Kinesins are a family of microtubule-interacting motor proteins that convert the chemical energy from ATP hydrolysis into mechanical work. Many kinesins are motile, walking along microtubules to fulfill different functions. Most kinesins are dimers, the monomer comprising a motor domain, a dimerizing stalk domain, and a tail domain. The motor domain contains both the nucleotide-binding site and the microtubule-binding site. I am interested in the molecular mechanism of kinesin's motility. In particular I want to establish the structural variations of the kinesin motor domain along with the mechanochemical cycle of this motor protein. During my thesis, I have focused my work on the human kinesin-1, also named conventional kinesin, which is the best characterized kinesin.I have studied two aspects of the kinesin mechanochemical cycle, by combining structural and mutational approaches. Both aspects rely on the binding of ADP-kinesin to a microtubule, which leads to the release of the nucleotide and to a tight kinesin-microtubule association. First I determined the crystal structure of nucleotide-free kinesin-1 motor domain in complex with a tubulin heterodimer, which is the building block of microtubule. This structure represented the main missing piece of the structural cycle of kinesin. Three subdomains in the kinesin motor domain can be identified through the comparison of my structure with ATP-analog kinesin-1-tubulin structure. The relative movements of these subdomains explain how ATP binding to apo-kinesin bound to microtubule triggers the opening of a hydrophobic cavity, 28 Å distant from the nucleotide-binding site. This cavity accommodates the first residue of the “neck linker”, a short peptide that is C-terminal to the motor domain, allowing the neck linker to dock on the motor domain. The docking of the neck linker is proposed to trigger the mechanical step, i.e. the displacement of the cargo and the stepping of the dimeric kinesin. By studying mutants of the neck linker, I have shown that, reciprocally, this peptide locks kinesin in the ATP state, which is also the conformation efficient for ATP hydrolysis. Doing so, it prevents the motor domain from switching back to the apo-state. It prevents also an untimely hydrolysis of ATP, before the mechanical step has occurred. These features are required for movement and processivity.Second, these structural data also suggest how the binding of ADP-kinesin to tubulin enhances nucleotide release from kinesin. To further study this step of the kinesin cycle, I studied the effect of kinesin-1 mutations. These mutations were designed in isolated kinesin to mimic the state when kinesin is bound to a microtubule. I identified two groups of mutations leading to a high spontaneous ADP dissociation rate, suggesting that there are two ways to interfere with ADP binding. Then I determined the crystal structures of the apo form of two mutants as well as that of the nucleotide-depleted wild type kinesin. It showed that apo-kinesin adopts either and ADP-like conformation or a tubulin-bound apo-like one. In the natural context, the second one is stabilized upon microtubule binding. Overall, the mutational and structural data suggest that microtubules accelerate ADP dissociation in kinesin by two main paths, by interfering with magnesium binding and by destabilizing the nucleotide-binding P-loop motif.