Metabolic profiling indicates impaired pyruvate dehydrogenase function in myalgic encephalopathy/chronic fatigue syndrome

Abstract

Myalgic encephalopathy/chronic fatigue syndrome (ME/CFS) is a debilitating disease of unknown etiology, with hallmark symptoms including postexertional malaise and poor recovery. Metabolic dysfunction is a plausible contributing factor. We hypothesized that changes in serum amino acids may disclose specific defects in energy metabolism in ME/CFS. Analysis in 200 ME/CFS patients and 102 healthy individuals showed a specific reduction of amino acids that fuel oxidative metabolism via the TCA cycle, mainly in female ME/CFS patients. Serum 3-methylhistidine, a marker of endogenous protein catabolism, was significantly increased in male patients. The amino acid pattern suggested functional impairment of pyruvate dehydrogenase (PDH), supported by increased mRNA expression of the inhibitory PDH kinases 1, 2, and 4; sirtuin 4; and PPARδ in peripheral blood mononuclear cells from both sexes. Myoblasts grown in presence of serum from patients with severe ME/CFS showed metabolic adaptations, including increased mitochondrial respiration and excessive lactate secretion. The amino acid changes could not be explained by symptom severity, disease duration, age, BMI, or physical activity level among patients. These findings are in agreement with the clinical disease presentation of ME/CFS, with inadequate ATP generation by oxidative phosphorylation and excessive lactate generation upon exertion.

Figure 1. Hypothetic mechanism of ME/CFS linked to amino acid catabolism.

According to this model, ME/CFS is caused by immune interference with an unidentified target, potentially a signaling factor, which ultimately causes metabolic dysfunction and induction of secondary rescue mechanisms. We hypothesize that aberrant PDK and SIRT4-mediated inhibition of PDH, and consequent obstruction(s) in central energy metabolism, occurs early during ME/CFS pathogenesis, followed by metabolic adaptations serving to maintain ATP production. The result will be increased consumption of amino acids not depending on PDH to fuel oxidative metabolism via the TCA cycle. Such a mechanism would expectedly change the serum amino acid profile in patients, depending on the different entry stages of the amino acids into the catabolic pathway. Accordingly, for the purpose of serum amino acid profiling, the 20 standard amino acids were assigned into three categories: category I amino acids that may convert to pyruvate (i.e., dependent on PDH; Gly, Ser, Thr, Cys, Ala); category II amino acids that may metabolize to acetyl-CoA and fuel the TCA cycle (i.e., independent of PDH; Lys, Leu, Ile, Phe, Tyr, Trp); and category III amino acids that are anaplerotic and serve to replenish TCA cycle intermediates (i.e., independent of PDH; His, Pro, Met, Val, Glu + Gln = Glx, Asp + Asn = Asx). The asterisk indicates amino acids that were significantly reduced in serum of ME/CFS patients compared with healthy controls in this study (see Table 1 and 2).

Figure 2. Serum amino acids in 153 nonfasting ME/CFS patients and 102 nonfasting healthy controls.

(A) Amino acids converted to pyruvate (category I). (B) Amino acids converted to pyruvate (category I) by sex. (C) Amino acids converted to acetyl-CoA (category II). (D) Amino acids converted to acetyl-CoA (category II) by sex. (E) Anaplerotic amino acids that may replenish TCA intermediates (category III). (F) Anaplerotic amino acids (category III) by sex. (G) Serum levels of 1-MHis (marker of dietary protein intake). (H) Serum levels of the 1-MHis by sex. (I) Serum levels of 3-MHis (marker of endogenous protein catabolism). (J) Serum levels of 3-MHis by sex. P values from unpaired t tests (equal variances not assumed) or from ANOVA when comparing ME/CFS patients versus healthy controls by sex. For 1-MHis, Mann-Whitney test for independent samples or Kruskal-Wallis test was used. Effect sizes estimated from Cohen’s d tests. Error bars indicate mean and SD (median and interquartile range [IQR] for 1-MHis). ME, ME/CFS patients; HC, healthy controls.

Figure 3. Quantitative RT-PCR for mRNA expression levels in peripheral blood mononuclear cells (PBMCs) of ME/CFS patients and healthy controls.

mRNA expression levels in PBMCs from 75 nonfasting ME/CFS patients and 43 nonfasting healthy controls, normalized according to coamplified internal β-actin (ACTB) in duplex qRT-PCR and calculated relative to the mean of healthy controls. (A) Pyruvate dehydrogenase kinase 1 (PDK1) mRNA in PBMCs from ME/CFS patients and healthy controls. Similar analyses are shown for (B) PDK2, (C) PDK3, (D) PDK4, (E) PPARAα (PPARA), (F) PPARδ (PPARD), (G) pyruvate dehydrogenase E1, subunit α (PDHA), (H) acyl-coenzyme A oxidase 1 (ACOX1), (I) mitochondrial pyruvate carrier 1 (MPC1), (J) MPC2, (K) sirtuin 4 (SIRT4), (L) HIF-1α (HIF1A), (M) PDK1 mRNA in PBMCs of ME/CFS patients versus sex, (N) PDK1 mRNA in PBMCs versus ME/CFS severity, (O) PDK1 mRNA in PBMCs versus ME/CFS duration, and (P) PDK1 mRNA in PBMCs versus steps (mean) per 24 hours in ME/CFS patients. P values were from Mann-Whitney U test for independent samples (A–N and P) and from Kruskal-Wallis test (O). Error bars indicate median with 95% CI. All samples in a qRT-PCR assay were run in triplicate on the same plate. Of 75 samples from patients and 43 from healthy controls, two samples for PDK1 and four samples for PDK4 were excluded due to unsuccessful amplification. For SIRT4, due to a low expression level, 11 samples from ME/CFS patients and 5 samples from healthy controls were excluded due to high SD (≥30%) among triplicates (see the Methods). Sensewear bracelet data for physical activity for 7 consecutive days were available from 62 of the 75 patients.

Figure 4. Correlation analyses between mRNA expression levels in peripheral blood mononuclear cells (PBMCs) of ME/CFS patients.

mRNA expression levels in PBMCs from nonfasting ME/CFS patients, normalized according to coamplified internal β-actin (ACTB) in duplex qRT-PCR and calculated relative to the mean of healthy controls. (A) Correlation of pyruvate dehydrogenase kinase 1 (PDK1) and PPARδ (PPARD) mRNA levels in PBMCs of ME/CFS patients. (B) Correlation of PDK1 and PPARα (PPARA). (C) Correlation of PDK1 and PDK4. (D) Correlation of PDK1 and sirtuin 4 (SIRT4). (E) Correlation of PDK4 and SIRT4. (F) Correlation of PPARD and SIRT4. (G) Correlation of PPARA and SIRT4. (H) Correlation of PPARD and PPARA. P values from Spearman correlation analyses. Of 75 samples from patients, two samples for PDK1 and four samples for PDK4 were excluded due to unsuccessful amplification. For SIRT4, due to a low expression level 11 samples from ME/CFS patients were excluded, due to high SD (≥30%) among triplicates (see the Methods). In three samples with SIRT4 mRNA data, PDK and PPAR mRNAs were not analyzed, leaving 61 samples for SIRT4 correlation analyses.

Figure 5. Effects of ME/CFS patient and healthy control serum on muscle cell metabolism.

Rates of oxygen consumption and lactate production were measured simultaneously in cultures of human muscle cells (HSMM) after exposure (6 days) to serum from healthy individuals (n = 12) or ME/CFS patients (n = 12). Glucose (GLC), oligomycin (OLIGO), carbonyl cyanide 3-chlorophenylhydrazone (CCCP), and rotenone/antimycin A (ROT/AMA) were administered sequentially during the analysis to assess specific properties of cellular energy metabolism. The consequent energetic states of the cells were classified as follows: I Resting (AAs), II Resting (AAs+GLC), III Anaerobic strain, and IV Aerobic strain, as described in the Methods. (A) Recordings of oxygen consumption rate are shown for muscle cells preexposed to healthy control (black) and ME/CFS patient (red) serum. The substance additions and the resultant energetic conditions of the cells are indicated (conditions I–IV). (B) Statistical analysis of the data in A. (C) Recordings of the lactate production rate from the same experiment as A and B. (D) Statistical analysis of the data in C. (E) Specific descriptors of mitochondrial respiration were calculated as indicated, based on the data in A. (F) Specific descriptors of inducible lactate production were calculated as indicated, based on the data in C. The analysis was performed with 5 replicate wells for each serum sample and is representative of 3 separate experiments. Statistical comparisons between healthy controls and ME/CFS samples were performed by Mann-Whitney U test for independent samples.