The power stroke pulling myosin along actin filaments during muscle contraction is achieved by a large rotation (Ϸ60°) of the myosin lever arm after ATP hydrolysis. Upon binding the next ATP, myosin dissociates from actin, but its ATPase site is still partially open and catalytically off. Myosin must then close and activate its ATPase site while returning the lever arm for the next power stroke. A mechanism for this coupling between the ATPase site and the distant lever arm is determined here by generating a continuous series of optimized intermediates between the crystallographic end-states of the recovery stroke. This yields a detailed structural model for communication between the catalytic and the force-generating regions that is consistent with experimental observations. The coupling is achieved by an amplifying cascade of conformational changes along the relay helix lying between the ATPase and the domain carrying the lever arm.chemo-mechanical coupling ͉ conformational transition ͉ Conjugate Peak Refinement ͉ muscle contraction ͉ power stroke T he myosin II head is a molecular motor that transforms chemical energy derived from the hydrolysis of ATP into mechanical work. The myosin head (or cross bridge) contains the catalytic activity, but the release of hydrolysis products is inhibited unless actin is bound. Lymn and Taylor (1) first proposed the cyclic scheme for the interaction between myosin and actin that produces motion (Fig. 1A). Underlying this cycle is myosin's ability to couple small changes in its catalytic ATPase site to large conformational changes in both the actin-binding and the distant force-generating domains with well defined communication mechanisms, which ensure that these changes are correlated so as to efficiently produce mechanical work. For instance, the communication mechanism between the ATP and actinbinding regions has been illuminated recently by crystal structures of the unconventional myosin V (2, 3). Here, we focus on one of the other essential communication pathways, the one for passing structural information between the ATPase site and the distant force-generating domain during the recovery stroke of the contractile cycle (i.e., going from states II to III in Fig. 1 A).The presence of that coupling first becomes apparent when comparing the crystallographic structures of the two end-states of the recovery-stroke ( Fig. 1 B and C) (4,5). As expected, the largest difference between these structures is in the orientation of the ''converter'' domain, which carries the lever arm and which is rotated by Ϸ60°relative to the rest of the head (referred to hereafter as the ''main body''). The other significant difference is in the ATP binding site, which is partially open before the recovery stroke, rendering the ATPase catalytically inactive (for example, see figure 6a in ref. 6). In contrast, after the recovery stroke, the ATP site is fully closed and the ␥-phosphate (␥P) group of the ATP forms an additional hydrogen bond with the amide of Gly-457 (Dictyostelium discoideum numberi...