The crystal structure has been determined of the F 1 -catalytic domain of the F-ATPase from Caldalkalibacillus thermarum, which hydrolyzes adenosine triphosphate (ATP) poorly. It is very similar to those of active mitochondrial and bacterial F 1 -ATPases. In the F-ATPase from Geobacillus stearothermophilus, conformational changes in the e-subunit are influenced by intracellular ATP concentration and membrane potential. When ATP is plentiful, the e-subunit assumes a "down" state, with an ATP molecule bound to its two C-terminal α-helices; when ATP is scarce, the α-helices are proposed to inhibit ATP hydrolysis by assuming an "up" state, where the α-helices, devoid of ATP, enter the α 3 β 3 -catalytic region. However, in the Escherichia coli enzyme, there is no evidence that such ATP binding to the e-subunit is mechanistically important for modulating the enzyme's hydrolytic activity. In the structure of the F 1 -ATPase from C. thermarum, ATP and a magnesium ion are bound to the α-helices in the down state. In a form with a mutated e-subunit unable to bind ATP, the enzyme remains inactive and the e-subunit is down. Therefore, neither the γ-subunit nor the regulatory ATP bound to the e-subunit is involved in the inhibitory mechanism of this particular enzyme. The structure of the α 3 β 3 -catalytic domain is likewise closely similar to those of active F 1 -ATPases. However, although the β E -catalytic site is in the usual "open" conformation, it is occupied by the unique combination of an ADP molecule with no magnesium ion and a phosphate ion. These bound hydrolytic products are likely to be the basis of inhibition of ATP hydrolysis.