Module 4: ATP Synthesis (F₀F₁)

By the late 1970s the chemiosmotic hypothesis (Module 3) had been accepted. What remained was the mechanism of the ATP-synthesizing enzyme itself. Efraim Racker had shown that the soluble “F₁” fragment of the enzyme, detached from membranes, catalyzed ATP hydrolysis rather than synthesis; reassembly with the membrane-bound “F₀” fragment restored coupled synthesis. The F₁F₀ structure was clearly two-part, and the coupling was mediated by protons.

Paul Boyer’s binding-change mechanism(reviewed in Boyer 1997, Annu. Rev. Biochem.) proposed that the three catalytic β-subunits of F₁ cooperate in a cyclic, rotary manner. Each turn of the γ-shaft drives each β through three states (open, loose, tight), generating 3 ATPs per revolution. The mechanism was confirmed by:

• Walker et al. (1994, Nature): atomic structure of bovine F₁ showing three different β-subunit conformations in one trimer (β_E, β_DP, β_TP), matching Boyer’s three-state prediction.
• Noji, Yasuda, Yoshida & Kinosita (1997, Nature): direct visualization of γ-shaft rotation by attaching an actin filament to F₁ and watching it spin under a microscope.
• Yasuda et al. (1998, Nature): quantitative measurement of 120° stepping with sub-steps at low [ATP].

The 1997 Nobel Prize in Chemistry was shared among Paul Boyer, John Walker, and Jens Skou (the discoverer of the Na⁺/K⁺-ATPase). Kinosita arguably should have been on the ticket too, but Nobel-committee rules capped the laureates at three.

The F₀ sector is embedded in the inner mitochondrial membrane and contains the proton-translocating c-ring plus the stator subunit a. The c-ring is a cylindrical stack of 8 to 15 identical c-subunits (depending on species), each contributing a pair of transmembrane helices and a central essential glutamate (Glu-58 in mammalian c) that accepts a proton from the half-channel in subunit a.

The proton-translocation mechanism is elegantly mechanical. Subunit a contains two half-channels that do not span the membrane: one opens to the IMS, one to the matrix. A proton enters from the IMS, binds Glu-58 on one c-subunit as the ring rotates, travels almost a full revolution through the hydrophobic lipid core, and exits via the matrix half-channel. Each cycle of the ring moves as many protons as there are c-subunits. The c-ring stoichiometrytherefore sets the H⁺/ATP ratio:

Different c-ring sizes represent an evolutionary tuning of the enzyme to the available PMF. Chloroplasts operate at lower \(\Delta\psi\) but higher\(\Delta\text{pH}\) with less total PMF per proton; their c-ring is larger, using more protons per ATP to compensate. Alkaliphilic bacteria, which operate with extreme \(\Delta\text{pH}\) and minimal \(\Delta\psi\), have 15-member c-rings.

The stator arm (peripheral stalk) prevents free rotation of the catalytic α₃β₃ head when subunit a pushes the c-ring. In mammals, the peripheral stalk is a long elongated coiled coil of subunits b, d, F₆, OSCP. Mutations that disrupt the stator cause severe ATP-synthase disorders (NARP syndrome, Leigh syndrome).

The F₁ head protrudes into the matrix and has the formula α₃β₃γδε. The three α and three β subunits alternate in a hexameric ring with C3 symmetry; each pair shares a nucleotide-binding pocket. The catalytic sites are on β, with α contributing structural residues. The central γ-shaft passes through the hexamer, bearing an asymmetric coiled coil that breaks the symmetry and imposes three different conformations on the three β-subunits at any instant.

Walker’s 1994 crystal structure (Nature 370, 621–628) was the Rosetta stone of the field. It showed the three β-subunits in three distinct conformations:

• β_E (empty): no nucleotide, open active site. Maps to Boyer’s O state.
• β_DP (ADP-bound): ADP + P_i, loose. Maps to L.
• β_TP (ATP-bound): ATP, tight, closed. Maps to T.

The three conformations are interconverted by 120° rotation of the γ-shaft, which forces each β-subunit to “feel” a different side of the shaft’s asymmetric cross-section as it sweeps past. This is the mechanical coupling at the heart of the binding-change mechanism.

The δ subunit anchors F₁to the peripheral stalk at the top of the head; the ε subunit is attached to the foot of the γ-shaft and mediates regulatory interactions with the c-ring. Bacterial F₁ additionally has the ε-subunit that acts as an intrinsic inhibitor, folding into the F₁ head when the enzyme runs backward.

Boyer’s key insight was that ATP synthesis on the enzyme is actually isoenergetic. Once ADP and P_iare bound tightly in the T state, the condensation reaction proceeds spontaneously,\(\Delta G \approx 0\) on the enzyme. The energy cost of ATP synthesis is all in releasing the tightly-bound ATP—that’s where the proton-motive force does its work, by mechanically forcing the γ-shaft to rotate and open the active site.

The key thermodynamic consequence: the enzyme doesn’t need a big conformational change during the bond-forming chemistry. All the conformational change happens in ATP release, which the rotary mechanism delivers by mechanical leverage.

Modern refinements from Kinosita’s single-molecule work subdivide the 120° step into an 80° sub-step (ATP-binding or ADP-release) and a 40° sub-step (Pi release or catalytic dwell). At very low [ATP], the 80° sub-step becomes rate-limiting and the sub-structure is directly visualized in the rotation traces. At saturating ATP, the two sub-steps merge.

The most dramatic experiment in 20th-century bioenergetics was done by Hiroyuki Noji, Ryohei Yasuda, Masasuke Yoshida, and Kazuhiko Kinosita Jr. at ERATO/RIKEN (1997–1998, Nature). They:

• Genetically engineered bacterial F₁ with a histidine tag on the β-subunit to immobilize it on a Ni-NTA-coated glass slide.
• Biotinylated a cysteine on the γ-shaft and attached a fluorescent actin filament (~1 µm) via streptavidin.
• Added ATP and watched the actin filament under a fluorescence microscope.
• The filament rotated unidirectionally at ~4 rev/s (per single enzyme).

The filament served as a giant indicator, amplifying a ~1 nm rotation of the central shaft to a micron-scale motion. Sub-step structure emerged at low [ATP], with dwells at 0°, 80°, 200°, and 320° positions within each 360° cycle.

From the hydrodynamic drag on the actin filament, they calculated the torque output: ~50 pN·nm. For comparison, typical cytoskeletal motors (kinesin, myosin) generate ~6–10 pN·nm torque equivalent. F₁ is thus a particularly powerful motor for its size (the γ-shaft is ~1 nm thick).

Kinosita’s group subsequently showed (Kinosita et al. 2004, PNAS) that F₁ operates at >90% mechanical efficiency near stall: nearly all of the chemical energy is converted to mechanical torque. Few man-made motors approach this efficiency at the nanoscale.

F₀F₁ is a fully reversible motor. Direction is determined by the balance between the PMF and the substrate free energy \(\Delta G_{\text{ATP}}\):

In ischemia, the PMF collapses rapidly. ATP synthase runs backward, hydrolyzing cytosolic ATP to pump protons and preserve the membrane potential. This wastes ATP rapidly—the cell must shut down the enzyme.

The natural inhibitor IF1(inhibitor factor 1) does this. IF1 is a ~10 kDa protein that binds the F₁head in its ATPase mode and blocks γ-shaft rotation. It is pH-sensitive: active (dimeric) at matrix pH < 6.8, inactive (tetrameric) at pH > 7.5. During ischemia, matrix acidification activates IF1 and stops futile ATP hydrolysis. Under healthy conditions (alkaline matrix), IF1 is disengaged and synthesis proceeds.

Pharmacologically, oligomycin binds the c-ring at F₀ and blocks proton translocation in both directions. Oligomycin-sensitive respiration is a gold-standard metric of the coupling state of mitochondria (the Seahorse extracellular flux analyzer exploits this).

The free-energy cost per ATP synthesized (under physiological conditions with [ATP] ~ 10 mM, [ADP] ~ 100 µM, [Pi] ~ 1 mM) is \(\Delta G_{\text{ATP}} \approx +50\) kJ/mol. The energy available from translocating\(n_H \approx 2.67\) protons at a PMF of 200 mV is\(2.67 \times 96485 \times 0.2 \approx 51.5\) kJ/mol—a close match that suggests ATP synthase operates close to equilibrium.

The enzyme’s near-equilibrium operation has deep consequences: small changes in the PMF or substrate ratios modulate ATP synthesis rates smoothly, and the enzyme can reverse direction instantaneously when conditions shift. This allows mitochondria to buffer short-term ATP demand transients without waiting for gene-expression responses.

ATP synthase does not sit randomly in the IMM. In dimers and rows along the cristae edges, it actively shapes the membrane. Each F₀F₁ dimer is bent by ~86° between monomers; packing dimers head-to-tail into rows forces the membrane to curve tightly.

Kühlbrandt’s group (Davies, Anselmi et al. 2012, PNAS) used cryo-ET to visualize these ATP-synthase rows along the tips of cristae: the enzyme is both a chemical catalyst and a structural scaffold. Subunits e, g, k (mammalian) mediate the dimer interface; loss of these subunits flattens cristae in yeast, producing massive reductions in respiratory capacity (Paumard et al. 2002).

The cristae-edge localization also concentrates the enzyme at the high-PMF end of the local membrane circuit, where protons pumped by complexes I/III/IV converge. This micro-domain organization is thought to enhance the local PMF seen by F₀F₁, increasing ATP synthesis yield.

Diseases that disrupt this assembly include:

• NARP syndrome / MILS (Leigh): mtDNA mutation T8993G in ATP6 (subunit a), disrupts the proton half-channel
• TMEM70 deficiency: assembly factor mutations, reduced F₀F₁ levels, Roma population founder variant
• ATPAF1, ATPAF2 assembly factor mutations: encephalocardiomyopathy

A human at basal metabolic rate consumes ~2000 kcal/day. Accounting for the ~40% efficiency of oxidation-to-ATP and the ~30 kJ/mol yield per ATP at cellular conditions, the total ATP turnover is

At any instant the body only holds ~50 g of ATP; the entire pool is cycled through approximately every minute. Each F₀F₁ spins ~100 times per second at typical respiration rates, contributing 300 ATP per second per enzyme. The human body contains perhaps 10¹⁹ individual F₀F₁ enzymes operating simultaneously.

Tissues differ dramatically in ATP demand:

• Heart: ~6 kg ATP/day (2% body mass, 10% of total ATP)
• Brain: ~8 kg ATP/day (2% body mass, 20% of oxygen)
• Resting skeletal muscle: ~15 kg/day; during exercise this can rise 100×
• Kidney: ~3 kg/day, for active Na⁺/K⁺ transport

Kucharczyk et al. (2009) reviewed inherited human ATP synthase diseases, which fall into several categories:

• NARP / Leigh syndrome from T8993G/C in mtDNA ATP6; threshold-dependent heteroplasmy.
• MT-ATP8 mutations (small subunit a): cardiomyopathy, ataxia.
• TMEM70 deficiency: nuclear-encoded assembly factor, neonatal encephalocardiomyopathy.
• ATP5F1A/F1B mutations: rare severe mitochondrial disease.

Senior (2002, J. Bioenerg. Biomembr.) reviewed the F₁-ATPase kinetic literature; the enzyme has been so central to 20th-century biochemistry that its mechanism is the most detailed of any molecular motor.

Therapeutic targeting of F₀F₁ has been considered in two opposite directions:

• Inhibitors for cancer: cancer cells can become addicted to mitochondrial ATP; bedaquiline (a TB drug) targets mycobacterial F₀F₁ and is explored for oncology.
• Activators / uncoupler-like: low-dose DNP analogs and UCP1-targeting compounds (metabolic diseases, fatty liver).

Oligomycin A is the classic F₀inhibitor: it binds the c-ring at a site that blocks proton translocation. The name derives from Streptomyces oligomycin—another gift from the fungal chemical arsenal that turned out to be essential for bioenergetics research.