For many years, John Kilmartin and colleagues have painstakingly tracked, using electron microscopy (EM), the duplication of the SPB, which is budding yeast's version of the centrosome. The painstaking part comes in because only a few sections happen to catch the tiny SPB in the act. The SPB has a protrusion called the half bridge. At some point this half bridge doubles in size. Another SPB accumulates on the far end of this structure, then the two SPBs split apart. Before the splitting occurs, the far ends of the two half bridges are too close to be resolved by regular light microscopy.
The group now takes a look at two half bridge proteins: Sfi1p, and the Cdc31p centrin. Sfi1p has ∼20 repeats that are thought to bind centrin. Crystal structures of 2–3 Sfi1p repeats with bound centrin now reveal that Sfi1p forms a straight rod of α-helix with one centrin bound to each repeat. The centrins line up along the long axis of the Sfi1p but also coil around it and interact with each other end-on-end. The coiling means that side-to-side interactions between filaments are infrequent, perhaps providing space for bridge-nucleated microtubules to sneak through the structure and access the rest of the cell.
An Sfi1p-centrin complex with 15 repeats, examined by EM, is also a long rod. Extrapolating its length to the in vivo situation is consistent with the rod determining the length of the half bridge. Finally, the group shows that Sfi1p's NH2 terminus is at the SPB whereas the COOH terminus is at the far end of the half bridge (or the center of the structure once the half bridge duplicates).
Kilmartin suggests that the half bridge duplicates when an unknown signal causes two copies of Sfi1p to link to each other via their COOH termini. Then the NH2 terminus of the new Sfi1p nucleates the growth of the new SPB. The signals controlling these events are not known. The test of this model is to see whether an Sfi1p with fewer repeats will make SPBs with shorter half bridges.