Metabolomics of Multiple System Atrophy Patient-Derived Striatal Medium Spiny Neurons

Abstract

In multiple system atrophy (MSA), the fatal movement disorder, cell populations of the striatum and other subcortical brain regions degenerate, leading to a rapidly progressive, atypical Parkinsonian syndrome. The pathophysiology of neurons and glial cells shows misfolding, aggregation, and increased release of the protein α-synuclein. In addition, neuronal hypoexcitability, a reduction in the activity of the mitochondrial respiratory chain, and a dysregulation of the enzymes involved in the biosynthesis of coenzyme Q10 were observed in human stem-cell models. In this study, untargeted and targeted metabolome analyses were performed with MSA patient-derived GABAergic striatal medium spiny neurons focusing on the citrate cycle and mitochondrial respiratory chain. The results indicate a significant decrease in succinate and ATP as well as an imbalanced NAD⁺/NADH ratio of MSA cell lines compared to matched healthy controls, suggesting alterations in mitochondrial processes which may facilitate neurodegeneration.

1. Introduction

Multiple system atrophy (MSA) is a sporadic, rare, rapidly progressive, neurological disease that results in the degeneration of multiple cell populations and brain regions, including the striatum and cerebellum. This movement disorder can lead to diverse autonomic, cerebellar, and Parkinsonian symptoms. Depending on the predominantly degenerated brain region, MSA can be distinguished into a Parkinsonian-type variant (MSA-P), a type that mainly affects mainly the striatum, and the cerebellar type (MSA-C) [1,2,3,4,5].

In the pathophysiology, the chaperone protein α-synuclein misfolds and aggregates in the cytoplasm of neurons and oligodendrocytes [2,4,6]. A recent in vitro study with striatal GABAergic medium spiny neurons showed an elevated release and neuronal distribution of α-synuclein in hypoexcitable MSA-P patient cell lines compared to matched healthy controls [7]. Neuroinflammation is also observed by the activation of microglia cells and the release of pro-inflammatory signals [1,2,4]. No specific pathogenic mutation in MSA patients has been described [1,2,4].

Dysfunctions of mitochondrial processes and autophagy play an important role in neurodegenerative disorders [8,9,10,11]. In MSA, the activity of the mitochondrial respiratory chain and the biosynthesis of coenzyme Q10 are altered [12,13,14]. Further metabolite alterations were found in the cerebrospinal fluids (CSFs), e.g., dopamine, lactate, citrulline, and norepinephrine [12,15,16], and in the plasma, e.g., succinate, lactate, arginine, formic acid, methionine, betaine, and urea [17,18] of MSA patients compared to controls, Parkinson’s disease (PD), or Progressive Supranuclear Palsy (PSP) patients. Usually, these findings originate from small patient and control groups, some of which were also obtained using different measurement techniques such as enzyme-linked immunosorbent assay (ELISA), high-performance liquid chromatography (HPLC), liquid chromatography (LC), and gas chromatography–mass spectrometry (GC-MS).

Nevertheless, metabolome studies help us to determine physiological conditions in both disease and health and lead us to understand pathophysiological mechanisms. For this purpose, untargeted and targeted metabolomics can be used to study the metabolome of biofluids from patients and healthy subjects, or from in vitro models using cells from humans. For unspecific and unknown metabolites, an untargeted analysis can provide extensive and complex data that can be validated and quantified in targeted analyses by incorporating specific and isotope-labelled metabolites [19,20].

Metabolome analyses of MSA patients or relevant disease models could provide new insights of the pathophysiological processes that are relevant for future therapeutic approaches. Furthermore, comparing metabolomic changes could aid in the search for early detection of biomarkers to distinguish MSA from neurodegenerative disorders with similar incipient symptoms, such as PD, PSP, and Dementia with Lewy bodies (DLB) [3,4,14,21].

In this study, the in vitro metabolomes of striatal GABAergic medium spiny neurons, which represent the main cell type in the striatum [22], differentiated from human induced pluripotent stem-cell lines (hiPSCs) of MSA-P patients (n = 3) and sex- and age-matched healthy controls (n = 3), were compared in an untargeted metabolome analysis. Subsequently, the citrate cycle and the mitochondrial functions were analysed in a targeted approach.

2. Materials and Methods

2.1. Cultivation and Differentiation of Human Induced Pluripotent Stem Cells (hiPSCs) into Striatal GABAergic Medium Spiny Neurons (MSNs)

In this study, we used cell lines from three MSA-P patients and three matched healthy controls (Table 1). After obtaining written consent and ethical approval from the Ethics Committee of Hannover Medical School (No. 8666_BO_K_2019, date of approval: 13 September 2019), skin punch biopsies were taken from MSA-P patients (cell lines P2 and P3) and healthy subjects (cell lines CTR1 and CTR3). These cell lines, including P1, were characterised in previous works [7,13].

Table 1.

List of hiPS cell lines used in this study. Control cell lines are shortened as CTR1, CTR2, and CTR3; MSA patient cell lines as P1, P2, and P3.

ID Code	Sex	Age at Biopsy	Number of Cell Passages	Origin/Reference
CTR1	M	59	P7—P33	Henkel et al. [7]
CTR2	F	62	P34—P48	StemBANCC consortium SFC084-03-02-01A
CTR3	F	62	P8—P20	Henkel et al. [7]
P1	F	78	P29—P45	Monzio Compagnoni et al. [13]
P2	M	52	P8—P34	Henkel et al. [7]
P3	F	56	P16—P36	Henkel et al. [7]

HiPSCs were cultivated in mTeSR Plus medium (STEMCELL Technologies, Vancouver, BC, Canada, 100-0276) with 1% Penicillin–Streptomycin (Thermo Fisher Scientific Inc., Waltham, MA, USA, 15140122) on a hESC-qualified Matrigel matrix (Corning, New York, NY, USA, 354277). For passaging, cells were detached as clumps with 0.5 mM EDTA (Thermo Fisher Scientific Inc., AM9260G) and reseeded in mTeSR Plus with 10 µM ROCK inhibitor (Y-27632, STEMCELL Technologies, 72304) for 24 h.

After adaptation from Staege et al. [23] and Henkel et al. [7], hiPSCs were differentiated as embryoid bodies (EBs) from day 0 to day 12 with two-day specific medium compounds. On day 0, the cells were passaged in mTeSR Plus with 1 µM dorsomorphin dihydrochloride (D, Abcam, Cambridge, UK, ab144821), 10 µM SB 431542 (SB, Abcam, ab291112), 1 µM Wnt Antagonist II (IWP-2, Merck KGaA, Darmstadt, Germany, 681671), and 10 µM Y-27632 for 48 h. From day 2, the N2 medium (KnockOut DMEM/F-12, 12660012; N-2 Supplement, 17502048; GlutaMAX Supplement, 35050038; 1% MEM Non-Essential Amino Acids Solution, 11140035; 1% Penicillin–Streptomycin, all from Thermo Fisher Scientific Inc.; 15 mM HEPES, Merck KGaA, H0887) was used as the basic medium. A total of 1 µM D, 10 µM SB and 1 µM IWP-2 were added on day 2; additionally, 1 µM purmorphamine (PMA, Abcam, ab120933) was added on day 4. On days 6 and 8, 1 µM IWP-2 and 1 µM PMA were supplemented. On day 10, the N2 medium was used without additives. Starting on day 12, the medium was replaced by a maturation medium (DMEM/F-12 with GlutaMAX Supplement, 31331093; Neurobasal medium, 21103049; N-2 Supplement; B-27 Supplement without Vitamin A, 12587010; 1% Penicillin–Streptomycin; 1% GlutaMAX Supplement; all from Thermo Fisher Scientific Inc.) with freshly added 20 ng/mL brain-derived neurotrophic factor (BDNF, Thermo Fisher Scientific Inc., 450-02), 10 ng/mL glial cell-derived neurotrophic factor (GDNF, Thermo Fisher Scientific Inc., 450-10), and 0.05 mM dibutyryl cyclic-AMP (dbcAMP, Merck KGaA, D0260).

From day 12 to 16, the EBs were seeded on hESC-qualified Matrigel matrix-coated plates. On days 16 to 18, the cells were detached with Accutase (Thermo Fisher Scientific Inc., A1110501) and reseeded in maturation medium supplemented with 10 µM Y-27632 for 48 h on poly-DL-ornithine hydrobromide (80 µg/mL, PORN, Merck KGaA, P8638) and laminin-coated (10 µg/mL, Thermo Fisher Scientific Inc., 23017015) dishes. Between day 24 and 30, the cells were split again on PORN and laminin-coated well plates and matured for up to 70 ± 7 days.

2.2. Immunocytochemistry (ICC)

On day 70 (±7), cells were fixed with 4% paraformaldehyde (PFA, Merck KGaA, 8.18715) for 20 min at room temperature and washed three times with phosphate-buffered saline (PBS, Thermo Fisher Scientific Inc., 14190169). For 60 min, the cells were treated with a blocking solution containing 1% bovine serum albumin (BSA, Merck KGaA, A7906), 5% goat serum (Thermo Fisher Scientific Inc., 16210072), and 0.3% Triton X-100 (Merck KGaA, T8787) diluted in PBS. Primary antibodies were prepared in the blocking solution and incubated overnight at 4 °C. Mouse anti-β III tubulin (TUJ1, 1:1000; Abcam, ab78078, RRID:AB_2256751), rabbit anti-γ-aminobutyric acid (GABA, 1:1000; Merck KGaA, A2052, RRID:AB_477652), and rat anti-COUP TF-1 interacting protein 2 (CTIP2, 1:300; Abcam, ab18465, RRID:AB_2064130) antibodies were used. After incubation, the cells were washed three times with PBS and afterwards, the secondary antibodies (goat anti-mouse Alexa Fluor 488, 1:1000; A-21131, RRID:AB_2535771; goat anti-rabbit Alexa Fluor 555, 1:1000; A-21428, RRID:AB_2535849; goat anti-rabbit Alexa Fluor 488, 1:500; A-11034, RRID:AB_2576217; goat anti-rat, 1:500; A-21434, RRID:AB_2535855; all from Thermo Fisher Scientific Inc.) were incubated for 2 h at room temperature. The cell nuclei were stained with 4’,6-diamidino-2-phenylindole (DAPI, 10 µg/mL, Thermo Fisher Scientific Inc., D1306) included in Mowiol 4-88 (Merck KGaA, 475904) as the mounting medium.

Fluorescence imaging was performed with the Olympus BX61 fluorescence microscope, the Olympus BX-UCB control box, the Olympus DP72 camera (Olympus, Hamburg, Germany), the X-Cite 120Q fluorescence lamp (Micrasys, Herborn, Germany), and the Olympus cellSens Dimension 1.18 analysis software. For each cell line and differentiation, four visual areas were randomly selected and stained cells were counted in ImageJ 1.54f (National Institutes of Health, NIH, Bethesda, MD, USA). Each staining with DAPI, TUJ1, and GABA or DAPI, GABA, and CTIP2 was analysed separately for each cell line and differentiation. Percentages relating to either DAPI, TUJ1, or GABA were calculated and averaged for each cell line. The three control and the three patient cell lines were summarised accordingly, and the percentages are shown.

2.3. Reverse Transcription Quantitative Real-Time Polymerase Chain Reaction (RT-qPCR)

RNA isolation of hiPSCs and MSNs (70 ± 7 days) was performed using RNeasy Mini Kit (QIAGEN, Venlo, The Netherlands, 74104) in combination with DNase digestion (RNase-free DNAse Set, QIAGEN, 79254) according to the manufacturer’s protocol. Before transcribing the RNA into cDNA using the QuantiTect Reverse Transcription Kit (QIAGEN, 205311), the RNA concentrations were determined in Nanodrop 2000c (Thermo Fisher Scientific Inc.). Real-time PCR was run in a StepOnePlus cycler (Applied Biosystems StepOnePlus Real-Time PCR System, StepOne Software v2.3, Thermo Fisher Scientific Inc.) using SYBR Green PCR Master Mix (Thermo Fisher Scientific Inc., 4367659), 1.75 µM forward and reverse primers, and 7 ng cDNA. The reaction started at 95 °C for 10 min, and continued as a cycle of 95 °C for 15 s and 60 °C for 1 min up to 40 rounds. Primers for neuronal (TUJ1; MAP2, microtubule-associated protein 2; FOXP1, forkhead box protein P1), GABAergic (GAD67, glutamic acid decarboxylase; FOXG1, forkhead box protein G1) and MSN-specific genes (CTIP2) were examined in this study. β2 microglobulin (B2M), glyceraldehyde-3-phosphate dehydrogenase (GAPDH), and β-actin (ACTB) were used as reference genes (Table S1) [7,23]. The PCRs were performed in triplicate.

For the comparative gene-expression analysis, the Ct values of the three reference genes (ΔCt Reference) were averaged and subtracted from the mean target genes (ΔCt Sample). For each cell line, the calibrator was calculated from the averaged Ct values of the hiPSCs (ΔCt Calibrator). Then, the ΔΔCt was calculated from the difference between the mean ΔCt Sample and the mean ΔCt Calibrator. The fold expressions (2^−ΔΔCt) are illustrated.

2.4. Cell Sample Preparation on Day 70 (±7)

For the following (metabolomic) investigations, the general harvest procedure started with cooling the cells on ice, removing the media, treating with either extraction reagent or assay buffer, and then scraping and vortexing. The samples for mass spectrum analysis were stored at −80 °C, then thawed and centrifuged for 10 min at 4 °C, 20,800× g. Those used for colorimetric assays were centrifuged and the supernatants separated before being stored at −20 °C. To normalise the data, the protein amounts in all corresponding pellets were determined.

2.5. Untargeted Metabolomics

Before being treated with 80% methanol solution (MeOH), cells were washed at least once with pre-cooled PBS.

Cell extracts were analysed in random order, along with quality controls (QC), on an ACQUITY UPLC I-Class/Vion IMS-QTOF high-resolution LC-MS system (Waters Corporation, Milford, MA, USA). QCs were assessed at the beginning and end of the sequence and regularly within the sequence.

For the chromatographic separation, reversed-phase chromatography was performed using a ZORBAX Eclipse XDB-C18 column (4.6 mm × 50 mm, 1.8 µm, Agilent Technologies, Santa Clara, CA, USA) maintained at a steady temperature of 30 °C. The analysis was conducted with a gradient elution procedure. Therefore, the initial mobile phase composition was set to 97% solvent A (water containing 0.1% formic acid) for the first 3 min. From 3 to 17 min, a gradient was applied to increase the proportion of solvent B (methanol with 0.1% formic acid) up to 97%. Between 17 and 22 min, the mobile phase was held constant with 97% solvent B. From 22 to 33 min, the system was re-equilibrated with 97% solvent A to restore the initial conditions. The total analysis time was 33 min per run, with a flow rate of 0.4 mL/min.

High-resolution mass spectrometry (MS) data was obtained using a Vion IMS-QT of mass spectrometer, equipped with an electrospray ionisation (ESI) source. The analysis was performed in both positive- and negative-ionisation modes. The capillary voltage was set to 3 kV in positive mode and 2.5 kV in negative mode. For both modes, the cone voltage was kept at 40 V, while the source temperature and desolvation gas temperature were set at 150 °C and 600 °C, respectively.

Data acquisition was controlled by UNIFI v1.9.4.0 software (Waters Corporation). The scan range was from m/z 50 to m/z 1000, with analyte fragmentation achieved by using nitrogen as the collision gas. A collision energy of 6 V was applied to generate a low-collision energy spectrum, while the collision energy was ramped from 21 to 41 V to obtain a high-collision energy spectrum.

Analysis of Untargeted Data Using MetaboAnalyst 6.0

For each ionisation mode, the raw abundance data was combined and uploaded as peak intensities into the one-factorial statistical analysis of MetaboAnalyst v6.0 (latest version: 23 October 2025). The data was normalised using Pareto scaling. For the volcano plots, the fold change (FC) threshold was set at 2.0 and the p-value (t-test and Wilcoxon Mann–Whitney test, applied by MetaboAnalyst) threshold at 0.05. Based on these statistically analysed data, potential metabolites were searched for in the Human Metabolome Database (HMDB v5.0). For this purpose, the masses were inserted into the LC-MS search, the adducts were selected for the positive (M+H, M+H-H₂O, M+Li, M+NH₄, M+Na and M+K) and negative-ion modes (M-H and M-H₂O-H), and 15 ppm was set as the molecular weight tolerance. The HMDB not only links one mass to one metabolite, but also different masses to the same metabolite or many metabolites to one mass.

Based on these HMDB metabolites, pathway analysis was performed using a hypergeometric test as the enrichment method and relative-betweenness centrality for the topology analysis.

In addition to this method, a metabolic enrichment analysis was performed using functional analysis [LC-MS1], which can serve as a further tool for identifying possible metabolites and metabolic pathways. For this purpose, the m/z values and p-values (ANOVA, based on the Progenesis QI software v1.0 (Waters Corporation)) were uploaded separately for negative- and positive-ion modes. The mass tolerance was set to 10 ppm, and the mummichog algorithm and Kyoto Encyclopedia of Genes and Genomes (KEGG) metabolic pathway library were selected.

2.6. Targeted Metabolomics of the Citrate Cycle

The targeted analysis of the citrate cycle was performed as previously described [24]. MSNs were treated once with ice-cold extraction reagents (acetonitrile (ACN), methanol (MeOH), water (H₂O); 2/2/1) including ¹³C-labelled substrates (0.25 µM citrate, 0.5 µM cis-aconitate, 0.25 µM itaconate, 2.5 µM lactate, 0.5 µM succinate) and washed twice with extraction reagents without internal standards. After thawing, the cell samples were centrifuged for 10 min at 4 °C, 20,800× g. The cell extracts were separated and measured using the high-performance liquid chromatography–tandem mass spectrometry (HPLC-MS/MS) method [24].

First, the sample was separated using a Shimadzu LC system (consisting of LC-30AD HPLC pumps, a SIL-30AC temperature-controlled autosampler, a DGU-20A5 degasser, a CTO-20AC oven, and a CBM-20A control unit; Shimadzu, Duisburg, Germany) on a Kinetex C18, 100 Å, 2.6 µm, 100 × 3 mm HPLC column (Phenomenex; Torrence, CA, USA). Mobile phase A consisted of water mixed with 0.2% formic acid. Mobile phase B was methanol (containing 0.2% formic acid). Chromatography started with a 99% proportion of mobile phase A. In order to achieve retention of the polar metabolites, these starting conditions were kept constant for 6 min. The proportion of mobile phase B was then increased to 90% within one minute. This composition was maintained for 1 min, followed by a 3 min re-equilibration to starting conditions. The entire analysis required 11 min at a flow rate of 0.4 mL/min. The column oven temperature was 30 °C.

After separation, the analytes were transferred to a QTRAP^® 5500 tandem mass spectrometer (Sciex, Framingham, MA, USA). Ionisation was performed using electrospray ionisation (ESI) operating in negative-ionisation mode. ESI parameters were as follows: ion spray voltage: −4500 V; curtain gas (CUR): 45 psi; collision gas (CAD): 9; temperature: 400 °C; gas 1: 60 psi; and gas 2: 75 psi. Analytes were detected in multiple reaction monitoring (MRM) mode. The mass transitions used for quantification were m/z 89/43 for lactate, m/z 191/111 for citrate, m/z 173/5 for cis-aconitate, m/z 117/73 for succinate, m/z 115/71 for fumarate, and m/z 133/115 for malate. The internal standards used were ¹³C₃-lactate (lactate), ¹³C₃-cis-aconitate (cis-aconitate), ¹³C₂-citrate (citrate), ¹³C₄-succinate (succinate and fumarate), and ¹³C₄-malate (malate).

The MS/MS chromatograms were detected by the mass spectrometer and then analysed using Analyst 1.7 software (Sciex, Framingham, MA, USA). The ratio of analyte peak areas and internal standard peak areas were used to obtain calibration curves for the metabolites, which were fitted with a quadratic regression using a weighting factor of 1/x.

2.7. Colorimetric Assays

Cells were scraped in assay buffer. For the enzyme activity assays of lactate dehydrogenase (Merck KGaA, MAK066), malate dehydrogenase (Merck KGaA, MAK196), and succinate dehydrogenase (Merck KGaA, MAK197), samples were centrifuged for 15 min at 4 °C, 10,000× g. Samples used for the colorimetric quantification of ATP (Abcam, ab83355) and NAD/NADH (Abcam, ab65348) were centrifuged twice for 10 min at 4 °C, 13,000× g. During the second centrifugation, the supernatants were filtered with a 10 kDa cut-off spin filter. All assays were performed according to the manufacturer’s protocols. The formulas specified in the manufacturer’s protocol were used to calculate enzyme activities, concentrations, and the NAD⁺/NADH ratio.

2.8. Quantification of the Total Cell Proteins

The thawed and overnight dried cell pellets were dissolved in 0.1 M sodium hydroxide solution (NaOH, Merck KGaA, 1091371000) by heating at 95 °C for 15 min. Protein concentrations were measured by using Pierce BCA Protein Assay Kit (Thermo Fisher Scientific Inc., A65453), Tecan Sunrise plate reader, and Magellan v7.2 SP1 software (Tecan Trading AG, Männedorf, Switzerland). The measurement wavelength was set to 570 nm; the reference wavelength to 630 nm.

2.9. Statistical Analysis

Statistical analyses were performed with GraphPad Prism v10.4.2. To allow for a more comprehensive illustration, the results were pooled into groups of CTR and MSA cell lines for comparison. An unpaired, two-tailed t-test was applied. Statistical significance was defined as p < 0.05. Outliers were tested with ROUT (Q = 1%). Data including mean and standard error of the mean (SEM) were displayed in bar graphs and scatter plots.

3. Results

3.1. Differentiation of hiPSCs into Striatal GABAergic MSNs

For the subsequent analyses, the cells were first examined for the expression of neuronal, GABAergic, and striatal markers in order to verify the 70-day differentiation to striatal GABAergic medium spiny neurons (MSNs). At least four independent differentiation approaches of three control and three MSA-P patient cell lines were evaluated by immunocytochemistry (ICC) staining and comparative gene-expression analysis (Figure 1). For a more comprehensive overview, the data of the controls (CTR) and the MSA cell lines (MSA) were grouped accordingly.

Figure 1.

Quality control of the striatal medium spiny neuron (MSN) differentiation after 70 ± 7 days. Representative images of ICC staining (showing the control cell line CTR2 in (A,B): 4’,6-diamidino-2-phenylindole (DAPI) in blue, neuron-specific βIII-tubulin (TUJ1) (A) and γ-aminobutyric acid (GABA) (B) in green, GABA (A) and COUP TF-1 interacting protein 2 (CTIP2) (B) in red (scale bar indicates 100 µm). Stained cells were counted and normalised to DAPI-, TUJ1- or GABA-positive cells from four to nine independent differentiations of three control (CTR) and three MSA-P patient cell lines (MSA). For each group, the independent differentiations of CTR (n = 15) and MSA (n = 19) were summarised and displayed as percentages (C). Gene expression of glutamic acid decarboxylase (GAD67) and forkhead box protein G1 (FOXG1) as GABAergic markers, CTIP2 as striatal marker, TUJ1, microtubule-associated protein 2 (MAP2) and forkhead box protein P1 (FOXP1) as neuronal markers was compared between MSNs and hiPSCs of each cell line (D). Five to ten independent differentiations were analysed by quantitative real-time PCR and summarised in fold expression (2^−ΔΔCt) as CTR (n = 21) and MSA (n = 19). Unpaired, two-tailed t-tests were applied (* p < 0.05, *** p < 0.001, **** p < 0.0001) (D). Graphs represent means ± SEM (C,D).

Representative images of 4’,6-diamidino-2-phenylindole (DAPI), neuron-specific βIII-tubulin (TUJ1), γ-aminobutyric acid (GABA), and COUP TF-1-interacting protein 2 (CTIP2) staining are presented in Figure 1A,B. For each group, the averaged percentages were summarised and displayed in Figure 1C. Of the DAPI-stained cells, approximately 96% were TUJ1-positive; 95% of CTR and 93% of MSA cells were also GABA-positive. Over 97% of TUJ1-stained neurons were also GABAergic (97% of MSA, 99% of CTR cell lines). The striatal marker CTIP2/DAPI was determined to be positive in about 87% of CTR and 86% of MSA cells and by CTIP2/GABA staining in 94% of CTR and 93% of MSA MSNs.

In the comparative gene-expression analysis, the fold expressions (2^−ΔΔCt) of MSNs were compared to their respective hiPSCs lines (Figure 1D). TUJ1, microtubule-associated protein 2 (MAP2), and forkhead box protein P1 (FOXP1) as neuronal markers, glutamic acid decarboxylase (GAD67) and forkhead box protein G1 (FOXG1) as GABAergic markers, as well as CTIP2, were significantly higher expressed in the differentiated target cell type.

Therefore, the results of the subsequent metabolome investigations originate mainly from homogenous cell cultures of differentiated striatal GABAergic medium spiny neurons.

3.2. Untargeted Metabolome Analysis

3.2.1. Statistical Analysis of Untargeted Metabolome Data

For the untargeted metabolome analysis, samples from four to five independent differentiations per cell line were measured in two experiments. In experiment 1, two MSA-P cell lines and two CTR cell lines were tested, whereas in experiment 2, one additional MSA-P cell line and one CTR cell line were compared separately due to logistical reasons.

The data obtained from the negative- and positive-ionisation modes were statistically tested in MetaboAnalyst v6.0 for both experiments (Figure 2). In experiment 1, 17 out of 20 samples clustered together in the scores plot of the principal component analysis (Figure 2A,C). In contrast, experiment 2 showed less overlap in the scores plot, indicating a deviation in the metabolome between the two groups (Figure 2E,G).

Figure 2.

Statistical analysis of untargeted metabolome data from MSA-P cell lines and control cell lines using MetaboAnalyst v6.0. Four to five independent differentiations per cell line were examined, with two MSA-P cell lines (n = 10) and two CTR cell lines (n = 10) measured and compared in experiment 1, and one MSA cell line (n = 5) and one CTR cell line (n = 4) studied in experiment 2. For each experiment, the raw abundance from the negative- and positive-ionisation modes was statistically analysed and displayed in the scores plot of the principal component analysis and in the volcano plot. The results from the negative-ionisation mode are shown in (A,B,E,F), while the results from the positive-ionisation mode are presented in (C,D,G,H). In the scatter plot, the red dots represent samples from CTR and the green dots samples from MSA. The fold changes in the volcano plots compared the peak intensities in the patient cell lines with those in the control cell lines and were considered significantly increased with log₂ > 1 and significantly decreased with log₂ < −1 in the MSA-P cell lines. The corresponding results with m/z values and p-values are provided in the Supplementary Materials (Tables S2–S5).

Both experiments resulted in significant differences in the comparison of peak intensities between the MSA group and the CTR group. In negative-ionisation mode, ten peaks were significantly reduced in the MSA group in experiment 1 (Figure 2B). Thirteen peaks were also found in positive-ionisation mode, which were also decreased in the MSA cell lines (Figure 2D). In experiment 2, both positive and negative changes were observed. When comparing the MSA group with the CTR group in negative-ionisation mode, fourteen increased and four decreased peak intensities were detected (Figure 2F). Most changes, 175 in total, were detected in positive-ionisation mode, with twelve displaying a negative fold change (Figure 2H). The corresponding m/z values with their fold changes and p-values can be found in the Supplementary Materials (Tables S2–S5).

3.2.2. Pathway Analysis of Untargeted Metabolome Data

Metabolites from the Human Metabolome Database (HMDB) were assigned to the respective m/z values that were examined in the pathway analysis for each experiment and ionisation mode. In experiment 1 (Figure 3A), five metabolic pathways were found in negative-ionisation mode, in which amino sugar and nucleotide sugar metabolism would be significantly affected (−log₁₀(p-value) = 6.2754).

Figure 3.

Pathway analysis of untargeted metabolome data from MSA-P cell lines and control cell lines using MetaboAnalyst 6.0. Four to five independent differentiations per cell line were examined, with two MSA-P cell lines (n = 10) and two CTR cell lines (n = 10) measured and compared in experiment 1, and one MSA cell line (n = 5) and one CTR cell line (n = 4) studied in experiment 2. For each experiment, the significant m/z values from the negative- and positive-ionisation modes were assigned to metabolites using Human Metabolome Database (HMDB) and used for pathway analysis. The results from the negative-ionisation mode are shown in (A,C), while the results from the positive-ionisation mode are presented in (B,D). p-values resulted from the hypergeometric test of the pathway enrichment analysis and indicate the probability that the observed number of metabolites from a specific metabolite set is more represented than would be expected by chance [25]. Pathway impact values (x-axis) are the sum of the importance values of the changed metabolites (from 0 to 1) obtained from the pathway topology analysis (relative-betweenness centrality). This value indicates how important the position of the altered metabolite is in this pathway based on the determination of the shortest paths of the metabolite to others [25,26]. Accordingly, the pathway impact values reflect the influence of the sum of these changed metabolites on the corresponding metabolic pathway. Lower p-values and higher pathway impact values are illustrated by the larger size (impact) and red colour (p-value) of the circles [25,26]. Higher p-values are gradually represented as yellow and orange circles.

In contrast, 27 metabolic pathways were identified in positive-ionisation mode (Figure 3B). With nine metabolites reduced, steroid biosynthesis would be significantly altered (−log₁₀(p-value) = 5.3395). Amino sugar and nucleotide sugar metabolism was also determined to be significant in positive-ionisation mode (−log₁₀(p-value) = 5.2461). Seven metabolites associated with significantly altered purine metabolism would also be reduced (−log₁₀(p-value) = 2.0995). Another metabolic pathway that could be altered in positive-ionisation mode would be pantothenate and CoA biosynthesis, with three reduced metabolites (−log₁₀(p-value) = 1.5536).

A total of 18 metabolic pathways were found in the negative-ionisation mode of experiment 2 (Figure 3C). Steroid hormone biosynthesis with seven elevated metabolites (−log₁₀(p-value) = 3.8627), and retinol metabolism with three elevated metabolites in the MSA group (−log₁₀(p-value) = 2.8037), would be involved.

As in the volcano plot from experiment 2 (Figure 3D), most changes were observed in the positive-ionisation mode. This resulted in 69 metabolic pathways, 9 of which were significantly altered. Again, steroid hormone biosynthesis was determined, but with 63 metabolites (−log₁₀(p-value) = 17.338). Other metabolic pathways included arachidonic acid metabolism with 28 metabolites (−log₁₀(p-value) = 5.9385), one carbon pool by folate with 16 metabolites (−log₁₀(p-value) = 3.3397), tryptophan metabolism with 20 metabolites (−log₁₀(p-value) = 2.3205), steroid biosynthesis with 20 metabolites (−log₁₀(p-value) = 2.3205), folate biosynthesis with 14 metabolites (−log₁₀(p-value) = 2.0395), linoleic acid metabolism with 4 metabolites (−log₁₀(p-value) = 1.5822), metabolism of xenobiotics by cytochrome P450 with 27 metabolites (−log₁₀(p-value) = 1.5046), and nicotinate and nicotinamide metabolism with 8 metabolites (−log₁₀(p-value) = 1.4114).

The complete list of metabolites and metabolic pathways is provided in the Supplementary Materials (Tables S6–S13).

3.2.3. Functional Analysis of Untargeted Metabolome Data

In addition to pathway analysis performed with HMDB metabolites from previous m/z mapping, those carrying out a functional analysis can skip this additional step and perform an enrichment analysis of the metabolic pathways based on the mass spectrum from the liquid chromatography–mass spectrometry (LC-MS). Using the mummichog algorithm and metabolic pathway library of the Kyoto Encyclopedia of Genes and Genomes (KEGG), the m/z values were assigned to possible metabolites and the significance of functional activity in the metabolic pathways was calculated [27]. Both significant and non-significant metabolites were considered for the enrichment analysis of metabolic pathways.

In experiment 1, a total of 24 metabolic pathways were found in negative mode (Figure 4A), with only the mannose type O-glycan biosynthesis being significantly enriched (–log₁₀(p-value) = 1.8663, enrichment factor = 17.0809).

Figure 4.

Functional analysis of untargeted metabolome data from MSA-P cell lines and control cell lines using MetaboAnalyst 6.0. Four to five independent differentiations per cell line were examined, with two MSA-P cell lines (n = 10) and two CTR cell lines (n = 10) measured and compared in experiment 1, and one MSA cell line (n = 5) and one CTR cell line (n = 4) studied in experiment 2. Based on the LC-MS mass spectrum, an enrichment analysis of the metabolic pathways was performed using the mummichog algorithm. The possible metabolites were assigned to the metabolic pathways and tested for local enrichment. Accordingly, a higher enrichment factor (x-axis) indicates the likelihood of a metabolic pathway reflecting the true activity [27]. The results from the negative-ionisation mode are shown in (A,C), while the results from the positive-ionisation mode are presented in (B,D). p-values resulted from Fisher’s exact test. Lower p-values and higher enrichment factors are illustrated by the larger size (enrichment factor) and red colour (p-value) of the circles [25,26]. Higher p-values are gradually represented as yellow and orange circles.

In contrast, 37 metabolic pathways were found in the positive mode (Figure 4B), none of which were statistically significantly enriched. Nevertheless, 11 of 16 metabolites were found in the citrate cycle (–log₁₀(p-value) = 0.1116, enrichment factor = 0.9239). The nicotinate and nicotinamide metabolism was only found in this experiment and mode (–log₁₀(p-value) = 0.5784, enrichment factor = 1.9710).

Eight metabolites from the citrate cycle were also identified in negative mode in experiment 2 (Figure 4C), which were likewise not significantly enriched (–log₁₀(p-value) = 0.2238, enrichment factor = 1.7576). However, of the 31 metabolic pathways determined, 3 metabolic pathways were significantly enriched. Three significant hits were observed in fatty acid biosynthesis (–log₁₀(p-value) = 2.9471, enrichment factor = 9.3741). Four metabolites were associated with linoleic acid metabolism (–log₁₀(p-value) = 2.3787, enrichment factor = 21.0911). A total of eight hits were identified for the terpenoid backbone biosynthesis.

Of 32 metabolic pathways, cysteine and methionine metabolism as well as the metabolism of xenobiotics by cytochrome P450 were determined to be significant in experiment 2 of the positive mode (Figure 4D). Nine hits were assigned to cysteine and methionine metabolism (–log₁₀(p-value) = 1.3610, enrichment factor = 1.1965). In the metabolism of xenobiotics by cytochrome P450 (–log₁₀(p-value) = 1.3269, enrichment factor = 1.5098), 34 of 68 metabolites were detected with 13 significances.

The metabolic pathways and significant metabolites are listed in the Supplementary Materials (Tables S14–S21).

3.3. Citrate Cycle Analysis

Although the citrate cycle was not significantly enriched in the analyses of the untargeted data, some metabolites associated with the citrate cycle were detected. Due to its important function in energy metabolism, due to the significant changes in the citrate cycle observed in a sporadic Parkinson’s cell model [28], and due to significant changes in mitochondrial processes found in MSA cell models [13,29], the citrate cycle was chosen as the starting point for a targeted metabolome analysis (Figure 5).

Figure 5.

Targeted metabolome analysis of the citrate cycle. The metabolites lactate (A), citrate (B), cis-aconitate (C), succinate (D), fumarate (E), and malate (F), which are involved in the citrate cycle (G), were measured in two to four independent differentiations of three control cell lines (CTR) and three MSA patient cell lines by HPLC-MS/MS. For each independent differentiation (biological replicate), two to three technical replicates were averaged. Each data point represents an averaged concentration of one independent differentiation, grouped by CTR (n = 8, blue circle) and MSA (n = 10, red square). Unpaired, two-tailed t-tests were used (* p < 0.05, ns = not significant). Means ± SEM are displayed (A–F).

The metabolites lactate (Figure 5A), citrate (Figure 5B), cis-aconitate (Figure 5C), succinate (Figure 5D), fumarate (Figure 5E), and malate (Figure 5F) were detected in the samples of three control cell lines and three MSA patient cell lines using high-performance liquid chromatography–tandem mass spectrometry (HPLC-MS/MS). Two to four independent differentiations with two to three technical replicates per cell line were analysed. The metabolite concentrations did not differ significantly between CTR and MSA cell lines for lactate (Figure 5A) citrate (Figure 5B), cis-aconitate (Figure 5C), fumarate (Figure 5E), and malate (Figure 5F). However, a decrease in succinate (Figure 5D) was observed in the MSA cell lines.

3.4. Enzyme Activity and Quantification Assays

In addition to the HPLC-MS/MS analysis of the citrate cycle, colorimetric assays were performed. For this purpose, three to five independent differentiations per cell line were analysed. The results were grouped according to the CTR or MSA cell lines shown in Figure 6A–H. One independent differentiation is represented as one data point.

Figure 6.

Enzyme activity and quantification assays. Three to five independent differentiations per cell line were used in the colorimetric assays to determine the enzyme activities of lactate dehydrogenase (LDH) (A), malate dehydrogenase (MDH) (B), and succinate dehydrogenase (SDH) (C), and the concentrations of adenosine-5’-triphosphate (ATP) (D), total nicotinamide adenine dinucleotide (NAD; NAD⁺ and NADH) (E), NAD⁺ (F), and NADH (G), as well as the NAD⁺/NADH ratio (H). For each enzyme activity assay (A–C), a total of 11 independent differentiations from three control cell lines (CTR, n = 11) and 12 from three patient cell lines (MSA, n = 12) were analysed. To quantify the ATP concentrations (D), 12 independent differentiations each from CTR (n = 12) and MSA (n = 12) groups were used. For all NAD concentrations (E–G) and corresponding NAD⁺/NADH ratios (H), 10 independent differentiations of the CTR (n = 10) and 14 of the MSA (n = 14) cell lines were investigated. Each differentiation is represented as one data point in the scatter plot. Unpaired, two-tailed t-tests were applied (* p < 0.05, ns = not significant); means ± SEM are displayed.

The enzyme activities of lactate dehydrogenase (LDH) (Figure 6A), malate dehydrogenase (MDH) (Figure 6B), and succinate dehydrogenase (SDH) (Figure 6C) were examined and showed no significant differences between CTR and MSA cell lines.

Further comparison of metabolite concentrations with CTR revealed a significantly lower adenosine-5’-triphosphate (ATP) concentration in MSA cell lines (Figure 6D). The concentrations of total nicotinamide adenine dinucleotide—NAD; NAD⁺, and NADH (Figure 6E); NAD⁺ (Figure 6F); and NADH (Figure 6G)—did not differ significantly between the cell lines. However, the NAD⁺/NADH ratios calculated from the concentrations showed an increase in the MSA group (Figure 6H).

The concentrations of total NAD, NAD⁺, and NADH, as well as the NAD⁺/NADH ratios, were also listed for each independent differentiation for comparability (Table 2).

Table 2.

Tabular overview of quantified NAD concentrations. The values from an independent differentiation were listed per row.

Group	Total NAD (ng/mg Protein)	NAD+ (ng/mg Protein)	NADH (ng/mg Protein)	NAD+/NADH Ratio
Control	1376.6	417.3	959.2	0.44
	5504.3	4155.4	1348.9	3.08
	9175.3	4313.4	4861.9	0.89
	1348.6	294.5	1054.1	0.28
	1301.1	374.5	926.6	0.40
	7437.0	3925.9	3511.1	1.12
	2694.2	1487.6	1206.5	1.23
	3236.9	1939.0	1297.9	1.49
	2594.4	1467.5	1126.9	1.30
	2266.0	1333.4	932.6	1.43
MSA	2393.8	934.1	1459.7	0.64
	6637.3	2450.1	4187.1	0.59
	2747.0	1009.8	1737.2	0.58
	3743.6	1611.8	2131.8	0.76
	1997.0	1376.8	620.2	2.22
	2590.0	1135.1	1454.9	0.78
	978.3	675.9	302.5	2.23
	825.2	615.3	210.0	2.93
	1300.0	1040.5	259.5	4.01
	1934.1	1541.2	393.0	3.92
	3020.8	2279.4	741.4	3.07
	2156.6	1603.7	552.9	2.90
	3213.8	2400.3	813.5	2.95
	4339.8	2862.8	1477.0	1.94

Abbreviations: NAD, nicotinamide adenine dinucleotide; MSA, multiple system atrophy.

To compare these results for total NAD, NAD⁺, NADH, and the NAD⁺/NADH ratio, the changes from other neurological and mitochondrial diseases were used as comparative examples (Table 3). In models using hiPSC-derived dopaminergic neurons from patients with sporadic Parkinson’s disease [28], GBA mutation [30], and LRRK2 mutation [31], a significant NAD⁺ decrease in LRRK2 cells, a significant NADH decrease in sporadic cells, and a significant NAD⁺/NADH ratio decrease in GBA cells were observed.

In late-onset Alzheimer’s disease (LOAD), a significant decline in total NAD, NAD⁺, and NADH levels was found in hiPSC-derived neural progenitor cells and astrocytes, as well as in fibroblasts, while NADH levels in fibroblasts were not significantly altered [32].

Leigh syndrome, a paediatric mitochondrial disorder, is also one of the neurodegenerative diseases that can be caused by numerous gene mutations [33]. A significant increase in NADH and the NAD⁺/NADH ratio were determined in the fibroblasts of patients with mutations in mitochondrial genes [34].

Further mitochondrial diseases with neurological symptoms can result from mutations in the POLG gene, which encodes the catalytic subunit of the mitochondrial DNA polymerase [35]. In neural stem cells derived from hiPSCs, homozygous and heterozygous POLG mutations showed a significant reduction in the NAD⁺/NADH ratio, with further significant decreases in NAD⁺ and NADH observed in cells with heterozygous mutations [36]. However, NAD⁺ was significantly increased in cells with homozygous mutations [36].

Table 3.

Comparison of the changes in direction of the total NAD, NAD⁺, NADH, and NAD⁺/NADH ratio in various neurological diseases. The changes in direction were shown for multiple system atrophy, Parkinson’s disease, late-onset Alzheimer’s disease, Leigh syndrome, and POLG-related disorders compared to controls.

Comparison	Total NAD	NAD+	NADH	NAD+/NADH Ratio
CTR vs. MSA
-HiPSC-derived striatal GABAergic medium spiny neurons	ns	ns	ns	sig. ↑
CTR vs. PD
-Sporadic, hiPSC-derived dopaminergic neurons [28]	nd	ns	sig. ↓	nd
-GBA, hiPSC-derived dopaminergic neurons [30]	nd	ns	nd	sig. ↓
-LRRK2, hiPSC-derived dopaminergic neurons [31]	nd	sig. ↓	nd	nd
CTR vs. LOAD [32]
-Fibroblasts	sig. ↓	sig. ↓	ns	nd
-HiPSC-derived neural progenitor cells	sig. ↓	sig. ↓	sig. ↓	nd
-HiPSC-derived astrocytes	sig. ↓	sig. ↓	sig. ↓	nd
CTR vs. LS [34]
-Fibroblasts	ns	ns	sig. ↑	sig. ↑
CTR vs. POLG-related disorders [36]
-Homozygous, hiPSC-derived neural stem cells	nd	sig. ↑	ns	sig. ↓
-Heterozygous, hiPSC-derived neural stem cells	nd	sig. ↓	sig. ↓	sig. ↓

Abbreviations: NAD, nicotinamide adenine dinucleotide; MSA, multiple system atrophy; PD, Parkinson’s disease; GBA, β-glucocerebrosidase; LRRK2, leucine-rich repeat kinase 2; LOAD, late-onset Alzheimer’s disease; LS, Leigh syndrome; POLG, DNA polymerase subunit gamma-1; ns, not significant; sig. ↑, significantly increased; sig. ↓, significantly decreased; nd, not determined.

3.5. Comparison of Untargeted and Targeted Metabolome Results

For comparison and clarification purposes, the results of the targeted and non-targeted analyses were summarised with the respective metabolites (Table 4). Fumarate and cis-aconitate were significantly elevated in the MSA group compared to the control group based on pathway analysis, but this could not be confirmed in the targeted analysis. The pathway analysis also showed that nicotinamide adenine dinucleotide (NAD⁺) was significantly elevated in the MSA group, which could not be verified by the NAD⁺ concentrations. However, a significant shift towards NAD⁺ was demonstrated based on the NAD⁺/NADH ratio.

Table 4.

Summary of metabolites examined in targeted metabolome analysis with results from untargeted metabolome analysis. Experiments 1 and 2 with negative- and positive-ionisation modes, and the analytical methods, pathway analysis, and functional analysis, are listed for the respective metabolites. The results from targeted metabolomics originate from the citrate cycle and colorimetric assays.

		Untargeted Metabolomics			Targeted Metabolomics
KEGG	Metabolite	Method	Experiment/Ionisation	Result	Result
C00002	ATP	Functional	2, negative	ns	sig. ↓
		Functional	2, positive	ns
C00003	NAD+	Pathway	2, positive	sig. ↑	ns
		Functional	1, positive	ns
C00022	Pyruvate	Functional	2, negative	ns	nd
C00024	Acetyl-CoA	Functional	1, positive	ns	nd
		Functional	2, negative	sig. ↑
C00026	2-Oxoglutarate	Functional	1, positive	ns	nd
		Functional	2, negative	ns
C00036	Oxaloacetate	Functional	1, positive	ns	nd
		Functional	2, negative	ns
C00042	Succinate			nd	sig. ↓
C00068	Thiamin diphosphate	Pathway	2, positive	sig. ↑	nd
C00122	Fumarate	Pathway	2, positive	sig. ↑	ns
		Functional	1, positive	ns
C00149	Malate	Functional	1, positive	ns	ns
		Functional	2, negative	ns
C00158	Citrate	Functional	1, positive	ns	ns
		Functional	2, negative	ns
C00311	Isocitrate	Functional	1, positive	ns	nd
		Functional	2, negative	ns
C00417	cis-Aconitate	Pathway	2, positive	sig. ↑	ns
		Functional	1, positive	ns
C01432	Lactate			nd	ns
C05125	2-(alpha-Hydroxyethyl)thiamine diphosphate	Pathway	2, positive	sig. ↑	nd
		Functional	1, positive	sig. ↑
		Functional	2, negative	ns
C05379	Oxalosuccinate	Pathway	2, positive	sig. ↑	nd
		Functional	1, positive	ns
C05381	Succinate semialdehyde-thiamin diphosphate	Functional	1, positive	ns	nd

Abbreviations: ATP, adenosine-5’-triphosphate; NAD⁺, nicotinamide adenine dinucleotide; KEGG, Kyoto Encyclopedia of Genes and Genomes; ns, not significant; sig. ↑, significantly increased; sig. ↓, significantly decreased; nd, not detected.

The comparison of the results highlights the difficulties and challenges involved in identifying metabolites in untargeted data. However, it was possible to show that there exist particular differences between healthy and patient groups, which can be investigated in more detail in future targeted studies.

4. Discussion

Multiple system atrophy (MSA) is a sporadic, fatal neurological disorder for which there is currently no approved disease-modifying treatment that could stabilise patients’ condition or at least significantly slow down the progression. Mitochondrial processes are known to be altered in neurodegenerative disorders such as MSA [8,9,10,11]. Reduced respiratory chain activity, increased mitochondrial mass, and elevated levels of the enzymes decaprenyl-diphosphate synthase subunit 2 (PDSS2), 4-hydroxy-3-methoxy-5-polyprenylbenzoate decarboxylase (COQ4), and aarF domain-containing kinase 3 (COQ8A), which are involved in the biosynthesis of coenzyme Q10 (CoQ10), were increased in MSA [13], possibly as compensatory mechanisms. Furthermore, our proteomic data from striatal GABAergic medium spiny neurons also showed significant changes in protein expression of mitochondrial proteins in MSA-P cell lines compared to healthy controls. The protein expressions of 2-methoxy-6-polyprenyl-1.4-benzoquinol methylase (COQ5), which is also involved in the biosynthesis of CoQ10 [13], the ubiquinol-cytochrome-c reductase complex assembly factor 1 (UQCC1), an assembly protein of complex III of the mitochondrial respiratory chain [37,38], and the large ribosomal subunit protein uL29m (MRPL47) were significantly upregulated, with at least a twofold increase in MSA cell lines [29].

Using untargeted metabolomics, significant differences were observed in the mass spectra. These were analysed in the pathway analysis using the previously matched metabolites from the Human Metabolome Database and in the functional analysis using the MetaboAnalyst database, which were tested for enrichment accordingly. Depending on the experiment, the ionisation mode of the mass spectrometer, and other analytical tools, various metabolic pathways were identified as significant. These included amino sugar and nucleotide sugar metabolism, purine metabolism, steroid biosynthesis, linoleic acid metabolism, retinol metabolism, the metabolism of xenobiotics by cytochrome P450, nicotinate and nicotinamide metabolism, fatty acid biosynthesis, and cysteine and methionine metabolism.

The untargeted analyses also revealed significantly altered metabolites from the citrate cycle, although the metabolic pathway did not differ significantly between MSA and control groups. In the subsequent targeted analysis, a significant reduction in succinate was observed in the striatal GABAergic medium spiny neurons of the MSA-P cell lines compared to the healthy control cell lines. In addition to the citrate cycle, succinate is also involved in complex II of the mitochondrial respiratory chain, where the anchored succinate dehydrogenase (SDH) oxidises succinate to fumarate and provides electrons from dihydroflavine-adenine dinucleotide (FADH₂) [39,40,41,42]. We also observed a significant change in the protein expression of succinate dehydrogenase [ubiquinone] iron–sulphur subunit (SDHB) in our proteome data, which was increased by a factor of 1.18 in the MSA cell lines, while succinate dehydrogenase [ubiquinone] flavoprotein subunit (SDHA), succinate dehydrogenase cytochrome b560 subunit (SDHC), and succinate dehydrogenase [ubiquinone] cytochrome b small subunit (SDHD) remained unchanged [29]. However, SDH activity showed no significant differences between control and MSA cell lines. Monzio Compagnoni [13] and colleagues could show that the complexes II and III were less active but more strongly expressed in the dopaminergic neurons of their MSA patient-derived cell lines, indicating compensatory mechanisms within the mitochondrial respiratory chain. The striatal GABAergic medium spiny neurons were also affected in mitochondrial processes, since we found a reduced ATP concentration in the MSA patient cell lines, which could be due to the lower succinate concentration and thus a reduced electron transfer from complex II to coenzyme Q10. Electrons in the respiratory chain are also provided by complex I from NADH supplied from the citrate cycle [39,40,41]. Our data showed a significantly higher NAD⁺/NADH ratio (Figure 6H) in MSA compared to control cell lines, with their concentrations tending to be lower in the patient cell lines (n.s.). The unbalanced NAD⁺/NADH ratio indicates a shift towards NAD⁺, suggesting possible NAD⁺ regeneration, which could take place at the NADH–ubiquinone oxidoreductase (complex I) point of the mitochondrial respiratory chain [43,44]. Similar findings were reported for dopaminergic neurons of sporadic Parkinson’s disease patients, where decreased ATP, NADH, succinate, fumarate, and malate were demonstrated in the patient cell lines [28]. In our targeted study analysis, malate and the MDH activity, which oxidises malate to oxalacetate by reducing NAD⁺ to NADH, were not affected. Another NAD⁺-coupled enzyme is LDH, oxidising lactate to pyruvate under aerobic conditions. LDH activity was also not significantly different between MSA and control cell lines. For these two enzymes, which serve as examples of the NAD⁺-dependent dehydrogenases, the enzyme activities remained unchanged.

The examples listed for total NAD, NAD⁺, NADH, and NAD⁺/NADH ratio demonstrate several aspects. Despite the underlying commonality of Parkinson’s disease, diverse genetic backgrounds led to different changes. The study on late-onset Alzheimer’s disease, using the same methodology within the study, revealed similar significant decreases in nicotinamide adenine dinucleotides in several cell types. Depending on mitochondrial genetic defects, the results of NAD studies may vary. MSA fits into this list. However, further studies on other cell types and patients are required to obtain a broader picture. A reduced succinate concentration could be the result of a decreased conversion of succinyl-CoA, which might be less available in the MSA patient cell lines. Succinyl-CoA can be synthesised from propionyl-CoA, which is derived from the amino acids leucine, isoleucine, valine, methionine, and threonine, odd-chain fatty acids, or cholesterol [39,42,45,46,47]. Furthermore, succinyl-CoA is also involved in the biosynthesis of porphyrin such as heme [39,42,45,46], so that further potential processes could be affected. The enzyme succinyl-CoA synthetase catalyses succinyl-CoA to succinate, whereby either adenosine diphosphate (ADP) is phosphorylated to ATP or guanosine diphosphate (GDP) is phosphorylated to guanosine triphosphate (GTP) [39,42,46], which may also have an effect on the ATP production. In further targeted studies, the involved metabolic pathways can be analysed to determine whether the decrease in succinate is due to alterations in succinyl-CoA concentration or processes that would explain the reduction in succinate and ATP in MSA cell lines. The other metabolic pathways investigated in this study can also be used as a guide for future targeted metabolome analyses in MSA.

Limitations

This study reports the metabolic results of three MSA and three control cell lines derived from six different individuals, leading to larger variations in the measured data which may contribute to the underestimation of potentially relevant findings due to statistical noise. To minimise individual effects, more cell lines could be added to the data, but the availability of MSA patient-derived cells is still very limited. Furthermore, the results focus on the striatal GABAergic medium spiny neurons, so the metabolic interplay with other cells is not fully represented. For this purpose, additional relevant cell types such as oligodendrocytes should be included in MSA stem-cell models of future studies. In addition, suitable MSA animal models can be used to investigate metabolomic mechanisms throughout the entire organism [48,49]. Finally, further targeted metabolome studies should be carried out on the basis of the investigated metabolic pathways in order to gain more mechanistic insights into the altered metabolism of the disease.

5. Conclusions

This study presents the first metabolome analyses of an iPSC model of MSA. By comparing striatal GABAergic medium spiny neurons from MSA-P patient-derived cell lines and matched healthy controls, metabolome alterations were detected in untargeted and targeted approaches. The screening for differences in peak intensities and the matching with metabolites found in databases led to the identification of several metabolites and metabolic pathways such as amino sugar and nucleotide sugar metabolism, purine metabolism, nicotinate and nicotinamide metabolism, and the citrate cycle in the untargeted analysis. The targeted investigation was carried out for the citrate cycle, showing a significantly decreased succinate concentration in the patient cell lines compared to controls. Further investigations revealed a markedly reduced ATP concentration and an imbalanced NAD⁺/NADH ratio in the MSA cell lines. These results indicate an insufficient electron supply to the mitochondrial respiration chain, leading to insufficient energy production and availability of ATP in the MSA patient-derived striatal medium spiny neurons.

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/biom16020190/s1, Table S1: Primer sequences used for RT-qPCR [7,23]; Table S2: Significant changes in the peaks detected in the negative-ionisation mode of the untargeted metabolome analysis in experiment 1; Table S3: Significant changes in the peaks detected in the negative-ionisation mode of the untargeted metabolome analysis in experiment 2; Table S4: Significant changes in the peaks detected in the positive-ionisation mode of the untargeted metabolome analysis in experiment 1; Table S5: Significant changes in the peaks detected in the positive-ionisation mode of the untargeted metabolome analysis in experiment 2; Table S6: Results of the pathway analysis using significant metabolites obtained in the negative-ionisation mode from the untargeted metabolome analysis in experiment 1 [25,26]; Table S7: Overview of metabolic pathways and significant metabolites from pathway analysis found in negative-ionisation mode in experiment 1 [25,26]; Table S8: Results of the pathway analysis using significant metabolites obtained in the positive-ionisation mode from the untargeted metabolome analysis in experiment 1 [25,26]; Table S9: Overview of metabolic pathways and significant metabolites from pathway analysis found in positive-ionisation mode in experiment 1 [25,26]; Table S10: Results of the pathway analysis using significant metabolites obtained in the negative-ionisation mode from the untargeted metabolome analysis in experiment 2 [25,26]; Table S11: Overview of metabolic pathways and significant metabolites from pathway analysis found in negative-ionisation mode in experiment 2 [25,26]; Table S12: Results of the pathway analysis using significant metabolites obtained in the positive-ionisation mode from the untargeted metabolome analysis in experiment 2 [25,26]; Table S13: Overview of metabolic pathways and significant metabolites from pathway analysis found in positive-ionisation mode in experiment 2 [25,26]; Table S14: Results of the functional analysis using mass spectra obtained in the negative-ionisation mode from the untargeted metabolome analysis in experiment 1 [27]; Table S15: Overview of metabolic pathways and significant metabolites from functional analysis found in negative-ionisation mode in experiment 1; Table S16: Results of the functional analysis using mass spectra obtained in the positive-ionisation mode from the untargeted metabolome analysis in experiment 1 [27]; Table S17: Overview of metabolic pathways and significant metabolites from functional analysis found in positive-ionisation mode in experiment 1; Table S18: Results of the functional analysis using mass spectra obtained in the negative-ionisation mode from the untargeted metabolome analysis in experiment 2 [27]; Table S19: Overview of metabolic pathways and significant metabolites from functional analysis found in negative-ionisation mode in experiment 2; Table S20: Results of the functional analysis using mass spectra obtained in the positive-ionisation mode from the untargeted metabolome analysis in experiment 2 [27]; Table S21: Overview of metabolic pathways and significant metabolites from functional analysis found in positive-ionisation mode in experiment 2.

Author Contributions

Conceptualisation, N.J.S., T.G., L.Y., M.K. and F.W.; methodology, N.J.S., H.B., T.G., S.G. and L.M.H.; formal analysis, N.J.S., H.B. and A.D.F.; investigation, N.J.S. and H.B.; resources, A.D.F. and L.M.H.; writing—original draft preparation, N.J.S.; writing—review and editing, H.B., T.G., L.Y., M.K., A.D.F. and F.W.; visualisation, N.J.S.; supervision, F.W.; funding acquisition, F.W. All authors have read and agreed to the published version of the manuscript.

Institutional Review Board Statement

The study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of Hannover Medical School (No. 8666_BO_K_2019 and date of approval: 13 September 2019).

Informed Consent Statement

Written informed consent has been obtained from the patients.

Data Availability Statement

All data are available from the corresponding author by reasonable request.

Conflicts of Interest

The authors declare no conflicts of interest.

Funding Statement

This study was financially supported by the Karlheinz-Hartmann-Stiftung, Hannover, Germany.