Correlations of blood and brain NMR metabolomics with Alzheimer's disease mouse models

Abstract

Alzheimer's disease (AD) is a complex, progressive neurodegenerative disorder, impacting millions of geriatric patients globally. Unfortunately, AD can only be diagnosed post-mortem, through the analysis of autopsied brain tissue in human patients. This renders early detection and countering disease progression difficult. As AD progresses, the metabolomic profile of the brain and other organs can change. These alterations can be detected in peripheral systems (i.e., blood) such that biomarkers of the disease can be identified and monitored with minimal invasion. In this work, High-Resolution Magic Angle Spinning (HRMAS) Nuclear Magnetic Resonance (NMR) spectroscopy is used to correlate biochemical changes in mouse brain tissues, from the cortex and hippocampus, with blood plasma. Ten micrograms of each brain tissue and ten microliters of blood plasma were obtained from 5XFAD Tg AD mice models (n = 15, 8 female, 7 male) and female C57/BL6 wild-type mice (n = 8). Spectral regions-of-interest (ROI, n = 51) were identified, and 121 potential metabolites were assigned using the Human Metabolome Database and tabulated according to their trends (increase/decrease, false discovery rate significance). This work identified several metabolites that impact glucose oxidation (lactic acid, pyruvate, glucose-6-phosphate), allude to oxidative stress resulting in brain dysfunction (L-cysteine, galactitol, propionic acid), as well as those interacting with other neural pathways (taurine, dimethylamine). This work also suggests correlated metabolomic changes within blood plasma, proposing an avenue for biomarker detection, ideally leading to improved patient diagnosis and prognosis in the future.

Fig. 1. Averaged HRMAS NMR spectra and hierarchical clusters.

A Averaged HRMAS NMR spectra with standard deviations presented as shaded areas for cortex, hippocampus, and plasma, with WT female (n = 8), AD female (n = 8) and AD male (n = 7) groups presented separately for each tissue type. B Unsupervised hierarchical clusters of ROIs (x-axis) are plotted against the three mice groups. High spectral intensity in the ROIs are represented by red boxes, while low, in blue. Heat maps show clear distinctions between AD and WT for all three sample types. Using the top 15 rows as a threshold, the sensitivities [93.3, 93.3, 86.6] and specificities [87.5, 87.5, 75.0] are calculated for cortex, hippocampus, and plasma, respectively.

Fig. 2. Spectral regions differentiating AD and WT groups.

Volcano plots of ln(Fold Change) vs. -ln(p), where Fold Change is the ratio of the AD mean over the control (i.e. WT) mean, and p values are calculated from Wilcoxon tests, for A. Cortex, B. Hippocampus, and C. Plasma, respectively. Solid circles represent results from all AD mice, including female and male, and open circles represent only female mice. Long dashed lines represent FDR significant levels for all AD animals, and short dashed lines represent FDR significant levels for only female animals. Fourteen spectral regions that were FDR significant based on Wilcoxon tests for all AD (solid circles) and female only (open circles) mice are labeled from 1–9 and a-e, representing various spectral ROIs in ppm: 1 3.98–3.96; 2 3.35–3.33; 3 2.50–2.48; 4 2.43–2.41; 5 2.02–2.00; 6 1.05–1.02; 7 3.20–3.16; 8 3.91–3.90; 9 3.89–3.88; a 3.83–3.81; b 3.79–3.75; c 3.50–3.49; d 3.48–3.46; e 3.45-3.44.

Fig. 3. Metabolomic ellipsoids differentiating AD from WT groups.

3D ellipsoids generated by the three most significant PCs as metabolomic profiles differentiating AD from WT animals as measured for A. Cortex, B. Hippocampus, and C. Plasma. Blue: WT, red: female AD, and purple: male AD.

Fig. 4. Principal components for WT and AD.

The five PCs present varied potentials of differentiating WT from AD for three tested tissue types. PC1 results are shown in A–D, PC2 in E–H, PC3 in I–L, PC4 in M–P, and PC5 in Q–T. A summary of all violin plots for each PC is shown in the first column A, E, I, M, Q, and results for each tissue type are presented in the other columns, namely, cortex B, F, J, N, R, hippocampus C, G, K, O, S, and plasma D, H, L, P, T. Statistically significant p-values (p < 0.05) calculated from Wilcoxon tests are presented in dark blue, and not significant p-values are in light blue. The most significant PCs are PC3 and PC4 for both cortex and hippocampus, and PC1 for plasma. Error bars represent standard deviation.

Fig. 5. Spectral ROI differentiations of WT and AD and their contributions to differentiating PCs.

Δ represents the difference of AD – WT. LF stands for PCA loading factor for a particular spectral ROI.