LC–MS-Based Urinary Metabolite Signatures in Idiopathic Parkinson's

Oct 1, 2014 - The results suggest that LC–MS-based urinary metabolomic profiling can reveal the metabolite signatures and related variations in meta...
7 downloads 10 Views 3MB Size
Download PDF
Article pubs.acs.org/jpr

LC−MS-Based Urinary Metabolite Signatures in Idiopathic Parkinson’s Disease Hemi Luan,†,# Liang-Feng Liu,‡,# Nan Meng,§ Zhi Tang,† Ka-Kit Chua,‡ Lei-Lei Chen,‡ Ju-Xian Song,‡ Vincent C. T. Mok,∥ Li-Xia Xie,⊥ Min Li,*,‡ and Zongwei Cai*,† †

Department of Chemistry, Hong Kong Baptist University, Science Tower T1304, Hong Kong SAR, China School of Chinese Medicine, Hong Kong Baptist University, 7 Baptist University Road, Hong Kong SAR, China § Beijing Genomics Institute-Shenzhen (BGI-Shenzhen), Build 11, Beishan Industrial Zone, Shenzhen 518083, China ∥ Department of Medicine and Therapeutics, Faculty of Medicine, The Chinese University of Hong Kong, 9/F., Lui Che Woo Clinical Sciences Building, Hong Kong SAR, China ⊥ Department of Pharmacy, Nanshan Affiliated Hospital of Guangdong Medical College, 89 Taoyuan Road, Shenzhen 518052, China ‡

S Supporting Information *

ABSTRACT: Increasing evidence has shown that abnormal metabolic phenotypes in body fluids reflect the pathogenesis and pathophysiology of Parkinson’s disease (PD). These body fluids include urine; however, the relationship between, specifically, urinary metabolic phenotypes and PD is not fully understood. In this study, urinary metabolites from a total of 401 clinical urine samples collected from 106 idiopathic PD patients and 104 normal control subjects were profiled by using high-performance liquid chromatography coupled to high-resolution mass spectrometry. Our study revealed significant correlation between clinical phenotype and urinary metabolite profile. Metabolic profiles of idiopathic PD patients differed significantly and consistently from normal controls, with related metabolic pathway variations observed in steroidogenesis, fatty acid beta-oxidation, histidine metabolism, phenylalanine metabolism, tryptophan metabolism, nucleotide metabolism, and tyrosine metabolism. In the fruit fly Drosophila melanogaster, the alteration of the kynurenine pathway in tryptophan metabolism corresponded with pathogenic changes in the alpha-synuclein overexpressed Drosophila model of PD. The results suggest that LC−MS-based urinary metabolomic profiling can reveal the metabolite signatures and related variations in metabolic pathways that characterize PD. Consistent PD-related changes across species may provide the basis for understanding metabolic regulation of PD at the molecular level. KEYWORDS: metabolite signatures, idiopathic Parkinson’s disease, urinary metabolomics, metabolic phenotype, drosophila

■

INTRODUCTION Parkinson’s disease (PD) is one of the most prevalent debilitating neurodegenerative disorders in the world afflicting ∼1% of people over 60 years old.1 In China, almost 2 million people currently suffer from PD, and the prevalence of PD is predicted to increase to 4.94% by 2030.2,3 The histopathological hallmark of PD is the significant loss of dopaminergic neurons in the substantia nigra pars compacta (SNpc) associated with the accumulation of alpha synuclein aggregates, which are the main component of Lewy bodies (LBs). The loss of dopaminergic neurons can result in a series of motor deficiencies including bradykinesia, resting tremor, and rigidity.4 In addition to the prominent motor symptoms, nonmotor symptoms such as gastrointestinal dysfunction, depression, anxiety, and sleep disorders5 also exist. Most of the nonmotor symptoms cannot be alleviated by dopamine replacement therapies. This is presumably because these symptoms are less correlated with loss of dopaminergic neurons in the SNpc.5 Such a broad array of symptoms © 2014 American Chemical Society

necessarily involves organs throughout the body. Thus, PD is increasingly recognized as a multisystem disorder.4 The pathogenesis of PD remains elusive, although great progress has been made during the past two decades. Undoubtedly, aging, which is the main risk factor of PD, is likely to contribute to the pathogenesis of PD by natural and chronic changes throughout the body. During the process of aging, alterations in energy metabolism, oxidative stress, inflammation, and corticosteroid signaling occur that could contribute to the onset of PD.6 Besides the main risk factor, aging, other risk factors such as mutation in SNCA (the alpha-synuclein gene) and exposure to environmental toxins are also linked to metabolic abnormalities involving neurotransmitter systems (dopamine, serotonin, GABA, and glutamate), fatty acids such as Special Issue: Environmental Impact on Health Received: August 7, 2014 Published: October 1, 2014 467

dx.doi.org/10.1021/pr500807t | J. Proteome Res. 2015, 14, 467−478

Journal of Proteome Research

Article

Table 1. Demographics and Clinical Measurements of Study Subjectsa idiopathic PD patients characteristic

normal controls

Gender female 61 male 43 Age (years) mean (sd) 59.8 (6.0) range 41−80 Hoehn and Yahr (H & Y) Score mean (sd) range UPDRS Motor Score mean (sd) range

stage 1 (S1)

stage 2 (S2)

stage 3 (S3)

40 66

37 56

44 54

63.1 (9.3)b 40−80

62.2 (9.2) 40−79

62.4 (9.4) 40−80

2.04 (0.63) 1−4

2.02 (0.61) 1−4

1.98 (0.64) 1−4

9.59 (6.95) 1−37

8.69 (6.18) 1−27

10.93 (7.88) 1−33

a

sd, standard deviation; UPDRS motor score, unified Parkinson’s disease rating scale motor score; S1, S2, and S3 denote urine samples collected at week 0, week 16, and week 32, respectively. bP < 0.05 versus control group. Statistical significance was determined using unpaired two-tailed Student’s t test.

subjects were recruited. Each idiopathic PD patient was scheduled for a visit in week 0 (PD S1), week 16 (PD S2), and week 32 (PD S3). Urine samples were collected (PD S1, 106 samples; PD S2, 93 samples; PD S3, 98 samples) at these three visits for diminishing the intraindividual variation. According to the written consent, patients were free to drop out at any stage. Some of the patients quit the study, resulting in different number of subjects at the three stages. After overnight fasting, morning midstream urine was collected, in a polypropylene container, then aliquoted into Eppendorf tubes and stored at −80 °C. Assessment with H&Y scale and UPDRS was also performed. Characteristics of the study subjects are shown in Table 1.

arachidonic acid-cascade, oxidative stress, and mitochondrial function.7−10 Several studies have explored the relationship between metabolic anomalies and the clinical symptoms of PD. They suggest that disturbances in the metabolic pathways related to oxidative stress, energy metabolism, and neurotransmitters are associated with PD.11−15 However, several limitations, such as lack of statistical power as a result of small sample size and inadequate number of analytes, prevent drawing definitive conclusions. Moreover, most of these studies focused on cerebrospinal fluid (CSF) and blood, which require invasive sampling.10,11,14−17 Urine, which contains most of the body’s metabolic end products, has seldom been studied with high-performance liquid chromatography coupled to highresolution mass spectrometry (LC−MS)-based metabolomics for PD research. In our study, we utilized LC−MS-based urinary metabolomics to investigate the urinary metabolite signatures in PD. More than 400 urine samples collected from 106 idiopathic PD patients and 104 normal control subjects were analyzed (Table 1). Forty-five altered metabolites were detected, derived from eight specific metabolic pathways (Table 2).

■

Urine Sample Preparation and Metabolic Profiling Spectral Acquisition

Urine samples were prepared and analyzed by LC−MS as previously reported, with slight modifications as follows.19,20 Briefly, the urine samples were thawed at room temperature. 100 μL of each thawed urine sample was precipitated by 100 μL of methanol. The mixture was then centrifuged under 14 000g for 10 min at 4 °C, and the supernatant was transferred to an autosampler vial. Each 10 μL aliquot of the supernatant was injected into the LC−MS. The sample preparation order was randomized, and sample analysis order was rerandomized in each analytical batch and across all batches. To ensure that data was of comparable high quality, a quality assurance strategy based on the periodic analysis of pooled samples (quality control or QC sample) together with the real samples was performed.21 The QC sample was prepared by mixing equal volumes of urine from PD patients and controls before sample preparation as they were aliquoted for analysis. This QC sample was utilized to estimate a “mean” profile representing all analytes encountered during the analysis. Metabolites in the urine were analyzed by LC−MS based on our previously published method with minor modifications.22 MS data was acquired using a Shimadzu Prominence LC system (Shimadzu) coupled online to an LTQ Orbitrap Velos instrument (Thermo Fisher Scientific, Waltham, MA) set at 30 000 resolution (at m/ z 400). Sample analysis was carried out under both positive and negative ion modes. The mass scanning range was 50−1000 m/ z and the capillary temperature was 350 °C. Nitrogen sheath gas was set at a flow rate of 30 L/min. Nitrogen auxiliary gas

MATERIALS AND METHODS

Clinical Samples Collection

Recruitment of subjects and sample collection was performed at the Hong Kong Baptist University Chinese Medicine Specialty Centre. The study was approved by the Ethics Committee of the Hong Kong Baptist University’s Institutional Review Board (code: HASC/09-10/09) with written information provided to and informed consent obtained from all subjects. Idiopathic PD patients were diagnosed by a movement disorder specialist according to the United Kingdom Parkinson’s Disease Brain Bank (UKPDBB) criteria.18 The inclusion criteria were UKPDBB clinical diagnostic criteria, stable treatment with levodopa, Hoehn and Yahr (H&Y) scale rating from 1 to 4, and normal liver and renal function. Exclusion criteria were one or more of the following: atypical or secondary Parkinsonism, use of antidepressants, Mini-Mental State Examination (MMSE) < 24, history of psychosis, or severe suicidal tendency. The volunteers without neurological or psychiatric problems were recruited as normal controls. A total of 234 patients were screened; finally, 106 patients together with 104 normal control 468

dx.doi.org/10.1021/pr500807t | J. Proteome Res. 2015, 14, 467−478

Journal of Proteome Research

Article

Table 2. List of Altered Metabolites between Idiopathic PD Patients (S1, S2, and S3) and Normal Controls PD S1 vs controls Steroidogenesis dihydrocortisolb 21-deoxycortisolb 11-deoxycortisolb hydroxyprogesterone dehydrocorticosterone cortisolb Tryptophan Metabolism acetylserotonin hydroxyanthranilic acidb tryptamineb 5-hydroxykynurenamine xanthurenic acidb 5-hydroxytryptophanb indolelactic acidb kynurenineb Glycine Derivatives hexanoylglycineb hydroxyphenylacetylglycine phenylacetylglycineb furoylglycineb tiglylglycineb glycineb Phenylalanine Metabolism phenylacetylglutamineb acetylphenylalanineb phenylacetic acid Tyrosine Metabolism hydroxyphenylacetic acidb glutamyltyrosine acetyltyrosine 5,6-dihydroxyindole Nucleotide Metabolism deoxyinosineb hypoxanthineb adenineb Fatty Acid Beta-Oxidation hydroxylauroylcarnitine malonylcarnitineb hydroxyhexanoycarnitine Histidine Metabolism histidineb urocanic acidb imidazoleacetic acidb Others porphobilinogenb acetylarginine phenylbutyric acidb spermidineb naphthol cholineb aminobutyric acidb trimethylamine N-oxideb pyridoxic acidb

adjusted P value

MDA

FC

1.59 0.76 1.00 2.01 1.00 2.94

1.97 1.55 1.72 2.23 0.64 2.23

1.83 1.12 9.54 7.99 1.05 1.27

× × × × × ×

0.43 1.00 2.57 1.63 1.22 0.00 3.23 1.05

1.96 1.94 2.74 1.98 1.80 2.62 3.36 3.15

2.09 8.44 2.26 1.46 5.18 9.19 2.15 2.15

1.41 2.00 1.75 0.00 2.95 2.32

1.67 2.08 2.20 3.28 2.18 2.74

1.33 4.88 1.50 1.50 1.16 1.04

2.07 2.68 3.49

2.02 3.21 2.36

1.84 3.18 1.93 1.88

2.16 16.86 2.45 2.17

0.00 0.00 0.00

2.16 1.77 1.67

1.00 0.73 1.00

PD S2 vs controls

PD S3 vs controls

adjusted P value

AUC

MDA

FC

10−9 10−6 10−5 10−11 10−4 10−12

0.77 0.79 0.70 0.83 0.68 0.83

1.77 0.00 1.50 1.85 1.00 0.68

2.03 1.82 1.92 2.52 0.72 2.01

4.07 1.08 2.25 4.43 2.01 1.86

× × × × × ×

× × × × × × × ×

10−9 10−7 10−14 10−9 10−7 10−11 10−16 10−16

0.79 0.79 0.83 0.78 0.75 0.83 0.84 0.82

0.51 1.46 3.54 0.00 1.91 1.63 1.66 2.13

2.28 2.29 3.31 2.19 2.08 3.05 3.31 3.27

1.24 5.22 2.68 2.13 9.60 5.18 1.39 1.39

× × × × × ×

10−5 10−12 10−15 10−15 10−12 10−8

0.65 0.84 0.81 0.80 0.82 0.78

0.00 1.79 2.55 1.54 0.69 1.00

1.70 2.76 2.96 4.34 1.89 1.97

7.06 2.16 9.28 9.28 3.56 2.51

2.79 × 10−9 9.28 × 10−16 1.50 × 10−15

0.78 0.88 0.86

2.02 1.46 3.12

10−11 10−11 10−13 10−11

0.79 0.84 0.83 0.82

3.53 × 10−7 2.26 × 10−6 1.21 × 10−6

1.86 2.02 1.86

1.00 1.96 2.12 0.00 1.27 1.00 0.73 1.40 1.00 1.93 0.78 0.06

adjusted P value

AUC

MDA

FC

10−11 10−9 10−7 10−12 10−2 10−10

0.79 0.83 0.75 0.84 0.59 0.80

0.21 1.61 1.00 0.77 0.76 0.29

1.74 1.52 1.75 2.34 0.78 1.89

6.09 8.40 2.29 1.95 6.68 5.39

× × × × × ×

10−7 10−7 10−4 10−8 10−3 10−9

0.73 0.80 0.68 0.79 0.61 0.77

× × × × × × × ×

10−10 10−9 10−16 10−10 10−10 10−14 10−13 10−13

0.80 0.83 0.85 0.78 0.79 0.86 0.81 0.84

1.00 1.57 1.00 1.34 1.63 1.55 1.95 2.53

1.94 1.96 2.22 1.90 1.61 2.40 2.43 3.01

4.25 2.27 1.49 2.93 3.19 8.33 1.35 1.35

× × × × × × × ×

10−8 10−6 10−9 10−7 10−5 10−11 10−11 10−11

0.77 0.80 0.73 0.73 0.79 0.84 0.74 0.79

× × × × × ×

10−6 10−14 10−17 10−17 10−10 10−5

0.70 0.86 0.85 0.86 0.85 0.71

0.00 1.00 2.38 1.00 2.26 1.50

1.58 2.04 2.19 3.25 1.74 1.42

3.92 3.02 9.44 9.44 3.04 1.48

× × × × × ×

10−5 10−10 10−15 10−15 10−8 10−2

0.63 0.81 0.80 0.82 0.73 0.67

2.25 3.65 2.71

3.61 × 10−10 5.96 × 10−15 9.28 × 10−17

0.78 0.86 0.87

1.15 4.18 1.89

1.93 3.38 2.45

3.84 × 10−9 1.41 × 10−16 9.44 × 10−15

0.75 0.86 0.80

1.41 3.38 1.67 0.94

2.08 17.0 2.85 2.48

5.26 5.22 4.95 3.69

10−10 10−15 10−15 10−13

0.76 0.86 0.86 0.86

1.75 5.21 2.19 1.70

1.96 5.18 2.48 2.20

3.22 4.30 3.91 6.21

10−9 10−12 10−13 10−10

0.73 0.79 0.81 0.80

0.72 0.72 0.74

0.81 1.28 0.00

2.44 2.24 1.92

1.00 × 10−9 1.67 × 10−10 5.76 × 10−9

0.77 0.78 0.76

1.09 1.31 0.00

1.93 1.91 1.67

1.62 × 10−6 1.07 × 10−7 1.93 × 10−6

0.68 0.72 0.72

4.29 × 10−4 1.68 × 10−10 4.29 × 10−4

0.65 0.80 0.69

0.00 1.00 0.00

2.47 2.32 2.47

1.86 × 10−6 7.09 × 10−11 1.86 × 10−6

0.70 0.79 0.70

0.00 1.92 0.00

1.95 2.00 1.40

4.52 × 10−4 5.33 × 10−9 4.52 × 10−4

0.63 0.76 0.66

1.76 2.06 1.60

1.32 × 10−5 2.94 × 10−10 1.46 × 10−9

0.71 0.80 0.84

0.72 1.00 1.62

2.44 2.25 1.68

2.56 × 10−9 5.59 × 10−11 9.07 × 10−11

0.76 0.80 0.81

1.41 1.00 1.51

1.87 1.71 1.70

3.32 × 10−6 5.81 × 10−7 6.84 × 10−9

0.72 0.74 0.74

1.58 1.41 1.81 2.77 2.67 1.38 3.67 2.37 1.96

5.10 2.54 4.05 4.05 7.60 6.12 1.46 3.40 5.78

10−5 10−4 10−9 10−9 10−10 10−6 10−9 10−10 10−6

0.70 0.68 0.79 0.72 0.82 0.71 0.78 0.83 0.75

0.00 0.41 0.99 0.62 1.37 0.00 3.00 1.00 0.00

1.68 1.91 1.72 2.67 2.76 1.61 4.71 3.05 3.28

4.53 2.20 2.53 2.53 1.12 9.39 1.93 4.09 7.46

10−5 10−7 10−8 10−8 10−8 10−7 10−11 10−11 10−10

0.70 0.73 0.78 0.72 0.79 0.71 0.79 0.84 0.82

1.39 1.00 0.08 0.32 1.00 0.97 1.82 1.41 1.35

1.41 1.30 1.54 2.20 2.29 1.47 3.53 2.31 2.52

2.10 5.19 1.50 1.50 6.47 4.64 4.98 1.33 4.23

10−3 10−3 10−7 10−7 10−8 10−6 10−8 10−8 10−6

0.66 0.64 0.77 0.64 0.78 0.70 0.74 0.79 0.95

4.21 1.41 2.89 9.99

× × × ×

× × × × × × × × ×

× × × ×

× × × × × × × × ×

× × × ×

× × × × × × × × ×

AUC

a

MDA: Mean decrease in accuracy. Adjusted P value: Adjusted for multiple testing (Wilcoxon−Mann U test) with FDR control. FC: Fold change. FC with a value >1.2 indicates a relatively higher concentration present in PD group, while a value WT α-synuclein was achieved by crossing elav-GAL4 with UAS-WT α-synuclein. One day’s male flies were collected and fed with standard corn meal for 35 days. On the 35th day, flies were snap-frozen and stored at −80 °C for further analysis.

Data Analysis

The LC−MS data was analyzed following our previously published method.22,23 Data pretreatment including feature detection, retention time correction, alignment, and annotation of isotope peaks was achieved using XCMS and CAMERA software implemented with the freely available R statistical language (v 2.13.1).24,25 To obtain consistent variables, the resulting matrix was further reduced by removing peaks with >80% missing values (those with ion intensity = 0) and with isotope ions from each group. The Quality Control−Robust Loess Signal Correction (QC-RLSC) algorithm21 for signal correction and integration of data from six analytical batches was also used. Each retained peak was normalized to the QC sample using QC-RLSC algorithm. A threshold of 30% was set for the relative standard deviation (RSD) values of metabolites in the QC samples, which was accepted for the assessment of repeatability in metabolomics data sets.26 After the Pareto scaling (mean centering and scaled to square root of variance), the filtered matrix was exported for multivariate statistical analysis using orthogonal projections to latent structures discriminant analysis (OPLS-DA) and random forest (RF) discriminant analysis. OPLS-DA, a supervised method, was used to obtain an overview of the complete data set and generate the maximum separation between the classes of normal controls versus PD S1, PD S2, and PD S3 groups. The RF machine learning algorithm was employed to classify urine samples as either normal controls or one of the three PD groups: PD S1, PD S2, and PD S3. RF was chosen because it is a high sensitive classifier with little or no over fitting.27 The tree number of the forest in this study was set to be 1000. Permutation-based mean decrease in accuracy (MDA) was used to measure the importance of each metabolite to the classification. For each metabolite peak reproducibly detected in whole samples, the nonparametric univariate method, Mann−Whitney U test, was applied to measure the significance of each metabolite, with results adjusted for multiple testing using false discovery rates (FDR) correction. Compound annotation was performed by comparing the MS/MS spectra and retention time of commercially available standard compounds or the accurate mass of compounds obtained from the Human Metabolome Database (www.hmdb. ca).

Protein Extraction and Western Blotting Analysis

Twenty fly heads were placed in sample tubes and then mixed with 50 μL of RIPA buffer. A precooled 7 mm stainless-steel bead was added to each tube. The samples were then homogenized using the Tissue Lyser LT (Qiagen, Germany) at 50 Hz for 2 min. Following homogenization, samples were centrifuged for 5 min (16 000g, 4 °C) to pellet the beads and tissue fragment. The supernatant was collected and cleared by an additional centrifugation for 10 min (16 000g, 4 °C). Protein concentration was quantified by BioRad BCA Reagent (BioRad Laboratories, Germany). For Western blot analysis, protein samples were diluted in loading buffer (5 × Laemmli buffer) and boiled (95 °C, 5 min), then loaded and separated with 15% SDS-PAGE gel (100 V, 2 h) and subsequent transfer onto PVDF membrane (300 mA, 90 min). The membranes were then blocked with skim milk (1 h, 5% skim milk in TBST: 0.05 M TBS containing 0.1% Tween-20). After blocking, membranes were incubated overnight at 4 °C with primary antibodies. The following primary antibodies were used in the study: mouse anti-α-synuclein (1:1000; BD Transduction Laboratories, USA), rabbit anti-tyrosine hydroxylase (1:1000; Millipore Chemicon, USA) and mouse anti-β-actin (1:1,000; Abcam, U.K.). Multistage Mass Spectrometry Analysis

Ten flies were homogenized in 100 μL of water containing 5% trichloroacetic acid and centrifuged at 16 000g for 5 min (4 °C). 80 μL of supernatant was precipitated by 160 μL of methanol. After centrifuging for 10 min at 16 000g at 4 °C, the supernatant was transferred to a 1.5 mL polypropylene tube and stored at −20 °C until analysis. For semiquantification of kynurenine (KYN) and kynurenic acid (KYNA) in multiplereaction monitoring (MRM) mode using positive electrospray ionization (ESI), a Waters ACQUITY ultra performance liquid chromatography (UPLC) system (Waters, Milford, MA) coupled to a Waters TQD triple quadrupole tandem mass spectrometer (Waters, Manchester, U.K.) was used. Metabolites were separated on a hydrophilic interaction liquid chromatography column (ZIC-HILIC, 5 μm particle size, 100 mm × 2.1 mm). The flow rate remained constant at 0.4 mL/ min. The mobile phases consisted of acetonitrile/water (50:50

■

RESULTS

Metabolic Phenotype of Urine from Idiopathic PD Patients

We carried out an LC−MS-based metabolomics analysis of 401 urine samples from 106 idiopathic PD patients (106 samples at PD S1, 93 samples at PD S2, 98 samples at PD S3) and 104 470

dx.doi.org/10.1021/pr500807t | J. Proteome Res. 2015, 14, 467−478

Journal of Proteome Research

Article

were significant differences between the first component (P[1]) scores of PD groups and normal controls, which demonstrated that diversity of urinary metabolite levels existed between PD groups and normal control groups (P < 0.0001, Mann− Whitney U test). In addition, a permutation test (n = 999) was also performed for the OPLS-DA model (Figure S2 in the Supporting Information). At the same time, no clear separation was found among the cohort samples of the three PD groups. This indicates that the urinary metabolic phenotype of PD patients’ urine was stable.

normal control subjects who were free of PD or other neurodegenerative diseases. Using the LC−MS analysis protocol and subsequent processes, we found 2790 (75.49%) retention time-exact mass pairs remaining in each sample profile. Figure S1 in the Supporting Information shows that no drift was observed in the principal component analysis (PCA) scores plot representation of QC samples. Thus, the metabolic features demonstrated acceptable reproducibility and stability. We used permutation multivariate analysis of variance (PERMANOVA) to test for association of clinical phenotypes, that is, Group, H&Y grade, unified Parkinson’s disease rating scale (UPDRS) motor score, age, gender, and metabolite profile. The indicators of PD or not (Group) and the severity of the PD disease (H&Y grade and UPDRS) showed marginally significant correlation with urinary metabolomics (P < 0.05; Supplementary Table S1 in the Supporting Information). In the OPLS-DA scores plot (R2X = 0.27, R2Y = 0.84, Q2 = 0.66), the normal controls are clearly separated from the PD S1, PD S2, and PD S3 groups, which clustered together (Figure 1). There

Abnormal Metabolites Associated with PD

Supervised analysis of the urinary metabolite profiles using the RF model yielded 94.8% classification accuracy for PD S1 versus controls, 95.4% classification accuracy for PD S2 versus controls, and 92.0% classification accuracy for PD S3 versus controls (Table 2 and Figure 2). Altogether 14 subjects from PD groups (13%) were misclassified in the three RF models, which might be associated with a clinical misdiagnosis rate of at least 10%.28 At the return visits 6 months later, the abovementioned subjects showed common clinical presentations mild disease or a relatively isolated tremor. The RF model produced a ranked list of metabolites responsible for the separation between each class of samples analyzed. Higher values of MDA indicate metabolites that are more important to the classification. Urinary metabolites passing the MDA threshold (MDA ≥ 0) in this model and the Mann−Whitney U test (P < 0.05) after FDR correction were annotated to discriminate these two groups. The three independent RF models showed a distinct separation between the metabolite profiles of matched pair groups (Figure 2A−C). In total, 45 metabolites that were determinative for the separation were identified. In addition, the AUC values and fold changes in each selected metabolite were also calculated for evaluating classification test accuracy (Table 2). The values of AUC range from 0.59 to 0.95 indicated the potential capacity of these urinary metabolites for distinguishing idiopathic PD patients from normal control subjects. A list of metabolic pathways related to the metabolites found, including steroidogenesis, tryptophan metabolism, phenylalanine metabolism, tyrosine metabolism, nucleotide metabolism, fatty acid beta-oxidation, and histidine metabolism, are also shown in Table 2.

Figure 1. Classification from OPLS-DA model of metabolites profiles in urine of idiopathic PD patients and normal controls. Scatter plot using two latent variables of OPLS-DA model (R2X = 0.27, R2Y = 0.84, Q2 = 0.66).

Figure 2. Random forest analyses of urine metabolite profiles from three stages of idiopathic PD patients and normal controls (Green, idiopathic PD patients; Brown, controls). (A) PD S1 versus controls. (B) PD S2 versus controls. (C) PD S3 versus controls. Several idiopathic PD patients misclassified as controls were labeled with ‘P’. 471

dx.doi.org/10.1021/pr500807t | J. Proteome Res. 2015, 14, 467−478

Journal of Proteome Research

Article

Figure 3. Correlation and pathway analysis of abnormal metabolites. Correlation matrix for urine metabolite levels (bottom left). Age- and genderdjusted Spearman correlation coefficients for metabolite values in idiopathic PD patients. Metabolome view plot (top right) shows the metabolic pathway impact on the x axis and the pathway enrichment analysis on y axis, which indicates the key nodes in metabolic pathways that have been significantly altered with PD.

versus controls, respectively. The first latent factor of RF models has shown that idiopathic PD patients could be statistically significantly distinguished from controls (P < 0.05, Mann−Whitney U test). An MDA score list of metabolites responsible for the separation between each class (PD S1, PD S2n and PD S3) is shown in Figure 4A. Furoylglycine (13.1) had the highest mean MDA score of the three RF models, followed by glutamyltyrosine (12.1), imidazoleacetic acid (12.1), hydroxyprogesterone (11.6), dihydrocortisol (11.1), and hydroxyphenylacetic acid (11.0). Coupling the receiver operating characteristic curve to its AUC is a widely used method to estimate the diagnostic potential of a classifier in clinical applications. The AUC values of 0.92, 0.92, and 0.89 for the three RF models (PD S1 versus controls, PD S2 versus controls, and PD S3 versus controls, respectively), indicates high predictive ability for idiopathic PD patients and normal control subjects (Figure 4B).

We assessed the correlations between concentrations of metabolites in idiopathic PD patients (Figure 3, bottom left). Mean correlations within groups of related molecules were highest for metabolites in histidine metabolism (age- and gender-adjusted r = 0.53, P < 0.001), steroidogenesis (r = 0.47, P < 0.001), phenylalanine metabolism (r = 0.47, P < 0.05), fatty acid beta-oxidation (r = 0.43, P < 0.001), tryptophan metabolism (r = 0.38, P < 0.05), nucleotide metabolism (r = 0.28, P < 0.05), and tyrosine metabolism (r = 0.17, P < 0.05). Additionally, we performed metabolome view analysis by using MetaboAnalyst 2.029 to analyze which pathways were altered with PD, confirming that tryptophan metabolism was significantly disturbed with a P value of 0.1 on the x axis in Metabolome view plot (Figure 3, top right). Selected Metabolite Combinations and Potential Prediction of PD

Altered Tryptophan Metabolism in Alpha-Synuclein Overexpressed Fly

We examined how well 45 selected metabolites distinguished between idiopathic PD patients and normal control subjects. RF models yielded 87.3, 89.4, and 82.0% classification accuracy for PD S1 versus controls, PD S2 versus controls, and PD S3

Tryptophan metabolism in fruit fly is highly conserved and relevant to human biology and disease.30,31 To determine whether the tryptophan metabolism variation in idiopathic PD 472

dx.doi.org/10.1021/pr500807t | J. Proteome Res. 2015, 14, 467−478

Journal of Proteome Research

Article

Figure 4. Identified metabolite combinations that distinguish between idiopathic PD patients and normal controls. (A) Random forest importance measure was used to rank metabolites according to their importance for classification. (B) Receiver operating characteristic curve coupled to its area under the curve (AUC) for three random forest models.

Figure 5. Altered tryptophan metabolism in alpha-synuclein overexpressed fly. (A) Wild-type alpha-synuclein is overexpressed in fly model of PD. (B) Dopaminergic neuron loss in the fly model of PD (n = 3; asterisks denotes P < 0.05). (C) Increased level of KYN in alpha-synuclein overexpressed fly contrast with the GAL4 control at 1 day and 35 days of age. (n = 3; asterisks denotes P < 0.05). (D) Increased content ratio of KYN/KYNA in alpha-synuclein overexpressed fly contrast with the GAL4 control at 1 day and 35 days of age. Data are shown as the mean ± s.e.m. Statistical significance was determined using unpaired two-tailed Student’ t test.

synuclein was overexpressed by elav-GAL4 in fly to simulate PD. Western blot result confirmed that wild-type alphasynuclein was overexpressed in fly (Figure 5A). Tyrosine

patients has biological relevance, we measured levels of KYN and KYNA in the wild-type alpha-synuclein overexpressed fly and the age-matched GAL4 control. Wild-type human alpha473

dx.doi.org/10.1021/pr500807t | J. Proteome Res. 2015, 14, 467−478

Journal of Proteome Research

Article

Figure 6. Box plot for the levels of representative metabolites. (A) Cortisol, (B) 11-deoxycortisol, (C) 21-deoxycortisol, (D) histidine, (E) urocanic acid, (F) imidazoleacetic acid, (G) KYN/KYNA ratio (KYN, kynurenine; KYNA, kynurenic acid), and (H) hydroxyphenylacetic acid. Asterisks denote P < 0.001.

spectrometry method was employed to investigate the metabolic profiles of urine and serum in a small number of PD subjects. Results indicate that changes in urine composition may be more helpful than those of serum.12 However, to our knowledge, in-depth study of the relationship between urinary metabolite profile and PD is still needed. In this study, we found a statistically notable relationship between the severity of PD (H&Y grade phenotype and UPRDS motor score) and urinary metabolite profile. RF models yielded >90% accuracy of classifications, which suggested that the urinary metabolic phenotype of idiopathic PD patients was significantly different from that of normal controls. Those 45 metabolites combinations also indicated the possibility of evidence for potential prediction of idiopathic PD. Of note, our results indicated that urine from idiopathic PD patients contains convergent points and end points of abnormal metabolic pathways in PD. The dynamic levels of urine metabolites associated with PD represent a snapshot of the dysfunctional metabolic network in idiopathic PD patients. Urine metabolite signatures may provide valuable information for assessing PD and evaluating the patient responses to therapeutic treatment. Our findings have defined a panel of molecules whose levels are altered in the urine of idiopathic PD patients. These metabolites play important roles in steroidogenesis, fatty acid beta-oxidation, histidine metabolism, phenylalanine metabolism, tryptophan metabolism, glycine derivation, nucleotide metabolism, and tyrosine metabolism. For steroidogenesis metabolism variation, several early studies recognized that PD patients were found to have significantly higher concentrations of cortisol in their blood and saliva.34−36 Cortisol released in response to stress could cross the blood−brain barrier to access glucocorticoids receptors within the central nervous system.

hydroxylase, a marker of dopaminergic neuron, was decreased in the PD flies compared with the GAL4 control (Figure 5B), which indicated dopaminergic neuron loss in the flies. Figure S3 in the Supporting Information showed the MRM chromatograms of a representative sample of fly. Our result showed the significantly increased level of KYN and raised content ratio of KYN to KYNA in alpha-synuclein overexpressed fly contrasted with the GAL4 control at day 1 and day 35 of age (Figure 5C,D), suggesting the alteration of the kynurenine pathway was linked with the pathogenic changes in the fly model of PD.

■

DISCUSSION Urinary metabolomics analysis of 401 urine samples from 106 idiopathic PD patients and 104 normal controls employing LC−MS-based metabolomics protocols and RF discriminant analysis was used to construct a panel of urinary metabolites that are altered in PD. Indeed, those 106 individuals were evaluated with routine medical examinations after 16 weeks. A growing number of studies have used MS-based metabolomics as a method of discovering biomarkers for diagnosing PD; however, these studies have largely used study groups of limited size and invasive sampling methods (i.e., CSF, blood, and tissue). Thus, the strengths of our investigation are two-fold: First, the statistically significant size of our PD samples and normal controls, and second, the use of body fluid that could be sampled without invasive procedures, namely, urine. Urine is a valuable source of biomarkers for neurodegenerative diseases not only because it can be collected noninvasively but also because it reflects terminal changes of blood.32 Urinary 8-hydroxy-2′(8-OHdG),33 a major product of DNA oxidative damage, has been associated with the progression of PD. A gas-chromatography-coupled-to-mass 474

dx.doi.org/10.1021/pr500807t | J. Proteome Res. 2015, 14, 467−478

Journal of Proteome Research

Article

KAT II in plasma of PD patients.58 Recently, the KYN pathway of tryptophan metabolism has been proposed to be a valuable drug target, affected by agents that change the balance of tryptophan catabolites, reducing excitotoxins and increasing neuroprotectants. Such kind of agents could possibly be developed into therapy for neurodegenerative disorders.56 Both 5-hydroxytryptophan and acetyl-serotonin are involved in serotonin pathways, which synthesize the monoamine neurotransmitters associated with depression, that is, serotonin and melatonin. The decreased serotonin in blood and CSF of patients with Parkinsonism has been recognized; however, the excessive excretion of these metabolites involved in the serotonin pathway still needs attention.59,60 Our data shows elevated phenylacetic acid and phenylacetyleglutamine levels in the urine of idiopathic PD patients. Urinary phenylacetyleglutamine could act as a dosing biomarker for patients with urea cycle disorders. The urea cycle may reduce ammonia in the body and avoid CNS disorder caused by the accumulation of ammonia.61,62 Hydroxyphenylacetic acid, an oxidative metabolite of tyramine in tyrosine metabolism, has been detected in the CSF and associated with neurological disorders.63 The increased excretion of hydroxyphenylacetic acid in the urine of PD patients has also been validated by Sandler et al. 64 Hydroxyphenylacetic acid is reduced significantly after the use of the antibiotic neomycin, which might indicate the interaction between metabolites and gut microbiota in PD patients64 (Figure 6H). The level of TMAO was found to be significantly increased in the urine of idiopathic PD patients compared with that of controls. Trimethylamine-N-oxide (TMAO) is converted from dietary carnitine by normal gut bacteria in the human microbiome. It could induce folding of alphasynuclein.65 In recent years, increasing evidence has suggested that gut microbiota can modulate brain development and behavior via metabolites in the gut−brain axis.66 Urine metabolite profiling has been widely used to indicate the importance of gut microbiota dynamics in disease progression.67,68 Urinary metabolomics appears to be a promising approach to revealing the connection between gut microbiota and the progression of PD. The fruit fly Drosophila has emerged as a valuable model for studying the molecular and cellular mechanisms of PD due to its high degree of conserved biological pathways.31 Our results showed that altered tryptophan metabolism is one of the metabolite signatures common between PD patients and alphasynuclein overexpressed fly model of PD and thus may be of great future value for developing potential markers of the disease process and evaluating the efficacy of novel therapeutic agents., but beyond that, other common metabolites and related biological pathways (e.g., histidine metabolism, steroidogenesis, and phenylalanine metabolism) across species may play important roles in PD and need to be further explored. Although a cohort of a relatively large number of samples ensures the statistical power and reliability of this study, the evaluation of some factors’ effects on altered metabolites, such as secondary PD, drug treatment, and other CNS disorders, still needs to be further investigated. Additional analytical technologies, for example, nuclear magnetic resonance and gas chromatography−mass spectrometry, may provide even more comprehensive metabolome analysis of urine.

Stress and elevated corticosteroid levels in PD animal models have been associated with inflammation-mediated dopaminergic neuron injury and motor handicap.37,38 The increased levels of urinary glucocorticoids including cortisol, 11-deoxycortisol, and 21-deoxycortisol may be valuable indicators for the progression of PD39 (Figure 6A−C). Urinary excretion of cortisol has been correlated with increased oxidative stress, which contributes to dopamine cell degeneration in PD.40,41 Some evidence suggests that a group of mutated components localized in mitochondria exacerbate PD.7 Mitochondrial dysfunction is closely associated with increased oxidative stress and reduced ATP levels.7,42 Impaired energy metabolism resulting from mitochondrial dysfunction might trigger neuronal degeneration in neurodegenerative diseases.43,44 Our data revealed increased levels of acyl-carnitines (hydroxylauroylcarnitine, hydroxyhexanoycarnitine, and malonylcarnitine) in urine of idiopathic PD patients, which might indicate the disturbance of mitochondrial fatty-acid beta-oxidation.45 Impaired mitochondrial carnitine transport system and failure of the mitochondria respiratory chain may contribute to accumulation of these acyl-carnitines in PD.46 Amino acid metabolism is involved in chemical neurotransmitter regulation in the nervous system. According to increasing evidence, the histaminergic system in the tuberomamillary nucleus of the posterior hypothalamus contributes to many brain functions such as sleep regulation, suppressive effects, and hormone secretion, which have been related to the nonmotor symptoms of PD.47 Both CSF and plasma histamine levels were increased in PD patients, although the activity of histidine decarboxylase is unchanged in the frontal cortex, caudate nucleus, hypothalamus, and substantia nigra.48,49 Coincident with the up-regulated level of histamine, histidine, urocanic acid, and imidazoleacetic acid involved in histidine metabolism showed increased levels in urine of idiopathic PD patients (Figure 6D−F). Urocanic acid, a breakdown (deamination) product of histidine in the outermost layer of skin, plays a role in UV-induced neuroimmune suppression.50 However, the mechanism of interaction between PD and urocanic acid as well as imidazoleacetic acid, a metabolite product of histamine, needs to be further explored. The neurotoxicity of tryptophan catabolites has been demonstrated in in vivo and in vitro studies.51−53 The imbalance of tryptophan catabolites plays an important role in mitochondrial disturbances, metabolic disturbances, and impairment of brain energy metabolism, which have been considered as culprits for neurodegenerative disorders.54,55 Metabolites including KYN, xanthurenic acid, and hydroxyanthranilic acid involved in tryptophan metabolism were significantly increased in the urine of idiopathic PD patients compared with that of controls. 60% of KYN in central nervous system (CNS) could be taken up from the blood and transported across the blood−brain barrier.52,56 The lower level of KYN in the CSF was reported previously,57 which may be associated with the excessive KYN released in urine. KYNA, proven to be a neuroprotective molecule as a nonselective glutamate receptor antagonist, is synthesized by the irreversible transamination of KYN by KYN aminotransferases (KAT I and KAT II). We observed increased KYN/KYNA ratios, but unchanged KYNA levels in both of the alpha-synuclein overexpressed fly and urine of idiopathic PD patients in this study (Figures 5C,D and 6G). Our results correlate well with the data of Hartai et al., who reported stable KYNA concentrations and down-regulated activities of KAT I and 475

dx.doi.org/10.1021/pr500807t | J. Proteome Res. 2015, 14, 467−478

Journal of Proteome Research

■

Article

(3) Dorsey, E.; Constantinescu, R.; Thompson, J.; Biglan, K.; Holloway, R.; Kieburtz, K.; Marshall, F.; Ravina, B.; Schifitto, G.; Siderowf, A. Projected number of people with Parkinson disease in the most populous nations, 2005 through 2030. Neurology 2007, 68 (5), 384−386. (4) Jenner, P.; Morris, H. R.; Robbins, T. W.; Goedert, M.; Hardy, J.; Ben-Shlomo, Y.; Bolam, P.; Burn, D.; Hindle, J. V.; Brooks, D. Parkinson’s disease–the debate on the clinical phenomenology, aetiology, pathology and pathogenesis. J. Parkinson’s Dis. 2013, 3 (1), 1−11. (5) Sprenger, F.; Poewe, W. Management of motor and non-motor symptoms in Parkinson’s disease. CNS Drugs 2013, 27 (4), 259−272. (6) Kaidanovich-Beilin, O.; Cha, D. S.; McIntyre, R. S. Crosstalk between metabolic and neuropsychiatric disorders. F1000 Biol. Repl 2012, 4, 14. (7) Henchcliffe, C.; Beal, M. F. Mitochondrial biology and oxidative stress in Parkinson disease pathogenesis. Nat. Clin. Pract. Neurol. 2008, 4 (11), 600−609. (8) Quinones, M. P.; Kaddurah-Daouk, R. Metabolomics tools for identifying biomarkers for neuropsychiatric diseases. Neurobiol. Dis. 2009, 35 (2), 165−176. (9) Lei, S.; Powers, R. NMR Metabolomics Analysis of Parkinson’s Disease. Curr. Metabolomics 2013, 1 (3), 191−209. (10) Chahine, L. M.; Stern, M. B.; Chen-Plotkin, A. Blood-based biomarkers for Parkinson’s disease. Parkinsonism Relat. Disord. 2014, 20 (Suppl 1), S99−S103. (11) Lewitt, P. A.; Li, J.; Lu, M.; Beach, T. G.; Adler, C. H.; Guo, L. 3-hydroxykynurenine and other Parkinson’s disease biomarkers discovered by metabolomic analysis. Mov. Disord. 2013, 28 (12), 1653−1660. (12) Michell, A. W.; Mosedale, D.; Grainger, D. J.; Barker, R. A. Metabolomic analysis of urine and serum in Parkinson’s disease. Metabolomics 2008, 4 (3), 191−201. (13) Bogdanov, M.; Matson, W. R.; Wang, L.; Matson, T.; SaundersPullman, R.; Bressman, S. S.; Flint Beal, M. Metabolomic profiling to develop blood biomarkers for Parkinson’s disease. Brain 2008, 131 (Pt 2), 389−396. (14) Poliquin, P. O.; Chen, J.; Cloutier, M.; Trudeau, L.-É.; Jolicoeur, M. Metabolomics and In-Silico Analysis Reveal Critical Energy Deregulations in Animal Models of Parkinson’s Disease. PLoS One 2013, 8 (7), e69146. (15) Musgrove, R.; Horne, J.; Wilson, R.; King, A.; Edwards, L.; Dickson, T. The metabolomics of alpha-synuclein (SNCA) gene deletion and mutation in mouse brain. Metabolomics 2014, 10 (1), 114−122. (16) Johansen, K. K.; Wang, L.; Aasly, J. O.; White, L. R.; Matson, W. R.; Henchcliffe, C.; Beal, M. F.; Bogdanov, M. Metabolomic profiling in LRRK2-related Parkinson’s disease. PLoS One 2009, 4 (10), e7551. (17) Roede, J. R.; Uppal, K.; Park, Y.; Lee, K.; Tran, V.; Walker, D.; Strobel, F. H.; Rhodes, S. L.; Ritz, B.; Jones, D. P. Serum metabolomics of slow vs. rapid motor progression Parkinson’s disease: a pilot study. PLoS One 2013, 8 (10), e77629. (18) Williams-Gray, C. H.; Foltynie, T.; Brayne, C. E.; Robbins, T. W.; Barker, R. A. Evolution of cognitive dysfunction in an incident Parkinson’s disease cohort. Brain 2007, 130 (Pt 7), 1787−1798. (19) Wang, X. J.; Zhang, A. H.; Han, Y.; Wang, P.; Sun, H.; Song, G. C.; Dong, T. W.; Yuan, Y.; Yuan, X. X.; Zhang, M.; Xie, N.; Zhang, H.; Dong, H.; Dong, W. Urine Metabolomics Analysis for Biomarker Discovery and Detection of Jaundice Syndrome in Patients With Liver Disease. Mol. Cell. Proteomics 2012, 11 (8), 370−380. (20) Cheng, Y.; Xie, G.; Chen, T.; Qiu, Y.; Zou, X.; Zheng, M.; Tan, B.; Feng, B.; Dong, T.; He, P.; Zhao, L.; Zhao, A.; Xu, L. X.; Zhang, Y.; Jia, W. Distinct urinary metabolic profile of human colorectal cancer. J. Proteome Res. 2012, 11 (2), 1354−1363. (21) Dunn, W. B.; Broadhurst, D.; Begley, P.; Zelena, E.; FrancisMcIntyre, S.; Anderson, N.; Brown, M.; Knowles, J. D.; Halsall, A.; Haselden, J. N.; Nicholls, A. W.; Wilson, I. D.; Kell, D. B.; Goodacre, R. Procedures for large-scale metabolic profiling of serum and plasma

CONCLUSIONS In summary, this study employed LC−MS technology to profile the metabolites in the urine of a cohort of 106 individual idiopathic PD patients (297 PD urinary samples) and of 104 individual controls over 32 weeks. LC−MS-based urinary metabolite profiling showed PD to be associated with a profound abnormality in metabolic phenotype. Our study produced a panel of altered metabolites associated with idiopathic PD. Metabolic pathways including steroidogenesis, fatty acid beta-oxidation, histidine metabolism, phenylalanine metabolism, tryptophan metabolism, glycine derivation, nucleotide metabolism, and tyrosine metabolism were found to be altered in PD patients. In addition to metabolic profiling analysis, correspondence between alteration of the kynurenine pathway in tryptophan metabolism and pathogenic changes was found for the first time in the alpha-synuclein overexpressed fly PD model. The results suggest that LC−MS-based urinary metabolomics profiling has great promise for discovering the urinary metabolite signatures and related metabolic pathways variations that characterize PD. Consistent PD-related changes across species may be used toward molecular understanding of metabolic regulation in PD.

■

ASSOCIATED CONTENT

S Supporting Information *

Data analysis. Results of PERMANOVA for PD patients and Controls (Table S1). Two-dimensional principal component analysis (2D PCA) scores plot for consecutively analyzed QC samples (Figure S1). Permutation test (n = 999) for the OPLSDA model (Figure S2). MRM chromatograms of a representative sample of fly (Figure S3). This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Authors

*M.L.: Tel: +852 3411 2919. Fax: +852 3411 2461. E-mail: [email protected]. *Z.C.: Tel: +852-34117070. Fax: +852-34117348. E-mail: [email protected]. Author Contributions #

H.L. and L.-F.L. contributed equally to this work.

Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS We thank Prof. Zhao-Xiang Bian, Dr. Feng Sun, and Ms. Vicky Keng from Clinical Division, School of Chinese Medicine, Hong Kong Baptist University for their support and help on this study. We also thank Dr. Martha Dahlen for her English editing. This work was supported by HHSRF08091111 from the Food and Health Bureau of Hong Kong Government, PuraPharm International (H.K.), Ltd., Interdisciplinary Research Match Scheme Grants (IRMS/12-13/1A) from Hong Kong Baptist University, and Dr. Ho Tzu-leung’s Foundation.

■

REFERENCES

(1) de Lau, L. M.; Breteler, M. M. Epidemiology of Parkinson’s disease. Lancet Neurol. 2006, 5 (6), 525−535. (2) Rocca, W. A. Prevalence of Parkinson’s disease in China. Lancet Neurol. 2005, 4 (6), 328−329. 476

dx.doi.org/10.1021/pr500807t | J. Proteome Res. 2015, 14, 467−478

Journal of Proteome Research

Article

using gas chromatography and liquid chromatography coupled to mass spectrometry. Nat. Protoc. 2011, 6 (7), 1060−1083. (22) Luan, H.; Chen, X.; Zhong, S.; Yuan, X.; Meng, N.; Zhang, J.; Fu, J.; Xu, R.; Lee, C.; Song, S.; Jiang, H.; Xu, X. Serum metabolomics reveals lipid metabolism variation between coronary artery disease and congestive heart failure: a pilot study. Biomarkers 2013, 18 (4), 314− 321. (23) Luan, H.; Meng, N.; Liu, P.; Feng, Q.; Lin, S.; Fu, J.; Davidson, R.; Chen, X.; Rao, W.; Chen, F.; Jiang, H.; Xu, X.; Cai, Z.; Wang, J. Pregnancy-Induced Metabolic Phenotype Variations in Maternal Plasma. J. Proteome Res. 2014, 13 (3), 1527−1536. (24) Kuhl, C.; Tautenhahn, R.; Bottcher, C.; Larson, T. R.; Neumann, S. CAMERA: an integrated strategy for compound spectra extraction and annotation of liquid chromatography/mass spectrometry data sets. Anal. Chem. 2012, 84 (1), 283−289. (25) Smith, C. A.; Want, E. J.; O’Maille, G.; Abagyan, R.; Siuzdak, G. XCMS: processing mass spectrometry data for metabolite profiling using nonlinear peak alignment, matching, and identification. Anal. Chem. 2006, 78 (3), 779−787. (26) Want, E. J.; Wilson, I. D.; Gika, H.; Theodoridis, G.; Plumb, R. S.; Shockcor, J.; Holmes, E.; Nicholson, J. K. Global metabolic profiling procedures for urine using UPLC−MS. Nat. Protoc. 2010, 5 (6), 1005−1018. (27) Houtkooper, R. H.; Argmann, C.; Houten, S. M.; Canto, C.; Jeninga, E. H.; Andreux, P. A.; Thomas, C.; Doenlen, R.; Schoonjans, K.; Auwerx, J. The metabolic footprint of aging in mice. Sci. Rep. 2011, 1. (28) Marti, M. J.; Tolosa, E. Parkinson disease: New guidelines for diagnosis of Parkinson disease. Nat. Rev. Neurol. 2013, 9 (4), 190−191. (29) Xia, J.; Mandal, R.; Sinelnikov, I. V.; Broadhurst, D.; Wishart, D. S. MetaboAnalyst 2.0–a comprehensive server for metabolomic data analysis. Nucleic Acids Res. 2012, 40 (Web Server issue), W127−W133. (30) Campesan, S.; Green, E. W.; Breda, C.; Sathyasaikumar, K. V.; Muchowski, P. J.; Schwarcz, R.; Kyriacou, C. P.; Giorgini, F. The kynurenine pathway modulates neurodegeneration in a Drosophila model of Huntington’s disease. Curr. Biol. 2011, 21 (11), 961−966. (31) Pandey, U. B.; Nichols, C. D. Human disease models in Drosophila melanogaster and the role of the fly in therapeutic drug discovery. Pharmacol. Rev. 2011, 63 (2), 411−436. (32) Eller, M.; Williams, D. R. Biological fluid biomarkers in neurodegenerative parkinsonism. Nat. Rev. Neurol. 2009, 5 (10), 561− 570. (33) Bolner, A.; Pilleri, M.; De Riva, V.; Nordera, G. P. Plasma and urinary HPLC-ED determination of the ratio of 8-OHdG/2-dG in Parkinson’s disease. Clin. Lab. 2011, 57 (11−12), 859−866. (34) Charlett, A.; Dobbs, R. J.; Purkiss, A. G.; Wright, D. J.; Peterson, D. W.; Weller, C.; Dobbs, S. M. Cortisol is higher in parkinsonism and associated with gait deficit. Acta Neurol. Scand. 1998, 97 (2), 77−85. (35) Djamshidian, A.; O’Sullivan, S. S.; Papadopoulos, A.; Bassett, P.; Shaw, K.; Averbeck, B. B.; Lees, A. Salivary cortisol levels in Parkinson’ disease and its correlation to risk behaviour. J. Neurol., Neurosurg. Psychiatry 2011, 82 (10), 1107−1111. (36) Hartmann, A.; Veldhuis, J. D.; Deuschle, M.; Standhardt, H.; Heuser, I. Twenty-four hour cortisol release profiles in patients with Alzheimer’s and Parkinson’s disease compared to normal controls: Ultradian secretory pulsatility and diurnal variation. Neurobiol. Aging 1997, 18 (3), 285−289. (37) Ros-Bernal, F.; Hunot, S.; Herrero, M. T.; Parnadeau, S.; Corvol, J. C.; Lu, L. X.; Alvarez-Fischer, D.; Carrillo-de Sauvage, M. A.; Saurini, F.; Coussieu, C.; Kinugawa, K.; Prigent, A.; Hoglinger, G.; Hamon, M.; Tronche, F.; Hirsch, E. C.; Vyas, S. Microglial glucocorticoid receptors play a pivotal role in regulating dopaminergic neurodegeneration in parkinsonism. Proc. Natl. Acad. Sci. U.S.A. 2011, 108 (16), 6632−6637. (38) Smith, L. K.; Jadavji, N. M.; Colwell, K. L.; Perehudoff, S. K.; Metz, G. A. Stress accelerates neural degeneration and exaggerates motor symptoms in a rat model of Parkinson’s disease. Eur. J. Neurosci. 2008, 27 (8), 2133−2146.

(39) Kibel, A.; Drenjancevic-Peric, I. Impact of glucocorticoids and chronic stress on progression of Parkinson’s disease. Med. Hypotheses 2008, 71 (6), 952−956. (40) Bagieva, G. K.; Fedorova, T. N.; Stvolinsky, S. L.; Boldyrev, A. A.; Ivanova-Smolenskaya, I. A.; Illarioshkin, S. N. Oxidative stress in the pathogenesis of Parkinson’s disease. Mov. Disord. 2004, 19, S266− S266. (41) Joergensen, A.; Broedbaek, K.; Weimann, A.; Semba, R. D.; Ferrucci, L.; Joergensen, M. B.; Poulsen, H. E., Association between Urinary Excretion of Cortisol and Markers of Oxidatively Damaged DNA and RNA in Humans. PLoS One 2011, 6, (6). (42) Abou-Sleiman, P. M.; Muqit, M. M. K.; Wood, N. W. Expanding insights of mitochondrial dysfunction in Parkinson’s disease. Nat. Rev. Neurosci. 2006, 7 (3), 207−219. (43) Beal, M. F. Does impairment of energy metabolism result in excitotoxic neuronal death in neurodegenerative illnesses? Ann. Neurol. 1992, 31 (2), 119−130. (44) Haorah, J.; Rump, T. J.; Xiong, H. G., Reduction of Brain Mitochondrial beta-Oxidation Impairs Complex I and V in Chronic Alcohol Intake: The Underlying Mechanism for Neurodegeneration. PLoS One 2013, 8, (8). (45) Vreken, P.; van Lint, A. E.; Bootsma, A. H.; Overmars, H.; Wanders, R. J.; van Gennip, A. H. Rapid diagnosis of organic acidemias and fatty-acid oxidation defects by quantitative electrospray tandemMS acyl-carnitine analysis in plasma. Adv. Exp. Med. Biol. 1999, 466, 327−337. (46) Reuter, S. E.; Evans, A. M. Carnitine and Acylcarnitines Pharmacokinetic, Pharmacological and Clinical Aspects. Clin. Pharmacokinet. 2012, 51 (9), 553−572. (47) Tiligada, E.; Kyriakidis, K.; Chazot, P. L.; Passani, M. B. Histamine pharmacology and new CNS drug targets. CNS Neurosci. Ther. 2011, 17 (6), 620−628. (48) Prell, G. D.; Green, J. P. Histamine metabolites and prosmethylimidazoleacetic acid in human cerebrospinal fluid. Agents Actions Suppl 1991, 33, 343−363. (49) Rinne, J. O.; Anichtchik, O. V.; Eriksson, K. S.; Kaslin, J.; Tuomisto, L.; Kalimo, H.; Roytta, M.; Panula, P. Increased brain histamine levels in Parkinson’s disease but not in multiple system atrophy. J. Neurochem. 2002, 81 (5), 954−960. (50) Walterscheid, J. P.; Nghiem, D. X.; Kazimi, N.; Nutt, L. K.; McConkey, D. J.; Norval, M.; Ullrich, S. E. Cis-urocanic acid, a sunlight-induced immunosuppressive factor, activates immune suppression via the 5-HT2A receptor. Proc. Natl. Acad. Sci. U.S.A. 2006, 103 (46), 17420−17425. (51) Speciale, C.; Hares, K.; Schwarcz, R.; Brookes, N. High-affinity uptake of L-kynurenine by a Na+-independent transporter of neutral amino acids in astrocytes. J. Neurosci. 1989, 9 (6), 2066−2072. (52) Fukui, S.; Schwarcz, R.; Rapoport, S. I.; Takada, Y.; Smith, Q. R. Blood-brain barrier transport of kynurenines: implications for brain synthesis and metabolism. J. Neurochem. 1991, 56 (6), 2007−2017. (53) Moroni, F. Tryptophan metabolism and brain function: focus on kynurenine and other indole metabolites. Eur. J. Pharmacol. 1999, 375 (1−3), 87−100. (54) Sas, K.; Robotka, H.; Toldi, J.; Vecsei, L. Mitochondria, metabolic disturbances, oxidative stress and the kynurenine system, with focus on neurodegenerative disorders. J. Neurol. Sci. 2007, 257 (1−2), 221−239. (55) Schuck, P. F.; Tonin, A.; Ferreira, G. D.; Viegas, C. M.; Latini, A.; Wannmacher, C. M. D.; Wyse, A. T. D.; Dutra, C. S.; Wajner, M. Kynurenines impair energy metabolism in rat cerebral cortex. Cell. Mol. Neurobiol. 2007, 27 (1), 147−160. (56) Vecsei, L.; Szalardy, L.; Fulop, F.; Toldi, J. Kynurenines in the CNS: recent advances and new questions. Nat. Rev. Drug Discovery 2013, 12 (1), 64−82. (57) Tohgi, H.; Abe, T.; Takahashi, S.; Takahashi, J.; Hamato, H. Concentrations of serotonin and its related substances in the cerebrospinal fluid of parkinsonian patients and their relations to the severity of symptoms. Neurosci. Lett. 1993, 150 (1), 71−74. 477

dx.doi.org/10.1021/pr500807t | J. Proteome Res. 2015, 14, 467−478

Journal of Proteome Research

Article

(58) Hartai, Z.; Klivenyi, P.; Janaky, T.; Penke, B.; Dux, L.; Vecsei, L. Kynurenine metabolism in plasma and in red blood cells in Parkinson’s disease. J. Neurol. Sci. 2005, 239 (1), 31−35. (59) Mayeux, R.; Stern, Y.; Cote, L.; Williams, J. B. Altered serotonin metabolism in depressed patients with parkinson’s disease. Neurology 1984, 34 (5), 642−646. (60) Karimova, M.; Aleuov, Sh. A. Changes in the blood level of serotonin in Parkinson disease. Vopr. Med. Khim. 1988, 34 (1), 45−47. (61) Mokhtarani, M.; Diaz, G. A.; Rhead, W.; Lichter-Konecki, U.; Bartley, J.; Feigenbaum, A.; Longo, N.; Berquist, W.; Berry, S. A.; Gallagher, R.; Bartholomew, D.; Harding, C. O.; Korson, M. S.; McCandless, S. E.; Smith, W.; Vockley, J.; Bart, P. S.; Kronn, D.; Zori, R.; Cederbaum, S.; Dorrani, N.; Merritt, J. L.; Sreenath-Nagamani, S.; Summar, M.; LeMons, C.; Dickinson, K.; Coakley, D. F.; Moors, T. L.; Lee, B.; Scharschmidt, B. F. Urinary phenylacetylglutamine as dosing biomarker for patients with urea cycle disorders. Mol. Genet. Metab. 2012, 107 (3), 308−314. (62) Mokhtarani, M.; Diaz, G. A.; Rhead, W.; Berry, S. A.; LichterKonecki, U.; Feigenbaum, A.; Schulze, A.; Longo, N.; Bartley, J.; Berquist, W.; Gallagher, R.; Smith, W.; McCandless, S. E.; Harding, C.; Rockey, D. C.; Vierling, J. M.; Mantry, P.; Ghabril, M.; Brown, R. S.; Dickinson, K.; Moors, T.; Norris, C.; Coakley, D.; Milikien, D. A.; Nagamani, S. C.; LeMons, C.; Lee, B.; Scharschmidt, B. F. Elevated phenylacetic acid levels do not correlate with adverse events in patients with urea cycle disorders or hepatic encephalopathy and can be predicted based on the plasma PAA to PAGN ratio. Mol. Genet. Metab. 2013, 110 (4), 446−453. (63) Kobayashi, K.; Koide, Y.; Yoshino, K.; Shohmori, T. Phydroxyphenylacetic acid concentrations in cerebrospinal fluid. No to Shinkei 1982, 34 (8), 769−774. (64) Sandler, M.; Karoum, F.; Ruthven, C. R.; Calne, D. B. mHydroxyphenylacetic acid formation from L-dopa in man: suppression by neomycin. Science 1969, 166 (3911), 1417−1418. (65) Uversky, V. N.; Li, J.; Fink, A. L. Trimethylamine-N-oxideinduced folding of alpha-synuclein. FEBS Lett. 2001, 509 (1), 31−35. (66) Heijtza, R. D.; Wang, S. G.; Anuar, F.; Qian, Y.; Bjorkholm, B.; Samuelsson, A.; Hibberd, M. L.; Forssberg, H.; Pettersson, S. Normal gut microbiota modulates brain development and behavior. Proc. Natl. Acad. Sci. U.S.A. 2011, 108 (7), 3047−3052. (67) Li, M.; Wang, B. H.; Zhang, M. H.; Rantalainen, M.; Wang, S. Y.; Zhou, H. K.; Zhang, Y.; Shen, J.; Pang, X. Y.; Zhang, M. L.; Wei, H.; Chen, Y.; Lu, H. F.; Zuo, J.; Su, M. M.; Qiu, Y. P.; Jia, W.; Xiao, C. N.; Smith, L. M.; Yang, S. L.; Holmes, E.; Tang, H. R.; Zhao, G. P.; Nicholson, J. K.; Li, L. J.; Zhao, L. P. Symbiotic gut microbes modulate human metabolic phenotypes. Proc. Natl. Acad. Sci. U.S.A. 2008, 105 (6), 2117−2122. (68) Marcobal, A.; Kashyap, P. C.; Nelson, T. A.; Aronov, P. A.; Donia, M. S.; Spormann, A.; Fischbach, M. A.; Sonnenburg, J. L. A metabolomic view of how the human gut microbiota impacts the host metabolome using humanized and gnotobiotic mice. ISME J. 2013, 7 (10), 1933−1943.

478

dx.doi.org/10.1021/pr500807t | J. Proteome Res. 2015, 14, 467−478