The chain and its test points: a testable model linking sensory throughput, thiol reserve and copper delivery in post-COVID illness, fatigue and sensory hypersensitivity

Version 1.5 (October 2026): This version adds a four-page supplement in German and English to fact sheet 1.4 (DOI 10.5281/zenodo.23084396), whose text is unchanged. The supplement proposes a mechanism for throughput, the factor the model places in front of the chain: stochastic resonance at the voltage sensor of voltage-gated calcium channels, by which a weak periodic signal that stays below threshold on its own can cross it once broadband noise is added. Device pulsing (VoLTE frames and comfort-noise packets, LTE paging, Wi-Fi beacons, 5G time-division duplex), broadband ambient noise and unintended electromagnetic radiation from satellite constellations are therefore treated as one joint input, which single-factor exposure assessments based on averages cannot capture. The pulse rhythms are placed in the grid of EEG bands, with the beta band as the one band in which no radio device pulses on its own and railway current at 16.7 Hz and the Schumann harmonics lie. Four transporters are named as measurable endpoints: KCC2, cleaved by calcium-activated calpain; EAAT2, inhibited by peroxynitrite; and xCT and LAT1, both bound to CD98hc by a disulfide bridge. A sixth way a survey runs towards zero is added to the section on null findings: few complaints occur at both ends of the reserve curve, with a full reserve and in overload. New are gap G19 (pulsing and broadband noise never tested factorially on the same cell, nor with endpoints at the thiol reserve), predictions P22 (an interaction term on the calcium signal that grows when glutathione is lowered) and P23 (a stronger change in sleep spindles and beta power in neurodivergent children with a switched-on phone at the pillow), and study designs S7, a two-by-two factorial cell experiment, and S8, a blinded within-person sleep-EEG crossover. Thirteen references support the supplement. Version 1.4 (October 2026): This version follows a complete comparison of the ten parts of the underlying series against the sheet. Three test points are added: the circle between viral reactivation and the thiol pool, which lies precisely between two published surveys — one measured glutathione, lipid peroxides and glutathione peroxidase 4 in ME/CFS and long COVID without the virus, the other collected Epstein-Barr and HHV-6 serology in the same individuals with a protein-oxidation marker but no thiols (gap G16, with prediction P19); the transport route behind two apparently separate findings, since the same carrier brings tyrosine and tryptophan into tissue while the evidence on serotonin after COVID is contested and neither study collected the transport itself (G17, with P20); and the level of transcript regulation in copper handling, shown for FDX1 and ATOX1 messenger RNA (G18, with P21). A new section answers the strongest objection to the sheet — why earlier surveys may well have found nothing — naming four ways a survey runs towards zero, the bias of exposure misclassification towards the null, the connection to allostasis and allostatic load, five non-invasive measures of reserve, and the CENT extension of CONSORT for within-person analysis. Eight reinforcements come from the series: peptidylglycine alpha-amidating monooxygenase as an eighth copper-dependent enzyme, the copper chaperone CCS with its disulfide bridge to SOD1, the identity of glutathione S-transferase zeta 1 with maleylacetoacetate isomerase in tyrosine catabolism, XPO1 and selinexor as an established precedent for functional loss of a guardian through logistics, CAT1 and ASCT2 as simultaneous virus receptors, mitochondrial DNA release and the cGAS-STING pathway, and two limitations the model names for itself. The sheet now states twenty-one falsifiable predictions and eighteen named gaps, with twenty-nine additional references. Versions 1.0 to 1.3 remain citable under the same concept DOI. Version 1.3 (October 2026): Three further test points are added, and the model closes into a circle a third and a fourth time, both times through selenium. The additions are: the circle between pathogen and pool, which has never been measured as a circle (gap G13, with prediction P16); the selenium-dependent reset system and what stalled copper does to it (G14, with P17); and the interval until recurrence as a measurement in its own right (G15, with P18). Thioredoxin reductase is a selenoenzyme and serves both resets in this model, that of the vitamin K circle and that of oxidised PTEN; raised copper lowers its activity in cell culture by way of selenoprotein synthesis. On the uptake side, the carrier that brings methionine and selenomethionine into tissue, LAT1 with CD98hc, is held together by a disulfide bridge — so a selenium deficit can arise while intake remains unchanged, the same form as the model's central claim about copper. Also included: a reinforcement of the ninth test point, since ATM not only is switched by the redox state but itself raises the cell's antioxidant capacity through the pentose phosphate pathway; a sixth study design that tests P18 retrospectively from existing longitudinal records; and a rewritten section 8, stating that the observations and records this work grew from were assembled from 2022 onwards. The sheet now states eighteen falsifiable predictions and fifteen named gaps, with thirteen additional references. Versions 1.0 to 1.2 remain citable under the same concept DOI. Version 1.2 (25 September 2026): Four further test points are added — the redox switch at PTEN and ATM (gap G9, with predictions P11 and P12), copper accumulation as an acquired state (G10, with P13), the sulfate exit at the endothelial glycocalyx (G11, with P14), and the shared draw on the thiol pool by pathogens that cannot synthesise it (G12, with P15). With them the model closes into a circle a second time, now around copper: delivery of copper to ATP7A and ATP7B is co-regulated by the glutathione system, ATP7A loads extracellular superoxide dismutase SOD3 on the glycocalyx, and superoxide is the fourth link of the chain from which peroxynitrite is formed. The thiol pool therefore also carries the defence against the very species that consumes it. Also included: four reinforcements of existing predictions, a self-correction to gap 5 (the earlier wording was drawn too widely; the gap stands for two different and sharper reasons), the extension of the series to ten parts, a fifth study design covering three of the new predictions in a single blood draw, and twenty-six additional references. The sheet now states fifteen falsifiable predictions and twelve named gaps. Versions 1.0 and 1.1 remain citable under the same concept DOI. Version 1.1 (13 September 2026): The vitamin K cycle is added as an eighth test point (gap G8) with two new falsifiable predictions (P9, P10) and fourteen additional references. With it the model closes into a circle: peroxynitrite, the endpoint of the chain as described in version 1.0, acts through S-nitrosylation of thioredoxin-like domains on vitamin K reduction, and through undercarboxylated matrix Gla protein back onto calcium distribution, where the chain begins. A figure of the closed chain has been added to section 1. Version 1.0 remains citable under the same concept DOI.