Step 5: Krebs Cycle - Enzyme-Level Failure Modes
1 Normal Function and ATP Accounting
The Krebs cycle generates NADH and FADH2 for the ETC via eight sequential enzymatic steps. Per glucose (two turns of the cycle): direct yield is +2 GTP, +6 NADH, and +2 FADH2; gated yield through the ETC is ~20 ATP (≈67% of total glucose-basis ATP). Each enzymatic step is a potential point of failure; the analysis below focuses on those most affected by the ME/CFS biology (ROS, calcium, cofactor supply).
2 Documented ME/CFS Findings and Failure Modes
The Krebs cycle generates NADH and FADH2 for the ETC via eight sequential enzymatic steps. Naviaux et al. found decreased TCA intermediates (citrate, isocitrate, malate, fumarate) in the ME/CFS metabolome, indicating globally reduced TCA flux . However, the specific enzymatic bottleneck(s) have not been identified.
Aconitase (step 2: citrate → isocitrate). Aconitase contains a [4Fe–4S] iron-sulfur cluster that is exquisitely sensitive to superoxide: a single molecule of superoxide inactivates the cluster and halts the enzyme. With Shankar et al. confirming elevated ROS and decreased MnSOD in ME/CFS lymphocytes , aconitase inactivation is a mechanistically predicted consequence. Yamano et al. (\(n = 67\) CFS, 66 controls) found citrate, isocitrate, and malate significantly reduced in CFS plasma, with the pyruvate/isocitrate ratio elevated, and explicitly proposed aconitase inactivation via chronic oxidative stress as the explanation . Ciregia et al. found aconitase protein (ACO2) upregulated 1.9-fold (\(p = 0.02\)) in a CFS cohort (\(n = 45 + 45\)) via proteomics (Ciregia et al. 2016) — paradoxically, this is consistent with compensatory overexpression of an enzyme that is being oxidatively destroyed: cells produce more aconitase to compensate for functional inactivation by ROS. Despite this convergent indirect evidence, aconitase enzymatic activity has never been directly measured in ME/CFS tissue or cells (G16).
Alpha-ketoglutarate dehydrogenase complex (KGDHC, step 4). KGDHC has identical cofactor requirements to PDC (thiamine, lipoic acid, riboflavin/FAD, NAD+) and shares the same E2/E3 architecture, making it vulnerable to the same failure modes. KGDHC is exquisitely sensitive to oxidative stress: Tretter & Adam-Vizi demonstrated 40–50% inhibition at 0.1–0.5 mM H2O2, irreversible HNE modification of E2 lipoic acid residues, peroxynitrite-mediated E1/E2 tyrosine nitration, and a positive feedback loop where elevated NADH/NAD+ drives E3 to generate additional ROS (Tretter and Adam-Vizi 2005) — every one of these inactivating conditions is independently documented in ME/CFS. A critically under-exploited connection: Fluge et al. documented SIRT4 upregulation in ME/CFS PBMCs but tested only PDC consequences. SIRT4 is a mitochondrial lipoamidase that also acts on the E2 subunit of KGDHC — its upregulation would impair KGDHC in parallel with PDC.
Indirect metabolomic evidence supports a functional KGDHC block. Che et al. (\(n = 106\) ME/CFS, 91 controls) found both \(α\)-ketoglutarate and succinate elevated in ME/CFS plasma (Che et al. 2022) — substrate accumulation upstream and product accumulation downstream of KGDHC, consistent with impaired flux through the step. More strikingly, Germain et al. found that \(α\)-ketoglutarate was the second most significant metabolite distinguishing ME/CFS from controls after exercise: controls showed an \(α\)-KG increase while ME/CFS patients showed a decrease (Germain et al. 2022) — an inverse trajectory consistent with selective KGDHC inhibition under metabolic stress, when NADH/NAD+ rises and oxidative burden increases. Despite this convergent indirect evidence, KGDHC activity has never been directly measured in ME/CFS tissue or cells (G17).
Succinate dehydrogenase (Complex II, step 6). Succinate dehydrogenase is the only enzyme that participates in both the TCA cycle and the ETC. It contains iron-sulfur clusters and is the entry point for FADH2 electrons into the ubiquinone pool. It is therefore simultaneously vulnerable to ROS (via iron-sulfur cluster damage) and to CoQ10 deficiency (via impaired electron acceptance).
Iron-sulfur cluster assembly as a systemic bottleneck. Aconitase, KGDHC, Complex I (which contains eight [Fe–S] clusters), Complex II, and ETF-QO all depend on iron-sulfur clusters assembled by the ISCU machinery. ISCU mutations cause exercise intolerance, myopathy, and lactic acidosis — a syndrome with significant clinical overlap with ME/CFS . ISCU function has not been studied in ME/CFS. MnSOD decrease (Shankar 2025) would accelerate superoxide-mediated destruction of all iron-sulfur proteins simultaneously, creating a multi-enzyme failure pattern that could amplify the effects of any primary Complex I dysfunction.
3 Research Gaps at Step 5
Six gaps at Step 5 focus on individual TCA-enzyme measurement and the iron-sulfur cluster assembly machinery: (G16) aconitase activity (predicted low given ROS-vulnerable [4Fe–4S] cluster); (G17) KGDHC activity (shares PDC cofactors, likely co-impaired); (G18) succinate dehydrogenase (Complex II) activity; (G19) isocitrate dehydrogenase and malate dehydrogenase (both NAD+-dependent, relevant to the NAD+ gap at Step 6); (G20) ISCU iron-sulfur cluster assembly machinery; and (G21) aconitase as a biomarker of mitochondrial superoxide.
| Gap | Essentiality | Worst-case impact | Tractability | Therapeutic leverage | Measurement | Priority |
|---|---|---|---|---|---|---|
| G16: Aconitase activity | High — sole TCA step 2; no isoform redundancy; indirect support from reduced isocitrate (Yamano 2016) and compensatory ACON protein upregulation (Ciregia 2016) | ~67% of gated yield (entire TCA downstream) | Moderate — aconitase activity assay is standardized; citrate/isocitrate ratio is a non-invasive proxy | Long-term — ROS reduction only; aconitase also serves as mitochondrial superoxide biomarker (G21) | Blood draw (citrate/isocitrate ratio) or muscle biopsy | 1 |
| G17: KGDHC activity | High — sole TCA step 4; indirect metabolomic support (elevated α-KG + succinate, Che 2022; inverse exercise α-KG trajectory, Germain 2022); SIRT4 E2 lipoamidase mechanism (Fluge 2016); 40–50% ROS inhibition at physiological H2O2 (Tretter 2005) | ~50% of gated yield downstream | Moderate — NADH production assay at 340 nm is standardized | Immediate — cofactor supplementation (thiamine, LA, riboflavin) | Muscle biopsy (NADH assay, 340 nm) | 1 |
| G18: Succinate dehydrogenase / Complex II | High — sole TCA step 6 and ETC entry point; open for ME/CFS specifically — Long COVID biopsy data show reduced Complex II flux (Colosio 2023 , already cited) but ME/CFS-specific SDH histochemistry has not been formally reported in modern cohorts | ~30% of gated yield | Moderate — SDH histochemistry is routine | Near-term — riboflavin | Muscle biopsy (SDH histochemistry) | 1 |
| G19: IDH + MDH activity | Medium — shared with NAD+ pool availability; IDH2 protein decreased in ME/CFS PBMCs by SWATH-MS proteomics (Sweetman 2020, \(n = 11\) (Sweetman et al. 2020)) — protein abundance only, not enzyme activity | Covered by Step 6 NAD+ gap | Moderate | Immediate — NAD+ precursors (NR/NMN) | Muscle biopsy | 2 |
| G20: ISCU iron-sulfur cluster assembly | High — upstream of aconitase, Complex I, Complex II, ETF-QO | Multi-step cascade; could zero-out ~80% of gated yield | Hard — muscle biopsy with [Fe-S] cluster reconstitution | Long-term | Muscle biopsy (Fe-S reconstitution) | 2 (high impact but hard to measure and no immediate therapy) |
| G21: Aconitase as ROS biomarker | N/A — biomarker repurposing | Dependent on G16 | Moderate | Near-term — could stratify ROS interventions | Blood draw (citrate/isocitrate proxy) or muscle biopsy | 2 |
G16, G17, and G18 are priority 1: each is a sole-route TCA enzyme with standardized activity assays and either immediate (cofactor) or near-term (riboflavin) therapeutic anchors. G20 (ISCU) is the highest-impact single gap but is demoted to priority 2 because the measurement is very difficult and no immediate therapy exists — resolving it would require a multi-year research programme whereas G17 could be closed in a single cohort study.
Step 5 as a whole is individually essential on the glucose-oxidation branch: there is no parallel path from acetyl-CoA to NADH/FADH2 for the ETC. Any severe TCA enzyme failure shuts down aerobic ATP production on carbohydrates. The chain has compensation through ketone-body oxidation (which bypasses glycolysis and PDC but still feeds into the TCA cycle at acetyl-CoA), so Step 5 remains essential even under the ketogenic bypass hypothesis discussed in Summary: Integrated Metabolic Model.
Certainty: 0.30. When superoxide attacks aconitase’s [4Fe-4S] cluster, the enzyme releases labile iron (Fe2+) into the mitochondrial matrix. This iron catalyses the Fenton reaction (Fe2+ + H2O2 → OH\(dot\) + OH- + Fe3+), generating hydroxyl radicals that attack inner mitochondrial membrane polyunsaturated fatty acids. If MnSOD is reduced (as documented by Shankar 2025 ), superoxide is elevated, aconitase continuously loses its cluster iron, and GPX4 (elevated 1.9× as a compensatory response but potentially enzymatically insufficient ) may be unable to clear the resulting lipid peroxides. This creates a mechanistic chain: reduced MnSOD → elevated O2\(dot -\) → aconitase [4Fe-4S] destruction → labile iron release → Fenton chemistry → lipid peroxidation → GPX4 overwhelmed → sub-lethal ferroptotic stress. Aconitase is not merely a TCA cycle casualty of oxidative stress — it may be the iron donor that converts a mitochondrial ROS problem into a membrane damage problem, bridging the energy metabolism deficit (this chapter) to the ferroptosis hypothesis.
A parallel cytosolic mechanism amplifies the effect. Cytosolic aconitase (ACO1) moonlights as Iron Regulatory Protein 1 (IRP1): when its [4Fe-4S] cluster is destroyed by oxidative stress, ACO1 gains RNA-binding activity, upregulating transferrin receptor (increasing iron import) while suppressing ferritin (reducing iron storage) and ferroportin (reducing iron export). The net result is intracellular labile iron accumulation despite normal or low serum ferritin — standard iron panels measure serum levels but not intracellular labile iron, creating a diagnostic blind spot through which Fenton-chemistry-driven damage could proceed undetected.
Testable predictions: (1) Mitochondrial labile iron (calcein-AM quenching) will be elevated in ME/CFS PBMCs and will inversely correlate with ACO2 enzymatic activity. (2) IRP1 RNA-binding activity (EMSA) will be elevated while ACO1 enzymatic activity is reduced. (3) Post-exercise C4a elevation (documented in ME/CFS) will correlate with aconitase inactivation markers.
Limitation: The entire chain is inferred from general biochemistry and indirect ME/CFS data; no step has been directly demonstrated in ME/CFS tissue. Alternative sources of mitochondrial labile iron exist — impaired ferritin synthesis, ISCU iron-sulfur cluster assembly failure (G20), and heme degradation — and any of these could drive Fenton chemistry independently of aconitase inactivation. The Ciregia 2016 compensatory protein upregulation (Ciregia et al. 2016) is consistent with but does not prove this mechanism.
Certainty: 0.35. The ME/CFS field has focused on PDC as the critical TCA entry-point lesion , but KGDHC may be more vulnerable and more consequential for three reasons:
- ROS sensitivity hierarchy: KGDHC is inhibited 40–50% at oxidative-stress-range H2O2 concentrations (0.1–0.5 mM, as used in neuronal oxidative stress models) (Tretter and Adam-Vizi 2005), making it the most oxidant-sensitive TCA enzyme. The KGDHC complex contains multiple redox-sensitive cysteine residues (Starkov et al. 2004), contributing to its ROS vulnerability.
- SIRT4 dual-targeting: Fluge 2016 documented SIRT4 upregulation in ME/CFS PBMCs . SIRT4 has demonstrated lipoamidase activity on PDC-E2 and likely acts on KGDHC-E2 by the same mechanism (both share the dihydrolipoamide acetyltransferase architecture, though KGDHC-E2 activity has not been directly tested). The field has discussed only the PDC consequence. If SIRT4 inhibits both complexes simultaneously, KGDHC impairment may be the larger metabolic consequence because KGDHC sits mid-cycle where the substrate (\(\alpha\)-ketoglutarate) is harder to replenish via anaplerosis, whereas PDC has glucose as an effectively unlimited upstream substrate.
- The “mid-cycle trap”: When PDC is impaired, pyruvate backs up but the cell can shunt to lactate (documented in ME/CFS). When KGDHC is impaired, \(\alpha\)-KG backs up but there is no analogous overflow valve mid-cycle — the cell must either reverse the cycle (thermodynamically unfavorable) or export \(\alpha\)-KG, creating a more severe metabolic trap.
Testable prediction: Paired PDC + KGDHC activity measurement in ME/CFS PBMCs or muscle will show KGDHC reduction exceeding PDC reduction in magnitude, particularly in patients with longer disease duration (where cumulative ROS damage would accumulate on redox-sensitive cysteine residues). Post-exercise KGDHC activity (PBMCs at 4h and 24h post-CPET) should decline in ME/CFS but not controls, with the magnitude of decline correlating with PEM severity.
Competing explanation: If PDC is the sole rate-limiting step (the current consensus framing), KGDHC activity should be normal or elevated (compensatory) rather than reduced. A simultaneous measurement of both enzymes would resolve this directly.
Limitation: KGDHC activity has never been measured in any ME/CFS study. The entire argument is from general enzymology and indirect metabolomics.
Certainty: 0.25. Chen et al. (Chen et al. 2016) demonstrated a duration-dependent switch in KGDHC regulation: acute KGDHC reduction increases glutathionylation by 114% (a protective modification), but chronic reduction depletes glutathionylation by 40% (removing protection). This maps onto ME/CFS natural history:
- Early phase (months to low-single-digit years): Viral trigger reduces KGDHC via ROS and SIRT4. Glutathionylation increases as a protective response. Some patients recover during this window — possibly because glutathionylation protection is still intact.
- Transition phase: Chronic KGDHC reduction exhausts glutathionylation capacity. The protective mechanism fails. ROS damage becomes cumulative rather than buffered.
- Chronic phase: Glutathionylation is depleted. KGDHC is now unprotected against ROS. Further inactivation occurs via a self-amplifying spiral: reduced KGDHC → reduced glutathionylation capacity → more ROS damage → further KGDHC reduction.
Testable prediction: KGDHC activity measured in PBMCs will show an inverse correlation with disease duration, with a non-linear acceleration point at some disease stage (the precise timing is unknown — the phase boundaries are conceptual, not empirically calibrated). The same intervention (cofactor repletion) will have different efficacy depending on disease duration — effective in the acute phase (where glutathionylation protection is intact) but ineffective in the chronic phase (where the protective mechanism is exhausted). Glutathionylation status would serve as a biomarker of treatment responsiveness.
Limitation: The Chen 2016 finding was in neuronal culture, not in ME/CFS tissue. The time-course mapping to years of disease is extrapolated from a cellular phenomenon lasting hours — the correspondence is conceptual, not quantitative.
Certainty: 0.25. Aconitase activity at TCA step 2 (citrate → isocitrate) may be impaired by two simultaneous, mechanistically independent mechanisms:
- Substrate diversion: ACOD1 (immune responsive gene 1) shunts cis-aconitate — the obligate intermediate between citrate and isocitrate — toward itaconate production. Thomas & Peterson (Thomas and Peterson 2025) demonstrated in a zebrafish model that ACOD1 overexpression alone produces a fatigue phenotype (preprint; not peer-reviewed; phylogenetically remote from human ME/CFS). If ACOD1 is upregulated in ME/CFS immune cells (consistent with the metabolic safe mode model in Chapter Integrative Models and Multi-System Pathophysiology), it directly reduces aconitase substrate availability.
- Enzyme inactivation: ROS destroys the aconitase [4Fe-4S] cluster, reducing enzymatic capacity (detailed in Aconitase Inactivation as the Iron Source for Mitochondrial Ferroptotic Stress above).
If both mechanisms produce partial rather than complete inhibition, the combined effect would be more severe than either mechanism alone. Critically, they operate through different pathways: the itaconate shunt is intentional immune metabolic reprogramming (part of the safe mode), while aconitase inactivation is collateral damage from ROS. This means they can be independently modulated — and failing to address both would produce incomplete therapeutic response.
Testable prediction: In ME/CFS PBMCs, itaconate levels will be elevated AND ACO2 activity will be reduced simultaneously. The citrate/isocitrate ratio will be more severely disrupted than predicted by either mechanism alone. ACOD1 suppression (genetic knockdown or validated pharmacological inhibitors, once available) will partially restore isocitrate production, but full restoration will additionally require ROS reduction.
Limitation: ACOD1 expression has not been measured in ME/CFS cells. The itaconate shunt is best characterized in macrophages; its relevance to muscle and neuronal aconitase is unclear.
Certainty: 0.25. Cells in ME/CFS may occupy a low-energy “safe mode” characterized by itaconate-mediated metabolic suppression and reduced TCA flux . Exercise forcibly increases metabolic demands, potentially driving cells out of this conserved state. This forced exit could produce a transient increase in metabolic activity—experienced subjectively as improved energy—followed by a deeper crash as the system re-enters safe mode with depleted reserves. If outcome questionnaires are administered during the “forced activation” window, patients may report improvement, while the subsequent crash would be missed by standard assessment timepoints. Not yet replicated—temporal metabolic dynamics during exercise in ME/CFS have not been profiled.
Testable prediction. Multi-timepoint metabolomics during an exercise trial should show transient mid-intervention aerobic marker increase, then deeper suppression at 2 weeks post-intervention versus baseline.