Molecular motors are perhaps best known for their ability to transport cargo around inside cells. To do so, these tiny vehicles move along defined intracellular 'tracks'; motor proteins of the kinesin family, for instance, move on microtubule filaments. But some kinesins have a quite separate function, at least in cultured cells: they can break down microtubules into their constituent parts. Writing in Cell (114, 229–239; 2003), Noriko Homma et al. show that the kinesin KIF2A has this activity in vitro — and they propose that this function is essential in vivo for the brain to develop normally.

The authors started by generating mice that lack KIF2A, and found that the animals died within a day of birth with severe brain abnormalities. Delving deeper, Homma et al. discovered defects in the collateral branches of neurons. During development, nerve cells extend projections — axons — to form appropriate connections in the brain. Sometimes, branches also extend outwards from the axons to form further connections. These collateral branches generally remain short until the primary axon has found its way.

But that doesn't appear to happen in KIF2A-deficient mice. The authors took neurons from the hippocampus of normal and mutant mice and cultured the cells for two days. The normal neurons (left-hand image) extended a single primary axon, with some short collateral branches. The mutant nerves, by contrast, developed abnormally long collateral branches, which rebranched several times (right-hand image).

Video analysis showed that, in wild-type neurons, any collateral branches that formed actively shrank back. But in the mutant cells the branches continued growing. So KIF2A seems to be necessary to stop the branches lengthening. How does it do this? Homma et al. found that the protein can depolymerize microtubules within extending neuronal protrusions, and that this activity is reduced in KIF2A-deficient cells. Moreover, when the authors studied the polymerization and depolymerization of microtubules in another type of cultured brain cell (it's difficult to do this in neurons), they found that, if KIF2A was missing, microtubules carried on growing when they reached the edge of the cell. But in the presence of KIF2A, microtubules either stopped growing or showed cycles of growth and shrinkage.

It seems, then, that microtubule growth drives the protrusion of collateral branches — possibly by pushing out the edge of the cell — and that KIF2A normally keeps this process under control by depolymerizing microtubules when they reach the edge. It remains possible that KIF2A has an indirect role: perhaps it carries microtubule-depolymerizing molecules to the edge. But the protein's in vitro activity would seem to argue against this.