Centrioles can form de novo.

Omnis centriolus e centriolo—every centriole comes from a centriole. This statement underlies the standard model of centrosome duplication, in which daughter centrioles that will form the foundation of a new centrosome must be patterned on some kind of template provided by a mother centriole. On page 1171, Khodjakov et al. now demonstrate that vertebrate cells are fully capable of assembling centrioles de novo, overturning this long-held belief and suggesting a different function for the mother centrioles.

A few specialized cell types, such as clam zygotes and rabbit blastomeres, previously have been shown to form centrioles de novo, but these examples were thought to be exceptions. The new work argues otherwise. The authors arrested CHO cells in S phase, and then destroyed their centrosomes with a laser. In this system, a loose cloud of pericentriolar material (PCM) forms, and within 24 h this PCM cloud becomes more compact, correlating with the appearance of new centrioles. The formation of the PCM clouds does not require microtubules, but centriole assembly does.

The de novo centriole synthesis occurred in every cell, indicating that it is very efficient, but the process produces a random number of centrioles—as many as 14 in one cell. This suggests that in normal cells, a mother centriole may act to limit the number of centrioles produced and prevent multipolar mitosis, rather than providing a necessary template for centriole formation. Preliminary analysis of other vertebrate cell lines suggests that the de novo pathway is a general mechanism, raising the possibility that it could be involved in cancerous transformation.

On page 1183, Dupuis-Williams et al. used a different model system to analyze the function of ε-tubulin, which specifically associates with centrosomes in human cells. The authors cloned the ε-tubulin gene of Paramecium, and found that it is required to stabilize the microtubule triplets that assemble into basal bodies in the protozoan. ▪