Myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS) is a heterogeneous, debilitating, complex, and controversial disorder with many uncertainties regarding both aetiology and delimitation towards other syndrome diagnoses such as fibromyalgia and postural orthostatic tachycardia syndrome (POTS) [1
]. ME/CFS is characterised by extreme mental and physical fatigue and fatigability associated with symptoms of pain, instability in the control of a broad range of organ systems, tendency towards hypersensitive sense modalities, and inability to respond appropriately to stressors of all sorts causing post-exertional malaise [3
]. As both pathophysiology and aetiology are undetermined and no definite biomarkers exist, the diagnosis of ME/CFS is based on exclusion criteria and symptoms [4
]. ME/CFS is often found to be preceded by infections or prolonged extreme physiological or psychological strain [6
]. The societal importance and impact of ME/CFS are considerable with an estimated prevalence of 0.2–0.8% [7
], with symptom severity comparable to other chronic and severe conditions such as depression, multiple sclerosis, and stroke [9
], and as few as 6% of diagnosed individuals returning to premorbid function [6
The lack of strict and univocal diagnostic criteria coupled with marked heterogeneity and lack of valid biomarkers has hampered diagnosis, research, and comparison of studies. Both the aetiology and pathophysiology of ME/CFS remain undetermined. However, in the last decade, several studies have shown diverse biological markers being significantly affected in ME/CFS patients revealing a disease state characterised by sustained autonomic nervous system (ANS) dysfunction, inflammation, altered metabolism, and mitochondrial dysfunction [3
]. The alteration observed in the ANS of ME/CFS patients is indicative of increased sympathetic and reduced parasympathetic regulation, especially autonomic regulation of the cardiovascular system [11
]. The resulting orthostatic intolerance (OI) is a core symptom of ME/CFS and is mirrored in the frequent comorbidity of ME/CFS and POTS [2
]. Moreover, studies have described the presence of G-protein coupled autoantibodies with agonistic effects on neuroendocrine receptors in the ANS in both ME/CFS and in associated conditions such as POTS and chronic regional pain syndrome (CRPS) [18
]. The neuroendocrine receptors mediate the control of ANS through crosstalk with mitochondrial function and metabolism [22
ME/CFS patients have shown to be metabolically reprogrammed with significant deviations in pathways affecting sphingolipids, phospholipids, purines, and amino acids with impaired metabolic responses to environmental stressors centred around mitochondrial function [26
]. Moreover, studies have shown dysfunction of mitochondrial bioenergetics and regulations in mitochondrial respiratory chain complexes to compensate for inefficient ATP production [31
]. Therefore, abnormal mitochondrial energy metabolism has become an area of interest in ME/CFS research in recent years [31
]. In this study, we aim to perform deep profiling of the mitochondrial function and evaluate its association with symptom burden in ME/CFS through a case–control design with a well-defined cohort of patients that have developed post-viral ME/CFS. Mitochondrial function was evaluated in peripheral blood mononuclear cells (PBMCs) that have shown their capabilities to assess systemic changes in mitochondrial function in other chronic diseases and the correlation between mitochondrial function and physical and cognitive dysfunction [35
2. Materials and Methods
2.1. Study Design
Six female patients between 30 and 50 years old were enrolled in the study, diagnosed with ME/CFS by a ME/CFS specialist, and recruited through the Danish ME association and with a mean illness duration of 12 years (Supplementary Materials, Table S1
). Participants, patients and controls were recruited with the following characteristics: female, age 30–50 years old, and non-smokers. The exclusion criteria for the participants were: known acute or chronic illness, presence of autoimmune disease, and known mental illness. During recruitment, all participants were asked about exercise and lifestyle to avoid excessive differences in physical activity between patients and controls. The patients were asked to abstain from medications—including supplements—a week prior to the examination. All the participants, patients and controls, were tested during 2016 and 2017. Blood samples were collected 5 hours after a light breakfast. Participants had abstained from medication intake and extreme exercise for the previous week. Standard blood tests were conducted to discard possible differential diagnosis. The questionnaires were answered within 2 months of the blood samples collection. The patients were submitted to a narrative, clinical interview and schematic questioning using a checklist of symptoms filled in by the patients at home and by the staff at the clinic to elucidate symptom burden and onset. The onset of symptoms was mainly after viral infections. Patients 1 and 6 described acute onset of symptoms after tonsillitis and a combination of physical and psychological stress, respectively. Patients 4 and 5 described subacute onset following mononucleosis and episodes of pyelonephritis, respectively. Finally, patients 2 and 3 described insidious symptoms after several viral infections. These evaluations were supplemented with three self-assessed questionnaires of well-being: Fatigue Scale of Motor and Cognitive function (FSMC), Medical Outcomes Study 36-Item Short-Form Health Survey (SF-36), and COMPASS-31. FSMC evaluates fatigue symptoms by providing a global fatigue severity score as well as subscales of motor and mental fatigue [39
]. SF-36 is a questionnaire that evaluates the health-related quality of life [40
]. COMPASS-31 specifies and quantifies symptoms and severity of autonomic dysfunction [41
2.2. Isolation of Human PBMCs Using a Vacutainer Cell Preparation Tube
Blood samples were collected by venipuncture in 8 mL citrate-containing Vacutainer cell preparation tubes (CPT) (Becton, Dickinson and Company, Frankline Lakes, NJ, USA) (Figure S1a
). PBMCs were isolated by differential centrifugation. The PBMC pellet was carefully resuspended in ammonium–chloride–potassium (ACK) lysis buffer (Gibco, Life Technologies Corporation, Grand Island, NY, USA) and incubated for 10 min to lyse any remaining red blood cells. Then, the PBMC pellet was washed with PBS (Sigma Aldrich, Steinheim, Germany) by centrifugation and was finally resuspended in Seahorse assay media (Agilent Technologies, Cedar Creek, TX, USA). The PBMCs yield and viability were determined using via-1 cassettes™ in image cytometer NC-3000 (Chemometec, Allerod, Denmark) according to the manufacturer’s instruction. The PBMCs viability threshold for inclusion was 90%. Cells were used immediately after isolation for bioenergetics analysis or frozen at −80 °C for further analysis.
2.3. Bioenergetics Profile
PBMCs (80,000 cells/well) were seeded in Seahorse 96-well (Seahorse XFe96, Agilent Technologies, MA, USA) pre-coated plates with 0.1 mg/mL Cell-tak™ (Corning, Discovery Labware Inc, Bedford, MA, USA). Media was freshly prepared with XF base media (Agilent Technologies, Cedar Creek, TX, USA) (pH 7.4) supplemented with 10 mmol/L glucose, 2 mmol/L sodium pyruvate, and 2 mmol/L glutamine (Sigma Aldrich, Hamburg, Germany). The bioenergetic profile comprised a total of 14 parameters (Table S2
). The Oxygen Consumption Rate (OCR) and Extracellular Acidification Rate (ECAR) were measured simultaneously in three cycles of mixing (3 min) and measuring (3 min) for each section: basal, injection 1, and injection 2. Three different groups of sequential injections were performed:
1 μmol/L Oligomycin (inhibitor of complex V); 2 μmol/LM Rotenone/2 μM Antimycin (inhibitor of complex I and complex III, respectively);
6 μmol/L FCCP (uncoupler); 2 μmol/L Rotenone/2 μM Antimycin;
1 μmol/L Oligomycin; 50 mmol/L 2-Deoxy-D-glucose (2-DG) (inhibitor of glycolysis).
The bioenergetic profile consists of 14 parameters obtained by combining the OCR and ECAR levels in basal conditions and in response to the different inhibitors; refer to Figure S1
for a schematic representation of the bioenergetics profile analysis. The parameters and their formulas are described in Table S2
. The different bioenergetic parameters were calculated for each well. Then, the replicates were pooled for each bioenergetic parameter to calculate the mean and the standard deviation.
To account for inter-experiment variations, we calibrated the Seahorse instrumentation before every experiment to account for day-to-day variation. Moreover, an inter-experiment blood sample from a healthy individual not related to this study was drawn on the same day as this study participants and was prepared and analysed simultaneously. We used this sample to assess overall assay performance, and it was not included in the bioenergetics profile calculations of the individuals in this study.
2.4. Protein Carbonyl Content
The amount of protein carbonylation was determined using the Oxiselect protein carbonyl ELISA kit from Cell Biolabs (San Diego, CA, USA), according to the manufacturer’s protocol. Briefly, each sample was analysed by triplicate with 1 μg of total protein lysate in each well. The proteins were adsorbed to the plate overnight at 4 °C, followed by washing and incubation with DNPH (2,4-Dinitrophenylhydrazine) at 0.04 mg/mL for 45 min. The excess DNPH was removed by washing with a mix of PBS/ethanol (1:1, v/v). The anti-DNP (2,4-dinitrophenyl) antibodies and the horseradish peroxidase-conjugated antibodies were incubated for 1 h with washing steps in between. The plate was read at 450 nm in a Synergy H1 microplate reader (Biotek, Winooski, VT, USA) using reduced BSA as a blank.
2.5. Protein Extraction and Sample Preparation
The frozen PBMC pellets were resuspended in 50 mmol/L HEPES lysis buffer (pH 7.4) with 0.3% SDS and protease-inhibitor cocktail tablets (Roche Diagnostic GmbH, Mannheim, Germany). The protein suspension was treated with ultrasonication (Branson Sonifier 250, Branson ultrasonics corp, Brookfield, CT, USA) on ice water, at output control 3 and 30% duty cycle for 3 rounds of 3 pulses with 1 minute on ice between each round. Insoluble proteins were removed by centrifugation at 16,000× g for 10 min at 4 °C.
2.6. Tandem Mass Tag Labelling, Isoelectric Focusing Separation, and Purification of Peptides
Equal protein amounts (80 μg) were processed according to the tandem mass tag (TMT) 11-plex manufacturer’s instructions (Thermo Fisher Scientific, Rockford, IL, USA). Briefly, each sample was sequentially reduced and alkylated with 200 mmol/L TCEP (tris(2-carboxyethyl)phosphine hydrochloride) and 375 mmol/L iodoacetamide, respectively. Then, the proteins were precipitated with acetone (−20 °C, overnight), followed by digestion with 2.5 μg trypsin. The peptides were labelled with different TMT labels for each sample. After TMT-labelling, the peptide samples were combined and subsequently purified using a strong cation exchange (Strata SCX, 55 µM, 70A, 100 mg/mL, Phenomenex, Torrence, CA, USA). The peptides were eluted with a mixture of 5% of ammonia and 30% methanol (Merck, Darmstadt, Germany) and subsequently vacuum-dried. The peptides were fractionated by isoelectric focusing (IEF) on a Multiphor II-unit (Pharmacia Biotech AB, Bromma, Sweden) using an 18 cm Immobiline Drystrip Gradient (IPG) pH 3–10 gel (GE Healthcare, Uppsala, Sweden). The sample was dissolved in rehydration solution containing 8 mol/L urea, 0.5% IPG buffer 3–10 (GE Healthcare, Uppsala, Sweden) and 0.002% bromophenol blue. The strip was then rehydrated with the sample at room temperature overnight. IEF was run at 59 kVh with the following program: 1 min gradient from 0–500 V, 1.5 h gradient from 500–3500 V followed by 16 h at 3500 V. The gel strip was cut in 10 pieces and peptides were extracted in three steps, of 1 hour each, with 100 μL 5% acetonitrile (AcN) (LC-MS grade, Merck, Darmstadt, Germany), 0.5% trifluoracetic acid (TFA) (Sigma Aldrich, Hamburg, Germany), and purified on PepClean C-18 Spin Columns (8 mg C18 resin, Pierce, Rockford, IL, USA) according to manufacturer’s protocol.
2.7. Nano-Liquid Chromatography and Mass Spectrometry (MS) Analysis
Each of the purified peptide mixtures was analysed twice by nano Liquid-Chromatography (nLC) (Easy-nLC 1000, Thermo Fisher Scientific, San Jose, CA, USA) coupled to a mass spectrometer (Q Exactive Plus, Thermo Fisher Scientific, Bremen, Germany) through an EASY-Spray nano-electrospray ion source (Thermo Fisher Scientific, Bremen, Germany). Pre-column (Acclaim PepMap 100, 75 µm × 2 cm, Nanoviper) and analytical column (EASY-Spray column, PepMap RSLC C18, 2 µm, 100 Å, 75 µM × 25 cm), both from Thermo Fisher Scientific (Vilnius, Lithuania) were used to trap and separate peptides using a 170-min gradient (5–40% Acetonitrile, 0.1% Formic acid). The MS was operated in positive mode and higher-energy collisional dissociation (HCD) with collision energy (NCE) of 32 for peptide fragmentation. Full-scan (MS1) resolution was 70,000, and AGC target set at 1 × 106 with scan range between 391 and 1500 m/z. Data-dependent analysis (DDA) was applied to fragment up to 10 of the most intense peaks in MS1. Resolution for fragment scans (MS2) was set at 35,000 with first fixed mass at 121 m/z and AGC target at 2 × 105. Dynamic exclusion was set at 19 seconds, and unassigned and +1 charge states were excluded. Furthermore, peptides with more than 10 peptide spectrum matches (PSMs) in the first analysis were excluded from the second analysis.
2.8. Proteomic Database Search and Statistical Analysis
The final database search was conducted in Proteome Discoverer 2.1 (Thermo Fisher Scientific, Bremen, Germany) with Mascot (Matrix Science, London, UK) on all raw files merged. Swiss-Prot was used as database with maximum two missed cleavages using trypsin as enzyme, and taxonomy was set for Homo sapiens
(20,350 reviewed sequences from uniprot.org, September 2019). Precursor and fragment mass tolerance were set at 10 ppm and 20 mmu, respectively. Oxidation of methionine was set as dynamic modification, and static modifications were carbamidomethylation of cysteines and TMT-labels on lysine and peptide N-terminus. Co-isolation threshold set at 50%. Identification false discovery rate was set to 0.01. Proteins with a minimum of two unique peptides and five quantitative scans were included for further statistical analysis. The mitochondrial proteins were determined using the Mitocarta list with a 7% false discovery rate [42
]. The proteomics data were filtered to obtain differentially altered proteins (DAPs) passing two criteria: fold change (FC) > 1.12 and <0.89 and student t
-value < 0.05. The FC criterion was established based on two times the median of the coefficient of variation of the control group. The group p
-value of the pathways was performed by student t
-test of the proteins involved in each pathway.
2.9. Targeted Metabolomics Profiling
Targeted metabolomics analysis of 408 metabolites in plasma samples was carried out using the AbsoluteIDQ®
p400 Kit (Biocrates Life Sciences AG, Innsbruck, Austria) and following the manufacturer’s instructions. Details on our analysis equipment and materials have been published previously [43
]; additional details are available in Supplementary Materials
. Of the 408 metabolites, 42 were acquired by liquid-chromatography (LC) coupled with high-resolution mass spectrometry (HRMS), namely, the biogenic amines and the amino acids, and 366 were acquired by flow-injection analysis (FIA) coupled with HRMS, namely, the acylcarnitines, glycerophospholipids, sphingolipids, hexoses, cholesterol esters, and glycerides (details in Supplementary Materials
Plasma samples were aliquoted in cryotubes and stored at −80 °C until analysis. Samples were extracted according to instructions (10 µL) and injected in a randomised order on the same analytical plate and along with a 7-point calibration curve (LC-HRMS only), three blanks, and three quality control levels (QC1-3), of which QC1 and QC3 in singlicates and QC2 in triplicates. The normalisation of results using the median value of the QC2 replicates in comparison with their target values (as recommended by the manufacturer) was performed using MetIDQ Carbon-2793 software (Biocrates Life Sciences AG, Innsbruck, Austria).
We performed extensive quality control and filtering of the measurements using the MeTaQuaC R-based package v0.1.30 [43
] (RStudio 1.3.1093, R 4.0.3) independently for LC and FIA measurements (as recommended). The automatically generated reports detail all parameters used and are available in the Electronic Supplementary Materials
. Preprocessed data (section 3.9 of the ME-CFS_MeTaQuaC_biocrates_qc_p400 reports in the Electronic Supplementary Materials
) were exported for further analysis. Statistical analyses were performed using MetaboAnalystR3.0.3 [45
] in RStudio (see Electronic Supplementary Materials
). Missing values (1.2% after preprocessing) were replaced by 1/5 of the minimal positive values of their corresponding variables. Concentrations were further glog (generalised logarithm) transformed and Pareto scaled. Analyses included FC analysis, t
-tests (false discovery rate, FDR, correction), principal component analysis (PCA), and hierarchical clustering heatmap. All original R scripts are available on https://github.com/SSI-Metabolomics/ME-CFS__SupplementaryMaterial
(Accessed on: 4 January 2021).
2.10. Acylcarnitines and Organic Acids Profile
The plasma samples consist of plasma isolated by centrifugation from EDTA blood. For the measurement of amino acids, we first measured the summed molarity of L-amino acids (termed total amino acids) by an enzyme-linked immunosorbent assay (ELISA) (catalogue no. ab65347; Abcam, Cambridge, UK). This kit determines concentrations of free l-amino acids but not protein-bound or D-amino acids. Second, we identified and quantified each of the free amino acids by using the MassTrak Amino Acid Analysis (AAA) Solution Kit, an Acquity UPLC system equipped with a C18 BEH column (1.7 µm; 2.1 × 150 mm) and an integrated photodiode array detector operating at 260 nm (all from Waters Corporation, Milford, MA, USA). The samples were prepared according to the manufacturer’s instructions. Briefly, plasma was deproteinised with an equal amount of sulphosalicylic acid (10%) containing norvaline as internal standard and thoroughly mixed. After centrifugation, part of the supernatant was alkalised with a borate buffer/NaOH solution, derivatised with 6-aminoquinolyl-N-hydroxysuccinimidyl carbamate (AQC), and analysed by UHPLC-UV. UHPLC conditions were in accordance with the manufacturer’s instructions, except: (1) instead of MassTrak AAA Eluent B, we used AccQ-Taq Ultra Eluent B (Waters) to enhance separation of amino acid eluting midway in the chromatogram (1); and (2) a steeper gradient curve between 2 and 5.5 min to enhance separation of arginine and glycine [46
]. Further details are available upon request.
The plasma concentration of free carnitine and 15 short-chain acylcarnitines, 6 medium-chain acylcarnitines, and 14 long-chain acylcarnitines were measured by ultra-HPLC tandem MS [47
]. In short, 250 mL of plasma was mixed with deuterium-labelled internal standard and deproteinised with 700 mL of AcN. After centrifugation, the supernatant was evaporated and resuspended in 100 mL 1:3 methanol/solvent A. We injected a 2 mL sample on an Acquity UPLCTM (Waters, Milford, MA, USA) equipped with an Acquity UPLCTM BEH C18 column (2.1 3 100 mm, 1.7 mm), using a flow rate of 0.35 mL/min and kept at 40 °C. The elution mobile phase consisted of a linear gradient using 0.2% heptafluorobutyric acid in 10 mM ammonium formate (solvent A) and 100% methanol (solvent B): 2.5–20% solvent B for 0.34 minutes, 20–90% solvent B for 18 min, and 90–100% solvent B for 1 min. Multiple reaction monitoring was performed on a Quattro MicroTM Tandem Mass Spectrometer (Waters Corporation, Milford, MA, USA) with an electrospray ionisation interface under positive-ion detection mode. The interface heater was maintained at 300 °C, and ion spray voltage was 3 kV. The dissolving gas (nitrogen), cone gas (nitrogen), and collision-activated dissociation gas (argon) were set at 700 L/h, 50 L/h, and 10 psi, respectively. The dwell time for each analyte was 75 ms. Optimisation of mass transitions and collision energy was performed by flow injection analysis of standard solutions.
2.11. Steroid Analysis
We used the targeted LCMS CHS MSMS Steroids Kit (PerkinElmer Inc., Waltham, MA, USA) to measure the concentration of 17-hydroxyprogesterone, testosterone, androstenedione, and cortisol in the plasma samples. More detailed methods are in Supplementary Materials
2.12. Statistical Analysis
All results are presented as means ± standard deviation (SD) unless stated otherwise. The patients’ clinical characteristics and steroid plasma levels were compared by student t
-test for the continuous variables. Bioenergetics values were compared between the two groups by t
-test with Bonferroni correction for multiple comparisons. Bivariate correlation analysis was performed using Pearson’s r coefficient. Statistical analysis was performed using R studio version 1.2.5033 [48
] and Excel (Microsoft, Redmond, WA, USA).
In this study, we describe a comprehensive molecular evaluation of a small cohort of ME/CFS patients presenting with severe fatigue (FSMC score > 63), both mental and physical. The FSMC scores are significantly higher than FSMC scores reported in patients with inherited mitochondrial diseases [55
], patients with stroke [56
], and multiple sclerosis [39
]. Moreover, the information collected about functional status and well-being (SF-36) showed limitation due to physical health, not emotional problems (Table 3
). Other studies have shown that the SF-36 scores for ME/CFS patients are lower than those of other chronic diseases such as type 2 diabetes, cardiovascular diseases, and multiple sclerosis [9
]. Regarding symptom severity mirroring ANS dysfunction, the patients had a score 27 times higher than the controls in the COMPASS-31 questionnaire, which is comparable to patients diagnosed with ANS failure [58
]. Interestingly, we found a correlation between FSMC score and COMPASS-31 score, suggesting that ANS dysfunction is a central driver of the patient’s fatiguing illness. Several of the symptoms evaluated by the COMPASS-31 can be explained by the possible dysregulation of autoantibodies directed against adrenergic and muscarinic receptors [18
]. Adrenergic signalling results in changes in mitochondrial function and the production of reactive oxygen species (ROS) [59
]. Future studies of a larger cohort should investigate the relationships between symptom score, autoantibodies, ANS, and mitochondrial dysfunction in ME/CFS.
The PBMCs bioenergetic profile of the ME/CFS patients showed a significantly lower coupling efficiency compared to controls in basal conditions. Coupling efficiency is a tightly regulated process that represents how much ATP is produced per molecule of oxygen. The lower coupling efficiency observed leads to a lower ATP production and could be related to the slightly higher proton leakage seen in the patients’ cells. We also observed a tendency of higher spare respiratory capacity reflecting how much of the maximal respiration is being used by the cells. This higher spare respiratory capacity could be a compensatory mechanism producing more mitochondrial ATP, especially under stress conditions such as higher physical or mental activity. However, we observed a tendency of decreased ATP-linked to maximal respiration, indicating that PBMCs from ME/CFS patients cannot produce as much ATP as the controls even with increased spare respiratory capacity. Similar to our study, immortalised lymphocytes from ME/CFS patients have shown increases in proton leakage and spare respiratory capacity with non-significant changes in basal respiration [31
], hereas another study on PBMCs from ME/CFS patients has shown no differences in coupling efficiency and a lower spare respiratory capacity [34
]. This discrepancy can be due to technical differences in the protocols used to measure mitochondrial respiration. This previous study used frozen PBMCs, while we used fresh PBMCs, and the third study used immortalised cells. These different procedures could lead to changes in mitochondrial respiration. Moreover, we used a modified protocol to avoid the sequential injections of oligomycin and FCCP that can lead to underestimation of the maximal respiration and spare respiratory capacity [61
]. This incongruity can also be explained by the heterogeneity of the ME/CFS population and reflects the importance of replicating biomarker studies in independent patient cohorts to validate their diagnostic sensitivity or usefulness for subgrouping of patients in future research studies.
To further evaluate these changes in mitochondrial respiration, we performed large-scale discovery proteomics. One of the causes of the changes in mitochondrial respiration can be changes in mitochondrial mass. However, the median of the 279 mitochondrial proteins quantified showed no differences. Nonetheless, when analysing the levels of mitochondrial proteins, we observed a significantly decreased level of PDPR that can lead to lower activation of the PDH. Pyruvate is metabolised from glucose in the cytoplasm and transported into the mitochondria by a specific transporter. Once in the mitochondrial matrix, it is metabolised by the PDH to acetyl-CoA to be incorporated into the citric acid cycle and mitochondrial respiration. In addition, we detected lower levels of VNN1 and PANK2, which are both involved in CoA biosynthesis and regeneration. VNN1 is responsible for the recycling of vitamin B5 [62
], an essential nutrient and precursor of CoA. PANK2 phosphorylates vitamin B5 in the first step of CoA biosynthesis. CoA is used in the mitochondrial oxidation of glucose, fatty acids, and some amino acids to acetyl-CoA, the major input to the citric acid cycle and is, thus, essential for mitochondrial energy production. A potential decrease in PDH, VNN1, and PANK2 activities could lead to lower mitochondrial ATP production and increased lactate production from glycolysis. Accordingly, patients’ cells showed higher levels of SLC16A3, the monocarboxylate transporter 4 (MCT4), which is upregulated in lactate-exporting glycolytic cells [63
], along with upregulation of phosphoenolpyruvate carboxykinase 2 (PCK2), which is involved in the recycling of lactate to support cellular biosynthesis in glycolytic cells [64
]. Dysregulated pantothenic acid and CoA metabolic pathways have already been shown in ME/CFS patients [28
]. Interestingly, reduced activation of the PDH has been reported in other studies, but by upregulation of the PDH kinase instead of downregulation of the PDH phosphatase [29
]. Recent studies in muscle cells from ME/CFS patients point to a PDH dysfunction as a potential cause of mitochondrial dysfunction [67
The higher levels of LAMTOR1 and LAMTOR5 indicate an upregulation of mTORC1 in PBMCs from ME/CFS patients [68
]. Activation of mTORC1 initiates a signalling cascade leading to mitochondrial biogenesis and upregulation of the expression of mitochondrial respiratory chain proteins [69
]. However, we observed no general upregulation of mitochondrial amount or respiratory chain proteins (Figure 3
B,C). We observed, however, lower levels of ATP5F1E, a complex V subunit. Previous studies in immortalised lymphocytes from ME/CFS patients have shown complex V deficiency and hyperactivation of mTORC1 [31
]. In addition, downregulation of SLC25A24 can also indicate disturbed ATP synthesis [71
]. One can hypothesise that the upregulation of mTORC1 is a compensatory mechanism for the inefficient mitochondrial ATP production, hampered by a disturbed mitochondrial respiratory chain structure or electron flow, as reflected by the lower coupling efficiency. Future studies should investigate the integrity of the respiratory chain in ME/CFS.
PBMC changes in mitochondrial proteins were not reflected in plasma metabolites. However, we found two compounds from the phosphatidylcholines (PC) class (PC(36:4) and PC-O(34:4)) with nominally significant lower levels in the patient group (before correction for multiple comparisons). PCs are known to support many of the body’s functions, ranging from fat metabolism, maintaining cell structure, and regulating the neurotransmitter acetylcholine in the brain [72
]. PCs have previously been found significantly downregulated in a larger cohort of ME/CFS patients [57
The heterogeneity of ME/CFS, the lack of definite biomarkers, and the difficult delineation towards other syndromes hamper the research into molecular mechanisms of the disease. Thus, in this study, we aimed to recruit a homogenous group of ME/CFS patients and applied strategies for personalised medicine by performing many tests in a few individuals to shed light on molecular mechanisms of disease in these individuals. Despite the small sample size, our data indicate a dysregulated mitochondrial metabolism centred on PDH and CoA metabolism, which supports findings from other and larger ME/CFS cohorts [29
]. Interestingly, between 35 and 40% of ME/CFS patients experience significant improvement in their health when treated with dichloroacetate, a well-known activator of PDH [74
]. These open-label studies need further follow-up; however, together with the present study and the previous metabolomics and proteomic studies of independent ME/CFS cohorts, they support dysregulated mitochondrial metabolism as an important feature of ME/CFS pathology. To some degree, abnormalities in bioenergetics parameters of blood cells have shown to correlate with the severity of ME/CFS symptoms [31
]. Similarly, coupling efficiency correlated with symptom severity and disease duration in our cohort of ME/CFS patients.