The Nervous System · Part One · How It Is Built and Fueled
Lesson 12 / 61
Mitochondria and ATP: From Acetyl-CoA to the Proton Gradient
Where oxidation becomes ATP, and what a neuron buys with it.
Mitochondria convert the oxidation of food into adenosine triphosphate, the molecule every neuron spends to hold its membrane potential and to fire. Fuel is oxidized to acetyl-CoA. The citric acid cycle strips its electrons onto carriers, and the electron transport chain uses them to pump protons across the inner membrane. Protons falling back through ATP synthase pay for the cell. The Unified Model of Tone reads that supply as the price of holding a state.
Respiratory chain
Five complexes, with I, III, and IV pumping protons
ATP synthase
Complex V, a bacterial F1 rotor of about 1 nm radius inside a 5 nm stator barrel
Electron carriers
NADH enters at complex I, FADH2 at complex II
Resting oxygen use
About 90 percent runs through mitochondria in mammals at rest
Mitochondrion
A double-membraned organelle present in every neuron. The outer membrane is smooth. The inner membrane folds into cristae and carries the electron transport chain. The enclosed matrix holds the citric acid cycle enzymes and the pyruvate dehydrogenase complex. Fuel is oxidized inside the matrix, and the energy released is captured on the membrane.
The gradient and tone
The inner membrane holds a voltage built by proton pumping and drawn down by ATP synthase. Its level reports the running balance between what the cell is making and what the cell is spending. A neuron works that voltage inside a range it defends. The Unified Model of Tone calls a defended range of that kind tone at the scale of a single machine, the same regulation it reads in a membrane or a reflex, one level down.
01Fuel converges on acetyl-CoA
Every fuel a neuron burns reaches the mitochondrion as acetyl-CoA
Oxidation runs in three stages, and the mitochondrion owns the last two. Digestion breaks food into monomeric sugars, fatty acids, and amino acids. Those monomers enter the cell and are oxidized in steps, first in the cytosol and then in the mitochondrial matrix. Stage one delivers acetyl coenzyme A, an activated carrier molecule. The citric acid cycle takes it from there, and oxidative phosphorylation banks the energy as ATP. Releasing that energy in measured increments keeps the cell from burning its fuel in one destructive step.
The tissue this machinery serves is expensive to run. Human brain sits at about 240 kcal per kilogram per day against 13 for skeletal muscle, an eighteenfold difference. Those coefficients were tested against whole-body MRI and indirect calorimetry in 80 young women Wang 2012. Heart and kidneys run higher still at 440, so the nervous system is among the most expensive tissues per kilogram without being the single most expensive. Cost per kilogram is not the same as total draw. What share of the body's fuel the brain actually takes, and how that fuel reaches the cell, is set out on Glucose and the brain.
Pyruvate reaches the cycle by two different routes
Pyruvate crosses into the matrix and the pyruvate dehydrogenase complex decarboxylates it to acetyl-CoA. Fatty acids reach the same molecule through beta-oxidation, and several amino acids join there. Pyruvate can instead enter through pyruvate carboxylase, which replenishes oxaloacetate directly and keeps the cycle stocked with the acceptor it needs to start a turn. Every neuron downstream, from cortical pyramidal cells to autonomic ganglia, depends on those routes staying open to hold its resting state and to fire on demand.
02Findings
What the research shows
03The citric acid cycle
The citric acid cycle strips electrons from carbon and loads them onto carriers
The citric acid cycle turns in the mitochondrial matrix and oxidizes every acetyl group the cell delivers to it. Citrate synthase joins acetyl-CoA to oxaloacetate, a four-carbon molecule, to form six-carbon citrate. Aconitase and isocitrate dehydrogenase carry citrate on to alpha-ketoglutarate. Succinate dehydrogenase, which is complex II, oxidizes succinate to fumarate. Fumarase adds water to give malate, and malate dehydrogenase oxidizes malate back to oxaloacetate. The acceptor is regenerated at the end of the turn and the carbon leaves as carbon dioxide.
Each turn yields three NADH, one FADH2, and one GTP. The cycle gives up little ATP directly and hands its real product, charged NADH and FADH2, to the respiratory chain lining the cristae of the inner membrane. The enzymes doing that stripping depend on cofactors the body must supply, which the B-vitamins and cofactors takes up in detail.
Cycle intermediates reach the release machinery directly
Losing one cycle enzyme damages synaptic transmission on its own. In Drosophila, loss of isocitrate dehydrogenase 3a lowered alpha-ketoglutarate and produced transmission defects resembling loss of synaptotagmin-1 Ugur 2017. Supplementing alpha-ketoglutarate suppressed those defects through a mechanism independent of ATP and of neurotransmitter levels. The metabolite promoted interaction between the C2 domains of synaptotagmin-1 and phospholipids. Metabolism therefore holds a second route into signaling that no accounting of ATP alone will find.
04The respiratory chain
Electron transport stores the energy of oxidation as a voltage across the inner membrane
Five protein complexes sit in the inner mitochondrial membrane and run oxidative phosphorylation. Complexes I, III, and IV pump protons out of the matrix. NADH is re-oxidized at complex I and FADH2 at complex II, feeding their electrons into the chain. Coenzyme Q and cytochrome c carry those electrons onward to oxygen, which is reduced to water at complex IV. Electrons lose energy in measured steps as they travel, and that released energy drives the pumping. Peter Mitchell named the coupling in 1961, proposing that phosphorylation joins to electron and hydrogen transfer by a chemi-osmotic mechanism Mitchell 1961.
Complex I pumps first, and its architecture has been resolved. The bacterial enzyme from Thermus thermophilus is a 536 kDa assembly of 16 different subunits with 64 transmembrane helices and 9 iron-sulfur clusters, solved at 3.3 Å Baradaran 2013. It translocates four protons per catalytic cycle. The quinone headgroup binds at the deep end of an enclosed chamber near cluster N2, and conformational changes propagate outward to four antiporter-like domains. The mammalian enzyme runs larger. A bovine structure at 4.2 Å holds 45 subunits, the 14 core subunits conserved from bacteria to humans plus 31 supernumerary ones Zhu 2016. Sorting those particles computationally separated structural classes matching the active-to-de-active transition the enzyme makes during hypoxia. Oxygen availability therefore changes the shape of the machine as well as its substrate supply.
Cytochrome c oxidase reduces oxygen to water, and ATP synthase turns as a rotary motor
Oxygen is reduced to water at cytochrome c oxidase, and the metal centers that hand it the electrons are known. The bovine 13-subunit enzyme was resolved at 2.8 Å Tsukihara 1996. Each monomer carries two hemes A and three copper atoms, plus one magnesium and one zinc. The dimer runs to 3,606 amino acid residues. Two possible proton pathways span the matrix and cytosolic surfaces, alongside possible channels for the chemical protons that make water and for oxygen itself. Complex V works as a rotary motor, the smallest one known. Direct observation of isolated F1-ATPase from a thermophilic bacterium showed a central rotor of about 1 nm radius turning inside a stator barrel of about 5 nm radius Noji 1997. A fluorescent actin filament attached to the gamma subunit made the turning visible. The filament rotated for more than 100 revolutions, and rotary torque reached more than 40 pN nm under high load.
The gradient is a range the neuron defends
Mitochondrial membrane potential in cultured rat cortical neurons sits at -139 mV at rest Gerencser 2012. It is regulated between -108 mV and -158 mV by concerted increases in ATP demand and calcium-dependent metabolic activation, with total calibration error under 11 mV. The membrane works across a range of roughly 50 mV that tracks what the cell is being asked to do. Output also depends on shape. In cultured mammalian cells, cristae shape determines the assembly and stability of respiratory chain supercomplexes and therefore respiratory efficiency Cogliati 2013. Two cells given identical fuel and identical oxygen can produce different amounts of ATP because their folds differ.
05Proton leak and the P/O ratio
The ATP a neuron receives is the gradient minus its leak
Only 80 percent of mitochondrial oxygen consumption is coupled to ATP synthesis, and the other 20 percent leaks back across the inner membrane on purpose. About 90 percent of mammalian oxygen consumption in the standard state is mitochondrial, and both proportions come from a whole-body budget for mammals at rest Rolfe 1997. Heat comes mainly from mitochondrial respiration, oxidative phosphorylation, and the leak on the inner membrane. Across the whole body at rest, protein synthesis takes 25 to 30 percent of the coupled share and the sodium-potassium ATPase takes 19 to 28 percent. Those are whole-body mammalian figures and not a brain budget. The itemized account of what neural signaling costs sits on the resting and action potential.
The cell tunes that leak deliberately. Proton leak across the inner membrane allows adjustment of coupling efficiency, and uncoupling proteins control how much escapes Divakaruni 2011. The cell sets how much of its gradient becomes chemical work and how much becomes heat, so coupling efficiency is a value it holds and can raise or lose.
Yield per carrier depends on where the electrons enter the chain
NADH and FADH2 are not worth the same. Measured in isolated mitochondria, P/O ratios run about 2.5 with NADH-linked substrates and about 1.5 with succinate Hinkle 2005. The fractional values follow from the coupling ratios of proton transport. ATP synthase structure implies a proton-to-ATP stoichiometry of 10/3 rather than 3, consistent with P/O values of 2.3 and 1.4. Net yield is a ratio set by pump and turbine stoichiometry, which is why a fixed count of ATP per glucose does not survive measurement.
A damaged chain loses rate before it loses efficiency
The coupling ratio is defended even when catalysis is compromised. Fourteen mutations of the Nqo13/ND4 membrane subunit were made in Paracoccus denitrificans, a bacterial model of the conserved complex I core. Catalysis slowed, and the enzyme still pumped exactly four protons per NADH oxidized, with no evidence of escape cycles that bypass blocked pumping steps Jarman 2022. A compromised chain delivers fewer ATP per unit time while its yield per electron holds. The same fuel then produces fewer molecules because the line runs slower, and the cell must burn more substrate or settle for a smaller budget.
06Signaling and nonsignaling spend
Signaling takes about 70 percent of the oxidative glucose burned by human gray matter
Awake resting cortical signaling takes about 70 percent of oxidative glucose consumption, and mass-dependent nonsignaling work takes about 30 percent Yu 2018. That is the gray matter figure. White matter runs in near opposite contrast, because nonsignaling demand dominates there. Inhibitory neurons and glia take 15 to 20 percent of that oxidative glucose consumption, and excitatory neurons take the rest. The human glutamatergic signaling rate is about 1.2 Hz, roughly a quarter of the rate calculated for rat cortex. The itemized four-way split of signaling energy, modeled from rodent gray matter, is set out on the resting and action potential.
Oxidative phosphorylation powers that spending, and glycolysis does not. In rat hippocampal slices, neuronal activity rapidly lowered extracellular oxygen and intracellular NADH Hall 2012. It did so with lactate dehydrogenase blocked to prevent lactate generation, and with only 20 percent superfused oxygen to mimic physiological levels. Pharmacological analysis in that preparation put 46 percent of oxygen use on postsynaptic glutamate receptors and 26 percent on postsynaptic action potentials. Presynaptic action potentials took 11 percent, and presynaptic calcium entry with transmitter release took 17 percent. Those are measured shares in a slice, and they land in approximate accord with the modeled budgets.
Mitochondria are anchored where demand is highest
In mature neurons only one-third of axonal mitochondria are motile, and the stationary remainder act as local energy sources and buffer intracellular calcium, a picture drawn from rodent axonal transport work Lin 2015. Neurons use dedicated mechanisms to hold mitochondria at distal sites of high demand such as synapses and axonal branches. The balance between motile and stationary pools responds quickly to changes in axonal and synaptic physiology, and defects in that transport appear in several major neurological disorders. In cultured rat hippocampal neurons, a nerve terminal holds about a million free ATP molecules at steady state, matched to the synaptic vesicle cycle that drives most of its activity-driven demand Rangaraju 2014. Even brief interruption of activity-stimulated ATP synthesis severely impairs presynaptic function.
07Energy failure and circuit tone
When mitochondrial output falls, conduction fails before transmitter release does
When mitochondrial output falls at the mouse calyx of Held, action potential propagation fails before synaptic vesicle recycling does. Sustained transmission at that mature synapse relies exclusively on mitochondrial ATP production supported by bath lactate, and not on glycolysis Lujan 2021. Transmission during short repetitive bursts was unaffected while either pathway remained intact, and before the onset of hearing both pathways were needed. At both ages propagation began to fail first, so energy failure arrives as a conduction problem before it arrives as a release problem.
Mitochondrial disease is common enough to sit inside general neurology. The minimum prevalence of mtDNA mutations among adults in North East England was 1 in 5,000, or 20 per 100,000 Gorman 2015. Cases were drawn from 1990 to 2014 in the population served by one referral center, and prevalence was evaluated for midyear 2011. Nuclear mutations accounted for clinically overt adult mitochondrial disease in 2.9 per 100,000 adults. Combined prevalence near 1 in 4,300 places mitochondrial disease among the commonest inherited adult neurological disorders. Fuel failure and degeneration follows what happens to a neuron once the supply stops.
Fatigue tracks the same variable in measurable form. Peripheral blood mononuclear cells from patients with myalgic encephalomyelitis or chronic fatigue syndrome showed significantly lower mitochondrial coupling efficiency Fernandez-Guerra 2021. Proteome alterations centered on pyruvate dehydrogenase and coenzyme A metabolism, and the cells showed a decreased capacity to provide adequate intracellular ATP. Severe fatigue correlated with the severity of autonomic dysfunction and with overall physical well-being. The cohort was six patients and the cells were blood cells, not neurons. The measured association still runs from an organelle setting to an integrative state.
Coupling efficiency and membrane potential are tone at organelle scale
Energy in this system is a balance the body defends. Coupling efficiency is a value the cell adjusts, and mitochondrial membrane potential holds a range near -139 mV instead of one fixed charge. The Unified Model of Tone reads both as tone appearing at organelle scale, held inside a window and moved by demand. That an inaccurate model of the body costs the brain measurable energy is already published with its sources on Load. This page does not claim it as new. What this page stakes is the organelle-level version, and it is directional. Raise a tissue's excitability and its mitochondria are drawn through more of their working range, with coupling efficiency adjusted to meet the draw. Drive excitability past what the tissue can supply and coupling efficiency falls and the range narrows. Both readings should move with excitability and recover with it, on an oxygen electrode and a potentiometric dye, in tissue whose firing is already being recorded. Tone carries the general commitment. The measurements above are what this page puts behind it.
Dysglycemia, poor oxygen delivery, and cofactor gaps converge on one organelle and degrade one currency. A terminal that cannot rebuild what a burst just spent stops behaving like a terminal with anything held back. Coupling efficiency falls. The working range narrows toward its floor, and the circuit settles at a resting excitability nobody chose for it. Fatigue that does not clear with rest, a shrinking tolerance for load, and conduction that gives out before release are what that settled state looks like in a person.
A neuron pays for its resting state before it pays for a single action potential.
08Tone
How this system expresses tone
A mitochondrion states its condition as a voltage. The inner membrane sits near -139 mV in a cultured rat cortical neuron and travels about 50 mV as demand rises, which is a working range and not a fixed charge.
Set point
Coupling is a ratio the cell sets. About 80 percent of mitochondrial oxygen use pays for ATP synthesis, and uncoupling proteins release the rest as heat.
Load
The tissue is expensive before anything happens in it. Human brain runs near 240 kcal per kilogram per day against 13 for skeletal muscle, an eighteenfold standing cost.
Constraint
Supply is local. A cultured rat hippocampal terminal holds about 1 million free ATP molecules, and a brief loss of local synthesis impairs release severely.
Gain: demand does not simply draw on the gradient but moves it, pulling the inner membrane across a 50 mV working range as ATP use and calcium entry rise. Oscillation: the cycle regenerates its own acceptor every turn, so output is a rate on a loop rather than a stock to be drawn down. Input quality: sugars, fatty acids and amino acids all converge on acetyl-CoA, so what the cycle can do is set by the carbon delivered to it. Time course: complex I switches from an active to a de-active conformation within minutes of losing oxygen. Coupling: cristae shape determines supercomplex assembly and respiratory efficiency in cultured mammalian cells, which makes structure and output one measurement. Prediction: neurons park mitochondria at synapses and axonal branches before the traffic arrives, and only one-third of axonal mitochondria in mature rodent neurons are still moving.
09Across the library
How this page relates to the rest of the library
Where fuel sits among the things a neuron must have before anything else in the system works, and why that ordering decides what to address first.
The itemized budget for neural signaling, modeled from rodent gray matter, and the cost of one extra spike per cortical neuron per second.
The fuel arriving at the gate. Transport across the endothelium, the astrocyte supply line, and what an uneven supply does to a tissue that stores almost nothing.
The other half of oxidative phosphorylation. How oxygen reaches complex IV, and what a fall in carrying capacity does to the tissue with the highest demand.
The cofactors that run the cycle enzymes named here, including the ones pyruvate dehydrogenase cannot turn without.
What follows once supply stops, from calcium entry to cell death, and why degeneration begins as an energy problem before it becomes a structural one.
Why an inaccurate model of the body costs the brain measurable energy, and how that standing demand is scored across a whole person on a mortality-linked composite.
10Frequently asked
Questions about this topic
What do mitochondria actually do in a neuron?
A neuron's mitochondria buy its excitability. They oxidize fuel and capture the released energy as ATP. The neuron spends that ATP on ion gradients, on transmitter recycling, and above all on the synaptic vesicle cycle, which drives most of the activity-driven demand at a terminal. Stationary mitochondria do a second job where they sit, buffering intracellular calcium, and two-thirds of axonal mitochondria stay parked in mature rodent neurons. A cultured rat nerve terminal holds only about 1 million free ATP molecules, so local synthesis must keep pace with local spending.
How is the citric acid cycle different from oxidative phosphorylation?
The citric acid cycle runs in the mitochondrial matrix and oxidative phosphorylation runs on the inner membrane. The cycle joins acetyl-CoA to oxaloacetate through citrate synthase. Carbon then passes through citrate and alpha-ketoglutarate, then succinate, fumarate, and malate. Oxaloacetate is regenerated at the end of the turn, and the direct ATP yield is small. The real product is charged carriers, NADH and FADH2. Oxidative phosphorylation then takes the electrons off those carriers, pumps protons with the energy released, and uses the returning protons to make ATP at complex V.
Why do NADH and FADH2 not produce the same amount of ATP?
The yield depends on where the electrons enter the chain. NADH is oxidized at complex I and its electrons pass three proton pumps. FADH2 enters at complex II and bypasses the first pump, so fewer protons cross per electron pair. Measured in isolated mitochondria, P/O ratios run about 2.5 with NADH-linked substrates and about 1.5 with succinate. The values are fractional because they follow the coupling ratios of proton transport. ATP synthase structure implies a proton-to-ATP stoichiometry of 10/3 rather than 3, giving P/O values near 2.3 and 1.4.
How large is the mitochondrial membrane potential in a neuron?
The mitochondrial membrane potential is large and it moves. In cultured rat cortical neurons it sits at -139 mV at rest and is regulated between -108 mV and -158 mV, with total calibration error under 11 mV. Concerted increases in ATP demand and calcium-dependent metabolic activation drive that movement, so the working range of roughly 50 mV tracks what the cell is being asked to do. The gradient is the immediate store from which every ATP is drawn, and its level reports the balance between supply and demand.
Why does a resting brain use so much energy?
Most of that energy holds states rather than changes them. In awake human gray matter about 70 percent of oxidative glucose consumption goes to signaling and about 30 percent to mass-dependent nonsignaling work, and the proportions invert in white matter. Inhibitory neurons and glia take 15 to 20 percent of the signaling share. The human glutamatergic signaling rate is about 1.2 Hz, roughly a quarter of the rate calculated for rat cortex. Ion gradients decay continuously and have to be rebuilt whether or not anything is happening.
What fails first when mitochondrial energy production falls?
Conduction fails before transmitter release does. At the mouse calyx of Held, action potential propagation began to fail before defects appeared in synaptic vesicle recycling, at both immature and mature ages. The mature synapse depends exclusively on mitochondrial ATP production for sustained high-frequency transmission, with glycolysis unable to carry that load on its own. Short repetitive bursts stayed unaffected while either pathway remained intact. A circuit therefore loses the ability to carry the signal down the axon before it loses the machinery that releases transmitter.
Is proton leak across the mitochondrial membrane a defect?
Proton leak is a regulated control rather than a fault. In a whole-body budget for mammals at rest, about 90 percent of oxygen consumption is mitochondrial. Of that share, roughly 20 percent is uncoupled by leak and 80 percent is coupled to ATP synthesis. The leak allows adjustment of coupling efficiency, and uncoupling proteins regulate how much escapes. Heat comes mainly from mitochondrial respiration and from that leak. A cell that let nothing escape would gain little, because that setting is how it trades yield for heat.
Why can fatigue persist when routine blood tests come back normal?
A standard panel measures what is in the blood, not how efficiently mitochondria turn oxygen into ATP. Immune cells from six patients with myalgic encephalomyelitis or chronic fatigue syndrome showed significantly lower mitochondrial coupling efficiency, with proteome changes centered on pyruvate dehydrogenase and coenzyme A metabolism. That was an exploratory study in blood cells rather than neurons. Inherited energy failure is not rare either, reaching a combined prevalence near 1 in 4,300 adults in one English population. Severe fatigue tracked the severity of autonomic dysfunction.
11The sources
References
Sources: primary literature, linked inline.