Cellular respiration: food, oxygen and ATP: sources & model

Charge a tiny membrane, disconnect its supply, and choose where the stored driving force goes. Discover how cells couple fuel chemistry to ATP—and why a bypass changes the result.

What supports the explanation?

What this model assumes

1. The workbench is an authored voltage-only finite-capacitor assay, not measured human physiology, molecular dynamics, cell timing or an ATP-per-glucose calculator.
2. Fixed chemical budgets, buffered equal pH and coarse full-turn packets omit activity changes, reverse catalysis, transport costs, kinetics and many metabolic routes.
3. Rotor motion, colors and membrane geometry are explanatory adaptations; the experimental coordinate models and frozen density data do not record that animation.
4. The buffered workbench does not give a complete elemental hydrogen/oxygen balance including nucleotide protonation, transport water and counterions.
5. Paper resets supply a fresh assumed condition externally. Moving cards repeatedly cannot create unlimited energy.
6. No running distance, health status, diet recommendation or exercise advice follows from these demonstrations.
7. A pathway, not one giant reaction: Glycolysis in the cytosol converts one glucose to two pyruvates and gives a net substrate-level ATP contribution; it does not directly consume oxygen. Pyruvate conversion releases CO2 and makes acetyl-CoA. The citric acid cycle regenerates its oxaloacetate pool while transferring reducing equivalents. Not all ATP is made by mitochondrial ATP synthase. Reactome’s human pathway entries are curated synthesis, with some molecular detail inferred from other species.
8. Carbon accounting is not atom tracing: The net oxidation account C6H12O6 + 6 O2 → 6 CO2 + 6 H2O balances matter overall. It does not identify the path of each oxygen atom or show every intermediate. Two CO2 are released in conversion of two pyruvates; complete oxidation through the cycle gives four more net CO2. Newly arriving acetyl carbons need not leave in the first turn. ATP turnover is a separate coupled account, not an extra carbon product in this equation.
9. Carriers and entries: NADH transfers reducing equivalents and returns to NAD+; it does not transform into ATP. In the scoped mammalian chain, complexes I, III and IV contribute to outward proton transfer. Complex II feeds electrons into the quinone pool without pumping protons. Its FAD cofactor is enzyme-bound, not a freely delivered FADH2 parcel.
10. Chemical protons and pumped protons: At canonical complex IV, four electrons reduce one O2 to two waters. Four chemical protons are consumed in that reaction and four additional protons are pumped. Our lumped two-NADH supply event consumes one O2, forms two redox waters and transfers 20 outward H+ equivalents across the idealized I/III/IV chain. Buffered chemical protons are separate from that pumped account.
11. Eight and three: Watt et al. resolved an eight-member c ring in bovine F1–c8, PDB 2XND. Its ideal complete turn translocates eight protons and supports three ATP catalytic events: intrinsic ratio 8/3. The full-turn packet is not three simultaneous reactions or a one-to-one match between 45° rotor increments and 120° catalytic steps. Ring sizes vary between organisms; transport and leaks add costs beyond the intrinsic ratio.
12. A structure is not a movie: PDB 6ZPO is a bovine ATP synthase model fitted to cryo-EM data, with an IF1 inhibitor in the experimental preparation. Its catalytic head faces the matrix. Source inspection retains the inhibitor; an uninhibited working animation is a schematic adaptation. The paper’s Results places proton entry on the intermembrane-space side; a conflicting “matrix” statement in its abstract must not reverse the teaching direction.
13. The deliberately chosen assay: The workbench isolates voltage by externally buffering both sides at equal pH. Capacitance C=10⁻¹⁴ F, supply budget B=440000/NA joules per two-NADH event and ATP demand G=50000/NA joules per ATP are authored conditions, not measurements from the pictured specimen. Finite inventories gate events while chemical activities and G are held fixed by assumption.
14. A finite electrical account: Let n count net outward monovalent charge equivalents, e=1.602176634×10⁻¹⁹ C and NA=6.02214076×10²³ mol⁻¹. Q=ne, V=Q/C and U=(ne)²/(2C). V is positive for the intermembrane-space side relative to matrix. The number n counts net stored charge, not every dissolved proton or pH. Exact physical constants do not turn the chosen capacitance into measured biology.
15. Check before committing: A supply event needs two NADH, one O2 and enough budget for U(n+20)−U(n). An ATP full-turn event needs at least eight stored charge equivalents, three ADP and three Pi, with U(n)−U(n−8)≥3G. Failed guards leave state unchanged. The full-turn abstraction can reject a packet even when a real enzyme could undergo smaller substeps.
16. Follow the remaining energy: The bypass returns up to 20 stored charge equivalents per event, sending the lost U to dissipation without changing ATP. ATP use returns one ADP and one Pi and transfers G to cellular use/dissipation. The balance includes stored electrical free energy, ATP synthesis free energy, dissipation and cellular use. A nonzero U is not necessarily enough for another forward synthesis event.
17. A separate fixed-force calculation: If an external apparatus holds the proton-motive force fixed, one ideal c8 turn can receive 8FΔp per mole of turns, with F=eNA≈96485.332 C/mol. At 0.15 V this is about 115.78 kJ/mol of turns; at 0.20 V it is about 154.38. Our chosen three-ATP demand is 150 kJ/mol of turns, so only the latter passes. This externally clamped comparison is not the discharging finite workbench.
18. Real membranes also have a chemical difference: Define favorable inward Δp=(ψIMS−ψmatrix)+(2.303RT/F)(pHmatrix−pHIMS). At 310 K, a 0.5 pH-unit difference contributes about 30.76 mV and represents a proton concentration ratio of about 3.16. The voltage-only workbench omits this term rather than adding it twice. Real enzymes may reverse under suitable conditions; reverse ATP hydrolysis and IF1 regulation are not simulated.
19. An experiment can separate two outputs: Jones et al. studied inverted bovine-heart membrane vesicles. Figure 2 reports fitted NADH-oxidation/ATP-synthesis slopes of 0.484/0.337 in control, 0.921/−0.003 with gradient dissipation, and −0.011/−0.015 with complex-I inhibition, in μmol min⁻¹ mg⁻¹. Preserve the small negative near-background fits; they are not negative molecular inventories. These are separate measured outputs, not fitted traces of our workbench or human exercise data.
20. Looking inside a frozen specimen: EMD-43064 is a tomogram from wild-type mouse embryonic fibroblasts in Fry et al.’s study. Moving between slices changes spatial depth, not time. The grayscale is reconstructed electron-scattering density, not voltage or visible moving protons. A fitted bovine protein structure, a mouse-cell tomogram and a hypothetical assay are different evidence layers.

What has been checked

Analytical reference cases, conservation or transition invariants, finite drawing commands, bounded setup parsing, discovery and route integrity are checked automatically. These checks do not establish anatomical fidelity, learner outcomes or browser/device compatibility. Independent subject review, learner trials, comprehensive accessibility review and browser video encoding checks remain pending.

Each source supports the associated claim. Sources do not certify this implementation or its visuals.

About the cover illustration

Actual Brytalearn browser render of smoothed heavy-atom envelopes from bovine ATP synthase PDB 6ZPO, Spikes et al. (2020), CC0 archive coordinates. Working adaptation omits IF1 and adds teaching colors and membrane cutaway; not measured molecular motion.