Centrioles and centrosomes have an important role in animal cell organization, but it is uncertain to what extent they are essential for animal development. The Drosophila protein DSas-4 is related to the human microcephaly protein CenpJ and the C. elegans centriolar protein Sas-4. We show that DSas-4 is essential for centriole replication in flies. DSas-4 mutants start to lose centrioles during embryonic development, and, by third-instar larval stages, no centrioles or centrosomes are detectable. Mitotic spindle assembly is slow in mutant cells, and approximately 30% of the asymmetric divisions of larval neuroblasts are abnormal. Nevertheless, mutant flies develop with near normal timing into morphologically normal adults. These flies, however, have no cilia or flagella and die shortly after birth because their sensory neurons lack cilia. Thus, centrioles are essential for the formation of centrosomes, cilia, and flagella, but, remarkably, they are not essential for most aspects of Drosophila development.

Centrosome amplification is a common feature of many cancer cells, and it has been previously proposed that centrosome amplification can drive genetic instability and so tumorigenesis. To test this hypothesis, we generated Drosophila lines that have extra centrosomes in approximately 60% of their somatic cells. Many cells with extra centrosomes initially form multipolar spindles, but these spindles ultimately become bipolar. This requires a delay in mitosis that is mediated by the spindle assembly checkpoint (SAC). As a result of this delay, there is no dramatic increase in genetic instability in flies with extra centrosomes, and these flies maintain a stable diploid genome over many generations. The asymmetric division of the larval neural stem cells, however, is compromised in the presence of extra centrosomes, and larval brain cells with extra centrosomes can generate metastatic tumors when transplanted into the abdomens of wild-type hosts. Thus, centrosome amplification can initiate tumorigenesis in flies.