Part 2 · Chapter 2.4

ATP Synthase: F₀F₁ Rotary Motor

A nanoscale turbine that converts the proton-motive force into ATP. Two coupled rotary engines—the membrane-embedded F₀ and the matrix-facing F₁—synthesize three ATP per 360° rotation via Boyer’s binding-change mechanism. Directly visualized at the single-molecule level by Yasuda and Noji in 1998.

Learning Objectives

▸Describe the architecture of F₀F₁ ATP synthase: rotor vs stator elements
▸Explain Boyer’s binding-change mechanism (open-loose-tight)
▸Derive H⁺/ATP stoichiometry from c-ring size and calculate torque
▸Analyze reversibility and the threshold PMF for synthesis vs hydrolysis
▸Identify inhibitors and their pharmacological applications

◆Architecture of the Molecular Turbine

ATP synthase is built from two coupled rotary machines:

F₁ — Catalytic Head (Matrix)

Stoichiometry: \(\alpha_3\beta_3\gamma\delta\epsilon\)
Three \(\alpha\beta\) dimers form a hexameric ring with nucleotide-binding sites at the \(\alpha\beta\) interfaces
Central \(\gamma\)-subunit is asymmetric, penetrates the \(\alpha_3\beta_3\) ring, and drives conformational changes
Three ATP synthesized per full rotation

F₀ — Proton Channel (Membrane)

Stoichiometry: ab₂c₁₀₋₁₂ (species-dependent)
c-ring rotates as protons hop onto and off each c-subunit carboxylate
Subunit a contains two half-channels (entry from IMS, exit into matrix)
Stator (b₂δ) keeps the \(\alpha_3\beta_3\) hexamer fixed while \(\gamma\) rotates

\[ n\,\text{H}^+_{\text{IMS}} + \text{ADP} + \text{P}_i \;\rightleftharpoons\; \text{ATP} + \text{H}_2\text{O} + n\,\text{H}^+_{\text{matrix}} \]

In bovine mitochondria the c-ring has 8 subunits and F₁ has 3 catalytic sites, giving n = 8/3 ≈ 2.67 H⁺ per ATP. Chloroplast ATP synthase uses c₁₄ (4.67 H⁺/ATP), yeast uses c₁₀(3.33 H⁺/ATP). The number is not strictly universal.

◆Molecular Diagram

◆Boyer’s Binding-Change Mechanism

Paul Boyer (Nobel Prize 1997, with John Walker) proposed in the 1970s that ATP synthesis does not require energy to form the P–O bond—chemistry of bond formation is essentially spontaneous inside a dehydrated active site. Instead, energy is consumed to release the already-formed ATP from the enzyme. The three \(\beta\) subunits interconvert synchronously among three conformations:

Open (O)

No bound nucleotide. Substrate (ADP + Pᵢ) can enter.

Loose (L)

Binds ADP + Pᵢ weakly; catalytically inactive.

Tight (T)

Catalytically active site; spontaneously forms ATP from ADP + Pᵢ.

Each 120° rotation of \(\gamma\) pushes all three subunits through the cycle O → L → T → O. At any moment, each site is in a different state. Yasuda and colleagues (1998) visualized this rotation directly by attaching an actin filament to \(\gamma\) and watching it spin under fluorescence microscopy—proof of a nanoscale rotary motor.

\[ \text{Free energy per 120}^\circ \text{ step} = \frac{\Delta G_{\text{ATP}}^{\text{cell}}}{3} \approx \frac{-50\;\text{kJ/mol}}{3} \approx 83\;\text{pN}\cdot\text{nm per step} \]

◆Torque and Single-Molecule Measurements

Yasuda’s single-molecule experiments (and subsequent refinements by Itoh, Noji, and others) revealed that the \(\gamma\)-subunit rotates counter-clockwise (as viewed from the membrane) during synthesis, and that each 120° step is subdivided into 80° (ATP binding) and 40° (hydrolysis/release) substeps. Peak torque measured is \(\approx 40{-}50\;\text{pN}\cdot\text{nm}\), essentially at the theoretical maximum given PMF of ~200 mV and \(n = 2.67\) H⁺/ATP. The motor operates near 100% thermodynamic efficiency—a rare and remarkable biological achievement.

\[ \tau \cdot 2\pi/3 \;=\; n_{H+\text{per step}} \cdot F \cdot \Delta p \quad\Longrightarrow\quad \tau \approx \frac{3\,n_{H+}F\,\Delta p}{2\pi} \]

For \(n = 2.67\) H⁺/ATP and \(\Delta p = 200\) mV, \(\tau \approx 41\;\text{pN}\cdot\text{nm}\), in excellent agreement with experiment.

◆Simulation 1: Binding-Change Kinetic Scheme

This model tracks the three \(\beta\) subunits as they cycle through O/L/T states synchronized with \(\gamma\) rotation. Output: cumulative ATP production and the torque-angle profile observed in single-molecule experiments.

Python

script.py139 lines

import numpy as np
import matplotlib.pyplot as plt
from scipy.integrate import odeint

# Boyer's binding-change mechanism: kinetic simulation of the three beta subunits
# Each subunit cycles through three states: Open (O), Loose (L), Tight (T)
# At every 120 degree rotation of gamma, states cycle: O -> L -> T -> O ...
# ATP is synthesized when a site transitions L -> T (accepts ADP + Pi, squeezed into ATP)
# ATP is released when T -> O

# We simulate a 12-state Markov model capturing the three beta subunits synchronously
# Rotation rate depends on proton flux through F0

t = np.linspace(0, 2.0, 2000)  # 2 seconds of simulation
tau = 0.01    # residence time per state (ms scale -> rotation ~100 Hz)
omega = 2*np.pi/(3*tau)  # angular velocity (rad/s); full rotation in 3*tau

# Angle of gamma subunit
theta = (omega*t) % (2*np.pi)

# For each beta (offset by 120 degrees), determine state as function of theta
offsets = [0, 2*np.pi/3, 4*np.pi/3]
states = np.zeros((3, len(t)), dtype=int)

for i, offs in enumerate(offsets):
    # phase 0-120: Open, 120-240: Loose, 240-360: Tight
    phi = (theta - offs) % (2*np.pi)
    s = np.zeros_like(phi, dtype=int)
    s[(phi >= 0) & (phi < 2*np.pi/3)] = 0  # Open
    s[(phi >= 2*np.pi/3) & (phi < 4*np.pi/3)] = 1  # Loose
    s[(phi >= 4*np.pi/3)] = 2  # Tight
    states[i] = s

# ATP synthesis events: every time a site transitions from L -> T (i.e. state 1 -> 2)
atp_events = 0
atp_time = []
for i in range(3):
    transitions = np.where((states[i,:-1] == 1) & (states[i,1:] == 2))[0]
    for tr in transitions:
        atp_time.append(t[tr])
atp_time = np.sort(atp_time)
atp_cum = np.arange(1, len(atp_time)+1)

# Compute torque: 50 pN*nm is required per 120 deg step, 3 ATP per full revolution
# Energy per ATP = integral of torque d(theta) = tau_torque * 2*pi/3
# delta G of ATP = 50 kJ/mol * 1.66e-21 = ~83 pN*nm
# so torque ~ 40-50 pN*nm is consistent with single-molecule measurements (Yasuda 1998)

fig, axes = plt.subplots(2, 2, figsize=(14, 9))

# Panel 1: rotation of gamma
ax = axes[0,0]
ax.plot(t, np.degrees(theta), color='#fbbf24', lw=1.5)
ax.set_xlabel('Time (s)', color='white')
ax.set_ylabel('\u03B3 rotation angle (degrees, mod 360)', color='white')
ax.set_title('F1 central stalk (\u03B3) rotation',
             color='white', fontsize=12, fontweight='bold')
ax.set_facecolor('#0a0a1a'); ax.tick_params(colors='white')
ax.grid(alpha=0.2, color='#fbbf24')
for s in ax.spines.values(): s.set_color('#fbbf24')

# Panel 2: state of each beta subunit
ax = axes[0,1]
state_labels = ['Open', 'Loose', 'Tight']
colors_state = ['#f87171','#fde047','#86efac']
for i in range(3):
    # Shade regions by state
    ax.step(t, states[i] + i*3.2, where='post', color=colors_state[0], lw=1, alpha=0)
    for s_idx, lbl in enumerate(state_labels):
        mask = states[i] == s_idx
        ax.fill_between(t, i*3.2-0.5, i*3.2+0.5, where=mask,
                        color=colors_state[s_idx], alpha=0.8,
                        label=lbl if i == 0 else None)
    ax.text(-0.05, i*3.2, f'\u03B2{i+1}', color='white', fontsize=12,
            va='center', ha='right', fontweight='bold')
ax.set_xlabel('Time (s)', color='white')
ax.set_ylabel('')
ax.set_title('Binding-change mechanism: three \u03B2 subunits cycle O \u2192 L \u2192 T',
             color='white', fontsize=12, fontweight='bold')
ax.legend(facecolor='#0a0a1a', edgecolor='white', labelcolor='white', loc='upper right')
ax.set_ylim(-2, 9)
ax.set_yticks([])
ax.set_facecolor('#0a0a1a'); ax.tick_params(colors='white')
for s in ax.spines.values(): s.set_color('#fbbf24')

# Panel 3: cumulative ATP production
ax = axes[1,0]
ax.plot(atp_time, atp_cum, color='#86efac', lw=2.5, drawstyle='steps-post')
rate = len(atp_time)/t[-1]
ax.set_xlabel('Time (s)', color='white')
ax.set_ylabel('Cumulative ATP molecules', color='white')
ax.set_title(f'ATP output: {rate:.0f} ATP/s \u2248 {rate/3*60:.0f} rpm rotation',
             color='white', fontsize=12, fontweight='bold')
ax.set_facecolor('#0a0a1a'); ax.tick_params(colors='white')
ax.grid(alpha=0.2, color='#86efac')
for s in ax.spines.values(): s.set_color('#86efac')

# Panel 4: torque-angle curve
ax = axes[1,1]
# Simulate torque during a single 120-deg step
theta_step = np.linspace(0, 2*np.pi/3, 500)
T0 = 50  # pN*nm peak torque
torque = T0 * (1 - 0.3*np.cos(3*theta_step))  # periodic substructure of 3 substeps (Yasuda)
ax.plot(np.degrees(theta_step), torque, color='#67e8f9', lw=2.5)
ax.fill_between(np.degrees(theta_step), torque, 0, color='#67e8f9', alpha=0.2)
ax.axhline(y=T0, color='white', ls=':', alpha=0.4, label=f'Peak \u03C4 = {T0} pN\u00B7nm')
# integrate work per 120 step
work = np.trapezoid(torque, theta_step)   # pN*nm (per step)
ax.set_xlabel('\u03B3 rotation angle (degrees)', color='white')
ax.set_ylabel('Torque \u03C4 (pN\u00B7nm)', color='white')
ax.set_title(f'Torque profile (Yasuda 1998): work per 120\u00B0 step = {work:.0f} pN\u00B7nm',
             color='white', fontsize=12, fontweight='bold')
ax.legend(facecolor='#0a0a1a', edgecolor='#67e8f9', labelcolor='white')
ax.set_facecolor('#0a0a1a'); ax.tick_params(colors='white')
ax.grid(alpha=0.2, color='#67e8f9')
for s in ax.spines.values(): s.set_color('#67e8f9')

fig.suptitle('F1 ATP synthase: Boyer binding-change mechanism',
             color='white', fontsize=15, fontweight='bold')
fig.patch.set_facecolor('#0a0a1a')
plt.tight_layout()
plt.savefig('output.png', dpi=120, bbox_inches='tight', facecolor='#0a0a1a')

# Summary energetics
kB = 1.381e-23; T_K = 310
thermal = kB*T_K/1e-21     # pN*nm of thermal energy
atp_energy = 83            # pN*nm per ATP synthesis at cellular conditions
print(f"Work per 120 deg step: {work:.0f} pN*nm")
print(f"ATP hydrolysis free energy at cell: ~{atp_energy} pN*nm")
print(f"Coupling efficiency: {atp_energy/work*100:.0f}%")
print(f"Thermal fluctuation scale kBT: {thermal:.2f} pN*nm (~{work/thermal:.0f} kBT per step)")
print(f"Simulated ATP production rate: {rate:.0f} molecules/s")
print()
print("Boyer (Nobel Prize 1997) insight:")
print("  Three beta subunits rotate sequentially through O, L, T states.")
print("  ATP synthesis does NOT require energy to form the bond (L -> T).")
print("  Energy is needed to RELEASE ATP (T -> O), driven by gamma rotation.")

Click Run to execute the Python code

Code will be executed with Python 3 on the server

◆Reversibility: Synthase vs Hydrolase

F₀F₁ is inherently reversible. If the proton-motive force falls below a threshold set by the cellular \(\Delta G_{\text{ATP}}\), the enzyme operates backward—hydrolyzing ATP to pump protons out of the matrix. In ischemia this is disastrous: the collapsing PMF causes F₀F₁ to rapidly deplete cellular ATP. Mitochondria defend themselves by expressing IF₁, a small inhibitor protein that binds F₁ under low PMF and blocks hydrolysis.

Python

script.py79 lines

import numpy as np
import matplotlib.pyplot as plt

# Reversibility of F0F1: synthesis vs. hydrolysis mode as a function of Delta p
F  = 96485.0
R  = 8.314
T_K = 310.0
dG_ATP_standard = 30.5e3   # J/mol (standard hydrolysis, +30.5 kJ/mol for synthesis)
Q_cellular = 1e5           # [ATP]/([ADP][Pi]) ~ 10^5 in cells
dG_ATP_cell = dG_ATP_standard + R*T_K*np.log(Q_cellular)   # ~ 50 kJ/mol synthesis cost

# Number of protons per ATP: 8c ring / 3 ATP ~ 2.67 (bovine) or 10c/3 ~ 3.3 (yeast)
n_H = 2.67

# Delta p (mV) range
pmf_mv = np.linspace(-100, 300, 500)
pmf_J = pmf_mv * 1e-3 * F   # J/mol per H+

# Free energy of the synthesis reaction: n * Delta_Gp(inward H+) + Delta_G_ATP_synth
# Negative means synthesis is favored
dG_system = -n_H*pmf_J + dG_ATP_cell   # J/mol

# ATP synthesis rate proxy: sigmoidal switch around dG_system = 0
rate_synthesis = 100.0/(1 + np.exp(dG_system/1500))    # fit scale
rate_hydrolysis = 100.0/(1 + np.exp(-dG_system/1500))

# Threshold PMF for synthesis
pmf_thresh = dG_ATP_cell/(n_H*F) * 1e3   # mV

fig, axes = plt.subplots(1, 2, figsize=(14, 5.5))

ax = axes[0]
ax.plot(pmf_mv, dG_system/1000, color='#67e8f9', lw=2.5)
ax.fill_between(pmf_mv, dG_system/1000, 0,
                where=dG_system<0, color='#86efac', alpha=0.3, label='Synthesis favored')
ax.fill_between(pmf_mv, dG_system/1000, 0,
                where=dG_system>=0, color='#f87171', alpha=0.3, label='Hydrolysis favored')
ax.axhline(y=0, color='white', lw=0.5)
ax.axvline(x=pmf_thresh, color='#fbbf24', ls='--', alpha=0.7,
           label=f'Threshold \u0394p = {pmf_thresh:.0f} mV')
ax.set_xlabel('Proton-motive force \u0394p (mV)', color='white')
ax.set_ylabel('\u0394G of synthesis reaction (kJ/mol)', color='white')
ax.set_title('Thermodynamic reversibility of ATP synthase',
             color='white', fontsize=13, fontweight='bold')
ax.legend(facecolor='#0a0a1a', edgecolor='#67e8f9', labelcolor='white')
ax.set_facecolor('#0a0a1a'); ax.tick_params(colors='white')
ax.grid(alpha=0.2, color='#67e8f9')
for s in ax.spines.values(): s.set_color('#67e8f9')

ax = axes[1]
ax.plot(pmf_mv, rate_synthesis, color='#86efac', lw=2.5, label='Synthesis (CW rotation)')
ax.plot(pmf_mv, rate_hydrolysis, color='#f87171', lw=2.5, label='Hydrolysis (CCW rotation)')
ax.axvline(x=pmf_thresh, color='#fbbf24', ls='--', alpha=0.5)
ax.set_xlabel('\u0394p (mV)', color='white')
ax.set_ylabel('Relative rate (% Vmax)', color='white')
ax.set_title('Bidirectional activity vs. PMF',
             color='white', fontsize=13, fontweight='bold')
ax.legend(facecolor='#0a0a1a', edgecolor='#86efac', labelcolor='white')
ax.set_facecolor('#0a0a1a'); ax.tick_params(colors='white')
ax.grid(alpha=0.2, color='#86efac')
for s in ax.spines.values(): s.set_color('#86efac')

fig.patch.set_facecolor('#0a0a1a')
plt.tight_layout()
plt.savefig('output.png', dpi=120, bbox_inches='tight', facecolor='#0a0a1a')

# Area under synthesis rate curve (productive zone)
productive = pmf_mv > pmf_thresh
synth_area = np.trapezoid(rate_synthesis[productive], pmf_mv[productive])
print(f"Cellular \u0394G for ATP synthesis: {dG_ATP_cell/1000:.1f} kJ/mol")
print(f"Threshold PMF for net synthesis (n = {n_H} H+/ATP): {pmf_thresh:.1f} mV")
print(f"Physiological PMF ~200 mV -> comfortably above threshold")
print(f"Integrated synthesis capacity above threshold = {synth_area:.0f} (rate*mV)")
print()
print("Reversibility is physiologically important:")
print("  Anoxia -> PMF collapses -> F0F1 runs backward -> depletes ATP rapidly")
print("  Protective mechanism: IF1 inhibitor protein binds F1 under low PMF")

Click Run to execute the Python code

Code will be executed with Python 3 on the server

◆Coupling Efficiency and State Transitions

Britton Chance (1955) classified respiring mitochondria into four states by their oxygen consumption kinetics. State 3 (substrate + ADP + O₂) is active phosphorylation; State 4 (substrate + O₂ but no ADP) is coupled respiration with O₂ use rate limited by proton leak. The ratio respiratory control ratio (RCR) = State 3 / State 4 is a sensitive readout of coupling: well-coupled mitochondria give RCR > 5, while damaged or uncoupled mitochondria show RCR < 2. Seahorse extracellular flux analysis (XF) in living cells performs an analogous measurement using sequential additions of oligomycin (isolates leak), FCCP (maximal respiration), and rotenone + antimycin A (non-mitochondrial rate).

◆Adenine Nucleotide Translocase and Phosphate Transport

ATP synthesis in the matrix is useless unless ATP can reach the cytosol where most ATP consumers live. This is accomplished by two dedicated transporters that complete the “oxidative phosphorylation supercomplex” in an energetic sense:

ANT (SLC25A4/5/6)

Electrogenic exchange: one ATP⁻⁴ exported for one ADP⁻³ imported. The net +1 charge moves out with \(\Delta\psi\), consuming some of the PMF. ANT is the most abundant protein of the inner membrane and a target of atractyloside and bongkrekic acid.

Pᵢ carrier (SLC25A3)

Electroneutral symport of H\(_2\)PO\(_4^-\) with H⁺, consuming another proton per ATP. Together, ANT + PiC cost ~1 H⁺ per ATP exported—baked into the 4 H⁺/ATP stoichiometry.

◆Daily ATP Turnover: Numbers That Astonish

A resting adult hydrolyzes approximately 40–75 kg of ATP per day—more than body weight—but the steady-state pool of ATP is only ~100 g at any moment. Every ATP molecule is synthesized and hydrolyzed hundreds of times per day. During strenuous exercise, the rate can rise tenfold. A single F₀F₁ ATP synthase makes ~100 ATP/s; a mitochondrion may have 10⁴ synthases; a cell may have 10\(^3\)mitochondria—so a single cell synthesizes ~10⁹ ATP molecules per second. Across the body’s \(10^{14}\) cells, the total production rate is astronomical. The sheer scale of this turnover underscores why defects in any step of OxPhos produce profound disease, and why drugs that modulate it (metformin, bedaquiline, oligomycin) have such broad biological effects.

◆Inhibitors

Oligomycin

Binds the F₀ c-ring, blocks proton flow, halts both synthesis and hydrolysis. Diagnostic for mitochondrial respiration studies.

DCCD (dicyclohexylcarbodiimide)

Covalently modifies the essential carboxylate (Glu/Asp) of c-subunits, locking the rotor.

Aurovertin B

Binds F₁ \(\beta\) subunits; fluorescent, useful as a probe of conformational change.

Bedaquiline

Targets mycobacterial ATP synthase c-ring; approved anti-tuberculosis drug (Sirturo).

◆Structural Evolution of ATP Synthase

ATP synthase is among the most ancient and conserved enzymes in biology. Its basic \(\alpha_3\beta_3\gamma\) topology is found in all three domains of life, suggesting it predates the divergence of archaea, bacteria, and eukaryotes. Several variants illustrate evolutionary and engineering principles:

Mitochondrial F-type

c₈–c₁₀ ring, dimerizes in cristae to shape membrane curvature, couples PMF to ATP synthesis. Uses both \(\Delta\psi\)and \(\Delta\text{pH}\).

Chloroplast CF₀CF₁

c₁₄ ring; driven primarily by pH gradient (pH 4.5 in lumen, 8 in stroma). The steeper \(\Delta\text{pH}\) compensates for smaller \(\Delta\psi\).

V-type (vacuolar)

Runs in reverse: hydrolyzes ATP to pump protons into lysosomes/vacuoles/synaptic vesicles. Homologous to F-type but regulated differently (reversible assembly of V₀/V₁).

A-type (archaeal)

Intermediate features; often uses Na⁺ instead of H⁺ as coupling ion in some species adapted to alkaline environments.

◆Dimerization and Cristae Morphology

Mitochondrial ATP synthase assembles into dimers in the inner membrane that further polymerize along the highly curved ridges of cristae. Cryo-EM tomography (Kuhlbrandt 2011, 2015) showed these dimer rows are responsible for bending the membrane and creating the characteristic cristae architecture. Deletion of dimerization subunits (e.g., subunit e, g) collapses cristae, impairs OxPhos, and shortens lifespan in yeast. The positive curvature at the dimer interface locally concentrates protons, possibly enabling direct proton uptake without equilibration with the bulk IMS—a kinetic advantage.

◆Coupling of OxPhos: The P/O Ratio Revisited

The “P/O ratio” is the moles of ATP synthesized per \(\tfrac{1}{2}\)mole of O₂ consumed (one O atom reduced = one electron pair). Historically taught as 3 for NADH and 2 for FADH₂, but careful measurement accounting for proton stoichiometry gives 2.5 and 1.5:

\[ \text{P/O} = \frac{n_{\text{H}^+_{\text{pumped per e-pair}}}}{n_{\text{H}^+\text{ per ATP}}} \]

NADH: 10 H⁺ pumped (I: 4, III: 4, IV: 2) / 4 H⁺ per ATP (including one for Pᵢ/ATP antiport) = 2.5 ATP. FADH₂: 6 H⁺ / 4 = 1.5 ATP. These values are not integers because the c-ring is not stoichiometrically coupled to F₁ (eg. c₈ vs \(\beta_3\)).

Clinical Relevance

NARP Syndrome / Leigh Disease

MT-ATP6 mutations impair proton channel function; neuropathy, ataxia, retinitis pigmentosa, maternally inherited.

Ischemia-Reperfusion Injury

F₀F₁ runs in reverse, hydrolyzing ATP and pumping protons out; contributes to myocardial/brain damage after stroke or heart attack.

Mitochondrial Permeability Transition

Under Ca²⁺ overload, ATP synthase may form the mPTP pore, triggering apoptosis (cytochrome c release).

Cancer Targeting

Surface ATP synthase on cancer cells is a therapeutic target (e.g., angiostatin binds it, inhibits tumor angiogenesis).

←Prev: 2.3 Electron Transport Chain Next: 2.5 Metabolic Integration→