complex by crosslinking the y subunit to one of the a or P subunits (60, 61) and recovery

after photobleaching experiments (62, 63). More convincing evidence was provided

when Duncan et al. crosslinked the y subunit to an unlabeled P subunit by disulfide bond

and then mixed the y-P complex with 35S-labeled P subunit (along with the a, 6 and a

subunits). When the disulfide bond was broken and ATP was added, the y subunit was

observed to switch from labeled to unlabeled P subunit (64). Finally, direct evidence was

achieved in single molecule experiments by attachment of a fluorescent actin filament to

the y subunit and observance of unidirectional rotation of the actin filament upon addition

of ATP (15). Direct observation of the rotating y subunit was soon followed by

observance of the rotation of the s and c subunits at the same speed and direction,

indicating that these three subunits rotate in synchrony, forming the central rotary

machinery of the enzyme complex (65-67). Until very recently, rotation has only been

observed in the direction of ATP hydrolysis. Direct evidence for the synthesis of ATP by

Fl has been shown by attaching a magnetic bead to the y subunit ofF1 fixed to a glass

surface and the rotating the bead, in the appropriate direction, using electrical magnets

(68).

 The first structural information obtained for the a subunit ofE coli was

accomplished by nuclear magnetic resonance (NMR) studies (69) and is good agreement

with the crystal structure solved at 2.3 A (70). In all previous crystal structures of the

rotor stalk, the portion of the rotor stalk' s y subunit protruding from the Fl ap hexamer

and the a subunit were disordered. Recently, the structure of the bovine homologs of the

rotor stalk y and a subunits has been solved and refined to 2.4 A+ (48). The structures of

the E. coli y and a subunits are remarkably similar with that from bovine Fl. When