Fig. 4. The kinesin-13 ATPase cycle. Kinesin-13 motors use ATP to depolymerise MTs and their ATPase cycle is controlled by whether depolymerisation can (at MT ends) or cannot (on the MT lattice) occur. The MT end is thought to have distinct structural and/or conformational properties that allow it to be recognised by kinesin-13 motors and +TIPs. (It is these distinct properties that are believed to be shared, to some extent, by individual tubulin heterodimers and explains why the ATPase activity of kinesin-13 motors can be stimulated by them.) In the diagram, terminal tubulins are depicted with dotted outlines. On the left side, the behaviour of a kinesin-13 motor core is illustrated. The nucleotide state of this motor core in solution is not known but when it binds the MT lattice, ADP is released; a conformational change that presumably corresponds to this release step is observed in cryo-EM maps (Moores et al., 2003). However, the ATPase activity of the motor is inhibited by the lattice and no further conformational changes are seen, which supports the idea that the ATPase cycle of the motor is blocked on the lattice prior to ATP binding (Moores et al., 2003). At the MT end, however, the motor ATPase activity is stimulated, which suggests that ATP can bind. The ATP binding step (mimicked by AMPPNP) is the only point at which bent tubulin intermediates, representing the active deformation of the terminal tubulins by the kinesin motor, are observed (Moores et al., 2002). By contrast, dimeric kinesin-13 (on the right) in solution contains ADP.Pi and undergoes 1D diffusion along the lattice in this nucleotide state; ATP is not required for this process (Helenius et al., 2006). Once at the MT end, because ATP is required for depolymerisation, presumably ADP and Pi must be lost from at least one motor domain before ATP can bind and depolymerisation can occur. Presumably, there is a similar coupling between ATP binding by the dimer and bending of the terminal tubulin (Desai et al., 1999).