Drosophila chibby is required for basal body formation and ciliogenesis but not for Wg signaling

The structure and protein composition of cilia are highly conserved across eukaryotes. A cilium is built on a basal body derived from the mother centriole. Centriole maturation into a basal body is a complex process that includes the progressive addition of different protein structures including transition fibers, basal feet, and ciliary rootlets (Hoyer-Fender, 2010; Ishikawa and Marshall, 2011). The prototypical centriole found in most organisms is a radially symmetric arrangement of nine microtubule triplets. However, Ecdysozoa such as Caenorhabditis elegans and Drosophila melanogaster show major differences in the ultrastructure of centriole/basal bodies. In C. elegans, the centriole is composed of singlet microtubules, not triplets. In Drosophila, centrioles in embryos and most tissues are composed of microtubule doublets. In spermatocytes of the Drosophila male germ line, centrioles do show the prototypical nine triplets but lack some of the maturation structures found in other eukaryotes.

Ciliogenesis starts with the docking of the basal body, either directly to the membrane at the cell surface or to cytosolic vesicles that will eventually fuse to the plasma membrane (Sorokin, 1968). An initial extension of microtubule doublets from the basal body constitutes a specialized transition zone (TZ) delineated by specific localized proteins and by structures such as transition fibers, which link the basal body to the membrane and form a selective barrier to transport into the cilium (Ishikawa and Marshall, 2011). Further extension of the axoneme and ciliary membrane above the cell surface requires intraflagellar transport (IFT), in which specialized kinesin and dynein motors move an IFT protein complex and associated cargoes from the base to the tip of the cilium and back (Rosenbaum and Witman, 2002; Ishikawa and Marshall, 2011). Two notable exceptions to the IFT mode of assembly are the flagella of Plasmodium and of Drosophila sperm (Han et al., 2003; Sarpal et al., 2003; Briggs et al., 2004). Drosophila sperm axonemes are first fully extended in a cytoplasmic syncytium, independent of IFT, and only then ensheathed by a plasma membrane, during spermatid individualization (Fuller, 1993). However, a form of centriole maturation does occur earlier in Drosophila spermatogenesis: in primary spermatocytes, all four centrioles extend a distal zone formed of nine doublets and ensheathed by a membrane cap. The elongated centrioles migrate to the plasma membrane, where the distal segment protrudes from the cell surface and hence resembles a short cilium (Tates, 1971; Fuller, 1993). The function of this transient primary cilium is unknown. unc mutants, in which it is shortened, can proceed through an apparently normal meiosis (Baker et al., 2004). At the onset of meiosis I, the pairs of elongated centrioles are engulfed together with their membrane caps and relocalized to the meiotic spindle poles (Tates, 1971). These observations strongly suggest that basal body and TZ-like structures assemble in Drosophila spermatocytes before meiosis and flagellar elongation.

In vertebrates, cilia are required during development to mediate signal transduction pathways such as the sonic hedgehog pathway (Goetz and Anderson, 2010). Defects in cilia or cilia-associated proteins are also associated with defective Wg-Int (WNT) and planar cell polarity signaling, although some of the data are contradictory, and the precise mechanism is far from understood (Ajima and Hamada, 2011; Wallingford and Mitchell, 2011). To date, reports of an association between cilia and Wingless (Wg) signaling are limited to vertebrates, with one notable exception: the CHIBBY (CBY) protein. CBY was first identified as an interactor with β-catenin in humans, and RNAi experiments targeting cby expression implied that CBY negatively regulates WNT signaling both in mammalian cell culture and in Drosophila (Takemaru et al., 2003). Identified as PIGEA-14, CBY was also shown to be associated with polycystin-2, a ciliary ion channel (Hidaka et al., 2004). Last, cby^−/− mutant mice showed defective airway function associated with impaired basal body docking, reduced number of motile cilia (Voronina et al., 2009; Love et al., 2010), and increased expression of β-catenin regulatory targets (Love et al., 2010).

We have identified Drosophila cby as a putative target of the RFX transcription factor, which activates transcription of many genes required for the assembly of cilia (Dubruille et al., 2002; Laurençon et al., 2007). Together, these observations suggested a possible link between Wg signaling components and cilia in Drosophila. We show here that Drosophila cby is expressed only in cells that harbor cilia, namely in ciliated type I sensory neurons and in male germ cells. CBY protein is associated with the maturing basal bodies and TZ in these two cell types. cby-null mutant flies show sensory transduction defects and male hypofertility but no evidence of Wg signaling defects. We show that CBY is required in ciliated neurons for proper TZ and axonemal assembly and for protein trafficking into or out of the cilia. In spermatogenesis, CBY is also required for centriole-to–basal body conversion, as revealed by defective basal body–associated uncoordinated (UNC) protein deposition. Last, we show that cby is required for axonemal assembly during spermiogenesis. Altogether our results imply that, in Drosophila, CBY functions in basal body/TZ biogenesis but not in WNT signaling.

CBY, a small protein with a C-terminal coiled-coil domain, is found in all ciliated unikont animals, with the exception of C. elegans and other nematodes (Fig. 1). CBY is also found in choanoflagellates, which can have a colonial life and are closely related to animals (King et al., 2008). Two probable CBY homologs are also present in the Excavate protist Trichomonas vaginalis and two possibly related proteins in the lower plant Selaginella moellendorffii (which has flagellated zoospores). Their occurrence suggests that a CBY homolog was present in the universal ancestral eukaryote. However, no CBY-related protein could be identified in any other bikont genomes, including well-studied ciliated organisms such as Chlamydomonas reinhardtii, Trypanosoma, or Paramecium. The Excavates have been proposed to stem from a very deep branching of the eukaryotic tree but with a closer association with bikonts than with unikonts (Hampl et al., 2009); the Trichomonas cby homolog instead favors a closer association with the unikonts.

Drosophila cby was identified in a screen for potential regulatory targets of the transcription factor RFX (Laurençon et al., 2007). To analyze cby expression in Drosophila, we constructed two transgenes including the cby promoter and entire coding sequence, fused at the C terminus to GFP or to Tomato (see Materials and methods). In transgenic embryos, expression of both constructs was restricted to type I sensory neurons of the peripheral nervous system (PNS), the only ciliated cells in the embryo. The fusion protein was found in the cell body and as a dot at the tip of the sensory process in all type I sensory neurons (Fig. 2 A). No expression was detectable in any other parts of the embryo (unpublished data). In an Rfx mutant background, cby expression was lost, confirming that in the PNS, cby is regulated by RFX (Fig. 2 A).

By coimmunolabeling with the centriole/basal body markers Asterless (ASL), pericentrin-like protein (PLP), and SAS-4, CBY was located more precisely at the tip of the basal body to a region that likely corresponds to the TZ between the basal body and the cilium proper (Fig. 2 B). To confirm this hypothesis, we looked at CBY location relative to B9D1/Meckel syndrome (MKS) 1–related (MKSR1), one of three interdependent B9 domain proteins involved in assembling the TZ (Bialas et al., 2009; Williams et al., 2011). Its Drosophila ortholog is encoded by CG14870 and is expressed in ciliated cells (Avidor-Reiss et al., 2004). We made a CG14870-MYC reporter construct, with the promoter and complete coding sequences fused to a MYC epitope, and found that CBY-Tomato and CG14870-MYC were indeed colocalized at the distal end of the basal body (Fig. 2 B). Hence, CBY is associated with the TZ of Drosophila sensory cilia. UNC, a basal body–associated protein required for axoneme structural integrity (Baker et al., 2004), also colocalized with CBY at the distal end of basal bodies in type I sensory neurons (Fig. 2 B).

cby was also expressed in larval and adult testis. CBY was associated with the duplicated elongating centrioles in G2 spermatocytes and with basal bodies in early spermatids (Fig. 3, A1–3 and 6–8). No cby expression was found associated with centrioles in dividing spermatogonia before the spermatocyte stage (unpublished data). In contrast to sensory neurons, cby expression in the testis does not require RFX (Fig. 3 B), consistent with RFX expression in spermatids only after the onset of cby expression (Vandaele et al., 2001).

During spermatid elongation, CBY was not maintained at the basal body but was observed at the tip of the growing axoneme until late spermatid differentiation (Fig. 3, A4, 5, 9, and 10). CBY was also found transiently associated with the mitochondria in onion-stage spermatids and early elongating spermatids (white arrowheads in Fig. 3, A8 and A4).

UNC is also located at both the basal body and growing tip of the axoneme (Baker et al., 2004; Wei et al., 2008). We observed that CBY-Tomato and UNC-GFP appeared together at the tip of the elongating centrioles in early G2 spermatocytes and remained there during spermatocyte maturation to spermatids, as the centrioles converted to basal bodies, with CBY localized in a zone that overlaps and extends more distal to UNC (Fig. 3, A6–8). However, during axoneme elongation, when CBY and UNC were both found at the tip of the growing axoneme, only UNC was maintained at the basal body (Fig. 3, A9 and 10).

Our observations show that CBY is localized at the distal end of maturing basal bodies in Drosophila. Interestingly, this localization is conserved in mammals in which CBY was found at the tip of the basal body of primary cilia in inner medullary collecting duct 3 (IMCD3) cells and above basal bodies of motile cilia in primary ependymal cells (Fig. S3).

To understand the precise function of CBY, we constructed a null mutation, cby¹, using homologous recombination (Maggert et al., 2008) to delete most of the cby coding region (Fig. 4 A). PCR performed with primers designed on Fig. 4 A show that recombination occurred correctly on the left and right arms of the construct and that cby sequences are deleted in the cby¹ allele (Fig. S1). In addition, we sequenced the entire recombinant region and confirmed that coding sequences of exon 2 are removed without affecting the adjacent genes.

RT-PCR on mRNA extracted from cby^1/1 flies detected no cby mRNA (Fig. S1), whereas expression of the two adjacent genes was not affected. In addition, cby^1/1 homozygotes and heterozygotes for cby¹ and Df(3R)BSC805, a complete deficiency of the region, showed similar survival, overall behavior, and fertility (see the following three paragraphs; Figs. 4 and S1), confirming that cby¹ is a null allele.

Homozygous cby^1/1 and cby¹/Df(3R)BSC805 mutants were viable, but their behavior was altered, and males appeared to be hypofertile (Figs. 4 and S1). Wild-type flies show negative geotaxis; they move upward on vertical surfaces. This can be assayed with a simple climb test in which the percentages of flies that climb to different heights in a tube are measured after tapping them to the bottom. Whereas most control flies reached and stayed at the top of the tube in the first minute, most cby^1/1 or cby¹/Df(3R)BSC805 flies stayed at the bottom of the tube (Fig. 4 B). This phenotype was rescued by one copy of the CBY-Tomato transgene, confirming that CBY is required for negative geotaxis.

This change in geotaxis behavior could be a result either of a defect in gravity perception or locomotor coordination or both. Gravity is mainly sensed in Drosophila by antennal chordotonal organs (COs; Kamikouchi et al., 2009; Sun et al., 2009), whereas general locomotor coordination relies principally on external mechanosensory organs, with some contributions by COs. The cby^1/1 behavioral phenotype was milder than the one observed in atonal mutants, which lack all COs (Jarman et al., 1993), or in transient receptor potential vanilloid channel or tilB mutants, which lack CO function (Eberl et al., 2000; Kavlie et al., 2010), and was much less severe than the extreme uncoordination seen in flies that lack mechanotransduction in bristles (Kernan et al., 1994) or lack all cilia (Sas-4 or Rfx mutant flies; Dubruille et al., 2002; Basto et al., 2006). This indicates that mechanosensation is not completely abolished in cby^1/1 flies.

We confirmed a defect in mechanosensation by recording antennal sound-evoked potentials (SEPs), which depend on mechanotransduction in the antennal CO (Eberl et al., 2000; Eberl and Boekhoff-Falk, 2007). For each antenna, we measured the maximum amplitude of the averaged responses to 10 standard stimuli that mimic the pulse phase of courtship song. The antennal response amplitude in cby^1/1 and in cby¹/Df(3R)BSC805 was strongly reduced compared with controls (Fig. 4 C). Further, whereas 95% of the control antennae showed a pulsatile response, many of the cby^1/1 and cby¹/Df(3R)BSC805 antennae lacked any response. These observations lead us to conclude that cby¹ mutants are partially defective in chordotonal-mediated mechanotransduction.

Because CBY was initially described as a negative regulator of Wg signaling in Drosophila (Takemaru et al., 2003), we carefully checked cby^1/1 flies for phenotypes associated with loss or gain of Wg signaling. We did not observe any expansion of engrailed expression domains (n = 21) nor any cuticular defects (n = 119) in offspring embryos from homozygous cby^1/1 male and female parents, which lack any parental or zygotic contributions (Fig. S2). Because Wg gain-of-function signaling leads to characteristic bristle phenotypes on the wing and the notum (Treisman et al., 1997), we also looked carefully at bristle number and distribution in cby^1/1 adult flies (n = 31) but found no significant alterations (Fig. S2 D). Together, these observations demonstrate that Drosophila CBY is needed for basal body–associated functions but not to mediate or regulate Wg signaling.

For a detailed understanding of the role of CBY in basal bodies, we examined ciliary ultrastructure in cby^1/1 flies and controls. CO units or scolopidia are composed of two or three neurons that each extends a cilium from the tip of their sensory process (Fig. 5, top left). EM of COs in adult fly legs or antenna showed that cilia were missing in 26% of the mutant scolopidia (Fig. 5, f2 and 3). When ciliary profiles were present, we observed characteristic ultrastructural defects that were never seen in controls. The most frequent defect was a reduced number of microtubule doublets (19%; Fig. 5, a2 and 3). In rare cases (1%), extra doublets were observed, or the symmetry of the axoneme was abnormal (Fig. 5, a4 and 5). The remaining 80% of cby^1/1 chordotonal cilia were ultrastructurally normal, which is in agreement with the relatively mild behavioral defects observed in cby^1/1 flies. We also investigated the TZ and basal body region in cby^1/1 flies. The two basal bodies in each chordotonal neuron are aligned along their center axes, one distal to the other (Fig. 5). However, basal bodies are embedded in electron-dense material, and individual microtubules cannot be visualized by EM. In control flies, all basal bodies of chordotonal neurons appear as doughnut-shaped profiles with fibers connecting the basal body to the dendritic membrane (Fig. 5, d1). In cby^1/1 mutant flies, some of the basal body structures were completely missing (22%; Fig. 5, d2 and 3). At the level of the TZ, which we believe to correspond to the Y link region (Fig. 5, cross-sections b and c), we could observe either completely interrupted symmetry (Fig. 5, b2 and 3) or more subtle discontinuities in the microtubule-associated structures of the TZ (black arrows in Fig. 5, c2 and 3). Discontinuities in the TZ could also be observed on longitudinal sections (white arrows in Fig. 5, e1 and 2) in addition to membrane bulges (black arrows in Fig. 5, e2 and 3) that were never observed in controls. Together, these results show that cby is required for correct axoneme assembly in Drosophila and that its absence leads to improper TZ formation. As we never observed missing basal bodies by γ-tubulin staining of ciliated neurons (unpublished data), the EM observations also suggest that basal bodies, though present in all neurons, are not positioned properly at the base of the cilia in some neurons.

Because ultrastructural defects were incompletely penetrant in cby^1/1 flies, we hypothesized that cby has a more subtle role in cilia biogenesis, as described for TZ proteins in other organisms (Ishikawa and Marshall, 2011). In C. elegans, TZ-associated proteins play an important role in admitting specific proteins into the cilia (Williams et al., 2011). Thus, we looked at the distribution of proteins known to traffic into the cilia. The first, NOMPB, is an IFT component (IFT88/Polaris) required for anterograde transport into the cilia (Han et al., 2003). We observed that NOMPB strongly accumulated in cby^1/1 embryonic cilia compared with discontinuous accumulation of NOMPB in control cilia (Fig. 6, A and B). Conversely, CG11356, the Drosophila ortholog of Arl13b, which couples anterograde and retrograde IFT (Caspary et al., 2007; Li et al., 2010b), showed reduced accumulation in cby^1/1 cilia compared with controls (Fig. 6, C and D). These results show that ciliary-directed protein trafficking is altered in cby^1/1 flies.

cby^1/1 males are hypofertile (Fig. S1), and CBY is associated with the maturing basal body during spermatogenesis (Fig. 3). To understand its function during spermatogenesis, we looked at the structure of mutant sperm cysts. Differential interference contrast microscopy of live squashes of Drosophila testis showed no cellular defects in spermatocyte-stage cysts; all were composed of 16 cells showing that cby is not needed for spermatogonial divisions (n = 12; unpublished data).

When cysts in cby^1/1 testes were examined by EM, we frequently observed (88%) a reduced number of spermatids in each mature cyst. Each set of 16 spermatocytes should give rise to 64 spermatids, and all control cyst sections do show 63–64 spermatid profiles (Fig. 7 A). An altered number of spermatids was sometimes seen in early spermatid cysts, suggesting a possible meiotic defect (four out of six cby^1/1 cysts had 62 or fewer spermatids, whereas five out of five control cysts had 63–64 cells; Fig. 7 B). But no significant variations were seen in the ratio of nuclei-to-mitochondrial (Nebenkern) derivatives in onion-stage cysts, even though some aberrant spermatids could sometimes be observed (two or three aberrant cells in 3 out of 14 cysts; unpublished data). Hence, meiosis defects do not seem to be prevalent in cby^1/1 testis.

In early cby^1/1 spermatid sections, mitochondrial derivatives with no axoneme could be observed (2%; n = 294; Fig. 7 B, white arrow) as well as axonemes with no visible mitochondria derivatives (3.4%; n = 294; Fig. 7 B, arrowhead), which were never observed in controls. We also observed partially penetrant (14%; n = 294) but reproducible defects of the axonemal ultrastructure. Incomplete axonemal structures were found (white arrows in Fig. 7, C and D) with a reduced number of microtubule doublets. Because these defects were visible in early spermatids, we hypothesize that CBY is required for normal axoneme assembly. Altogether, our observations indicate that CBY plays an important function during spermatid differentiation in Drosophila.

Because UNC is a basal body–associated protein that is also required for axonemal assembly in Drosophila, we looked at UNC distribution in cby mutant testis. Remarkably, we observed that UNC deposition during spermatocyte maturation was severely affected, with UNC being deposited on a longer and more variable length of the elongating centrioles compared with controls (Fig. 8, A and B). The paired elongating centrioles in cby^1/1 testes often showed unequal extents of UNC protein (Fig. 8 B, white arrow) in contrast to the symmetrical UNC deposition in the controls. This aberrant UNC distribution is maintained in early spermatids (Fig. 8, C, D [white arrows], E, and F). Later, during spermatid differentiation, UNC labeling at the basal body was reduced in cby^1/1 testis compared with controls (Fig. 8, G and H). In contrast, no differences were observed between cby mutants and controls in γ-tubulin labeling (unpublished data). Hence, altered UNC deposition could either reflect a direct action of CBY on UNC deposition or indirectly reflect a more general basal body formation defect. Altogether, these results show that CBY is required for proper centriole-to–basal body conversion in Drosophila spermatocytes.

We show that cby plays an important function in the two Drosophila cell types in which centrioles mature to basal bodies. In sensory cilia, CBY is required for proper TZ assembly and modulates protein trafficking into or out of the cilium. In the testis, we observed that CBY is required for centriole conversion to a basal body and for proper completion of spermatogenesis. In conclusion, our work demonstrates that CBY has an ancestral function in basal body biogenesis. Notably, we find no evidence for a CBY function in Wg signaling in Drosophila.

CBY is associated with the TZ in cilia. The TZ is a molecularly and ultrastructurally distinct axoneme domain located just distal to the basal body. The ultrastructure of the TZ differs between primary cilia and motile cilia and also between species (for a comprehensive review, see Fisch and Dupuis-Williams [2011]). Nevertheless, several components of the TZ are conserved in many ciliated organisms. Several groups of TZ proteins are associated with human diseases, such as nephronophthisis (NPHP), MKS, and Joubert syndrome (Hodges et al., 2010). TZ proteins can be grouped in functional modules by their pattern of cooccurrence across species and because mutations in each group cause similar sets of symptoms in humans. The MKS functional module is conserved in most of the ciliated organisms, but the NPHP module is not found in all species and in particular not in Drosophila (Hodges et al., 2010). We show here that CBY is a novel component of the TZ, which colocalizes with CG14870 (B9D1/MKSR1), a component of the MKS module (Bialas et al., 2009). CBY is found in animals with motile cilia but not in C. elegans, suggesting a specific function in TZs of motile cilia in animals. In agreement with this hypothesis, cby mutant mice show defects in biogenesis of motile cilia as a result of defective docking of the basal bodies at the apical membrane (Voronina et al., 2009). However, in Drosophila, CBY is found at the tip of the basal bodies, both in cells with motile cilia or flagella (chordotonal neurons and spermatids) and also in neurons that innervate external sensory (ES) organs such as bristles. ES cilia lack the canonical axonemal cytoskeleton, and dynein arms are tightly ensheathed by a surrounding matrix and, so, are unlikely to be motile. We do not know whether CBY has any function in ES cells. No behavioral defects were seen when cby mutant larvae were challenged with chemical or olfactory stimuli (unpublished data), nor did mutant adults show the severe uncoordination that characterizes bristle loss-of-function mutations, suggesting that CBY is not essential for the function of ES cilia. CBY may be more essential in motile CO cilia than in ES organs. Or, the moderate structural defects in cby mutant axonemes may be more consequential for the function of motile cilia than for static ES cilia. Perhaps, as for the MKS module proteins in nematode sensory cilia (Williams, et al., 2011), ES phenotypes may become evident when other TZ proteins are also mutated.

In mouse, a ciliary requirement for CBY may also be limited to motile cilia because cby-deficient embryos, unlike mutants with complete ciliogenesis defects, survive until birth and mostly show defects in airway motile ciliogenesis. However, we observed that CBY was localized at the TZ of primary cilia in IMCD3 cells (Fig. S1), suggesting a more general function of CBY at the TZ. A function for CBY in primary cilia could explain the observation that cby mutant mice show defects in lung development that are thought to result from WNT signaling defects (Love et al., 2010). In agreement with a function for CBY in primary cilia, injection of a truncated dominant form of CBY in LLC-PK1 cells reduces the number of cells with primary cilia (Hoffmeister et al., 2011).

Several TZ proteins have been shown to be required for protein trafficking into the cilia and for IFT-dependent processes. This has clearly been demonstrated for NPHP1, NPHP4, and NPHP6 in C. elegans for several of the MKS-associated proteins and for NPHP1 in mammalian photoreceptors (Jauregui et al., 2008; Jiang et al., 2009; Williams et al., 2011). We show here that at least two proteins involved in IFT are abnormally distributed in cby^1/1 chordotonal cilia, NOMPB/IFT88 accumulating in cilia to higher levels than normal, and Arl13 being depleted. This could either result from an altered balance between anterograde and retrograde IFT transport or from altered gating of proteins into the cilia. We do not know precisely all the TZ structures that are affected in cby^1/1 cilia, but we observe some defects at the level of what likely corresponds to the Y-linker region of the TZ also called champagne glass structures (Fisch and Dupuis-Williams, 2011). In C. elegans, mutants lacking MKS/MKSR/NPHP proteins show missing connecting links, suggesting that these proteins are either associated with components of the connecting Y links or are required for their stability (Williams et al., 2011). In addition, nematodes with defects in MKSR1 and NPHP4 show ultrastructural defects of the cilia, such as a reduction in the number of microtubule doublets, that are similar to the defects observed in cby^1/1 flies (Williams et al., 2011). Altogether, these observations suggest that CBY plays a function at the TZ similar to the one of MKS and NPHP complexes. MKS proteins belonging to the B9 domain family (MKS1, B9D1, and B9D2) are conserved in Drosophila, but most of the NPHP proteins are not. Conversely, CBY is not found in C. elegans, where most of the NPHP proteins are conserved. Hence, these different modules likely operate independently. Such variations in the architecture and function of the TZ between organisms still need to be understood.

Only few other proteins associated with the TZ in Drosophila have been functionally analyzed. Among them, UNC is also a coiled-coil–containing protein but is found in only Dipteran species, where it diverged rapidly. It has also been suggested that UNC could be functionally equivalent to human OFD1, a centrosomal and basal body coiled-coil protein mutated in the ciliopathy oro-facial-digital syndrome 1 (Baker et al., 2004). We observed that loss of CBY function affects UNC protein accumulation during centriole-to–basal body conversion in spermatocyte. This could reflect an altered basal body structure in cby^1/1 testes but could also reflect that UNC and CBY functionally interact during TZ assembly. In support of this latter hypothesis, the basal bodies in unc^−/−-deficient sensory neurons show defects similar to cby mutant flies with broken basal body/TZ rings, missing cilia, or disrupted cilia (Baker et al., 2004). unc^−/− flagella also have incomplete axonemes like in cby^1/1 testis. Thus, UNC and CBY could work together in a functional module required for centriole maturation in Drosophila testis. The unc^−/− phenotype is more penetrant than the cby^1/1 phenotype, suggesting that unc is more limiting than cby for basal body/TZ assembly.

Interestingly, a novel protein, DILATORY (DILA), was identified recently as interacting with UNC in Drosophila chordotonal neurons (Ma and Jarman, 2011). This suggests that CBY, UNC, and DILA may function together at the TZ. However, in chordotonal neurons of dilatory mutants, ultrastructural defects were restricted to the cilia but did not apparently affect the TZ. Moreover, no flagellar axonemal defects are associated with dila mutations (Ma and Jarman, 2011), suggesting that CBY acts differently from dila in basal body function. Investigating the genetic and biochemical interactions between cby and dila will help dissect the pathway by which basal bodies mature in Drosophila and in other animals in which both cby and dila are conserved.

We observed that CBY, like UNC and DILA, is deposited at the tip of the maturing centriole in Drosophila spermatocytes. The maturing centrioles are associated with cytoplasmic vesicles before reaching the plasma membrane (Tates, 1971; unpublished data), where they grow small cilium-like extensions. Altogether, these observations suggest that a structure molecularly similar to the TZ assembles during centriole maturation in Drosophila spermatocytes and is maintained during meiosis and spermatid elongation. Indeed, both UNC and CBY are found to be associated with the tip of the growing axoneme, suggesting that this TZ-like structure is maintained during flagellar assembly. TZ assembly does not require IFT (Williams et al., 2011), which is consistent with the observations that IFT is not required for Drosophila spermatogenesis (Han et al., 2003; Sarpal et al., 2003). Therefore, we propose that the primary cilium-like structure assembled in Drosophila spermatocytes more likely corresponds to an elongated TZ from the maturing centriole, which hence behaves like a basal body. Therefore, we hypothesize that flagellar growth in Drosophila is similar to TZ assembly, which would explain also why flagella assembly is IFT independent in Drosophila.

Interestingly, in UNC-deficient spermatocytes, the primary cilia/TZs are shorter, showing that UNC is involved in elongating these primary cilia/TZ (Baker et al., 2004). Whereas primary cilia-like extensions were easily observed in wild-type testis, we did not observe any centrioles that were anchored to the plasma membrane in cby^1/1 flies. We also did not manage to observe enough centrioles to describe precisely their distal ultrastructure. However, because UNC deposition is severely affected in cby^1/1 spermatocytes, CBY appears to be required to control either directly or indirectly the assembly of the TZ in these cells.

The function of the spermatocyte primary cilia is unknown. Centrioles are dispensable for postembryonic development in Drosophila (Basto et al., 2006) but are required for male meiosis, as observed in sas-4 or sak/plk4-deficient flies (Rodrigues-Martins et al., 2008; Riparbelli and Callaini, 2011), even though spermatids can still form without axonemes (Bettencourt-Dias et al., 2005; Basto et al., 2006). BLD10 is required in Drosophila both for centriole elongation spermatocytes and to complete meiosis, suggesting a possible meiotic requirement for centriole elongation (Mottier-Pavie and Megraw, 2009; Carvalho-Santos et al., 2010). However, no precise description of the primary cilia ultrastructure in spermatocytes has been described in Bld10 mutants. Among the other proteins that are required for basal body biogenesis in Drosophila (Gogendeau and Basto, 2010), only UNC is associated with specific defects in spermatocyte primary cilia. Therefore, it will be interesting to examine in more detail primary cilia assembly in mutants for other TZ proteins and to finely examine meiosis and early spermatid differentiation defects.

The reduced number of early spermatids in cysts and aberrant mitochondrial derivative numbers suggest that cby might play a role in meiosis. We know that this phenotype is not a result of improper centriole distribution, as γ-tubulin staining showed that all spermatocytes and spermatids had the right number of centrioles. Our results could also implicate CBY in early spermatid survival or in elongation, which would lead to a reduced number of spermatid profiles in some cyst sections. In Drosophila, spermatid elongation relies on microtubule-dependent mitochondria remodeling (Noguchi et al., 2011). CBY is transiently associated with mitochondria in onion-stage and early elongating spermatids. Spermatids do elongate in cby^1/1 flies, but we do not know whether elongation is as complete as in control testis. The function of CBY in mitochondria during spermatid elongation needs to be more thoroughly investigated.

In mammals, CBY has at least two functions. Analysis of cby^−/− mice demonstrates that cby is required for basal body anchoring during multiciliated cell differentiation, illustrating that cby has conserved this anchoring function throughout evolution. Experiments in cell cultures indicate that CBY is involved in WNT signaling in different cell types, essentially by controlling nuclear–cytoplasmic shuttling of β-catenin, the WNT signaling pathway effector (Takemaru et al., 2003, 2009; Li et al., 2007, 2008, 2010a; Singh et al., 2007; Mokhtarzada et al., 2011). CBY directly interacts with β-catenin, and WNT signaling defects have been described during lung development of cby knockout mice (Love et al., 2010). A previous RNAi-based study suggested that cby could also be involved in Wg signaling in Drosophila (Takemaru et al., 2003). However, Wg signaling and development appear normal in cby^1/1-null flies, ruling out a function of cby in Wg signaling in Drosophila. Off-target RNAi effects may have caused the observed defects in the previous study. We do not know whether CBY interacts with β-catenin in Drosophila, but, if so, such an interaction would have consequences only in the ciliated cells in which cby is expressed.

Therefore, we conclude that CBY function in WNT signaling has either been acquired during chordate evolution or lost in the Dipteran lineage. In mammals, CBY interacts with cis-Golgi–associated proteins and is implicated in polycystin-2 distribution in cilia, even though this capacity remains controversial (Hidaka et al., 2004; Hoffmeister et al., 2011). A function in vesicular transport from the Golgi is consistent with CBY function in TZ assembly, which starts when centrioles fuse to vesicular membranes likely derived from the Golgi apparatus inside the cytosol. CBY could be required early for basal body–to-membrane attachment and also for vesicular trafficking to the cilia. Whereas CBY is not found associated with most of the cis-Golgi apparatus in cultured mammalian cells, a small pool of the cis-Golgi protein GMAP210 localizes next to CBY at the base of motile cilia (Fig. S3). This is compatible with a possible function in cis-Golgi vesicular trafficking to the cilia, as demonstrated for polycystin-2 transport (Hoffmeister et al., 2011).

Together, our data demonstrate that CBY is required for the assembly of the ciliary TZ in Drosophila both in sensory neurons and in male germ cells.

Comparison of cby cDNA and genomic sequences in different Drosophila strains showed a consistent insertion of two base pairs in the 3′ end of the cby coding sequence compared with the gene annotation on the FlyBase database. The consequence of this insertion (in italics) is a slight change in the C-terminal protein sequence starting at amino acid 121. Five amino acids are different (italic) and are followed by eight additional amino acids at the C terminus: 5′-CAT­GCGACGGAACT­AAGTGAGCTGA­AGCCAAAGGAA­AAGTGA-3′ (H A T E L S E L K P K E K *).

BLAST searches used as a query the cby gene sequence from FlyBase, modified as described in the previous section, using the basic local alignment search tool from the National Center for Biotechnology Information. The sequences obtained were then aligned using SeaView software.

All primer sequences are described in Table S1.

A 1,350-bp fragment including cby coding sequences and upstream regulatory sequences was amplified by PCR from wild-type Drosophila genomic DNA using the following primers designed from the noncorrected cby sequence: Cby-pro5 and Cby-pro3/BamHI. The resulting PCR fragment was cloned into the EcoRI and BamHI sites of the pW8-GFP plasmid (Loppin et al., 2005), in frame with the Gfp sequence.

The Tomato sequence was obtained by PCR on the Zeo-TomatoNT-2 (a gift from R. Basto, Institut Curie, Paris, France) with the introduction of an AgeI site at the 5′ end and a STOP and NotI site at the 3′ end. The resulting PCR fragment was cloned in the BglII and NotI sites of the pattB plasmid. The complete cby 3′ untranslated region (UTR) was cloned downstream of the Tomato sequence in the pattB-Tomato plasmid obtained above using XhoI and XbaI restriction sites introduced in the 3′ UTR PCR fragment.

cby coding and upstream regulatory sequences (1,375 bp) were amplified using the Cby-pro5 forward primer and a Cby-PRO3/AgeI reverse primer based on the corrected cby sequence with the addition of an AgeI site at the 3′ end. The resulting PCR product was cloned in the EcoRI and AgeI sites of the pattB-Tomato-3′cbyUTR plasmid in frame with the Tomato sequence. Transgenic lines were established by P element–mediated germline transformation (Spradling, 1986) or by phiC31-mediated germline transformation for the pattB plasmid–derived constructs (Bischof et al., 2007).

A 2,942-bp fragment covering CG14870 coding and upstream regulatory sequences was amplified by PCR on wild-type Drosophila genomic DNA with primers F-14870/BamHI and R-14870/NotI. The resulting PCR product was cloned in frame with the 6xMycTag in the modified pattB plasmid containing the 6xMycTag and SV40polyA sequences. Transgenic lines were obtained from BestGene Inc.

A 4,230-bp fragment covering CG11356 coding sequences and upstream regulatory sequences was amplified by PCR on wild-type Drosophila genomic DNA with the primers CG 11356-PRO3 and CG 11356-PRO5. The resulting PCR fragment was cloned into the pW8-Gfp plasmid. Transgenic lines were obtained by P element transformation.

The cby gene was disrupted by ends-out homologous recombination (Maggert et al., 2008). The left and right arms of the targeting construct were amplified by PCR on genomic DNA of the y[1] w[*]; P{ry[+t7.2] = 70FLP}11 P{v[+t1.8] = 70I-SceI}2B noc[Sco]/CyO, S[2] stock using primers F-3′cby/SphI, R-3′cby/NotI, F-5′cby/BsiWI, and R-5′cby/AscI. The resulting construct enabled the deletion of most of exon 2, including the stop codon, and the acceptor splice site of intron 1 (Fig. 4 A). All the coding regions from the cloned 5′ and 3′ homologous fragment were verified by sequencing. Transgenic lines were established by phiC31-mediated germline transformation (Bischof et al., 2007) and used for homologous recombination, as previously described (Maggert et al., 2008).

We thank J. Schmitt and P. Morales (both from Centre de Génétique et de Physiologie Moléculaire et Cellulaire, Université Claude Bernard Lyon 1, Villeurbanne, France) for Drosophila medium and stock maintenance. The chromosomes and alleles P{unc-GFP}^34a, nompB¹, and P{GFP:nompB} were previously described (Han et al., 2003; Baker et al., 2004). P{unc-GFP}^34a is inserted on the second chromosome and encodes the entire UNC protein fused to GFP in the C terminus. nompB¹ is a loss-of-function allele of nompB that deletes 1,100 bp of exon 3 and part of exon 4. P{GFP:nompB} is inserted on the third chromosome and encodes the full-length NOMPB protein fused to GFP at its N terminus. {Ubq55:RFP:sas4} was a gift from R. Basto. Df(3R)BSC805 was obtained from the Bloomington Stock Center and uncovers the 93F13-94A6 cytological interval. All the other following fly strains were constructed in the laboratory: w; P{cby::GFP}^F23/CyO; P{cby::GFP}^M27, w; P{cby::GFP}^F23/CyO; rfx²⁵³ e/TM3, P{GAL4-twi.G}^2.3, P{UAS-2xEGFP}^AH2.3, w; P{cby::GFP}^F23/CyO; rfx⁴⁹/TM3, P{GAL4-twi.G}^2.3, P{UAS-2xEGFP}^AH2.3, w; Bl/CyO; P{cby::Tomato}attP62E1^M1F2/TM6B, w; P{cby::Tomato}attP62E1 rfx²⁵³ e/TM6B, st rfx¹⁶ e ca/TM6B Tb Dr, w; cby¹/TM3, P{GAL4-twi.G}^2.3, P{UAS-2xEGFP}^AH2.3, w; P{cby::GFP}^F23/CyO; cby¹/TM6B, w; P{cby::GFP}^F23/CyO; P{cby::GFP}^M27 cby¹/TM6B, w; P{unc-GFP}^34a/CyO; P{cby::Tomato}attP 62E1^M1F2/TM6B, w; P{unc-GFP}^34a/CyO; cby¹/TM3, P{GAL4-twi.G}^2.3, P{UAS-2xEGFP}^AH2.3, y w P{CG11356::GFP}^F6M1/FM7, w, w; P{GFP::nompB} cby¹/TM3, P{GAL4-twi.G}^2.3, P{UAS 2xEGFP}^AH2.3, w; P{CG14870-myc}attP59D3/CyO; cby¹/TM3, P{GAL4-twi.G}^2.3, P{UAS 2xEGFP}^AH2.3, and w; P{CG14870-myc}attP59D3/CyO; P{cby::Tomato}attP62E1^M1F2/TM6B.

Extracellular compound potentials were recorded with tungsten electrodes placed in the antenna and head, as previously described (Eberl et al., 2000; Eberl and Kernan, 2011). Each fly was mounted in a micropipette tip and its proboscis fixed in wax so that its antennae were exposed and free to move. A computer-generated sound stimulus (five pulses of a 500-Hz sine wave at 35-ms intervals) was amplified by an audio amplifier (Realistic MPA-30; RadioShack) and an 8-in, 4-Ω speaker and brought to the fly via 0.25-in Tygon tubing, with its opening placed close enough to the antennae to ensure near-field sound conditions. To record antennal nerve potentials, one electrolytically sharpened tungsten wire electrode was placed between the first and second antennal segments and a second in the head. The differential alternating current signal from these was amplified (5,000× gain) by a BMA-200 Bioamplifier (CWE, Incorporated) and digitized and analyzed using an Instrunet 100B A/D converter and SuperScope II software (GW Instruments, Inc). The same stimulus amplitude was used for all recordings. Flies were chilled briefly at 0°C for mounting. Flies that were sorted under CO₂ anesthesia were allowed to recover for >40 h before recording. Mutant (cby^1/1 or cby¹/Df) flies were recorded alternately with their age-matched control sibs. For each antenna, we measured the maximum amplitude of the averaged responses to 10 sound stimuli. P-values were calculated from the Mann–Whitney rank sum test, with all responses lacking peaks tied at the lowest rank.

The climbing test was performed on female flies of the same age. 30 flies were placed in graduated tubes and banged on the table at t = 0. Pictures were taken every minute for 5 min and analyzed by counting the number of flies present in each interval every minute and for each genotype.

Late pupae were dissected once in PBS 1×. The dissected testes were transferred on a slide in a drop of PBS 1× containing 8.3 mg/ml Hoechst 33342 and then squashed under a coverslip before being observed live using a microscope (Imager Z1; Carl Zeiss).

Precisely staged embryos were sorted and collected in PBS 1×/0.1% Triton X-100 and dechorionated in bleach. Embryos were rinsed in PBS 1×, extracted from the vitelline membrane using a tungsten needle, transferred to a polylysine-coated slide, and dissected. Embryos were fixed in 4% PFA/PBS 1× (20 min), rinsed, and blocked in PBT (PBS 1×, 5% BSA, and 0.1% Triton X-100). Embryos were incubated with primary antibodies diluted in PBT washed in PbT (PBS 1×, 0.1% BSA, and 0.1% Triton X-100) and incubated with secondary antibodies diluted in PbT. Embryos were washed in PTW (0.1% Tween 20 in PBS 1×) and mounted (Cytomation; Dako).

Whole-mount testis preparation was performed as follows: testes were fixed 15 min in 4% PFA/PBS 1× followed by a 15-min treatment in PBS 1× and 0.1% Triton X-100, blocked in 1% BSA in PBS 1×, and incubated with primary antibodies overnight at 4°C. Testes were incubated with secondary antibodies for 2 h at room temperature. Testis squashes were performed as follows: testis (0–2-d adults) were fixed 20 min at room temperature in 3.7% PFA/PBS 1× and flattened under a coverslip on a microscope slide pretreated with 10% polylysine solution. Slides were quick frozen in liquid nitrogen, and coverslips were removed, fixed in chilled 95% ethanol for 5 min, washed twice in PBT (PBS 1× containing 0.1% Triton X-100), and blocked for at least 1 h in PBTB (PBT with 3% BSA) at room temperature. Samples were incubated overnight at 4°C with primary antibodies diluted in PBTB and 2 h at room temperature in secondary antibodies in PBTB. Testes were mounted in mounting medium (Cytomation).

The antibodies used are as follows: rabbit anti-GFP (1:500; Invitrogen), anti-HRP (1:3,000; Jackson ImmunoResearch Laboratories, Inc.), anti-ASL (1:500; provided by R. Basto), anti-DPLP (1:1,000; provided by R. Basto), anti–γ-tubulin (used on embryos; 1:400; Sigma-Aldrich), mouse monoclonal anti-GFP (1:1,000; Roche), anti-22C10 (1:250; provided by S. Benzer, California Institute of Technology, Pasadena, CA), anti-myc (1:3,000; Euromedex), and anti–γ-tubulin (used on testes; 1:100; Sigma-Aldrich). The secondary antibodies used, diluted at 1:1,000, are as follows: goat anti–mouse Alexa Fluor 488 conjugated, 546 conjugated, and 647 conjugated; goat anti–rabbit Alexa Fluor 488 and 555 conjugated (Invitrogen); and goat anti–rabbit Cy5 conjugated (GE Healthcare).

Slides were analyzed at room temperature on a microscope (Imager Z1) equipped with 20× Plan-NEOFLUAR (0.5 NA) and 40× Plan-NEOFLUAR (0.75 NA) objectives or a 63× Plan-NEOFLUAR (NA 1.25) objective. Epifluorescence images were acquired with a charge-coupled device camera (CoolSNAP HQ2, Roper Scientific) and MetaView software (Molecular Devices). Confocal stacks were acquired with a confocal microscope (LSM 510 META; Carl Zeiss) and software equipped with 63× Plan-Apochromat (1.4 NA), 40× Plan-NEOFLUAR (1.3 NA), and 25× Plan-NEOFLUAR (0.8 NA) objectives. Image brightness and contrast were adjusted by using ImageJ (National Institutes of Health), and separate panels were assembled with Photoshop software (CS4; Adobe). Capture times and adjustments were identical for compared images on one figure. γ was set to 1 in all figure panels.

Antennae, legs, and testes were dissected in PBS 1× and fixed in 2% glutaraldehyde, 0.5% PFA, and 0.1 M Na-cacodylate, pH 7.4, for 48 h at 4°C. After three washes of 15 min in 0.15 M sodium cacodylate, pH 7.4, samples were postfixed in 1% OsO4 for 4 h for antennae and 1 h for testes. Samples were then dehydrated through ethanol series and propylene oxide and embedded in Epon medium (Fluka). Ultrathin sections were cut on an ultramicrotome (Leica). Sections were contrasted in Ultrastainer (Leica) in aqueous uranyl acetate and lead citrate. Contrasted sections were observed on a transmission electron microscope (CM120; Philips). Images were acquired with a 2k × 2k digital camera (ORIUS 200; Gatan) and digital micrograph software. Images were processed with Photoshop software (CS4). Only contrast and brightness were adjusted.

Mouse primary ependymal cell cultures were derived from mouse newborn brains (OF1 strain), as previously described (El Zein et al., 2009). IMCD3 cell lines (a gift from P. Goggolidou, Mammalian Genetics Unit, Medical Research Council, Harwell, Oxfordshire, England, UK) were cultured in DME/F12 medium containing 10% FBS, 100 U/ml penicillin, 100 µg/ml streptomycin, and 1× nonessential amino acids (reagents from GE Healthcare). Ependymal and IMCD3 cells were seeded on polylysine- and laminin-coated glass coverslips. Confluent ependymal cells were fixed the day of serum deprivation or after 12 d of serum deprivation, whereas IMCD3 cells were fixed after 2 d of culture in 0.5% FBS containing medium. Cells were fixed in methanol for 5 min at −20°C or in 4% PFA for 15 min at room temperature when using the anti-Arl13b primary antibody. After permeabilization in 0.1% Triton X-100 for 10 min, cells were blocked 2 h in 5% FBS in PBS 1× followed by an overnight primary antibody incubation using the following: mouse anti-Cby antibody (1:200; Santa Cruz Biotechnology, Inc.), rabbit anti-Arl13b (1:500; a gift from T. Caspary, Emory University, Atlanta, GA), rabbit anti–γ-tubulin (1:100; Sigma-Aldrich), and rabbit anti-GMAP210 (1:4,000; a gift from M. Bornens, Institut Curie, Paris, France). Immunostaining was revealed with goat anti–mouse Alexa Fluor 488 and goat anti–rabbit Alexa Fluor 555–conjugated antibodies (1:500; Invitrogen). The slides were mounted in the presence of DAPI or DRAQ5 in fluorescence mounting medium (Dako). Slides were visualized under a microscope (AxioImager Z1; Carl Zeiss) and a confocal microscope (LSM 510 META).

Total RNA of w¹¹¹⁸ and cby^1/1 testes was extracted using TRIZOL (Invitrogen) and treated with DNase (Invitrogen). cDNA was obtained using 1 µg of DNase-treated RNA, 200 ng of random primers (Promega), and 200 U of RevertAid H Minus M-MuLV reverse transcription (Thermo Fisher Scientific) in a final volume of 30 µl. 1 µl cDNA was used for PCR reactions.

Fig. S1 shows the molecular characterization of the cby¹ allele and quantification of cby^1/1 male fertility. Fig. S2 shows the absence of any Wg-associated phenotype in cby¹ flies. Fig. S3 shows the distribution of mouse CBY protein in IMCD3 or ependymal cells. Table S1 lists all primers used for creating reporter constructs and the cby¹ allele.

We thank Alice McGarry for technical assistance with electrophysiological recordings in flies. We thank Jérôme Schmitt and Patricia Morales for Drosophila medium and stock maintenance. We thank Tamara Caspary for the Arl13b antibody. EM was performed at the Centre Technologique des Microstructures of the Université Claude Bernard Lyon 1.

This work was supported by grants from the Fondation pour la Recherche Médicale (équipe FRM 2009), the Agence Nationale de la Recherche (Ciliopath-X), and the Région Rhône-Alpes (Cible 2008). C. Enjolras and F. Soulavie were supported by a doctoral fellowship from the Ministère de l’Enseignement Supérieur et de la Recherche and from the Région Rhône-Alpes, respectively. B. Chhin was supported by a postdoctoral fellowship from the Fondation pour la Recherche Médicale.