The Drosophila centriole – conversion of doublets into triplets within the stem cell niche

Centrioles are small cylindrical organelles that play the dual role of centrosome organizers and axoneme constructors in cilia and flagella of vertebrate and invertebrate cells (Bettencourt-Dias and Glover, 2007). Dysfunctions in these organelles are associated with a wide range of inherited diseases and cancer (Nigg and Raff, 2009; Bettencourt-Dias et al., 2011). Vertebrate cells usually begin interphase with a centrosome that contains one pair of centrioles that have distinctive morphologies and different abilities to nucleate microtubules (Bornens and Gönczy, 2014). The mother centriole recruits more centrosomal materials and displays characteristic distal appendages involved in microtubule binding and membrane docking. Only the mother centriole can assemble the primary cilium in non-dividing cells (Nigg and Stearns, 2011). The daughter centriole will acquire the distal appendages at the end of interphase and the ability to organize a primary cilium a cell cycle later. Parent centrioles in Drosophila lack a distinct structural dimorphism like that observed in vertebrate centrioles (Callaini et al., 1997). They can only be recognized by specific markers such as centrobin and asterless or by the accumulation to the mother centriole of some centrosomal proteins (Reina and Gonzalez, 2014).

Usually the centriole structure remains constant within different tissues of the same organism. Although certain Drosophila tissues present centrioles with doublets, whereas others show triplets, it is actually unclear if these structural differences are correlated with different centriolar lineages. Here, we show that somatic centrioles are usually composed by microtubule doublets whereas triplets characterize the centrioles of the germ cell line. We also found that the old mother centriole, which stays in the apical cytoplasm of the male GSC (Yamashita et al., 2007), is invariably composed of triplets, whereas its daughter is always built from mixed doublets and triplets. This difference represents the first documentation of a structural asymmetry between mother and daughter centrioles in Drosophila GSCs and might reflect a correlation between the architecture of the parent centrioles and the unequal distribution of centrosomal proteins. Moreover, the mother centriole is bound to the cell membrane by discrete links that might help to maintain its apical positioning during centrosome separation.

Although the structural features of the animal centriole have been carefully investigated (Winey and O'Toole, 2014), the available evidence regarding the proper architecture of the centrioles in the early Drosophila syncytial embryo is rather contradictory (Gonzalez et al., 1998). Hence, we performed a detailed analysis and defined that mature embryonic centrioles were invariably built by nine doublet microtubules and a distinct cartwheel (Fig. 1A). We also found centrioles with nine singlet microtubules (Fig. 1B) often showing lateral hooks that might correspond to forming B-tubules. Therefore, these structures might represent intermediate stages in centriole assembly that will acquire the final architecture later in interphase. The overall morphology of the embryonic centrioles was conserved during further developmental stages (Fig. 1C,D). Centrioles of cultured cells also lack the C-tubule (Cunha-Ferreira et al., 2009). We never found structural differences between mother and daughter centrioles in the somatic cells examined. Parent centrioles could be only recognized by their reciprocal disposition: the daughter centriole is orthogonal to the proximal end of the mother centriole. By contrast, parent centrioles were unusually aligned in tandem and easily distinguishable in auditory and olfactory sensilla that contained sensory neurons (Fig. 1E). The distal centriole that nucleated the ciliary axoneme was built by nine doublet microtubules (Fig. 1F), whereas the proximal one was composed by singlets and one or two doublets (Fig. 1G). This suggests that the daughter centrioles remained incomplete during development of the sensory organ.

The centriole of the germline stem cell (GSC) daughter, the gonioblasts, had nine triplet microtubules (Fig. 1H–J). Sections of its distal end often displayed isolated A-tubules together with doublets and triplets (Fig. 1J), suggesting that tubule elongation is asynchronous. The B- and C-tubules are added sequentially and grow from the basal region of the centriole, unlike in cultured vertebrate cells in which the growth of the B- and C-tubules is bidirectional (Guichard et al., 2010). Triplet microtubules elongated further during the spermatogonial mitoses (Fig. 1K–M) and throughout prophase of the first meiosis (Fig. 1N,O) to reach their full length at the onset of the first prometaphase. We were unable to find ultrastructural details to distinguish parent centrioles among them. Rather, the ability of both the centrioles within the same pair to nucleate a ciliary axoneme points to the loss of daughter characteristics upon the acquisition of mother properties. Often parent centrioles were linked together by a dense fibrillar material (Fig. 1K,N) that might play a role in centriole engagement.

Why would the centriole architecture be different in somatic and germ line cells? It has been suggested that the evolutionary pressure applied to the ciliary and flagellar axoneme might affect the architecture of the basal bodies (Carvalho-Santos et al., 2011). Thus, the structural differences we observed might be explained by the special role of the germline centrioles to nucleate the cilium-like projections of the primary spermatocytes, which at the onset of spermatid differentiation elongate in the sperm axoneme (Gottardo et al., 2013; Basiri et al., 2014). However, the female germ cells also have centrioles with triplets (Mahowald and Strassheim, 1970) but lack ciliary projections. Moreover, the presence of the C-tubule seems to be dispensable in Drosophila for the proper assembly of an axoneme since the sensory cilia of auditory and olfactory sensilla are assembled from basal bodies composed of only doublets. However, when the C-tubule dynamics is defective in unc Drosophila mutants the growth of centrioles and axonemes is severely affected (Gottardo et al., 2013). Thus, at least during Drosophila male gametogenesis, the C-tubule seems to be essential for the proper dynamics of centrioles and to the correct assembly of the ciliary projections.

Given that the precursors of the germ line, the pole cells, have centrioles with doublet microtubules (data not shown), whereas the centrioles of the gonioblasts have triplets (Fig. 1I), the modification of the centriole architecture might occur within the stem cell niche (Fig. 2A,B). Thus, we analyzed the ultrastructure of the centrioles in interphase GSCs, when the centriole pairs are found near the hub boundary (Fig. 2C,D). We easily distinguished parent centrioles among them by their reciprocal position. In images showing the mother centriole in cross section, the daughter centriole was always visible in a longitudinal orientation (Fig. 2E–G), whereas images showing cross sections of the daughter never contained the mother centriole (Fig. 2H,I). This is because the daughter centriole is always orthogonal to the basal end of the mother. We found three classes of mother centrioles: centrioles with nine triplets (Fig. 2E; 15 out of 39 centrioles), centrioles with nine doublets (Fig. 2F; 5 out of 39 centrioles) and centrioles with mixed doublets and triplets (Fig. 2G; 19 out of 39 centrioles).

Centrosomes of different ages have been suggested to have different functions during the asymmetric divisions of neuroblasts (Conduit and Raff, 2010; Januschke et al., 2011), and male (Yamashita et al., 2007) and female (Salzmann et al., 2014) germ stem cells in Drosophila. Specifically in male gametogenesis, the mother centrosome remains in the GSC close to the hub boundary, whereas the daughter one is displaced into the differentiating gonioblast. These centrosomes are distinct from each other owing to their age and to their different ability to organize a functional microtubule-organizing center (MTOC), which might enable the mother centriole to maintain its apical position (Yamashita et al., 2007). Therefore, to define whether the different structural patterns of mother centrioles were correlated with their age, we took serial sections throughout the apical cytoplasm of the GSCs, the region where centriole duplication occurred. We expected one centriole pair in early interphase and two distinct centriole pairs in late interphase. We found that mothers with triplets were usually associated with fully elongated daughters that had a smaller diameter (Fig. 2E) likely due to the incomplete C-tubule. Serial sections failed to reveal additional centrioles in this region. Therefore, stem cells were at the beginning of interphase and the centriole pairs examined were inherited from the previous cell cycle. Mothers with doublets or mixed tubules were always associated with short procentrioles (Fig. 2F,G) and were found within the apical cytoplasm of the GSCs together with an additional centriole pair. Daughter centrioles never displayed a complete set of nine triplets when they resided in the apical side of the GSC, but had triplets mixed with doublets that held lateral hooks of various lengths (Fig. 2H). Procentrioles consisting of nine single A-tubules and a central tube were often found (Fig. 2I). Centrioles with triplets were found in early interphase and might represent the older mothers, whereas centrioles with mixed tubules are likely to be the daughters that are turned into new mothers at the beginning of centriole duplication. Our observations suggest that the daughter centrioles acquire the complete set of microtubule triplets during the migration to the opposite pole of the nucleus. By contrast, the newborn daughter associated with the stationary mother fully elongates, but seems to complete its microtubule wall during the next cell cycle (Fig. 2L).

The failure to complete the microtubule set does not prevent centriole duplication in Drosophila, unlike daughter centrioles in vertebrate cells, which cannot support the assembly of a new procentriole until completion of their maturation process. However, only centrioles with a full set of nine triplets organize functional centrosomes in Drosophila GSCs. Therefore, the acquisition of the C-tubule and the completion of the triplet set might represent an aspect of the maturation process of the male germline centrioles. This could explain why the mother centrosome supports the nucleation of a robust aster of microtubules that might enable its anchoring at the apical side of the GSC (Yamashita et al., 2007). In this respect, the distinct projections that link the old mother centriole to the cell membrane (Fig. 2J,K) might play an important role in keeping the centriole to the apical position.

Male GSCs are flanked in the niche by a pair of cyst stem cells (CySCs) (Fig. 3A). The asymmetric divisions of these cells were slightly uncoupled, so that each GSC daughter had his accompanying somatic cell before its division was completed (Fig. 2B; Fig. 3B). The centrioles in CySCs looked like those of the somatic centrioles, and were built from nine doublet microtubules with a distinct cartwheel (Fig. 3B, inset).

The centrioles of the hub cells (Fig. 3C) resembled those of GSCs because they were built by mixed doublets and triplets (Fig. 3D,E), although they lacked lateral hooks and a distinct cartwheel. Despite examining serial sections of different hub cells (n=21), we only found single centrioles and we never observed procentrioles. Therefore, we assessed whether the centriole number was reduced by staining larval, pupal and adult testes with antibodies against Drosophila Pericentrin-like protein (D-PLP) (Martinez-Campos et al., 2004) and γ-tubulin. Unlike GSCs and CySCs, which showed one or two spots, a few hub cells displayed a distinct staining (Fig. 3F,G) with these antibodies. The missing cartwheel, the incomplete triplets, the lack of daughters and the unsuccessful recruitment of γ-tubulin suggests that the hub centrioles are not functional. This agrees with the general view that the hub is composed of post-mitotic cells that are terminally differentiated. We, indeed, never found mitotic figures within the hub regions examined (n=87). However, the conversion of quiescent hub cells into functional CySCs has been recently reported (Hétié et al., 2014), suggesting that the hub cells might be able to rescue their mitotic activity. Although this possibility contrasts with the findings of abnormal centrioles, we cannot exclude the possibility that these unusual centrioles convert into functional ones or that new centrioles can assemble ex novo.

Oregon R flies were raised on standard Drosophila medium at 24°C. Testes were dissected from larvae, pupae and adults in PBS and fixed as previously reported (Riparbelli et al., 2009). We used the following antibodies: mouse anti-α-tubulin (1:100; Sigma-Aldrich), mouse anti-γ-tubulin GTU88 (1:100; Sigma-Aldrich); anti-D-PLP (1:1500; Rodrigues-Martins et al., 2007) and rabbit anti-Vasa (1:400; Lasko and Ashburner, 1988). After washing in PBS with 0.1% bovine serum albumin (BSA) the samples were incubated for 1 h at room temperature with the appropriate Alexa-Fluor-488- and -555-conjugated secondary antibodies. DNA was visualized by incubation for 3–4 min in Hoechst 33258 (Sigma-Aldrich). Samples were observed by using an Axio Imager Z1 (Carl Zeiss) microscope equipped with an AxioCam HR camera (Carl Zeiss). Grayscale digital images were collected separately and then pseudocolored and merged using Adobe Photoshop V.7.0 software (Adobe Systems).

Testes dissected from pupae were fixed overnight in 2.5% glutaraldehyde in PBS at 4°C and processed as described previously (Gottardo et al., 2013). Ultrathin sections were stained with uranyl acetate and lead citrate, and then observed with a Tecnai G2 Spirit EM (FEI) equipped with a Morada CCD camera (Olympus).

The authors would like to thank Paul Lasko (Department of Biology, McGill University, Canada) and Ana Rodrigues-Martins (Instituto Gulbenkian de Ciência, Portugal)  for generously providing the antibodies.