Identification of altered metabolic pathways in plasma and CSF in mild cognitive impairment and Alzheimer's disease using metabolomics

Abstract

Alzheimer's Disease (AD) currently affects more than 5 million Americans, with numbers expected to grow dramatically as the population ages. The pathophysiological changes in AD patients begin decades before the onset of dementia, highlighting the urgent need for the development of early diagnostic methods. Compelling data demonstrate that increased levels of amyloid-beta compromise multiple cellular pathways; thus, the investigation of changes in various cellular networks is essential to advance our understanding of early disease mechanisms and to identify novel therapeutic targets. We applied a liquid chromatography/mass spectrometry-based non-targeted metabolomics approach to determine global metabolic changes in plasma and cerebrospinal fluid (CSF) from the same individuals with different AD severity. Metabolic profiling detected a total of significantly altered 342 plasma and 351 CSF metabolites, of which 22% were identified. Based on the changes of >150 metabolites, we found 23 altered canonical pathways in plasma and 20 in CSF in mild cognitive impairment (MCI) vs. cognitively normal (CN) individuals with a false discovery rate <0.05. The number of affected pathways increased with disease severity in both fluids. Lysine metabolism in plasma and the Krebs cycle in CSF were significantly affected in MCI vs. CN. Cholesterol and sphingolipids transport was altered in both CSF and plasma of AD vs. CN. Other 30 canonical pathways significantly disturbed in MCI and AD patients included energy metabolism, Krebs cycle, mitochondrial function, neurotransmitter and amino acid metabolism, and lipid biosynthesis. Pathways in plasma that discriminated between all groups included polyamine, lysine, tryptophan metabolism, and aminoacyl-tRNA biosynthesis; and in CSF involved cortisone and prostaglandin 2 biosynthesis and metabolism. Our data suggest metabolomics could advance our understanding of the early disease mechanisms shared in progression from CN to MCI and to AD.

Figure 1. Heat map analysis of metabolites in plasma and CSF samples from CN, MCI and AD patients.

Metabolite perturbations were calculated based on the median for each metabolite level of three independent biological replicates of plasma and CSF samples from each study participant. Each row represents a metabolite, and each column depicts a subject. The study groups are color-coded as follows: AD is denoted in red, MCI is denoted in blue, and CN is denoted in maroon. The fold change in metabolite levels is color-coded: red pixels, up regulation; blue, down regulation; yellow, no significant change.

Figure 2. Plasma and CSF samples have distinct metabolomic profiles between AD, MCI and CN groups.

Two-dimensional score plots of unsupervised principal component analysis (PCA) of the plasma (A) and CSF (B) samples, and orthogonal two partial least squares-discriminant analysis (O2PLS-DA) of plasma (C) and CSF (D) samples from AD (red), MCI (blue) and CN (green) patients. Each sample is labeled with a triangle.

Figure 3. Unsupervised Principal Component Analysis (PCA) of plasma and CSF samples from CN, AD and MCI subjects.

Each dot corresponds to an individual sample. (A, B): AD – red; CN – blue; (C, D): MCI – red; CN – blue; (E, F): AD – red; MCI- blue.

Figure 4. Altered metabolic pathways and process networks in plasma and CSF of MCI vs. CN subjects.

The significance of the pathways was evaluated using P values and false discovery rate <0.05. Pathways that are affected in both fluids are colored in red. The ratio depicts the number of affected metabolites to the total number of metabolites in the pathway. TCA – tricarboxylic acid (Krebs) cycle; GABA - γ-Aminobutyric acid; MB – metabolism.

Figure 5. Canonical pathways and process networks affected in plasma and CSF of AD vs. CN subjects.

The significance of the pathways was evaluated using P values and false discovery rate <0.05. Pathways that are affected in both fluids are colored in red. The ratio depicts the number of affected metabolites to the total number of metabolites in each pathway. FXR - farnesoid X receptor; CFTR - cystic fibrosis transmembrane conductance regulator; VDR - vitamin D receptor; DGAT1 - diacylglycerol acyltransferase 1; nNOS – neuronal nitric oxide synthase; UMP - Uridine monophosphate; HETE - hydroxyeicosatetraenoic acid; HPETE - hydroperoxyeicosatetraenoic acid; BS – biosynthesis.

Figure 6. Altered metabolic pathways and process networks specifically affected either in plasma or in CSF of AD vs. MCI subjects.

The significance of the pathways was evaluated using P values and false discovery rate <0.05. GABA - γ-Aminobutyric acid; nNOS – neuronal nitric oxide synthase; CFTR - cystic fibrosis transmembrane conductance regulator; VDR - vitamin D receptor; HETE - hydroxyeicosatetraenoic acid; HPETE - hydroperoxyeicosatetraenoic acid.

Figure 7. Venn diagram illustrating shared and uniquely affected pathways in plasma and CSF of MCI, AD and CN individuals.

Common pathways are defined.