Omics data integration suggests a potential idiopathic Parkinson's disease signature

Abstract

The vast majority of Parkinson's disease cases are idiopathic. Unclear etiology and multifactorial nature complicate the comprehension of disease pathogenesis. Identification of early transcriptomic and metabolic alterations consistent across different idiopathic Parkinson's disease (IPD) patients might reveal the potential basis of increased dopaminergic neuron vulnerability and primary disease mechanisms. In this study, we combine systems biology and data integration approaches to identify differences in transcriptomic and metabolic signatures between IPD patient and healthy individual-derived midbrain neural precursor cells. Characterization of gene expression and metabolic modeling reveal pyruvate, several amino acid and lipid metabolism as the most dysregulated metabolic pathways in IPD neural precursors. Furthermore, we show that IPD neural precursors endure mitochondrial metabolism impairment and a reduced total NAD pool. Accordingly, we show that treatment with NAD precursors increases ATP yield hence demonstrating a potential to rescue early IPD-associated metabolic changes.

Fig. 1. Transcriptomic and metabolic profiles reveal neurodevelopmental and metabolic alterations in IPD neural precursor cells.

a Heatmap of scaled gene counts of significantly differentially expressed genes (p < 0.05) between IPD and control NESCs. b Log2FC of the top significantly expressed genes (FDR < 0.05) displaying gene expression difference in IPD NESCs compared to control cells. c The most dysregulated metabolic pathways, selected by the average of Log2FC < −1 or >1 of genes annotated in each pathway. The color represents the Log2FC. d Principal component scores plot of non-polar metabolites detected by untargeted GC-MS analysis. e Principal component scores plot of polar metabolites detected by untargeted GC-MS analysis.

Fig. 2. IPD neural precursors show reduced ability to metabolize various metabolic substrates and impaired mitochondrial respiratory capacity.

a Scatter plot of the maximal metabolic rate of different substrates. Each dot represents a unique substrate placed in the plot according to the metabolic rate by which it has been metabolized by the IPD (y-axis) and control (x-axis) NESCs. The maximal metabolic rate is normalized by background subtraction and cell density in the respective well. The median rate between the three lines of each condition is considered. Substrates metabolized with the normalized maximal rate above 1 are labeled. b Oxygen consumption rate (OCR) over time representing mitochondrial respiratory capacity. Basal respiration is measured until the injection of oligomycin which inhibits complex V activity, resulting in a decrease in respiration, which is linked to ATP production. FCCP injection disrupts ATP synthesis and mitochondrial membrane potential, allowing measurement of maximal respiration and spare respiratory capacity. The final injection is a mixture of complex I and complex III inhibitors—rotenone and antimycin A. Here mitochondrial respiration is shut down, enabling the calculation of nonmitochondrial respiration. Statistics: unpaired t-test. Significance asterisks represent *P < 0.05, **P < 0.01, ***P < 0.001. Error bars represent mean + SD. N = 3 independent experiments. c Bar graphs of mitochondrial respiratory capacity features. Statistics: non-parametric Mann-Whitney test. Error bars represent mean + SD. N = 3 independent experiments, each data point represents a measurement of a single well of the assay. d Extracellular acidification rate (ECAR) over time representing glycolytic function. Before glucose injection, ECAR shows non-glycolytic acidification caused by processes in the cell other than glycolysis. The first injection of glucose enables measurement of the rate of glycolysis under basal conditions. The second injection of oligomycin, a complex V inhibitor, enhances the energy production via glycolysis, revealing the maximum glycolytic capacity. The final injection of 2-deoxy-glucose (2-DG), a glucose analog that inhibits glycolysis, allowing to measure glycolytic reserve. N = 3 independent experiments. e Pyruvate dehydrogenase activity measured based on nmol of generated NADH from NAD+ over time, which is proportional to enzyme activity. Statistics: non-parametric Mann-Whitney test. Error bars represent mean + SD. N = 3 independent experiments. f Total pool of NAD (NAD+ and NADH) concentration in pmol relative to the control samples. Statistics: non-parametric Mann-Whitney test. Error bars represent mean + SD. N = 3 independent experiments. g NAD+/NADH ratio relative to the control samples. Statistics: non-parametric Mann-Whitney test. Error bars represent mean + SD. N = 3 independent experiments.

Fig. 3. Metabolic modeling.

a Unsupervised clustering of the top 50 reactions with the greatest fold change of flux relative to the mean flux of control models. Flux distribution was determined using eFBA. b Metabolic pathways to which the top 50 reactions are assigned, demonstrating the percentage of the top 50 reactions belonging to each pathway. c The top five subsystems with the highest reaction count were determined by assigning the top 50 most changed reactions to the Recon 3D subsystems. d Unsupervised clustering of NAD+ metabolism reactions based on the fold change of flux relative to the mean flux of control models. Flux distribution was determined using eFBA. e Unsupervised clustering of estimated flux for the NAD+ involving reactions in the mitochondria. Flux distribution was determined using eFBA. f Unsupervised clustering of estimated flux for the NAD+ involving reactions in the cytosol. Flux distribution was determined using eFBA.

Fig. 4. Data integration.

a Heatmap of the top discriminant features between the IPD and control NESCs. b Circos plot showing the correlation between features contributing to the variation of the component 1. Correlation threshold: r = 0.9. c Correlation network of features contributing to the variation of the component 1. Correlation threshold: r = 0.95. d Relative abundance of glycerol-3-phosphate. Statistics: Welch’s t-test. Error bars represent mean + SD. N = 3 biologically independent samples. e Graphical representation of glycerol-3-phosphate (G3P) as an intermediate metabolite in glycolysis, lipid metabolism and oxidative phosphorylation and its role in NAD metabolism. G3P synthesis from dihydroxyacetone phosphate (DHAP) by cytosolic glycerol-3-phosphate dehydrogenase (cGPDH) regenerates cytosolic NAD+ from the NADH that is generated by glyceraldehyde-3-phosphate dehydrogenase in glycolysis. The G3P shuttle also facilitates electron transport between cytosol to mitochondria. Flavin linked mitochondrial glycerol-3-phosphate dehydrogenase (mGPDH) oxidases G3P at the same time reducing flavin adenine dinucleotide (FAD) to FADH₂ and transferring electrons to ubiquinone pool of the electron transport chain (ETC)_. ETC oxidazes NADH generated in TCA to replenish the mitochondrial NAD+ pool. G3P can also be produced from glycerol, which is the end product of lipolysis. f Intracellular ATP levels measured in relative light units (RLU) and normalized to the cell number in samples treated with vehicle, 20 nM quinolinic acid (QA) and 5 mM nicotinic acid (NA). Statistics: non-parametric Mann-Whitney test. Error bars represent mean + SD. N = 3 independent experiments.

Fig. 5. Metabolites corresponding to significant IPD model reactions compared with metabolites associated with PD diagnosis from a meta-analysis of clinical metabolomic studies.

a The proportion of metabolites with changes reported in more than one study (replicated metabolite). b The number of studies reporting increased or decreased concentrations for replicated metabolites.