The dogma that the cytoskeleton — the filamentous microtubule- and actin-based architecture of all our cells — evolved in eukaryotes has finally been overturned. In recent years, the microtubule-based cytoskeleton was traced back to bacteria with the discovery that FtsZ, a bacterial homologue of tubulin, forms filaments in the cell that mediate cell division. But the quest for the origin of its actin counterpart has proved more elusive. Now, the search has come to an end as two papers from Cell and Nature uncover the prokaryotic ancestor of actin — MreB.

There are several actin-like factors in bacteria, but none so far have been found to have related functions. MreB and Mbl are two such proteins from Bacillus subtilis that mediate cell shape, with the loss of these genes causing the normally rod-shaped cells to become round or twisted. Earlier this year in Cell, Jones and colleagues set out to characterize their function. When they looked at the subcellular localization, they saw that MreB and Mbl appeared to form filamentous helical bands around the cortex of the cell surface. By searching the bacterial genome databases, they found that MreB is present in non-spherical bacteria only. From these findings, they proposed that MreB-like factors form filamentous actin-like structures in non-spherical bacteria that determine cell shape.

But if MreB is a bona fide homologue of actin, can it self-assemble to form filaments? Reporting in Nature, van den Ent and colleagues show that it can. They found that Thermotaga maritima MreB can form polymers in vitro, and when they used electron microscopy to take a closer look, they found that these polymers form thin 'protofilaments' that line up in pairs or parallel arrays. Importantly, the longitudinal spacing of these subunits was 51 Å, similar to the 55-Å spacing seen between eukaryotic actin subunits. The final crunch came when they determined the atomic crystal structure of MreB — this showed that MreB and actin are highly similar in their three-dimensional atomic structure. Moreover, the MreB proteins self-assembled into protofilaments, which allowed a first glimpse at the interface between subunits.

One revelation to come from this work is that MreB does not tend to form helical protofilaments like eukaryotic actin but is straighter, contesting the long-held belief that the natural shape of actin is helical. Together, these papers have provided the missing link needed to establish the prokaryotic origin of our actin cytoskeleton. But many questions have now been raised. How does MreB (and related factors such as Mbl) control cell shape? Bacterial motor proteins have not yet been found and so, at least for now, we must envisage another model by which MreB regulates cell shape, perhaps one in which MreB filaments exert force on the cell membrane.