Step 3: Pyruvate Dehydrogenase Complex (PDC)
1 Normal Function and ATP Accounting
Pyruvate produced by glycolysis crosses both mitochondrial membranes via the mitochondrial pyruvate carrier (MPC) and undergoes oxidative decarboxylation by PDC to yield acetyl-CoA, CO2, and NADH. PDC is the irreversible commitment step from glycolysis to the TCA cycle. Direct ATP yield is zero (the step produces +2 NADH per glucose, =+5 ATP equivalent gated via the ETC). Gated yield is ~25 ATP per glucose (the aerobic portion); the glycolytic 2 ATP remain available even with PDC blocked. PDC activity is tightly regulated by phosphorylation: PDK1–4 inactivate PDC by phosphorylating the E1 subunit; PDP1–2 reactivate it.
2 Documented ME/CFS Findings and Failure Modes
Fluge et al. found significantly increased mRNA expression of PDK1, PDK2, and PDK4 (but not PDK3) and of the mitochondrial lipoamidase SIRT4 in peripheral blood mononuclear cells from 200 ME/CFS patients versus 102 healthy controls . Patient serum applied to healthy muscle cells induced increased mitochondrial oxygen consumption while causing excess lactate excretion — a pattern consistent with functional impairment of PDC and compensatory anaerobic metabolism. SIRT4 deacetylates and inhibits the E3 subunit of PDC (dihydrolipoamide dehydrogenase), providing a second inhibitory mechanism parallel to PDK-mediated phosphorylation. Study: (\(n=200\) ME/CFS \(+\) 102 controls; transcriptomic and cell-based assay; certainty: 0.50 — findings not yet independently replicated in muscle tissue; partially replicated by downstream metabolomic patterns).
The PDK upregulation finding (Fluge 2016) has not been directly replicated in an independent cohort using muscle biopsy; the evidence base is transcriptomic (PBMCs) and cell-based. The closest tissue-level evidence comes from Tomas et al. (\(n = 9\) CFS, 11 controls), who cultured primary skeletal muscle cells from vastus lateralis biopsies and found that CFS muscle cells cannot oxidize glucose normally while galactose and fatty acid oxidation were preserved — localising the defect upstream of the TCA cycle at the pyruvate→acetyl-CoA link step (Tomas et al. 2020). The authors explicitly name PDK as a candidate, but direct PDC activity was not measured. Downstream metabolomic evidence (reduced TCA intermediates, elevated lactate) from multiple independent groups is consistent with PDC impairment but does not isolate this step. In long COVID, muscle biopsies show depleted TCA cycle metabolites, glycolytic fibre shift, and worsened mitochondrial function after PEM provocation (Appelman et al. 2024); 31P MRS demonstrates severely prolonged phosphocreatine recovery (\(τ_\text{PCr}\) = 92.5 vs 51.9 s, \(p ≤ 0.001\)) indicating in vivo OXPHOS impairment (Finnigan et al. 2024). ME/CFS-specific muscle biopsies show preferential subsarcolemmal mitochondrial damage by electron microscopy, distinct from the post-COVID pattern (Bizjak et al. 2024).
PDC requires four cofactors, each representing an independent failure mode: thiamine (vitamin B1; cofactor for the E1/TPP subunit), lipoic acid (E2 subunit), riboflavin/FAD (E3 subunit, dihydrolipoamide dehydrogenase), and NAD+ (final electron acceptor from E3). Functional depletion of any one cofactor can impair PDC independently of PDK regulation.
Certainty: 0.35. High-dose thiamine and riboflavin are established treatments for thiamine-responsive PDH-deficiency and riboflavin-responsive Complex II deficiency respectively . Naviaux et al. identified riboflavin pathway abnormalities among the 20 dysregulated metabolic pathways in ME/CFS . Yet neither high-dose thiamine (600–1800 mg/day) nor riboflavin as standalone interventions have been trialled in ME/CFS . Combined cofactor replacement targeting all four PDC cofactors simultaneously represents an uninvestigated therapeutic approach.
Testable prediction: Erythrocyte transketolase activation ratio (functional thiamine status) and riboflavin loading tests will be abnormal in a subgroup of ME/CFS patients, and supplementation will normalize PDC-flux indicators (lactate/pyruvate ratio, acylcarnitine profiles).
Limitation: Cofactor replacement addresses only the substrate availability arm of PDC inhibition, not the PDK overactivation arm; combined PDK inhibition plus cofactor repletion may be required.
Two small open-label studies examined sodium dichloroacetate (DCA), a direct PDK inhibitor that restores PDC activity by blocking PDK-mediated phosphorylation of the E1 subunit (Comhaire 2018a) (Comhaire 2018b). Approximately 37–45% of treatment-resistant ME/CFS patients responded favorably, and six pre-treatment characteristics predicted response with AUC = 0.92, suggesting that only a subgroup has PDC inhibition as a dominant mechanism. DCA carries neurotoxicity risk (peripheral neuropathy) with chronic use and is not approved for ME/CFS; these reports are hypothesis-generating only.
The DCA reports (Comhaire 2018, \(n=22\) and \(n=35\)) are open-label single-investigator studies published in Medical Hypotheses (low editorial threshold). DCA carries risk of peripheral neuropathy with prolonged use. This evidence is insufficient to support clinical use; it provides mechanistic support for the PDK/PDC hypothesis only.
A mechanistic bridge between ME/CFS immune dysfunction and muscle-level PDK4 overactivation is provided by Park & Jeoung, who demonstrated in C2C12 skeletal muscle myoblasts that LPS-induced inflammation upregulates PDK4 (but not PDK1/2/3) via JNK phosphorylation, with JNK inhibition preventing both PDK4 upregulation and lactate excess (Park and Jeoung 2016). If chronic low-grade inflammation in ME/CFS (elevated TNF-α, IL-6, IL-1β) activates the JNK pathway in muscle, this would drive PDK4-mediated PDC inhibition in the metabolically relevant tissue without requiring the PBMC-to-muscle generalization that limits the Fluge 2016 finding.
The PDK upregulation finding is conventionally framed as a pathological lesion driving energy deficit. An alternative interpretation — that PDK overactivation may be a protective cellular response to compromised ROS handling rather than a primary cause — is developed in the Cross-Cutting section (Section PDK Upregulation May Be Protective, Not Pathological below). These framings are reconcilable by MnSOD status: patients with intact antioxidant capacity and elevated PDK represent pathological inhibition (PDK inhibition would help); patients with low MnSOD/GPX4 represent protective restriction (PDK inhibition without antioxidant support may worsen). A stratified DCA trial with baseline antioxidant biomarkers would discriminate between subgroups — but only if the biomarkers used reflect muscle and brain MnSOD/GPX4 status, not only lymphocyte levels. The Shankar 2025 finding is currently lymphocyte-based; a validation study establishing lymphocyte-to-tissue correspondence would be a prerequisite for interpretable stratification.
Certainty: 0.30. The Godlewska/Syan 2025 7T MRS finding creates a striking puzzle: elevated brain lactate (pgACC 1.52 vs 1.22 mM) alongside normal resting muscle metabolites . If PDK upregulation were systemic, both tissues should show resting lactate elevation. Three explanations resolve the dissociation:
- Tissue-selective PDK isoform expression: PDK4 dominates in skeletal muscle while PDK2 and PDK3 are more highly expressed in brain. The Fluge 2016 PBMC finding (PDK1/2/4 all upregulated ) may map differently onto each tissue. Brain PDK2/3 upregulation could produce resting lactate elevation; muscle PDK4 upregulation may manifest only under exertion when pyruvate flux increases — explaining normal resting muscle metabolites.
- Activity-gated threshold: PDK4 is acutely upregulated by fasting, exercise, and fatty acid availability. In ME/CFS, the baseline may be near-normal but the PDK4 response to even mild exertion may be exaggerated and recovery delayed — a metabolic trap that produces the characteristic PEM pattern.
- Different metabolic reserve margins: Brain has essentially zero glycogen reserve and relies on continuous glucose oxidation; even modest PDC inhibition immediately elevates lactate. Muscle has substantial glycogen and creatine phosphate buffers that absorb mild PDC inhibition at rest, decompensating only under load.
Testable predictions: (1) Exercise-provoked muscle 31P-MRS or venous lactate will show dramatically exaggerated PDK-signature (high lactate, slow PCr recovery) compared to the mild resting abnormality; a paired brain MRS + exercise muscle MRS study would demonstrate the divergence. Consistent with this prediction’s premise, dynamic 31P-MRS under an exercise/recovery protocol has already shown reduced muscle oxidative capacity (prolonged recovery time-constant and reduced maximal flux) in COVID-19 and multiple sclerosis patients versus controls (Naëgel et al. 2023), while resting muscle 31P-MRS PCr/ATP does not differ between ME/CFS, long COVID, and controls (Godlewska et al. 2025) — indicating the deficit is load-gated/extrinsic rather than resting. (2) A muscle biopsy finding only PDK4 elevated (not PDK1/2) would implicate inflammation-driven JNK→PDK4 mechanisms (Park and Jeoung 2016) rather than cell-autonomous reprogramming.
Limitation: The brain-vs-muscle comparison is a cross-study inference from different populations and modalities. DCA crosses the BBB but carries neurotoxicity risk — paradoxically, the brain compartment showing more resting dysfunction is the one most vulnerable to DCA side effects. No tissue-selective PDK inhibitor delivery method currently exists; pharmacological PDK inhibition cannot be targeted to muscle while sparing brain.
3 Research Gaps at Step 3
PDC sits at a sole-route bottleneck between glycolysis and the TCA cycle, so every Step 3 gap is high essentiality by construction. The open gaps are: (G7) muscle-biopsy replication of the Fluge 2016 PDK1/2/4 upregulation finding ; (G8) direct PDC activity assay in ME/CFS muscle (not just transcriptomic inference); (G9) simultaneous measurement of all four PDC cofactors (thiamine, lipoic acid, riboflavin/FAD, NAD+) within the same cohort; and (G10) MnSOD-stratified DCA trial design to test the PDK-as-protective reinterpretation (PDK Upregulation May Be Protective, Not Pathological).
G9 is a confirmed open gap despite existing partial evidence. The closest published approach is Heap et al. 1999 (\(n = 12\)), who simultaneously measured erythrocyte transketolase (thiamine), glutathione reductase (riboflavin), and aspartate aminotransferase (pyridoxine) — finding all three reduced in CFS — but omitted CoQ10, NAD+, magnesium, and lipoic acid (Heap, Peters, and Wessely 1999). Myhill et al. 2012 co-measured red cell NAD, plasma CoQ10, and L-carnitine in a clinical audit (\(n ≈ 34\) subset) and found only 11% of patients had all three simultaneously within reference ranges — but the study lacked a control group for cofactor data and the ATP Profile methodology was not reproduced by Tomas et al. 2019. Castro-Marrero et al. 2015 measured both CoQ10 and NAD+/NADH in the same RCT (\(n = 73\)) but only these two cofactors . A systematic review of 45 studies on vitamins and minerals in CFS found little evidence for individual deficiencies but critically excluded CoQ10 and NAD+ from the primary analysis despite their being the cofactors with the strongest mechanistic rationale (Joustra et al. 2017). A 2025 supplementation review (\(n = 809\) across 14 studies) identified CoQ10+NADH as the combination with the strongest fatigue-reduction signal but noted that no study measured all relevant cofactors at baseline (Dorczok et al. 2025). Lipoic acid and iron-sulfur cluster status have never been measured in any ME/CFS study.
| Gap | Essentiality | Worst-case impact | Tractability | Therapeutic leverage | Measurement | Priority |
|---|---|---|---|---|---|---|
| G7: Muscle-biopsy replication of Fluge 2016 PDK upregulation | High — confirms or refutes the anchor finding; indirect tissue support from Tomas 2020 (glucose oxidation defect at PDC link step) and Bizjak 2024 (subsarcolemmal mito damage) | Determines whether the current PDC framing is correct; up to ~83% of gated yield | Hard — muscle biopsy qPCR + PDK Western; inflammation→PDK4 mechanism via JNK available (Park 2016) | Near-term — DCA is available (Comhaire 2018) | Muscle biopsy (qPCR + Western) | 1 |
| G8: Direct PDC activity assay in ME/CFS muscle | High | Same as G7 | Hard — radioactive pyruvate decarboxylation assay requires specialist lab | Near-term | Muscle biopsy (specialist assay) | 1 |
| G9: Combined cofactor measurement (thiamine + LA + riboflavin + NAD+) | High — any single cofactor failure inactivates PDC; partial data exist (Heap 1999: 3 B vitamins; Myhill 2012: NAD+CoQ10+carnitine; Castro-Marrero 2015: CoQ10+NAD) but no study covers all four PDC cofactors | Up to 83% if any cofactor is limiting | Easy (individual assays all standardized) to moderate (integration); lipoic acid and Fe-S cluster status have never been measured in ME/CFS | Immediate — all four cofactors are OTC supplements; CoQ10+NADH shows strongest supplementation signal (Dorczok 2025 systematic review) | Blood draw (standardized assays) | 1 |
| G10: MnSOD-stratified DCA trial | Dependent on PDK Upregulation May Be Protective, Not Pathological | Intervention test, not a mechanism gap | Moderate — requires stratified recruitment | Immediate — DCA is orphan-drug available | Blood draw (stratified RCT) | 2 |
All four gaps at Step 3 are high-essentiality because PDC is a sole-route bottleneck. G7–G9 are priority 1; G10 is priority 2 as an interventional follow-up contingent on the protective/pathological reinterpretation being tested. PDC is one of the two individually essential steps in the chain (the other is Step 7 ANT export) — failure cannot be compensated by any parallel route.