Metabolic Profiling of Brain Tissue and Brain-Derived Extracellular Vesicles in Alzheimer's Disease

Abstract

Alzheimer´s disease (AD) is the most frequent neurodegenerative disorder in the world and is characterised by the loss of memory and other cognitive functions. Metabolic changes associated with AD are important players in the development of the disease. However, the mechanism underlying these changes is still unknown. Extracellular vesicles (EVs) are nano-sized particles that play an important role in regulating pathophysiological processes and are a non-invasive manner to obtain information of the cell that is secreting them. The analysis of brain-derived EVs (bdEVs) will provide new insights in the metabolic processes associated with AD. To characterize bdEVs in AD, we optimised a method to isolate them from tissue of different brain regions, obtaining the highest enrichment in isolations from the temporal cortex. We performed unbiased untargeted metabolomics analysis on post-mortem human temporal cortex tissue and bdEVs from the same region of AD patients and healthy controls. Both, univariate and multivariate statistical analysis were used to determine the metabolites that influence the separation between AD patients and controls. Interestingly, a clear separation between control and AD groups was obtained with bdEVs, which allowed to select 12 relevant features by a validated PLS-DA model. Furthermore, comparison of tissue and bdEVs identified 68 common features. The pathway enrichment analysis of the common metabolites showed that the alanine, aspartate and glutamate pathway and the arginine, phenylalanine, tyrosine pathway were the most significant ones in the separation between the AD patients and controls. The phenylalanine, tyrosine and tryptophan pathway, still had a very high influence in the separation between groups, albeit not significant. Notably, some metabolites were identified for the first time in bdEVs. For example, the N-acetyl aspartic acid (NAA) metabolite present in bdEVs was suitable to differentiate AD patients from healthy controls. Furthermore, the analysis of the hippocampus, midbrain, temporal and entorhinal cortex and their respective bdEVs indicated that the metabolic profiles of different brain areas were distinct and showed some correlation between the metabolome of the tissue and its respective bdEVs. Thus, our study highlights the potential of bdEVs to understand the metabolic fingerprint associated with AD and their potential use as diagnostic and therapeutic targets.

Keywords: Alzheimer's disease; brain; extracellular vesicles; mass spectrometry; metabolomics.

FIGURE 1

Western blotting analysis of samples from four different brain regions in AD patients. F3‐5 were enriched in EV markers (CD63, TSG101, Flotillin 1, Caveolin, CD9, CD81, Rab8) and void of cellular markers (GRP78, Calnexin, COX IV). Neuron‐specific enolase (NSE) was present in F8‐10, but not in F3‐5. BH, brain homogenate; BHC, brain homogenate after treatment with collagenase; P10k, pellet after 10,000 × g centrifugation; S10k, supernatant after 10,000 × g centrifugation; F3‐5, SEC‐fractions EV‐enriched; F8‐10, SEC‐fractions protein‐enriched. Brain regions: H = hippocampus, TC = temporal cortex, M = midbrain, EC = entorhinal cortex. Molecular weights are shown in kDa.

FIGURE 2

Characterization of EVs isolated from human temporal cortex by SEC. (A) Particle concentration (left y‐axis) and protein concentration (right y‐axis) of each fraction were measured by NTA and Bradford assay, respectively. The mean and standard deviation of three samples are shown (n = 3). (B) Particle concentration of EV preparations (F3‐5) was measured by NTA, comparing both AD and CTRL groups. The mean and standard deviation of two samples per group are shown (n = 2). U‐Mann Whitney test, p value 0.3333 (ns). (C) Representative Western blotting membranes from AD and CTRL samples. F3‐5 were enriched in EV markers (Flotillin 1, CD63, CD81, Rab8) and void of cellular markers (GRP78, Calnexin, COX IV). BH, brain homogenate; BHC, brain homogenate after treatment with collagenase; P10k, pellet after 10,000 × g centrifugation; S10k, supernatant after 10,000 × g centrifugation; F3‐5, SEC‐fractions EV‐enriched; F8‐10, SEC‐fractions protein‐enriched. Molecular weights are shown in kDa. (C) Representative images of cryo‐EM of EVs from AD and CTRL samples present in fraction 4. The scale bar represents 200 nm.

FIGURE 3

Untargeted metabolomics analysis of temporal‐brain tissue and bdEVs analysed at positive (ESI+) and negative (ESI–) ionization modes. (A) PCA scores plot of temporal‐brain tissue samples obtained at ESI+ (R2X = 0.486, Q2 = 0.216, 2 PCs and pareto scaling); (B) PCA scores plot of temporal‐brain tissue samples obtained at ESI‐ (R2X = 0.467, Q2 = 0.076, 2 PCs and pareto scaling). The red dot represents the outlier CTRL7; (C) PCA scores plot of bdEV samples obtained at ESI+ (R2X = 0.318, Q2 = 0.071, 2PCs, and pareto scaling); (D) PLS‐DA scores plot of bdEV samples obtained at ESI+ (R2X = 0.203, R2Y = 0.906, Q2 = 0.383, 2PCs, pareto scaling and CV‐ANOVA p value = 0.0298); (E) Volcano plot obtained from validated PLS‐DA model with selected features highlighted in blue (VIP > 1 and |0.5 < p(corr) > 0.6|) or red (VIP > 1 and |p(corr) > 0.6|); (F) Venn diagram of features obtained in temporal‐brain tissue and bdEVs at ESI+; (G) Venn diagram of features obtained in temporal‐brain tissue and bdEVs at ESI‐ and (H) Pathway analysis overview where higher pathway impact values represent the relative importance of the pathway, the size of the circle indicates the impact of the pathway and the color the significance (the more intense the red color, the lower the p value).

FIGURE 4

Untargeted metabolomics analysis of brain tissue and bdEVs isolated samples obtained from the four brain regions at positive (ESI+) ionization mode. (A) PCA scores plot colored based on brain region (ENTOR = entorhinal cortex, HIPO = hippocampus, MESEN = midbrain, and TEMP = temporal cortex) (_T = tissue, _F35 = bdEV) (R2X = 0.776, Q2 = 0.441, 4 PCs and autoscaling); (B) PCA scores plot colored based on disease (AD = Alzheimer's disease patient, CTRL = healthy control individual, and QC = quality control samples); (C) PLS‐DA scores plot containing the four‐brain region (R2X = 0.505, R2Y = 0.275, Q2 = 0.0631, 2PCs, autoscaling and CV‐ANOVA p value = 0.99), and (B) scores plot containing only three regions (R2X = 0.610, R2Y = 0.314, Q2 = –0.028). ENTOR, entorhinal cortex; HIPO, hippocampus; MESEN, midbrain; TEMP, temporal cortex; _T, tissue, _F35, bdEV.