From a new Science paper:
Centrioles duplicate once in each cell division cycle through so-called templated or canonical duplication. SAK, also called PLK4 (SAK/PLK4), a kinase implicated in tumor development, is an upstream regulator of canonical biogenesis necessary for centriole formation. We found that overexpression of SAK/PLK4 could induce amplification of centrioles in Drosophila embryos and their de novo formation in unfertilized eggs. Both processes required the activity of DSAS-6 and DSAS-4, two molecules required for canonical duplication. Thus, centriole biogenesis is a template-free self-assembly process triggered and regulated by molecules that ordinarily associate with the existing centriole. The mother centriole is not a bona fide template but a platform for a set of regulatory molecules that catalyzes and regulates daughter centriole assembly.
First it was shown that flies don't need centrosomes for development (they die soon after tey are born), now it seems like centrioles can be made de novo - in other words from scratch. Yes, the mother centriole does not act as a template to form a daughter centriole (i.e. semi-conservative replication of centrosomes) ... so either:
1) Centrioles are not that important and can be made at will
2) Flies are just weird when it comes to centrioles. Other organisms may still have semi-conservative replication of centrioles and may still require centrioles to progress through the cell cycle. Centrioles may have features (such as centriolar RNA) that must be duplicated and these features were lost in some eukaryotic lineages.
So is the theory of semi-conservative replication of centrioles dead?
What is the phylogenetic distribution of centrioles? Does it match that of cilia / flagella?
Wow, that's cool! Do we have any indications in any other group (I mean, we know Drosophila are weird in other ways too)? I'd love to see some vert work on this.