Mutants and transgenic fly constructs:

The generation of asl, ana1, sas-6, and bld10 fused with GFP or tdTomato was described previously (Blachon et al. 2008). SAS-4-GFP was done in a similar way. The sas-4 gene, including 2-kb upstream elements immediately adjacent to the predicted initiator methionine and up to the stop codon, was cloned into the p{UAST} vector using EcoRI and NotI. GFP was fused at the C terminus between NotI and XbaI. plk4^c06612, sas-6^c02901, sas-4^S2214, and cnn^HK21 mutant flies were obtained from the Bloomington Stock Center. bld10^c04199 and poc1^c06059 mutant flies were obtained from Exelixis collection (Thibault et al. 2004). Ana1^mecB mutant flies were identified in a genetic screen of a collection of ∼600 potential mechanosensory mutants (pmm) (Avidor-Reiss et al. 2004).

Tissue processing for electron microscopy:

EM analysis of macrochaetae and fly testis were done as previously described (Eberl et al. 2000; Avidor-Reiss et al. 2004).

Immunofluorescence staining and imaging:

Embryos aged between 0 and 3 hr were collected on grape agar plates. They were dechorionated and fixed according to Rothwell and Sullivan (2000). Whole-mount preparations of pupa or adult testis were imaged in PBS containing 1 μg/ml DAPI using phase-contrast light microscopy. For staining, dark pupa or adult testes were dissected in PBS and fixed 5 min in formaldehyde (3.7% in PBS). After squashing the tissue, the coverslip was removed using liquid nitrogen and slides were placed in methanol for 2 min. After washing in PBS, the preparations were permeabilized with PBS 0.1% TritonX-100 for 10 min and saturated with PBS, 1% BSA, and TritonX-100 0.1% for 45 min. Antibody staining with mouse anti-γ-tubulin (1:200; GTU88, Sigma) or with anti-Cnn (1:200; Megraw and Kaufman 2000) was performed for 1 hr at room temperature followed by three washes with PBS. Secondary antibody rhodamine goat anti-mouse or rhodamine goat anti-rabbit (1:200; Jackson Immunoresearch) was applied for 1 hr at room temperature. DAPI (at 1 μg/ml; D9564, Sigma) was used to stain DNA. The slides were then mounted in mounting media (PBS, 50% glycerol, 0.5% N-propyl-gallate) and examined using a Leica TCS SP5 scanning confocal microscope. Images were processed using Adobe Photoshop. Third instar larva brains were treated in the same way as in the testis, and the primary antibodies used were rabbit anti-GFP (1:200; Fitzgerald Industries), mouse anti-γ-tubulin (Sigma), and rat anti-α-tubulin (1:200; Chemicon). The corresponding secondary antibodies were used at 1:200 dilution, FITC goat anti-rabbit, rhodamine goat anti-mouse, and Cy5 donkey anti-rat (Jackson Immunoresearch). Anti-Ana1 (1191, 1:200, homemade) immunostaining was performed according to Varmark et al. (2007). We used rhodamine goat anti-rabbit or FITC goat anti-rabbit (1:200, Jackson Immunoresearch) as secondary antibodies. Basal body-length measurements were made using the Leica LAS AF software, and statistical analysis was done with Excel.

Antibody:

Rabbit anti-peptide antibodies (1191) to Ana1 were generated against the peptide CSSTTASSPERRPRKSR by Immunology Consultants Laboratory (Newberg, OR) (Figure 1, A and E).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.—

Ana1 is a centriolar protein required for centriole and cilia formation. (A) Predicted structure of two isoforms (PA and PB) of the Ana1 protein. Red indicates the recognition site of the antibody in the C-terminal part of the protein. The ana1^mecB mutation results in a C-to-T transition at position 3358 (relative to ATG), which changes the Q 1120 codon to a stop codon. (B) Centrosomal labeling is reduced in ana1^mecB mutants. In the control, GFP–PACT (Martinez-Campos et al. 2004) labels the centriole and γ-tubulin the PCM of mitotic cells from larva brain. In ana1^mecB, only 5/25 cells show foci stained by both γ-tubulin and GFP–PACT (arrows), the majority of the cells (18/25) have only γ-tubulin foci, which look smaller than those of the control (double arrows). Few cells (2/25) show a total absence of centrosomal staining. (C) Ana1^mecB mutants (grown in 18°) present all the characteristics of ciliary mutants. Flies are uncoordinated, their legs are crossed, and their wings extend upward. EM cross sections of mechanosensory neurons in the control show the tubular body (TB) surrounded by the dendritic sheath (DS). Control images were previously published in Blachon et al. (2008). In ana1^mecB, the tubular body is absent but the dentritic sheath is present. At the base of the cilium, basal body (BB) is observed in the control whereas it is absent in ana1^mecB. Similarly, in sperm tails no axoneme is detected in ana1^mecB whereas in the control they are clearly visible (arrows) sitting near the mitochondria derivatives (arrowheads). (D) In embryos, Ana1-GFP localizes to the centriole surrounded by γ-tubulin. (E) Before meiosis, each spermatocyte contains four giant centrioles arranged by pairs as a V-shape. In primary spermatocytes expressing the centrosomal protein Asl-GFP (Varmark et al. 2007; Blachon et al. 2008), the anti-Ana1 antibody labels the V-shaped giant centrioles. Some nonspecific signal is detected in the nucleus, but in the ana1^mecB mutant no giant centriole labeling is detected.

Genetic screening:

A total of 150 fly lines from the Exelixis collection (Thibault et al. 2004) and 150 fly lines from the Zucker collection (Koundakjian et al. 2004) were crossed to a fly line containing w; Cyo/Sco; MKRS/TM6B. Males with w; +/Cyo; mutant/TM6B were selected and crossed to w; Ana1-GFP; MKRS/TM6B. Siblings with w; Ana1-GFP/Cyo; mutant/TM6B from this cross were crossed with each other to establish a stable stock. Testis from homozygote mutants were dissected as described above and screened for the absence of PCL using light microscopy.

Molecular biology:

For DNA sequencing, DNA was extracted from whole-body flies using Promega Wizard SV genomic DNA purification system. DNA samples were then amplified by PCR and purified using the QIAGEN gel extraction kit. DNA sequencing was done by Biopolymers Facility at Harvard Medical School. For mRNA experiments, RNA was extracted from whole-body flies using the Promega SV total RNA isolation system. cDNA sample was then generated by RT–PCR with oligo(dT). Primers used in the experiments were as follows: Actin, 5′-GCTGCTCTGGTCGTTGAC-3′ and 5′-CGCGATTTGACCGACTACC-3′; poc1, 5′-CCGCATCCTAGACCTGC-3′ and 5′-GGATGTCGAACTGCCTTCGC-3′.

Ana1 is a pan-centriolar protein involved in centriole/cilium formation:

In a screen for genes involved in centriole/cilium formation, we identified mechanosensory mutant B (mecB), a mutant of Ana1 (Avidor-Reiss et al. 2004). We found that the ana1^mecB mutation changes C to T at position 3368 in the gene CG6631 (Figure 1A), leading to an early stop codon. Blast analysis finds that insect Ana1 bears significant similarity in its N terminus to two related vertebrate genes, KIAA1731 and Ddc8. KIAA1731 was previously identified as a candidate centriolar protein in a proteomic study for human centrosomes (Andersen et al. 2003), and Ddc8 is a gene expressed during spermatogenesis (Catalano et al. 1997). Fly Ana1 was previously identified by Goshima et al. (2007) in an RNA interference (RNAi) screen in S2 cells. RNAi depletion of Ana1 results in a decrease of SAS-6 at the spindle poles (Goshima et al. 2007). Consistent with this is our observation of a reduction of centrosome labeling by GFP–pericentrin/AKAP450 centrosomal targeting (PACT) and γ-tubulin in neuroblasts of ana1^mecB mutants (Figure 1B). In addition, ana1^mecB mutants exhibit all the features of a ciliary mutant (Figure 1C): flies are uncoordinated due to the absence of cilia in mechanosensory neurons. Additionally, axonemes are missing in sperm tails and the sperm is immotile. These results suggest that Ana1 has a role in centriole and cilium formation.

We made a GFP fusion of the Ana1 protein and confirmed its specific localization to the centriole in embryos (Figure 1D), which is consistent with Goshima's results that localize it to S2 cell centrioles. In addition, we have shown previously that Ana1 is a component of mature basal bodies in spermatids and sensory neurons (Blachon et al. 2008), suggesting that Ana1 is a core component of the centriole. We took advantage of the ubiquity and specificity of Ana1 labeling of the centriole to use it as a pan-centriolar marker.

We also generated an antibody against the C terminus of Ana1 and checked its specificity using the ana1^mecB mutant. As expected, in primary spermatocytes, the antibody labels the giant centriole stained by the centrosomal protein Asl-GFP (Varmark et al. 2007; Blachon et al. 2008); however, no centriolar staining was detected in ana1^mecB spermatocytes (Figure 1E).

Ana1 labels a novel structure that associates with the sperm basal body and is distinct from the centriolar adjunct:

In fly spermatids, the centriole elongates to form the basal body, which is the base for the flagellum axoneme (Figure 2A). The basal body in this system is basically a giant centriole that measures 2.6 μm and therefore is easily observed using light microscopy. This system lends itself to cytological analysis as distinct stages in centriole biogenesis are associated with morphological changes of the cell and are organized in chronological order along the testes (Tates 1971; Cenci et al. 1994) (Figure 2A). While studying the localization of centriolar protein Ana1 using the Ana1-GFP reporter, we noted that it labels a novel structure in spermatids. We use phase-contrast imaging of live or fixed preparations to precisely identify the stage of spermatid development. Early spermatids (or onion stage, S13) are recognized as round cells containing a white, round nucleus (N) and dark, round mitochondria derivatives (M) of similar size (Figure 2B, left). In these cells, Ana1-GFP labels the giant centriole (also called the basal body) uniformly all along its length. Intermediate spermatids (S15) have short protrusions at one side and a white, round nucleus (Figure 2B, middle). In intermediate spermatids, the Ana1-GFP labeling of the basal body is no longer uniform and a single bulge appears on its proximal surface. Late spermatids are organized in a tight bundle of elongated cells and at that time the bulge disappears and a distinct Ana-1-labeled structure appears more toward the center of the giant centriole (Figure 2B, right). We measured the length of the fluorescence signal and found that the structure width is 0.44 ± 0.06 μm (N = 34), which is similar to the width of the giant centriole, 0.54 ± 0.04 μm (N = 41). We conclude that this novel structure first appears on the proximal surface of the giant centriole and later migrates more distally from the giant centriole during spermatid differentiation.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.—

Ana1 labels a novel structure appearing near the mother centriole in spermatids. (A) Diagram depicting the different stages of spermatid development based on the observations of Tates (1971). (M, mitochondria; N, nucleus; Ax, axoneme). The basal body or giant centriole (Cen) is surrounded by the centriolar adjunct (CA) and, near it, we can follow the formation of the PCL. (B) We use phase-contrast pictures (unfixed testis) to determine the spermatid stage. The onion stage (stage S13) is characterized by a round nucleus (N) of the same size as the mitochondrial derivatives (M). The cell body of intermediate spermatids (stages 15 and 16) elongates, forming short protrusions (arrows), but the nucleus remains round. In late spermatid development (stage 17), the nucleus becomes oval. Ana1-GFP labels the giant centriole (Cen), and in intermediate spermatids a bulge forms on one side and becomes individualized as PCL in late spermatid development. (C) Staining with anti-γ-tubulin antibody shows that the PCL labeled by Ana1 is an entity different from the γ-tubulin collar that is reminiscent of the centriolar adjunct (CA). (D) Antibody against Ana1 labels the V-shape pair of giant centrioles in primary spermatocytes (left) and the giant centriole and PCL in spermatids (right) in flies expressing Ana1-GFP. (E) In wild-type primary spermatocytes, anti-Ana1 antibody stains the endogenous protein in the giant centrioles and colocalizes with γ-tubulin staining (left). In spermatids, the antibody labels the PCL, demonstrating that its formation is not due to centriolar protein overexpression (right).

We then studied the relationship between this novel structure and the PCM marker γ-tubulin. Although no direct demonstration has been done using immuno-EM on γ-tubulin during spermatogenesis, γ-tubulin staining shows a remarkably close correlation with the morphogenesis of CA, therefore it is accepted that it most likely represents the centriolar adjunct (Wilson et al. 1997; Callaini et al. 1999; Baker et al. 2004; Texada et al. 2008). In early spermatids, γ-tubulin labels the vicinity of the giant centriole along most of its length (Figure 2C). As the spermatids continue to differentiate, γ-tubulin labeling gets shorter and is shaped into a collar that surrounds the centriole on one side (Figure 2C). The open side of the γ-tubulin collar faces the Ana1-GFP-labeled focus; clearly, the two signals label two different structures. Later, γ-tubulin fades away while Ana1-GFP staining is maintained (Figure 2C). This demonstrates that Ana1-GFP labels a novel structure that is clearly distinct from the centriolar adjunct. Since this structure forms near the proximal side of the giant centriole, is labeled by a centriolar protein, and is distinct from the centriolar adjunct, it presumably represents the fly equivalent of the vertebrate proximal centriole, which is why we called this proximal centriole-like (PCL).

In electron microscopy, the centriolar adjunct appears very dark, which possibly hinders the capacity to distinguish structures that are closely associated with it (Anderson 1967; Tates 1971; Tokuyasu 1975a,b). In an attempt to visualize the PCL using electron microscopy, we performed serial electron microscopy of fixed tissue in late spermatids when the centriolar adjunct is absent (Figure S1 and Figure S2). We have observed only one centriolar structure, the basal body, suggesting that the PCL is hard to detect structurally.

Strong overexpression of centriolar proteins generate ectopic structures (Kleylein-Sohn et al. 2007; Peel et al. 2007; Rodrigues-Martins et al. 2007; Strnad et al. 2007). To minimize this effect, Ana1 and other centriolar proteins used in this study are driven by their own promoters. Each of the centriolar protein GFP fusion constructs studied in this work show a strict specificity to centriolar structures. Additionally, to ensure that the formation of the PCL is not a consequence of the expression of Ana1-GFP, we used the antibody against Ana1. As expected, the antibody recognizes the centriole and the PCL in testes from flies expressing Ana1-GFP (Figure 2D). Importantly, in spermatids, which do not overexpress any centriolar proteins, the Ana1 antibody recognizes the PCL found near the centriole that is surrounded by the γ-tubulin collar (Figure 2E).

Altogether, these results indicate that Ana1 is a component of a novel centriolar structure forming near the giant centriole. In vertebrates, the distal and proximal centrioles are mother and daughter centrioles, respectively (Krioutchkova and Onishchenko 1999). In numerous animals including fish, starfish, annelids, and mollusks, the proximal centriole is smaller than the distal centriole (Krioutchkova and Onishchenko 1999). The proximal centriole-like structure resembles a daughter centriole as it exhibits two features that are universally shared when daughter centrioles are formed: it associates with the vicinity of the proximal end of the mother centriole and its number is restricted to one. As a result, PCL formation likely depends on the same molecular mechanism that triggers daughter centriole formation.

Proximal centriole-like structure formation is mediated by a conserved molecular pathway:

Recently, studies in Caenorhabditis elegans using RNAi established the chronology of a module of proteins responsible for the formation of the daughter centriole (Delattre et al. 2006; Pelletier et al. 2006). For a recent review, see Strnad and Gonczy (2008). The consensus resulting from these studies indicates that a Zyg-1/PLK4 family of kinases is required for the early recruitment of SAS-6 and SAS-5. Electron tomography made it possible to show that centriole assembly in C. elegans starts with the formation of a central tube that depends on SAS-6 and SAS-5, followed by the assembly of a microtubule wall under the control of SAS-4 (Pelletier et al. 2006). In addition, it was found recently that, while SAS-6 incorporates into a stable structure early on, SAS-4 remains in dynamic equilibrium until later when it is stably incorporated in a step requiring γ-tubulin and mediating centriole wall assembly (Dammermann et al. 2008).

This module of proteins is conserved in other organisms; for example, in humans the homolog of Zyg1, the kinase SAK/PLK4, is required early on in centriole formation (Kleylein-Sohn et al. 2007). The role of SAS-6 in centriole duplication has been confirmed in humans, and in addition it is found localized in the centriole cartwheel in humans (Kleylein-Sohn et al. 2007), in Chlamydomonas (Nakazawa et al. 2007), and in Tetrahymena (Kilburn et al. 2007). Finally, SAS-4 has two homologs in humans; one of them, CPAP, is a centriolar protein that is able to bind microtubules (Hung et al. 2000; Hsu et al. 2008). This fits with a model where, in a first step, early players such as SAK/PLK4 and SAS-6 build the centriole core, and at a later time SAS-4, via its interaction with microtubules, assembles the centriolar wall. As in other model organisms, Drosophila SAK/PLK4, SAS-6, and SAS-4 are essential for centriole formation (Bettencourt-Dias et al. 2005; Basto et al. 2006; Rodrigues-Martins et al. 2007).

To test our hypothesis on the nature of the PCL, we studied the phenotypes of available mutants in Drosophila—plk4/sak, sas-6, and sas-4—and followed the localization of the corresponding proteins. In flies, the hypomorphic mutation of plk4^c06612 contains a piggyBac insertion near the end of the gene that affects centriole duplication. It was reported that, in this mutant, 72% of the neuroblast cells have fewer than two centrioles (Bettencourt-Dias et al. 2005). Similarly, a piggyBac insertion in the second exon of the sas-6 gene leads to a decrease of centriole number, and 82% of neuroblast cells have fewer than two centrioles (Rodrigues-Martins et al. 2007). Consistent with this, using Ana1-GFP as a reporter, we noted that plk4^c06612 and sas-6^c02901 have a reduced number of centrioles in spermatocytes (data not shown). More importantly, as expected from our hypothesis, similar reduction in PCL formation is observed in plk4^c06612 and sas-6^c02901 spermatids (Figure 3, A and B, and Table 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.—

PCL formation depends on early players in centriole duplication. (A and B) PCL formation is impaired in plk4 and sas-6 mutants (for quantification see Table 1). (C) In primary spermatocytes, we can distinguish the daughter centriole (Dau) sitting perpendicularly to the mother centriole (Mo) (based on the γ-tubulin staining). SAS-6-GFP localizes to the proximal part of both mother (Mo) and daughter (Dau) giant centrioles in spermatocytes (left) but is enriched in the daughter. In spermatids, SAS-6-GFP localizes to the proximal side of the giant centriole (Cen) and in addition a second dot, probably in the PCL, appears (right). (D) SAS-6-GFP is localized to the proximal part of the V-shaped pair of giant centrioles in primary spermatocytes (left) stained with anti-Ana1 antibody. In spermatids (right), the staining with anti-Ana1 antibody confirms the colocalization of the SAS-6-GFP second dot with the PCL. (E) We follow the fate of the maternally contributed centriole (for explanation see Blachon et al. 2008) in sas-4 mutants. In sas-4, centrioles are shorter but it does not affect PCL formation, demonstrating that SAS-4 is not required for PCL formation. (F) Like SAS-6-GFP, SAS-4-GFP is localized to the proximal part of the giant centriole in primary spermatocytes. In early spermatids, SAS-4-GFP is still present at the proximal part of the centriole and a second dot can be distinguished. Later, the SAS-4-GFP signal decreased and no SAS4-GFP was detected in the PCL. (G) The PCM protein Cnn shows a similar localization to γ-tubulin (Blachon et al. 2008) in meiosis. However, after meiosis, it is restricted to the proximal part of the giant centriole in early spermatids; subsequently, it is colocalized transiently with the forming PCL in intermediate spermatids. Later, it disappears in late spermatid development. (H) In cnn mutant spermatids, we still observe the PCL labeled by Ana1-GFP, but the γ-tubulin pattern is disrupted. Some spermatids show a very faint or an absence of γ-tubulin labeling (left) whereas others have abnormal localization of the γ-tubulin collars (right).

Examination of the localization of SAS-6 tagged with GFP that shows a specific localization to the centriole in neuroblast cells (Blachon et al. 2008) demonstrates that, in early spermatocytes, SAS-6-GFP labels the center of the centrosome and that, in late spermatocytes after centriole elongation, only the proximal end of the centriole is labeled (Figure 3C). This localization is different from previous research, which reported that SAS-6-GFP also labeled the distal end of the centriole (Rodrigues-Martins et al. 2007), which is likely to be due to the strong ubiquitin promoter used to drive SAS-6-GFP expression in previous studies. In spermatids, SAS-6-GFP maintains its labeling at the proximal end of the giant centriole (Figure 3C). However, we noted that, in addition to this signal, a second SAS-6-GFP focus appears in early spermatids and remains in intermediate spermatids. To determine whether this focus corresponds to the PCL, we stained the testes with Ana1 antibody (Figure 3D). We found that the second SAS-6-GFP focus colocalizes with the Ana1-labeled PCL in intermediate spermatids. This result confirms that the PCL is labeled by SAS-6, which is consistent with the role of SAS-6 in PCL formation.

It was previously shown that, in sas-4^s2214 flies, centriole numbers are reduced in the development of the fly until they are completely missing in adult flies (Basto et al. 2006). Since in sas-4^s2214 flies adult spermatids lack new centrioles, we analyzed the formation of the PCL at the vicinity of maternally contributed centrioles. This method (Blachon et al. 2008) is based on following the fate of centrioles built during embryonic development using the wild-type maternal proteins accumulated in the egg (maternal contribution). After the maternal proteins are depleted, we can study this preassembled centriole in the mutant context. We noted that sas-4^s2214 centrioles are abnormally short (1.1 ± 0.3 μm, N = 11, vs. 2.3 ± 0.4 μm, N = 41, for wild type; Blachon et al. 2008) (Figure 3E). This result is consistent with a previous report that showed that the partial inactivation of SAS-4 in C. elegans leads to the formation of smaller centrioles with an abnormal structure (Kirkham et al. 2003). Together, this suggests that SAS-4 has a role in centriole stability. Despite the fact that the giant centriole is abnormal, the PCL is present in spermatids (Figure 3E). This result suggests that SAS-4 is not involved in PCL formation and that PCL formation can take place even if the mother centriole integrity is compromised.

To determine whether SAS-4 plays a nonessential role, we analyzed SAS-4-GFP localization. We found that, like SAS-6-GFP, SAS-4-GFP in spermatocytes and spermatids labels the proximal end of the centriole. Again, this is different from previous reports that used a strong ubiquitin promoter to drive SAS-4 expression (Basto et al. 2006). In early spermatids (S13), similarly to SAS-6-GFP, SAS-4-GFP shows two dots. However, in intermediate and late spermatids, SAS-4-GFP does not label the PCL (Figure 3F), supporting the observation that SAS-4 is not essential for PCL formation. Consistent with results in worms, it seems that, while SAS-4 and SAS-6 are recruited together to an early structure during the PCL formation, only SAS-6, and not SAS-4, is essential at that time (Pelletier et al. 2006; Dammermann et al. 2008). SAS-4 in worms is essential later for the formation of the centriole wall (Dammermann et al. 2008), leading us to hypothesize that the PCL might not have a microtubule-containing wall, which explains why it is so hard to find it using electron microscopy.

Most of the more established markers of centrosomes in flies are components of the PCM and are not essential for centriole duplication (Kellogg et al. 1995; Megraw et al. 1999; Butcher et al. 2004). We have tested the localization of one of these proteins: centrosomin (Cnn), which has been identified as a PCM protein (Li and Kaufman 1996). In early spermatocytes, Cnn is localized to the centriole and, like γ-tubulin (Blachon et al. 2008), it is reorganized to the PCM during meiosis (Figure 3G, left). In early spermatids, Cnn is localized at the extremities of the centriole with some enrichment at the proximal part. In intermediate spermatids when the PCL starts to form, Cnn colocalizes with the bulge (Figure 3G, right) and later disappears. It has recently been shown that Cnn is required for the connection between the centriole and the PCM (Lucas and Raff 2007); consequently, in cnn mutants, PCM recruitment to the centriole is impaired and centrioles segregate abnormally. Since Cnn colocalizes with the forming PCL, we wondered if PCL formation would be impaired in cnn mutants. We found that in cnn spermatids the PCL forms (Figure 3H); however, the γ-tubulin collars were largely reduced and almost absent (Figure 3H, left) or showed a totally abnormal localization (Figure 3H, right). These results demonstrate that, as in normal daughter centriole formation, Cnn is dispensable for PCL formation. We propose that, similar to its role in recruiting the PCM, Cnn localization to the forming PCL is required for the proper localization of the γ-tubulin collar.

Bld10p is recruited late to the PCL but is not required for its formation:

The above findings suggest that the PCL is accessible for genetic analysis. Therefore, we first tested whether this system could be used to determine the involvement of known centriolar protein in the initiation of the daughter centriole. Bld10p/Cep135 is a centriolar protein that was found in Chlamydomonas to be localized just internally to the wall of microtubules. In human cells, it was shown to be a component of the centriole wall (Hiraki et al. 2007; Kleylein-Sohn et al. 2007). For this reason, it is possible that Bld10p acts between SAS-6 and SAS-4 in the centriole assembly pathway. To test this, we examined the localization of a Bld10p-GFP reporter (Blachon et al. 2008). We found that Bld10p-GFP labels the centriole in early spermatocytes and labels them along their length after elongation (Figure 4A). This localization is maintained in spermatids. In addition, Bld10p-GFP labels the PCL; however, this labeling appears later in spermatid differentiation when the PCL is more distal and farther away from the centriole. To directly observe the sequential order of Ana1 and Bld10p in centriole assembly, we generated flies expressing both Bld10p-GFP and Ana1-tdTomato in a dbld10 mutant background as a mean to maximize the Bld10p signal (Figure 4B). We confirmed that Bld10p localizes to the PCL only after Ana1 is already there, demonstrating that Bld10p is recruited later in PCL formation (Figure 4B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.—

Bld10p is recruited late to the PCL and is not required for its formation. (A) Bld10p-GFP labels the centriole in spermatocytes (left) and the giant centriole (Cen) in spermatids (right). Only later, in late spermatids, is Bld10p recruited to the PCL. The centriolar adjunct (CA) and the nucleus (N) are visible in phase-contrast pictures (middle). (B) Colocalization of Bld10p-GFP with Ana1-tdTomato confirms that Bld10p localizes to the PCL only later (arrow). The double arrowhead points to the PCL labeled by Ana1-tdTomato and not yet by Bld10p-GFP. To maximize the signal, Bld10p-GFP was expressed in a bld10 mutant context (rescue flies). (C) In the bld10 mutant, centrioles are shorter but PCL formation is not affected, consistent with the role of Bld10p later in the process.

To test for the role of Bld10p in initiating daughter centriole formation, from the Exelixis collection we obtained the mutant allele dbld10^c04199 (Thibault et al. 2004), which has an insertion in the middle of the gene resulting in Bld10p truncation, and most likely represents a severe loss of function. We analyzed flies that are heterozygous for this allele over the chromosomal deletion Df(3L)ED218, which covers this gene. We found that males are sterile, have shorter centrioles than the control, and occasionally have mislocalized PCLs (Figure 4C). Consistent with the late recruitment of Bld10p to the PCL, Bld10p is not essential for its formation. This indicates that Bld10p acts after the initiation of PCL formation.

Centriolar markers of the giant centriole and PCL disappear in mature sperm:

Sperm centrosomes are known to undergo a process in which they lose many of their markers; this process is known as centrosome reduction (Schatten 1994). In flies it was reported that centrosomal proteins such as γ-tubulin, Cnn, centrin, and CP190 are not found in mature sperm centrosomes and centrioles (Wilson et al. 1997; Callaini et al. 1999). To test what happened to the PCL at the final stages of spermatogenesis, we analyzed mature sperm for PCL markers. We were unable to detect any of the markers that we tested (Figure S3). As a consequence, until now, it was not possible to follow the fate of the PCL after fertilization to investigate its function.

A forward genetic screen identified a new player in PCL formation—POC1:

Next we were interested in testing if we could use forward genetics in this system to identify novel centriolar mutants involved in the initiation of daughter centriole formation. Mutations that specifically affect the centriole and cilia, such as plk4, sas-6, and sas-4, develop into adult flies (Bettencourt-Dias et al. 2005; Basto et al. 2006; Rodrigues-Martins et al. 2007). Therefore, we expected that mutations that specifically affect the PCL would not cause lethality early in development. We screened 150 fly lines randomly selected from the Exelixis collection (Thibault et al. 2004) and 150 lines from the Zuker collection (Koundakjian et al. 2004) that either had a ciliary/centriolar phenotype producing uncoordinated flies or had a known mutation in a candidate ciliary gene (Avidor-Reiss et al. 2004). To observe the centriole and the PCL, we introduced the Ana1-GFP reporter into the genetic background of each of these mutants. A single mutation found in the Zuker collection, named k245, had a complete loss of Ana1-GFP (Figure 5A) and Bld10p-GFP (Figure 5B) labeling of the PCL. Other mutants from the Zuker collection had clear PCL labeling, indicating that the lack of PCL is due to a particular mutation in the k245 line (data not shown). In addition to the absence of PCL labeling, k245 had a shorter centriole in spermatocytes (Figure 5C) but was able to generate viable and fertile flies.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.—

POC1, a conserved centriolar protein, is involved in PCL formation. (A) A forward genetic screen identified a new mutation, k245, that impairs the formation of the PCL as judged by Ana1-GFP labeling. (B) Absence of a normal PCL in k245 is confirmed using Bld10p-GFP. (C) Statistical analysis of the basal body length on poc1 alleles. The deficiency used is Df(3L)Ly. Each experiment was performed on three different flies and 60 basal bodies were measured. The columns marked by a star (*) are statistically different from the control; additional pairs were also found significantly different on the basis of P < 0.01 using Student's t-test. (D) Complementation test with several deletions shows that k245 maps to an interval containing the gene CG10191. Solid bars: deletions that did not complement the k245 phenotype; shaded bars: deletions that did complement k245. (E) CG10191 encodes the conserved centriolar protein POC1. k245 mutation changes a G to an A at position 880 (in red), which is predicted to eliminate a splicing site. poc1^c06059 contains an insertion in poc1 promoter (in red). POC1 protein is composed of 7 WD domains known to form a β-propeller structure and contains a coiled-coil domain in its C terminus. The k245 mutation is predicted to induce a frameshift resulting in a truncated protein containing only the WD domains. (F) RT–PCR on mRNA shows that in wild-type flies mRNA is spliced to give a smaller product, which is absent in poc1^k245 flies, confirming that the k245 mutation eliminates a splicing site. poc1^c06059 shows a significantly lower amount of mRNA than wild type. Actin mRNA is shown as a control for loading.

Positional cloning found that k245 mapped to an interval containing 10 genes (Figure 5D). A single gene in the interval, CG10191, was identified previously as a candidate ciliary gene in a bioinformatics screen (Avidor-Reiss et al. 2004) and was later reported in centriolar proteomic studies to be the ortholog of Chlamydomonas and Tetrahymena: proteome of centrioles 1, or POC1 (Keller et al. 2005; Kilburn et al. 2007). We found that a piggyBac insertion allele (c06059) generated by Exelixis (Thibault et al. 2004) in the predicted gene CG10191 failed to complement k245 (Figure 5E), demonstrating that k245 is a poc1 allele. POC1 is a 391-amino-acid protein composed of a highly conserved N-terminal domain with 7 WD repeats (amino acids 1–300) and a less conserved C terminus of 90 amino acids that contains a coiled-coil domain of 42 amino acids at its end (Figure 5E). Seven WD repeats are known to form a β-propeller structure (Neer and Smith 1996), and this domain is encoded by the first exon that is separated from the rest of the protein by a single intron (Figure 5E). We sequenced poc1^k245 and found that it contains the mutation-changing nucleotide G880A (relative to ATG), which eliminates the predicted 5′ intron splice site (Figure 5E). This was confirmed by RT–PCR of mRNA from poc1^k245 and wild-type control flies (Figure 5F). Therefore, the poc1^k245 is predicted to result in a frameshift after the first exon and to generate a truncated protein consisting of only the WD β-propeller structure. poc1^c06059 has a piggyBac insertion in the promoter of the gene (67 bases upstream to the ATG) (Figure 5E) that results in a reduced mRNA level (Figure 5F). poc1^c06059, like dpoc1^k245, had centrioles that were shorter than normal in spermatocytes (Figure 5C) and was able to generate viable and fertile flies.

The identification of POC1 as a protein essential for PCL formation suggests that POC1 plays a central role in centriole formation. This is consistent with the localization study in Tetrahymena, which shows that POC1 associates with the site of daughter centriole assembly (Kilburn et al. 2007). In addition, a very recent study shows that, in Chlamydomonas, POC1 localizes to the forming basal body (pro-basal body) and that, in human cells, RNAi against POC1 reduces centriole overduplication (Keller et al. 2008). This demonstrates that studying PCL formation can serve to identify novel mutation in conserved centriolar genes.

Conclusion:

We have found that topological, molecular, and genetic criteria point to the presence of a daughter centriole in fly spermatids. By analogy with other animal groups, the PCL is likely the fly counterpart of the proximal centriole found in vertebrates. PCL formation depends on SAK/PLK4 and SAS-6, which are proteins that act early in centriole duplication, but is independent of proteins such as SAS-4 or Bld10p, which act later. This indicates that the PCL represents an early step of the centriole formation that has stopped in its development. Unlike the vertebrate procentriole, the PCL did not conserve the distinctive structure of mature centrioles, suggesting that its unknown function does not require maturation in spermatids. Consequently, the PCL is an easily accessible model for studying genetically the early steps of centriole formation.