The Rotary Enzyme of the Cell: The Rotation of F1-ATPase

See allHide authors and affiliations

Science  04 Dec 1998:
Vol. 282, Issue 5395, pp. 1844-1845
DOI: 10.1126/science.282.5395.1844

Is there really a rotating enzyme within the cell? Every day, our body synthesizes approximately its own weight in ATP, which acts as an energy source for cells. ATP synthase (1), the enzyme responsible for synthesizing most of this ATP, is located on the inner membrane of the mitochondria. This common enzyme is also found in the plasma membrane of bacteria and the thylakoid membrane in chloroplasts. For ATP synthesis, it utilizes a proton electrochemical potential generated by biomembranes containing respiratory chains and photochemical systems. ATP hydrolysis in turn generates this potential when the ATP synthase pumps protons in the opposite direction. Its reverse reaction, proton transport, is accompanied by ATP hydrolysis. ATP synthesis and degradation are performed primarily by the portion of the enzyme exposed on the membrane (termed F1-ATPase); alone, F1-ATPase exhibits high ATP hydrolytic activity. Proton translocation is made possible by the F0 portion of the enzyme, which crosses the membrane. Electron microscopy (2) has shown that the catalytic site of F1-ATPase is separated by approximately 80 Å from the F0 proton-transport region, and the two are in contact with the F1-ATPase γ subunit. Energy transformation between ATP synthesis-hydrolysis and proton translocation is thus communicated by the γ subunit. Therefore, analysis of structural changes in the γ subunit can illuminate the mechanism of energy transmission for the enzyme.

Visualizing rotation

Experimental system for the observation of the rotation of the γ subunit in F1-ATPase (α3β3γ subcomplex). The His-tagged F1-ATPase was immobilized, and fluorescent actin filaments were attached to the subcomplex through streptavidin for observation with an epifluorescence microscope. When ATP was infused into the chamber, the actin filaments always rotated counterclockwise. In the presence of low concentrations of ATP, they rotated in discrete 120° steps.

On the basis of extensive kinetic analyses from a large number of researchers, Boyer proposed a model for the catalytic mechanism of F1-ATPase. Among other properties, this “binding-change mechanism” predicted that the energy was transmitted through rotation of the γ subunit in the center of the F1-ATPase molecule (3). Boyer's model was almost entirely substantiated by experimental results, with the exception of the rotation hypothesis. In 1994, Walker and colleagues reported the (at the time biggest) x-ray crystal structure of F1-ATPase (4), which strongly indicated that the enzyme acts as a “motor.” The structure that they described was a ring formation consisting of three α subunits alternating with three β subunits having catalytic sites. The rod-shaped γ subunit spanned the center of the ring. This x-ray crystal structure provided a specific and easily understood image that permitted rotational movement of the γ subunit, a concept that had initially been considered unconventional. Does this enzyme really rotate? At the time, I remained unconvinced.

Unchanged Secondary Structure of γ

What might be another possiblity besides rotational motion? Because Jagendorf and colleagues reported the use of a tritium exchange reaction to study secondary structural changes within F1-ATPase (5), I first focused on the “coiled-coil” configuration by which the two longest helices are wrapped in the γ subunit, as could be seen in the x-ray crystal structure. I took this approach because in recent years there have been reports of major changes in the coiled-coil configuration of proteins. Perhaps this structure of the γ subunit changed greatly during the process of the catalytic reaction. To test this hypothesis, I determined whether the stability of the helixes within F1-ATPase could be assessed by an amide proton exchange reaction (6). After labeling isolated γ subunits with tritium, I added α and β subunits to reconstitute the molecule. Subsequently, I compared the amount of tritium remaining in the γ subunit before and after hydrolysis. However, no difference in tritium content before and after the catalytic reaction could be found. In other words, the γ subunit did not undergo major secondary structural changes during ATP hydrolysis. So was the rotation yet the correct model? I could think of no other that was consistent with the x-ray crystal structure image.

Observation of Rotational Motion

Most biochemical experiments are based on the observation of the behavior of many molecules. In case of the detection of the rotation, however, it is impossible to detect rotation from the average behavior of all molecules. Nevertheless, by using some very elaborate methods, several groups reported results supporting the rotation (7, 8). Simply showing that the γ subunit moved, however, did not prove that it actually rotated. It was a case of “seeing is believing.” Only the direct observation of the single rotating F1-ATPase molecule could provide this evidence. Since the rotation of the γ subunit itself was far too small to be detected directly, I decided to attach a giant probe (fluorescently labeled actin filaments) to the portion of the γ subunit protruding on the F0 side of the membrane. In addition, to eliminate rotational Brownian movement throughout the whole molecule, I introduced a histidine tag to the amino-terminal end of the three β subunits opposite to the probe, and fixed the entire molecule to a glass substrate coated with Ni-nitrilloacetic acid (NI-NTA). After preparing this experimental system, I observed the motion of the actin filaments using a fluorescent microscope. The actin filaments attached to the F1-ATPase only rotated in the presence of Mg-ATP (9). Moreover, the direction of rotation was without exception counterclockwise, and continued for at least 10 min. This was not Brownian movement, but rotation of the γ subunit. In addition, the direction of rotation corresponded to the direction that was predicted from the x-ray crystal structure. Specifically, the rotational movement I observed provided evidence that the three catalytic sites in the three β subunits catalyze synchronously and also that the x-ray crystal structure indeed corresponds to a reaction intermediate state.

By observing this rotation at extremely low ATP concentrations, I found that the rotation occurs in increments of 120°, one step per molecule of ATP hydrolyzed (10). The orientation of the γ subunit changes with each ATP hydrolysis, occurring sequentially at β subunits positioned 120° apart on the molecular ring.

In these measurements, the actin filaments attached to the γ subunit rotated against the viscous resistance of water. The amount of rotational torque required for this movement could be estimated from the length of the actin filaments and the rate of rotation to provide a mean value of approximately 40 pN nm. For a single step of 120° and, hence, a single ATP hydrolysis, this amount to work of 80 pN nm. This value is in close agreement with those calculated for the release of free energy from the hydrolysis of a single ATP molecule under physiological conditions. The addition of ADP and Pi caused a shift in the free energy change for a single ATP molecule to 110 or 90 pN nm, but the work required for 120° rotation remained unchanged. These findings clearly indicate that the energy conversion efficiency for this enzyme is extremely high. This may be a necessary characteristic that allows the reverse reaction to occur.

The results of these experiments provided direct proof of Boyer's rotational model, which had once been considered unconventional. My findings show this enzyme to be a newly recognized motor protein with a conformation very similar to that of the portions of other proteins, such as myosin, which are also involved in structural changes accompanying ATP hydrolysis. It seems probable that these motor proteins share the same fundamental energy conversion mechanism (11). I hope to continue my research in comparing these motor proteins that convert chemical energy to mechanical energy and vice versa. In the long run, this might deepen our understanding of how specific properties could be acquired starting from a common principle, as it is seen so often throughout the biosphere.


View Abstract

Navigate This Article