Metabolic Biomarkers and Neurodegeneration: A Pathway Enrichment Analysis of Alzheimer's Disease,Parkinson's Disease,and Amyotrophic Lateral Sclerosis

Abstract

Neurodegenerative diseases such as Alzheimer's disease (AD), Parkinson's disease (PD), and amyotrophic lateral sclerosis (ALS) lack robust diagnostics and prognostic biomarkers. Metabolomics is a postgenomics field that offers fresh insights for biomarkers of common complex as well as rare diseases. Using data on metabolite-disease associations published in the previous decade (2006–2016) in PubMed, ScienceDirect, Scopus, and Web of Science, we identified 101 metabolites as putative biomarkers for these three neurodegenerative diseases. Notably, uric acid, choline, creatine, L-glutamine, alanine, creatinine, and N-acetyl-L-aspartate were the shared metabolite signatures among the three diseases. The disease-metabolite-pathway associations pointed out the importance of membrane transport (through ATP binding cassette transporters), particularly of arginine and proline amino acids in all three neurodegenerative diseases. When disease-specific and common metabolic pathways were queried by using the pathway enrichment analyses, we found that alanine, aspartate, glutamate, and purine metabolism might act as alternative pathways to overcome inadequate glucose supply and energy crisis in neurodegeneration. These observations underscore the importance of metabolite-based biomarker research in deciphering the elusive pathophysiology of neurodegenerative diseases. Future research investments in metabolomics of complex diseases might provide new insights on AD, PD, and ALS that continue to place a significant burden on global health.

Introduction

Neurodegenerative diseases represent a heterogeneous group of disorders that are characterized by the progressive degeneration of the structure and function of the central or peripheral nervous system. Since neurons do not reproduce and replace themselves, their degeneration or death induces incurable and debilitating conditions such as ataxia or dementia. Alzheimer's disease (AD), Parkinson's disease (PD), and amyotrophic lateral sclerosis (ALS) are three major neurodegenerative diseases that lack robust diagnostics or prognostic biomarkers.

AD, being the most prevalent neurodegenerative disease, is the best known source of dementia, especially the adult-onset dementia (Ferri et al., 2006; Paglia et al., 2016), and it may contribute to 60–70% of cases (WHO, 2016). It is estimated that more than 80 million people will be suffering from this disease until 2040 (Cornutiu, 2015). AD is a complex and progressive disorder. The formation of intraneuronal neurofibrillary tangles and extracellular amyloid plaques are the two histopathological signatures of the AD (Bloom, 2014; Multhaup et al., 2015; Paglia et al., 2016).

The second most common neurodegenerative disorder is PD (Tahmasian et al., 2015). Hereditary and environmental factors play important roles in the development of PD (Delenclos et al., 2016; Hatano et al., 2016). In addition, aging is an important risk factor for disease development (Rodríguez-Nogales et al., 2016). It was estimated that 2% of the population with PD are older than 65 (Ghio et al., 2016). Hence, the percentage of aged people in the general population is increased gradually, with PD becoming a major global health burden (Delenclos et al., 2016). There are two main clinical findings of PD: (1) a progressive loss of dopaminergic neurons in the substantia nigra (S. Nigra) pars compacta located in the midbrain, and (2) aberrant aggregation and folding of the α-synuclein protein, which creates inclusions (known as Lewy bodies) that cause neuronal dysfunctions in the influenced brain areas (Ghio et al., 2016; Hatano et al., 2016). Apart from tremor, patients may experience bradykinesia, disorders of speech, swallowing, sleep, and cognition (Andalib et al., 2014).

ALS is another neurodegenerative disease lacking robust diagnostics. ALS is characterized by degeneration of upper and lower motor neurons in the brain, brainstem, and spinal cord, usually leading to death in about 2–5 years of onset. ALS pathophysiology is not clearly understood, because multiple mechanisms are associated with the motor neuron death, including mitochondrial dysfunction, oxidative stress, glutamate toxicity, apoptosis, and aggregation of misfolded proteins (Chio et al., 2013; Mitchell et al., 2009). Phenotypic heterogeneity and lack of diagnostic biomarkers pose difficulties in establishing the diagnosis in very early stages. Indeed, the interval time between the appearance of first symptoms and the diagnosis is more than 1 year and often, it is greater (Blasco et al., 2015, 2016).

Collectively, degenerative diseases of the nervous system pose substantial global health burdens on populations worldwide. The prevalence and incidence of these diseases rise dramatically with age; thus, the number of cases is expected to increase for the foreseeable future, as life spans in many countries continue to increase. Although there has been considerable progress with respect to understanding the triggers of neurodegenerative diseases, causal contributions from genetic and environmental factors are, with some exceptions, poorly understood. Here, the importance of early diagnosis comes into prominence. One of the most important techniques used for detection of biomarkers for diagnosis of neurodegenerative diseases is metabolic profiling or the field of metabolomics.

Metabolomics is an emerging specialty of postgenomics inquiry that examines the global changes in metabolites and metabolic networks underpinning common complex disease pathogenesis and/or mechanisms of host–environment interactions with a view to precision medicine. Metabolomics has found recent broad applications in both preclinical and clinical research (Embade et al., 2016; Kim et al., 2011; Lindeque et al., 2015; Lopes et al., 2015). The metabolome, one of the end products of gene-environment interactions immediately upstream to their phenotypic impacts, can help characterize and discriminate pathophysiological signatures of AD (Sapkota et al., 2015), PD (Hatano et al., 2016), and ALS (Blasco et al., 2016). Still, only a limited number of target molecules are associated with neurodegenerative disease pathogenesis or prognosis (Blasco et al., 2010, 2016; Sussmuth et al., 2008; Tumani et al., 2008).

The present research was undertaken by using pathway enrichment analysis to evaluate the proposed metabolic biomarkers of three common neurodegenerative diseases noted earlier, and to discern the similarities and differences of their molecular signatures from a systems perspective, also analyzing the significantly altered metabolic pathways.

Materials and Methods

Data curation

A comprehensive literature search was performed for the hitherto proposed metabolic biomarkers of AD, PD, and ALS through electronic searches including PubMed, ScienceDirect, Scopus, and Web of Science. The studies to be included in the analyses were selected in accordance with the following criteria: (1) The articles must have been published within the previous decade (since 2006); (2) the studies must be published as a review article or an original research paper, including low-throughput experimentation in human blood (serum, plasma), urine, cerebrospinal fluid (CSF), or S. Nigra; and (3) the studies must investigate the association between metabolites and the investigated disease(s).

A literature search with the criteria mentioned earlier resulted in a total of 101 metabolites that were proposed as biomarkers for the three neurodegenerative diseases. In each case, the following set of data were extracted: the name and Kyoto Encyclopedia of Genes and Genomes (KEGG) compound ID of the metabolite, associated disease, the alteration in the expression pattern of the metabolite, the sample type used, the experimental method performed throughout the analysis, and the citation (first author and date) of publication.

Comparative analyses

To find similarities and differences of disease signatures at metabolome level, all possible comparisons were performed and results were represented by Venn diagrams via Jvenn, an interactive Venn diagram viewer (Bardou et al., 2014). In addition, to evaluate any possible correlation between the expression patterns of the metabolites, a heat map was constructed.

Pathway enrichment analyses of metabolites

Statistically significant metabolic pathways associated with proposed metabolic biomarkers were elucidated through metabolites biological role (MBRole), which performs an overrepresentation (enrichment) analysis of categorical annotations (pathways, biological roles, enzyme interactions, diseases, etc.) for a set of compounds of interest (Chagoyen and Pazos, 2011). The KEGG database was used as the annotation source, the KEGG compound ID's of metabolites were used as input, and significantly enriched (p < 0.05) metabolic and signaling pathways were identified. A hypergeometric test was employed to evaluate the significance of the enrichments: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*} P ( X = k ) = { \frac { { K \choose k } { { N - K } \choose { n - k } } } { { N \choose n } } } \end{align*} \end{document}

where N is the total number of metabolites, K is the total number of metabolites associated with the pathway, n is the number of metabolites in the set, and k is the number of pathway-associated metabolites in the set. The p values were corrected by using the Benjamini–Hochberg method.

Construction of disease-metabolite-pathway associations

Disease-metabolite and metabolite-pathway associations were constructed through literature survey and pathway enrichment analyses, respectively. These analyses then led to the coupling of diseases with metabolic pathways. In addition, for each disease, linkages between metabolic pathways were constructed by using shared metabolites. Here, we assumed that (1) if two pathways share at least one common metabolite (associated with the inspected disease), they should be linked, and this linkage may be associated with disease; and (2) if the number of common metabolites increases, the significance of the linkage increases as well. With regard to these assumptions, the pathway interaction graph was constructed by using Cytoscape (Shannon et al., 2003) for each disease. The topological analysis of the graphs was performed by using CytoHubba plugin (Smoot et al., 2011), and degree metrics were used to define the significance of nodes in the graph. In visualization, node color and size expressed the p value significance and pathway impact, respectively. Colors with sliding through light to dark gray showed increased statistical significance, whereas larger circle diameters corresponded to higher degree values.

Results

We identified a total of 101 metabolites that were proposed as biomarkers for the three neurodegenerative diseases. These included 58, 54, and 46 metabolites for AD, PD, and ALS, respectively (Table 1).

Table 1.

Metabolite Markers for Three Neurodegenerative Diseases

Metabolites' names	KEGG ID	Diseases	Expression pattern	Sample type	Analysis method	Reference
Acetate	C00033	PD	Decreased	Plasma	¹H-NMR	Ahmed et al. (2009)
		ALS	Increased	Serum/CSF	¹H-NMR	Blasco et al. (2015), Kumar et al. (2010), Toczylowska et al. (2013).
Acetone	C00207	ALS	Increased	Serum/CSF	NMR	Blasco et al. (2015), Kumar et al. (2010).
N-acetyl-L-aspartate	C01042	AD	Decreased	Brain	¹H-NMR	Kantarci et al. (2000), Kumar et al. (2010)
		PD		S. Nigra	MRS	Gröger et al. (2014)
		ALS		Serum	MRS	Kumar et al. (2010)
Adenosine	C00212	AD	Increased	Serum	GC-MS	González-Domínguez et al. (2015)
Alanine	C01401	AD	Increased	Blood/CSF	UPLC-ToF-MS	Trushina and Mielke (2014)
		PD	Decreased	CSF	¹H-NMR	Ahmed et al. (2009)
		ALS	Increased	CSF	¹H-NMR	Toczylowska et al. (2013)
Amine	C00706	ALS	Decreased	Plasma/CSF	GC-MS/¹H-NMR	Lawton et al. (2012)
γ-aminobutyric acid	C00334	PD	Increased	S. Nigra	3D-MRI	Gröger et al. (2014)
Anandamide	C11695	AD	Decreased	Plasma	UPLC-MS-MS	Voyle et al. (2016)
Arginine	C02385	PD	Decreased	Serum	¹H-NMR	Ahmed et al. (2009)
Ascorbate	C00072	PD	Decreased	Plasma	¹H-NMR	Ahmed et al. (2009)
		ALS	Increased/decreased	CSF	¹H-NMR/LC-HRMS	Blasco et al. (2010), Spasojević et al. (2010)
Asparagine	C16438	PD	Decreased/increased	Serum/CSF/blood	CE-MS/UPLC-ToF-MS	Ahmed et al. (2009), González-Domínguez et al. (2014b), Trushina and Mielke (2014)
Aspartate	C16433	PD	Increased	CSF	¹H-NMR	Ahmed et al. (2009)
		ALS		CSF/blood	UPLC-ToF-MS	Trushina and Mielke (2015)
Betaine	C00719	ALS	Decreased	Plasma	UHPLC-MS-MS	Lawton et al. (2012)
Beta-alanine	C00099	AD	Increased	CSF/plasma	UPLC-ToF-MS	Trushina and Mielke (2014)
Caffeine	C07481	ALS	Decreased	Plasma	UHPLC-MS-MS	Lawton et al. (2012)
Carnitine	C00487	AD	Decreased	Serum	CE-MS	González-Domínguez et al. (2014b)
		ALS	Increased	Plasma	UHPLC-MS-MS	Lawton et al. (2012)
Choline	C00114	AD	Increased	Serum	CE-MS	Gröger et al. (2014)
		PD	Decreased	S. Nigra	DIES-MS	González-Domínguez et al. (2014a)
		ALS	Increased	CSF	¹H-NMR	Toczylowska et al. (2013)
Citrate	C00158	PD	Decreased	Plasma	¹H-NMR	Ahmed et al. (2009)
Creatine	C00300	AD	Increased	CSF	3D-MRI	Jukarainen et al. (2008)
		PD	Decreased	S. Nigra	3D-MRI	Gröger et al. (2014)
		ALS	Increased/decreased	Plasma/CSF	UHPLC-MS-MS/GC-ToFMS	Lawton et al. (2012), Wuolikainen et al. (2011)
Creatinine	C00791	AD	Increased	Serum	CE-MS	González-Domínguez et al. (2014b)
		PD	Decreased	CSF	GC-ToFMS	Trupp et al. (2014)
		ALS	Decreased/increased	CSF/plasma, CSF/serum	GC-MS/UHPLC-MS-MS/¹H-NMR	Botosoa et al. (2012), Lawton et al. (2012), Trupp et al. (2014), Toczylowska et al. (2013)
Cysteine	C00736	AD	Increased	CSF/plasma	GC-MS/LC-MS-MS/UPLC-ToF-MS	Trushina and Mielke (2014)
Cystine	C01420	PD	Decreased	Plasma	LC-QToF-MS	Wang et al. (2014)
Desmosterol	C01802	AD	Decreased	Plasma	LC/MS	Sato et al. (2012)
Dopamine	C03758	PD	Decreased	S. Nigra	3D-MRI	Gröger et al. (2014)
Ethanolamine	C00189	PD	Decreased	Plasma	¹H-NMR	Ahmed et al. (2009)
Fatty acids (saturated and polysarurated)	C00162	PD	Increased	Serum	DIES-MS	González-Domínguez et al. (2014a)
Formate	C00058	ALS	Increased	Serum	¹H-NMR	Kumar et al. (2010)
Galacticol	C01697	PD	Decreased	Plasma	¹H-NMR	Ahmed et al. (2009)
Gluconate (D-gluconic acid; D-gluco-hexonic acid)	C00257	PD	Decreased	Plasma	¹H-NMR	Ahmed et al. (2009)
Glucose	C00031, C00293	ALS	Increased	CSF	¹H-NMR	Toczylowska et al. (2013)
Glutamic acid	C00302	PD	Decreased	Plasma	LC-QToF-MS	Wang et al. (2014)
Glutamate	C00025	PD	Decreased/increased	CSF/plasma	¹H-NMR	Ahmed et al. (2009)
		ALS	Decreased/increased	Plasma/CSF	¹H-NMR/GC-ToFMS/LC-HRMS	Blasco et al. (2013), Kumar et al. (2010), Wuolikainen et al. (2011)
L-glutamine	C00064	AD	Increased	Urine	LC-QToF-MS	Cui et al. (2014)
		PD	Decreased	CSF	¹H-NMR	Ahmed et al. (2009)
		ALS	Increased	CSF	¹H-NMR	Blasco et al. (2015)
5-L-glutamylglycine (5-L-glutamyl-L-cysteinylglycine)	C00051	AD	Increased	Urine	LC-QToF-MS	Cui et al. (2014)
Glutarate	C00489	PD	Decreased	Plasma	¹H-NMR	Ahmed et al. (2009)
Glutathione	C00051	PD	Decreased	S. Nigra	3D-MRI	Gröger et al. (2014)
Glycine	C00037	PD	Increased	CSF	¹H-NMR	Ahmed et al. (2009)
		ALS				Toczylowska et al. (2013)
Glycolate	C00160	PD	Decreased	Plasma	¹H-NMR	Ahmed et al. (2009)
Glycerol	C00116	PD	Decreased	Plasma	¹H-NMR	Ahmed et al. (2009)
Histidine	C00768	AD	Decreased	Serum	CE-MS	González-Domínguez et al. (2014b)
		ALS	Increased	CSF	¹H-NMR	Kumar et al. (2010), Toczylowska et al. (2013)
Hippurate	C01586	ALS	Decreased	Plasma	UHPLC-MS-MS	Lawton et al. (2012)
Homocysteine	C05330	ALS	Increased	Plasma	NMR	Zoccolella et al. (2008)
2-hydroxybutyrate	C05984	ALS	Increased	Plasma	UHPLC-MS-MS	Lawton et al. (2012)
β-hydroxybutyrate	C01089	ALS	Increased	Serum	NMR	Kumar et al. (2010)
24S-hydroxycholesterol (24S-OH)	C13550	AD	Increased	CSF	SPE-IDMS	Leoni and Caccia (2011)
		PD	Decreased	Plasma	GC-MS/GC-ToFMS	Lee et al. (2009), Trupp et al. (2014)
3-hydroxyisovaleric acid	C01013	PD	Decreased	CSF	GC-ToFMS	Trupp et al. (2014)
2-hydroxyphenylacetic acid	C05852	AD	Increased	CSF	LCECA	Kaddurah-Daouk et al. (2013)
4-hydroxyphenyllactic acid	C00642	AD	Increased	CSF	LCECA	Kaddurah-Daouk et al. (2013)
5-hydroxytryptophan	C01017	AD	Increased	PMV-CSF	LCECA	Trushina and Mielke. (2014)
Homovanillic acid	C05582	PD	Increased	S. Nigra	3D-MRI	Gröger et al. (2014)
Inositol	C00137	ALS	Increased	CSF	¹H-NMR	Blasco et al. (2015)
Isoleucıne	C16434	PD	Increased	Serum	¹H-NMR	Ahmed et al. (2009)
Isocitrate	C00311	AD	Increased	Serum	GC-MS	González-Domínguez et al. (2015)
		PD	Decreased	Plasma	¹H-NMR	Ahmed et al. (2009)
α-ketoglutarate	C00026	AD	Increased	Serum	GC-MS	González-Domínguez et al. (2015)
		ALS	Increased	Plasma	UHPLC-MS-MS	Lawton, et al. (2012)
Ketone	C01450	ALS	Increased	Serum	¹H-NMR	Kumar et al. (2010)
Lactate	C1432	ALS	Increased	CSF/plasma/serum	¹H-NMR	Botosoa et al. (2012), Gray et al. (2015), Lawton et al. (2012), Toczylowska et al. (2013)
Lipid hydroperoxides	C01025	PD	Increased	Plasma	Chemiluminescence	de Farias et al. (2016)
Lysine	C16440	PD	Decreased	CSF	¹H-NMR	Ahmed et al. (2009),
		ALS	Increased			Toczylowska et al. (2013)
Lysophosphatidylcholine (2-lysolecitine)	C04230	AD	Decreased	Serum	LC-QToF-MS	Cui et al. (2014)
Malate	C00149, C00711	PD	Decreased	Plasma	¹H-NMR	Ahmed et al. (2009)
Malondialdehyde (malonaldehyde)	C19440	PD	Increased	Plasma	HPLC	de Farias et al. (2016)
Mannose	C00159	ALS	Increased	Plasma	UHPLC-MS-MS	Lawton et al. (2012)
Methionine	C00073	AD	Decreased	Serum	CE-MS	González-Domínguez et al. (2014b)
	C01733	PD	Decreased	Serum/PMV-CSF	¹H-NMR/LCECA	Ahmed et al. (2009), Trushina and Mielke (2014)
3-methoxytyramine (3-methoxy-4-hydroxyphenethylamine)	C05587	AD	Decreased	PMV-CSF	LCECA	Trushina and Mielke (2014)
Methylamine	C00218	PD	Decreased	Plasma	¹H-NMR	Ahmed et al. (2009)
Methylmalonate	C02170	PD	Decreased	Plasma	¹H-NMR	Ahmed et al. (2009)
Myoinositol	C00137	PD	Decreased/increased	S. Nigra/plasma	3D-MRI/¹H-NMR	Ahmed et al. (2009), Gröger et al. (2014)
N-acetyl derivative (N-acetyl-X)	C01042	ALS	Decreased	Serum	¹H-NMR	Kumar et al. (2010)
Norephinephrine (noradrenaline)	C00547	AD	Decreased	PMV-CSF	LCECA	Trushina and Mielke (2014)
Oxaloacetate	C00036	ALS	Increased	Plasma	UHPLC-MS-MS	Lawton et al. (2012)
2-oxoisocaproate (ketoleucine) (4-methyl-2-oxopentanoat)	C00233	PD	Increased	Plasma	GC-ToFMS	Trupp et al. (2014)
1-palmitoleoylglycero-phosphocholine	C04102	ALS	Increased	Plasma	GC-MS	Lawton et al. (2012)
Paraxanthine	C13747	ALS	Decreased	Plasma	UHPLC-MS-MS	Lawton et al. (2012)
Pentothenate	C00864	ALS	Increased	Plasma	UHPLC-MS-MS	Lawton et al. (2012)
Phenylalanine	C02057	ALS	Increased	CSF	¹H-NMR	Toczylowska et al. (2013)
3-phenylpropionate (benzyl acetate)	C15513	ALS	Decreased	Plasma	UHPLC-MS-MS	Lawton, et al. (2011)
Pipecolic acid	C00408	AD	Decreased	Serum	GC-MS	González-Domínguez et al. (2015)
Proline	C16435	PD	Increased	Plasma	LC-QToF-MS	Wang et al. (2014)
Propylene glycol	C00583	PD	Increased	Plasma	¹H-NMR	Ahmed et al. (2009)
Pyroglutamate	C01879	AD	Increased	Serum	GC-MS	González-Domínguez et al. (2015)
		PD		Plasma	GC-ToFMS	Trupp et al. (2014)
Pyruvate	C00022	PD	Increased	Plasma	¹H-NMR/ UHPLC-MS-MS	Ahmed et al. (2009), Lawton et al. (2012)
		ALS		Serum/CSF	¹H-NMR	Blasco et al. (2015), Blasco et al. (2010), Kumar et al. (2010)
Riluzole	C07937	ALS	Increased	Plasma	GC-MS	Lawton et al. (2012)
Sorbitol (glucitol)	C00794	PD	Increased	Plasma	¹H-NMR	Ahmed et al. (2009)
Sphinganine-1-phosphate	C01120	AD	Increase	Serum	UHPLC-MS	Liang et al. (2016)
Stearoyl sphingomyelin	C00550	ALS	Increased	Plasma	GC-MS	Lawton et al. (2012)
Suberate	C08278	PD	Decreased	Plasma	¹H-NMR	Ahmed et al. (2009)
Succinate (unmedicated)	C00042	PD	Decreased	Plasma	¹H-NMR	Ahmed et al. (2009)
Superoxide dismutase	K04564	PD	Increased	Plasma	HPLC	de Farias et al. (2016)
Theobromine	C07480	ALS	Decreased	Plasma	UHPLC-MS-MS	Lawton et al. (2012)
Theophylline	C07130	ALS	Decreased	Plasma	UHPLC -MS-MS	Lawton et al. (2012)
Threonate	C01620	PD	Decreased	Plasma	¹H-NMR	Ahmed et al. (2009)
Trimethylamine	C00565	PD	Decreased	Plasma	¹H-NMR	Ahmed et al. (2009)
Tryptophan	C00078	PD	Decreased	CSF	GC-ToFMS	Trupp et al. (2014)
	C00806	ALS		Plasma/CSF	UHPLC-MS-MS	Lawton et al. (2012), Toczylowska et al. (2013)
Tyrosine	C00082, C01536	ALS	Increased	CSF	¹H-NMR	Toczylowska et al. (2013)
Urate	C00366	ALS	Decreased	Plasma	UHPLC-MS/GC-MS	Lawton et al. (2014)
Uric acid	C00366	AD	Decreased	Plasma	Uricase colorimetry	Kim et al. (2006), Abraham et al. (2014)
		PD
		ALS
Uridine	C00299	AD	Decreased	CSF	UPLC-ToF-MS	Trushina and Mielke (2014)
Urinary biopyrrin	C00486	PD	Increased	Urine	ELISA	Luan et al. (2015)
Xanthine	C00385	AD	Increased	CSF	LCECA	Kaddurah-Daouk et al. (2013)
Valine	C16436	PD	Increased	Serum	¹H-NMR	Ahmed et al. (2009)
		ALS		CSF		Toczylowska et al. (2013)
Vanillylmandelic acid	C05584	AD	Increased	CSF	LCECA	Kaddurah-Daouk et al. (2013)

AD, Alzheimer's disease; ALS, amyotrophic lateral sclerosis; CSF, cerebrospinal fluid; KEGG, Kyoto Encyclopedia of Genes and Genomes; PD, Parkinson's disease; PMV, post-mortem ventricular; S. Nigra, substantia nigra.

The majority of the biomarkers belonged to acids (31%, 32 metabolites), and amino acids and their derivatives (24%, 25 metabolites) categories (Fig. 1). Apart from these, 6 lipids (including sphingolipids and fatty acids), 6 neurotransmitters (such as dopamine and norephinephrine), 6 alkoloids (including caffeine, betaine, and xanthine derivatives), 6 alcohols (such as glycerol, inositol, and sorbitol), and 20 other metabolites, including amines, vitamins, carbohydrates, ketones, and cholesterols, were proposed as metabolic biomarkers for three neurodegenerative diseases.

FIG. 1.

Distribution of metabolic biomarkers in biochemical categories.

These metabolites were detected in six different fluids: plasma, serum, CSF, post mortem ventricular (PMV)-CSF, urine, and S. Nigra. Different detection techniques (mostly, gas or liquid chromatography coupled with mass spectrometry) were employed in analyses (Table 1).

Six acids (urate, lactate, oxaloacetate, pentothenate, 2-hydroxybutyrate, and 3-phenylpropionate), five alkaloids (caffeine, paraxanthine, theobromine, theophylline, and betaine), two lipids (1-palmitoleoylglycerophosphocholine and stearoyl sphingomyelin), homocysteine, amine, mannose, and riluzole were determined in plasma of patients having ALS disease (Lawton et al., 2014; Zoccolella et al., 2008). Among those, amine and lactate, as well as the amino acids, phenylalanine and tyrosine, glucose, acetone, and inositol were determined in CSF of ALS patients (Blasco et al., 2015; Gray et al., 2015; Toczylowska et al., 2013). Besides CSF and plasma, lactate was determined in serum of patients having ALS disease (Botosoa et al., 2012; Lawton et al., 2014). Ketone, acetone, formate, beta-hydroxybutyrate, and N-acetyl derivatives were also determined in serum of ALS patients (Kumar et al., 2010).

As for AD, the lipids, sphinganine-1-phosphate and lysophosphatidylcholine (2-lysolecitine), together with pipecolic acid and adenosine were determined in serum of patients (González-Domínguez et al., 2015; Liang et al., 2016; Voyle et al., 2016).

Three acids (4-hydroxyphenyllactic acid [4-HPLA], 2-hydroxyphenylacetic acid [2-HPA], vanillylmandelic acid [VMA]), two amino acids (cysteine and beta-alanine), xanthine, and uridine were determined in CSF of AD patients (Kaddurah-Daouk et al., 2013; Trushina and Mielke, 2014). Besides CSF, two amino acids, cysteine and beta-alanine, were also found in plasma of AD patients (Trushina and Mielke, 2014). The neurotransmitters, norephinephrine (noradrenaline) and 3-methoxytyramine, and the amino acid, 5-hydroxytryptophan, were determined in PMV-CSF (Trushina and Mielke, 2014). Anandamide and desmosterol were found in plasma, whereas 5-L-glutamylglycine (5-L-glutamyl-L-cysteinylglycine) was found in urine of AD patients (Cui et al., 2014; Sato et al., 2012; Voyle et al., 2016).

Eight acid salts (suberate, glutarate, malate, succinate, gluconate, methylmalonate, glycolate, and citrate), four alcohols (sorbitol, glycerol, propylene glycol, and galacticol), three amines (methylamine, trimethylamine, and ethanolamine), there amino acids (proline, cysteine, and glutamic acid), the threonate vitamin, lipid hydroperoxides (LOOH), malondialdehyde (MDA), superoxide dismutase (SOD), and 2-oxoisocaproate were detected in plasma of PD patients (Ahmed et al., 2009). The amino acids, isoleucine, arginine, and asparagine, as well as fatty acids were found in serum (Ahmed et al., 2009; González-Domínguez et al., 2014a, 2014b). In addition, asparagine was also determined in CSF of PD patients together with the 3-hydroxyisovaleric acid (Ahmed et al., 2009; González-Domínguez et al., 2014a; Trupp et al., 2014; Trushina and Mielke, 2014). Urinary biopyrrin was detected in urine (Luan et al., 2015); whereas two neurotransmitters, dopamine and γ-aminobutyric acid, were found in S. Nigra of PD patients as well as glutathione, homovanillic acid, and myoinositol (Gröger et al., 2014).

When the expression patterns (i.e., up- or downregulation) of metabolites were examined, we observed that the majority of the ALS biomarkers (riluzole, 1-palmitoleoylglycerophosphocholine, phenylalanine, glucose, tyrosine, ketone, oxaloacetate, mannose, 2-hydroxybutyrate, pentothenate, homocysteine, formate, β-hydroxybutyrate, inositol, stearoyl sphingomyelin, sphinganine-1-phosphate, lactate, and acetone) were over-expressed in the diseased state; whereas N-acetyl derivatives, 3-phenylpropionate, amine, betaine, and urate levels were decreased. In addition, low levels of alkaloids (caffeine, paraxanthine, theobromine, and theophylline), and hippurate were observed in ALS patients (Blasco et al., 2015; Botosoa et al., 2012; Gray et al., 2015; Kumar et al., 2010; Lawton et al., 2012, 2014; Toczylowska et al., 2013; Zoccolella et al., 2008).

Among the AD biomarkers, three acids (4-HPLA, 2-HPA, and VMA), two amino acid derivatives (5-L-glutamylglycine and 5-hydroxytryptophan), xanthine, and adenosine were over-expressed in the diseased state; whereas three neurotransmitters (anandamide, norepinephrine [NE], and 3-methoxytyramine), two amino acids (cysteine and beta-alanine), pipecolic acid, desmosterol, uridine, and lysophosphatidyl choline levels were decreased (Cui et al., 2014; González-Domínguez et al., 2015; Kaddurah-Daouk et al., 2013; Liang et al., 2016; Sato et al., 2012; Trushina and Mielke, 2014; Voyle et al., 2016).

Increases in levels of lipid hydroperoxides, and saturated or polysaturated fatty acids together with isoleucine, proline, γ-aminobutyric acid, homovanillic acid, MDA, SOD, 2-oxoisocaproate, and urinary biopyrrin were observed in PD patients. However, the amino acids (arginine, glutathione, cysteine, and glutamic acid), amines (methylamine, trimethylamine, and ethanolamine), alcohols (galacticol and glycerol), acid salts (suberate, glutarate, gluconate, malate, glycolate, succinate, citrate, and methylmalonate), 3-hydroxyisovaleric acid, threonate, and dopamine levels were decreased in PD. Moreover, myoinositol was decreased in S. Nigra, but it increased in blood plasma. Asparagine was increased in CSF and blood plasma, but it decreased in serum in PD patients. Elevated sorbitol and propylene glycol levels were also observed in PD patients (Ahmed et al., 2009; de Farias et al., 2016; González-Domínguez et al., 2014a; Gröger et al., 2014; Luan et al., 2015; Marklund and Marklund, 1974; Trupp et al., 2014; Wang et al., 2014).

To discern the similarities and differences of disease signatures at metabolome level, all possible comparisons were performed and results were represented by a Venn diagram (Fig. 2). In addition, a heat map (Fig. 3) was constructed to evaluate any possible correlation between the expression patterns of the metabolites.

FIG. 2.

Venn diagram representing the metabolic signature similarities and differences among the three neurodegenerative diseases examined in the present study. AD, Alzheimer's disease; ALS, amyotrophic lateral sclerosis; PD, Parkinson's disease.

FIG. 3.

A heat map analysis displaying the correlations according to the direction of change (increased, decreased, or both) in the regulation of the putative metabolite biomarkers of the three neurodegenerative diseases examined in the present study.

The comparative analyses of metabolic biomarkers indicated that there were seven mutual metabolites (uric acid, choline, creatine, L-glutamine, alanine, creatinine, and N-acetyl-L-aspartate) among three neurodegenerative diseases. Among these metabolites, uric acid and N-acetyl-L-aspartate levels were decreased in all diseases (Fig. 3). The choline, L-glutamine, and alanine levels were increased in AD and ALS, but they were decreased in PD. The creatine and creatinine levels were also increased in AD and decreased in PD, but their expression patterns in ALS were not stable and fluid dependent (Abraham and Drory, 2014; Ahmed et al., 2009; Blasco et al., 2015; Botosoa et al., 2012; Cui et al., 2014; González-Domínguez et al., 2014a; Gröger et al., 2014; Jukarainen et al., 2008; Kantarci et al., 2000; Kim et al., 2006; Kumar et al., 2010; Lawton et al., 2012; Toczylowska et al., 2013; Trupp et al., 2014; Trushina and Mielke, 2014; Wuolikainen et al., 2011).

In addition, 24S-hydroxycholesterol, pyroglutamate, isocitrate, and methionine were the mutual metabolites for both AD and PD (Fig. 2). Methionine levels were decreased in both diseases, and pyroglutamate levels were increased. On the other hand, 24S-hydroxycholesterol and isocitrate levels were increased in AD, but they were decreased in PD.

Three metabolites (histidine, α-ketoglutarate, and carnitine) were determined as mutual biomarkers of AD and ALS. α-ketoglutarate levels were increased in both diseases. However, elevated levels of histidine and carnitine were observed in ALS; whereas the expressions of these metabolites were decreased in AD.

Nine metabolites were indicated as mutual biomarkers of PD and ALS. Among those, glycine, aspartate, valine, and pyruvate levels were increased, and tryptophan levels were decreased in both diseases. On the other hand, lysine and acetate levels were elevated in ALS, but they were decreased in PD (Fig. 3). The levels of ascorbate and glutamate were dependent on the fluid. For instance, glutamate levels were decreased in the CSF of PD patients and the plasma of ALS patients; however, elevated glutamate levels were observed in the plasma of PD and the CLS of ALS patients.

Besides the mutual metabolites, the majority (78 of 101) of the biomarkers were disease specific (Figs. 2 and 3): 17, 34, and 27 metabolites came into prominence in only AD, PD, and ALS, respectively. The biological roles of these metabolites were determined via metabolic pathway enrichment analyses (Table 2). Five pathways were significantly enriched for all diseases: three amino acid pathways (arginine and proline metabolism; alanine, aspartate, and glutamate metabolism; and glycine, serine, and threonine metabolism), ATP binding cassette (ABC) transporters, and purine nucleotide metabolism.

Table 2.

Significantly Enriched Metabolic Pathways for Three Neurodegenerative Diseases (p Value <0.05)

KEGG pathway ID	Pathway	p value
AD-specific
hsa04080	Neuroactive ligand-receptor interaction	0.0002
hsa00350	Tyrosine metabolism	0.0003
hsa04270	Vascular smooth muscle contraction	0.0053
hsa00360	Phenylalanine metabolism	0.0172
hsa00240	Pyrimidine metabolism	0.0275
PD-specific
hsa00630	Glyoxylate and dicarboxylate metabolism	0.0002
hsa00052	Galactose metabolism	0.0002
hsa00020	Citrate cycle (TCA cycle)	0.0003
hsa00561	Glycerolipid metabolism	0.0013
hsa00640	Propanoate metabolism	0.0018
hsa00250	Alanine, aspartate, and glutamate metabolism	0.0117
hsa00350	Tyrosine metabolism	0.0151
hsa00410	Beta-alanine metabolism	0.0192
hsa00620	Pyruvate metabolism	0.0204
hsa00680	Methane metabolism	0.0228
hsa00650	Butanoate metabolism	0.0310
hsa00280	Valine, leucine, and isoleucine degradation	0.0324
hsa00053	Ascorbate and aldarate metabolism	0.0417
hsa00071	Fatty acid degradation	0.0466
ALS-specific
hsa00232	Caffeine metabolism	<0.0001
hsa00072	Synthesis and degradation of ketone bodies	0.0004
hsa00052	Galactose metabolism	0.0014
hsa00250	Alanine, aspartate, and glutamate metabolism	0.0076
hsa00910	Nitrogen metabolism	0.0089
hsa00010	Glycolysis/gluconeogenesis	0.0125
hsa00620	Pyruvate metabolism	0.0133
hsa00640	Propanoate metabolism	0.0166
hsa00630	Glyoxylate and dicarboxylate metabolism	0.0243
hsa00360	Phenylalanine metabolism	0.0264
Common
hsa00330	Arginine and proline metabolism	0.0004
hsa00250	Alanine, aspartate, and glutamate metabolism	0.0009
hsa00260	Glycine, serine, and threonine metabolism	0.0037
hsa02010	ABC transporters	0.0121
hsa00230	Purine metabolism	0.0126

ABC, ATP binding cassette; TCA, tricarboxylic acid.

When disease-specific metabolic pathways were analyzed, several amino acid pathways were significantly enriched in all diseases: in AD, tyrosine and phenylalanine; in PD, beta-alanine; and in ALS, phenylalanine. Moreover, pyrimidine nucleotide metabolism, the neuroactive ligand-receptor interaction pathway, and the vascular smooth muscle contraction process were significantly enriched in AD. On the other hand, seven pathways of carbohydrate metabolism (citrate cycle, pyruvate metabolism, ascorbate and aldarate metabolism, galactose metabolism, glyoxylate and dicarboxylate metabolism, propanoate metabolism, and butanoate metabolism) and two pathways from lipid metabolism (glycerolipid metabolism and fatty acid degradation) came into prominence in PD. Similar to PD-specific pathways, ALS were also enriched with pathways of carbohydrate metabolism (glycolysis/gluconeogenesis, galactose metabolism, pyruvate metabolism, propanoate metabolism, glyoxylate and dicarboxylate metabolism) and lipid metabolism (synthesis and degradation of ketone bodies). However, caffeine metabolism and nitrogen metabolism pathways were specific to ALS (Table 2).

Ranking the disease-metabolite-pathway associations according to their statistical significance pointed out the importance of membrane transport (through ABC transporters), and arginine and proline amino acids in all diseases (Fig. 4). Neuroactive ligand-receptor interaction and tyrosine metabolism came into prominence in AD. Glutamate metabolism, citrate cycle (tricarboxylic acid [TCA] cycle), glycoxylate and dicarboxylate metabolism, and butanoate metabolism were significantly enriched in PD; whereas pyruvate metabolism, alanine, aspartate and glutamate metabolism, glycosis/glucogenesis, and glycoxylate and dicarboxylate metabolism were distinguished in the case of ALS.

FIG. 4.

A schematic representation of the disease-metabolite-pathway associations according to their significance for the three neurodegenerative diseases examined in the present study. Node color: the p value significance (sliding through light to dark gray showed increased statistical significance), circle diameter: pathway impact (larger circle diameters correspond to higher degree values).

We also analyzed the metabolites within the human serum metabolomics network (Yazdani et al., 2016b) and identified causal network parameters (in and out degrees) to determine the effects of metabolites within the networks (Yazdani et al., 2016a). Interestingly, common metabolites between PD and ALS (tryptophan, glycine, valine, and lysine) had higher influence in the metabolomics network in comparison with common metabolites between AD and PD or AD and ALS (Table 3). However, neither high influential metabolites (active metabolites) nor metabolites in modules in the metabolomics causal network were among the common metabolites between these diseases.

Table 3.

Causal Network Parameters for the Common Metabolites Among Alzheimer's Disease, Amyotrophic Lateral Sclerosis, and Parkinson's Disease

Name of the metabolite	Out degree	In degree
Common metabolites of all diseases (AD, PD, and ALS)
Alanine	1	1
Choline	4	3
Common metabolites of AD and PD
Pyroglutamine	1	1
Citrate	1	0
Common metabolites of PD and ALS
Tryptophan	5	2
Glycine	4	2
Valine	3	2
Lysine	3	1
Carnitine	1	1

Discussion

Metabolomics offers the promise to identify at a systems scale the molecular signatures of cellular processes, molecular interactions, and metabolic pathways that are attendant to complex diseases such as neurodegenerative disorders with multi-factorial underpinnings. In the present study, we have undertaken research to evaluate the hitherto proposed metabolic biomarkers of three common neurodegenerative diseases (AD, PD, and ALS) by using pathway enrichment analyses and as seen through a lens of systems approach. We analyzed not only metabolites but also the significantly altered metabolic pathways.

The comparative analyses of metabolic biomarkers pointed out to the presence of seven mutual metabolites (uric acid, choline, creatine, L-glutamine, alanine, creatinine, and N-acetyl-L-aspartate) among three neurodegenerative diseases, whereas the majority (78 of 101) of the biomarkers was disease specific. Among the mutual metabolites, uric acid and N-acetyl-L-aspartate levels were decreased in all diseases: Choline, L-glutamine, and alanine levels were increased in AD and ALS but they were decreased in PD. Creatine and creatinine levels were increased in AD and were decreased in PD, but their expression levels in ALS were not stable and sample fluid dependent (Fig. 3). The disease-specific trends in expression profiles of these metabolites best suggested them as a set of biomarkers (i.e., systems biomarkers instead of single biomarkers).

The in-depth analyses of biomarkers indicated the importance of amino acids and derivatives in the diagnosis and treatment of the three neurodegenerative diseases. Twenty-five amino acids and their derivatives were identified as biomarkers for these three diseases. Among them, alanine, L-glutamine, and N-acetyl-L-aspartate were common to all three diseases. N-acetyl-L-aspartate levels were decreased in all diseases, whereas alanine and L-glutamine levels were increased in AD and ALS in contrast to PD. In addition, pathway enrichment analysis results indicated the dominance of three amino acid pathways in all diseases: arginine and proline metabolism; aspartate and glutamate metabolism; and glycine, serine, and threonine metabolism. Moreover, phenyalanine metabolism was significantly altered in AD and ALS.

The potential of proline-rich polypeptides as therapeutic agents in neurodegenerative diseases was previously reported (Gladkevich et al., 2007). Nitric oxide, synthesized from L-arginine, affects brain functions by generating nitroxidative stress, and its elevation may play a causative role in the development of neurodegenerative diseases, including AD and PD (Virarkar et al., 2013). Many neurodegenerative diseases were associated with neurotransmitter levels, and glutamate is the most prevalent neurotransmitter in the brain. Glutamate concentration was raised in ALS patients, and glutamate inhibitors (for example, riluzole) were presented as an alternative cure for ALS patients (Sako et al., 2016). In addition, aberrant glutamate signaling in the basal ganglia may be involved in PD development (Gasparini et al., 2013). Glycine behaves as a neurotransmitter (López-Corcuera et al., 2001), and phenylalanine have a significant role in neurotransmitter regulation (Humphries et al., 2008). Therefore, as neurotransmitters, or neurotransmitter regulators, the role of amino acids (glutamate, glycine, and phenylalanine in particular) in neurodegenerative diseases is evident.

According to metabolic pathway enrichment analyses, ABC transporters were significantly enriched in all diseases. Forty-nine different genes encoding ABC transporters were identified in the human genome, and they were classified into seven sub-groups (named ABCA to ABCG). ABC transporters mediate the export of numerous substances (peptides, drugs or drug conjugates, and hydrophobic molecules) through the blood–brain barrier, and they are expressed in each cell type in the brain (Pahnke et al., 2014; Schumacher et al., 2012). The relationship between ABC transporters and neurodegenerative diseases was commonly based on the function of ABC transporters or impairment of their function in the blood–brain barrier. The aberrant ABC transporter expression has been reported in several neurodegenerative diseases, including AD, PD, and ALS. ABC transporters play an essential role in AD disease initiation and development by facilitating Aβ exports from the brain barriers, which can cause inadequate Aβ export (Pahnke et al., 2014). In addition, the expression of ABC transporters (and thereby their function) was increased at the blood–brain and blood–spinal cord barriers in ALS patients (Jablonski et al., 2015).

The links between diseases and altered metabolism are supported by many observations. For example, metabolic changes can affect disease progression, response to therapy, and patients' clinical outcome. Pathways that have essential roles in disease progression of ALS and PD generally belong to carbohydrate metabolism. Pathway enrichment analyses in the present study indicated that glycoxylate and dicarboxylate metabolism, galactose metabolism, citrate cycle (TCA cycle), glycolysis/gluconeogenesis, and pyruvate metabolism were the most influenced metabolic pathways in regards to disease pathogenesis. These pathways share mutual hub metabolites, like pyruvate and oxaloacetate, which are the main compounds for TCA cycle, gluconeogenesis, and glycoxylate and dicarboxylate metabolism (Garrett and Grisham, 2002). Pyruvate and pyruvate-oxaloacetate have neuroprotective effects on neuro-metabolism via increasing the brain-to-blood glutamate efflux (Carvalho et al., 2011; Zilberter et al., 2015; Zlotnik et al., 2007).

Since accumulation of glutamate can cause neuro-excitotoxic effects, its accumulation was repressed via scavenging of it by pyruvate and oxaloacetate (Castillo et al., 2016; Danbolt et al., 2001; Zlotnik et al., 2012). The weight of the evidence suggests that particularly carbohydrate and energy metabolism are the main components of PD, AD, and ALS development. Fluctuations in carbohydrate metabolism may also lead to an imbalance of glucose tolerance and insulin resistance. Indeed, the metabolic homeostasis of leptin, ghrelin, insulin, and IGF-1 provides a neuroprotective and neurotrophic effect in AD and PD (Procaccini et al., 2016).

The levels of these compounds were directly associated with carbohydrate and energy metabolisms in our analyses. Insulin resistance impairs carbohydrate metabolism and mitochondrial function, which results in elevated oxidative stress and partial cognitive functions during the progression of AD (Abolhassani et al., 2016). Therefore, it can be hypothesized that metabolic dysregulation and neurodegeneration have a strong relationship in the pathophysiology of neurodegenerative diseases.

When AD-specific pathways were considered, tyrosine metabolism and neuroactive ligand-receptor interactions were significantly altered in AD in the present study. The metabolites 3-methoxythramine (3-MT), VMA, 4-HPLA, and 2-HPA were components of the tyrosine metabolism. Sphinganine-1-phosphate, anandamide, beta-alanine, adenosine, and NE were enriched with neuroactive ligand-receptor interactions.

The levels of 3-MT, which is a catabolic intermediate of dopamine, were decreased 92% in the striatum of PD patients (Loeffler, 1995). VMA (also known as 3-Methoxy-4-hydroxy-mandelic acid) is one of the end products of the catabolism of catecholamines (epinephrine, NE, and dopamine). VMA levels in CSF were found to be dramatically increased in AD, whereas NE levels were significantly decreased. Kaddurah-Daouk et al. (2013) emphasized the correlations between VMA and amyloid-β (Ab42); between VMA, 4-HPLA, 2-HPA, and total tau; and between VMA, 4-HPLA, and phosphorylated tau. 2-HPA is involved in 4-hydroxyphenylpyruvate dioxygenase activity, which is a key enzyme in the degradation of tyrosine.

The hydroxyphenylacetate group was used as a urinary biomarker for diagnosis of neuropsychiatric disorders such as major depressive disorder (Zheng et al., 2013). 4-HPLA, which was increased in AD, is a tyrosine metabolite formed by gut microbes, such as lactobacilli (Mu et al., 2010). 4-HPLA was significantly reduced in Huntington disease (HD) in comparison to both normal control and premanifest HD subjects (Rosas et al., 2015). Hence, its reduction in AD might further indicate an AD/gut interaction like in HD.

Sphinganine 1-phosphate, which is an intermediate of glycosphingolipid and sphingolipid metabolisms, is one of the key components for the neuroactive ligand-receptor interactions. Sphingolipid metabolism was effective in neural cells (van Echten-Deckert and Herget, 2006). Studies have shown alterations in phospholipid metabolite levels in AD, and fluctuations in phospholipid metabolite levels were correlated with neuro-pathology that was indicative of the disease and cognitive decline (Cuénod, 1995; Klunk, 1996; Pettegrew, 1987). They may be elevated due to loss of spingosine-1-phosphate, which is a protective factor (Couttas, 2014) that is employed in neuroactive ligand-receptor interaction pathways. Levels of ceramide, the pivotal molecule of sphingolipid metabolism with sphingosine-1-phosphate in neural signaling and function (Colombaioni and Garcia-Gil, 2004), were also reduced.

Anandamide, beta-alanine, and adenosine were other components of the neuroactive ligand-receptor interaction pathway, which were significantly altered in neurodegenerative diseases. Ab42-dependent brain anandamide (AEA) mobilization contributes to cognitive dysfunction in AD (Jung, 2012). It was suggested that there is a relationship between AEA and Notch signaling, which is essential for neural cell maintenance in the adult brain, and prolonged formation and development of nerve tissues throughout life. AEA might offer neuroprotection of neurodegeneration. Enhancement of AEA might enhance Notch signaling and could provide a therapy to ease symptoms and hold off progression of neurodegeneration (Khaspekov, 2004; Ramírez, 2005; Tanveer, 2012).

Evaluation of AD-specific pathways of neuroactive ligand-receptor interaction and tyrosine metabolism together suggested that depletion of NE, 3-methoxytyramine, and metabolites within the purine pathway were associated with neurofibrillary tangle formation and/or Aβ deposition (Trushina and Mielke, 2014). If we sum up all, these results indicated the perturbations of tyrosine metabolism, neuroactive ligand-receptor interactions, and phospholipids metabolism in AD, which may be important for clinical diagnosis and treatment. They could be helpful for an understanding of metabolic pathogenesis of AD.

In the present study, according to an enrichment analysis of common metabolites, lipid metabolism was also responsible for PD. Two pathways from lipid metabolism (glycerolipid metabolism and fatty acid degradation) came into prominence in PD. Lipid metabolism was affected from levels of glycerol, propylene glycol, and fatty acid compounds in PD. Glycerol levels were decreased (Ahmed et al., 2009); however, propylene glycol and fatty acid levels were increased in PD (González-Domínguez et al., 2014b). These findings were supported by the previous studies that reported elevated levels of polyunsaturated fatty acids in the brain (Dias et al., 2013; Liu et al., 2008). Highly unsaturated fatty acids can cause lipid peroxidation and promote generation of toxic products under oxidative stress circumstances, which results in neural damage and promotes PD improvement.

According to pathway enrichment analysis, sorbitol, glycerol, myoinositol, and galacticol were also members of the galactose metabolism, which may be considered the alternative way to overcome mitochondrial dysfunction and to create energy. Moreover, the elevated levels of fatty acids in serum of PD (González-Domínguez et al., 2014b) may also be considered as the usage of fatty acids as an alternative carbon source to increase energy demand (Prins, 2008).

Similar to PD-specific pathways, ALS was also enriched with pathways from lipid metabolism (synthesis and degradation of ketone bodies). However, caffeine metabolism and nitrogen metabolism pathways were specific to ALS.

Ketone bodies (acetoacetate, beta-hydroxybutyrate, and acetone) are produced when fatty acid is broken down in excess (Fukao et al., 2004; Sass, 2012). Acetoacetate and beta-hydroxybutyrate are synthesized from acetyl-CoA, mainly in the mitochondria of liver cells when carbohydrates are so scarce that energy must be obtained from breaking down of fatty acids; acetone is formed by spontaneous decarboxylation of acetoacetate (Fukao et al., 2004; Sass, 2012). Ketone body synthesis in the liver is effectively irreversible, because the enzyme that catalyzes the conversion of acetoacetate to acetoacetyl-CoA is not present in liver cells (Fukao et al., 2004; Sass, 2012). During prolonged starvation, ketone bodies can provide up to two-thirds of the brain's energy requirements (Sass, 2012). When the expression patterns of metabolites were examined, we observed that the majority of the ALS biomarkers (acetone, beta-hydroxybutyrate) were over-expressed in the diseased state (Kumar et al., 2010).

Formate, which was significantly increased in ALS disease, is a byproduct in the production of acetate. It provides both metabolic acidosis and damaged mitochondrial electron transport and energy production by inhibiting cytochrome oxidase activity, the terminal electron acceptor of the electron transport chain (Kumar et al., 2010; Nicholls et al., 1975). Since cytochrome oxidase inhibition by formate leads to partial depletion of ATP or reducing energy concentration, cell functions cannot be maintained and cell death occurs (Kumar et al., 2010). In addition to this, the increased production of cytotoxic reactive oxygen species (ROS) derived from inhibition of cytochrome oxidase by formate may also cause cell death (Kumar et al., 2010). ROS, which leads to the disruption in mitochondrial electron transport and energy production, also increases levels of formate (Kumar et al., 2010).

Caffeine was determined in plasma of patients having ALS disease (Lawton et al., 2012). Caffeine affects multiple mechanisms in the brain such as adenosine, ryanodine, and γ-aminobutyric acid receptors and cyclic nucleotide phosphodiesterase isoenzymes. Caffeine also stimulates the activity of the central nervous system, and it is associated with changes in the rates of cerebral energy metabolism (Luong and Nguyen, 2015). Several studies suggested that caffeine has significant effects on ALS through non-genomic factors, including glutamate, Na⁺-K⁺-ATPase, the mammalian family of mitogen-activated protein kinases, prostaglandins and cyclooxygenase-2, and reactive oxygen species (Luong and Nguyen, 2015). On the other hand, a further investigation of the effect of caffeine in neurodegenerative diseases is needed.

A further understanding about metabolic programming is definitely necessary for efficient therapy progression. In as much as, any pathophysiological mechanism lead by disease will predictably cause associated changes in the concentrations of particular metabolites; therefore, nowadays, exploration of metabolites to evaluate disease(s) detection, monitoring, and treatment is obviously a very promising and effective direction. Furthermore, the metabolic biomarkers will lead to develop new strategies that enable more precise diagnosis of three neurodegenerative diseases, and should be regarded as subjects of drug repositioning studies, which employ known drugs and compounds to treat new indications.

Conclusions

The current analysis identified 101 metabolites as potential biomarkers of AD, PD and ALS. The comparative analyses indicated the mutual and disease-specific metabolites, as well as metabolic pathways. In the light of these data, we defined a set of biomarkers (uric acid, choline, creatine, L-glutamine, alanine, creatinine, and N-acetyl-L-aspartate) as putative systems biomarkers, which may inform the future strategies for diagnostic or prognostic biomarkers. The importance of amino acids, ABC transporters, neuroactive ligand-receptor interactions, and caffeine metabolism as well as the carbohydrate, lipid, and energy metabolism were notable as an outcome of the present study. For these neurodegenerative diseases, to overcome inadequate glucose supply and energy crisis in neurodegeneration, alanine, aspartate and glutamate metabolism, and purine metabolism appear to be alternative pathways. Further analysis of metabolites and enzymes of the proposed pathways could be helpful for an understanding of metabolic pathogenesis and therapy of AD, PD, and ALS diseases.

These new observations collectively underscore the importance of metabolite-based biomarker research in deciphering the elusive pathophysiology of neurodegenerative diseases. Future research investments in post-genomics fields as metabolomics would provide new insights on AD, PD, and ALS that continue to place a significant burden on global health.

Footnotes

Acknowledgments

The authors thank the anonymous reviewers and the editor for their comments to improve this article. No funding was received in support of this research.

Author Disclosure Statement

The authors declare that no conflicting financial interests exist.

Abbreviations Used

References

Abolhassani

, Leon

, Sheng

, et al. (2016). Molecular pathophysiology of impaired glucose metabolism, mitochondrial dysfunction, and oxidative DNA damage in Alzheimer's disease brain. Mech Ageing Dev [Epub ahead of print]; DOI: 10.1016/j.mad.2016.05.005.

Abraham

, and Drory

. (2014). Influence of serum uric acid levels on prognosis and survival in amyotrophic lateral sclerosis: A meta-analysis. J Neurol, 261, 1133–1138.

Ahmed

SSSJ

, Santosh

, Kumar

, and Christlet

HTT

. (2009). Metabolic profiling of Parkinson's disease: Evidence of biomarker from gene expression analysis and rapid neural network detection. J Biomed Sci, 16, 1–12.

Andalib

, Vafaee

, and Gjedde

. (2014). Parkinson's disease and mitochondrial gene variations: A review. J Neurol Sci, 346, 11–19.

Bardou

, Mariette

, Escudié

, Djemiel

, and Klopp

. (2014). Jvenn: An interactive Venn diagram viewer. BMC Bioinform, 15, 293.

Blasco

, Corcia

, Moreau

, et al. (2010). H-NMR-based metabolomic profiling of CSF in early amyotrophic lateral sclerosis. PLoS One, 5, e13223.

Blasco

, Corcia

, Pradat

, et al. (2013). Metabolomics in cerebrospinal fluid of patients with amyotrophic lateral sclerosis: An untargeted approach via high-resolution mass spectrometry. J Proteome Res, 12, 3746–3754.

Blasco

, Nadal‐Desbarats

, Pradat

, et al. (2015). Biomarkers in amyotrophic lateral sclerosis: Combining metabolomic and clinical parameters to define disease progression. Eur J Neurol, 23, 346–353.

Blasco

, Patin

, Hounoum

, et al. (2016). Metabolomics in amyotrophic lateral sclerosis: How far can it take us?. Eur J Neurol, 23, 447–454.

10.

Bloom

. (2014). Amyloid-β and tau the trigger and bullet in Alzheimer disease pathogenesis. JAMA Neurol, 71, 505–508.

11.

Botosoa

, Zhu

, Marbeuf-Gueye

, et al. (2012). NMR metabolomic of frontal cortex extracts: First study comparing two neurodegenerative diseases, Alzheimer disease and amyotrophic lateral sclerosis. IRBM, 33, 281–286.

12.

Carvalho

ASR

, Torres

, Persike

, et al. (2011). Neuroprotective effect of pyruvate and oxaloacetate during pilocarpine induced status epilepticus in rats. Neurochem Int, 58, 385–390.

13.

Castillo

, Loza

, Mirelman

, et al. (2016). A novel mechanism of neuroprotection: Blood glutamate grabber. J Cereb Blood Flow Metab, 36, 292–301.

14.

Chagoyen

, and Pazos

. (2011). MBRole: Enrichment analysis of metabolomic data. Bioinformatics, 27, 730–731.

15.

Chio

, Logroscino

, Traynor

, et al. (2013). Global epidemiology of amyotrophic lateral sclerosis: A systematic review of the published literature. Neuroepidemiology, 41, 118–130.

16.

Colombaioni

, and Garcia-Gil

. (2004). Sphingolipid metabolites in neural signalling and function. Brain Res, 46, 328–355.

17.

Cornutiu

. (2015). The epidemiological scale of Alzheimer's disease. J Clin Med Res, 7, 657–666.

18.

Couttas

, Kain

, Daniels

, et al. (2014). Loss of the neuroprotective factor Sphingosine 1-phosphate early in Alzheimer's disease pathogenesis. Acta Neuropathol Commun, 2, 1.

19.

Cuénod

, Kaplan

, Michot

, et al. (1995). Phospholipid abnormalities in early Alzheimer's disease: In vivo phosphorus 31 magnetic resonance spectroscopy. Arch Neurol, 52, 89–94.

20.

Cui

, Liu

, Wang

, et al. (2014). Lysophosphatidylcholine and amide as metabolites for detecting alzheimer disease using ultrahigh-performance liquid chromatography-quadrupole time-of-flight mass spectrometry-based metabonomics. J Neuropathol Exp Neurol, 73, 954–963.

21.

Danbolt

. (2001). Glutamate uptake. Prog Neurobiol, 65, 1–105.

22.

de Farias

, Maes

, Bonifácio

, et al. (2016). Highly specific changes in antioxidant levels and lipid peroxidation in Parkinson's disease and its progression: Disease and staging biomarkers and new drug targets. Neurosci Lett, 617, 66–71.

23.

Delenclos

, Jones

, McLean

, and Uitti

. (2016). Biomarkers in Parkinson's disease: Advances and strategies. Parkinsonism Relat Disord, 1, 106–110.

24.

Dias

, Junn

, and Mouradian

. (2013). The role of oxidative stress in Parkinson's disease. J Parkinsons Dis, 3, 461–491.

25.

Embade

, Mariño

, Diercks

, et al. (2016). Metabolic characterization of advanced liver fibrosis in HCV patients as studied by serum 1H-NMR spectroscopy. PLoS One, 11, e0155094.

26.

Ferri

, Prince

, Brayne

, et al. (2006). Global prevalence of dementia: A Delphi consensus study. Lancet, 366, 2112–2117.

27.

Fukao

, Lopaschuk

, and Mitchell

. (2004). Pathways and control of ketone body metabolism: On the fringe of lipid biochemistry. Prostaglandins Leukot Essent Fatty Acids, 70, 243–251.

28.

Garrett

, and Grisham

. (2002). Principles of Biochemistry with a Human Focus. Brooks/Cole, Thomson Learning, USA, 578–585.

29.

Gasparini

, Di Paolo

, and Gomez-Mancilla

. (2013). Metabotropic glutamate receptors for Parkinson's disease therapy. Parkinsons Dis, 2013, Article ID 196028.

30.

Ghio

, Kamp

, Cauchi

, Giese

, and Vassallo

. (2016). Interaction of α-synuclein with biomembranes in Parkinson's disease—Role of cardiolipin. Prog Lipid Res, 61, 73–82.

31.

Gladkevich

, Bosker

, Korf

, Yenkoyan

, Vahradyan

, and Aghajanov

. (2007). Proline-rich polypeptides in Alzheimer's disease and neurodegenerative disorders—Therapeutic potential or a mirage?. Prog Neuropsychopharmacol Biol Psychiatry, 31, 1347–1355.

32.

González-Domínguez

, García

, García-Barrera

, Barbas

, and Gómez-Ariza

. (2014a). Metabolomic profiling of serum in the progression of Alzheimer's disease by capillary electrophoresis-mass spectrometry. Electrophoresis, 35, 3321–3330.

33.

González-Domínguez

, García-Barrera

, and Gómez-Ariza

. (2014b). Using direct infusion mass spectrometry for serum metabolomics in Alzheimer's disease. Anal Bioanal Chem, 406, 7137–7148.

34.

González-Domínguez

, García-Barrera

, and Gómez-Ariza

. (2015). Metabolite profiling for the identification of altered metabolic pathways in Alzheimer's disease. J Pharm Biomed Anal, 107, 75–81.

35.

Gray

, Larkin

, Claridge

, Talbot

, Sibson

, and Turner

. (2015). The longitudinal cerebrospinal fluid metabolomic profile of amyotrophic lateral sclerosis. Amyotroph Lateral Scler Frontotemporal Degener, 16, 456–463.

36.

Gröger

, Kolb

, Schaöfer

, and Klose

. (2014). Dopamine reduction in the substantia nigra of Parkinson's disease patients confirmed by in vivo magnetic resonance spectroscopic imaging. PLoS One, 9, e84081.

37.

Hatano

, Saiki

, Okuzumi

, Mohney

, and Hattori

. (2016). Identification of novel biomarkers for Parkinson's disease by metabolomic technologies. Neurosurg Psychiatry, 87, 295–301.

38.

Humphries

, Pretorius

, and Naudé

. (2008). Direct and indirect cellular effects of aspartame on the brain. Eur J Clin Nutr, 62, 451–462.

39.

Jablonski

, Miller

, Pasinelli

, and Trotti

. (2015). ABC transporter-driven pharmacoresistance in amyotrophic lateral sclerosis. Brain Res, 14, 1–14.

40.

Jukarainen

, Korhone

, Laakso

, et al. (2008). Quantification of 1H NMR spectra of human cerebrospinal fluid: A protocol based on constrained total-line-shape analysis. Metabolomics, 4, 150–160.

41.

Jung

, Astarita

, Yasar

, et al. (2012). An amyloid β 42-dependent deficit in anandamide mobilization is associated with cognitive dysfunction in Alzheimer's disease. Neurobiol Aging, 33, 1522–1532.

42.

Kaddurah-Daouk

, Zhu

, Sharma

, et al. (2013). Alterations in metabolic pathways and networks in Alzheimer's disease. Transl Psychiatry, 3, e244.

43.

Kantarci

, Jack

Jr. , Xu

, et al. (2000). Regional metabolic patterns in mild cognitive impairment and Alzheimer's disease a 1H MRS study. Neurology, 55, 210–217.

44.

Khaspekov

, Brenz Verca

, Frumkina

, Hermann

, Marsicano

, and Lutz

. (2004). Involvement of brain-derived neurotrophic factor in cannabinoid receptor‐dependent protection against excitotoxicity. Eur J Neurosci, 19, 1691–1698.

45.

Kim

, Taylor

, Ganti

, Guo

, Osier

, and Weiss

. (2011). Urine metabolomic analysis identifies potential biomarkers and pathogenic pathways in kidney cancer. OMICS, 15, 293–303.

46.

Kim

, Pae

, Yoon

, et al. (2006). Decreased plasma antioxidants in patients with Alzheimer's disease. Int J Geriatr Psychiatry, 21, 344–348.

47.

Klunk

, Panchalingam

, McClure

, and Pettegrew

. (1996). Quantitative 1H and 31P MRS of PCA extracts of postmortem Alzheimer's disease brain. Neurobiol Aging, 17, 349–357.

48.

Kumar

, Bala

, Kalita

, et al. (2010). Metabolomic analysis of serum by H NMR spectroscopy in amyotrophic lateral sclerosis. Clin Chim Acta, 411, 563–567.

49.

Lawton

, Brown

, Alexander

, et al. (2014). Northeast ALS Consortium. Plasma metabolomic biomarker panel to distinguish patients with amyotrophic lateral sclerosis from disease mimics. Amyotroph Lateral Scler Frontotemporal Degener, 15, 362–370.

50.

Lawton

, Cudkowicz

, Brown

, et al. (2012). Biochemical alterations associated with ALS. Amyotroph Lateral Scler, 13, 110–118.

51.

Lee

CYJ

, Seet

, Huang

, Long

, and Halliwell

. (2009). Different patterns of oxidized lipid products in plasma and urine of dengue fever, stroke, and Parkinson's disease patients: Cautions in the use of biomarkers of oxidative stress. Antioxid Redox Signal, 11, 407–420.

52.

Leoni

, and Caccia

. (2011). Oxysterols as biomarkers in neurodegenerative diseases. Chem Phys Lipids, 164, 515–524.

53.

Liang

, Liu

, Zhang

, Jiang

, Xing

, and Zhang

. (2016). Discovery of serum metabolites for diagnosis of progression of mild cognitive impairment to Alzheimer's disease using an optimized metabolomics method. RSC Adv, 6, 3586–3591.

54.

Lindeque

, Jansen van Rensburg

, Louw

, et al. (2015). Obesity and metabolomics: Metallothioneins protect against high-fat diet-induced consequences in metallothionein knockout mice. OMICS, 19, 92–103.

55.

Liu

, Yamada

, Maruyama

, and Osawa

. (2008). Formation of dopamine adducts derived from brain polyunsaturated fatty acids: Mechanism for Parkinson disease. J Biol Chem, 283, 34887–34895.

56.

Loeffler

, LeWitt

, DeMaggio

, Juneau

, Milbury

, and Matson

. (1995). Markers of dopamine depletion and compensatory response in striatum and cerebrospinal fluid. J Neural Transm Park Dis Dement Sect, 9, 45–53.

57.

Lopes

, Geloneze

, Pareja

, Calixto

, Ferreira

, and Marsaioli

. (2015). Blood metabolome changes before and after bariatric surgery: A (1)H NMR-based clinical investigation. OMICS, 19, 318–327.

58.

López-Corcuera

, Geerlings

, and Aragón

. (2001). Glycine neurotransmitter transporters: An update. Mol Membr Biol, 18, 13–20.

59.

Luan

, Liu

, Tang

, Mok

, Li

, and Cai

. (2015). Elevated excretion of biopyrrin as a new marker for idiopathic Parkinson's disease. Parkinsonism Relat Disord, 21, 1371.

60.

Luong

KVQ

, and Nguyen

LTH

. (2015). The role of caffeine in neurodegenerative diseases: possible genetic and cellular signaling mechanisms. In: Watson

, and Preedy

(eds): Bioactive Nutraceuticals and Dietary Supplements in Neurological and Brain Disease. Westminster, CA: Academic Press, 261–279.

61.

Marklund

, and Marklund

. (1974). Involvement of the superoxide anion radical in the autoxidation of pyrogallol and a convenient assay for superoxide dismutase. Eur J Biochem, 47, 469–474.

62.

Mitchell

, Freeman

, Randazzo

, et al. (2009). A CSF biomarker panel for identification of patients with amyotrophic lateral sclerosis. Neurology, 72, 14–19.

63.

, Yang

, Jia

, Zhang

, and Jiang

. (2010). Production of 4-hydroxyphenyllactic acid by Lactobacillus sp. SK007 fermentation. J Biosci Bioeng, 109, 369–371.

64.

Multhaup

, Huber

, Buée

, and Galas

. (2015). Amyloid precursor protein (APP) metabolites APP intracellular fragment (AICD), Aβ42, and Tau in nuclear roles. J Biol Chem, 290, 23515–23522.

65.

Nicholls

. (1975). Formate as an inhibitor of cytochrome c oxidase. Biochem Biophys Res Commun, 67, 610–616.

66.

Paglia

, Stocchero

, Cacciatore

, et al. (2016). Unbiased metabolomic investigation of Alzheimer's disease brain points to dysregulation of mitochondrial aspartate metabolism. J Proteome Res, 15, 608–618.

67.

Pahnke

, Langer

, and Krohn

. (2014). Alzheimer's and ABC transporters—New opportunities for diagnostics and treatment. Neurobiol Dis, 72, 54–60.

68.

Pettegrew

, Withers

, Panchalingam

, and Post

. (1987). 31P nuclear magnetic resonance (NMR) spectroscopy of brain in aging and Alzheimer's disease. J Neural Transm Suppl, 24, 261–268.

69.

Prins

. (2008). Cerebral metabolic adaptation and ketone metabolism after brain injury. J Cereb Blood Flow Metab, 28, 1–16.

70.

Procaccini

, Santopaolo

, Faicchia

, et al. (2016). Role of metabolism in neurodegenerative disorders. Metabolism, 65, 1376–1390.

71.

Ramírez

, Blázquez

, del Pulgar

, Guzmán

, and de Ceballos

. (2005). Prevention of Alzheimer's disease pathology by cannabinoids: Neuroprotection mediated by blockade of microglial activation. J Neurosci, 25, 1904–1913.

72.

Rodríguez-Nogales

, Garbayo

, Carmona-Abellán

, Luquin

, and Blanco-Prieto

. (2016). Brain aging and Parkinson's disease: New therapeutic approaches using drug delivery systems. Maturitas, 84, 25–31.

73.

Rosas

, Doros

, Bhasin

, et al. (2015). A systems‐level “misunderstanding”: The plasma metabolome in Huntington's disease. Ann Clin Transl Neurol, 2, 756–768.

74.

Sako

, Abe

, Izumi

, Harada

, and Kaji

. (2016). The ratio of N-acetyl aspartate to glutamate correlates with disease duration of amyotrophic lateral sclerosis. J Clin Neurosci, 27, 110–113.

75.

Sapkota

, Tran

, Huan

, et al. (2015). Metabolomics analyses of salivary samples discriminate normal aging, mild cognitive impairment, and Alzheimer's disease groups and produce biomarkers predictive of neurocognitive performance. Alzheimers Dement, 11, 654.

76.

Sass

. (2012). Inborn errors of ketogenesis and ketone body utilization. J Inherit Metab Dis, 35, 23–28.

77.

Sato

, Suzuki

, Nakamura

, Bernier

, Aoshima

, and Oda

. (2012). Identification of a new plasma biomarker of Alzheimer's disease using metabolomics technology. J Lipid Res, 53, 567–576.

78.

Schumacher

, Krohn

, Hofrichter

, et al. (2012). ABC transporters B1, C1 and G2 differentially regulate neuroregeneration in mice. PLoS One, 7, e35613.

79.

Shannon

, Markiel

, Ozier

, et al. (2003). Cytoscape: A software environment for integrated models of biomolecular interaction networks. Genome Res, 13, 2498–2504.

80.

Smoot

, Ono

, Ruscheinski

, Wang

, and Ideker

. (2011). Cytoscape 2.8: New features for data integration and network visualization. Bioinformatics, 27, 431–432.

81.

Spasojević

, Stević

, Nikolić-Kokić

, Jones

, Blagojević

, and Spasić

. (2010). Different roles of radical scavengers–ascorbate and urate in the cerebrospinal fluid of amyotrophic lateral sclerosis patients. Redox Rep, 15, 81–86.

82.

Sussmuth

, Brettschneider

, Ludolph

, and Tumani

. (2008). Biochemical markers in CSF of ALS patients. Curr Med Chem, 15, 1788–1801.

83.

Tahmasian

, Bettray

, van Eimeren

, et al. (2015). A systematic review on the applications of resting-state fMRI in Parkinson's disease: Does dopamine replacement therapy play a role?. Cortex, 73, 80–105.

84.

Tanveer

, Gowran

, Noonan

, Keating

, Bowie

, and Campbell

. (2012). The endocannabinoid, anandamide, augments Notch-1 signaling in cultured cortical neurons exposed to amyloid-β and in the cortex of aged rats. J Biol Chem, 287, 34709–34721.

85.

Toczylowska

, Jamrozik

, Liebert

, and Kwiecinski

. (2013). NMR-based metabonomics of cerebrospinal fluid applied to amyotrophic lateral sclerosis. Biocybern Biomed Eng, 33, 21–32.

86.

Trupp

, Jonsson

, Ohrfelt

, et al. (2014). Metabolite and peptide levels in plasma and CSF differentiating healthy controls from patients with newly diagnosed Parkinson's disease. J Parkinsons Dis, 4, 549–560.

87.

Trushina

, and Mielke

. (2014). Recent advances in the application of metabolomics to Alzheimer's diseases. Biochim Biophys Acta, 1842, 1232–1239.

88.

Tumani

, Teunissen

, Süssmuth

, Otto

, Ludolph

, and Brettschneider

. (2008). Cerebrospinal fluid biomarkers of neurodegeneration in chronic neurological diseases. Expert Rev Mol Diagn, 8, 479–494.

89.

van Echten-Deckert

, and Herget

. (2006). Sphingolipid metabolism in neural cells. Biochim Biophys Acta, 1758, 1978–1994.

90.

Virarkar

, Alappat

, Bradford

, and Awad

. (2013). L-arginine and nitric oxide in CNS function and neurodegenerative diseases. Crit Rev Food Sci Nutr, 53, 1157–1167.

91.

Voyle

, Kim

, Proitsi

, et al. (2016). Blood metabolite markers of neocortical amyloid-β burden: Discovery and enrichment using candidate proteins. Transl Psychiatry, 6, e719.

92.

Wang

, Zhou

, Huang

, et al. (2014). Plasma metabolite profiles of Alzheimer's disease and mild cognitive impairment. J Proteome Res, 13, 2649–2658.

93.

WHO (World Health Organization) Report: Dementia, Fact Sheet (2016). http://www.who.int/mediacentre/factsheets/fs362/en/

94.

Wuolikainen

, Moritz

, Marklund

, Antti

, and Andersen

. (2011). Disease-related changes in the cerebrospinal fluid metabolome in amyotrophic lateral sclerosis detected by GC/TOFMS. PLoS One, 6, e17947.

95.

Yazdani

, Yazdani

, and Boerwinkle

. (2016a). A causal network analysis of the fatty acid metabolome in African-Americans reveals a critical role for palmitoleate and margarate. OMICS, 20, 480–484.

96.

Yazdani

, Yazdani

, and Boerwinkle

. (2016b). A causal network analysis in an observational study identifies metabolomics pathways influencing plasma triglyceride levels. Metabolomics, 12, 104.

97.

Zheng

, Wang

, Chen

, et al. (2013). Identification and validation of urinary metabolite biomarkers for major depressive disorder. Mol Cell Proteomics, 12, 207–214.

98.

Zilberter

, Gubkina

, and Ivanov

. (2015). A unique array of neuroprotective effects of pyruvate in neuropathology. Front Neurosci, 9, 1–5.

99.

Zlotnik

, Gurevich

, Tkachov

, Maoz

, Shapira

, and Teichberg

. (2007). Brain neuroprotection by scavenging blood glutamate. Exp Neurol, 203, 213–220.

100.

Zlotnik

, Sinelnikov

, Gruenbaum

, et al. (2012). Effect of glutamate and blood glutamate scavengers oxaloacetate and pyruvate on neurological outcome and pathohistology of the hippocampus after traumatic brain injury in rats. Anesthesiology, 116, 73–83.

101.

Zoccolella

, Simone

, Lamberti

, et al. (2008). Elevated plasma homocysteine levels in patients with amyotrophic lateral sclerosis. Neurology, 70, 222–225.