Centrosome duplication must be coupled to the main cell cycle to ensure that each cell has precisely two centrosomes at the onset of mitosis. Supernumerary centrosomes are commonly observed in cancer cells, and may contribute to tumorigenesis. Drosophila skpA, a component of SCF ubiquitin ligases, regulates the link between the cell and centrosome cycles. Lethal skpA null mutants exhibit dramatic centrosome overduplication and additional defects in chromatin condensation, cell cycle progression and endoreduplication. Surprisingly, many mutant cells are able to organize pseudo-bipolar spindles and execute a normal anaphase in the presence of extra functional centrosomes. SkpA mutant cells accumulate higher levels of cyclin E than wildtype cells during S and G2, suggesting that elevated cdk2/cyclin E activity may account for the supernumerary centrosomes in skpA cells. However, centrosome overduplication still occurs in skpA;cycE mutant animals, demonstrating that high cyclin E levels are not necessary for centrosome overduplication. These data suggest that additional SCF targets regulate the centrosome duplication pathway.

Introduction

The centrosome serves as the major microtubule-organizing center in animal cells, helping to create polarity and organization within the cell. Each G1 cell contains a single centrosome, consisting of pericentriolar material organized around a pair of centrioles, which nucleates microtubule polymerization. During cell division, the centrosome must be duplicated precisely once and the resulting two centrosomes help to organize the bipolar spindle, which segregates the chromosomes (reviewed by Andersen, 1999).

Recent studies have begun to elucidate how the cell initiates centrosome duplication. In most cell types, centriole duplication begins near the onset of S phase (reviewed by Sluder and Hinchcliffe, 1998), suggesting that it may be controlled by part of the pathway that initiates DNA synthesis, such as cyclin E bound to cyclin-dependent kinase-2 (Cdk2-E). In somatic cells, levels of cyclin E rise in late G1 and the resulting rise in Cdk2-E kinase activity is necessary and sufficient to drive cells into S phase(Knoblich et al., 1994; Strausfeld et al., 1996). Centrosome duplication is blocked by inhibitors of Cdk2 activity(Hinchcliffe et al., 1999; Lacey et al., 1999; Matsumoto et al., 1999; Meraldi et al., 1999), and constitutive expression of cyclin E results in centrosome duplication beginning prematurely in early G1 (Mussman et al., 2000). In Swiss 3T3 cells, Cdk2-E phosphorylates nucleophosmin, a component of unduplicated centrosomes, and expression of a nonphosphorylatable form of nucleophosmin blocks centrosome duplication(Okuda et al., 2000). Thus,Cdk2-E activity is necessary to initiate centrosome duplication, in part through the phosphorylation of nucleophosmin.

Little is known about the regulatory mechanism that ensures centrosome duplication occurs only once in each cell cycle. Cells apparently lack a cell cycle checkpoint to detect the presence or production of excess centrosomes(Sluder et al., 1997). Conversely, it is not known if cells will efficiently proceed into mitosis in the absence of centrosome duplication. Thus, the fidelity of centrosome production relies largely on regulating the duplication process itself, rather than by using checkpoints to monitor the fidelity of the process afterwards(reviewed by Hinchcliffe and Sluder,2001). The observation of supernumerary centrosomes (≥3 centrosomes in a cell) has frequently been used as evidence for misregulation of centrosome duplication, suggesting that genes such as p53, Brca1,Brca2, p21, ATR and others are part of the pathway that regulates centrosome duplication (reviewed by Meraldi and Nigg, 2002). However, recent studies suggest that many instances of supernumerary centrosomes, including those in p53–/– cells,arise through failed cell division resulting in tetraploid cells with twice the normal number of centrosomes (Borel et al., 2002; Meraldi et al.,2002). Consequently, our understanding of the pathway controlling centrosome duplication remains murky.

Many diverse cellular processes are regulated by the SCF family of ubiquitin ligases, which target specific proteins for proteolysis (reviewed by Deshaies, 1999). SCF complexes are found in all eukaryotes and consist of an invariant core containing Skp1,Cul1 and Rbx1/Roc1 complexed with one member of a large family of F-box proteins. Substrate recognition typically occurs through a protein interaction motif in the F-box protein, and the rest of the complex acts to recruit a ubiquitin-conjugating enzyme that catalyzes the assembly of a polyubiquitin chain on the substrate, thus targeting it for degradation by the proteasome(Feldman et al., 1997; Skowyra et al., 1997). These biochemical studies suggest that mutations in SCF complex genes will disrupt the regulated degradation of many substrates in the cell. Several SCF components have been localized to centrosomes in vertebrate cells(Freed et al., 1999; Gstaiger et al., 1999), and supernumerary centrosomes have been reported in cells mutant for the F-box proteins skp2 (mouse) (Nakayama et al., 2000) and slimb (Drosophila)(Wojcik et al., 2000). However, many of the mammalian studies are confounded by high frequencies of polyploidy which have made it difficult to ascribe a direct role for SCF function in regulating centrosome duplication.

Here I demonstrate that null mutations in Drosophila skpA, a homolog of Skp1, result in centrosome overduplication and defective endoreduplication, chromatin condensation and cell cycle progression. SkpA mutant cells accumulate elevated levels of cyclin E after entering S phase; however, genetic epistasis experiments demonstrate that high cyclin E levels are not necessary for centrosome overduplication to occur. Thus, the accumulation of other SCF substrates probably accounts for centrosome overduplication. One of these targets may function as a centrosome-licensing factor to restrict centrosome duplication to once per cell cycle.

Materials and Methods

Molecular analyses

A portion of the skpA gene was amplified from Drosophilagenomic DNA using degenerate primers (GCGCGAATTCTTYGAGYTNATHYTNGCNGC,CGTCTAGAATYTCYTCNGGNGTYTTNCC), and used to identify a BAC clone (BACR09A14)that includes the skpA gene. The skpA ORF was then used to probe an arrayed library that identified P1s containing skpB(DS03515, DS04791) and skpC and -D (DS01693). Other Skp1 homologs were identified in Genbank using TBLASTN. SkpAcDNA clones were identified from the Drosophila EST database(Berkeley Drosophila Genome Project, personal communication) and by filter screening of an embryonic cDNA library.

mRNA was purified from staged embryos, larvae and adults using Purescript RNA Isolation (Gentra Systems) and Oligotex mRNA Isolation kits (Qiagen). For each sample 2 μg was electrophoresed through a 1% formaldehyde gel,transferred to a Hybond-N membrane (Amersham Pharmacia Biotech), and hybridized at 68°C with 32P-labeled, double-stranded DNA probes prepared by random priming.

Genetic manipulations

The EP(X)1423 P element was remobilized in EP(X)1423white+ w/FM7a, wa B; TMS,Δ 2,3 Sb/+ females and crossed to FM0, w B/Y males,yielding EP(X)1423white– w/FM0, w B progeny with precise or imprecise excisions of the P element. After establishing individual lines, DNA from single flies was amplified by PCR to test for the presence or absence of each end of the P element and the neighboring sequences. Select lines were further characterized by Southern analyses and inverse PCR according to standard protocols.

Lethal skpA alleles failed to complement two deficiencies{Df(1)svr, Df(1)su(s)83}, and were covered by the duplication Dp(1;Y) y2 sc. For rescue experiments, a 4 kb Hind III fragment was cloned into P{HZ-Casper} and transformed into y w embryos using a through-the-chorion injection protocol(Miller et al., 2002). Three independent insertions rescued the lethality associated with the four skpA alleles.

Larvae were collected in one-day intervals (unless otherwise noted) and raised on grape juice-and-agar plates supplemented with wet yeast in individual humid chambers at 23°C. SkpA larvae used for phenotypic analyses were identified as GFP-progeny (skpA/Y)from the cross skpA1/FM7, GFP × FM7, GFP/Y. Other skpA alleles exhibited phenotypes similar to skpA1. SkpAcycE larvae were identified as GFP lacZ progeny from the cross skpA1/FM7, lacZ; cycEk05007/CyO, GFP× FM7, lacZ; cycEk05007/CyO, GFP. Clones of skpA cells were generated in female larvae of the genotype w hsFLP tubP-GAL80 FRT-19A/w skpA FRT-19A;UAS-mCD8::GFP/+; tubP-GAL4/+. FLP-induced mitotic recombination produces clones of skpA cells that no longer express the GAL4-inhibitor GAL80, resulting in the expression of membrane-bound GFP in the mutant cells.

Cytological analyses

All cytological steps were performed at 23°C unless otherwise noted. Larval central nervous systems (CNSs) were prepared for immunofluorescence using slightly modified standard techniques. For squashes, CNSs from two wildtype and three mutant larvae were dissected on a slide in Robb's saline,fixed in a drop of 3.6% formaldehyde in Robb's saline for 7 minutes, rinsed briefly in 45% acetic acid, extracted in 60% acetic acid for 3 minutes,squashed under a silanized coverslip (Hampton Research), and frozen in liquid nitrogen. The coverslip was then removed with a razor blade and the slide immersed in methanol overnight. For whole mounts, larval heads were partially dissected in Robb's saline, fixed in 1 mL 3.2% paraformaldehyde in Robb's saline with 0.5 μM paclitaxel for 30 minutes, rinsed in PT (1×phosphate-buffered saline, 0.5% Triton X-100) for over 30 minutes, and treated with RNase A (500 μg/mL in PBS) for over 30 minutes.

After fixation, samples were blocked in PT + 5% normal goat serum (PTG) for 1 hour, incubated with 1° antibody in PTG overnight at 4°C, washed in PT for over 4 hours, incubated with 2° antibody in PTG overnight at 4°C, washed, and rinsed in PBS before mounting in Vectashield (Vector Laboratories). Antibody incubations were performed in 40 μl under coverslips for squashes and in over 200 μl for whole mounts. DNA was stained with TOTO-3 (0.5 μM, Molecular Probes) or DAPI (0.5 μg/mL,Sigma) for 5 minutes. Primary antibodies were used at the following dilutions:α-tubulin (mouse DM 1A, Sigma), 1:1000; γ-tubulin (mouse GTU-88,Sigma), 1:1000; cnn (rabbit), 1:1000; phosphorylated-histone H3 (rabbit,Upstate Biotechnology), 1:5000; BrdU (mouse B44, Becton Dickinson), 1:20;cyclin E (rabbit), 1:200; SKPa (guinea pig), 1:100. Alexa-488, Alexa-568, FITC and Cy5 conjugated 2° antibodies (Molecular Probes and Jackson Immunoresearch) were used at 1:200.

Triple staining with α-tubulin, cnn and P-histone H3 antibodies required sequential antibody incubations and additional blocking steps. Samples were sequentially incubated overnight at 4°C with 1) rabbit anti-cnn, 2) Alexa-568 goat anti-rabbit, 3) 10% normal rabbit serum, 4)unlabeled goat anti-rabbit IgG Fab fragments (130 μg/mL, Jackson Immunoresearch), 5) rabbit anti-P-histone H3 and mouse anti-α-tubulin,and 6) Alexa-488 anti-mouse and Cy5 anti-rabbit at the normal dilutions. Occasionally the anti-cnn antibody was not completely blocked, resulting in weak centrosome staining visible in the Cy5 (P-histone H3) channel.

For BrdU incorporation, larval heads were partially dissected in Grace's medium, and incubated for one hour in 1 mL of Grace's medium containing 10μM BrdU (Sigma). Samples were then fixed in 3.6% formaldehyde in Robb's saline for 30 minutes, washed in PT for over 4 hours, denatured in two changes of 2.2 N HCl + 0.1% Triton X-100 for 15 minutes each, neutralized with two changes of 100 mM sodium tetraborate for 2 minutes each, and washed in PT for over 4 hours. Samples were then RNase treated, blocked, and stained with anti-BrdU as described above.

TUNEL labeling was performed using the Apoptag Fluorescein Direct In Situ Apoptosis Detection Kit (Intergen). Larval heads were dissected and fixed as described above. Samples were then washed in equilibration buffer, incubated with 400 μl working-strength TdT enzyme for 1 hour at 37°C, washed in stop/wash buffer for 10 minutes, and further washed in PT overnight at 4°C. Samples were then RNase treated, counterstained with TOTO-3, and mounted as described above.

Squashed samples were imaged on a Zeiss Axioplan II microscope with a Quantix cooled-CCD camera run with IP Lab Spectrum (Scanalytics). Whole mounts were imaged on a Leica TCS/NT or TCS/SP2 laser-scanning confocal microscope. Cell cycle analyses were performed with the segmentation and analysis features of IP Lab Spectrum to quantify the total DAPI and average cyclin E intensity after background subtraction of more than 2000 nuclei per sample. DNA content was normalized to the signal from diploid mitotic cells. Centrosome clustering was determined by plotting the position of each centrosome in a mitotic cell from confocal sections using the 3D Tracer function of IP Lab Spectrum, and determining the distance to its nearest neighbor using Excel (Microsoft). Images were prepared for publication using Photoshop (Adobe Systems) and Canvas (Deneba Systems).

Samples were prepared for electron microscopy according to standard protocols. CNSs from 5.5 days after egg deposition (AED) larvae were fixed in glutaraldehyde, post-fixed with osmium tetroxide, and stained en bloc with uranyl acetate. After dehydration in graded ethanol samples were embedded in epoxy resin, cut and mounted onto copper grids which were stained with lead citrate, and examined on a Philips Tecnai 12 electron microscope.

Biochemical analyses

For antibody production, SKPa protein was purified from E. coliusing the QIAexpressionist system (Qiagen). The skpA ORF was PCR cloned into pQE-30 and sequence verified. The His6-SKPa fusion protein was expressed in M15 cells by IPTG induction for 4 hours. A denatured, cleared cell lysate was incubated with 4 mL of Ni-NTA agarose,loaded onto an FPLC column, washed and step-eluted with 125, 250 and 500 mM imidazole. His6-SKPa fusion protein was further purified on a 14%acrylamide gel, and the excised band was injected into guinea pigs (0.8 mg/animal) for antibody production (BAbCO).

For western analyses, embryos or larvae were homogenized in SDS-PAGE sample buffer, boiled and run on a 14% acrylamide gel. Gels were either stained with Coomassie Blue to assay protein concentration, or electro-transferred to PVDF. Immunodetection was performed with the ECL Western Blotting System (Amersham Pharmacia Biotech). Guinea pig SKPa and rabbit cycE antibodies were used at a 1:5000 dilution; HRP-conjugated secondary antibodies (Jackson Immunoresearch and Amersham Pharmacia Biotech) were used at 1:2000.

Results

Drosophila skpA is required for viability

Six Drosophila skp1-related genes were identified and named skpA through skpF (Fig. 1A). SkpA was chosen for further analysis because it is the most widely expressed (Fig. 1B,C) and shares the highest identity with human and yeast Skp1 (76% and 45%, respectively). cDNA analyses revealed that the skpA transcript is alternatively spliced, but all of the splice forms encode the same 162 amino acid protein(Fig. 2A).

Fig. 1.

The Drosophila genome contains six Skp1-related genes. (A) Partial phylogenetic tree of eukaryotic Skp1 homologs. ORFs homologous to human Skp1 and the related ElonginC gene were identified from the genomes of Saccharomyces cerevisiae (Sc), Schizosaccharomyces pombe (Sp), Caenorhabditis elegans (Ce), Arabidopsis thaliana (At), Drosophila melanogaster (Dm), and humans (Hs), and the predicted protein sequences were aligned using ClustalW and displayed as a phenogram. The six Drosophila Skp1-related proteins share 43-76%identity with human Skp1. Only one human Skp1 gene is known to be complete and expressed. Clusters of Skp1-related genes restricted to a single species have been grouped for clarity. (B) Expression of Drosophilaskp genes during development. Northern blots of mRNA from different developmental stages (0-2- and 2-20-hour-old embryos, 3rd instar larvae, adult females, and adult males, respectively) were probed for skpA, skpB or a mixture of skpC and skpD (combined because the two transcripts could not be reliably distinguished by hybridization). Signal from the skpB probe in embryos may result from cross-hybridization with the abundant skpA transcripts. Two images of each blot are shown,with the signal scaling increased 100-fold in the bottom images to compensate for the differences in signal levels. Relative amounts of mRNA in each lane were determined by quantifying the signal obtained using an rp49 probe. (C)Representation of Drosophila skp genes in EST databases. cDNAs corresponding to five of the six Drosophila skp genes were identified, indicating that they are bona fide genes, although skpAESTs were 10- to 80-fold more common than the other skp genes.

Fig. 1.

The Drosophila genome contains six Skp1-related genes. (A) Partial phylogenetic tree of eukaryotic Skp1 homologs. ORFs homologous to human Skp1 and the related ElonginC gene were identified from the genomes of Saccharomyces cerevisiae (Sc), Schizosaccharomyces pombe (Sp), Caenorhabditis elegans (Ce), Arabidopsis thaliana (At), Drosophila melanogaster (Dm), and humans (Hs), and the predicted protein sequences were aligned using ClustalW and displayed as a phenogram. The six Drosophila Skp1-related proteins share 43-76%identity with human Skp1. Only one human Skp1 gene is known to be complete and expressed. Clusters of Skp1-related genes restricted to a single species have been grouped for clarity. (B) Expression of Drosophilaskp genes during development. Northern blots of mRNA from different developmental stages (0-2- and 2-20-hour-old embryos, 3rd instar larvae, adult females, and adult males, respectively) were probed for skpA, skpB or a mixture of skpC and skpD (combined because the two transcripts could not be reliably distinguished by hybridization). Signal from the skpB probe in embryos may result from cross-hybridization with the abundant skpA transcripts. Two images of each blot are shown,with the signal scaling increased 100-fold in the bottom images to compensate for the differences in signal levels. Relative amounts of mRNA in each lane were determined by quantifying the signal obtained using an rp49 probe. (C)Representation of Drosophila skp genes in EST databases. cDNAs corresponding to five of the six Drosophila skp genes were identified, indicating that they are bona fide genes, although skpAESTs were 10- to 80-fold more common than the other skp genes.

Fig. 2.

Null mutations in skpA result in larval lethality. (A) Structure of the skpA gene, transcripts and mutations. The skpAtranscript structures, site of the EP(X)1423 P-element insertion, and extent of the skpA deletion alleles are shown relative to the skpA genomic region. The skpA1 deletion breakpoints were precisely mapped by sequencing; the extents of the other deletions are not known. P-element transformation of a 4 kb genomic fragment completely rescued the lethality associated with all four skpAalleles. (B) Survival of skpA larvae. Four-hour collections of embryos from the cross skpA1/FM7, GFP× FM7, GFP/Y were aged from 1 to 5 days AED, and scored for the percentage of GFP larvae. Approximately 25% of larvae from a viable precise excision allele are GFP(skpA+/Y). Similarly, ∼25% of larvae from skpA mutant crosses are GFP immediately after hatching; however, this percentage steadily declines, indicating that skpA/Y larvae die at various points during larval development. (C) Growth of skpA larvae. The average cross-sectional area of GFP larvae from the previous cross are graphed (error bars equal one standard deviation). Wildtype and GFP+ sibling larvae increased in size 17-fold over 4 days, whereas skpA larvae only grew 6-fold.

Fig. 2.

Null mutations in skpA result in larval lethality. (A) Structure of the skpA gene, transcripts and mutations. The skpAtranscript structures, site of the EP(X)1423 P-element insertion, and extent of the skpA deletion alleles are shown relative to the skpA genomic region. The skpA1 deletion breakpoints were precisely mapped by sequencing; the extents of the other deletions are not known. P-element transformation of a 4 kb genomic fragment completely rescued the lethality associated with all four skpAalleles. (B) Survival of skpA larvae. Four-hour collections of embryos from the cross skpA1/FM7, GFP× FM7, GFP/Y were aged from 1 to 5 days AED, and scored for the percentage of GFP larvae. Approximately 25% of larvae from a viable precise excision allele are GFP(skpA+/Y). Similarly, ∼25% of larvae from skpA mutant crosses are GFP immediately after hatching; however, this percentage steadily declines, indicating that skpA/Y larvae die at various points during larval development. (C) Growth of skpA larvae. The average cross-sectional area of GFP larvae from the previous cross are graphed (error bars equal one standard deviation). Wildtype and GFP+ sibling larvae increased in size 17-fold over 4 days, whereas skpA larvae only grew 6-fold.

Null mutations in skpA were generated by imprecise excision of a P-element localized to the first intron of skpA(Fig. 2A) (Berkeley Drosophila Genome Project, personal communication). Four alleles were recovered with deletions of either the skpA ORF or promoter. The skpA1 deletion completely removes the P element and 1782 bp 3′ of the original insertion, including the entire skpA ORF,and is therefore a null allele.

All four skpA alleles are homozygous lethal when crossed to skpA1 or larger deficiencies. This lethality was completely rescued in transgenic flies expressing skpA(Fig. 2A and data not shown),indicating that the lethality results from loss of skpA function. SkpA mutant embryos develop normally and hatch at wildtype frequencies (Fig. 2B and data not shown), potentially because of perdurance of maternally loaded mRNA and protein (Fig. 1B and data not shown). Most mutant larvae die within four days after hatching(Fig. 2B), and surviving mutant animals proceed through larval development but fail to pupate and grow significantly slower than wildtype (Fig. 2C). These results indicate that skpA function is required for larval growth and viability.

SkpA mutants have defects in cell cycle progression

SkpA larvae show pronounced defects in all proliferating tissues. The imaginal discs are rudimentary or absent, and the central nervous system (CNS) shows little increase in size past three days after egg deposition (AED). To further investigate these defects, I compared various cell cycle parameters in the CNS from mutant larvae with wildtype controls (a precise excision of the P element in the same genetic background).

SkpA cells exhibit a dramatic decrease in cell proliferation. The proportion of mitotic cells is comparable to wildtype shortly after hatching, but is dramatically reduced as early as 3.5 days AED and continues to decrease in surviving older animals(Fig. 3A). The proportion of cells in S phase is similarly reduced (Fig. 3B,F). These data suggest that skpAcells have a lengthened G1 and/or G2 phase of the cell cycle. To measure this more directly, the DNA content of individual nuclei was quantified to determine if they were in G1, S or G2 phase. No change was observed in the ratio of G1 to G2 cells in the CNS from young larvae; however, older mutant animals showed a dramatic increase in the proportion of G1 cells(Fig. 3C). Taken together, loss of skpA function results in a lengthening of the cell cycle by approximately twofold and ultimately a delay or arrest in G1.

Fig. 3.

skpA mutants exhibit cell cycle defects. (A-E) Quantification of cell cycle parameters. The CNSs from wildtype and skpA larvae of the indicated ages were assayed for different cell cycle parameters. (A) Mitotic index is the number of phosphorylated-histone H3 (P-histone H3)-positive cells per squashed field observed with a 100× objective. (B) S phase index is the number of BrdU-incorporating cells relative to the volume of DNA staining determined from whole mount confocal analyses. (C) 2C:4C ratio is the ratio of cells with a 2C versus a 4C DNA content, determined from CCD images of DAPI-stained squashed CNSs. (D) Apoptosis index is the number of TUNEL-positive cells relative to the volume of DNA staining. (E) S phase (%) is the percentage of BrdU-incorporating salivary gland or fat body nuclei relative to the total number of nuclei from larvae 3.5 days AED. Error bars in B, D and E represent one standard deviation (n≥4 CNSs). (F-H) Examples of BrdU incorporation and TUNEL-labeling in wildtype and mutant tissues. Projections of confocal sections through entire CNSs (F,G) or salivary glands and fat bodies (H) are shown. Note the skpA imaginal disc(I.D.) with many TUNEL+ cells (G′). No BrdU-incorporating nuclei were observed in skpA fat bodies (F.B.), in contrast with the neighboring salivary glands (S.G.) (H′). Bar, 30 μm.

Fig. 3.

skpA mutants exhibit cell cycle defects. (A-E) Quantification of cell cycle parameters. The CNSs from wildtype and skpA larvae of the indicated ages were assayed for different cell cycle parameters. (A) Mitotic index is the number of phosphorylated-histone H3 (P-histone H3)-positive cells per squashed field observed with a 100× objective. (B) S phase index is the number of BrdU-incorporating cells relative to the volume of DNA staining determined from whole mount confocal analyses. (C) 2C:4C ratio is the ratio of cells with a 2C versus a 4C DNA content, determined from CCD images of DAPI-stained squashed CNSs. (D) Apoptosis index is the number of TUNEL-positive cells relative to the volume of DNA staining. (E) S phase (%) is the percentage of BrdU-incorporating salivary gland or fat body nuclei relative to the total number of nuclei from larvae 3.5 days AED. Error bars in B, D and E represent one standard deviation (n≥4 CNSs). (F-H) Examples of BrdU incorporation and TUNEL-labeling in wildtype and mutant tissues. Projections of confocal sections through entire CNSs (F,G) or salivary glands and fat bodies (H) are shown. Note the skpA imaginal disc(I.D.) with many TUNEL+ cells (G′). No BrdU-incorporating nuclei were observed in skpA fat bodies (F.B.), in contrast with the neighboring salivary glands (S.G.) (H′). Bar, 30 μm.

Cell cycle defects may ultimately induce mutant cells to undergo programmed cell death; therefore, skpA cells were tested for changes in apoptosis by TUNEL-labeling. No increase in apoptotic cells was observed (Fig. 3D,G); in fact,significantly fewer cells were undergoing apoptosis which may result from disruption of the normal schedule of programmed cell death in the CNS. In contrast, virtually all cells in the few rudimentary imaginal discs observed were undergoing apoptosis (Fig. 3G), which probably accounts for the lack of imaginal discs in most mutant larvae.

Larval growth occurs primarily through increasing cell size supported by nuclear endoreduplication; consequently, many cell proliferation mutants do not cause lethality until the beginning of pupation(Gatti and Baker, 1989). To test if skpA lethality may result from a defect in endoreduplicating tissues, endoreduplication was assayed in larval salivary glands and fat bodies by BrdU incorporation. Comparable levels and frequencies of endoreduplication were observed in wildtype and skpA larval salivary glands; however, mutant fat body nuclei contained less DNA than wildtype and rarely underwent endoreduplication (Fig. 3E,H). Similarly, gut nuclei contained less DNA and had an abnormal morphology in skpA larvae (data not shown). Thus, skpAis required for endoreduplication in some larval tissues, perhaps by regulating promoters or inhibitors of S phase.

Loss of skpA results in centrosome overduplication

Previous studies have asserted roles for SCF components in regulating the separation or duplication of centrosomes(Freed et al., 1999; Nakayama et al., 2000; Wojcik et al., 2000). To determine if skpA plays a role in controlling centrosome duplication,centrosomes were stained in wildtype and skpAneuroblasts with antibodies against γ-tubulin or centrosomin, two components of the pericentriolar matrix(Heuer et al., 1995; Zheng et al., 1991). As expected, nearly all mitotic wildtype cells contained two centrosomes that labeled with both γ-tubulin and centrosomin. In contrast, three or more centrosomes were frequently observed in mitotic skpA cells (Fig. 4C-E). Supernumerary centrosomes were found in 4% of cells as early as 1.5 days AED, and in most mitotic cells in older animals (at days AED: 3.5, 60%; 5.5, 78%; 7.5, 85%) with as many as 17 centrosomes observed in a single diploid cell. SkpA interphase(phospho-histone H3 negative) nuclei also frequently showed aberrant chromatin condensation (Fig. 4B), which was especially pronounced in CNS cells from older animals. Clonal analyses demonstrated that the supernumerary centrosomes, delayed cell cycle and abnormally condensed chromatin are caused by cell autonomous defects in skpA function (data not shown).

Fig. 4.

SkpA larval CNS cells accumulate supernumerary centrosomes. (A-D) DNA and centrosome staining in wildtype and skpA cells. (A-B) SkpA interphase cells 3.5 days AED have abnormally condensed chromatin (arrowhead) compared to wildtype. DNA that labeled with P-histone H3 antibodies is marked (M). (C) skpA metaphase cell with four centrosomes labeled with antibodies against γ-tubulin and centrosomin (cnn). (D) Polyploid(∼16C) skpA nucleus from larva 5.5 days AED that partially labeled with P-histone H3 antibodies and contained 24 centrosomes. Bars, 5 μm. (E) Quantification of centrosomes in mitotic wildtype and skpA cells. The distribution of P-histone H3-positive cells with different numbers of centrosomes is shown.(F) Quantification of polyploidy in mitotic wildtype and skpA cells. Polyploid P-histone H3-positive cells were identified from measurements of their total DNA content. Error bar represents one standard error. (G) Electron micrographs of skpA cell from 5.5 days AED larval CNS with four pairs of centrioles in nine 105 nm sections. A low magnification section with two pairs of centrioles is shown, with high magnification views of centrioles from this or adjacent sections shown in the insets. Bars, 500 nm.

Fig. 4.

SkpA larval CNS cells accumulate supernumerary centrosomes. (A-D) DNA and centrosome staining in wildtype and skpA cells. (A-B) SkpA interphase cells 3.5 days AED have abnormally condensed chromatin (arrowhead) compared to wildtype. DNA that labeled with P-histone H3 antibodies is marked (M). (C) skpA metaphase cell with four centrosomes labeled with antibodies against γ-tubulin and centrosomin (cnn). (D) Polyploid(∼16C) skpA nucleus from larva 5.5 days AED that partially labeled with P-histone H3 antibodies and contained 24 centrosomes. Bars, 5 μm. (E) Quantification of centrosomes in mitotic wildtype and skpA cells. The distribution of P-histone H3-positive cells with different numbers of centrosomes is shown.(F) Quantification of polyploidy in mitotic wildtype and skpA cells. Polyploid P-histone H3-positive cells were identified from measurements of their total DNA content. Error bar represents one standard error. (G) Electron micrographs of skpA cell from 5.5 days AED larval CNS with four pairs of centrioles in nine 105 nm sections. A low magnification section with two pairs of centrioles is shown, with high magnification views of centrioles from this or adjacent sections shown in the insets. Bars, 500 nm.

Supernumerary centrosomes may arise from any of four mechanisms: (1) failed cytokinesis, (2) segregation of both centrosomes to the same daughter cell,(3) aberrant centriole splitting or fragmentation, or (4) formation of additional centrosomes in a single cell cycle, either de novo or from reduplication of the existing centrosomes. The extra centrosomes observed in skpA cells are unlikely to occur by the first three mechanisms for several reasons. First, few skpAcells are polyploid (Fig. 4F),indicating that most cells complete cytokinesis. Second, all skpA anaphase cells have centrosomes at both poles,suggesting that skpA cells do not assemble acentrosomal spindles that would allow both centrosomes to (randomly)segregate to the same daughter cell. Third, most of the centrosomes observed in skpA cells were of uniform size and morphology,suggesting that they had not arisen from centrosome fragmentation. Furthermore, serial section electron microscopy of the CNS from skpA larvae found one cell with at least four pairs of centrioles, three of which had incomplete daughter centrioles(Fig. 4G) suggesting that they were undergoing assembly and that extra centrosomes arise from normal centriole duplication. Taken together, these data suggest that loss of skpA function results in the formation of extra centrosomes through multiple rounds of centrosome duplication in the same cell cycle.

Supernumerary centrosomes function as microtubule organizing centers

Cells with three or more centrosomes typically form multipolar spindles that ultimately lead to chromosome missegregation and aneuploidy(Heneen, 1975; Pera and Rainer, 1973; Rieder et al., 1997; Sluder et al., 1997). Surprisingly, all of the skpA anaphase cells with supernumerary centrosomes were segregating their chromosomes to only two poles, suggesting that the additional centrosomes may not be completely functional. This possibility was also raised for the supernumerary centrosomes observed in Drosophila slimb mutants, which encodes another SCF component. Therefore, I performed a detailed analysis of the functional properties of skpA supernumerary centrosomes.

Confocal analyses of skpA cells stained for centrosomes, microtubules and chromosomes revealed that skpA supernumerary centrosomes are competent to nucleate microtubules and attach to chromosomes. All of the centrosomes in skpA prophase and prometaphase neuroblasts appeared to nucleate similar numbers of microtubules and were equally spaced around the nuclear periphery, suggesting that the microtubule arrays were actively positioning the centrosomes relative to one another(Fig. 5A). However, once the chromosomes had attached to the spindle, most of the centrosomes were typically found clustered into two poles and formed a pseudo-bipolar spindle with the chromosomes positioned at a normal metaphase plate(Fig. 5B,C). Anaphase cells retained a bipolar configuration, with the majority of centrosomes clustered at the two poles (Fig. 5E). Three-dimensional quantification of centrosome positioning in young larval CNSs revealed that supernumerary centrosomes were 2.5-fold more likely to be within 2 μm of another centrosome in metaphase and anaphase than earlier in the mitotic cycle, even though progression through the mitotic cycle was not significantly altered and similar numbers of centrosomes were seen at all stages (Fig. 5G,H, and data not shown). Therefore, the extra centrosomes are dynamically repositioned during mitosis allowing formation of pseudo-bipolar spindles and progression to anaphase.

Fig. 5.

Supernumerary centrosomes in skpA cells nucleate microtubules. (A-F) Projections of confocal sections of mitotic skpA neuroblasts (A-E) or ganglion mother cell (F)labeled for DNA (P-histone H3 or TOTO-3), centrosomes (cnn) and microtubules(α-tubulin). Cells have the following number of centrosomes: (A) 4,prometaphase; (B) 10, metaphase; (C) 12, metaphase; (D) 8, multipolar prometaphase; (E), 4 and 6, anaphase; and (F) 6, metaphase. Note the two chromosomes pulled off the metaphase plate by supernumerary centrosomes(arrowheads, B), and the spindle pole not associated with any centrosomes(arrow, E). Bar, 2 μm. (G) Quantification of centrosome clustering during mitotic stages. Supernumerary centrosomes in P-histone H3-positive cells from 3.5-day AED larvae were scored as clustered if they were within 2 μm of another centrosome. Mitotic stages were determined from chromosome and spindle morphology. Error bar represents one standard error. The average number and range (solid line) of centrosomes per cell is also shown for each stage. (H)Ratios of metaphase to anaphase cells from larvae of different ages.

Fig. 5.

Supernumerary centrosomes in skpA cells nucleate microtubules. (A-F) Projections of confocal sections of mitotic skpA neuroblasts (A-E) or ganglion mother cell (F)labeled for DNA (P-histone H3 or TOTO-3), centrosomes (cnn) and microtubules(α-tubulin). Cells have the following number of centrosomes: (A) 4,prometaphase; (B) 10, metaphase; (C) 12, metaphase; (D) 8, multipolar prometaphase; (E), 4 and 6, anaphase; and (F) 6, metaphase. Note the two chromosomes pulled off the metaphase plate by supernumerary centrosomes(arrowheads, B), and the spindle pole not associated with any centrosomes(arrow, E). Bar, 2 μm. (G) Quantification of centrosome clustering during mitotic stages. Supernumerary centrosomes in P-histone H3-positive cells from 3.5-day AED larvae were scored as clustered if they were within 2 μm of another centrosome. Mitotic stages were determined from chromosome and spindle morphology. Error bar represents one standard error. The average number and range (solid line) of centrosomes per cell is also shown for each stage. (H)Ratios of metaphase to anaphase cells from larvae of different ages.

Although all centrosomes were equally competent to nucleate microtubules in prophase and prometaphase, only a subset of centrosomes were associated with the bulk of the spindle microtubules in later mitotic stages(Fig. 5B,C) and some spindle poles appeared to be detached from any centrosomes(Fig. 5E), suggesting that some centrosomes may be inactivated or have decreased microtubule retention capacity. Centrosome inactivation is a normal characteristic of wildtype ganglion mother cells (Bonaccorsi et al.,2000), which are descended from neuroblasts, suggesting that the reduced microtubule nucleation/retention of some supernumerary centrosomes may result from the normal developmental switch to ganglion mother cell characteristics. Nevertheless, all supernumerary centrosomes were associated with at least a few microtubules, even in mitotic ganglion mother cells(Fig. 5F), and some formed functional kinetochore attachments that could either displace chromosomes from the metaphase plate (Fig. 5B)or generate multipolar spindles in a few cases(Fig. 5D). Furthermore,anaphase cells were increasingly rare and had fewer centrosomes than metaphases in older animals (Fig. 5H and data not shown), suggesting that cells with many centrosomes delayed or arrested in metaphase. These cells may ultimately forgo cytokinesis and account for the small increase in polyploid cells in older animals.

In conclusion, the supernumerary centrosomes in skpA cells can act as functional microtubule organizing centers, but neuroblasts can partially compensate for this aberrant microtubule nucleation by either clustering extra centrosomes together or partially inactivating them in later mitotic stages. These compensation mechanisms are sufficient to allow some cells to divide normally, although older skpA cells delay or arrest in metaphase. These mitotic defects may ultimately induce the observed delay or arrest in G1; alternatively, the defect in progression into S phase may be independent of the accumulation of extra centrosomes.

SKPa protein is present throughout the cytoplasm and nucleus

Skp1 is localized to centrosomes throughout the cell cycle in vertebrate cells (Freed et al., 1999; Gstaiger et al., 1999),suggesting that it may act directly at the centrosome to regulate duplication. To determine if SKPa shows a similar localization pattern in Drosophila cells, two polyclonal antibodies were raised against recombinant SKPa. Both antisera predominantly recognize a single 24 kDa band in embryo, larval and adult extracts which was absent from skpA larval extracts(Fig. 6A and data not shown),indicating that the antisera specifically recognize SKPa.

Fig. 6.

SKPa protein is found in the cytoplasm and nucleus. (A) Western analysis of extracts from 4-8-hours-old embryos or newly hatched skpA+or skpA larvae with SKPa antisera. Approximately 30 larvae or 100 μl of embryos were used to make each extract. Relative loading was determined from a coomassie-stained gel prepared with the same samples. (B-D) Localization of SKPa in CNS and salivary gland cells. (B)Clones of skpA cells generated using FLP-mediated mitotic recombination were positively marked with GFP using the MARCM system,and stained with SKPa antisera. Only background staining is detected in skpA cells. (C) SKPa is found in the cytoplasm and nucleus of CNS cells, and fails to co-localize with centrosomes (cnn), DNA(TOTO-3) or mitotic chromosomes (arrowhead). SKPa is partially concentrated in the nucleus of some cells (arrow). (D) SKPa is found exclusively in the nucleus of salivary gland cells. B and D are single confocal sections, C is a projection of 5 μm of confocal sections. Bar, 30 μm (B,D) or 5 μm(C).

Fig. 6.

SKPa protein is found in the cytoplasm and nucleus. (A) Western analysis of extracts from 4-8-hours-old embryos or newly hatched skpA+or skpA larvae with SKPa antisera. Approximately 30 larvae or 100 μl of embryos were used to make each extract. Relative loading was determined from a coomassie-stained gel prepared with the same samples. (B-D) Localization of SKPa in CNS and salivary gland cells. (B)Clones of skpA cells generated using FLP-mediated mitotic recombination were positively marked with GFP using the MARCM system,and stained with SKPa antisera. Only background staining is detected in skpA cells. (C) SKPa is found in the cytoplasm and nucleus of CNS cells, and fails to co-localize with centrosomes (cnn), DNA(TOTO-3) or mitotic chromosomes (arrowhead). SKPa is partially concentrated in the nucleus of some cells (arrow). (D) SKPa is found exclusively in the nucleus of salivary gland cells. B and D are single confocal sections, C is a projection of 5 μm of confocal sections. Bar, 30 μm (B,D) or 5 μm(C).

Immunofluorescence analyses revealed that SKPa is localized throughout the cytoplasm and nucleus in diploid tissues such as the CNS and during embryogenesis (Fig. 6B,C and data not shown). Triple labeling cells for SKPa, DNA and centrosomes suggested that SKPa does not preferentially associate with centrosomes or chromosomes at any point during the cell cycle (Fig. 6C). SKPa was slightly more concentrated in the nucleus of some diploid cells, and was predominantly nuclear in salivary gland and fat body cells (Fig. 6C,D). No signal was observed in skpA cells from mosaic or homozygous mutant larvae (Fig. 6B and data not shown), demonstrating that the staining pattern specifically represents the localization of SKPa in CNS cells.

The relatively uniform distribution of Drosophila SKPa in diploid tissues is dramatically different from the centrosomal localization observed in vertebrate cells. In Drosophila, SKPa may regulate centrosome duplication by transiently associating with the centrosome; alternatively, it may function indirectly by acting on cytoplasmic or nuclear proteins. In contrast, the pronounced localization of SKPa to the nucleus of polyploid cells suggests that it may function directly on nuclear proteins involved in endoreduplication.

SkpA cells accumulate cyclin E

Extensive biochemical analyses in yeast and vertebrates have shown that Skp1 homologs primarily function as part of SCF complexes that regulate the ubiquitination and subsequent degradation of various proteins in the cell. One known target is cyclin E, which is degraded via an SCF complex in vitro and in vivo and is necessary for centrosome duplication in some vertebrate cell assays (Dealy et al., 1999; Hinchcliffe et al., 1999; Koepp et al., 2001; Lacey et al., 1999; Moberg et al., 2001; Strohmaier et al., 2001; Wang et al., 1999). Together,these data suggest a model in which skpA cells accumulate high levels of cyclin E that drive extra rounds of centrosome duplication.

To test this model, I quantified cyclin E levels at different points of the cell cycle using immunofluorescence correlated with nuclear DNA content and morphology. Wildtype cells in G1 phase (2C DNA content) showed low levels of cyclin E staining that were, on average, greater than the background staining observed in cycE cells(Fig. 7A,C), which probably reflects cyclin E beginning to accumulate in late-G1 cells. Cyclin E staining intensity was increased in S- and G2-phase cells, and was highest in mitotic cells (2.3-fold higher than in G1). Because cyclin E levels are low in G1 phase, cyclin E must normally be degraded at the end of mitosis (after anaphase) in the CNS.

Fig. 7.

skpA cells accumulate cyclin E. (A-B) Cyclin E staining in wildtype and skpA cells from larvae 3.5 days AED. Stronger cyclin E staining was detected in mitotic cells (M), and in some skpA cells (arrowheads). Most cytoplasmic proteins were extracted with these fixation conditions; samples prepared without extraction also had elevated levels of cyclin E in the cytoplasm. Cyclin E did not co-localize with centrosomes in either wildtype or skpA cells. Equivalent exposures of cyclin E staining are shown. Bar, 5 μm. (C) Quantification of cyclin E intensity relative to cell cycle stage. DNA content was determined from the total DAPI intensity of each nucleus and correlated with the average cyclin E intensity for the corresponding region. Mitotic cells with a 4C DNA content are shown separately. Significantly higher levels of cyclin E were detected in skpA cells with a 3C or more DNA content. Almost no signal was detected in cells from cycEk05007 larvae. (D)Western analysis of cyclin E levels in skpA+ and skpA larvae. Cyclin E was detected on a blot identical to the one used for Fig. 6A. Both maternal (77 kDa) and zygotic (67 kDa) cyclin E are detected in extracts from 4-8-hours-old embryos.

Fig. 7.

skpA cells accumulate cyclin E. (A-B) Cyclin E staining in wildtype and skpA cells from larvae 3.5 days AED. Stronger cyclin E staining was detected in mitotic cells (M), and in some skpA cells (arrowheads). Most cytoplasmic proteins were extracted with these fixation conditions; samples prepared without extraction also had elevated levels of cyclin E in the cytoplasm. Cyclin E did not co-localize with centrosomes in either wildtype or skpA cells. Equivalent exposures of cyclin E staining are shown. Bar, 5 μm. (C) Quantification of cyclin E intensity relative to cell cycle stage. DNA content was determined from the total DAPI intensity of each nucleus and correlated with the average cyclin E intensity for the corresponding region. Mitotic cells with a 4C DNA content are shown separately. Significantly higher levels of cyclin E were detected in skpA cells with a 3C or more DNA content. Almost no signal was detected in cells from cycEk05007 larvae. (D)Western analysis of cyclin E levels in skpA+ and skpA larvae. Cyclin E was detected on a blot identical to the one used for Fig. 6A. Both maternal (77 kDa) and zygotic (67 kDa) cyclin E are detected in extracts from 4-8-hours-old embryos.

As predicted by the model, some skpA cells accumulated higher levels of cyclin E than wildtype(Fig. 7B,C). SkpA cells in G1 and early S phase had cyclin E levels similar to wildtype; however, cells in late S, G2 and M phase stained 1.5 to 2-fold more intensely than similarly-staged wildtype cells. As in wildtype, cyclin E levels were highest in mitotic cells (4-fold higher than in G1). Cyclin E levels were also measured in extracts of newly-eclosed wildtype and skpA larvae by western blotting(Fig. 7D). Fivefold more cyclin E was detected in skpA larval extracts than wildtype, confirming that loss of skpA function results in the accumulation of cyclin E.

These results indicate that skpA function is required to properly regulate cyclin E levels in the CNS. However, the overall pattern of cyclin E accumulation during S, G2 and M phases and subsequent degradation at the end of mitosis is not perturbed. One possibility is that many skpA cells are arrested in G1 with low cyclin E levels while a subpopulation of cells manage to proceed through the cell cycle and produce two G1 cells with high cyclin E levels; however, this is unlikely to be the case because no small population of G1 cells with high cyclin E levels was observed. Therefore, CNS cells probably have a skpA-independent mechanism to degrade cyclin E at the end of mitosis,and only require skpA to prevent the accumulation of abnormally high levels of cyclin E during the cell cycle.

SkpA-induced centrosome overduplication is not suppressed by a mutation in cyclin E

If the elevated levels of cyclin E in skpA cells are necessary to induce centrosome overduplication, then mutations in cyclin E should suppress the skpA phenotype. To test this prediction, I reduced cyclin E levels with a P element allele{l(2)k05007, hereafter referred to as cycEk05007}that results in larval lethality and a growth defect similar to, but more severe than, loss of skpA function. Immunofluorescence staining of cyclin E verified that the cycEk05007 mutation dramatically reduces levels of cyclin E in all cells of the CNS(Fig. 7C). Most CNS cells in cycEk05007 larvae had a 2C DNA content suggesting that they are arrested in G1 phase. However, a few cells were able to proceed through the cell cycle and enter mitosis(Fig. 8C); these cells had slightly higher levels of cyclin E than seen in G1 phase cells(Fig. 7C), suggesting that some cycEk05007 cells still produce a small amount of cyclin E that is sufficient to proceed through the cell cycle. Therefore, cycEk05007 is a strong hypomorphic mutation that dramatically reduces but does not eliminate cyclin E from CNS cells.

Fig. 8.

A cyclin E mutation fails to suppress skpA-induced centrosome overduplication in the CNS from larvae 3.5 days AED. (A-B) Supernumerary centrosomes in skpA1; cycEk05007mitotic cells. Shown are cycEk05007 and skpA1; cycEk05007 mutant cells with two and six centrosomes, respectively. (C) Quantification of mitotic index and centrosome numbers in different genotypes. Mitotic index and centrosome numbers were calculated as in Fig. 3A and 5G, respectively. Bars, 5μm.

Fig. 8.

A cyclin E mutation fails to suppress skpA-induced centrosome overduplication in the CNS from larvae 3.5 days AED. (A-B) Supernumerary centrosomes in skpA1; cycEk05007mitotic cells. Shown are cycEk05007 and skpA1; cycEk05007 mutant cells with two and six centrosomes, respectively. (C) Quantification of mitotic index and centrosome numbers in different genotypes. Mitotic index and centrosome numbers were calculated as in Fig. 3A and 5G, respectively. Bars, 5μm.

Centrosome staining and quantification in cycEk05007and skpA1; cycEk05007 cells revealed that low levels of cyclin E are sufficient for centrosome overduplication. CycEk05007 cells that had proceeded into mitosis invariably had two centrosomes and replicated chromosomes(Fig. 8A,C), indicating that the low levels of cyclin E in cycling cycEk05007 cells are sufficient for centrosome duplication to occur. Mitotic cells in skpA1; cycEk05007 larvae were even rarer than in cycEk05007 larvae(Fig. 8C). Nonetheless,supernumerary centrosomes were observed in 56% of skpA1;cycEk05007 mitotic cells 3.5 days AED, similar to the 60%frequency seen in skpA1 cells(Fig. 8B,C). The difficulty of generating skpA1; cycEk05007 larvae precluded direct measurements of cyclin E levels; however, the fact that loss of skpA function did not increase the frequency of cycling cells compared to cycEk05007 larvae suggests that cyclin E levels are still limiting for entry into S phase and must be lower than in wildtype cells. Therefore, the elevated levels of cyclin E found in skpA cells are not necessary for centrosome overduplication to occur.

Discussion

The typical eukaryotic genome encodes a large repertoire of SCF components,suggesting that SCF complexes will regulate many diverse cellular and developmental processes. This study characterizes several cellular roles of skpA, the most abundantly expressed Drosophila skp1-related gene. SkpA performs essential roles in regulating centrosome duplication, endoreduplication, chromatin condensation, cell cycle progression and cyclin E accumulation. Undoubtedly, these functions represent only a subset of the processes regulated by skpA, and provide a framework for understanding the variety of pathways regulated by SKPa-based complexes.

SKPa functions as part of SCF ubiquitin ligases

Three lines of evidence suggest that SKPa primarily acts as part of multiple SCF ubiquitin ligase complexes. First, SKPa is highly similar to human and yeast Skp1, which form multiple SCF complexes in vitro and in vivo(Deshaies, 1999). Second, SKPa interacts with the Drosophila SCF homologs Cullin1 (Cul1),Supernumerary Limbs (Slimb) and Partner of Paired (Ppa) by in vitro or yeast two-hybrid assays (Bocca et al.,2001; Raj et al.,2000), indicating that it can form at least two types of SCF complexes. Third, mutations in the Drosophila F-box genes archipelago (ago) and slimb induce elevated cyclin E levels and centrosome overduplication, respectively(Moberg et al., 2001; Wojcik et al., 2000), similar to portions of the skpA mutant phenotype reported here. Thus, SKPa probably functions as a core component of SCFago,SCFslimb, SCFppa and potentially other SCF complexes in mediating the poly-ubiquitination and subsequent degradation of specific target proteins.

SkpA regulates centrosome duplication

SkpA mutant cells accumulate dramatic numbers of supernumerary centrosomes from multiple rounds of centrosome duplication in each cell cycle. Supernumerary centrosomes are first observed in some cells within one day after hatching, soon after the maternal supply of SKPa protein has been exhausted and before any growth defects or lethality are detected. Furthermore, centrosome overduplication occurs in mitotic clones demonstrating that it results from a cell autonomous function of skpA. Thus, extra centrosomes most probably accumulate directly from loss of SCF function and not as a secondary consequence of another skpA function such as cell cycle progression.

Several groups have proposed that centrosome overduplication in cancer cells may arise from aberrant accumulation of cyclin E(Hinchcliffe et al., 1999; Matsumoto et al., 1999; Nakayama et al., 2000). This hypothesis was attractive because centrosome duplication requires cdk2 function, activated by cyclin E or in some cells cyclin A(Hinchcliffe et al., 1999; Lacey et al., 1999; Matsumoto et al., 1999). Furthermore, overexpressed cyclin E associates with and is ubiquitinated by an SCF complex in human and Drosophila cells(Koepp et al., 2001; Moberg et al., 2001; Nakayama et al., 2000; Strohmaier et al., 2001; Yeh et al., 2001). Constitutive cyclin E overexpression in cultured mammalian cells induced little or no centrosome overduplication(Mussman et al., 2000; Spruck et al., 1999); however,the immortal cell lines used in the studies may have accumulated mutations which suppressed aberrant centrosome duplication, as is seen in p53–/– mouse epithelial cells in late passages(Chiba et al., 2000).

I have directly tested the role of cyclin E in centrosome overduplication by genetically manipulating cyclin E levels in wildtype and skpA cells. Strikingly, drastically reducing cyclin E levels with a near-null allele does not suppress centrosome overduplication in cycling skpA cells. One possibility is that cyclin E is not required for centrosome duplication in Drosophila. This seems unlikely, because Drosophila cdk2 does not associate with cyclin A and lacks in vitro kinase activity when immunoprecipitated from cyclin E-deficient embryos, and other functions of cdk2 are conserved between Drosophila and vertebrates(Knoblich et al., 1994; Lane et al., 2000; Sauer et al., 1995). In any case, centrosome overduplication occurs independently of SCF control of cyclin E accumulation.

How do SCF components regulate centrosome duplication? One possibility is that simply lengthening the cell cycle introduces enough time for multiple cycles of centrosome duplication to occur. Although this model cannot be ruled out, it seems unlikely given that a centrosome must duplicate in as little as 55 minutes in a cycling neuroblast but does not reduplicate in the 12-hour cycle of an imaginal wing disc cell(Neufeld et al., 1998; Truman and Bate, 1988; Vidwans et al., 2003). Furthermore, slowing the cell cycle in abdominal histoblasts by overexpressing the Drosophila retinoblastoma-family protein RBF is not sufficient to induce centrosome overduplication (Fung et al., 2002).

Instead, I favor the idea that a target of SCF-mediated degradation acts as a Centrosome Licensing Factor (CLiF) that limits centrosome duplication to once per cell cycle. CLiF would be expressed early in the cell cycle, loaded onto centrosomes, and excess CLiF would be targeted to the proteasome by an SCF complex. One cycle of centrosome duplication could then be triggered by Cdk2-E activity, but the daughter centrosomes would not be relicensed until the next cell cycle. SCF mutants would fail to degrade excess CLiF, allowing duplicated centrosomes to relicense and reduplicate in the course of a single cell cycle. One candidate CLiF is nucleophosmin/B23, which is phosphorylated by Cdk2-E and associates specifically with unduplicated centrosomes(Okuda et al., 2000). Future experiments will need to determine if nucleophosmin/B23 or other candidate CLiFs are targeted for degradation by an SCF complex.

In Xenopus, antibody-addition experiments using an in vitro assay suggested that Skp1 is required for centriole separation(Freed et al., 1999). The results presented here clearly demonstrate that centrioles can separate and duplicate in the absence of Drosophila SKPa. These contrasting results may reflect a functional difference between Drosophila and Xenopus, potentially related to the difference in SKPa/Skp1 localization to the centrosome in these two organisms. Alternatively, Skp1 antibodies may block centriole separation in a way that does not reflect an in vivo requirement for SCF activity. I favor this second possibility because immunodepletion of Skp1 from Xenopus extracts did not block centriole separation (Freed et al.,1999). Determining if Skp1 serves an additional role in vertebrate centriole separation will require genetic analyses in a vertebrate model system.

Pseudo-bipolar spindles form with extra centrosomes

Remarkably, the large numbers of supernumerary centrosomes in skpA cells typically do not generate multipolar spindles in mitosis. The extra centrosomes are probably not defective, because most centrosomes can efficiently nucleate microtubules in prometaphase and one cell examined by electron microscopy had multiple centrioles apparently undergoing duplication. Furthermore, anaphase cells were increasingly rare and had fewer centrosomes than metaphases in older animals, suggesting that cells with many centrosomes delayed or arrested in metaphase. This differs from many cell types in which extra centrosomes frequently led to the formation of multipolar spindles (Heneen,1975; Pera and Rainer,1973; Rieder et al.,1997; Sluder et al.,1997), although mouse neuroblastoma (N115) cells and p53–/– mouse embryonic fibroblasts with extra centrosomes typically form bipolar spindles(Fukasawa et al., 1996; Ring et al., 1982). Also, one or two extra centrosomes in sea urchin zygotes or PtK1 cells do not delay anaphase onset (Sluder et al.,1997).

How do skpA neuroblasts form bipolar spindles with extra centrosomes? The centrosomes appear to be dynamically rearranged during the mitotic cycle so that the majority are clustered into two cooperative poles, potentially through the action of microtubule bundling proteins such as the nuclear mitotic apparatus protein (NuMa) and the kinesin Ncd (Gaglio et al., 1997; Matthies et al., 1996). This ability to rearrange centrosomes into two poles does not require additional genetic mutations, as has been proposed for mammalian cells(Hinchcliffe and Sluder,2001). Instead, it may reflect an inherent preference for Drosophila neuroblasts to form bipolar spindles; alternatively, loss of skpA function may result in the upregulation of compensatory proteins. It is unclear how the presence of many centrosomes delays anaphase onset. Further studies are needed to determine if this indicates a novel way to activate the spindle assembly checkpoint or the presence of another checkpoint governing anaphase onset.

SCF function and cancer

The roles of SCF complexes in governing centrosome duplication and the cell cycle may be important for understanding tumorigenesis. Many solid tumors accumulate supernumerary centrosomes which are thought to contribute to cancer progression (reviewed by Brinkley,2001), suggesting that upregulation of the proposed centrosome-licensing factor may be oncogenic. Recently, the human homolog of the F-box gene ago, hCdc4, was reported to be mutated in several human breast and ovarian cancer cell lines with high cyclin E levels(Moberg et al., 2001; Strohmaier et al., 2001). Levels of the F-box protein Skp2 are upregulated in some oral carcinomas and inversely correlate with levels of the tumor suppressor p27(Gstaiger et al., 2001). Future studies will need to determine if other human SCF components including Skp1 are also mutated in cancer cells. Further analyses of the functions of Drosophila skpA will help to elucidate how SCF-mediated protein degradation may be a key mechanism governing centrosome duplication, cell proliferation and cancer progression.

Acknowledgements

I thank Craig Garafola and Cameron Kennedy for providing valuable technical assistance; Tom Kaufmann, Doug Koshland, Todd Laverty, Christian Lehner,Helena Richardson, Pernille Rørth, Allan Spradling and Yixian Zhang for supplying antibodies and fly stocks; and Audrey Huang and Doug Koshland for useful comments on the manuscript. I am indebted to Gary Karpen in whose lab these studies were initiated. This work was funded by the Carnegie Institution of Washington and the NIH.

References

Andersen, S. S. (
1999
). Molecular characteristics of the centrosome.
Int. Rev. Cytol.
187
,
51
-109.
Bocca, S. N., Muzzopappa, M., Silberstein, S. and Wappner,P. (
2001
). Occurrence of a putative SCF ubiquitin ligase complex in Drosophila.
Biochem. Biophys. Res. Commun.
286
,
357
-364.
Bonaccorsi, S., Giansanti, M. G. and Gatti, M.(
2000
). Spindle assembly in Drosophila neuroblasts and ganglion mother cells.
Nat. Cell Biol.
2
,
54
-56.
Borel, F., Lohez, O. D., Lacroix, F. B. and Margolis, R. L.(
2002
). Multiple centrosomes arise from tetraploidy checkpoint failure and mitotic centrosome clusters in p53 and RB pocket protein-compromised cells.
Proc. Natl. Acad. Sci. USA
99
,
9819
-9824.
Brinkley, B. R. (
2001
). Managing the centrosome numbers game: from chaos to stability in cancer cell division.
Trends Cell Biol.
11
,
18
-21.
Chiba, S., Okuda, M., Mussman, J. G. and Fukasawa, K.(
2000
). Genomic convergence and suppression of centrosome hyperamplification in primary p53–/–cells in prolonged culture.
Exp. Cell Res.
258
,
310
-321.
Dealy, M. J., Nguyen, K. V., Lo, J., Gstaiger, M., Krek, W.,Elson, D., Arbeit, J., Kipreos, E. T. and Johnson, R. S.(
1999
). Loss of Cul1 results in early embryonic lethality and dysregulation of cyclin E.
Nat. Genet.
23
,
245
-248.
Deshaies, R. J. (
1999
). SCF and Cullin/Ring H2-based ubiquitin ligases.
Annu. Rev. Cell Dev. Biol.
15
,
435
-467.
Feldman, R. M., Correll, C. C., Kaplan, K. B. and Deshaies, R. J. (
1997
). A complex of Cdc4p, Skp1p, and Cdc53p/cullin catalyzes ubiquitination of the phosphorylated CDK inhibitor Sic1p.
Cell
91
,
221
-230.
Freed, E., Lacey, K. R., Huie, P., Lyapina, S. A., Deshaies, R. J., Stearns, T. and Jackson, P. K. (
1999
). Components of an SCF ubiquitin ligase localize to the centrosome and regulate the centrosome duplication cycle.
Genes Dev.
13
,
2242
-2257.
Fukasawa, K., Choi, T., Kuriyama, R., Rulong, S. and Vande Woude, G. F. (
1996
). Abnormal centrosome amplification in the absence of p53.
Science
271
,
1744
-1747.
Fung, S. M., Ramsay, G. and Katzen, A. L.(
2002
). Mutations in Drosophila myb lead to centrosome amplification and genomic instability.
Development
129
,
347
-359.
Gaglio, T., Dionne, M. A. and Compton, D. A.(
1997
). Mitotic spindle poles are organized by structural and motor proteins in addition to centrosomes.
J. Cell Biol.
138
,
1055
-1066.
Gatti, M. and Baker, B. S. (
1989
). Genes controlling essential cell-cycle functions in Drosophila melanogaster.
Genes Dev.
3
,
438
-453.
Gstaiger, M., Jordan, R., Lim, M., Catzavelos, C., Mestan, J.,Slingerland, J. and Krek, W. (
2001
). Skp2 is oncogenic and overexpressed in human cancers.
Proc. Natl. Acad. Sci. USA
98
,
5043
-5048.
Gstaiger, M., Marti, A. and Krek, W. (
1999
). Association of human SCF(SKP2) subunit p19(SKP1) with interphase centrosomes and mitotic spindle poles.
Exp. Cell Res.
247
,
554
-562.
Heneen, W. K. (
1975
). Kinetochores and microtubules in multipolar mitosis and chromosome orientation.
Exp. Cell Res.
91
,
57
-62.
Heuer, J. G., Li, K. and Kaufman, T. C. (
1995
). The Drosophila homeotic target gene centrosomin (cnn) encodes a novel centrosomal protein with leucine zippers and maps to a genomic region required for midgut morphogenesis.
Development
121
,
3861
-3876.
Hinchcliffe, E. H., Li, C., Thompson, E. A., Maller, J. L. and Sluder, G. (
1999
). Requirement of Cdk2-cyclin E activity for repeated centrosome reproduction in Xenopus egg extracts.
Science
283
,
851
-854.
Hinchcliffe, E. H. and Sluder, G. (
2001
). `It Takes Two to Tango': understanding how centrosome duplication is regulated throughout the cell cycle.
Genes Dev.
15
,
1167
-1181.
Knoblich, J. A., Sauer, K., Jones, L., Richardson, H., Saint, R. and Lehner, C. F. (
1994
). Cyclin E controls S phase progression and its down-regulation during Drosophila embryogenesis is required for the arrest of cell proliferation.
Cell
77
,
107
-120.
Koepp, D. M., Schaefer, L. K., Ye, X., Keyomarsi, K., Chu, C.,Harper, J. W. and Elledge, S. J. (
2001
). Phosphorylation-dependent ubiquitination of cyclin E by the SCFFbw7ubiquitin ligase.
Science
294
,
173
-177.
Lacey, K. R., Jackson, P. K. and Stearns, T.(
1999
). Cyclin-dependent kinase control of centrosome duplication.
Proc. Natl. Acad. Sci. USA
96
,
2817
-2822.
Lane, M. E., Elend, M., Heidmann, D., Herr, A., Marzodko, S.,Herzig, A. and Lehner, C. F. (
2000
). A screen for modifiers of cyclin E function in Drosophila melanogaster identifies cdk2 mutations, revealing the insignificance of putative phosphorylation sites in cdk2.
Genetics
155
,
233
-244.
Matsumoto, Y., Hayashi, K. and Nishida, E.(
1999
). Cyclin-dependent kinase 2 (Cdk2) is required for centrosome duplication in mammalian cells.
Curr. Biol.
9
,
429
-432.
Matthies, H. J., McDonald, H. B., Goldstein, L. S. and Theurkauf, W. E. (
1996
). Anastral meiotic spindle morphogenesis: role of the non-claret disjunctional kinesin-like protein.
J. Cell Biol.
134
,
455
-464.
Meraldi, P., Honda, R. and Nigg, E. A. (
2002
). Aurora-A overexpression reveals tetraploidization as a major route to centrosome amplification in p53–/– cells.
EMBO J.
21
,
483
-492.
Meraldi, P., Lukas, J., Fry, A. M., Bartek, J. and Nigg, E. A. (
1999
). Centrosome duplication in mammalian somatic cells requires E2F and Cdk2-cyclin A.
Nat. Cell Biol.
1
,
88
-93.
Meraldi, P. and Nigg, E. A. (
2002
). The centrosome cycle.
FEBS Lett.
521
,
9
-13.
Miller, D. F. B., Holtzman, S. L. and Kaufman, T. C.(
2002
). Customized microinjection glass capillary needles for P-element transformations in Drosophila melanogaster.
BioTechniques
33
,
366
-375.
Moberg, K. H., Bell, D. W., Wahrer, D. C., Haber, D. A. and Hariharan, I. K. (
2001
). Archipelago regulates Cyclin E levels in Drosophila and is mutated in human cancer cell lines.
Nature
413
,
311
-316.
Mussman, J. G., Horn, H. F., Carroll, P. E., Okuda, M.,Tarapore, P., Donehower, L. A. and Fukasawa, K. (
2000
). Synergistic induction of centrosome hyperamplification by loss of p53 and cyclin E overexpression.
Oncogene
19
,
1635
-1646.
Nakayama, K., Nagahama, H., Minamishima, Y. A., Matsumoto, M.,Nakamichi, I., Kitagawa, K., Shirane, M., Tsunematsu, R., Tsukiyama, T.,Ishida, N. et al. (
2000
). Targeted disruption of Skp2 results in accumulation of cyclin E and p27Kip1, polyploidy and centrosome overduplication.
EMBO J.
19
,
2069
-2081.
Neufeld, T. P., de la Cruz, A. F., Johnston, L. A. and Edgar, B. A. (
1998
). Coordination of growth and cell division in the Drosophila wing.
Cell
93
,
1183
-1193.
Okuda, M., Horn, H. F., Tarapore, P., Tokuyama, Y., Smulian, A. G., Chan, P. K., Knudsen, E. S., Hofmann, I. A., Snyder, J. D., Bove, K. E. et al. (
2000
). Nucleophosmin/B23 is a target of CDK2/cyclin E in centrosome duplication.
Cell
103
,
127
-140.
Pera, F. and Rainer, B. (
1973
). Studies of multipolar mitoses in euploid tissue cultures. I. Somatic reduction to exactly haploid and triploid chromosome sets.
Chromosoma
42
,
71
-86.
Raj, L., Vivekanand, P., Das, T. K., Badam, E., Fernandes, M.,Finley, R. L., Brent, R., Appel, L. F., Hanes, S. D. and Weir, M.(
2000
). Targeted localized degradation of Paired protein in Drosophila development.
Curr. Biol.
10
,
1265
-1272.
Rieder, C. L., Khodjakov, A., Paliulis, L. V., Fortier, T. M.,Cole, R. W. and Sluder, G. (
1997
). Mitosis in vertebrate somatic cells with two spindles: implications for the metaphase/anaphase transition checkpoint and cleavage.
Proc. Natl. Acad. Sci. USA
94
,
5107
-5112.
Ring, D., Hubble, R. and Kirschner, M. (
1982
). Mitosis in a cell with multiple centrioles.
J. Cell Biol.
94
,
549
-556.
Sauer, K., Knoblich, J. A., Richardson, H. and Lehner, C. F.(
1995
). Distinct modes of cyclin E/cdc2c kinase regulation and S-phase control in mitotic and endoreduplication cycles of Drosophilaembryogenesis.
Genes Dev.
9
,
1327
-1339.
Skowyra, D., Craig, K. L., Tyers, M., Elledge, S. J. and Harper,J. W. (
1997
). F-box proteins are receptors that recruit phosphorylated substrates to the SCF ubiquitin-ligase complex.
Cell
91
,
209
-219.
Sluder, G. and Hinchcliffe, E. H. (
1998
). The apparent linkage between centriole replication and the S phase of the cell cycle.
Cell Biol. Int.
22
,
3
-5.
Sluder, G., Thompson, E. A., Miller, F. J., Hayes, J. and Rieder, C. L. (
1997
). The checkpoint control for anaphase onset does not monitor excess numbers of spindle poles or bipolar spindle symmetry.
J. Cell Sci.
110
,
421
-429.
Spruck, C. H., Won, K. A. and Reed, S. I.(
1999
). Deregulated cyclin E induces chromosome instability.
Nature
401
,
297
-300.
Strausfeld, U. P., Howell, M., Descombes, P., Chevalier, S.,Rempel, R. E., Adamczewski, J., Maller, J. L., Hunt, T. and Blow, J. J.(
1996
). Both cyclin A and cyclin E have S-phase promoting (SPF)activity in Xenopus egg extracts.
J. Cell Sci.
109
,
1555
-1563.
Strohmaier, H., Spruck, C. H., Kaiser, P., Won, K. A., Sangfelt,O. and Reed, S. I. (
2001
). Human F-box protein hCdc4 targets cyclin E for proteolysis and is mutated in a breast cancer cell line.
Nature
413
,
316
-322.
Truman, J. W. and Bate, M. (
1988
). Spatial and temporal patterns of neurogenesis in the central nervous system of Drosophila melanogaster.
Dev. Biol.
125
,
145
-157.
Vidwans, S. J., Wong, M. L. and O'Farrell, P. H.(
2003
). Anomalous centriole configurations are detected in Drosophila wing disc cells upon Cdk1 inactivation.
J. Cell Sci.
116
,
137
-143.
Wang, Y., Penfold, S., Tang, X., Hattori, N., Riley, P., Harper,J. W., Cross, J. C. and Tyers, M. (
1999
). Deletion of the cul1 gene in mice causes arrest in early embryogenesis and accumulation of cyclin E.
Curr. Biol.
9
,
1191
-1194.
Wojcik, E. J., Glover, D. M. and Hays, T. S.(
2000
). The SCF ubiquitin ligase protein Slimb regulates centrosome duplication in Drosophila.
Curr. Biol.
10
,
1131
-1134.
Yeh, K. H., Kondo, T., Zheng, J., Tsvetkov, L. M., Blair, J. and Zhang, H. (
2001
). The F-Box protein SKP2 binds to the phosphorylated threonine 380 in cyclin E and regulates ubiquitin-dependent degradation of cyclin E.
Biochem. Biophys. Res. Commun.
281
,
884
-890.
Zheng, Y., Jung, M. K. and Oakley, B. R.(
1991
). Gamma-tubulin is present in Drosophila melanogaster and Homo sapiens and is associated with the centrosome.
Cell
65
,
817
-823.