Mechanical Rotation of the c Subunit Oligomer in ATP Synthase (F0F1): Direct Observation

See allHide authors and affiliations

Science  26 Nov 1999:
Vol. 286, Issue 5445, pp. 1722-1724
DOI: 10.1126/science.286.5445.1722


F0F1, found in mitochondria or bacterial membranes, synthesizes adenosine 5′-triphosphate (ATP) coupling with an electrochemical proton gradient and also reversibly hydrolyzes ATP to form the gradient. An actin filament connected to a c subunit oligomer of F0 was able to rotate by using the energy of ATP hydrolysis. The rotary torque produced by the c subunit oligomer reached about 40 piconewton-nanometers, which is similar to that generated by the γ subunit in the F1 motor. These results suggest that the γ and c subunits rotate together during ATP hydrolysis and synthesis. Thus, coupled rotation may be essential for energy coupling between proton transport through F0 and ATP hydrolysis or synthesis in F1.

The proton-transporting ATP synthase, F0F1, consists of a catalytic sector, F1 or F1–adenosine triphosphatase (ATPase) (α3β3γ1δ1ɛ1), and a proton pathway, F0(a1b2c12) (1, 2). The crystal structure of the bovine α3β3γ complex indicates that the α and β subunits are arranged alternately around the NH2- and COOH-terminal α helices of the γ subunit (3). The isolated F1hydrolyzes ATP, followed by γ subunit rotation, which is driven by conformational changes of the catalytic subunits (4). The γ subunit rotation in F1 has been suggested by biochemical experiments (5) and has been observed directly as counterclockwise rotation of an actin filament connected to the γ subunit (6, 7).

The γ subunit rotation in F1 should be transmitted to the membrane sector, F0, in order to complete the ATP hydrolysis–dependent proton transport. The detailed underlying mechanism of the energy transmission between F0and the γ subunit remains unknown. If the c subunit oligomer rotates counterclockwise (the same direction as γ) in the membrane, the ATP hydrolysis–dependent γ subunit rotation could be connected mechanically to the F0 sector. In this regard, c subunit rotation has been proposed (2, 8). However, to the best of our knowledge, this possibility of energy coupling has not been studied.

We designed several experimental systems to examine this possibility. The γ and ɛ complex is shown to be a rotor (6–9) and the α, β, δ, a, and b complex is proposed to be a stator in F0F1 (8). Therefore, we fixed F1 α (or β) subunits on a glass surface to demonstrate the rotation of an actin filament connected to the F0 c subunit, or conversely, the c subunits were fixed and the rotation of α or β was examined. ATP-dependent rotation was only successfully observed with the system described below (10). Escherichia coliF0F1 was immobilized on a coverslip through a His tag linked to the NH2-terminus of each α subunit (Fig. 1). A c subunit Glu2 was replaced by cysteine and then biotinylated to bind streptavidin and a fluorescently labeled actin filament. The γ subunit cysteine residues were replaced with alanine (11) in order to avoid direct binding of the actin filament to this subunit. Thus, cysteine is present only in the c subunit of the presumed rotor complex of the engineered F0F1. Specific biotinylation of the c subunit in F0F1 was confirmed by protein immunoblotting with streptavidin (12).

Figure 1

Observation system for the c subunit rotation in F0F1 established in this study. The fluorescently labeled actin filament–biotin-streptavidin complex was connected to the cysteine residue introduced at position 2 of the c subunit. In this system, all cysteine residues in the γ subunit were replaced by alanine, and the ɛ subunit does not contain cysteine. Therefore, an actin filament cannot bind to the γ and ɛ subunits, which form a rotor with the c subunit (shown in this study). cGlu2Cys, cGlu2 → Cys2.

After the addition of Mg ATP (13), the actin filament that was connected to the c subunit rotated. This rotation required Triton X-100 (14). Similar to the γ subunit in the F1 sector, the c subunit rotated counterclockwise when viewed from the membrane side (15). Nuclear magnetic resonance structure and biochemical or genetic analyses suggested that 12 copies of the c subunit form a symmetrical cylinder in which the NH2-terminus including Glu2 faces the periplasmic surface of the central pore of ∼1 nm in diameter (16). Therefore, the actin filament may be connected in the vicinity of the central pore.

Video images of the filaments connected to the c subunits were processed through centroid analysis (13); thus, the time course of the rotation could be obtained (Fig. 2A). Filaments connected to the c subunit continued to rotate for up to 2 min after the addition of Mg ATP, whereas filaments connected to the γ subunit in F1 often rotated for more than 10 min (7, 12). The filaments connected to the c subunits ceased rotations abruptly and disappeared from the glass surface in all cases, possibly because of the dissociation of the F0 sector from the F1sector.

Figure 2

Effects of actin filament length on the rotation of a filament connected to the c subunit. (A) The rotations (rounds) of actin filaments (1.5, 2.2, 2.9, and 3.6 μm) were recorded in the presence of 5 mM Mg ATP. (B) Rotational rate versus length of the actin filament. Rotating filaments connected to the c subunit at one end were analyzed. Linear segments having R 2 values of >0.96, except those defined as pauses (18), were selected from traces with an expanded time scale (examples shown in Fig. 3A), and then rotational rates were calculated. The average values for the rotational rates (∼20 data points) are plotted with standard deviations (error bars) against filament length (solid circles). Frictional torque T was calculated with T = (4π/3)ωηL 3/[ln(L/2r) − 0.447], where ω is angular velocity; η is 10−3N·s m−2, the viscosity of the medium; L is the length of the actin filament; and r is 5 nm, the radius of the actin filament (24). The dotted line represents the calculated rotational rates of the filaments with a constant torque value of 40 pN·nm. For comparison, the rotational rates of the γ subunit in F1 are plotted (open circles). For the assay, 5 mM Mg ATP was used.

The rotational rates varied slightly during the video recording, when the scales for the rotation and assay time were expanded (see Fig. 3A, upper left trace, for an example). The rates of single molecules were obtained, and the average values with deviations were plotted against the filament length (Fig. 2B). The results indicate that the c subunit rotation generates an average frictional torque of 40 pN·nm, which is similar to the value obtained for the γ subunit in F1 (7). Thus, mechanical energy transmission from the γ subunit to the c oligomer occurs essentially with no energy loss.

Figure 3

Effect of venturicidin on the rotation of an actin filament connected to the c subunit of F0F1. Rotational movements of filaments were followed on a subsecond scale. During video recording for the rotating filaments, 20 μl of the reaction mixture containing 70 μM venturicidin (provided by R. H. Fillingame) was slowly (∼5 μl/s) introduced into the flow cell, and the movement was further recorded. The approximate volume of the flow cell was 10 μl. (A) Typical examples of the rotations before (left trace) and 2 s after (right trace) the addition of venturicidin. Upper traces indicate the filament (2.0 μm) connected to the c subunit [c(F0F1)], and lower traces indicate the filament (1.6 μm) connected to the γ subunit [γ(F1)]. Red lines show records obtained at a resolution of 33 ms; black lines show the same records passed through a nonlinear median filter of rank 5 (133-ms width) (25). Arrowheads (blue) pointing down and up indicate the beginning and end of pauses, respectively. (B) Increase of pauses after venturicidin addition. The rotations of five actin filaments (∼1.5 to ∼2.0 μm) connected to the F0F1 c or F1 γ subunit were recorded, and numbers of pauses (per rotation, or round) after and before the venturicidin addition were counted. Ratios of the event numbers (pauses after venturicidin/pauses before venturicidin) and final venturicidin concentrations (0, 7, and 70 μM) are shown. Error bars indicate standard deviations.

We examined whether venturicidin, an E. colimembrane F0F1 inhibitor (17), could affect the c subunit rotation. After the addition of the antibiotic, the rotations frequently paused and started again on a subsecond scale (Fig. 3A, upper traces). In contrast, the γ subunit rotation in F1 remained unchanged (Fig. 3A, lower traces). The effects of venturicidin on the rotations of c and γ subunits were statistically analyzed by counting the number of pauses (18). In the c subunit rotation, the pauses after the venturicidin addition increased fivefold in comparison with those before the addition, whereas the antibiotic had no effect on the γ subunit rotation (Fig. 3B). The inhibitory effect on the c subunit rotation was dependent on the concentration (19) (Fig. 3B), indicating that the antibiotic must be binding and then dissociating from the c subunit oligomer. These results are consistent with a previous suggestion that venturicidin binding sites are located around c subunit Asp61 (E. coli), which is essential for proton transport (20).

Our results indicate that the c subunit oligomer rotates with the γ subunit during ATP hydrolysis by F0F1. In the reverse direction, proton transport should drive rotation of the c subunit oligomer, which in turn would drive rotation of the γ subunit to promote ATP synthesis. Our study demonstrates that the mechanical rotation of the γ and c subunit complex is an essential feature for the energy coupling between proton transport through the F0sector and ATP hydrolysis or synthesis in the F1 sector. Analysis of a series of E. coli F0F1mutants (21), based on the progress of single molecule biomechanics (22), will contribute to the further understanding of the motor mechanism.

  • * Present address: Department of Molecular Physiology and Biological Physics, University of Virginia, Charlottesville, VA 22906–0011, USA.

  • To whom correspondence should be addressed. E-mail: m-futai{at}


View Abstract

Navigate This Article