H22O4 (1.4, 0.7) 261.04 C6H13O9P (-0.1, 1.1)

13 14 15 16 17 18 19 20 21 22 23 24

4-dodecylphenol Linoleyl carnitine Unknown 1 Unknown 2 Unknown 3 Unknown 4 Unknown 5 Unknown 6 Unknown 7 Unknown 8 Unknown 9 Unknown 10

1.72 1

8.9 1.67 1.67 9.34 9.71 7.65 9.33

9.80

3.88 76.08 C3H9NO (0.9, 0.0) 2.37 138.06 C7H7NO2 (1.3, 0.0)

21.85 263.24 C18H30O (1.4, 0.1) 424.34 C25H45NO4 (-1.1, 1.4) 163.06 195.09 225.15 20.82 239.16 18.5 240.16 293.11 426.36 17.56 439.2 18.79 476.28 20.47 540.37

path way

5w 9w 13w 20w 5w 9w 13w 20w

Glycine f vv Benzoate degradation v v via hydroxylation

Ubiquinone biosynthesis Pentose phosphate pathway

HS-F5

v vv

vv v

v v

v vv

v v

v v

f f vvv vv

V V vvv v

VVV VVV vvv v

VVV VVV vvv f

vvv vvv f V VVV

VVV

f v vvv f f vv

v v

v v

v v

vv v

v v

V f f vvv vvv

VVV V V vvv vvv

f V V vvv vvv

f VVV VVV vvv vvv

f

f

vvv vvv vvv VVV VVV VVV

f VVV

vvv vvv vvv VVV VVV VVV v vv vv

vvv VVV f

vvv vvv

v VVV f f vvv vvv

VVV V V vvv vvv

vv vv VVV VV

v

vv

VVV VVV VVV vv vvv

VVV VVV VVV

f V

f V

v

f

f

vv

a The arrows of increase (v, vv, and vvv) denote the ratio (RI) of MS intensities of each compound in ddY-H to ddY-L [RI ) ddY - H/ddY - L]. The arrows of decrease (V, VV, and VVV) denote the ratio (RD) of MS intensities of each compound in ddY-L to ddY-H [RD ) ddY - L/ddY - H]. f, RI or RD ) 0.8-1.2; v, RI ) 1.2-2.0; vv, RI ) 2.0-3.0; vvv, RI ) >3.0; V, RD ) 1.2-2.0; VV, RD ) 2.0-3.0; VVV, RD ) >3.0.

diabetic states, many different compounds in both groups were extracted from the S-plots of the OPLS-DA. Tables 3-5 list the compounds that increased and decreased in the ddY-H group along with the ages. The arrow denotes the increase and 3918


decrease in the metabolites compared to those in the ddY-L group mice. The amount of each compound variously changed at the age. The determination of exact amounts is essentially very difficult by metabonomic study, because large number of


research articles

Figure 5. Metabolite pathway map of Lysine biosynthesis. Arrow denotes increase (v) and decrease (V) in the metabolite. The large arrow also means a significant difference in the increase and decrease. (O) denotes the detected metabolites.

the compounds is detected on each chromatogram, and the peaks are very crowded, as the results. Thus, the fluctuation of the compounds between ddY-H and ddY-L groups is defined as low (single arrow, 1.2-2.0 times), medium (double arrows, 2.0-3.0 times) and high (triple arrows, more than 3.0 times) in Tables 3-5. The structural identification of the biomarker candidates was carried out by the search of several databases. Only 20% less of all the compounds were fit with the compounds in the databases, such as ChemSpider and HMDS. The metabolic pathways including these compounds were studied by KEGG. Several compounds were involved in the metabolic pathway maps of the lysine biosynthesis and lysine degradation. The lists of the fit compounds on the pathways are shown in Tables 3 and 4. The concentration change in the metabolites seemed to be higher at 9 weeks than for the other weeks. The increase and decrease in these metabolites are shown in the pathway maps (Figures 5 and 6). Although the concentration of several metabolites in the pathway maps decreased or increased, the metabolites in the lysine biosynthesis pathway tended to decrease, whereas the metabolites in the lysine degradation pathway generally increased. Because the glucose concentration in the blood is high in the 13-week old mice (Table 2), the 9-week old mice are speculated to be the prediabetic state. Therefore, the concentration change in the metabolites for the lysine biosynthesis and lysine degradation pathways seems to correlate with the prediabetic state. This means that these metabolites are possibly powerful biomarkers of the prediabetic state. Although the elemental composition, mass error and i-FIT values, detected from the database search, suggested the compounds, the structures are further identified by the retention times and mass spectra of the authentic compounds. Other biomarker candidates in the diabetes mellitus are also shown

in Table 5. Several compounds were involved in the metabolic pathway maps, such as the pentose phosphate and ubiquinone biosynthesis. Furthermore, several unknown compounds, which may be the possible potential biomarker, were also detected by the metabolite profiling study. Several examples of the fluctuation are shown in Figure 7. The structures of unknown compounds could not be still identified. The amounts of the compounds detected in the plasma samples seem to be very small quantities, judging from MS intensity. Therefore, the determination using NMR after the peak fractionation may be very difficult. MS/MS analysis seems to be practical means for the structural elucidation of the unknown compounds. Thus, the detailed experiments utilizing tandem TOF-MS/MS (QTOF-MS) are planned, and the results will be reported elsewhere. As described herein, several metabolite pathways seem to correlate with the diabetic mellitus, judging from the identified compounds. Because a couple of metabolites were correlated with the prediabetic state at 9 weeks, the detection of the metabolites on the metabolite pathway maps, such as the lysine biosynthesis and lysine degradation may be useful for the diagnosis of the prediabetic state. The activity and the amount of the enzyme related to the production may also be biomarker candidates. We plan the determination of these biomarker candidates in the human plasma of normal, prediabetic state and diabetic patients. Wilson et al.42-45 report some fatty acids and phospholipids as the biomarker candidates of diabetes in Zucker rats. However, the concentration difference of these compounds was not identified in our study using ddY mice. According to the data of animal supplier, the blood levels of cholesterol, fatty acids and phospholipids increased with increase the age of Zucker rats. In contrast, no remarkable change of those levels determined from other methods (Kits of Wako Pure Chemical Journal of Proteome Research • Vol. 9, No. 8, 2010 3919

research articles

Tsutsui et al.

Figure 6. Metabolite pathway map of Lysine degradation. Arrow denotes increase (v) and decrease (V) in the metabolite. The large arrow also means a significant difference in the increase and decrease. (O) denotes the detected metabolites.

Figure 7. Fluctuation of unknown metabolites between ddY-L and ddY-H mice at 5-20 weeks. Each column plus bar represents mean ( SE (n ) 10).

Industries) was observed in plasma of ddY mice used in this study. (unpublished results). Therefore, the difference seems 3920


to be due to the animals species (Zucker rats versus ddY mice) and not conflict with the results by Wilson et al.


Conclusion In this study, we reported an analytical approach for the identification of potential biomarkers in the prediabetic and diabetic states using plasma metabolite profiling and chemometric analysis. The outline for the identification of potential biomarkers in the predisease and disease states based upon the metabonomic study was described. The detection of the predisease state in humans is generally very difficult because of lifestyle differences and ethical considerations. Therefore, animal experiments were carried out in this study. Since several biomarker candidates concerning the prediabetic state were identified in the present research, the metabolite profiling study could be helpful for understanding the abnormal state in various diseases. Further studies for the resolution of the predisease state are currently underway in our laboratory.

Acknowledgment. This work was supported in part by the Global COE Program from the Ministry of Education, Culture, Sports, Science and Technology of Japan. References (1) Nicholson, J. K.; Lindon, J. C.; Holmes, E. ‘Metabonomics’: understanding the metabolic responses of living systems to pathophysiological stimuli via multivariate statistical analysis of biological NMR spectroscopic data. Xenobiotica 1999, 29, 1181– 1189. (2) Woo, H. M.; Kim, K. M.; Choi, M. H.; Jung, B. H.; Lee, J.; Kong, G.; Nam, S. J.; Kim, S.; Bai, S. W.; Chung, B. C. Mass spectrometry based metabolomic approaches in urinary biomarker study of women’s cancers. Clin. Chim. Acta 2009, 400, 63–69. (3) Dettmer, K.; Arnov, P. A.; Hammock, B. D. Mass spectrometrybased metabolomics. Mass Spectom. Rev. 2007, 26, 51–78. (4) Lenz, E. M.; Wilson, I. D. Analytical strategies in metabonomics. J. Proteome Res. 2007, 6, 443–458. (5) Lenz, E. M.; Bright, J.; Knight, R.; Wilson, I. D.; Major, H. A metabonomic investigation of the biochemical effects of mercuric chloride in the rat using H-1 NMR and HPLC-TOF/MS: time dependant changes in the urinary profile of endogenous metabolites as a result of nephrotoxicity. Analyst 2004, 129, 535–541. (6) Idborg-Bjo¨rkman, H.; Edlund, P.-O.; Kvalheim, O. M.; SchuppeKoistinen, I.; Jacobsson, S. P. Screening of biomarkers in rat urine using LC/electrospray ionization-MS and two-way data analysis. Anal. Chem. 2003, 75, 4784–4792. (7) Plumb, R. S.; Granger, J. H.; Stumpf, C. L.; Wilson, I. D.; Evans, J. A.; Lenz, E. M. Metabonomic analysis of mouse urine by liquidchromatography-time of flight mass spectrometry (LC-TOFMS): detection of strain, diurnal and gender differences. Analyst 2003, 128, 819–823. (8) Plumb, R. S.; Granger, J. H.; Stumpf, C. L.; Johnson, K. A.; Smith, B. W.; Gaulitz, S.; Wilson, I. D.; Castro-Perez, J. A rapid screening approach to metabonomics using UPLC and oa-TOF mass spectrometry: application to age, gender and diurnal variation in normal/Zucker obese rats and black, white and nude mice. Analyst 2005, 130, 844–849. (9) Hodson, M. P.; Dear, G. J.; Roberts, A. D.; Haylock, C. L.; Ball, R. J.; Plumb, R. S.; Stumpf, C. L.; Griffin, J. L.; Haselden, J. N. A genderspecific discriminator in Sprague-Dawley rat urine: The deployment of a metabolic profiling strategy for biomarker discovery and identification. Anal. Biochem. 2007, 362, 182–192. (10) Wilson, I. D.; Plumb, R.; Granger, J.; Major, H.; Williams, R.; Lenz, E. M. HPLC-MS-based methods for the study of metabonomics. J. Chromatogr., B 2005, 817, 67–76. (11) Govorukhina, N. I.; Reijmers, T. H.; Nyangoma, S. O.; Van der Zee, A. G. J.; Jansen, R. C.; Bischoff, R. Analysis of human serum by liquid chromatography-mass spectrometry: Improved sample preparation and data analysis. J. Chromatogr. A 2006, 1120, 142– 150. (12) Bijlsma, S.; Bobeldijk, I.; Verheij, E. R.; Ramaker, R.; Kochhar, S.; Macdonald, I. A.; Van Ommen, B.; Smilde, A. K. Large-scale human metabolomics studies: A strategy for data (pre-) processing and validation. Anal. Chem. 2006, 78, 567–574. (13) Yang, J.; Xu, G.; Zheng, Y.; Kong, H.; Pang, T.; Lv, S.; Yang, Q. Diagnosis of liver cancer using HPLC-based metabonomics avoiding false-positive result from hepatitis and hepatocirrhosis diseases. J. Chromatogr., B 2004, 813, 59–65.

research articles (14) Idborg, H.; Zamani, L.; Edlund, P.-O.; Schuppe-Koistinen, I.; Jacobsson, S. P. Metabolic fingerprinting of rat urine by LC/MS Part 1. Analysis by hydrophilic interaction liquid chromatographyelectrospray ionization mass spectrometry. J. Chromatogr., B 2005, 828, 9–13. (15) Idborg, H.; Zamani, L.; Edlund, P.-O.; Schuppe-Koistinen, I.; Jacobsson, S. P. Metabolic fingerprinting of rat urine by LC/MS Part 2. Data pretreatment methods for handling of complex data. J. Chromatogr., B 2005, 828, 14–20. (16) Dunn, W. B.; Ellis, D. Metabolomics: Current analytical platforms and methodologies. TrAC Trends Anal. Chem. 2005, 24, 285–294. (17) Lindon, J. C.; Holmes, E.; Nicholson, J. K. Pattern recognition methods and applications in biomedical magnetic resonance. Prog. NMR Spectrosc. 2001, 39, 1–40. (18) Nicholson, J. K.; Connelly, J.; Lindon, J. C.; Holmes, E. Metabonomics: a platform for studying drug toxicity and gene function. Nature Rev. Drug. Discovery 2002, 1, 153–161. (19) Hollywood, K.; Brison, D. R.; Goodacre, R. Metabolomics: Current technologies and future trends. Proteomics 2006, 6, 4716–4723. (20) Jordan, K. W.; Cheng, L. L. NMR-based metabolomics approach to target biomarkers for human prostate cancer. Expert Rev. Proteomics 2007, 4, 389–400. (21) Reo, N. V. NMR-based metabolomics. Drug Chem. Toxicol. 2002, 25, 375–382. (22) Toyo’oka, T. Determination methods for biologically active compounds by ultra-performance liquid chromatography coupled with mass spectrometry: Application to the analyses of pharmaceuticals, foods, plants, environments, metabonomics, and metabolomics. J. Chromatogr. Sci. 2008, 46, 233–247. (23) Ma, S. G.; Chowdhury, S. K.; Alton, K. B. Application of mass spectrometry for metabolite identification. Curr. Drug Metab. 2006, 7, 503–523. (24) Inagaki, S.; Noda, T.; Min, J. Z.; Toyo’oka, T. Rapid determination of histamine and its metabolites in mice hair by ultra-performance liquid chromatography with time-of-flight mass spectrometry. J. Chromatogr., A 2007, 1176, 94–99. (25) Yuan, K. L.; Kong, H. W.; Guan, Y. F.; Yang, J.; Xu, G. X. A GCbased metabonomics investigation of type 2 diabetes by organic acids metabolic profile. J. Chromatogr., B 2007, 850, 236–240. (26) Benavente, F.; van den Heijden, R.; Tjaden, U. R.; van den Greef, J.; Hankemeier, T. Metabolite profiling of human urine by CEESI-MS using separation electrolytes at low pH. Electrophoresis 2006, 27, 4570–4584. (27) Chen, J.; Zhao, X.; Fritsche, J.; Yin, P.; Schmitt-Kopplin, P.; Wang, W.; Lu, X.; Haring, H. U.; Schleicher, E. D.; Lehmann, R.; Xu, G. Practical approach for the identification and isomer elucidation of biomarkers detected in a metabonomic study for the discovery of individuals at risk for diabetes by integrating the chromatographic and mass spectrometric information. Anal. Chem. 2008, 80, 1280–1289. (28) Coulier, L.; Bas, R.; Jespersen, S.; Verheij, E.; van der Wert, M. J.; Hankemeier, T. Simultaneous quantitative analysis of metabolites using ion-pair liquid chromatography - Electrospray ionization mass spectrometry. Anal. Chem. 2006, 78, 6573–6582. (29) Gika, H. G.; Theodoridis, G. A.; Wilson, I. D. Hydrophilic interaction and reversed-phase ultra-performance liquid chromatography TOF-MS for metabonomic analysis of Zucker rat urine. J. Sep. Sci. 2008, 31, 1598–1608. (30) Mohamed, R.; Varesio, E.; Ivosev, G.; Burton, L.; Bonner, R.; Hopfgartner, G. Comprehensive analytical strategy for biomarker identification based on liquid chromatography coupled to mass spectrometry and new candidate confirmation tools. Anal. Chem. 2009, 81, 7677–7694. (31) Zimmet, P.; Alberti, K. G. M. M.; Shaw, J. Global and societal implications of the diabetes epidemic. Nature 2001, 414, 782–787. (32) Yoon, K. H.; Lee, J. H.; Kim, J. W.; Cho, J. H.; Choi, Y. H.; Ko, S. H.; Zimmet, P.; Son, H. Y. Epidemic obesity and type 2 diabetes in Asia. Lancet 2006, 368, 1681–1688. (33) Knowler, W. C.; Barrett-Connor, E.; Fowler, S. E.; Hamman, R. F.; Lachin, J. M.; Walker, E. A.; Nathan, D. M. Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin. N. Engl. J. Med. 2002, 346, 393–403. (34) Tuomilehto, J.; Lindstrom, J.; Eriksson, J. G.; Valie, T. T.; Hamalainen, H.; Ilanne-Parikka, P.; Keinanen-Kiukaanniemi, S.; Laakso, M.; Louheranta, A.; Rastas, M.; Salminen, V.; Unsitupa, M.; Aunola, S.; Cepaitis, Z.; Moltchanov, V.; Hakumaki, M.; Mannelin, M.; Martikkala, V.; Sundvall, J. Prevention of type 2 diabetes mellitus by changes in lifestyle among subjects with impaired glucose tolerance. N. Engl. J. Med. 2001, 344, 1343–1350. (35) Noge, I.; Kagawa, Y.; Maeda, T. A new diabetic mouse model derived from the ddY strain. Biol. Pharm. Bull. 2010, 33, 988–992.

Journal of Proteome Research • Vol. 9, No. 8, 2010 3921

research articles (36) Yin, P.; Mohemaiti, P.; Chen, J.; Zhao, X.; Lu, X.; Yimiti, A.; Upur, H.; Xu, G. Serum metabolic profiling of abnormal savda by liquid chromatography/mass spectrometry. J. Chromatogr., B 2008, 871, 322–327. (37) Spagou, K.; Tsoukali, H.; Raikos, N.; Gika, H.; Wilson, I. D.; Theodoridis, G. Hydrophilic interaction chromatography coupled to MS for metabonomic/metabolomic studies. J. Sep. Sci. 2010, 33, 716–727. (38) Lai, L.; Michopoulos, F.; Gika, H.; Theodoridis, G.; Wilkinson, R. W.; Odedra, R.; Wingate, J.; Bonner, R.; Tate, S.; Wilson, I. D. Methodological considerations in the development of HPLC-MS methods for the analysis of rodent plasma for metabonomic studies. Mol. BioSyst. 2010, 6, 108–120. (39) Gika, H. G.; Theodoridis, G.; Extance, J.; Edge, A. M.; Wilson, I. D. High temperature-ultra performance liquid chromatography-mass spectrometry for the metabolomic analysis of Zucker rat urine. J. Chromatogr., B 2008, 871, 279–287. (40) Granger, J. H.; Williams, R.; Lenz, E. M.; Plumb, R. S.; Stumpf, C. L.; Wilson, I. D. A metabonomic study of strain- and age-related differences in the Zucker rat. Rapid Commun. Mass Spectrom. 2007, 21, 2039–2045. (41) Williams, R. E.; Lenz, E. M.; Evans, J. A.; Wilson, I. D.; Granger, J. H.; Plumb, R. S.; Stumpf, C. L. A combined 1H NMR and HPLC-

3922


Tsutsui et al.

(42)

(43)

(44)

(45)

MS-based metabonomic study of urine from obese (fa/fa) Zucker and normal Wister-derived rats. J. Pharm. Biomed. Anal. 2005, 38, 465–471. Plumb, R. S.; Johnson, K. A.; Rainville, P.; Shockcor, J. P.; Williams, R.; Granger, J. H.; Wilson, I. D. The detection of phenotypic differences in the metabolic plasma profile of three strains of Zucker rats at 20 weeks of age using ultra-performance liquid chromatography/orthogonal acceleration time-of-flight mass spectrometry. Rapid Commun. Mass Spectrom. 2006, 20, 2800–2806. Williams, R.; Lenz, E. M.; Wilson, A. J.; Granger, J.; Wilson, I. D.; Major, H.; Stumpf, C.; Plumb, R. A multi-analytical platform approach to the metabonomic analysis of plasma from normal and Zucker (fa/fa) obese rats. Mol. BioSyst. 2006, 2, 174–183. Loftus, N.; Miseki, K.; Iida, J.; Gika, H. G.; Theodoridis, G.; Wilson, I. D. Profiling and biomarker identification in plasma from different Zucker rat strains via high mass accuracy multistage mass spectrometric analysis using liquid chromatography/mass spectrometry with quadrupole ion trap-time of flight mass spectrometry. Rapid Commun. Mass Spectrom. 2008, 22, 2547–2554. Major, H. J.; Williams, R.; Wilson, A. J.; Wilson, I. D. A metabonomic analysis of plasma from Zucker rat strains using gas chromatography/mass spectrometry and pattern recognition. Rapid Commun. Mass Spectrom. 2006, 20, 3295–3302.

PR100121K

Practical Analytical Approach for the Identification of Biomarker

Recommend Documents