Step 1: Substrate Delivery - Glucose, Fatty Acid, and Oxygen Uptake
1 Normal Function
Before mitochondria can produce ATP, substrate must reach the cell. Two fuel-delivery chains run in parallel: glucose via GLUT transporters and fatty acids via albumin-bound transport followed by CPT-mediated membrane uptake. Oxygen delivery — the third substrate — depends on vascular perfusion, red blood cell deformability, and capillary diffusion distance.
2 ATP Accounting
Substrate delivery is a binary gate: 0 ATP produced directly, but ~30 ATP per glucose are gated on it. Without substrate and oxygen, no downstream step operates. Partial delivery failure — which is the clinically relevant case — produces proportionally reduced capacity across every subsequent step, not a graceful degradation at any single one.
3 Documented ME/CFS Findings and Failure Modes
Before mitochondria can produce ATP, substrate must reach the cell and enter the cytoplasm. Two delivery chains run in parallel: glucose via GLUT transporters and fatty acids via albumin-bound transport followed by membrane uptake.
Glucose delivery. Cellular glucose uptake depends on GLUT transporter expression and, in insulin-sensitive tissues (skeletal muscle, adipose), on insulin-stimulated GLUT4 trafficking from intracellular vesicles to the plasma membrane. Mandarano et al. demonstrated that ME/CFS CD4+ and CD8+ T cells show increased GLUT1 expression in PBMCs — suggesting a compensatory attempt to increase glucose uptake — but that hexokinase I (which phosphorylates glucose to trap it inside the cell) is simultaneously downregulated . The result is reduced basal glycolysis despite elevated transporter expression: substrate can enter but cannot be efficiently phosphorylated and committed to the glycolytic pathway.
Partially characterized: GLUT4 trafficking in ME/CFS skeletal muscle has not been directly examined, though Brown et al. showed that glucose uptake (a functional GLUT4 output) is impaired in cultured ME/CFS muscle cells, with the defect localized proximal to AMPK (Brown et al. 2018). The malonyl-CoA/CPT1 switch — the molecular gate that controls whether cells preferentially oxidize glucose or fatty acids — has not been characterized in ME/CFS, though Reuter and Evans inferred reduced CPT-I activity from long-chain acylcarnitine deficiency (\(n = 44\) CFS vs 49 controls, 30–40% lower C18:1/C18:2, \(p < 0.0001\)) . AMPK activators have been tested in vitro: metformin and compound 991 rescued both AMPK activation and glucose uptake in ME/CFS muscle cells (Brown et al. 2018). In Long COVID, the REVIVE-TOGETHER adaptive RCT (\(n = 399\)) halted its metformin 750 mg BID arm for futility at 60 days while the fluvoxamine arm showed significant fatigue reduction , and the earlier COVID-OUT trial showed early metformin prevented Long COVID by ~41% but did not treat established fatigue (Bramante et al. 2023). No AMPK-activator trial has been conducted in primary ME/CFS. Fineberg et al. (2025) have proposed metformin as a weight-neutral therapeutic candidate for ME/CFS — in contrast to GLP-1 RAs that cause substantial lean mass loss, metformin does not promote muscle catabolism and may preserve insulin sensitivity during periods of enforced inactivity (Fineberg, Moreau, and Schneider-Futschik 2025).
Fatty acid delivery. Acylcarnitines and acylcholines, which reflect mitochondrial fatty acid utilization, are consistently reduced in ME/CFS across two independent cohorts studied by Germain et al. , indicating impaired flux of long-chain fatty acids into beta-oxidation. Reduced serum carnitine is a documented finding , further limiting fatty acid transport across the inner mitochondrial membrane via the CPT1/CPT2 carnitine shuttle.
4 Microcirculation and Oxygen Delivery
The energy crisis in ME/CFS begins upstream of mitochondrial ATP synthesis. Multiple independent research groups have documented vascular and microcirculatory dysfunction that directly impairs substrate and oxygen availability, creating a pre-mitochondrial bottleneck.
4.1 Endothelial Dysfunction
Scherbakov et al. found endothelial dysfunction (ED) in 51% of ME/CFS patients using peripheral arterial tonometry (EndoPAT). This early indicator of vascular disease correlates with systemic atherosclerosis risk and predicts cardiovascular events. Haffke et al. (Haffke et al. 2023) measured flow-mediated dilation (FMD), the gold-standard non-invasive endothelial function test, and found severely impaired responses in ME/CFS patients (5.1% vs 8.2% in healthy controls, \(p < 0.0001\)). This indicates reduced nitric oxide (NO) bioavailability and impaired vasodilatory capacity.
Blauensteiner et al. (Blauensteiner et al. 2021) identified five endothelial-regulating microRNAs (miR-21, miR-34a, miR-92a, miR-126, miR-200c) with altered expression in up to 60% of ME/CFS patients. All five microRNAs are key regulators of the NO pathway and endothelial function. This suggests dysfunction at the molecular level controlling vascular biology.
4.2 Red Blood Cell Deformability and Rheology
Red blood cell function is critical for oxygen delivery: RBCs must deform to traverse narrow capillaries and must maintain membrane fluidity to exchange gases efficiently. Saha et al. studied ME/CFS patient RBCs using ektacytometry and rheological analysis, finding:
- Higher capillary entry time (approximately 12% longer)
- Lower transit velocity through capillaries (approximately 17% reduced)
- Lower elongation index on deformation (approximately 14% reduced membrane deformability)
- Lower membrane fluidity (approximately 30% reduced)
- Higher reactive oxygen species (ROS) production (approximately 30% elevated)
Critically, recovering ME/CFS patients showed normalization of these parameters, suggesting that RBC dysfunction is a reversible consequence of the disease rather than a primary genetic defect. The combination of reduced deformability and elevated ROS creates a cascade: stiff RBCs move slower through capillaries, allowing more time for oxidative damage, further reducing deformability and oxygen delivery.
4.3 Capillary Structural Pathology
Wüst, Charlton, Slaghekke et al. (Charlton et al. 2025) conducted the largest muscle biopsy study in ME/CFS to date (n=26 ME/CFS pre-2020 dx, n=24 Long COVID, n=30 healthy controls) with vastus lateralis biopsies analysed by immunohistology and electron microscopy, plus a unique comparator group: 10 healthy volunteers after strict 60-day bed rest. This design tests the deconditioning hypothesis head-to-head against the microvascular pathology hypothesis.
Key structural findings:
- Capillary basement membrane (BM) thickening (\(p < 0.0001\)) — the most striking finding, with a near-complete separation between patients and controls: maximum BM thickness in healthy controls was 62.7%, while minimum BM thickness in patients was 63.2%. This is a structural anatomical obstruction, not a functional deficit.
- Decreased capillary tortuosity (\(p < 0.0001\)) — reduced capillary winding reduces the total surface area available for gas exchange
- Decreased capillary contact length (\(p < 0.0001\)) — less capillary surface in direct contact with muscle fibers, increasing the oxygen diffusion distance
- Lower capillarization (\(p < 0.0005\)) — but only in ME/CFS, not Long COVID (discordant with Agergaard 2023 (Agergaard et al. 2023) who found reduced capillarization in Long COVID; possibly reflecting different biopsy sites — vastus lateralis vs biceps brachii)
- EM ultrastructural abnormalities — basement membrane thickening confirmed at the ultrastructural level, plus microvacuolization, endothelial hypertrophy, and signs of endothelial degeneration
Deconditioning rebuttal. The bed rest comparator is critically informative. Bed rest produced the expected detraining phenotype: muscle atrophy plus reduced OXPHOS capacity. Patients did not show atrophy. Bed rest altered cardiovascular and respiratory responses at both submaximal and maximal exercise; patients showed alterations only at submaximal intensity — a pattern inconsistent with simple deconditioning. Physical inactivity alone cannot explain the lower exercise capacity (Charlton et al. 2025).
The capillary BM thickening finding has been independently replicated in three countries: Amsterdam (Charlton/Slaghekke (Charlton et al. 2025) — vastus lateralis), Berlin (Aschman et al. 2023 (Aschman et al. 2023) — vastus lateralis, n=11 PCS), and Aarhus (Agergaard et al. 2023 (Agergaard et al. 2023) — biceps brachii + anterior tibialis, n=18 LC). Three independent cohorts, two muscle groups, consistent core finding.
These are not functional abnormalities but anatomical obstructions — capillaries structurally remodelled such that oxygen diffusion is mechanically limited independent of mitochondrial capacity. This represents a pre-mitochondrial mechanical bottleneck.
The electron microscopy finding of endothelial hypertrophy, microvacuolization, and endothelial degeneration (Charlton et al. 2025) adds a further dimension to the structural pathology. Endothelial cell swelling narrows the capillary lumen from the inside. In the narrowest capillaries (~5–7 µm, already a tight fit for an ~7 µm RBC), even modest endothelial hypertrophy (~1 µm swelling) reduces luminal diameter to 3–5 µm — functionally excluding RBC transit. Combined with documented RBC stiffness in ME/CFS (approximately 14% reduced deformability; see Red Blood Cell Deformability above), narrowed capillaries become impassable for already-stiff RBCs. The result is heterogeneous perfusion: some capillaries are perfused while anatomically adjacent capillaries are functionally excluded. Tissue pO₂ becomes a patchwork — well-perfused regions are normoxic, excluded regions are severely hypoxic — and whole-muscle NIRS averages across both, potentially masking the true severity of micro-regional hypoxia. This mechanism is distinct from BM thickening (diffuse diffusion resistance) and from arteriolar vasoconstriction (reduced total inflow). Here, total flow may be normal but its spatial distribution is pathologically heterogeneous due to stochastic luminal occlusion by endothelial swelling.
Certainty: 0.40. The EM finding of endothelial hypertrophy is from a single study (Charlton/Slaghekke 2025, preprint). The link to heterogeneous perfusion is mechanistically plausible but not directly measured — no study has mapped capillary-level perfusion heterogeneity in ME/CFS muscle. (Evidence source: vastus lateralis biopsy EM — Inference target: capillary-level perfusion distribution. Link is indirect — no perfusion mapping at single-capillary resolution in ME/CFS exists.)
(Origin: brainstorm.)
Falsifiable predictions: + EM-measured capillary luminal diameter distribution should have wider variance in ME/CFS vs controls — indicating heterogeneous narrowing — and the fraction of capillaries with luminal diameter \(< 4\) µm should correlate with NIRS spatial heterogeneity (variance in tissue O₂ saturation across muscle regions) + Pentoxifylline (RBC deformability enhancer) should improve NIRS-measured O₂ extraction more in patients with higher RBC stiffness — but only if luminal narrowing is not so severe that even flexible RBCs cannot transit + Falsified if: luminal diameter distribution is identical between ME/CFS and controls — endothelial hypertrophy is EM artifact or functionally inconsequential, in which case perfusion heterogeneity is not the rate-limiting mechanism
Consequence: The swelling of cells lining capillaries in muscle may be creating a patchwork of oxygen delivery — some areas receive oxygen normally, adjacent areas are starved. This hidden patchwork could explain why patients feel muscle burning and weakness even when whole-muscle oxygen measurements look near-normal. An oxygen sensor on the skin averages across the patchwork and misses the silent suffocation of individual muscle regions.
The capillary BM thickening and endothelial pathology documented to date are limited to limb skeletal muscle (vastus lateralis, biceps brachii, anterior tibialis). No data exist for diaphragm, cardiac muscle, smooth muscle (GI tract, bladder), or skin capillaries. If BM thickening is restricted to limb skeletal muscle, the mechanism is limited to exertional symptoms and cannot explain CNS, GI, cardiac, or cutaneous symptoms. If BM thickening is present in all capillary beds, muscle biopsy findings reveal a systemic microvascular disease of which skeletal muscle is merely the most accessible manifestation. The distinction determines whether BM thickening is a consequence of muscle-specific factors (disuse, repetitive micro-injury, metabolite accumulation) or a systemic endothelial/BM disorder: same-patient multi-site biopsy or skin capillary BM measurement is needed to resolve this. Consequence: We don’t yet know whether this is a muscle problem or a whole-body capillary disease — the answer determines whether treatments targeting capillary structure need to be systemic or can be muscle-localized. Patients with primarily muscle symptoms and no systemic involvement may need a different treatment strategy than patients whose every organ’s capillaries are affected.
4.4 Brain Microvascular Insufficiency
Godlewska et al. used 7-Tesla magnetic resonance spectroscopy (7T MRS) in 24 ME/CFS patients to measure brain lactate levels at high spatial resolution. They found elevated brain lactate in the pregenual anterior cingulate cortex (pACC) and dorsal anterior cingulate cortex (d ACC)—regions critical for mood, pain processing, and autonomic regulation. Elevated brain lactate reflects anaerobic metabolism in brain tissue, consistent with localized microvascular insufficiency in these regions.
4.5 Integration: Substrate Delivery Failure and Vascular-Primary Causation
Together, these findings identify a pre-mitochondrial energy bottleneck: endothelial dysfunction reduces perfusion pressure, RBC deformability limits capillary transit, structural capillary pathology blocks oxygen diffusion, and localized brain insufficiency impairs central regulatory systems. Cells cannot generate energy efficiently when they cannot access substrates, regardless of mitochondrial capacity.
A critical question for both mechanistic understanding and therapeutic targeting is whether vascular oxygen delivery failure causes mitochondrial dysfunction or results from it. The 2022–2025 literature increasingly supports vascular pathology as primary, with mitochondrial damage as the downstream consequence of chronic ischemia/reperfusion.
Scheibenbogen and Wirth (2024) proposed an explicit temporal model: early post-COVID syndrome is dominated by microvascular disturbance (microclots, inflammatory capillary changes) with mitochondrial damage appearing later and selectively in patients who develop ME/CFS. The proposed causal chain is: capillary hypoperfusion → ischemia/reperfusion → anaerobic metabolism → proton accumulation → NHE1 sodium influx → NCX1 reversal → mitochondrial Ca2+ overload → mitochondrial structural damage. Once mitochondria are damaged, the loop becomes self-sustaining via ROS-mediated worsening of endothelial function and reduced ATP availability for ion pump maintenance (Section WASF3–DRP1 Convergence: Two Routes to the Same Mitochondrial Failure).
Three independent lines of interventional and structural evidence support vascular primacy:
- Structural precedence: Electron microscopy shows capillary basement membrane thickening (collagen IV deposition) in ME/CFS skeletal muscle — a structural change that precedes and imposes mechanical limits independent of mitochondrial state
- Cell-autonomous RBC oxygen sensing failure: ME/CFS RBCs fail to sense and respond to tissue hypoxia by accelerating capillary transit (Guo et al. 2025), a defect that is independent of mitochondrial function in the target tissue
- Interventional evidence: Pharmacological correction of preload failure (pyridostigmine) improves peak VO2 by +0.9~mL/kg/min (\(p = 0.002\)) in ME/CFS (Joseph et al. 2022) — a vascular-targeted intervention producing functional improvement without directly targeting mitochondria. Joseph et al. used invasive CPET (arterial and venous catheters) — the gold standard — demonstrating that the exercise limitation is peripheral O2 extraction (0.69 vs 0.77 in controls, \(p < 0.001\)), not cardiac output or pulmonary gas exchange
- NIRS confirmation: Near-infrared spectroscopy during maximal cycling exercise in the Charlton/Slaghekke cohort (Charlton et al. 2025) confirmed lower tissue O2 uptake (\(p = 0.001\)) and lower vasodilatory capacity (\(p = 0.011\)) in both ME/CFS and Long COVID patients, despite no differences in maximal heart rate or ventilatory equivalents — direct functional evidence that capillary-level structural pathology translates to impaired oxygen extraction during exercise
- Deconditioning excluded: The 60-day bed rest comparator (above) definitively excluded physical inactivity as the sole or primary explanation for reduced VO2max — the patient phenotype (capillary BM thickening + glycolytic shift + no atrophy) is structurally distinct from the detraining phenotype (atrophy + reduced OXPHOS), confirming the exercise limitation has a disease-specific structural basis
These data do not exclude mitochondrial primary dysfunction in all patients, but they challenge the assumption that vascular abnormalities are secondary epiphenomena. For therapeutic purposes, the distinction matters: a patient whose primary lesion is capillary ischemia requires a different intervention strategy than one whose primary lesion is mitochondrial supercomplex disruption. The acquired ischemic mitochondrial myopathy (AIMM) framework proposed by Scheibenbogen and Wirth frames ME/CFS as arising from a primary vascular event producing acquired — and therefore potentially reversible — mitochondrial damage, with important implications for ion homeostasis restoration therapies (Section WASF3–DRP1 Convergence: Two Routes to the Same Mitochondrial Failure).
5 Research Gaps at Step 1
Three substrate-delivery mechanisms remain uncharacterized in ME/CFS despite being standard targets in diabetes, mitochondrial myopathy, and obesity research: GLUT4 trafficking in skeletal muscle, the malonyl-CoA/CPT1 fuel-switch, and AMPK-activator trials (metformin, berberine) for GLUT4 restoration. The three are enumerated below, with an explicit assessment of essentiality, worst-case impact if the mechanism fails, and a priority tier.
5.1 Prioritization Framework
To decide which gaps to study first, the following four criteria are applied to each gap:
- Essentiality: Is the mechanism a sole pathway (no redundancy) or one of several parallel routes? A sole pathway’s failure is catastrophic; a redundant pathway’s failure is compensable.
- Worst-case impact: If the mechanism is fully failed, what fraction of the ~30 ATP/glucose total is lost downstream? Expressed as a percentage of Step 1’s gated yield (100% of ~30 ATP).
- Tractability: How hard is the measurement to perform in patients? Scored as easy (plasma/serum, existing assay), moderate (PBMC or imaging, specialist lab), hard (muscle biopsy, research-only assay), or very hard (tissue that requires invasive sampling or post-mortem).
- Therapeutic leverage: Is a correction already available, or would development be required? Immediate (FDA-approved drug exists for analogous indication), near-term (off-label repurposing with reasonable prior), long-term (new drug development required).
A gap is priority 1 (highest) if its essentiality is high, worst-case impact is ≥50% of step-gated yield, and tractability is easy or moderate. A gap is priority 2 if any one of those is downgraded. A gap is priority 3 if essentiality is low, worst-case impact is modest, or tractability is hard. The framework is deliberately biased toward actionable measurements — a gap that would take a decade to measure and would not change treatment is explicitly de-prioritised against one that could be resolved in months with an existing assay.
Exception for sole-route catastrophic gaps: a gap that is high-essentiality (sole route, no redundancy) AND high worst-case impact (≥80%) may be assigned priority 1 even with hard tractability, because the cost of not knowing is strictly higher than the cost of measuring. This exception applies to a small number of Step 3 (PDC), Step 4 (ETF/ETF-QO), and Step 10 (PGC-1α, mitophagy flux) gaps where the mechanism is both sole-route and catastrophic.
Intervention gaps (those asking whether a particular trial has been run, rather than whether a mechanism has been measured) are evaluated on a separate axis: priority is driven by whether the trial would change clinical practice conditional on a positive mechanistic signal, and by drug availability. Intervention gaps marked “N/A” on essentiality are conditional gaps — their priority reflects the prior probability that the upstream mechanistic gap will turn out to be positive.
5.2 Essentiality Note
None of the three listed Step 1 gaps is individually essential to cellular energy production. Step 1 as a whole is essential (it gates 100% of downstream ATP), but within Step 1 there are parallel delivery routes (glucose via GLUT1/2/3/4, fatty acids via CPT1/2, and oxygen via vascular perfusion — all independently reviewed in this Step). Failure of one substrate route forces compensation through the others, not catastrophic shutdown. This is different from Steps 3 (PDC) and 7 (ANT), which are sole-route bottlenecks and therefore intrinsically high-priority. At Step 1, prioritization must weigh the combined leverage of a mechanism across multiple subroutes, not just its individual necessity.
5.3 Gap-by-Gap Assessment
| Gap | Essentiality | Worst-case impact | Tractability | Therapeutic leverage | Measurement | Priority |
|---|---|---|---|---|---|---|
| G1: GLUT4 trafficking (skeletal muscle) | Medium — one of 4 GLUT isoforms; GLUT1/3 compensate partially at rest but not under insulin-stimulated demand; glucose uptake impaired in ME/CFS muscle cells with defect proximal to AMPK (Brown et al. 2018) | ~30–60% of gated yield in insulin-sensitive tissues during exertion; negligible at rest | Hard — muscle biopsy with GLUT4 immunofluorescence or surface-biotinylation assay | Immediate — metformin, berberine, exercise-mimetics (AICAR research-grade) | Muscle biopsy (IF) | 2 |
| G2: Malonyl-CoA/CPT1 fuel switch | High — sole gate between glucose-preferential and fatty-acid-preferential metabolism; no parallel switch; Reuter 2011 inferred reduced CPT-I from 30–40% lower long-chain acylcarnitines | ~20–40% of gated yield if the switch is locked in either mode (prevents substrate-flexible compensation during exertion or fasting) | Moderate — plasma malonyl-CoA (specialist LC-MS) + PBMC ACC/AMPK phospho-Western | Near-term — AMPK activators (metformin); ACC inhibitors (firsocostat) in phase II for NASH | Blood draw (specialist LC-MS) | 1 |
| G3: AMPK-activator trials (metformin, berberine) | N/A — intervention gap; in vitro proof-of-concept exists (Brown 2018: metformin rescues AMPK and glucose uptake in ME/CFS muscle cells (Brown et al. 2018)); Long COVID RCT null for established fatigue (Reis 2026: metformin halted for futility ); prevention-stage evidence from COVID-OUT (41% Long COVID reduction (Bramante et al. 2023)) | Unknown pending G1/G2; Long COVID RCT null weakens the case for established fatigue | Easy — metformin is FDA-approved, cheap, and widely tolerated; RCT designs for ME/CFS are straightforward | Immediate — metformin is off-patent and available | Blood draw (RCT) | 3 (downgraded: in vitro positive but clinical null in Long COVID fatigue) |
5.4 Prioritization Summary
G2 (malonyl-CoA/CPT1 switch) is the highest-priority Step 1 gap: it is the sole gate between carbohydrate and fatty-acid oxidation, tractable via plasma and PBMC assays, and has an immediate therapeutic anchor (metformin). G1 (GLUT4 trafficking) is priority 2 because its measurement requires muscle biopsy — the main barrier is tractability, not impact. G3 (metformin/berberine RCT) is priority 3: Brown 2018 provided in vitro proof-of-concept but the REVIVE-TOGETHER Long COVID RCT found metformin null for established fatigue , weakening the clinical case. The priority could be upgraded if a positive mechanistic signal emerges from G1 or G2 in ME/CFS specifically.
None of the three gaps is essential to the chain in the way that PDC (Step 3) or ANT (Step 7) is. Fully resolving all three would close a significant diagnostic and therapeutic gap in insulin-sensitive tissues, but Step 1 has enough parallel redundancy (GLUT1/3, alternative fuel sources, passive oxygen diffusion at low demand) that its failure mode tends to be graded capacity loss under exertion rather than absolute shutdown at rest. This is consistent with the clinical pattern of ME/CFS: preserved baseline function with catastrophic failure under demand (Section Selective Energy Dysfunction: The CNS-Dependency Hypothesis).
The same four-criterion framework is applied to each subsequent step’s research gaps in the subsections below. Step 1 establishes the scoring rubric; Steps 2–10 apply it without restating the criteria. See Summary: Uninvestigated Failure Modes with Therapeutic Implications for the chapter-wide summary list of uninvestigated failure modes.