A multiscale approach reveals the molecular architecture of the autoinhibited kinesin KIF5A

Kinesin-1 is a microtubule motor that transports cellular cargo along microtubules. KIF5A is one of three kinesin-1 isoforms in humans, all of which are autoinhibited by an interaction between the motor and an IAK motif in the proximal region of the C-terminal tail. The C-terminal tail of KIF5A is ∼80 residues longer than the other two kinesin-1 isoforms (KIF5B and KIF5C) and it is unclear if it contributes to autoinhibition. Mutations in KIF5A cause neuronal diseases and could affect autoinhibition, as reported for a mutation that skips exon 27, altering its C-terminal sequence. Here, we combined negative-stain electron microscopy, crosslinking mass spectrometry (XL-MS) and AlphaFold2 structure prediction to determine the molecular architecture of the full-length autoinhibited KIF5A homodimer, in the absence of light chains. We show that KIF5A forms a compact, bent conformation, through a bend between coiled-coils 2 and 3, around P687. XL-MS of WT KIF5A revealed extensive interactions between residues in the motor, between coiled-coil 1 and the motor, between coiled-coils 1 and 2, with coiled-coils 3 and 4, and the proximal region of the C-terminal tail and the motor in the autoinhibited state, but not between the distal C-terminal region and the rest of the molecule. While negative-stain electron microscopy of exon-27 KIF5A splice mutant showed the presence of autoinhibited molecules, XL-MS analysis suggested that its autoinhibited state is more labile. Our model offers a conceptual framework for understanding how mutations within the motor and stalk domain may affect motor activity.