Asterless Is a Centriolar Protein Required for Centrosome Function and Embryo Development in Drosophila

Asterless Is a Centriolar Protein Required for Centrosome Function and Embryo Development in Drosophila Hanne Varmark, 1,6,7 Salud Llamazares, 2,6 Elena Rebollo, 2 Bodo Lange, 3 Jose Reina, 2 Heinz Schwarz, 4 and Cayetano Gonzalez 2,5, * 1Cell Biology and Biophysics Programme European Molecular Biology Laboratory Meyerhofstrasse D-69117 Heidelberg Germany 2Cell Division Group Institute for Research in Biomedicine (IRB-Barcelona) Parc Cientific Barcelona C/ Josep Samitier 1-5 08028 Barcelona Spain 3Department of Vertebrate Genomics Max Planck Institute for Molecular Genetics Ihnestrasse 73 D-14195 Berlin Germany 4Electron Microscopy Unit Max Planck Institute for Developmental Biology Spemannstr. 35 D-72076 Tubingen Germany 5Institucio Catalana de Recerca i Estudis Avanc¸ ats (ICREA)Passeig Lluis Companys 23 08010 Barcelona, Spain Summary Background: Centrosomes, the major organizers of the microtubule network in most animal cells, are com-posed of centrioles embedded in a web of pericentriolar material (PCM). Recruitment and stabilization of PCM on the centrosome is a centriole-dependent function. Com-pared to the considerable number of PCM proteins known, the molecular characterization of centrioles is still very limited. Only a few centriolar proteins have been identified so far in Drosophila, most related to centriole duplication. Results: We have cloned asterless (asl) and found that it encodes a 120 kD highly coiled-coil protein that is a constitutive pancentriolar and basal body component. Loss of asl function impedes the stabilization/maintenance of PCM at the centrosome. In embryos deficient for Asl, development is arrested right after fertilization. Asl shares significant homology with Cep152, a protein described as a component of the human centrosome for which no functional data is yet available. Conclusions: The cloning of asl offers new insight into the molecular composition of Drosophila centrioles and a possible model for the role of its human homolog. In addition, the phenotype of asl-deficient flies reveals that a functional centrosome is required for Drosophila embryo development. Introduction As the major organizer of the microtubule cytoskeleton in most animal cells, the centrosome provides essential functions required for cell proliferation, differentiation, and development [1–3]. Centrosomes consist of a pair of centrioles surrounded by a matrix of pericentriolar material (PCM) that contains the proteins involved in microtubule nucleation and other cellular processes regulated by the centrosome. A significant number of PCM proteins have been identified in different experimental species by genetic analysis, via antibodies raised against purified centrosomes, or by homology with centrosomal proteins identified in other species [4]. In Drosophila, examples include the Polo and Aurora-A kinases [5, 6], founding members of the Polo/Plk and Aurora families; CP190 and CP60 [7, 8]; Cnn [9, 10]; DTACC and Msps [11–13]; and components of the gTURC [14–18]. Data on the molecular composition of centrioles and the pathways that control their assembly are very limited [19]. Most of our current knowledge on centrioles comes from studies in C. elegans where genome-wide RNAi and genetic screens have identified a number of proteins essential for centriole duplication. These include SAS-4, SAS-5, SAS-6, ZYG-1, and SPD-2 [20, 21]. Some of these proteins seem to be conserved in evolution. Drosophila Sak and its human orthologs Plk4 are related to C. elegans ZYG-1 [22, 23], and DSas-4 and CenpJ/CPAP are the suspected orthologs of C. elegans SAS-4 in Drosophila and humans [24]. Orthologs of SAS-6 have also been identified in humans [25] and Drosophila [26]. Other proteins necessary for centriole duplication in Drosophila are Ana1 and Ana2 that were identified in an RNAi screen in S2 cells [27]. In addition to these, the only other known centriolar proteins in Drosophila are Unc and D-Plp, reported to be required mostly for ciliogeneis [28, 29]. The gene asterless (asl) was identified by B. Wakimoto in a screen for mutants that affect male fertility [30]. Cytological studies showed that in larval neuroblasts and spermatocytes mutant for asl, gTUB accumulation and aster nucleation were highly defective [30–32]. We have cloned asl and found that it encodes a constitutive pancentriolar protein. The asl mutant phenotype reveals that Asl function is required for PCM recruitment and that a functional centrosome is mandatory for embryo development. The asl gene encodes a large, highly coiled-coil protein that shares significant homology with Cep152, one of the proteins identified in a proteomic analysis of human centrosomes [33]. No functional data have yet been published on CEP152. The cloning of asl now offers a new tool to further characterize centriole function in Drosophila and a possible model for the role of its human homolog. Results Drosophila Asl Is a Highly Coiled-Coil Protein and Shares Homology with the Human Centrosomal Protein Cep152 The gene asl was originally mapped by meiotic recombination 0.03 cM proximal to Ki (47.3), within region 83A1-83D3 of the polytene chromosomes [30]. Three alleles of asl have been described: asl1 (or asl), asl2, and asl3, all three EMS induced [30]. As a first step toward cloning the asl gene, we decided to map it more precisely by P element-induced recombination of the asl1 allele in the male germline [34]. By taking advantage of a series of available P elements inserted in this region, we were able to narrow down the position of asl to a short stretch of 11.735 base pairs, between the insertion points of EP(3)3591 and EP(3)3713 (Figure 1A). Only two genes have been predicted to be encoded by this region: CG1427 and CG2919 (Figure 1A). No mutant alleles of these genes have been described. A fragment of a third gene, Rga, is also within this region, but it is truncated by EP(3)3713 and it is not allelic to asl. We therefore focused our attention on CG1427 and CG2919. We found that P[aslGR], a transgene carrying a copy of the genomic region that spans from 300 nucleotides upstream to 500 nucleotides downstream of the CG2919 coding sequence (Figure 1, aslGR), restored viability of the lethal asl transheteroallelic combinations, as well as male and female fertility in all sterile combinations (Table S1 in the Supplemental Data available online). Because only a truncated form of CG1427, which lacks all the 50UTR and the first 45 nucleotides of the ORF, is present in P[aslGR], the rescue of asl mutant phenotypes by the P[aslGR] transgene strongly suggested that CG2919 was the asl gene. To confirm this possibility, we tested another transgene, P[YFP-AslFL], which expresses a fusion between YFP and the coding sequence of CG2919 under the control of the poly-Ubiquitin promoter. We found that P[YFP-AslFL] was also able to rescue viability and fertility in all transheteroallelic combinations, including those over Df(3R)ED1557 that uncovers CG2919 (Table S1). An-other transgene, P[YFP-AslD44], similar to P[YFP-AslFL] but lacking the sequences encoding the first 44 amino acids and the first intron of CG2919, did not rescue any of the asl phenotypes. These results confirmed that Asl is encoded by CG2919. In extracts made from S2 cells or from embryos de-rived from wild-type females, western blotting with Rb5110, an antibody raised against the C-terminal 16 amino acid peptide of CG2919, recognizes one band of about 120 kD that corresponds to the MW predicted from the asl coding sequence and a second band of about 95 kD. A third band with a rMW of 54 kD is recognized in extracts from embryos, but not in S2 cells (Figure 1B, lanes 1 and 2). The 120 kD band is absent in embryos laid by asl1/Df(3R)ED1557 females (Figure 1B, lanes 3 and 5), as well as in embryos laid by females carrying any of the viable transheteroallelic combinations asl1/asl1, asl1/asl2, or asl1/asl3 (not shown). In embryos derived from transgenic asl1/ Df(3R)ED1557 females expressing the genomic rescue construct P[aslGR], the 120 kD band is restored to endogenous levels (Figure 1B, lane 4). In embryos derived from asl1/Df(3R)ED1557 females expressing the YFP-aslFL fusion, the 120 kD band is replaced by a slightly overexpressed, higher molecular weight band of the size expected for this fusion (Figure 1B, lane 5), which can also be observed, together with the 120 kD band in embryos derived from wild-type females expressing the YFPaslFL fusion (Figure 1B, lane 6). These results strongly suggest that the 120 kD protein recognized by the Rb5110 antibody is Asl. Upon sequencing of the genomic region of the asl1 and asl2 alleles covered by the P[aslGR] transgene, no significant polymorphism were identified. This is a surprising result given the fact that aslGR contains all sequences required to provide asl function. However, a 71 base pair deletion (from bp 2381 to 2452) was observed in the coding sequence of asl3. This deletion shifts the reading frame and intro-duces a premature stop codon at position 795. CG2919 encodes a 994 amino acid protein with six predicted coiled-coil motifs that span 86.7% of the protein (Figure 1C). Stretches of low sequence complexity link these domains. Comparison with public protein databases by BLAST identified Asl homologs in different Drosophila species like D. pseudoobscura (64% sequence identity) as well as in mosquito (A. gambiae; 32% identity; Figure 1D). PSI-BLAST search based on the cluster of highly conserved Asl homologs in insects identified the human centrosomal protein Cep152 [33] as an Asl homolog in humans (E = 6e-20; see Experimental Procedures). In addition, Drosophila Asl and human Cep152 were identified as putative orthologs by reciprocal best hit analysis with BLAST, as well as by the orthology prediction software of the Ensembl Genome Browser and the eukaryotic ortholog database Inparanoid (see Experimental Procedures). Identity throughout the entire protein sequence between Asl and Cep152 is 13% (Figure 1D). Asl Is a Constitutive Centriolar Protein To determine the subcellular localization of Asl, we first immunostained wild-type testes with the Asl antibody Rb5110. PCM and chromatin were counterstained with gTUB antibodies and DAPI. Figure 2A summarizes the main stages of PCM reorganization that take place through spermatogenesis. During the four rounds of mitosis in spermatogonia, two dots of gTUB signal, one at each side of the metaphase plate, reveal the PCM of each centrosome in these cells (Figure 2A, mitosis). At this stage, Asl signal is located at the core of each centrosome, surrounded by the PCM. During prophase of meiosis I, the PCM is dramatically reshaped and enlarged (Figure 2A, pro-I), appearing as two rods, about 2 mm in length each, joined to form a V-shaped structure. At this stage, Asl largely colocalizes with the PCM. A second reorganization of the PCM is observed during prometaphase, as the PCM coalesces back into a com-pact mass (Figure 2A, prometa-I). At this stage, Asl remains as a V-shaped figure, the vertex of which overlaps with the PCM. After completion of meiosis, early spermatids contain a spot of gTUB that is close to the nucleus (Figure 2A, spermatid) and is made of two do-mains of different gTUB concentrations, the proximal being more heavily labeled by the gTUB antibodies than the distal. In these cells, the Asl antibody stains one rod that spans the entire length of the low-gTUB density domain and partially colocalizes with the high-gTUB density domain. Asl was also detected in individualized, fully mature sperm (not shown) where, as previously reported, gTUB cannot be observed [35]. EM studies have shown that in spermatocytes, before meiosis, the centriole pair acquires a distinctive V shape that is large enough to be observed by light microscopy and remains such until the end of meiosis I, when each of the two rods are segregated apart in preparation for meisosis II [36]. Thus, throughout spermatogenesis, Asl colocalizes with centrioles and remains basal body bound in fully matured sperm. No signal could be detected by immunofluorescence with the Asl antibody Rb 5110 in spermatocytes from asl1 males (Figure S1). We then followed YFP-AslFL localization in embryos during early cleavage mitosis. At metaphase, YFP-AslFL (Figure 2B, yellow) reveals a core structure surrounded by PCM (Figure 2B, red). The centrosome cycle during these nuclear divisions, which proceed very rapidly without intervening gap phases, is somewhat different from the canonical cycle. In these syncytial cycles, centrosomes split apart and start to segregate by late anaphase [37]. At this stage, YFPAslFL labels two dots at the core of the replicating centrosomes (Figure 2B, double arrows). Asl also localizes to the centrioles throughout the cell cycle in larval discs and brains (not shown; [38]). To determine the precise localization of Asl at the ultrastructural level, we used immuno-electronmicroscopy. Three different experimental conditions were tested. First, we immunostained Schneider’s S2 cells with the Asl antibody Rb5110 (Figure 2C). Second, we immunostained sections of testes expressing YFP-AslFL with an GFP antibody (Figure 2D). Finally, we immunostained whole-mounted embryo centrosomes purified by centrifugation through sucrose gradients with the Asl antibody (not shown). All three tests revealed that Asl and the YFP-AslFL fusion are closely bound to centrioles, largely located on the periphery of the centriole barrel along its entire length. All centrosomes studied, whether by immunofluorescence or immuno-EM, contained Asl. Altogether, these data strongly suggest that Asl is a constitutive component of Drosophila centrioles and basal bodies and that the centriolar localization of Asl is not cell cycle dependent. FRAP analysis in living embryos expressing YFP-AslFL showed that photobleached centrioles recover endogenous levels of centriole-bound YFP-AslFL signal with a half-turnover rate of w5 min (Figure 2E). A similar rate of YFP-AslFL turnover was observed in larval neuroblasts (not shown). This dynamic behavior of Asl is markedly different from that of the other two centriolar proteins subjected to FRAP analysis so far in Drosophila. In the case of PACT-GFP, a fusion between the Pericentrin/AKAP450 centrosomal targeting (PACT) domain of Drosophila, FRAP of the PCM-bound signal occurs rapidly, with a half-turnover rate of 1–2 min, while the centriolebound signal recovers only in the following round of centriole replication [29]. The same applies to Unc-GFP that shows no sign of recovery 1 hr after photobleaching [28]. Asl Function Is Required for PCM Recruitment and Centriole Orientation To further characterize the function of Asl, we analyzed asl1 mutant spermatocytes expressing the centriolar marker PACT-GFP [29] and immunostained with anti-bodies against gTUB to reveal the PCM. In wild-type spermatocytes from prophase to prometaphase, PACT-GFP reveals the centrioles as two pairs of rods that partially overlap with the PCM (Figure 3A). In asl1 spermatocytes, where no Asl protein could be detected (Figure S1), the centrosomes were severely perturbed in two regards (Figure 3A). First, two PACT-GFP-deco-rated, V-shaped centriole pairs were rarely observed. Rather, the PACT-GFP signal in most asl1 spermatocytes appeared as one cluster of irregularly shaped structures (Figure 3A). In some unfixed, living asl1 spermatocytes where the PACT-GFP signal is sharper, these clusters seems to be made of rods like those seen in wild-type cells, but randomly arranged (Figure S2). Consistent with these observations, nonserial EM sectioning of these cells showed that most asl1 spermatocytes contained a single cluster of up to four centrioles (Figure 3B). Centriole ultrastucture was normal in most cells (Figure S3), with only a very small fraction of the sections showing minor alterations at this level (Figure S3). Therefore, the irregularly shaped centriolar material revealed by PACTGFP in asl1 spermatocytes, rather than reflecting gross structural centriole abnormalities, seem to correspond to groups of up to four clustered centrioles that fail to segregate and may have lost the geometric arrangement stereotypical of wild-type centriole pairs. Loss of D-plp has also been reported to compromise the orthogonal orientation of mother and daughter centrioles [29]. The centrosomes in asl spermatocytes were also abnormal in that the amount of PCM material associated with the centrioles was highly reduced: in 70% of the cells, the gTUB signal was below 25% of the average signal observed in control cells, the remaining 30% of the cells showing no significant accumulation of gTUB around the centrioles (Figures 3A and 3C). Such loss of PCM recruitment and the resulting failure to nucleate microtubules have direct consequences in spindle assembly (Figure 3C). In cells where the centrosomes do not recruit detectable amounts of PCM, microtubule organization is anastral and chaotic (Figure 3C, asterisk). In cells where some PCM is recruited, loss of centrosome segregation results in either one single aster (Figure 3C, arrow) or two (Figure 3C, double arrows) that remain very close to each other. In a few instances, the asters separate and bipolarity is established (Figure 3C, arrowheads). Thus, the expressivity of the asl1 phenotype varies from virtually acentrosomal cells to cells that contain two segregated asters before NEB. However, even in those cases where PCM was recruited and microtubule asters were organized, the asters were always much smaller and less dense than in control cells. Video recording in living asl1 spermatocytes expressing GFP-a-tubulin revealed that in the few cells that initiated spindle assembly with well-separated centrosomes, biastral bipolar spindles were organized (Figure S4). In the case of cells with nonseparated centrosomes, the microtubule arrays got organized into a monastral spindle. Monastral spindle figures were the main spindle type observed in asl1 spermatocytes (Figure S5). Thus, the abnormal centrosome function caused by the asl mutation results in reduced microtubule nucleation and severe defects in meiotic spindle assembly, which in turn results in a high incidence of aneuploidy (Figure S5) [30]. A Functional Centrosome Is Required to Initiate Development Fertilization contributes the first centriole of the developing Drosophila embryo via the sperm basal body [37]. We found Asl to be a constitutive element of the sperm basal body and hence to be paternally contributed (not shown). However, Asl is also maternally provided in quantities that significantly exceed the amount of paternal centriole-bound Asl. This fact, together with the high turnover rate of centriolar Asl that we have documented before, results in the fast replacement of the basal body-bound Asl by the maternal pool. Thus, in unlabelled eggs fertilized by males expressing the YFP-AslFL fusion, the basal body loses its YFP signal soon after fertilization and becomes untraceable by fluorescence microscopy (not shown). Likewise, when YFP-AslFLexpressing females are fertilized by wild-type males, the basal body incorporates YFP-AslFL immediately after sperm entry. This is illustrated in Figure 4A in which the basal body-derived centriole can be seen as a rod-like YFP-AslFL-labeled structure, significantly larger than the dotshaped newly synthesized centriole. In the following telophase, after a further round of replication, the rod-like, basal body-derived centriole can still be seen together with the three dotlike new centrioles (Figure 4B). Indeed, the long centriole that originates during spermatogenesis perdures and serves as a centriole beyond the first zygotic mitosis (not shown). These observations show that Asl is associated with centrioles from the first stages of development and that paternal basal body-bound Asl is quickly ex-changed with the maternally provided Asl pool. We then decided to determine whether Asl plays a function during these first stages of zygotic develop-ment. After sperm entry and activation, female meiosis is resumed and four haploid nuclei are produced in wild-type eggs [37]. The most internal of these nuclei, which is the closest to the sperm nucleus, becomes the female pronucleus, while the others, usually clustered into a polar body, remain inactive and eventually disappear. Recruitment of PCM around the paternally contributed centriole results in the assembly of the first zygotic centrosome (Figure 4C) and in the organization of a prominent aster, which is thought to mediate male and female pronuclear fusion. Soon afterwards, the duplicated centrosomes migrate apart over the male pronucleus, and fusion with the female pronucleus takes place (Figure 4D). The first mitotic spindle is then assembled (Figure 4E), and repeated rounds of nuclear division cycles result in the exponential proliferation of syncytial nuclei (Figure 4F). In embryos derived from asl1/ Df(3R)ED1557 mothers, female meiosis proceeds normally: distinct polar bodies and female pronuclei can be identified, sperm entry takes place, and upon fertilization, all five nuclei are present and arranged in a seemingly wild-type configuration. Cnn accumulates at a point near the male pronucleus, presumably around the paternally provided basal body that is wild-type for Asl, but a functional MTOC is not organized, develop-ment is brought to a halt, and the first zygotic mitosis never occurs (Figure 4G). Instead, nonfunctional, anastral spindle-shaped structures are organized around the chromatin (Figure 4H). These spindles, which do not segregate chromosomes, persist in embryos aged for 1–2 hr (Figures 4I and 4J), a stage at which wild-type embryos contain hundreds of nuclei (Figure 4F). Thus, maternal Asl is needed to facilitate the centrosome function required for initiation of cleavage cycles in the fly. This terminal phenotype is indistinguishable from the phenotype of embryos derived from gTUBTW1 homozygous females, which lack the maternal gTUB37C gene [39]. Discussion Asl Is a Conserved Centriolar Protein that Mediates PCM Assembly We have found that the gene previously identified by genetic analysis as asl corresponds to CG2919. By using several fluorescence microscopy and EM assays, we have found Asl to be a centriolar protein. The number of centriolar proteins identified so far in Drosophila is very limited. Aside from Unc (whose expression is restricted to certain developmental stages and is specifically required for basal body function [28]) and the PACT domain-containing protein D-plp (which is found in the PCM as well and whose main function seems to be ciliogenesis [29]), all centriolar proteins identified so far in Drosophila are mainly involved in centriole duplica-tion. Such is the case of a number of proteins recently identified, including D-SAS4, D-SAS6, Ana1, Ana2, and SAK/PLK4 [22–24, 26, 27]. Asl is a ubiquitous constitutive component of centrioles and basal bodies. Even in cell types such as the developing oocyte, where centrioles are eventually disposed of, Asl remains centriole bound until the last traces of these organelles disappear. No significant accumulation of Asl is detected in the PCM or at the pole of mitotic spindles. Moreover, no Asl is found in the poles of female meiosis I or II spindles, which are known to be acentriolar, and none of these meiotic spindles is affected by loss of Asl function. Finally, loss of Asl severely diminishes PCM stabilization around centrioles. Aside from their role as basal bodies in cells with cilia or flagella, clustering of the PCM at the centrosome is the main function assigned so far to centrioles [40]. Asl is one of the first identified molecular components that mediate this key process. Asl contains 994 amino acid residues, of which more than 80% are within coiled-coil motifs. High rMW and coiled-coil content are common traits of centrosomal proteins [4]. Relatively low levels of conservation are also common among these large coiled-coil proteins, suggesting that their function can tolerate many liberal substitutions in the coding sequence. For instance, in the cases of Pericentrin and SAS4, sequence identity between the Drosophila and human proteins is a mere 17% in both cases [29, 41]. Asl is no exception. The Asl protein of Drosophila melanogaster shares 64% identity with its homolog in the closely related species D. pseudoobscura and only 32% with its homolog in the mosquito A. gambiae. Iterated BLAST based on the cluster of highly conserved Asl sequences in insects identifies Cep152 as a likely human homolog of Asl (13% sequence identity; Evalue = 6e-20). In addition, orthology prediction algorithms based on reciprocal best-hit analysis suggest that Asl and Cep152 are putative orthologs. Cep152 was identified during a proteomic characterization of purified centrosomes, and a Cep152-GFP fusion has been shown to localize to the centrosome [33]. No functional data on CEP152 has yet been reported. Centrosomes in Development In terms of the possible roles that centrioles may play during development, our results show that the first zygotic division never occurs in a cytoplasm deficient for Asl, strongly suggesting that functional centrosomes are needed for embryogenesis in Drosophila. The same conclusion was suggested by the observation that in eggs derived from females lacking PCM components like gTUB37C or D-TACC, the first mitotic division does not take place [11, 39]. However, the possibility remained that this early developmental arrest in gTUB37C or D-TACC-deficient embryos could be a downstream consequence of the meiotic defects caused by mutations in these genes [13, 42]. This caveat is now largely circumvented by the phenotype of embryos derived from asl mutant females in which both meiotic divisions proceed normally. Thus, although we cannot rule out a possible noncentrosomal function of Asl, the phenotype of embryos derived from asl mutant females is consistent with the hypothesis that centrosomes are required for Drosophila embryo development. Previous reports have shown that zygotic loss of key centrosomal proteins such as D-plp, Sas4, Sak/Plk4, or Cnn does not block progression of development into adult flies [10, 13, 22, 24, 29, 43]. However, the centrosome-less females that hatch are sterile, strongly suggesting that eggs defective for these centrosomal components cannot support embryogenesis. How development can proceed in zygotic loss-of-function conditions for these genes is not entirely clear. However, the initial stages of development of individuals homozygous for mutations in these centrosomal proteins are likely to be sustained by the wild-type RNA/protein contributed to the egg by the heterozygous females from which they derive. Thus, until a certain stage that is hard to specify, development in these mutant individuals actually takes place when cells still have centrosomes. The hatching of adults that have undergone the last stages of development without centrosomes certainly proves a certain level of centrosome dispensability in Drosophila development, even though such adult flies are uncoordinated and sterile and die only hours after eclosion. Loss of centrosome function has been reported to impair a number of developmental stages in vertebrates. In humans, for instance, abnormal centrosomes have been linked to impaired neuronal migration, hereditary spastic paraplegia, Bardet-Biedl syndrome, develop-ment of cystic kidneys, perturbed left-right asymmetry, microcephaly, and cancer [1, 44]. In mice, lack of function for Sak/Plk4 is a lethal condition, and haploinsuficiency for this gene results in a high incidence of tumors [45, 46]. The molecular dissection of centrioles in Drosophila may help to model the cellular basis of some of these processes. Experimental Procedures Drosophila Stocks and Fly Culture Fly stocks were obtained from Bloomington and Szeged. The alleles asl1, asl2, and asl3 were obtained from M. Gatti [30]. Df(3R)ED5177 was purchased from Szeged. Generation of Transgenic Flies Genomic construct P[aslGR]: A genomic fragment spanning from 300 bp upstream the initial ATG to 500 bp downstream of the STOP codon of CG2919 was cloned in the transformation vector pW8. YFP-AslFL construct: The entire open reading frame of the CG2919 gene was amplified from an embryonic cDNA library with primers 50ATTTGCGGCCGCTATGAACACGCCAGGTATAAGCCTC TTTC-30 and 50ATTTGCGGCCGCTTAGCTGTGACCATTGCCTTT GGG-30 and cloned into the EYFP-C1 vector (Clontech) with NotI. The EYFP-C1 vector was modified by introducing additional NotI and NheI sites between the BglII and EcoR1 sites. The resulting fu-sion was then cut with NheI and cloned into the SpeI site of the Dro-sophila polyubiquitin transformation vector. YFP-AslD44 construct: This was constructed as YFP-AslFL except that primers were de-signed so that the resulting fusion protein contains a deletion of the 44 N-terminal amino acids. Transgenic flies were generated by standard P element-mediated transformation. Sequencing Genomic DNA was isolated from the different asl alleles and used as template for PCR with the High Fidelity PCR System (Roche). PCR products were subcloned into PBSK (Stratagene) prior to sequencing. At least two clones from different PCR reactions were sequenced. Identification of Asl Homologs and Sequence Alignments BLAST search with the Asl sequence CG2919 (AAF51993.2) identified homologs in other insect species with highly significant E-values (E < 1e280), including the Anopheles gambiae ENSANGP 00000013679 (EAA08928). PSI-BLAST, a standard algorithm for identification of remote homologs, was then used to search for potential human homologs. A first round of PSI- BLAST search with a threshold value of 1e215 identified, aside from CG2919 itself, Anopheles gambiae ENSANGP00000013679 (E = 2e281) and Aedes aegypti EAT35837 (E = 2e287). These sequences were then used in a second round of iterative search. The three resulting top hits from this search were two isoforms of the centrosomal protein Cep152 from chimpanzee (E = 8e221; accession# XP523070) and human Cep152 (E = 6e220; accession# O94986). In addition, standard BLAST searches with Drosophila Asl or human Cep152 confirmed EAA08928 as the best match among Anopheles gambiae proteins (E = 2e281). Orthology queries were run in the Ensembl Genome Browser and the eukaryotic ortholog database Inparanoid [47]. Homology alignments were carried out by Clustal IW [48]. Antibody Production The Asl antibody Rb5110 was raised against a synthetic peptide corresponding to the 16 Cterminal amino acids of CG2919 (SIGMA-GENOSYS) and affinity purified against the immunogenic peptide. Western Blotting of Embryo Lysates and S2 Cell Extracts Embryo lysates: 0–2 hr embryos were collected, and the chorions were removed and boiled in SDS sample buffer. S2 cell extracts: Approximately 2 3 106 cells were pelleted, resuspended, and boiled in SDS sample buffer. Samples were then subjected to PAGE in 7% polyacrilamide gels and blotted onto a nitrocellulose filter. The blot-ted filter was incubated with affinity-purified Rb5110 diluted 1:100, washed three times in 0.1% PBS-T, incubated with HRP-labeled anti-rabbit IgG (Jacksons), washed as before, and treated with the ECL kit (Amersham) to reveal peroxidase activity. Immunocytochemistry Testes were dissected in PBS and fixed in methanol/acetone as described [30]. This fixation preserved the GFP fluorescence in cells expressing GFP-a-tubulin or PACT-GFP [29]. Fixed tissue was incubated 1 hr with primary antibodies diluted in 0.1% PBS-T containing RNAase (Boehringer), followed by three washes in 0.1% PBS-T for 5 min, a 45 min incubation with secondary antibody diluted in 0.1% PBS-T, and a final wash in 0.1% PBS-T for 5 min. DNA was labeled with 5 nM TOTO3 (Molecular Probes) in 0.1% PBS-T for 15 min. Embryos and ovaries were fixed as previously described [49]. a-TUB was detected with DM1a antibody (SIGMA). Cnn anti-bodies were kindly provided by T. Kauffman and E. Schejter. Other antibodies used were monoclonal mouse anti-g-tubulin (clone GTU88, Sigma) diluted 1:50; rabbit anti-Asl Rb 5110 diluted 1:200; Alexa 594 labeled goat anti-rabbit IgG (Molecular Probes) diluted 1:200; and Alexa 488 labeled goat anti-mouse IgG (Molecular Probes) diluted 1:200. Fluorescence images were acquired in a Leica TSC SP2 confocal microscope equipped with a 633 oil objective (NA = 1.4) and further processed in Adobe Photoshop. FRAP Analysis Syncytial Drosophila embryos expressing YFP-AslFL were recorded in a Leica TSC SP2 confocal microscope equipped with a 633 oil inmersion objective (NA = 1.4). Photobleaching was performed by a 15 s pulse of the 488 nm laser line at full power. Florescence recovery was imaged at a rate of 1 frame/10 s for 30 min. Florescence intensities were analyzed with NIH Image. Electron Microscopy and Immunogold Labeling Testes were dissected in PBS and fixed in 2.5% glutaraldehyde in PBS at 4oC overnight. The tissue was then post-fixed in 1% osmium and 1% uranyl acetate (UA) on ice, followed by room Figure 3. PCM Stabilization, Centrosome Segregation, and Microtubule Nucleation Require asl Function (A)Projection of confocal sections from spermatocytes expressing PACT-GFP (green) and immunolabelled with g-tubulin antibodies (red). Wildtype meiotic prophase I spermatocytes contained four centrioles organized into two pairs of orthogonally oriented centrioles. In asl mutants, PACT-GFP reveals a cluster of centriolar signal. gTUB localization in these clusters is highly irregular, ranging from undetectable to one or a few small dots over the much larger PACT-GFP signal. (B) Ultrastructural analysis of centrioles in wild-type and asl primary spermatocytes. In wild-type premeiotic spermatocytes, the Vshaped cen-triole pairs are positioned with the distal part of the orthogonally oriented centrioles protuding slightly under the plasma membrane (arrow). In asl spermatocytes, the two centriole pairs are clustered so that the distal end of a centriole may be found in close contact with the centriole of the other pair (arrowheads). (C) Confocal projections of intact cysts of wild-type and asl spermatocytes that expressed GFP-a-tubulin (green) and were immunolabeled with antibodies against g-tubulin (red). In the wild-type, during prometaphase, centrosomes have segregated and organize dense microtubule asters. In asl cells, g-tubulin accumulation and aster nucleation was highly reduced and the extent of centrosome separation varied greatly among the cells within a cyst. In some cells, one centrosome cluster producing a single aster was seen (single arrow). In others, the two centrosomes were slightly separated (double arrows). A few cells had well-separated centrosomes positioned at opposite sides of the nucleus prior to nuclear envelope breakdown (arrowheads). In some cells, no g-tubulin accumulation or aster nucleation could be detected (asterisk). Figure 4. Maternally Provided Asl Is Required for Embryo Development (A) Male pronucleus derived from wild-type unlabelled sperm in an egg derived from a fly expressing YFP-Asl. Because of the relatively high turn-over of the Asl protein, the sperm basal body incorporates the YFP-Asl label immediately after fertilization and can be seen serving as a centriole in one of the centrosomes (long green rod). (B) After centrosome duplication, the paternally derived centriole is still visible during the first telophase as the longest of the four YFP-Asl-labeled structures. (C) In the wild-type, the basal body-derived centrosome (cnn, red, white arrow) associated with the male pronucleus (DAPI, blue) nucleates an aster (a-Tub, green) while the female pronucleus is still at a distance. (D and E) After pronuclear fusion and centrosome duplication (D), the first mitotic spindle is assembled (E). This first zygotic division is gono-meric: the maternal and paternal chromosome sets are still clearly separated. (F) In less than an hour, exponential nuclear proliferation fills the wild-type embryo with hundreds of nuclei. (G–I) Assembly of the sperm aster, centrosome duplication, and pronuclear fusion are absent in embryos derived from asl1/Df(3R)ED1557 females fertilized by wild-type males (G). An anastral, barrel-shaped microtubule array is formed around the female-derived chromosomes (H and I), but chromosome segregation was never observed. (J) As a consequence thereof, development was completely arrested at this early stage in asl1/Df(3R)ED1557 embryos. Figure S1. Asl Does Not Localize to Centrosomes in asl1 Mutant Cells Projection of confocal sections from spermatocytes during prophase I immunostained for Asl (green) and counterstained for DNA (blue). The Asterless antibody recognizes the centrioles in wild-type spermatocytes but not in asl1 spermatocytes. Figure S2. asl Cells Contain Clusters of Centrioles where Orthogonal Arrangement May Have Been Lost Projection of confocal sections from unfixed, living spermatocytes expressing PACT-GFP. During meiotic prophase I, centriole pairs are orthogonally arranged and can be observed as a V-shaped structure in most wild-type spermatocytes. In contrast, V-shaped structures are rare in asl mutant cells, suggesting that the orthogonal arrangement may have been lost. Figure S3. The Ultrastructure of Centrioles in Wild-Type and asl Spermatocytes Is Largely Similar Centriole structure in Drosophila spermatocytes varies along its length [S1]. (A) The proximal end is made of a central tubule surrounded by nine peripheral triplet tubules. (C) The distal end contains nine peripheral duplet tubules plus a hook and is surrounded by a membranous sheath. (E) Wild-type axonemes (arrow) contain two central tubules surrounded by nine duplets plus a hook and are attached to one major and one minormitochondrial derivative (asterisks). (B, D, F) In most asl cells, centriole and axoneme ultrastructure is not affected, except that because of defective meiosis, the resulting spermatids have more than one axial filaments (F, arrow) and fused, abnormally shaped nebenkern derivatives (F, asterisks). (G–L) However, with a very low frequency, malformed centrioles (G–I) and axonemes (J–L) with missing duplets and hooks can be observed in asl cells. Figure S4. Spindle Assembly in Living asl1 Spermatocytes Time-lapse video microscopy of spindle assembly during meiosis I in spermatocytes expressing GFP-a-tubulin [S2]. Time is shown in min:s. (A) In wild-type cells, prominent asters localized at opposite sides of the nuclear envelope can be seen at the time of nuclear envelope breakdown(NEB, 0:00). After NEB, elongated astral microtubules enter the nucleus (6:35 and 8:55). During prometaphase, thick microtubule bundles are assembled (15:59, white arrows), and by the end of prometaphase, a bipolar spindle with prominent asters is formed (30:14). (B) In rare occasions, asl cells enter prometaphase with fully separated asters (0:00). The size and density of the asters are highly reduced relative to the wild-type (7:25). As prometaphase proceeds, the nucleoplasm fills up with microtubules (17:00), and a bipolar microtubule array is formed between the two weak asters (31:51). (C) Most asl cells initiate spindle assembly with nonseparated or adjacent asters (0:00). In these cells, microtubules are nucleated in all directions from the single or adjacent asters (8:00). During prometaphase, microtubules elongate in the nuclear region (26:10), and monastral microtubule arrays are eventually formed (59:10). Figure S5. Defective Meiosis in asl Males Is Associated with Abnormal Spindle Organization (A) Phase-contrast cell microscopy of unfixed preparations from wild-type and asl1 testes. Meiosis I and II: Wild-type cells form bipolar, biastral spindles (bb), whereas asl1 spindles are frequently monastral (ma) and some are highly abnormal (o). Arrowheads point to astral poles and open arrows points to the phase dense mitochondrial derivative that decorate a subset of the spindle microtubules. Spermatids: In the wild-type, completion of meiosis results in cysts of spermatids with nuclei (arrow) and Nebenkerns (open arrow) of homogenous size. On the contrary, asl early spermatids have multiple, variably sized nuclei (arrows) resulting from irregular chromosome segregation and differentially sized Nebenkerns (open arrows) because of asymmetrical cytokinesis. (B) Quantification of the abnormal spindle morphologies observed in living asl1 spermatocytes during meiosis I and II. n(wt, meiosis I) = 67, n(wt, meiosis II) = 181, n(asl, meiosis I) = 151, n(asl, meiosis II) = 221.