Flies hooked on coffee? Well, Drosophila melanogaster does need cappuccino — for oocyte and embryonic polarity. It also requires spire . Both genes contribute to polarity by affecting the actin cytoskeleton. The cappuccino gene product, Capu, nucleates actin-filament formation through its formin homology (FH) domains, but Quinlan et al. now report that Spire (Spir) nucleates filaments by a totally new mechanism.

Spir contains four WASP-homology-2 (WH2) domains and a stretch of acidic residues, which indicated that it might activate the actin-nucleating Arp2/3 complex. But Spir induced the formation of filamentous actin clusters in the absence of Arp2/3. This was mediated by its N-terminal half, which contains the WH2 repeats and the acidic domain, and occurred by nucleating actin filaments de novo with growth from the barbed end of the filament.

Next, the authors compared the nucleation activities and mechanisms of Spir, Capu and Arp2/3. The FH1 and FH2 domains of Capu and the N terminus of Spir nucleated actin polymerization at similar rates — slower than that of Arp2/3. And whereas the Arp2/3 complex generated crosslinked filaments, filaments that were formed by Spir and Capu weren't crosslinked. But, similar to Arp2/3, the N terminus of Spir capped the pointed ends of filaments.

Further investigation showed that each WH2 domain and the sequences linking them have varying roles in actin nucleation. Mutating all four WH2 domains didn't completely abolish nucleation activity, and Quinlan et al. found that linker-3 (L-3) showed weak nucleation activity, which they propose stabilizes two actin monomers to promote actin-dimer formation.

The authors then tested their hypothesis that, to form a new filament, Spir binds several actin monomers before assembling them into a filament nucleus. Rod-like Spir–actin complexes of four actin monomers aligned along their length were seen, consistent with WH2 domains binding and aligning actin monomers end to end.

Because the last two WH2 domains connected by L-3 seemed to comprise the functional core of Spir, the authors propose that this region mediates the formation of an actin dimer — the main kinetic hurdle to nucleation. A third and fourth monomer are then added by the first two WH2 domains. Spir might then stack monomers to form a nascent filament, and rapid filament elongation from the stable nucleus follows. This brings the number of classes of actin-nucleation factor to three.